T6c gid: Joseph F. Hair, JR.

- Marko Sarstedt
Christian M. Ringle - Siegfried P. Gudergan
Bi6n dich: Nguyen Quang Anh

N h u n g van __

MO HiNH PHUONG T I^ H
CAU TRUC BiNHd
TOI TH1EU PHAN

dvanced Issues in
Partial Least Squares
Structural Equation Modeling

NHA XUAT BAN TA I C H IN H

 Advanced Issues
in P artial Least Squares
S tru ctu ral Equation
M odeling

,o v
Joseph F. Hair, Jr.
University o f South Alabama, USA

M arko Sarstedt
Otto-von-Guericke-University
Magdeburg, Germany, and University
o f Newcastle, Australia

Christian M. Ringle
Hamburg University o f Technology,
Germany, and University o f
Newcastle, Australia
/ * V
Siegfried P. Gudergan
University o f Newcastle, Australia

tS A G E
Los Angeles | London | New Delhi
Singapore | Washington DC | Melbourne
 tf

< ?
<o

<b
 Nhung van de nang cao
trong mo hinh phUo'ng trinh
cau true binh phUo'ng toi thieu
rieng phan

Thanh vien tham gia Bien dich:

h:
■(C'
1. TS. Nguyen Quang Anh
2. TS. Nguyln Ngoc Tuyen
3. TS. Pham Quoc Luyen
4. TS. Nguyen Ngoc Duy Phuong
5. TS. Nguyen Thanh Vu
TS. Cao Quoc Viet
ThS. Huvnh Trong Hien

nh A x u a t b a n tA i c h in h
 sS>
s v
&

< ?
<o

<b
Loi gioi thieu

Trong 5 nam gan day PLS-SEM dbpc sb dung rong rai trong cac
nghien cbu ve thi trbbng, m arketing, du lich lb hanh, quan tri chien
lbpc, he thong thong tin va db lieu lbn (big data). Cac nghien cbu gan
day tap trung nghien cbu viec hieu ro hon v& giai thich vai trb cua cac
bien so, moi quan he can thiep cung n h b ngau nhien giba cac bien so m a
cac nha nghien cuu quan tarn. Sb chu y lorn hon den nhbng vain d e bat
ngb diioc phan anh trong nhiing khac biet lien quan den sii p h an loai
cac phan nhom thuoc ve con ngiicti, to chtic hoac m oi tniang. De hieu
cac tin h h u ong bat ngb n h ii vay doi h 6i phai dan h gia stt khong dong
n hat quan sat diipc hoac khong quan sat diipc.
Sit xuat hien cua cac yeu cau ve m o hinh hoa phiic tap h o n se di
cung va nhain m anh tarn quan trong cua cac phifPng phap phan tich
nang cao. NhQng tien bo dang ghi nhan trong PLS-SEM, bao gom phan
tich bo td khang dinh de danh gia thUc nghiem m o hin h do Itfdng, ty
so heterotrait-m onotrait cua cac moi tiiong quan nhtf m ot each tiep
can m bi de kiem dinh gia trj phan biet, phan tich phan doan theo dinh
hiidng dit bao de xac dinh va xu Iy sb khong dong n hat khong quan sat
diipc, cac dang phan tich da nhom khac nhau va kiem djnh bait bien
bang phbong phap bat bien do liibng theo each ket hop cac m o hinh.

Nhieu ky th u at p h an tich cao cap nhii xb Iy sp khong dong nhat

khong quan sat dppc, so sanh da nhom , bat bien va phan tich bo tb
khang djnh da cb trong SEM dpa tren hiep phbong sai (CB-SEM),
nhbng gan day chi cb cac nha nghien cbu ve p hbo n g phap m bi bat dau
gibi th ieu chung trong PLS-SEM. Dong thbi, cac p h b o n g phap tiep can
k h a e n h b phan tichbieu do tarn quan trong - hieu suait va cac phan tich
dieu tiet vbi bien lien tuc chi diipc th b c hien duy nhat qua p h a n mem
PLS-SEM.

Lpi ich cua viec cb cac phbo n g p h ap tiep can PLS-SEM nang cao vo
cung lbn, vi cac dang phan tich nay ho trp danh gia cac bbc lb p n g PLS-
SEM. Dong thbi, viecap dung cac p h b o n g p h ap tiep can PLS-SEM nang
cao nay doi hoi phai hieu dbpc n hbng phbc tap va biet khi nao chung cb
the ho trp phan tich d b lieu mot each co y nghla sao cho cac b n g dung
phu hop vbi ngb canh nghien cuu.

v
 C hung toi xin gtii Idi cam o n ve nhQng thong tin quy bau va nhan
xet chan th anh tii cac dong nghiep. D ac biet, chung toi m uon giii Idi
cam On sau sSc nhat den N ha xuat ban SAGE da dong y va tao dieu kien
thuan loi cho chung toi phat h a n h cuon sach nay.

M ac dii co rat nhieu co gang nhiing chung toi cung rat can cac gop
y va p h an hoi tii cac n h a nghien cd u va ban dQC de cac thuat ngfl duoc
dich sat h o n va de hieu hon. M pi gop y va phan hoi cua Q uy doc gia xin
vui long giii ve dia chi email: nqanh76@ gm ail.com

Rat m ong n h an diipc nhieu y kien gop y va p h an hoi cua Quy doc
gia de cac lan phat hanh tiep theo diipc sifa chufa va cap nhat to t hPn.

Tran trpng cam on./.

TM. N hom Bien dich

TS. N guyen Q uang Anh

Noi dung tom tat

L a i g ia i thieu
Leri noi dau
G iai thieu ve cac tac gia
V C/
Chu'o’ng 1: Tdng quan ve s ir phat trien gan day va
dang noi len trong PLS-SEM
Chu'o'ng 2: Mo hinh nang cao
Chu*o*ng 3: Danh gia mo hinh nar o
Chu'o'ng 4: Mo hinh hoa tinh khong dong nhat quan
sat d ira c
Chu'o’ng 5: Mo hinh hoa tinh khong dong nhat khong
quan sat du’o’c

Giai th ich thuat ngu*

Tai lieu tham khao
Noi dung chi tiet

L<yi gid*i thieu ............................................................................ v

Ld'i noi d a u ................................................................................xiii
Gio'i thieu ve cac tac g ia ................ ..................................

Chu’o'ng 1: Tong quan ve s ir phat trien gan day va 0

dang noi len trong PLS-SEM ............................................. 1
Tong quan chucmg ................................................................... I
N guon goc va sir tien trien cua mo hinh phuong trinh cau
true binh phuong t6 i thieu rieng phan ........................................2
Do l u o n g .............................................................. 5
Cac bien khai niem va bien dai d i e n , ........................................5
V I
Cac mo hinh do l u o n g ...............................................8
U*6c luong mo h i n h ..........................................................15
PLS-SEM hay C B - S p 4 ^ * W . ...............................................................19
T rilt ly do lu o n g ........................................................................... 19
M uc tieu cua phan tich ........................................................................ 22
C ach to chuc cac chucmg con l a i ............................................................24
Tinh huong nghien cuu m inh hoa ......................................................... 27
Mo hinh Danh tieng cong t y .............................................................. 27
Phan m em P L S-SE M ...........................................................................30
T hiet lap m o hinh trong S m artP L S ..................................................31
m t4t.......................................................................................................... 33
C au hoi on ta p ............................................................................................. 35
Cau hoi tu duy phan b ie n ..........................................................................35
Thuat n g u ......................................................................................................36
Tai lieu doc them .......................................................................................37

Chu'o’ng 2: Mo hinh nang c a o .............................................. 39

Tong quan c h u o n g ..................................................................................... 39
Cac mo hinh thanh phan thu b a c ............................................................40
Thuat ngu va D ong luc thiic d a y ...................................................... 40
 Nhirng can nhic v£ mat khai n ie m ..................................................43
N hung can n h ic vS mat ky th u a t......................................................50
Tong q u a n ....................................................................................... 50
Phicong phap tiep can cac bien quan sa t tai la p ....................50
Cdch tiep can hai giai d o q n ......................................................... 56
Nhirng de xuat va quy tic kinh n g h iem ..........................................57
Tinh huong nghien cuu m inh h o a ................................................
H C M ket qua - k it qua........................................................
H C M k ct qua - nguyen nhan ........................................ (!?}}..... 65
Nhirng moi quan he phi tuyin.
M o hinh hoa tac dong bac hai trong PLS-SEP
Danh gia cac tac dong phi tu y in ... as
D iin giai ket q u a ..............................
Tinh huong nghien cuu minh h o a ......
Tom ta t.....................................................|
Cau hoi on ta p ............................... ^ .....
Cau hoi tu duy phan bien.,1^ ......... .
Thuat n g u ......................4*
Tai lieu doc them ...J ...v\

Chu'o'ng 3: Danh gia Mo hinh nang c a o .............................. 91

T ong quan c h u c m g ................................................................................... 9 1
Phan tich bo tir khang djnh .....................................................................92
Phuong p h a p ........................................................................................ 92
Tinh huong nghien cuu m inh hoa - Phan tich bo tu khang
d in h ............................................................................................ 101
Phan tich Bieu do tarn quan trong - hieu s u a t..................................109
Tong quan ...........................................................................................109
Thuc hien IPM A mot each he th o n g ..............................................113
B u d c 1: K iem tra cac y e n cau ................................................. 113
B udc 2: Tinh toan cac gia tri ve hieu s u a t............................ 114
B udc 3 : Tinh toan cac g ia tri ve 1am quan tro n g .................119
B udc 4: Tao bieu do tarn quan trong - hieu su a t.................121
B udc 5: M o rong IPMA d cap do bien quan s a t..................123
Tinh huong nghien cuu minh hoa - Phan tich IP M A ................125
 Tom tat 135
Cau-h o io n t|^,,™nww*mmnBmnsnm-.sraBranm-................:............. 136
Cau hoi tu duy phan b ie n ........................................................................137
Thuat n g u ...................................................................................................137
Tai lieu doc th e m ......................................................................................138

Chu'o'ng 4: Mo hinh hoa tinh khong dong nhat quan

sat d u 'o 'c ..............................................................................139
Tong quan c h u o n g ........................................................................... 139
Tinh khong dong nhat quan sat dugc va khong d o n g nhat J
khong quan sat d u g c .........................................................!..... .........140
Kiem djnh bat bien mo hinh do lu o n g ............................. *...........144
B udc 1: Bat bien cau h in h ............ . .................................. 147
B udc 2: B at bien thanh p h a n ......................................................... 148
B udc 3: T uong duong trj trung binh tdng hgp va phuong s a i... 151
Phan tich da n h o m ............................................................................ 152
Kiem djnh tham so ..........................................153
Kiem djnh phi tham so ..........................................................155
P L S -M G A ..........................................................155
Kiem dinh hoan v jf .......................................................................156
So sanh nhieu hon hai nhom ............................................................ 159
K iem dinh s y khac biet tong t h e .............................................. 160
So sanh theo c a p ............................................................................162
Tinh huong nghien cmi m inh hoa - danh gia bat bien va phan
tich d a n h o m ........................................................................................ 164
Tom t a t ...................................................................................................... 175
Cau hoi on t a p .......................................................................................... 177
Cau hoi tu duy phan b i e n .......................................................................177
<& Thuat n g u .................................................................................................. 177
Tai lieu doc them ................................................................................... 179

Chu'o'ng 5: Mo hinh hoa tinh khong dong n h a t khong

quan sat d u 'o 'c .................................................................... 181
Tong quan c h u o n g .................................................................................. 18 1
Kham pha tinh khong dong nhat khong quan sat dugc
trong cac mo hinh dudng dan P L S ...........................................182
 B u o c 1: Chay thu tuc F IM IX -P L S ................................................ 186
B uoc 2: Xac dinh so luong phan d o a n ......................................... 188
B uoc 3: Chay thu tuc P L S -P O S .....................................................191
B uoc 4: Giai thich cau true phan doan tiem a n .......................... 197
B uoc 5: U oc luong cac m o hinh phan doan cu t h e ...................198
T inh huong nghien cuu m inh hoa - phan tich lop tiem a n 200
B uoc 1: Chay thu tuc F IM IX -P L S ................................................200
B uoc 2: Xac djnh so luong phan d o a n ........................................ 203
B uoc 3: Chay thu tuc P L S -P O S ........................................ 206
B uoc 4: Giai thich cau true phan doan tiem a n ................ 208
B uoc 5: Uoc luong cac m o hinh phan doan cu t h e .................. 211
Tom t a t ................................................................... ~y«.........*................215
Cau hoi on t a p ............................................................... 217
Cau hoi tu duy phan b i e n .................. y v O s ...............................^18
Thuat n g u ................................................................. 218
Tai lieu doc them

Giai thich thuat ngu*. / T '

Tai lieu tham khao

 sS>
s v
&

< ?
<o

<b
Loi noi dau

Cac n h a nghien ciiu trong linh vUc quan ly, tiep thi, he thong thong tin
quan ly va cac nganh khoa hoc xa hoi khac ngay cang tap tru n g tim hieu
sau hem ve m oi quan he phiic tap tao thanh “hop den” cua nhieu khi'a
canh thuoc ve to chiic va hanh vi. Mo hin h phUdng trin h cau tru e binh
philctng toi thieu rieng phan (PLS-SEM) da cho phep cac nha nghien
ciiu ti&c tinh dong then cac m oi lien he phiic tap lien quan d en nhieu
cau true va cac chi so v6 i cac moi quan he triic tiep, gian tiep hoac kiem
d in h m a neu khong co no se khong the d l dang got roi va kiem tra. Tuy
nhien, cac nghien ciiu gan day tap trung vao viec hieu ro h d n va giai
thich vai tro cua cac bien so, m oi quan he can thiep cung nhU ngau
nh ien giQa cac bien so m a cac nha nghien ciiu quan tarn. Vi du, sU chu
y Ion hon da dupe dat vao nhUng van de bat ngcJ duqc phan an h trong
nhQng khac biet lien quan den sU phan loai cac phan nhom thuoc ve con
ngUOi, to chdc hoac moi trUong. De hieu cac tinh huong bat ngb nhU
vay doi hoi phai danh gia sU khong dong nhat duqc quan sat hoac khong
dUOc quan sat. Trong mot dac tin h tu o n g tU, m ot khai niem chung nhan
ra rang cac hieu ling khong nhat thiet phai lien tuc nhUng no co the lam
giam hoac tang len nhU vay m a cac nha nghien ciiu can phai vUOt ra
ngoai m o hinh tuyen tinh den mo hinh phi tuyen.
SU xuat hien cua cac yeu cau ve m o hinh hoa phiic tap h o n se di
cung va nhan m anh tarn quan trong cua cac phUOng phap phan tich
nang cao. Nhflng tien bo dang ghi nhan trong PLS-SEM bao gom , vi du:

• phan tich bo tii khang dinh (confirm atory tetrad analysis) de

danh gia thUc nghiem m o hin h do ludng (G udergan, Ringle,
VVende, & Will, 2008),

• ty so heterotrait-m onotrait cua cac m oi tUdng quan nhU mot

each tiep can m di de kiem dinh gia tri phan biet (Henseler,
Ringle, & Sarstedt, 2015),

• phan tich phan doan theo dinh hudng du bao de xac dinh va
xd ly sU khong dong nhat khong quan sat duqc (Becker, Rai,
Ringle, & Volckner, 2013),
. cac dang phan tich da nhom khac nhau (Sarstedt, Henseler, &
Ringle, 2011), va

xiii
xiv NhCmg van (Jb nang cao trong PLS-SEM

• kiem dinh bat bien bang phbdng phap do bat bien do lbbng
theo cach-ket hop cao m o-binh-(-Henseler, Ringle, &-Sarstedt,
2016).
N hieu ky thuat phan tich cao cap n h b xb ly sb khong dong nhat
khong dboc quan sat, so sanh nhieu nhom , bat bic'n, va phan tich bo
tb khang dinh da co trong SEM dba tren hiep p h b d n g sai (CB-SEM),
n h b n g gan day chi co cac nha nghien ciiu ve p h b d n g phap bat dau gicfi
th ieu chung trong m ot ngii canh PLS- SEM. D ong thbi, cac phiidng
phap tiep can khac n h d phan tich bieu do tarn q u an tro n g - hieu suat v&
cac phan tich dieu tiet vbi bien lien tuc chi dbdc th d e hien duy nhat qua
phan m em PLS-SEM.
Ldi ich cua viec co cac phiidng phap tiep can PLS-SEM nang cao vo
cung lbn, vi cac dang phan tich nay ho trd danh gia cac libc lbdng PLS-
SEM va ngay cang dbdc cac nha bien tap va phan bien yeu cau. Dong thbi,
viec ap dung cac phiidng phap tiep can PLS-SEM tien tien nay doi hoi
phai hieu diidc n hbng phbc tap cua chung va biet k h i nao ho co the ho
trd p h an tich db lieu m ot ca ch co y nghia sao cho cac bng dung phu hop
vbi ngb canh nghien cbu.
^ y
C ung vbi sb phat trien cua cac ky thuat cao cap, cac nghien cbu gan
day da chbng kien m ot cuoc tranh luan ngay cang tang ve ldi the tbdng
ddi cua PLS-SEM so vbi cac phbong phap SEM khac. N h b n g cuoc tranh
luan khoa hoc n h b vay rat quan trong vi chbng d o n g vai tro n h b m ot
chat xuc tac lam toa ra m ot sb kiem tra ky lbbng h d n ve cac tinh chat
cua p h b o n g phap. M ac du co sb hieu biet sau hon ve n h b n g thuan ldi
va bat ldi, cac nghien cbu bo sung va nhbng tien bo ve phbdng phap
luan cung bat nguon tb cac cuoc thao luan khach quan va m ang tinh
xay d b n g giba cac hoc gia, nham muc dich thuc d ay tien bo khoa hoc.
Tuy nhien, gan day cong dbng khoa hoc da chbng k ien m o t m bc do thao
luan dang ngac n h ien cua thong tin lien quan den PLS-SEM. Antonakis,
Bendahan, Jacquart va Lalive (2010, trang 1103) da d e xuat rang “khong
co ich cho PLS va do do khuyen khich cac nha n g h ien cbu tb bo no”. Cac
nha nghien cbu khac cung gdi y rang viec sb d u n g PLS-SEM “rat kho
phan xet” (Ronkko 8c Everm ann, 2013, trang 443). Do do, viec bo qua
n h bng gipng dieu nay va cac tuyen bo tbdng tb, n h a m m uc dich dap tat
bat ky cuoc thao luan nao cho thay mot sb thieu h ie u biet thau dao ve
khoa hoc. Cac n ha nghien cbu da chi ra rang cac n h a phe binh ve PLS-
SEM th b b n g dba ra ly do khong chinh xac va k h o n g co can cb de ne
tranh phb d n g p h ap nay (Rigdon, 2016). N h b n g q u an niem sai lam vba
neu cb nguon gbc tb sb thieu hieu biet ve co sb k h a i n iem cua phbo n g
phap va dac biet la triet ly do lbbng m a no dba vao (v i du: Rigdon, 2012;
Sarstedt, Hair, Ringle, Thiele, 8c Gudergan, 2016).
 Lbi noi dau xv

“Phiidng phap tiep can PLS va ML-LISREL doi vbi cac m o hinh
d bdng dan vbi cac bien tiem an diidc quan sat gian tiep bbi cac bien da
dang diidc bo sung chb khong phai la canh tranh. Sb khac biet chinh
cua PLS lien quan den LISREL la bbc lbdng ro rang cac gia tri trbbng
hdp cua cac bien tiem an” (Wold, 1982, trang 5). C hinh xac hdn, PLS
la m ot each tiep can dba tren sb tong hdp doi vbi SEM trong do no
ket hd p tuyen tinh cac bien quan sat de tao th an h cac bien tong hdp
(Lohmoller, 1989), lam dai dien cho cac khai niem dang dbdc dieu tra
(Rigdon, 2016). Cach tiep can nay khac vbi CB-SEM, dba tren cac yeu
to chung va do do xem xet cac khai niem nghien cbu n h b cac yefu to
chung m a yeu to nay giai thich sb dbng bien doi giba cac bien quan
sat lien quan cua chung. Mac du sb phan biet nay da dbdc ghi nhan tb
lau (vi du: Joreskog & Wold, 1982; Schneeweifi, 1991), chi m bi day co
nghien cbu bat dau lam ro n h b n g ham y cua sb khac biet nay doi vbi
viec sb dung PLS-SEM (Rigdon, 2012; Sarstedt, H air, Ringle, Thiele, &
Gudergan, 2016; Hair, Hult, Ringle, Sarstedt, & Thiele, dang in). Dbng
thbi, cac nghien cbu gan day da d b a ra n h b n g each tiep can khac nhau
de dieu chinh cac bbc lbdng PLS-SEM giong n h b CB-SEM khi bbc tinh
cac m b hinh yeu tb chung (vi du: Bender & H uang 2014; Dijkstra &
Henseler 2015a, 2015b; Dijkstra 8c Schermelleh - Engel, 2014).

Cb nhieu ly do de chung toi da viet cuon sach nang cao nay ve PLS-
SEM n hb la mot phan m b rang cua cuon “Primer on Partial Least Squares
Structural Equation Modeling (PLS-SEM)”, an ban lan 2 (Hair, Hult,
Ringle, 8c Sarstedt, 2017). Thb nhat, each sb dung PLS-SEM da dbdc nang
cap trong nhbng nam gan day. Khi nhbng tien bo ve phbdng phap luan
nay dang phat trien rat nhanh, tuy nhien, khong cb sb h bbng dan ap dung
cac each tiep can mbi nay mot each hdp ly, dac biet la db b i dang sb dung
mot each de dang. Cuon sach nay nham trang bi cho cac nha nghien cbu
nhbng hieu biet va nang lbc ap dung phbdng phap PLS-SEM nang cao.
Thb hai, vi cac cau hoi nghien cbu dang trb nen phbc tap hdn va db lieu
de tiep can hdn, cac nha nghien cbu can phai ap dung cac phan tich SEM
cao cap hdn rat nhieu lan vbi PLS-SEM. Cuon sach nay khong chi gibi
thieu va lam ro cac tien bo ve phbdng phap luan m a con phac thao mot
each quan trong n hbng phbdng phap PLS-SEM nang cao khi nao co the
ap dung va khi nao khong the ap dung. Thb ba, co rat nhieu sb hieu lam
ve each tiep can PLS-SEM phu hdp hdn so vbi phbdng phap SEM khac
va ngbdc lai. Su hieu lam n h b vay cb the dan den cac each tiep can SEM
khong dung each dbdc ap dung va giai thich. Do do, cac n h a nghien cbu
co the thieu tb tin trong viec sb dung PLS-SEM noi chung va phbdng
phap m b hinh nang cao lien quan den PLS-SEM.
xvi Nh&ng vbn be nang cao trong PLS-SEM

Tom lai, cu on sach nay cho phep cac nha nghien cbu phat trien sb
tin-tbdng de_ap.jdung-phbdng phap. tiep can PLS-SF.M nang cao khi ho
danh gia no ph u h d p va th b c h ien dung. No lam ro cac boi canh nghien
cbu m a PLS-SEM khong nhat th iet phai la each tiep can SEM thich hdp,
ma con boi canh m a 6 do PLS-SEM khong chi la p h b d n g phap hoan
toan chinh xac, no thbc sb la p h b d n g phap SEM phu h d p duy n h a t hien
co. Cuoi cung, cuon sach cho phep cac nha nghien cbu viet cac bao cao
nghien cbu va cac bai bao cong b o tren cac tap chi khoa hoc yeu cau ap
dung chinh xac cac phbdng phap PLS-SEM nang cao khi ho danh gia no
thich hdp, va truyen dat n h b n g phat hien cua ho hieu qua hdn. Xuyen
suot cuon sach, chung toi nh an m anh rang bat ky p h b o n g phap tiep can
PLS-SEM nang cao nao cung khong nen dbdc ap dung m ot each m u
quang, no phai phu hdp vbi ngb canh nghien cb u va dac diem d b lieu
thich hdp la nen tang cua nghien cbu.

P h b d n g phap tiep can cua cuon sach nay dbdc theo sat cuon “Primer
on Partial Least Squares Structural Equation Modeling (PLS-SEM), ain
ban lan 2 (H air va cong sb, 2017) va dba tren n h ieu nam nghien cbu va
giang day cua chung toi cung n h b m ong m uon cua chbng toi truyen dat
ca hai nen tang va tien bo cua p h bdng phap PLS-SEM tbi m ot doi tbdng
rong hdn. De hoan thanh m uc tieu nay, chung toi da gibi han sb nhan
m anh vao p h b d n g trin h , cong thbc, cac ky hieu Hy Lap,... la n hbng
dac trb n g cua hau het cac cuon sach va bai bao ve chu de nay. Thay vao
do, chung toi se giai thich chi tiet cac nguyen tac cd ban cua cac phan
tich PLS-SEM n an g cao va dba ra n h b n g quy tac d e co the sb dung lam
h b b n g dan ch u n g de hieu va danh gia ket qua ap dung.
T rong m ot n o lbc hdn n b a de tao thuan ldi ch o hoc tap, chung toi
sb dung m ot nghien cuu tinh huong duy nhat xuyen suot cuon sach.
Tinh huong nay dbdc rut ra tb m ot nghien cbu da cong bo ve danh
tieng cua cong ty (Eberl, 2010) va chbng toi tin ra n g sb hieu biet cua
cac n h a nghien cbu den tb n h ieu linh vbc khoa h o c xa hoi va cac llnh
vbc khac la h o an toan khong ch in h xac, do do ca n phai hieu them ve
phbong ph ap nay. Tat ca cac vi du trong cuon sach sb dung m ot goi
phan m em duy n hat, SmartPLS 3 (w w w .sm artpls.com ), phan m em nay
cb the dbdc sb dung khong chi de hoan th an h cac bai tap m a con de
thbc hien nghien cbu cua chinh ngbdi doc. Cac cau hoi danh gia va tb
duy p h an bien dbdc dat ra b cuoi chbdng cung n h b cac th u at ngb chinh
dbdc d in h n g hla de giup ngbdi doc hieu ro hdn ve cac khai niem . Cuoi
cung, n h b n g de xuat ve tai lieu tham khao can d o c va m b rong them
dbdc cung cap de cac nha nghien cbu quan tarn co the danh gia m ot
p h am vi n an g cao cua chu de cao hon.
 Lbi noi dau xvii

Cac chbdng sach va cac g*i ho trd hoc tap dbdc sap xep xung quanh
cac kSt qua hoc tap dbdc trin h bay b dau m oi chbdng. H dn the n ba, thay
vi m ot ban tom tat duy nhat vao cubi m oi chbdng, chung toi trin h bay
m ot ban tom tat rieng cho tilng ket qua hoc tap. Cach tiep can nay lam
cho cuon sach trb nen d l hieu va de sb dung h d n cho ca sinh vien va
giang vien. Trang web cua cuon sach (www.pls-sem .com ) cung bao gom
cac tai lieu ho trd khac de tao sb thuan tien trong hoc tap va ap dung
p h b d n g phap PLS-SEM.

C hung toi m uon gbi lbi cam dn vi nhieu thong tin sau sac va gdi y cua
m ot so dong nghiep cung nhu sinh vien cba chbng toi. Q uan trong nhat,
chung toi gbi lbi cam on den Sonke Albers (Dai hoc Kuhne Logistics),
Haya Ajjan (Dai hoc Elon), M urad Ali (Dai hoc King Abdulaziz),
Necmi Avkiran (Dai hoc Q ueensland), Barry J. Babin (Dai hoc C ong
nghe Louisiana), Jan-Michael Becker (D ai hoc Cologne), D iogenes de
Souza Bido (Universidade Presbiteriana M ackenzie), Roger C alantone
(Dai hoc bang M ichigan), Gabriel C epeda C arrion (Dai hoc Seville),
Wynne W. C hin (Dai hoc Houston), Alain Yee Loong Chong (Dai hoc
Nottingham ), Adam antios D iam antopoulos (Dai hoc Vienna), Theo
Dijkstra (Dai hoc Groningen), M arkus Eberl (TNS Infratest), A ndreas
Eggert (Dai hoc Paderborn), Vincenzo Esposito Vinzi (Trbbng Kinh
doanh ESSEC), David Garson (Dai hoc bang Bac C arolina), Oliver Gotz
( Dai hoc Reutlingen), Anne Gottfried (Dai hoc Nam M ississippi), Lars
G ronholdt (Trbbng Kinh doanh C openhagen), K arl-W erner H ansm ann
(Dai hoc H am burg), Sven H auff (Dai hoc Ham burg), Jorg H enseler
(Dai hoc Twente)), Roland Holten (Dai hoc Goethe Frankfurt), Lucas
Hopkins (Dai hoc bang Florida), Geoffrey S. Hubona (R-Courseware),
G. Tomas M. Hult (Dai hoc bang M ichigan), Ida Rosnita Ismail (Dai hoc
Kebangsaan Malaysia), Dave Ketchen (Dai hoc A uburn), Hengky Latan
(Dai hoc D iponegoro), Marcel Lichters (Dai hoc O tto-von-G uericke
Magdeburg), A nne M artensen (Trbbng Kinh doanh C openhagen),
Annette Marie Mills (Dai hoc C anterbury), A rthur Money (Trbbng
Kinh doanh Henley), Erik Mooi (Dai hoc M elbourne), C hristian Nitzl
(Dai hoc Bundeswehr M unchen), Arun Rai (Dai hoc bang Georgia),
Sascha Raithel (Dai hoc Freie Berlin), Ih u rasam y Ramayah (D ai hoc
Sains Malaysia), S. Mostafa Rasoolimanesh (Dai hoc Sains M alaysia)),
Nicole F. Richter (Dai hoc Nam Dan M ach), Edward E. Rigdon (Dai
hoc bang Georgia), James A. Robins (Dai hoc Kinh te va Kinh doanh
Vienna), Jose Luis Roldan (Dai hoc Seville), Alexander Rossm ann
(Dai hoc Reutlingen) ty), Phillip Sam ouel (Dai hoc Kingston), G aston
xviii Nhung v ln d£ nang cao trong PLS-SEM

Sanchez (Dai hoc California, Berkeley), R ainer Schlittgen (Dai

hoc H am burg), Tobias Schutz (Tracing Kinh doanh ESB), M anfred
Schw aiger- T D a f'h o c EudwTg: M axim iIians-Uhiversitat M iinchen),
H arjit Sekhon (Dai hoc C oventry), Abdel M onim Shaltoni (Dai hoc
Alfaisal), M ac (W en-Lung) Shiau (Dai hoc M ing C huan), Galit Shmueli
(Dai hoc quoc gia Tsing H ua), Rudolf R. Sinkovics (Lien m inh trUdng
kinh d o an h M anchester), D o n n a Sm ith (Ryerson) D ai hoc), D etm ar
W. Straub (Dai hoc bang G eorgia), M ichel T enenhaus (H EC Paris),
Nelly T odorova (Dai hoc C anterbury), Laura T rinchera (T radng Kinh
doanh NEOM A), Sven W ende (SmartPLS G m bH ), va A nita W hiting
(Dai hoc bang Clayton) cho n h an xet gia tri cua ho. Ngoai ra, chung toi
cam cfn nhCtng ngUdi danh gia sau day vi nhOng nhan xet rat htfu ich
cua ho: David H em sw orth (Dai hoc Nipissing), Rajesh Iyer (Dai hoc
Bradley), Jay M em m ott (Dai hoc Nam Dakota), B rent Sm ith (Dai hoc
Saint Joseph) va James N. Smith (Dai hoc Dong N am Nova). Ben canh
do, chung toi cam on doi ngu nghien ciiu sinh va nghien ciiu vien tai
Dai hoc C ong nghe H am burg (T U H H ) va Dai h o c O tto-von-G uericke
M agdeburg — cu the la: Kati Barth, Katrin Engelke, A ndreas Fischer,
A m ra K ram o-C aluk, Frauke Kuhn, Anke Lepthien, D oreen N eubert,
Jana Rosenbusch, Victor Schliwa, Sandra S chubringva Kai O liver Thiele
— vi sU ho trp chan thanh cua hp. Cuoi cung, tai SAGE, chung toi cam
o n to an bo nhom xuat ban, dang chii y nhat la Kelly DeRosa, Sarah J.
Duffy, Leah Fargotstein va Yvonne McDuffee.

C hung toi hy vong cuon sach nay se m d rong kien thiic ve cac kha
n an g va lpi ich cua PLS-SEM c h o m ot nh o m cac n h a n g h ien ciiu va cac
nha thiic hanh. C uoi cung, neu ban co bat ky n h a n xet, de xuat hoac y
tudng nao de cai thien cuon sach nay, vui long lien he vdi chiing toi.
C hung toi danh gia cao bat ky phan hoi nao ve ly thuyet va noi dung
cua cuon sach!
Jo se p h F. H a ir Jr., D a i hoc South Alabama
M ark o S arsted t, O tto-von-G uericke - Dai hoc Magdeburg,
Diic va D ai hoc Newcastle, Uc
C h ris tia n M . R ingle, Dai hoc Cong nghe Hamburg,
Diic va D a i hoc Newcastle, Uc
S iegfried P. G u d e rg an , D ai hoc Newcastle, Uc
Gioi thieu ve cac tac gia

Joseph F. Hair, Jr. la giao sd m arketing, giam doc DBA, va C hu tich

kinh d o an h Cleverdon tai Trdcfng Cao dang Kinh doanh Mitchell, Dai
hoc Nam Alabama. Trdcfc day ong tdng la hoc gia cao cap, chifcmg trinh
DBA, trd d n g Cao dang Kinh doanh Coles, Dai hoc bang Kennesaw, va
gid ghe C hu tich chilcfng trinh Von Khdi nghiep cua C opeland cung
n h d la giam doc Vien D oanh nhan tai Trdotag Q uan trj Kinh doanh
Ourso, Dai hoc Bang Louisiana. Ong la tac gia cua h o n 60 cuon sach,
bao gom Phan tich DU lieu Da bien (an ban lan 7, 2010), M K T G (an ban
lan 10, 2016), Cac PhUOngphap Nghien ciiu Kinh doanh can ban (2016)
va Nghien ciiu Marketing can ban (an ban lan 4,2017). O ng cung da xuat
ban nhieu bai bao trong cac tap chi hoc thuat va dddc cong n h an la nha
giao due m arketing cua Hoc vien Khoa hoc M arketing cua nam . Mot
giang vien noi tieng, giao sd H air th d d n g trinh bay cac hoi thao ve ky
thuat nghien cdu, phan tich dd lieu da bien va cac van de m arketing cho
cac to chdc 6 chau Au, Uc, Trung Quoc, An Do va Nam My.
Marko Sarstedt la giao sd m arketing tai Dai hoc O tto-von-G uericke
M agdeburg (Ddc) va la d6ng giao sd d Khoa Kinh doanh va Luat tai Dai
hoc Newcastle (Uc). Trddc day ong tdng la pho giao sd ve cac phddng
phap dinh lddng trong m arketing va quan ly tai Dai hoc Ludwig-
M axim ilians-M unich (Ddc). N ghien cdu chinh cua ong la dng dung cua
cac phddng phap mo hinh cau tru e nang cao de hieu ro hdn ve hanh vi
cua ngdbi tieu dung va de cai thien viec ra quyet dinh m arketing. Nghien
cdu cua ong da dddc xuat ban tren cac tap chi n h d Journal of M arketing
Research, Journal o f the Academy o f Marketing Science, Organizational
Research Methods, M IS Quarterly, International Journal o f Research in
Marketing, Long Range Planning, Journal o f World Business, va Journal of
Business Research. Theo xep hang Handelsblatt, Giao sd Sarstedt la m ot
trong ba n ha nghien cdu tiep thi tre tuoi hang dau 6 Ddc, Ao va Thuy
Si. O ng thdcmg xuyen giang day cac hoi thao tien si ve thong ke da bien,
m o hinh phd d n g trinh cau true, va do lddng tren toan the gidi.

xix
 xx Nhung van <36 nang cao trong PLS-SEM

Christian M. Ringle la giao sd quan tri ch in h cua Dai hoc C ong nghe
H am burg (Dbc)-va-la dbng-giao sd.cua.Khoa_Kinh.doanh va Luat tai Dai
hoc Newcastle (Uc). O ng co bang thac sy quan tri kinh doanh cua Dai
hoc Kansas va nhan bang tien si cua Dai hoc H am burg (Ddc). N ghien
cdu duqc xuat ban rong rai cua ong giai quyet vain de ve quan ly cac to
chdc, chien luqc va quan ly nguon nhan ldc, m arketing va cac p h u o n g
phap d in h luong cho nghien cdu kinh d o an h va thi trdbng. Ong la dong
sang lap cua SmartPLS (www.sm artpls.com ), m ot cong cu phan m em
vdi giao dien do h o a cho viec ap dung p h d o n g phap m o h in h hod rieng
phan nho nhat (PLS-SEM). Ben canh h o trq td van va h o trd cac tap
doan quoc te, ong th d d n g xuyen giang day cac hoi thao tien si ve thong
ke da bien, phd ong phap PLS-SEM, va viec sd dung SmartPLS. Thong
tin them ve Giao sd Ringle va danh m uc cac an p ham cua ong co th e tim
thay tai w w w .tuhh.de/hrm o/team /prof-dr-c-m -ringle.htm l.
Siegfried P. Gudergan la giao sU chinh ve chien lu q c tai Trdbng Kinh
doanh Newcastle cua Dai hoc Newcastle (Uc). O ng la ngdbi ddng
dau cua trd b n g va la giao sU m arketing cua trd b n g K inh doanh UTS
thuoc trd b n g Dai hoc Cong nghe, Sydney (Uc). O ng co bang tien sy
cua T rdbng Q u an tri K inh doanh Uc do Dai h o c Sydney va Dai hoc
New South Wales tra o tang. N ghien cdu cua ong tap tru n g vao viec ra
quyet d in h quan ly va du q c ap dung cho cac van de ve quan ly chien
luqc. M ot so cong trin h cua ong lien quan den PLS-SEM da xuat hien
tren cac tap chi n h d Journal o f the Academ y o f M arketing Science, Long
Range Planning, vd Journal o f Business Research. O ng la co van chinh
c h o n g h ien cd u sinh tien si da d d q c tra o tang giai th d d n g ve nghien cdu,
thi du, m o t de cb danh du cho Giai th d d n g Luan van cua K hoa C hinh
sach & Chibn luqc K inh doanh cua Hoc vien Q u a n tri va Giai th d d n g
Bai bao Tien sy xuat sac nhat cua Hiep hoi Q uan ly C hien ldOc.

c§ >

uy cap website cua cdc dbng tac gia

ac du* lieu va cac tai nguyen hqc tap khac
o qua trinh nghien cdu cua ban ve PLS-SEM!
ii studyj.agepub.cQm^
 CHUONG 1

Tong quan ve
sU phat trien gan day
va dang noi len.O
trong PLS-SEM*

< s T
CAC KET QUA HOC TAP

1. Hieu nguon goc va su tien trien cua PLS-SEM.

2. Hieu dupe cac nguyen ly va su phat trien gan day trong

3. Tim hieu sii khac biet co ban gitfa PLS-SEM va CB-SEM va

hieu khi nao stfdung tdng phuong phap.
i .i.w «. up

TONG QUAN CHUONG

Trong nhftng nam gan day, nhieu su phat trien da xay ra trong m o hin h
phiidng trin h cau true binh phuo n g toi thieu rieng phan (PLS-SEM).
Trong so nhUng phat hien noi bat nhat la phien ban m di nhat cua phan
meni SmartPLS (Ringle, Wende, & Becker, 2015). Phan m em than thien
v6i ngUoi svt dung lam cho cac nha nghien cdu khoa hoc xa hoi dat dupe
cac m uc tieu nghien ciiu va cong bo hieu qua hon. Dieu nay dac biet
dung khi phan m em bao gom cac tuy chon de hoan thanh viec phan tich
cap nhat nhat co the tuong dng vdi m ot phuong phap thong ke cu the.

Trong chuong nay, chung toi cung cap tong quan ve n guon goc va
sii tien trien cua PLS-SEM de thiet lap m ot nen tang de hieu ro hon ly

1
2 Nhirng van <5e nang cao trong PLS-SEM

do tai sao phiidng phap nay cham diidc chap n h an nhiing da difpc ap
dung- ngay-cang nhieu-trong nhieu-nam -qua trong nh ieu nganh khoa
hoc xa hoi, dac biet la cac linh vpc q u an ly kinh d o an h khac nhau. Sau
do ch u n g toi tom tat cac kha n an g cua p h a n m em n h am tao dieu kien
de dang dn g dung ky th u a t m Pi no i nay va lam noi bat n h d n g p hat trien
gan day ve p h iid ng phap luan trong PLS-SEM. Be hieu diidc PLS-SEM
khac vdi SEM d pa theo hiep p h u o n g sai (CB -SEM ), chung ta se thao
luan cac each tiep can khac n h a u de do liidng cac k hai niem va lam noi
bat p h iid n g phap nao phu h o p h d n cho cac kieu m o h in h nhat dinh.

NGUON GOC VA Si/ TlfEN TRIEN CUA MO H IN H PHl/ONG

TRINH CAU TRUC BiNH PHl/ONG TOI THIEU RIENG PHAN
Cac tien than cua phddng phap PLS-SEM la hai th u tuc lap lai sd dung
lidc liipng b in h phiidng nho n h a t de phat trien cac giai phap cho cac
mo h in h do n va m o hin h da th an h phan cung nhii cho moi tiidng quan
do n gian (Wold, 1966). Sii phat trien tiep theo cua cac th u tuc nay cua
Herm an W old dan den thuat toan binh phiidng toi thieu rieng phan tai
lap phi tuyen tin h (NIPALS) (Wold, 1973). M ot p h ien ban tong quat
sda doi cua thuat toan PLS da tap trung vao viec th ie t lap va bao gom
cac bien tiem an trong cac m o hin h diidng dan (Lohm oller, 1989; Wold,
1980, 1982, 1985).

M ot so thu tuc PLS da p h at trien tii thuat to a n binh phiidng toi

thieu khai quat cua Wold (M ateos-Aparicio, 2011). M ot thu tuc la hoi
quy thanh phan chinh (p rin cip a l com ponent regression), thiic hien
phan tich th an h phan chinh tren cac bien doc lap v a trong do cac thanh
phan chinh diipc sd dung lam cac bien dp bao / giai thich cho bien phu
thuoc. Tuy nhien, hoi quy thanh phan chinh chu yeu tap trung vao viec
giiim chieu hiiPng cua cac bien doc lap m a khong tin h den moi quan he
gida cac bien doc lap va phu thuoc.

M ot th u tuc khac la hoi quy binh phddng to i thieu rieng phan

(PLS-R), ban dau diidc th iet ke de giam van de da co n g tuyen trong cac
mo h in h hoi quy. PLS-R la m ot phiidng phap tap tru n g vao viec giam
chieu hiiPng cua cac bien doc lap trong m ot m o h in h hoi quy vdi m uc
tieu loai bo da cong tuyen td cac bien dp bao, n h iin g lam vay nhii mot
each de toi du hoa phiidng sai th u diidc td cac bien d o c lap trong khi toi
da hoa dong thdfi phiidng sai giai thich trong cac bien phu thuoc. Noi m ot
each chinh xac hdn, PLS-R dpa vao m pt phan tich th a n h phan chinh rut
Chu’o'ng 1: Tong quan v i sir phat trien gan day va dang noi len trong PLS-SEM 3

tri'ch ra td cac tong hop tuyen tinh cua cac bien doc lap va cac diem so
tdong dng cua chung, lam giam chieu hddng cua cac bien doc lap, va xem
xet moi qu an he giufa cac bien doc lap va phu thuoc, do do toi da hoa giai
thich cua phdong sai trong bien phu thuoc. Ket qua la PLS-R cho phep
cac n h a nghien cdu ddc ldpng cac m o hin h co nhieu bien doc lap hon cac
quan sat trong tap dd lieu (Valencia & Diaz-Llanos, 2003).

Thu vi thay, PLS-R da khong diipc phat trien bdi H erm an Wold,
nhiing con trai cua ong, Svante Wold da phat trien chung vao dau n hdng
nam 1980. Svante Wold da lam viec trong llnh vdc h o a hoc phan tich m a
ngay nay diipc goi la hoa liipng (chemometrics) - linh vdc ap d u n g cac
phdong phap th o n g k e cho dd lieu hoa hoc. C ung vdi Harald M artens,
ong da dieu chinh NIPALS de phan tich d d lieu h o a hoc, va them vao
do la giai quyet van de da cong tuyen trong cac m o h in h hoi quy da bien,
phdong phap cua ho giai quyet dupe van de phat sinh khi so ldpng cac
bien lPn hon so phan hoi cua dap vien. ^ %

Mot thu tuc thd ba xuat hien td thuat toan PLS tong quat cua Wold
la m o hinh hoa ddPng dan binh p h d o n g toi thieu rieng phan, sau nay
trd thanh PLS-SEM (Hair, Ringle, va Sarstedt, 2011). PLS-SEM xac dinh
cac tham so cua m ot tap cac phdo n g trinh trong m ot m o hinh ddPng
dan bang each ket hop cac phan tich thanh phan chinh de danh gia cac
mo h in h do lddng vdi phan tich ddP ng dan de ddc ldpng cac m oi quan
he gida cac bien tiem an. Wold (1982) da de xuat “thiet ke co ban m o
hinh m em ” cua PLS-SEM n h d la m ot giai phap thay the cho SEM dda
tren hiep phddng sai (CB-SEM) cua Joreskog (1973). CB-SEM doi khi
ddpc goi la SEM dda tren nhan td va ddpc danh dau la m o hinh cdng
vi cac gia d in h nghiem ngat hon nhieu ve phan phoi dd lieu va kich cd
mau. Dieu quan trong la “khong phai khai niem cung khong phai la cac
m o hinh cung khong phai la cac ky thuat ddc ld p n g ‘m em ’, chi cd cac
gia dinh phan phoi” (Lohmoller, 1989, trang 64). M ac du ca hai phddng
phap tiep can da ddpc phat trien cung thdi diem , CB-SEM da trd nen
pho bien rong rai hon do sd hien dien cua phan m em LISREL ke td
cuoi n h d n g nam 1970. N gdpc lai, phan m em dau tien cua PLS-SEM la
LVPLS xuat hien vao gida n h d n g nam 1980 (Lohm oller, 1984) nhdng
khong than thien lam. Cho den khi phan mem PLS-Graph cua C hin
(1994) xuat hien vao gida n hdng nam 1990, PLS-SEM bat dau ddpc
ap dung rong rai hon. Vdi viec phat hanh SmartPLS 2 vao nam 2005
(Ringle, Wende, & Will, 2005), cac dng dung PLS-SEM da tang theo cap
so nhan.
4 Nhung v ln de ndng cao trong PLS-SEM

Tu lieu 1.1

ro
ro
>»
ro
a
</}
*D
3
ro
2 40
UJ
500 -□ro
CO E
•J
a. 3
C
ro
>
ro
3

00 |
o

1980 1985 1990 1 2005 2010 2015

________ year
Ghi chu: Cac nghien ciiu ve L S-SE M ditac x u a t ban trong M IS Q uarterly chi diidc x em
x et tit nam 1992 trd di.

T d lieu 1.1 tom tat viec ap dung PLS-SEM tro n g cac tap chi hang
dau ve m arketing (Hair, Sarstedt, Ringle, & M ena, 2012) va quan ly chien
ldpc (Hair, Sarstedt, Pieper, 8c Ringle, 2012) cung n h d M IS Quarterly,
tap chi hang dau ve nghien cd u he thong th o n g tin quan ly (Ringle,
Straub, 2012). Tuy nhien, viec sd dung PLS-SEM m d rong
llnh vdc khac nhau n h d quan ly chuoi c u n g dng, quan ly hoat
va nghien cdu kinh d o an h gia din h (vi du: D o Valle & Assaker,
16; Hair, Hollingsworth, R andolph, 8c C hong, 2017; Henseler, Ringle,
8c Sarstedt, 2012; Kaufm ann 8c Gaeckler, 2015; Lee, Petter, Fayard, 8c
Robinson, 2011; Nitzl, 2016; Peng 8c Lai, 2012; Richter, Sinkovics,
Ringle, 8c Schlagel, 2016; Sarstedt, Ringle, Smith, R eam s, 8c Hair, 2014).
Mot so tap chi n h d Journal o f M arketing Theory and Practice (H air va
cong sd, 2011), Long Range Planning (H air, Ringle, 8c Sarstedt, 2013;
Robins, 2012, 2014), Journal o f Business Research (C arrion, Henseler,
Ringle, 8c Roldan, 2016) va International M arketing Review (R ichter va
cong sd, 2016) cong bo cac van de dac biet ve p h d d n g phap PLS-SEM.
Chifccng 1: Tong quan v§ sir phat trien gin day va dang noi len trong PLS-SEM 5

Ben canh sb pho bien cua no trong nghien ciiu kinh doanh, viec sii
dung PLS-SEM nhii cac b li bao cong bo gan day da m o rong sang cac
llnh vbc nghien ciiu khac nhii sinh hoc (Kansky, Kidd & K night, 2016),
ky thuat (vi du: Kwofie, Adinyira, & Fugar, 2015; Liu, Zhao, & Yan, 2016;
M inkm an, Rutten, & van der Sanden, 2016), y hoc (vi du: Hsu, C hang, &
Lai, 2016; Pai, 2016; Pedrosa, Rodrigues, Oliveira, & Alexandre, 2016),
va tarn ly hoc (vi du: Jisha, & Thomas, 2016; Willaby, Costa, Burns,
M acCann, & Roberts, 2015).

Viec phat hanh phien ban 3 cua phan m em PLS-SEM diigtc ap dung
rong rai nhat, SmartPLS, bao gom nhieu tinh nang mbi. Thii nhat, viec
co db d c m ot giai phap cO ban tb phien ban 2 (Ringle va cong sii, 2005)
cua phan m em la d l tiep can va van tiep tuc dung vbi phien ban 3. Tuy
nhien, m ot vai phep xb ly de danh gia cac giai phap PLS-SEM khi sb
dung phien ban 2 phai la dbdc tinh bang tay h o ac da hoan thanh vbi
m ot gbi phan m em khac, chang han n h u Excel, SPSS hoac Statistica.
SmartPLS 3 tb dong thiic hien cac chbc nang nay, bao gom he so tac
dong f , danh gia da cong tuyen, van van. Tuy nhien, quan tro n g h on la
viec bo sung cac Ida chon nhii phan tich bo tb khang dinh (G udergan,
Ringle, Wende, & Will, 2008), tieu chi heterotrait-m onotrait m bi cua
cac m oi td o n g quan (HTMT) de kiem dinh gia tri phan biet (Henseler,
Ringle, Sarstedt, 2015), phan doan theo dinh hiibng db bao (Becker,
Rai, Ringle, & Volckner, 2013), phan tich bien dieu tiet (Henseler &
C hin, 2010), cac loai phan tich da nhom khac nhau (Sarstedt, Henseler,
& Ringle, 2011) va kiem dinh sii bat bien bang phiidng phap bat bien do
liidng cua p h b d n g phap tiep can ket hop cac m o h in h (Henseler, Ringle,
& Sarstedt, 2016). N hbng ldi ich cua viec da Ida chon de dang nay la vo
cung lbn, vi n h b n g loai phan tich nay ngay cang dbdc yeu cau bbi cac
bi£n tap vien cua tap chi va ngbbi phan bien. N h b n g phat trien nay rat
quan trong doi vbi cac nha nghien cbu khoa hoc xa hoi bbi vi n hieu gia
thuyet khong the thbc hien dbdc vbi CB-SEM do n h bng gia dinh h an
che cua no, trong khi nhung nghien cbu khac khong dbdc lap trin h de
tao thuan ldi cho viec sb dung.

DO LUttNG
Cac bien khai niem va bien dai dien
C ho du cac nha nghien cbu co theo each tiep can nghien cbu dien dich
hoac quy nap, tai m ot so diem nao do - trong sb tim toi cua ho de hieu
6 Nhbng van <J§ nang cao trong PLS-SEM

va giai thich ly thuyet to t hdn - h o giai quyet cac m o hinh ly thuyet va

cac bierr khai-m e mv M ot m d h tn h ty tln jy lf r p h a ir a n h mQt tap h d p cac
m oi quan he cau true, th b b n g d b a tren m ot tap h d p cac p h b d n g trin h
ket noi cac bien khai n iem m a sb ket noi nay hoan chinh hoa tb m ot ly
thuyet va sb ket noi nay dai dien cho cac m oi quan he (Bollen, 2002). Vi
la cac yeu to cua cac m o hinh ly thuyet, cac khai niem dai dien cho cac
y tbdng rong lbn hoac suy nghi ve cac khai niem khai quat m a cac nha
n ghien cb u thiet lap va de xuat do lbdng trong n g h ien cb u cua ho (vi du,
khai niem sb hai long cua khach hang).
Cac khai niem nghien cbu dai dien cho cac bien khai niem trong cac
m o hinh thong ke n h b trong m ot m o hinh phbdng trin h cau true. C hung
dbdc dung cho cac phep kiem dinh thbc nghiem cac gia thuyet lien quan
den cac bien khai niem (Rigdon, 2012) va dbdc xac d in h cac thuoc tinh va
doi tb d n g (vi du: MacKenzie, Podsakoff, & Podsakoff, 2011). Thuoc tinh
xac dinh loai tinh chat chung m a khai niem de cap den, chang han nhb
thai do (vi du nhb thai do doi vdi quang cao), cam n h an (vi du: nhan thbc
de sb dung ve cong nghe), hoac y dinh ve hanh vi (vi du: y dinh m ua). Doi
tbdng trong tarn la m ot thbc the m a tinh chat dbdc ap dung. Vi du, trong
tarn cua sb quan tarn co the la sb hai long cua khach hang vbi san pham ,
sb hai long vbi djch vu, va sb hai long vbi gia ca. Trong n hbng vi du nay,
sb hai long tao thanh thuoc tinh, trong khi san pham , djch vu va gia ca dai
dien cho cac doi tbdng trong tarn.

Thiet lap m ot dinh nghla cua khai niem nghien cbu cung bao gom
viec xac d inh cac chieu hbdng m o ta bien khai niem , vbi m bi chieu
khac nh au bib’u dien m ot khia canh khac nhau cu a bien khai niem . Vi
du, k h ai n iem sb cam ket co the d b q c xem la m o t k h a i niem bao gom
ba khia canh k h ac n h au (Meyer & Allen, 1991), d b d c xac d in h va do
lbbng m o t each khac nhau: cam ket tinh cam (nghla la gan ket ve m at
tin h cam ), cam ket lien tuc (nghla la ldi ich so vbi m at m at khi lam viec
trong m ot to chbc), va cam ket quy pham (nghla la d b a tren cam xuc cb
nghla vu). M bi m ot trong cac chieu h b b n g nay doi h o i m ot din h nghla
hoat hoa rieng biet. M ot bien khai niem khong p h a i la m ot dac tin h
rieng b iet n h b ddn h b b n g hoac da h bbng, cho phep n o i d en hai, ba hoac
b on h b b n g (Bollen, 2011). Thay vao db, no p h u th u b c vao dinh nghla cu
the theo ngb canh cua bien so khai niem va bieu h ie n di kem vbi no. Ve
nguyen tac, co the goi n h b n g bieu hien la vo han vi bien the khai niem
tb d n g tb cb the bieu dien cac m b c do trb u tb d n g ly thuyet khac n h au
g iba cac n g b canh (D iam antopoulos, 2005; Law & W ong, 1999). D inh
Chuong 1; Tong quan ve sy phat trien gin day va dang noi len trong PLS-SEM 7

nghla hoat hoa phu thuoc vao bbi canh trong do m ot bien khai niem
d b fc kiem dinh va dinh nghia cb the thay doi tb nghien cbu nay sang
nghien cbu khac. Do do, m ot dinh nghla hoat dong cb the khac nhau
ve chieu hbbng va doi tb o n g quan tarn, phu thuoc vao bbi canh nghien
cbu. Vi du, thoa m an cua khach hang vbi djch vu co the dboc chia thanh
nhieu ph an cu the hon, chang han nhb sb hai long vbi toe do dich vu,
khung canh djch vu va nhan vien. Khia canh th b hai cb the dbefe chia
thanh nhieu phan cu the hon n h b sb hai long vbi sb than thien, nang
lbc, va hinh dang ben ngoai cua nhan vien dich vu. Ve m at nguyen tic,
m oi khia canh tren co the dboc chia nhb thanh cac phan nhb hon.

Do dinh nghla khai niem nghien cbu lam ro m bc do khai quat,

bien khai niem phai cb lien quan den cac do lbbng co the quan sat dbctc,
no cflng giup hieu dboc each thbc, trong cac m o h in h phbo n g trinh
cau true, cac bien khai niem dboc bieu dien bbi cac khai niem nghien
cbu (con goi la cac bien tiem an). Cac khai niem nghien cbu khong
dboc trb c tiep quan sat n hbng dboc suy ra (toan hoc) tb cac bien bieu
hien db o c quan sat (dboc do lbbng trbc tiep). Cac bien bieu hien nay
th b b n g dboc gpi la cac bien do lbbng (item) hoac la cac bien quan sat
(indicator). Vi du, cac bien quan sat tbong bng vbi cac cau hoi trong m ot
cuoc khao sat th u thap nhan thbc, thai do va hanh vi cua ngbbi tra lbi.
Dieu quan trong la cac bien quan sat khong can phai la db lieu d b a tren
khao sat m a con cb thb la db lieu thb cap (vi du: xep hang m bc do hai
long tb trang web danh gia). Khi m ot khai niem nghien cbu m ang tinh
da hbbng, moi hbbng can phai dbpc bieu dien bbi m o t cau true rieng
biet, dbpc van dung vbi m ot bo bien quan sat cu the. Vi du, neu mot
nghien cbu xem xet ca ba khia canh cua cam ket thi se co ba bien tiem
an rieng biet trong m o h in h phbong trinh cau true.

D inh nghia cua khai niem nghien cbu giai thich each thbc bien khai
niem lien quan den cac bien quan sat tb goc do ly thuyet. Tuy nhien, cac
khai niem nghien cbu khong dai dien cho cac bien khai niem hoan hao
n h b bat ky khai niem va bat ky dinh nghla hoat dong cb m ot sb m bc do
mO ho lien quan den no. Hon nba, cac khai niem nghien cbu bat nguon
tb db lieu va do db chia sb n h b n g dac diem rieng cua db lieu (Cliff, 1983;
M acCallum , Browne, & Cai, 2007; Rigdon, 2012), dieu nay tiep tuc tach
chung ra khoi cac khai niem m a chung djnh dai dien. Trong bbi canh
nay, Michell (2013, trang 20) lbu y rang cac khai niem nghien cbu dboc
tao ra theo each tach khoi cau true hien tai cua cac hien tbOng kiem
din h va dboc gib nguyen tai cho bang mot loat cac phbong phap dinh
8 Nhung v ln d i nang cao trong PLS-SEM

lbpng, chang han n h b phan tich nhan to, phan tich nay cho phep dinh
Ib p n g c a u truc-m pt-rach-vo-cb-honTasuy-luan-nb tb cac hien tb p n g co
lien quan.

C hong lai lap luan nay, Rigdon (2012, trang 343-344) ket luan rang
cac khai niem nghien cbu nen dbpc xem n h b la “cai gi do dboc tao ra tb
cac db lieu thbc nghiem nham de kiem tra thbc nghiem cac gia thuyet
de xuat lien quan den khai niem ”. Tbc la tat ca cac do lbbng cua bien
khai niem la xaip xi hoac dai dien cho cac bien khai niem , doc lap vbi
each chung dbpc bat nguon (vi du W ickens, 1972). Do db, khong phan
biet chat lb b n g m a m ot bien khai niem ve ly thuyet d b p c chbng m inh va
dbpc xac dinh ve m at hoat dong va do nghiem ngat bao gom viec phat
trien m o hinh do lbbng, bat ky sb do lbbng nao trong cac m o hin h cau
true chi dai d ien cho cac bien tiem an (Rigdon, 2 0 12). D anh gia nay phu
hop vbi sb phat trien cua tat ca cac loai cong cu d o i hoi phai do lbbng co
ban ciing m ot khai niem nghien cbu, m ac d u th b b n g it co cb hoi chuyen
doi cac do lb b n g cua m ot cong cu nay sang bat ky d o lbbng cua cac cong
cu khac (Salzberger, Sarstedt, va D iam antopoulos, 2016). Vi du, nghien
cbu kinh d o an h va thbc tien da dba ra vo so cac cong cu do lbbng cho
dan h tieng cua cong ty, dba tren cung m ot d in h nghia ve khai niem
nh bng khac ve co ban khai niem va cac bien q u a n sat dbpc do lbbng
(Sarstedt, W ilczynski, Melewar, 2013).

Cac mo hinh do lbbng

Theo sau din h nghia cua khai niem nghien cbu, b b b c tiep theo la khai
niem hoa m ot m o hinh do lbbng, the hien each thbc do lbbng khai
niem nghien cbu bang m ot tap cac bien quan sat. N o i chung, co hai each
ph o bien de khai niem hoa cac m o h in h do lbbng. C ach tiep can d lu
tien dbpc goi la do lb b n g ket qua. Trong m ot m o h in h do lb b n g k e t qua
cac bic'n qu an sat dbpc coi la cac bien bieu hien k h o n g tran h khoi sai so
cua m ot khai niem nghien cbu p hu thuoc vbi cac m o i quan he di tb khai
niem nghien cbu den cac bien quan sat cua no (T b lieu 1.2). Moi quan
he giba m ot bien quan sat dbpc va khong quan sa t dbpc th b b n g dbpc
m o h in h hoa n h b the hien trong phbong trin h sau:

x = 1- Y + e

trong db x la bien quan sat, Y la bien tiem an, he s b tai I la he so hoi quy
dinh lbpng do m anh cua moi quan he giba moi b ien quan sat x va khai
niem Y cua no; e dai dien cho cac sai so ngau n h ien .
 Chuong 1: Tong quan v§ su phat trien gan day va dang noi len trong PLS-SEM 9

T u lieu 1.2 m o ta m ot m o hinh do ludng ket qua cho m ot bien

tiem an , du'o'c do bang bon bien quan sat x x, x 2, x i va x r O day, cac
m oi quan he Z, den Z4 la ket qua cua bon hoi quy dcfn, n h u dUqtc mo ta
trong phUctng trinh trildc, vdi m oi bien quan sat x { den ,r4 dudi dang
cac bien phu thuoc va khai niem nghien ciiu 7, n h u la bien doc lap.
Cac bien quan sat ket qua (doi khi duqc goi la cac bien quan sat mang
tinh ket qua trong ly thuyet tarn ly hoc do liidng) co the d u q c xem nhu
m o t m au dai dien cua tat ca cac bien do liidng co the cd tro n g pham vi
khai niem cua khai niem nghien cUu (N unnally 8c Bernstein, 1994). Vi
m b h inh do lUdng ket qua chi ra riing tat ca cac bien do ludng p h an anh
cung m ot khai niem nghien ciiu, cac bien quan sat gan lien vdi m o t khai
niem nghien cUu cu the phai cd m oi tUong quan cao vdi nhau. Ngoai ra,
cac bien do ludng rieng le cd the hoan doi cho nhau, va bat ky m ot bien
quan sat nao cung cd the dilbc bo di ma khong thay doi y nghia cua cau
true, m ien la cau true cd du do tin cay. Thiic te rang moi quan he nay di
til khai niem nghien cUu den cac bien quan sat cua no ham y rang neu
sU danh gia ve tinh chat tiem an thay doi (vi du: do thay doi trong tieu
chuan so sanh), tat ca cac bien quan sat se thay doi dong thdi - it nhat d
m ot m dc do nao dd.

ic, loai mo hinh do luting

\g u y e n nhan

Bi6n quati sat k it qua Bien quan sat nguyen nhan Bien quan s a t to n g hgrp

z, = 0

Xi xt *t xt

et e-y ej e4

Dang m b h inh do ludng khac la m o hinh do ludng nguyen nhan.

Trong mo hinh do ludng nguyen nhan cac bien quan sat tao thanh
10 Nhdng van de nang cao trong PLS-SEM

khai niem nghien ciiu bang cac ket hop tuyen tinh. M ot sd thay doi gia
tri ciia b ien quan sat, vi du nhii sd thay doi trong sii d an h gia cua m ot
ngddi tra ldi ve dac tinh ddpc n am bat bofi sii thay doi gia tri cua bien
quan sat nay lam thay doi gia tri cua khai niem nghien ciiu. Nghia la, cac
thay doi tro n g cac bien quan sat di trd d c sii thay doi trong bien tiem an
(Borsboom , M ellenbergh, & van H eerden, 2003), co nghia la, theo dinh
nghia, cac khai niem nghien ciiu vdi m o hinh do liidng nguyen nhan
g a n c h a tv d ic a c d o liidng cu a chung (D iam antopoulos, 2006). B en canh
sii khac biet trong m oi quan he gida (cac) bien q u an sat va khai niem
nghien ciiu, cac m o hin h do liidng nguyen nhan khong doi hoi cac bien
quan sat phai tiiong quan.

Mac du cd sii khac biet ve khai niem m ot cbch ro rang, viec xac
d jnh cac m o h in h do liidng ket qua hoac nguyen nhan khong ro rang
trong th iic te, vi ca c b ie n khai n iem khong tuan thCi th eo logic do liidng
ket qua hay nguyen nhan. Thay vao dd, nha nghien ciiu cd the linh hoat
de xac d in h cac bien dai dien nhii the nao se diipc d d a ra trong m ot m o
hinh do lddng dda tren din h nghia cua khai niem nghien cdu m a nha
nghien cdu chi dinh. Vi du, xem xet khai niem chi phi chuyen doi cam
nhan. Jones, M othersbaugh va Beatty (2000, trang 262) xac dinh chi phi
chuyen doi cam nhan la “cam n h an cua ngddi tieu dung ve thdi gian,
tien bac va no ldc gan lien vdi thay doi cac nha cu n g cap djch vu”. Cach
tiep can do lddng cua ho trong cac dich vu ngan hdng dda tren ba bien
quan sat (“N oi chung, dd se la n h d n g ngan hang th ay doi rac roi”, “Se
m at rat nhieu thdi gian va no ldc thay doi ngan h an g ”, va “Doi vdi toi,
cac chi phi ve thdi gian, tien bac, va no ldc de chuyen doi ngan hang la
cao”). Do dd, cac tac gia ngam cho rang cd m ot khai niem ve chi phi
chuyen doi cam nhan, cd the d d p c dien ta bang each tru y van m ot tap
hp p ba b ien quan sat). M at khac, B arroso va P icon (2012, trang 532),
xem cac chi phi chuyen doi cam nhan la “m ot khai niem tong hpp tiem
an ddpc bieu hien nhd la m ot th an h phan dai sd c u a cac khia canh khac
nhau cua n o ”. Cac tac gia nay xac d in h m ot tap h p p sau khia canh gom
(chi p h i to n th at lpi ich, chi phi to n th at lien q u a n d e n con ngddi, chi
phi rui ro kinh te, chi phi d an h gia, chi phi thiet lap, va chi phi to n that
tien te), bieu hien cho m ot so dac diem cu the, m oi khia canh bao tru m
m ot phan doc lap cua khai niem chi phi chuyen doi cam nhan. N hd vay,
khai niem cua Barroso va Picon ve cac chi phi chuyen doi cam nhan la
m ot lap luan cua m b hin h do ldcJng nguyen nhan. Tat n h ien , cac bien
do ld d n g ph u thuoc deu cd quan he tddng quan th d e nghiem , va cd le
lien quan den quan he nhan qua, n h d n g chung th d e sd khong the trao
Chu'o’ng 1: Tong quan v l sy phSt trien gin day va dang noi len trong PLS-SEM 11

doi lan nhau theo each m a khai niem m o hinh do liidng ket qua gia
dinh (Rigdon, Preacher va cong sii, 2011). Co n g h ia la, sii tiio n g quan
gida cac bien quan sat khong to n tai vi khai niem nghien cdu ve chi phi
chuyen doi cam nhan diipc gia din h la nguyen n h an chung cua chung.

Them phan kho khan trong viec quyet dinh quan diem do liidng
do la, thpc te khong co m ot dang mo hinh do liidng nguyen nhan -
nhii de nghi trong cac nghien cdu ban dau ve do liidng nguyen nhan
(D iam antopoulos & Winklhofer, 2001) va viec sd dung cac m o hinh do
liidng nguyen nhan trong phan tich thong ke (H air va cong sii, 2011).
H on the, co hai loai bien quan sat ton tai trong cac m o hinh do lddng
nguyen nhan: cac bien quan sat nhan qua va cac bien quan sat tong
hp p (Bollen, 2011; Bollen & Bauldry, 2011). N hd ddpc ngu y bdi ten
cua chung, cac bien quan sat nhan qua ddpc gia d in h la tao ra khai
niem nghien cdu. Do dd, cac bien quan sat can cd sd thong n hat khai
niem tro n g dd tat ca cac bien quan sat tdPng dng vdi d in h nghla cua
nha nghien cdu ve khai niem nay (Bollen & D iam antopoulos, 2016).
Do rong bao phu cua pham vi khai niem la cdc ky quan trong de dam
bao rang cac noi dung cua khai niem ddpc nam bat day du - bo qua cac
bien q u an sat q u an trong ngu y bo qua m ot p h a n cua bien khai niem
m a khai niem the hien. Vi cac bien quan sat n h an qua ddpc ky vong
bao quat tat ca cac khia canh cua pham vi noi dung (Bollen & Bauldry,
2011), cac khai niem nghien cdu ddpc do bang cac bien quan sat nhan
qua (khai niem V, vdi cac bien quan sat td x5 den x g trong T d lieu 1.2)
cd m ot khoan sai so (z 2 trong T d lieu 1.2). Khoan sai so nay nam bat tat
ca cac nguyen nhan khac cua khai niem nghien cdu khong cd trong m o
h in h (D iam antopoulos, 2006). Dieu quan trong la cac bien quan sat nay
ddpc gia d inh la khong cd sai so. P hdong trinh sau dai dien cho m ot m o
h in h do lddng vdi cac bien quan sat nguyen nhan, trong dd w. chi ra sd
gop cua x. (i = 1,..., I) den Y, va z la khoan sai so lien quan den T:
/

i= l

Loai bien quan sat khac, cac bien quan sat tong hpp, cac bien quan
sat nhan qua tdong doi chat che ngoai trd m ot chi tiet nhd. T ddng td nhd
khai niem nghien cdu ddpc do bang cac bien quan sat nhan qua, cac khai
niem nghien cdu ddpc do bang cac bien quan sat tong hpp (bien nghien
cdu Vj vdi cac bien quan sat td x 9 den x ,2trong Td lieu 1.2) cung cd khoan
sai so, nhdng khoan sai sd nay ddpc dat th an h zero (z3 = 0 trong T d lieu
12 Nhung vln nang cao trong PLS-SEM

1.2). Sii khac biet nay co y nghla thiet yeu doi vdi viec m o ta cac m o
hinh do liidng nguyen nhan (Henseler va cong sii, 2014), vi cac bien quan
sat tong hop ho^t dqng nhii n hdng yeu to dong gop cho m ot khai niem
nghien cdu hon la thde sd gay ra no (Bollen, 2011; Bollen & Bauldry,
2011). C hung hinh th an h sd tong hop dai dien day du cho khai niem
nghien cdu b in g cac sd ket hop tuyen tinh, do dd dam bao rang khai niem
nghien cdu khong cd sai sd. Phdong trinh sau m inh hoa m ot m o hinh do
lddng vdi cac bien tong hop, trong dd 71a sd ket hop tuyen tinh cua cac
bien quan sat x vdi trong so tv (Rigdon, 2012):
i
Y=
i=i
cP
Theo truyen thong, cac bien quan sat tong h d p da dddc xem n h d
m ot phdo n g tien de ket hop nhieu bien tddng trd n g cho m ot thde the
m di cd y nghia ddoc xac dinh bang viec Ida chon cac bien quan sat (vi
du: Bollen, 2011; Henseler, Hubona, & Ray, 2016). C hang han nhd, m ot
khai niem m o hinh do lddng cua cac hoat dong tim kiem thong tin co
the dda tren tong sd cac hoat dong m a khach han g tham gia khi tim
kiem thong tin td cac dai ly, tai lieu quang cao, In tern et va cac nguon
khac. Mot n h a nghien cd u khac co the chon m ot b o cac bien khac nhau
de tao th an h m ot th d d c do cua cac hoat dong tim kiem thong tin. Do
dd, cac bien do ld d n g cuoi cu n g x ac d in h y nghia c u a khai n iem nghien
cdu, ngu y la th em hoac bo qua m ot bien quan sat cd the lam thay doi
ban chat cua khai niem nghien cdu. Do dd, theo Bollen (2011), cac
bien qu an sat to n g h op khong can phai chia se sd th o n g n h at khai niem
n h d n g cd the cd m ot chu de tddng td.

Tuy nhien, trong thde te, van con chda ro rang d cho nao de ve ra
mQt dddng gida cac bien quan sat cd sd thong n h a t ve khai niem va
cd sd chia se m ot chu de tddng td. That sd la cac n h a nghien cdu chap
n h an m o t each td n hien m ot din h nghia cua khai niem nghien cd u va
Ida chon cac bien quan sat trong viec van h an h cac m o hinh do lddng
phu hdp vdi din h nghia da dddc m o ta, bat ke khai niem do lddng th d e
te cd dda tren cac bien quan sat ket qua, nguyen n h a n hoac tong h d p
hay khong. N ghia la ho xd ly cac khai niem nghien cdu trong nghien
cdu cua ho n h d la cac th d e the ddn nhat, giong n h d B arroso va Picon
(2012) th d e hien khi d d a ra mot cd sd ly thuyet d a n h gia chuyen sau ve
ban chat va khia canh cua khai niem chi phi chuyen doi cam nhan, trd d c
khi dda ra cac bien quan sat van hanh mo h in h do ld d n g cua khai niem
Chuong 1: Tong quan ve sy phat trien gin day va dang noi len trong PLS-SEM 13

nghien cdu. Do do, ho thde hien tot nhat viec do lddng bang each phat
trien m ot khung khai niem hddng dan va m ot tap hop cac bien quan
sat, m a n h d n g bien quan sat nay m ang y nghia ly thuyet cua khai niem
nghien cdu.
Gia sd rang cac nha nghien cdu sd dung cac bien quan sat tong hop
nh d la nhdng tom ldoc thuan tien cua dd lieu - nhd m ot so nha nghien
cdu thde hien - ham y rang thde tien pho bien cua viec tong hop cac bien
quan sat nhd la sd tong hop de the hien khai niem nghien cdu, mac du
each thde hien thong thddng trong thde tien cho thay tat ca cac nghien
cdu khong phai SEM trong tat ca cac linh vdc nghien cdu (vi du, khi sd
dung bien tong ho p lam dau vao cho phan tich hoi quy), khong cd bat
ky ly thuyet danh gia va lam suy yeu cac nguyen tac cO ban cua do lddng
thich hop. Tuy nhien, cac do lddng rat giong nhau da ddoc phat trien va
thd nghiem can than theo cac hddng dan danh gia m o hinh do lddng pho
bien - hddng dan nay da ddoc ghi chep rong rai trong cac so tay thang
do lddng tieu chuan (vi du: Bearden, Netemeyer, & Haws, 2011; Bruner,
James, & Hensel, 2001). Do do, chinh hoat dong cua cac thang do tong
hop hinh thanh td cac thang do lddng da dddc kiem chdng danh gia cac
bien quan sat tong hop va cac quan sat nhan qua. Vi vay, viec keu goi td
bo PLS-SEM bdi vi no sd dung bien tong hop nhd la sd bieu hien cho
cac bien khai niem (Guide & Ketokivi, 2015) mang ham y bo di tat ca cac
nghien cdu thde nghiem lien quan den cac bien tiem an ve m at thde tien,
bao gom cac nghien cdu ddpc xuat ban trong quan ly dieu hanh, tiep thi,
chien lddc, va cac linh vdc khoa hoc xa hoi khac.

Cac bien q u an sat tong hop khong chi cung cap m ot each thuan
tien de tong hop d d lieu m a con cd the ddoc sd dung de do bat ky loai
tinh chat nao ma khai niem tieu diem de cap den, bao gom thai do, cam
nhan va y d inh hanh vi. Tuy nhien, vdi bat ky dang khai niem do lddng
nao, cac nha nghien cdu can dda ra m ot dinh nghia ro rang va xac dinh
cac bien quan sat tdong dng vdi dinh nghia nay - nghia la chung phai
chia se sd thong nhat khai niem. Ngoai ra, cac mo hin h do lddng vdi cac
bien quan sat tong hop co the ddoc hieu nhd la m ot quy dinh de giam
chieu hddng cua khai niem, trong dd muc dich la co dong cac do lddng
de chung bao quat m ot each day du cac dac tinh noi bat cua m ot bien
khai niem (D ijkstra & Henseler, 2011). Vi du, nha nghien cd u cd the
quan tarn den viec do lddng cac khia canh noi bat cua chi phi chuyen
doi cam nhan bang ba bien quan sat (tong hop), bao gom cac khia canh
dac biet lien quan den nghien cdu hien tai (vi du: chi phi danh gia, chi
phi thiet lap, chi p h i mat m at tien te).
 14 Nhung vim de nang cao trong PLS-SEM

T d lieu 1.3 m d rong sd th e hien cac khai niem m o hin h do lddng

ket qua va 3o liIongliguy^nlT hahT n3o^day 3e~bao gom cac'bieh khai
niem m a chung tim each dai dien (Rigdon, 2012). Bien khai niem la cac
hin h tron n hd d dau set do. Cac bien dai dien ddc lddng dddc hien thi
dddi dang bieu tddng giong ngoi sao tarn canh n h o h o n d gida cac h in h
tron 16n ho n . Cac m ui ten hai dau th in g ddng cho thay m ile do m a cac
bien dai dien nay la nhiing do ld d n g co gia tri - n h d n g m ui ten d d n g nay
cang ngan, cac bien dai dien cang co gia tri.

Tif lieu 1.3 fa i»tfng va cac bten kh«1i mem

K et qua Nguyen nhan

Bidn quan sat ket qua B ien quan sat nguyen nh3n B ien quan sat tong hop

23= 0

Vi tri cua bien dai dien trong hin h tron cang ldn la m ot chi so do
tin cay cua cac phep do lddng khai niem nghien cdu, vi tri d tru n g tarn
phan anh do tin cay hoan hao. Ldu y rin g chung ta phai xem xet cac loai
do tin cay khac nhau, tuy thuoc vao loai m o h in h . Cac thong ke ve do
tin cay nhat quan noi tai n h d C ronbach’s alpha ap dung cho cac m o hin h
do ld d n g ket qua. N hdng cac thong ke nay k h o n g thich hop khi danh
gia cac m o hinh do lddng nguyen nhan vi cac b ien quan sat nhan qua va
tong hop khong nhat thiet phai co moi tdOng quan. Trong cac m d hinh
do lddng nguyen nhan, do tin cay qua quy trin h kiem din h - tai kiem
dinh la phdong tien duy nhat de kiem dinh do tin cay cua cac do lddng.
Tuy nhien, vi n h d n g van de da dang trong viec d a n h gia do tin cay qua
q ua trinh kiem dinh - tai kiem dinh (vi du, C am pbell & Stanley, 1966),
Chuang 1: Tong quan v i su phat trien gin (Jay va dang noi len trong PLS-SEM 15

hau h et cac n ha nghien ciiu th d d n g bo qua danh gia do tin cay khi ddc
tinh cac m o hinh do lddng nguyen nhan.

1/CfC LLfdNG MO HiNH

Cac phan trd d c m o ta cac loai m o hinh do lddng khac nhau va cac bien
qu an sat cua chung. Cau hoi tiep theo la lam the nao de d d c lddng cac
m o h in h do ld d n g va cac m oi quan he cua chung vdi n h d n g th d khac
trong m o h in h p h d o n g trin h cau true. Cac nha nghien cdu th d d n g sd
dung hai phddng phap tiep can de ddc lddng m d hinh cau true: CB-
SEM (Bollen, 1989; Diam antopoulos, 1994; Joreskog, 1978) va PLS-
SEM (Hair, Hult, Ringle, & Sarstedt, 2017; Lohmoller, 1989; Wold,
1982). Mac du ca hai phddng phap ddc lddng cac mdi quan he gida
cac khai niem nghien cdu va cac bien quan sat, chung khac nhau ve
each tiep can cua hai nhom nghien cdu trong viec thde hien dieu dd
(Joreskog 8c Wold, 1982).
CB-SEM ban dau phan chia phddng sai cua moi bien quan sat
th an h 2 phan: ( 1) phddng sai chung (common variance), dddc ddc
lddng td phdd n g sai chia se vdi cac bien quan sat khac trong m d hinh do
lddng cua mot khai niem nghien cdu, va (2) phddng sai rien g (unique
variance), gom cd ca phddng sai dac thu (specific variance) va phddng
sai sai so (error variance). P hddng sai dac thu dddc gia din h la co he
thong va tin cay, trong khi phddng sai sai sd dddc gia djnh la ngau nhien
va k h ong tin cay (sai s6 do do lddng, sai sd do lay m au, sai sd do dac tinh
ky thuat). CB-SEM ban dau tinh toan hiep p h ddng sai cua m ot tap cac
bien (phddng sai chung) va chi co p h d d n g sai dd dddc tin h toan trong
cac giai phap dd d c suy ra tit no. Do dd, CB-SEM tuan theo each tiep can
md hinh nhan to chung (common factor model) trong viec ddc ldpng
cac do lddng khai niem nghien cdu, gia thuyet dat ra dd la p h ddng sai
cua m ot tap cac bien quan sat cd the dddc giai thich m ot each hoan hao
bdi sd ton tai cua m ot bien khong quan sat dddc (nhan td chung) va sai
sd ngau nhien rieng biet (Spearman, 1927;Thurstone, 1947). Cach tiep
can ddc lddng m d hinh nhan td chung hinh th an h day du nen triet ly do
lddng vdi ten goi cac md hinh do lddng phan anh ket qua.

Ve nguyen tic, CB-SEM cung cd the thich dng vdi cac m d hinh do
lddng nguyen nhan vdi cac bien quan sat nhan qua, n h d n g de d&m bao
viec xac dinh md hinh, cac nha nghien cdu phai tuan theo cac quy tdc ky
thuat m d hinh m a doi hoi n hdng rang buoc nhat dinh doi v d i m d hin h
16 NhCWig vein nang cao trong PLS-SEM

(Bollen & Davis, 2009; D iam antopoulos & Riefler, 2011). M acCallum
va Browne (1993) c u n g v d i n h P n g ngdOi k h a c d a giai quyet chi tie tc a c
van de dac diem ky thuat CB-SEM nay. Tuy nhien, nhii Hair, Sarstedt,
Ringle, va M ena (2012, trang 420) liiu y, “(nhPng rang buoc n a y th d d n g
m au th u in vdi nhP ng can nhac ly thuyet va cau hdi dat ra la lieu thiet ke
m o hinh co nen hPdng theo ly thuyet hay ngiipc lai hay khong]”. Thay
vao do, cac hoc gia ve CB - SEM da de xuat m o h in h d a nguyen n h a n
va d a bien q u an sat (M IM IC ) (m ultiple indicators and multiple causes),
bao gom ca cac bien quan sat ket qua va bien quan sat nguyen n h an (vi
du: D iam antopoulos & Riefler, 2011; D iam antopoulos, Riefler, & Roth,
2008). Trong khi cac mo h in h M IM IC c h o p h e p cac nha nghien cPu giai
quyet van de nhan dien, chung khong vdpt qua dilpc van de m a cac m o
hinh do ltfdng nguyen n h an vdi cac bien q u an sat n h a n q ua, khong the
hien diipc phdcrng sai trong k hai niem nghien cd u , bofi vi cac bien quan
sat tUcfng q u an diipc yeu cau bori m o h in h n h an to chung CB-SEM tao
ra m o t dai d ien co gia tri va tit do de dai dien m p t each tiiong thich cho
m o t bien k h ai niem. T heo Lee va Cadogan (2013, tran g 243) liiu y, “cac
n h a nghien ciiu khong nen lam tiidng rang viec dat diipc nhan dang
m ang tinh thong ke cho phep chung ta co dope thong tin ve sp bien
thien cua m ot bien tiem an m ang tinh nguyen n h a n ”. Noi tom lai, viec
Png dung phiiong phap tiep can M IM IC bao gom cac khai niem nghien
ciiu diipc do bang cac bien quan sat nhan qua tro n g CB-SEM luon diia
ra sp sai lech tro ng cac bien dai dien va trong cac giai phap SEM. Ro
rang la CB-SEM chi tot n hat khi no cho phep ilcfc liipng cac m o h in h do
liidng nguyen n h a n gan d u n g vcri cac bien q u an sat n h a n qua.

Tiiong tp, CB-SEM co the thich Png vdi cac m o hinh do liidng
nguyen nh an vdi cac bien quan sat tong hop (B ollen & D iam antopoulos,
2017). Vi cac khai niem nghien cPu diipc do bang cac cac bien quan sat
tong hop diipc xac dinh bdi phiiong sai bang khong, viec xac dinh sai so
p h d o n g sai cua khai niem nghien cPu khong phai la van de. Tuy n hien,
cac van de p h at sinh lien quan den viec xac din h ta t ca cac dodng dan
n ay sin h tP k h ai niem nghien cPu. Grace va Bollen (2007, tran g 206) liiu
y rang van de nay co the d p p c giai quyet bang each xac d in h m ot m oi
q u an he d P d n g d an di vao hoac di ra duy n h a t b an g 1,0 (vi du, m ot m oi
quan he tP cac bien quan sat den bien tong h p p tr e n m oi m o h in h do
lPdng). M ac du cac dac diem ky th u at nhii vay v P p t qua cac van de n h an
dang tham so, nhiing chung han c h e k h a nang d ie n giai cac Pdc lPpng
m o h in h cau true lien quan den tarn quan trong va d p Idn cua cac dPdng
dan co dinh. H on nPa, cac khai niem nghien cPu vdi cac bien quan sat
Chirong 1: Tong quan vb so phat trien gan day va dang noi len trong PLS-SEM 17

tong hop chi co the diipc dua vao nhii cac bien tiem an ngoai sinh trong
khi cung luc yeu cau it nhat mot m oi quan he vdi m ot bien tiem an noi
sinh diipc do liidng phan anh ket qua. Ca hai van de nay gidi han tinh
hOu ich cua chung trong cac ling dung CB-SEM.

Khac vdi CB-SEM, PLS-SEM khong phan chia phiicrng sai thanh
phiiong sai chung va phiiong sai rieng. C u the hem, m yc tieu cua PLS-
SEM la tinh den phiiong sai tong the trong cac bien quan sat hem la
chi giai thich cac m oi tiiong quan gifla cac bien quan sat (vi du: Chin,
1998; Tenenhaus, Esposito Vinzi, Chatelin, & Lauro, 2005). Cach xti
ly khac biet ve phiiong sai trong PLS-SEM so vdi CB-SEM tiiong dng
vdi sit phan biet gida phan tich thanh phan chinh (principal component
analysis) va phan tich nhan to chung (common factor analysis). Trong
khi phan tich thanh phan chinh danh gia phiicrng sai tong va cac lidc
liicrng cac nh an to (thanh phan) bang so ket hop tuyen tinh cac bien
quan sat, phan tich nhan to chung chi danh gia phan phiiong sai chia se
(chung) (Kline, 2015). Lap luan cua each tiep can PLS-SEM do la trong
viec udc liicrng cac m oi quan he cua m o hinh, tat ca phiicrng sai (chung,
rieng va sai so) m a cac bien ngoai sinh co diem chung vdi cac bien noi
sinh nen diicfc diia vao. Cac khai niem co ban cua PLS-SEM la cac chi
so co the diiqc ket hop (m ot each tuyen tinh) de tao thanh cac bien tong
hop la dai dien toan dien cua cac bien tiem an, va cac ket h o p tuyen tinh
nay la cac bien dai dien co gia tri cua cac bien khai niem dang nghien
ciiu. Do do, PLS-SEM tuan theo phiiOng phap mo hinh hon hop trong
viec lidc liicrng cac do liidng cua khai niem nghien ciiu. Thiet ke cua
PLS-SEM diia tren sii tong hop chi de cap den phiiong phap dai dien
cho khai niem nghien ciiu ma phiiong phap nay chi xap xi vdi cac bien
khai niem tro n g m ot mo hinh. Tuy nhien, trong khi PLS-SEM diia tren
cac bien tong h o p ma viec sii dung theo truyen thong diicrc coi la phu
h o p vdi cac m o h inh do liidng nguyen nh an va khong phu h o p vdi cac
m b h inh do liidng ket qua, thiic te la phiiong phap nay sSn sang dap
dng ca hai kieu m o hinh do liidng (vi du, Roldan & Sanchez-Franco,
2012). Tuy nhien, trong cac mo hinh do liidng nguyen nhan, PLS-SEM
xd ly tat ca cac bien quan sat thanh cac bien tong hop. Nghia la, phiiong
phap nay khong cho phep mo hinh the hien khoan sai so cua khai niem
nghien ciiu diipc do liidng vdi cac bien quan sat n h an qua (tiic la, khoan
sai so z3 tro n g cac Til lieu 1.2 va 1.3 la bangO ). Ket qua la tiiong tii nhii
CB-SEM, PLS-SEM chi cho phep cac m o hinh do liidng nguyen nhan
xap xi vdi cac bien quan sat nhan qua.
18 Nhung van d§ nang cao trong PLS-SEM

C uoi cung, de ddc ldpng cac tham so cua m o hinh, PLS-SEM sii
dung-hai m o hinh,-lien quanden-ph-U ang-phapdbc ldpng trong so bien
quan sat dai dien cho sit dong gop cua moi bien quan sat doi vdi bien
tong hop. Mo h in h A tiidng ling vdi trong so tifdng quan xuat phat til
moi tiidng quan hai chieu gida moi bien quan sat va khai niem nghien
cdu; m o h in h B tiidng dng vdi tro n g so hoi quy, tieu chuan trong phan
tich hoi quy binh phiidng toi thieu thong thiidng. Trong so hoi quy
k hong chi tin h den sd tiidng quan gida m oi bien quan sat va khai niem
nghien cdu m a con la m oi tddng quan gida cac bien quan sat. C ho du
phddng phap danh gia trong so bien quan sat nao dddc sd dung, ket qua
bien tiem an luon luon dddc mo h in h xem n h d la m ot bien tong hdp
(Henseler, H ubona, & Rai, 2016). T dc la, tat ca cac d o ld d n g da bien
dd d c chuyen th an h cac th an h phan tro n g so - ngay ca trb n g m o h in h
A - PLS-SEM tin h toan cac th an h phan bang each ket h d p tuyen tin h
cac bien quan sat.
PLS-SEM m ac din h sii dung m o h in h A cho cac khai niem do lddng
ket qua va m o h in h B cho cac khai niem do lddng nguyen nhan. Tuy
nhien, cac nghien cdu gan day cho thay Ida chon p h d d n g thiic trong sd
xap xi doi hoi each tiep can chu dao hon. C u the, Becker, Rai va Rigdon
(2013) chi ra rang doi vdi cac khai niem do lddng nguyen nhan, ddc
lddng theo m o hinh A m ang lai ket qua d d bao to t hdn doi vdi m iu cd
kich th ddc ldn h d n 100 va khi gia tri R2 di td tru n g binh den ldn (nghia
la R2 > 0,30). Doi vdi kich cd m au ld n va gia tri R2 ldn, m d hin h A va
m d h in h B th d e hien tot n h d n h a u ve m at d d bao ngoai m iu . Ve do
ch inh xac cac th o n g sd tro n g m d h in h cau true, m o h in h A cho ket qua
tot nhat khi kich th d d c m au hoac gia tri R2 di td n h d den tru n g binh.
Doi vdi cac k ich th d d c m au ldn hcfn hoac gia tri R 2 ldn h o n , ddc ldpng
theo m d h in h A va B khong khac nh au ve do ch in h xac cua thong sd.

Thy chon sd dung cac m d hinh ddc ldpng khac nhau cho m oi khai
niem nghien cdu cho phep dieu ch in h ddc ldpng m d h in h p h u hpp vdi
m uc tieu cua nghien cdu. TdOng td n h d vay, trong sd bien quan sat cua
cac do lddng khai niem nghien cdu nhat din h cung cd the ddpc xac dinh
trddc (Howell, Breivik, & Wilcox, 2013), do do d u y tri sd on dinh theo
thdi gian trong tarn quan trong tdong doi cua m oi bien quan sat trong
sd h inh th an h bien tong hpp. Mac d u bddc nay tao dieu kien thuan Ipi
cho sd dien giai nhat quan trong cac lan danh gia lap lai cua cung m ot
m d hinh, n h d n g trong sd cd djnh lam can bang bat ky sd khac biet nao
trong tarn q u an trong cua moi bien quan sat de tao th a n h bien tong hpp
Chirceng 1: Tong quan ve sir phat trien gan day v£ dang noi len trong PLS-SEM 19

trong m ot ngfl canh cu the. Tuy nhien, viec hieu cac trong so cua tdng
bien quan sat cung cap nhflng hieu biet quan trpng trong boi canh cu
the cua phan tich. Vi du, khi do lddng khai niem sd hai long doi vdi dich
vu, cac trong so xac dinh cac bien quan sat rieng le nao co tarn quan
trpng dac biet trong viec dinh hinh sd hai long doi vdi dich vu. Mac du,
trpng so cua cac bien quan sat don le co th ek h ac nhau theo ngfl canh, vi
tat ca cac bien quan sat deu co sd chia se tinh thong nhat cua khai niem,
y nghia thdc tien cua khai niem nghien cdu (vi dp: Howell, Breivik, &
W ilcox, 2007) se khac biet trong pham vi dinh nghia cua khai n i ^ i do.

PLS-SEM HAY CB-SEM?

Cac n ha nghien ciiu tfl lau da thda nhan sit khac nhau gifla PLS-
SEM va CB-SEM, lam noi bat cac tinh huong itu tien sCt dung m ot trong
hai phdd n g phap tren (vi du: Joreskog & Wold, 1982). TilPng td, nhieu
nghien cdu da so sanh thdc nghiem sd khac biet trong cac ddc ldpng
m o h inh cua CB-SEM so vdi PLS-SEM. Nhflng sd so sanh nay tap trung
vao kha nang cua PLS-SEM de co ddpc nhflng ket qua td o n g td nhd
CB-SEM. Tuy nhien, PLS-SEM d dang ban dau (Wold, 1982; Lohmoller,
1989) khong ddpc tao ra de bdt chddc CB-SEM m a dung hon la theo
m ot triet ly khac ve do ldbng va m uc dich phan tich. Trong phan tiep
theo, chung toi thao luan hai tieu chi chinh n^y m a chu yeu do la phan
biet PLS-SEM vdi CB-SEM. H air va cong sd (2017) cung cap m ot thao
luan chi tiet hon ve cac dac tinh m o hinh va do lddng lien quan den Ida
chon gifla PLS-SEM va CB-SEM. Cuoi cung, Rigdon (2016) va Sarstedt,
Ringle, va Hair (2017) thao luan cac ly do sfl dung PLS-SEM va nhflng
lap luan con thieu sot ung ho phddng phap nay.

Triet ly Do lddng
M ot sd khac biet ve mat khai niem rat quan trpng gifla PLS-SEM va CB-
SEM lien quan den each thdc m di phddng phap xfl ly cac bien tiem an
trong m o hinh. Vi CB-SEM chi xem xet cac phddng sai chung cua cac
bien quan sat, nen phddng phap xfl ly cac khai niem nghien cflu la cac
yeu to chung. Mat khac, PLS sfl dung tong phddng sai cfla cac bien quan
sat trong phddng phap tao ra cac ket hop tuyen tin h cac bien quan sat
de bieu dien cac khai niem nghien cflu, theo do tao thanh m ot each tiep
can m o hinh tong hop doi vdi SEM.

Mot so nghien cflu m o phong da chi ra rang ddc lddng m o hinh

do lddng PLS-SEM (he sd tai) la Idn hon va ddc lddng m o hinh cau
20 NhCmg v in de nang cao trong PLS-SEM

tru e (he so dddng dan) nhd hon tiiong doi so vdi so lieu thu diiqc bdi
p h u o n g phap-CB--SEM (v-i du;. Reinartz, H aenlein, & Henseler, 2009).
Tuy nhien, cac nghien cdu m o phong nay da xac dinh ddqc m ot each
chung chung cac tong the yeu to chung va thu thap d d lieu td cac tong
the nay (M arcoulides & C hin 2013). Do do, ho dan h gia PLS-SEM tren
co sd cac tong the d d a tren yeu to chung (com m on factor-based) khong
phu hop vdi triet ly do lddng cua phdong phap. That vay, thde te m oi
phiiong phap th o n g ke se hoat dong kem h o n khi m o h in h la khong
chinh xac, va PLS-SEM cung khong ngoai le ve m at nay. Trong thde te,
CB-SEM khong hoat dong tot khi m o hinh dang diiqc ddc ldqng khong
ph u h o p vdi tong the, chang han nhii khi d d lieu ddqc tao ra td m ot mo
h inh tong ho p (Becker, Rai, & R igdon,2013; H enseler va cong sii, 2014).
Becker, Rai va Rigdon (2013) xac dinh cac tong the diia tren sii tong
hop, va cho thay cac ildc ldqng tham sd PLS-SEM cung giong n h d ddc
ldqng cua CB-SEM tren cac gia trj dd khi kich th d d c m lu tang len. Hair,
Hult, Ringle, Sarstedt, va Thiele (2017) cung cap hb trq cho n h d n g phat
hien nay trong danh gia so sanh cac phiiong phap SEM d d a tren sii tong
hop. Do dd, thuat ngd tinh nhat quan dien rong kh i ap dung cho cac ket
qua PLS-SEM n h d la the hien sd sai lech la m ot sd nham lan.

Ket qua CB-SEM d d a tren m b hinh yeu td chung, bao gom cac he
sd dddng dan, sd td tdong quan gida cac khai niem nghien cdu va he sd
tai cua bien quan sat, ddqc danh dau la “diem sd thde” (Bollen, 1989).
N h d n g dieu gi khien chung trd nen khac vdi n h d n g gi chung da td n g gqi
trd d c day? D iem sd that sq khi do cac khai niem n h d thai do, nh an thde,
hoac y dinh la gi? Cac n h a nghien cdu td lau da can h bao rang m b hinh
yeu td chung It khi ton tai trong nghien cdu dng d u n g (Schonem ann &
Wang, 1972). Vi du, trong sd 72 bai bao ddqc xuat ban nam 2012 trong
dd Atinc, Sim m ering va Kroll (2012) xem xet b o n tap chi quan ly hang
dau da kiem din h m ot hoac nhieu m o hinh yeu td chung, it hon 10% co
m b h in h yeu td chung m a khong phai bi td choi. Bdi vi n h d n g ket qua
nay tdong td, khong co m o h in h nao ddqc coi la c o y nghla Idn h o n mb
h in h khac ve sd ton tai hoac ban chat cua cac b ie n khai niem (Rigdon,
2016). Bat ky sd hoat dong cua khai niem nghien cdu nao - bat ke dda
tren m ot yeu td chung hoac cac m b hinh tdng h g p logic - di kem vdi
sd mO ho lien quan den dinh nghla khai niem (vi du: D iam antopoulos,
2005; Gilliam & Voss 2013), td ngd cua bien q u a n sat, va sd ldqng cac
bien quan sat can thiet de nam bat pham vi khai n iem (vi du: DeVellis,
2011). T dqng td n h d vay doi vdi viec xac nhan gia tri do lddng, rat cu
the theo bdi canh, theo do tan dung cac dac tinh c u a d d lieu. N hdng m o
Chiromg 1: Tong quan ve sy phat trien gin day vS dang noi len trong PLS-SEM 21

ho nay, se hop ly hon do la xem cac ddc ldpng cua ca CB-SEM va PLS-
SEM lam dai dien cho cac khai niem dang ddpc nghien ciiu va khong co
gi hon (Rigdon, 2012).

Bat ke cac m oi quan tarn ve n hdng khai niem nay, sd khac biet ma
PLS-SEM tao ra khi ddc ldpng cac m o hinh nhan to chung la rat n h o
n eu cac m o h in h do lddng dap dng cac tieu chuan dd p c de cap toi thieu
ve so ldpng cac bien quan sat va he so tai cua cac bien quan sat, va m o
h in h dd p c xac dinh chinh xac. Cu the, khi cac m o h in h do ld d n g co bon
h oac nhieu bien quan sat va he so tai bien quan sat dap d n g cac tieu
chuan ddpc de nghj, cd rat it sd sai chech trong cac ddc ldpng th a m so
khi d dc ld p n g m o hinh nhan to chung vdi PLS-SEM, n h d Reinartz va
cong sd (2009) da chi ra.

N hdng nghien cdu trddc day cua R bnkko, M cIntosh, A ntonakis,

va Edwards (2016) da tap tru n g vao sd khac biet gida PLS - SEM va CB
- SEM. R einartz va cong sd (2009) da tap tru n g vao su k h ac biet m ang
tin h m o ta gida cac gia tri tong the v& cac ddc ldpng tham so, khong
ban den khai niem cd ban cua thong ke suy dien. R einartz va cong sd
(2009, trang 318) ghi chu cac m o ta ket qua cua tat ca cac dieu kien m o
phong, “cac ddc ldpng tham so khong khac nhau dang ke td cac gia tri
ly thuyet doi vdi CB - SEM dda tren ML (p-value gida 0,3963 va 0,3621)
hoac PLS- SEM (p-value gida 0,1906 va 0,3449)”. Chi khi ddc ldpng
m o h in h dda tren cd mau ldn (N = 10000) va cac m d h in h do lddng
cd nhieu bien quan sat cd he sd tai cao, sd khac biet cd y nghla thong
ke gida 0,58% va 3,23% xay ra. Q uan trong nhat, khi ddc ldpng cac md
hinh nh an td chung, sd khac biet gida ket qua PLS-SEM va CB-SEM rat
it lien quan th d e te doi vdi phan ld n cac dng dung (vi du: Astrachan,
Patel, & W anzenried, 2014). Thde te, cac nghien cdu gain day cho thay
sd sai lech ma CB-SEM tao ra (khong chinh xac) khi sd dung phdong
phap ddc ldpng cac m d hinh tong hpp cao hon nhieu so vdi sd sai lech
ma PLS-SEM tao ra (khong chinh xac) khi sd dung p h d o n g phap de
ddc ldpng cac m d hinh nhan td chung (Sarstedt, Hair, Ringle, Thiele,
va Gudergan, 2016). Nghien cdu tdong td cho thay sai lech cua tham
sd trong PLS-SEM ro rang thap h o n khi sd dung dung p h d d n g phap
de ddc ldpng cac md hinh tong hpp h o n la sd sai lech cua CB-SEM khi
sd dung phdong phap de ddc ld p n g cac md h in h nhan td chung, dac
biet khi kich cd m au nhd. Tom lai, PLS-SEM dda ra khong sai lech khi
ddc ldpng dd lieu td m ot tong th e m d h in h h d n hpp, bat ke m d h in h
do lddng cd d d a tren cac bien quan sat tong h p p hoac cac bien quan sat
22 Nhflng v ln de nang cao trong PLS-SEM

phan anh ket qua hay khong. Sai lech nay phan nao cao hern doi vdi tong
the m o h inh n h an to, n hdng thap xet ve ban than. Ro rang, PLS-SEM
la toi du de ddc lddng cac m o h in h tong hop dong thdi cho phep ddc
lddng cac m o h inh nhan to chung vdi cac bien quan sat ket qua vdi thdc
te khong gidi han (Sarstedt, H air va cong sd, 2016).

Sd khac biet gida PLS-SEM d d a tren bien tong va CB-SEM dda

tren cac yeu to chung khi dO lieu phu thuoc dddc dda ra td m ot m o
hinh nh an to chung doi hoi phai co “sd dieu ch in h ” cua phddng phap
PLS-SEM khi ddc lddng cac m o hinh nhan to. Cu the, trong nd ldc sap
xep cac n h an to chung va cac p h ddng phap SEM d d a tren bien tong hop,
D ijkstrava H enseler (2015a, 2015b) gan day da gidi thieu phdd n g phap
PLS n h a t q u a n (PLSc) (xem B ender & Huang, 2014). PLSc tuan theo
logic m o h in h hoa tong hop n hdng m o phong m o h in h n h a n to chung.
De lam dddc dieu nay, dau tien p h ddng phap ddc lddng cac tham so m o
hinh sd dung thuat toan PLS-SEM chuan va sda lai nhflng ddc lddng
nay giam di bang each sd dung h e so do tin cay n h a t quan p A Sd hieu
chinh nay chi ap dung cho cac m o hinh do lddn g ket qua; ddc lddng m o
hin h do lddng nguyen nhan cua cac bien quan sat to n g h d p van khong
thay doi. Do la, khi ddc lddng cac m d hinh do lddng ket qua, PLSc dda
tren gia thuyet dang nghi ngd rang ket qua CB-SEM la diem chuan de
dan h gia ket qua PLS-SEM. Mot so n h a nghien cdu da ldu y rang cac he
so tai dda tren PLS-SEM dang co sai lech tang len, n hdng do la so vdi
he so tai tfl CB-SEM. Sd khac biet cua cac ddc ld d n g tham so PLS-SEM
khong nen dddc coi la sai lech, n h d n g m ot ket qua khac dda tren m ot
thuat toan theo m ot triet ly khac ve do lddng. Noi chung, rat kho tranh
luan tai sao ngddi dung PLS-SEM lai m uon p h d d n g phap nay cho ra
nhflng ket qua td d n g td n h d CB-SEM, vi phddng p h ap dd da d d d c thda
nhan va chap n h an rong rai n h d m ot each tiep can de co dddc cac giai
phap cho nghien cflu dddc thiet ke xung quanh m o h in h nhan to chung.
M ot ly do tiem an co the la mo h in h khong dddc xac dinh ro hoac ddc
lddng cfla nd dan den cac van de hoi tu trong CB-SEM. Ngoai ra, cac
n ha nghien cflu co le m uon bao gom cac n h an to ch u n g va n h an to tong
hdp trong cung m ot m o h in h va dam bao rang m oi loai dddc ddc lddng
phu hdp vdi khai niem hoa cfla nd.

Muc tieu cua phan tich

Ngoai nhflng khac biet ve triet ly do lddng va each xfl ly khac nhau ve
cac bien tiem an, ban chat diem so bien tiem an d o PLS-SEM va CB-
Chucmg 1: Tong quan vb sy phat trien gbn day va dang noi len trong PLS-SEM 23

SEM tao ra cung co n hdng ket qua doi vdi cac linh vdc ap dung cua cac
p h d d n g phap. Trong CB-SEM, cac diem so bien tiem an k h o n g phai la
duy nhat, nghia la co m ot so lddng vo han cac tap diem so bien tiem
an khac nhau phu hdp vdi m o hinh. Mot he qua quan trong cua do b a t
d in h (so) n h a n to nay la sd tddng quan gida m ot nhan to chung va bat
ky bien nao ben ngoai m o hinh nhan to la khong xac dinh (G uttm an,
1955). Nghia la, moi td d n g quan co the cao hoac thap, tuy thuoc vao tap
diem nhan to ma ban chon. Do do, gidi han nay lam cho CB-SEM cvtc
ky khong phu hop de dd bao (vi du: Dijkstra, 2014). Ngddc lai, PLS-
SEM luon tao ra m ot diem so cu the (tdc la xac dinh) cho moi hon h o p
cho tdng quan sat, m ot khi cac trong so dddc thiet lap. Sd dung cac dai
dien nay lam dau vao de co ddOc m ot giai phap, PLS-SEM ap d u n g hoi
quy b in h phdOng nhd n hat thong th d d n g vdi m uc tieu to id a h o ag ia tri
R2 cua cac khai niem noi sinh. Do dd, PLS-SEM la phdOng phap d d d c d a
thich h o n k h i muc dich phan tich la d d b a o (Albers, 2010, Rigdon, 2014;
Sarstedt, Ringle, Henseler, va Hair, 2014).
Cac nha nghien cdu nen Idu y rang cac gia tri R2 trong ca hai loai
SEM khong the dien giai giong nhau vdi m o hinh hoi quy boi. Gia tri
R2 trong SEM la phdong sai ddOc giai thich trong cac diem khai niem
nghien cdu, va khong phai la phdOng sai dddc giai thich trong cac bien
quan sat. Vi diem so khai niem chi the hien m ot phan cua phddng sai
trong bien qu an sat, gia tri R2 trong SEM la lddng phddng sai dddc giai
thich tro n g khai niem nghien cdu. L d d n g p h d d n g sai bao gom ben trong
diem so khai niem dddc the hien bang p h d o n g sai tru n g binh dddc rut
trich (AVE), la m dc do ma khai niem nghien cdu giai thich phdong sai
cua cac bien quan sat. Do do gia tri R3 trong SEM la phdong sai ddOc
giai thich tro n g AVE.

Hay xem xet nhdng dieu sau ve m ot vi du doi vdi gia tri R2 trong
SEM. CB-SEM chi dda tren phddng sai chung. Neu ban gia dinh rang
p h d o n g sai chung dai dien cho cac bien quan sat cua khai niem nghien
cdu noi sin h la 6f % tong phddng sai, va AVE la 0,60 thi p h d d n g sai cua
AVE thde sd la 36% p h dong sai ban dau trong tong p h d d n g sai (60 %
cua 60% = 36%). Do do, neu gia tri R2 trong CB-SEM td d n g doi cao,
vi d u 0,50, thi ty 1? cua tong phddng sai d d d c d d bao trong khai niem
nghien cd u noi sinh chi la 18% tong phddng sai tro n g cac bien quan sat.
N gddc lai, hay xem xet mot vi du td d n g td vdi PLS-SEM dda tren tong
p h d d n g sai. Neu AVE cua mot khai niem noi sin h la 0,60, va gia tri R2
tro n g PLS-SEM la 0,50, thi ty le cua tong phdong sai dddc d d bao trddc
24 Nhung v ln de nang cao trong PLS-SEM

trong khai niem noi sinh la 30% tong phiictng sai trong cac bien quan
sat eua -khai n ie m n d i sinh, gan -n h d gap-ddi.-PLS-SEM dda tren tong
phiidng sai, va cac chi so AVE luon cao h o n trong CB-SEM, vi vay gia
tri R2 tro n g PLS-SEM co y nghia n h ii m ot d d bao ve sd khac biet trong
cac bien q u an sat cua khai niem noi sinh h d n la gia tri R2 trong CB-SEM.
Tom lai, khi m uc tieu nghien cdu la dd bao, viec Ida chon cac phddng
phap la ro rang - nen thde hien PLS-SEM.

CACH TO CHU’C CAC CHl/ONG c6 n LAI

Cac chddng con lai cung cap thong tin chi tiet h d n ve cac phan tich
nang cao bang each sd dung PLS-SEM, bao gom cac vi du cu the ve
each sd d u n g phan m em SmartPLS 3 cho cac p h a n tich n h d vay. Trong
cuon sach nang cao nay, chung toi m d rong Bddc 7 cua qua trin h ap
dung PLS-SEM (T d lieu 1.4) tii cuon Primer on Partial Least Squares
Structural Equation Modeling, 2nd edition (PLS-SEM)) (H air va cong sd,
2017). Phan tich PLS-SEM nang cao se cho phep b an hieu ro h o n va giai
thich ket q ua nghien cdu, va cung cap cac dang p h an tich va chan doan
so lieu m a cac bien tap vien yeu cau ngay cang tang. T d lieu 1.5 xac dinh
cac chddng va chu de dddc de cap trong cuon sach nay.

C hd d n g 2 gicfi thieu ve cac chu de m o h in h n an g cao, bat dau vdi

cac m o h in h th an h phan th d bac. M ot m o hin h th d n h phan th d bac the
hien m o t khai ni?m nghien cdu tong quat hdn, d d d c do lddng d m dc
khai quat cao hdn, ddng thdi bao gom m ot so th a n h phan con, gom cac
dac diem cu the h d n cua bien khai niem dddc the h ien trong khai niem
nghien cdu nay. Vdi suf phde tap ngay cang gia tang cua cac ly thuyet
va cac m o h in h nhan - qua trong cac nganh k h o a hoc xa hoi, cac nha
nghien cdu th d d ng sd dung cac m d hin h phde tap nay trong cac nghien
cdu PLS-SEM cua ho (vi du: Ringle va cong sd, 2012). T ddng td n h d
vay, cd sd quan tarn tang-len-trong viec-mo phong cac m oi quan he phi
tuyen gida cac khai niem nghien cdu. Khi m oi q u a n he gida hai khai
niem nghien cdu la phi tuyen, m dc do tac dong giCta hai khai ni?m phu
thuoc khong chi vao m dc do thay doi trong khai niem ngoai sinh ma
con ve gia tri cua no. Trong phan th d hai cua C h d d n g 2, chung toi gidi
thieu cac nguyen ly cua m d hinh phi tuyen va m d t a each chay cac phan
tich td d n g dn g trong SmartPLS 3.
Chuccng 1: Tong quan ve su phat then gin day va dang noi len trong PLS-SEM 25

Tu’ lieu 1.4 Quy trinh he thong ap dung PLS-SEM

Budc 1 Xac djnh m6 hinh clu true

' ' ■- • , . ' ..........

Budc 2 Xac dinh cac m6 hinh do ludng

B udc 3 Thu thSp kiim tra d u li§u

Budc 4 U’dc Itttjmg md hinh dudng d§n PLS

. K ! Oanhgi
Buoc 5a
b V .j'b **

Danh gid ket qua PLS - SEM cua mo hinh do .wWiu

Buoc 5b Hr : : “ WBm ^ v i " *•-
b -A nguyen nhan

Budc 6 i Danh gia ket||ua PLS -SEM ctia rrio hinh cau true I

Budc 7 : *5 Phdn tich P LS - SEM n£ng cao Pi

jm

Budc Suy d iin ket qua va rut ra k it ludn

Trong C h d d n g 3, chung ta thao luan hai dang danh gia m o hinh

nang cao, bat dau vdi phan tich bo td (tetrad) khang din h (CTA-PLS),
o phep danh gia thde nghiem lieu d d lieu cd ho trd dac diem m o hin h
o ld d n g nguyen nhan hoac dac diem md hinh ket qua hay khong. Tiep
theo, chung toi gidi thieu ve phan tich bieu do tarn quan tro n g - hieu
suat (IPM A), la sd m d rong ket qua m o hinh cau true chuan cua cac ddc
ldong he so dddng dan bang each them m ot chieu hddng (khia canh) de
xem xet cac gia tri tru n g binh cua cac diem so bien tiem an.
26 Nhung v in de nang cao trong PLS-SEM

Td lieu 1.5 j C i c ch u o n g .v a C hu M i r o n * tu
' ....'is®
ChuOng Chu de
Mo hinh nang cao
- M o h in h th a n h p h a n th d b a c (H ierarchical com ponent
2
m odels)
- Cac m oi quan h e phi tuyen (N onlinear relationships)
D anh gia m o hin h nang cao
3 - Phan tich bo td khang dinh (CTA-PLS)
- Phan tich bieu do tarn quan trong * hieu suat (IPMA)
Mo hinh khong dong nhat quan sat dd q c (M odeling
O bserved H eterogeneity)
4
- D anh gia bat bien do lddng (M IC O M )
- Phan tich da n h o m (M ultigroup analysis)

Mo hin h khong dong nhat k h o n g quan sat ddqc

(M odeling U nobserved H eterogeneity)
5
- PLS h o n help h d u han (FIM IX-PLS)
- Phan doan din h h d d n g dd bao (PLS-POS)

Hai chddng ke tiep de cap den cac khai niem khac nhau cho phep
cac nha nghien culu co th£ xd ly d d lieu vdi m o h in h co d d lieu khong
dong nhat. Chdcfng 4 cung cap goc nhin khai q u at ve sd khong dong
nhat quan sat ddqc va khong quan sat ddqc, cho thay viec khong quan
tarn den cau true dd lieu khong dong nhat co the gay ra ket qua sai lech.
Tiep theo, chung ta thao luan ve bat bien do lddng, dd la m oi quan tarn
chinh trddc khi so sanh cac nhom d d lieu. C h d d n g 4 ket thuc bang viec
gidi thieu cac loai phan tich da nhom khac n h a u ddqc sd dung de so
sanh cac th o n g so (thddng la cac he so d d d n g d a n ) gida hai hay nhieu
nhom dd lieu. Trong khi cac p h d o n g phap nay cho phep tin h toan doi
vdi sd khong dong nhat quan sat ddqc, th d d n g h ay h o n khong, n h d n g
tinh huong tro n g dd sd khac biet lien quan den tin h k hong dong nhat
khong quan sat ddqc ngan can khong cho ra ket q u a chinh xac khi phan
tich tren cap do d d lieu tong hop cac cap do d d lieu theo nhom . ChdOng
5 gidi thieu hai p h ddng phap, h 6 n h q p h d u haul PLS (FIMIX-PLS) va sd
phan doan theo dinh hddng dd bao trong PLS (PLS-POS), cho phep cac
Chuang 1: Tong quan ve su phat trien g§n day va dang noi len trong PLS-SEM 27

n ha nghien ciiu xac dinh va xd ly sd khong dong nhat khong quan sat
d d d c trong cac mo hin h ditdng dan PLS.

TINH HUONG NGHIEN CUD M IN H HOA

Mo hinh danh tieng cong ty

Cach hieu qua n h at de hoc each sir dung phdd n g phap thong ke la
ap dung no vao m ot bo dd lieu. Xuyen suot cuon sach nay, chung toi sir
dung m ot vi du duy nhat cho phep ban lam dieu do. Vi du nay d d d c rut
ra tir m ot loat cac nghien ciiu dddc cong bo ve d an h tieng cua cong ty,
dieu nay la dii de cac nha nghien cdu hieu dddc td nhieu llnh vdc khac
nh au , qua do giup tao dieu kien cho viec hieu cac phan tich dddc trin h
bay tro n g cuon sach nay. C h in h xac hdn, chung toi dda vao m o h in h
danh tieng cong ty cua Eberl (2010), vi du nay da d d d c H air va cong
sd (2017) sd dung trong cuon sach c0 ban PLS-SEM. M uc dich cua m o
hin h la giai thich cac anh hddng cua dan h tieng cong ty len sd hai long
cua khach hang (CU SA) va cuoi cim g la long tru n g th a n h cu a khach
hang (CUSL). Danh tieng cong ty dai d ien cho m ot danh gia tong the
cua cong ty biri cac ben lien quan (Helm, Eggert, & Garnefeld, 2010).
Theo Schwaiger (2004), dan h tieng cong ty dddc do bang hai khia canh.
M ot khia canh dai dien cho cac danh gia n h an th d c cua cong ty va do
lddng khai niem bang cdch mo ta n an g ldc cua cong ty (COMP). Khia
canh thd hai thu thap cac phan doan tinh cam va danh gia sd nhan thdc
ve do yeu thich cua cong ty (LIKE). Schwaiger tiep tuc xac d in h bon yeu
to tac dong den d an h tieng: chat lddng (QUAL), hieu qua (PERF), tinh
hap dan (ATTR) va trach nhiem xa hoi cua doanh nghiep (CSOR). T d
lieu 1.6 trin h bay m o hinh danh tieng cua cong ty.

Cac m o hinh do lddng cua cac khai niem LIKE, COMP, va C USL co
ba bien quan sat k et qua, trong do CUSA dddc do bang m o t bien quan
sat duy nhat. Ldu y rang ban nen tran h sd d u n g cac bien ddn le kieu
nay, dac biet la trong cac ph an tich PLS-SEM (vi du: D iam antopoulos,
Sarstedt, Fuchs, Kaiser, & Wilczynski, 2 0 12;Sarstedt, Di am antopoulos, &
Salzberger, 2016; Sarstedt, Diam antopoulos, Salzberger, & Baum gartner,
2016). Tuy nhien, chung toi da dda ra bien quan sat ddn nay cho m uc
dich m inh hoa. Cuoi cung, bon khai niem ngoai sinh dddc do lddng
bang tong cong 21 bien quan sat nguyen nhan (formative indicator). T d
lieu 1.7 cung cap tong quan ve tat ca cac bien quan sat va noi dung cua
no. N hdng ngddi dd d c hoi d an h gia cac phat bieu lien quan den thang
28 Nhdng v in de nang cao trong PLS-SEM

do Likert 7 diem , vdi diem so cao hcfn bieu thi m d c do dong y cao hdn
vdi m o ttu y e n bd cu. the. T rong-trddng-hdp-bien-quan sat cusa, diem
so cang cao hd n bieu thi m dc do hai long cang cao. Sd hai long va long
tru n g th anh da dddc do ld d n g d d a tren quan diem danh gia cua ngddi
dddc hoi doi vdi nha cung cap dich vu rieng cua ho.

Cach tiep can do lddng da dddc hieu chinh ta i cac quoc gia khac
n h au va dddc ap dung trong nhieu nghien cdu (Eberl & Schwaiger,
2005; Raithel & Schwaiger, 2014; Raithel, W ilczynski, Schloderer, &
Schwaiger, 2010; Schloderer, Sarstedt, & Ringle, 2014). N ghien cdu da
chi ra rang, so vdi cac do lddng danh tieng khac, each tiep can nay thdc
hien thuan ldi ve m at gia tri hoi tu va gia tri d d b ao (Sarstedt va cong
sd, 2013). Tap dd lieu d d d c sd d u n g cho ta t ca cac p h a n tich trong cuon
sach nay bat nguon td H air va cong sd (2017) va co 344 quan sat.

N ang ld c (COMP)
[Cong ty...] la m ot doi thu c^inh tra n h hang dau tren thi
com p_l
trd d n g cua nd.
Theo n h d toi biet, [cong ty...] d d d c nh an biet tren toan
comp_2
the gidi.
comp__3 Toi tin rang [cong ty...] hoat dong of m ot dang cap cao.
 ,. -pong quan v§ su phat trien gSn day va dang noi len trong PLS-SEM 29
Chuo^g1-

Bien quan sat va noi dung ca c bien

D o yeu thich (LIKE)
[Cong ty ... ] la m ot cong ty ma toi co the nhan dang tot
lik e j h o n vdi h o n so vdi cac cong ty. khac
[Cong ty ... ] la m ot cong ty toi se hoi tiec nhieu neu no
like_2 khong con ton tai vi toi m uon no h o n cac cong ty khac.
Toi quan tarn [cong ty...] nhd la m ot cong ty dang de
like_3
yeu thich.
Long trung thanh cua khach hang (CUSL)
Toi se gidi thieu [cong ty ...] den ban be va ngddi than
cusl_l cua toi.

Toi m uon gidi thieu [cong ty] den ban be va ngiidi than.
N eu toi phai chon m ot lan nfla, toi se chon [cong ty...]
cusl_2
la nha cung c^p djch vu dien thoai di dong cua toi.

Toi se van la mot khach hang cua [cong ty ... ] trong

cusl_3
td o n g lai. v
Sd hai long cua khach hang (CUSA)
Neu ban xem xet trai nghiem cua ban vdi [cong ty...],
cusa m dc do hai long cua ban doi vdi [cong ty...] n h d the
n ao i>
C hatlddng (QU AL)
V- San pham /dich vu ddOc cung cap bdi [cong ty...] co
qual_l
chat ldong cao.
[Cong ty] la m ot nha doi m di chd khong bat chddc vdi
quaU
sd ton trong [nganh... ].
quaU San pham /dich vu cua [congty...] co gia tot.
qual_4 Dich vu (cong ty ... | tot.
Cac quan tarn cua khach hang ddOc danh gia cao tai
qual_5
[congty...].
qual_6 [Cong ty...] la doi tac dang tin cay doi vdi khach hang.
q u a lj7 [ C ong ty... ] la m ot cong ty dang tin cay.
qual__8 Toi rat ton trong [cong ty...].
30 Nhung v in dd nang cao trong PLS-SEM

Bien quan sat va noi dung c a c bien

Sp hieu qua (PERF)
perf.,1 (Cong ty ... ] la m ot cong ty diipc quan ly tot.
perf_2 [Cong ty...] la m ot cong ty c o k in h te on dinh.
Rui ro kinh doanh cua [cong ty...] la khiem ton nhat so
Perf - 3
vdi cac doi th u canh tranh.
perf_4 [Cong ty...] co tiem nang tang triidng.
p erf.5 [Cong ty... ] co tarn n h in ro rang ve tiiong lai cua cong ty.
Trach nhiem xa hoi cua cong ty (C SO R)
csor_l [Cong ty ...] cP xP theo each co y th d c v l m at xa hoi.
[Cong ty...] thang than trong vide cung cap thong tin
csor_2
cho cong chung.
[Cong ty...] co thai do cong bang doi vdi cac doi thu
csor_3
canh tranh.
csor_4 [Cong ty ...] quan tam ddn viec b ao ve m oi triidng.
csor_5 [Cong ty...] khong chi quan tam ve lpi nhueui.
Stfthu hut (A T T R )
[Cong ty...] th an h cong trong viec thu hu t nh an vien
attr_l
chat liipng cao.
attr_2 Toi co the thay m inh dang lam viec tai [cong ty...].
Toi thich dang ve ben ngoai cua [cong ty...] (cong ty,
attrJ3
toa nha, cPa hang, v.v.).

sm PLS-SEM
Phan mer
De thiet lap va tide lPpng cac m o hinh dPdng dan PLS, ngiidi dung
the Ipa chon tP m ot-loat cac ch iio n g trin h p h a n m em . M ot vi du
pho bien dau tien cua m ot chiiong trin h phan m em PLS-SEM la PLS-
G raph (C hin, 1994), la m ot giao dien do hoa vdi LVPLS cua Lohm oller
(1987), the he phan m em PLS dau tien. So vdi LVPLS, m ot phan m em
yeu cau ngiidi dung nhap cac lenh thong qua mQt trin h soan thao van
ban, PLS-Graph cho thay m ot sp cai tien d an g ke, dac b iet la ve sp than
th ien vdi ngiidi sd dung. Tuy n h ien , PLS-G raph da khong diipc p hat
trie n th e m tro n g n h d n g nam gan day. Tiiong tp n h ti vay doi vdi m ot so
chiiong trin h phan m em nhP VisualPLS (Fu, 2006) va SPAD-PLS (Test
Chirong 1: Tong quan vd sy phat trien gin day va dang noi len trong PLS-SEM 31

& Go, 2006); xem them danh gia cua Temme, Kreis, va H ildebrandt
(2010). Vdi sd pho bien ngay cang nhieu cua PLS-SEM trong cac llnh
vile khac nhau, mot so chdOng trinh khac co giao dien do hoa than
thien vdi ngddi dung da ddpc gidi thieu ra thi trd d n g , nhii goi PLSPM
cua XLSTAT, Adanco (H enseler & Dijkstra, 2016); PLS-GUI (H ubona,
2015); va dac biet la SmartPLS (Ringle va cpng sii, 2015; Ringle va cong
sii, 2005). Cuoi cung, ngiidi dung co kinh nghiem trong m oi triidng
phan m em thong ke R cung co the thde hien tren cac goi n h d semPLS
(M onecke & Leisch, 2012) va plspm (Sanchez, Trinchera, & Russolillo,
2015), giup phan tich linh hoat cac m o hinh dddng d an PLS. Cho den
nay, SmartPLS 3 la chdcfng trin h toan dien va tien tien nhat trong linh
vdc nay va la co sd cho tat ca cac vi du nghien cdu tinh huong m inh hoa
trong cuon sach. Cac phien ban dan h cho sinh vien cua ph an m em co
san m ien phi tai vw w .sm art-pls.com . Phien ban sinh vien cung cap hau
n h d tat ca chdc nang cua p h ien ban day du n h d n g dd p c gidi han trong
bd dd lieu co toi da 100 quan sat. Tuy nhien, vi tap d d lie u ddpc sd dung
trong cuon sach nay co hon 100 quan sat (chinh xac la 344), ban nen sd
dung phien ban chuyen nghiep cua SmartPLS, co san dddi dang phien
ban dung thd 30 ngay tai www.smartpls.com. Sau thdi gian dung thd,
ap dung m ot khoan phi sd dung. Phi ap dung cho cac khoang thdi gian
khac nhau (vi dy: 2 thang, 1 nam , 3 nam ) va co the m ua thong qua trang
web cua SmartPLS. Trang web SmartPLS bao gom khu vdc tai ve phan
m em , bao gom phien ban cu SmartPLS 2 (Ringle va cong sd, 2005) va
nhieu tai nguyen b6 sung khac n h d giai thich ng an ve PLS-SEM va cac
chu de lien quan d£n phan m em , danh sach cac tai lieu ddpc de xuat,
cau tra ldi cho cac cau hoi thddng gap, video hddng d in de b it dau sd
dung phan m em va d iln dan SmartPLS, cho phep ban thao luan cac chu
de PLS-SEM vdi n h d n g ngddi dung khac.

Thiet lap mo hinh trong SmartPLS

Trddc khi ban ve m o h inh cua ban trong SmartPLS 3, ban can co dd
lam cd sd cho viec chay m o hinh. D d lieu chung toi sd d ung vdi mo
hinh danh tieng co the ddpc tai xuong dddi dang file cd duoi (.csv) hoac
tap tin van ban (.Lxt) trong phan tai xuong tai w w w .pls-sem .com . N hap
vao Save Target As ... de ldu d d lieu vao m ot th d m uc tren o cdng cua
ban va sau do nhap Close. SmartPLS cd the sd d u n g ca hai dinh dang
tep d d lieu (tdc la .csv hoac .txt). Lam theo hiidng dan tren m an hinh
de ldu m ot trong hai tep nay tren o cdng cua ban. Vdi d d lieu nay, n h d
diipc giai thich trong C hiiong 2 cuon sach cua H air va cong sU (2017),
ban cd the tao ra mo hinh d iidngdanP L S nhd the hien trong Td lieu l.8.
32 Nhirng v in 6b nang cao trong PLS-SEM

Ngoai ra, ban co the tai tap tin D anh tieng cong ty (Corporate
Reputation - Advanced PLS-SEM B ook.zip) cho SmartPLS til www.
pls-sem.com. LUu file tap tin d u an nay tren m ay tinh cua ban (hoac
thiet bi khac). Bay gib chay phan m em SmartPLS 3 bang each nhap vao
bieu tu'Ong m ay tinh de ban co san sau khi cai dat ph an m em tren m ay
tin h cua ban. Mot kha nang khac la vao th ii m uc ncfi ban da cai dat phan
m em SmartPLS tren m ay tin h cua ban. N hap vao tep chay SmartPLS va
sau do tren tab Run de bat dau p h an m em . De n h a p tap tin cong trin h
danh tieng cua cong ty vao phan m em SmartPLS, sii dung tuy chon
File Import Project from Backup File trong th a n h m enu SmartPLS.
Sau do, ban se thay cong trin h dan h tieng cua cong ty trong Project
Explorer SmartPLS. M b ra cong trin h nhii the h ien trong T u lieu 1.9 va
nhap dup vao Extended m odel. Sau do, m o hin h d ubng dan PLS n h u
trong TU lieu 1.8 xuat hien tro n g cba so m o h in h SmartPLS.
Tuan thu cac quy trin h cb tin h he thong de ap d u n g PLS-SEM dUOc
trin h bay trUbc trong cuon cO ban cua H air va cong sii (2017), cac bubc
tiep theo se danh gia cac m o h in h do ket qua va nguyen nhan, tiep theo
la dan h gia mb h inh cau true. NgUbi doc nen th a m khao cuon Primer
on Partial Least Squares Structural Equation M odeling (PLS-SEM) (H air
va cpng sU, 2017) de thao luan chi tiet va m in h h o a cac bubc phan tich
nay. Cac nghien cbu tinh huong m in h hoa trong cac ch u o n g sau se bat
Chinang 1: Tflng quan ve sir phat trien g§n day va dang noi len trong PLS-SEM 33

dau tit day, gia d inh rang chat lddng cua cac m o hinh do lddng goc va
m o hinh cau true cua m o hinh cau true da dflcrc thiet lap.

Td lieu 1.9 Giao dien SmartPLS Project Ex

ST|T| Project Explorer: ^

a Corporate Reputation - Advanced PLS-SEM Book]
J l, Extended model
HCM (reflective-formative)
r i x- i v i v i c i i c c u v c ' i u i m o i i Y c j

^ HCM (reflective-reflective)
Inadmissible HCM (reflective-formative)
Simple model
Q Corporate reputation data case wise deletion [335 records]
:'r j Corporate reputation data (344 records]
----------------------------------------------------------------------------

TOM TAT
tr v i
• H ieu nguon goc va sit tien triefn cua PLS-SEM. Cac tien than
cho phiiong phap PLS-SEM la hai thu tuc lap (tdc la hoi quy
th anh phan chinh va PLS-R) sd dung ddc lilpng bin h phiidng
nh d nhat de phat trien cac giai phap cho cac m o hinh don va m o
h in h da thanh p h an va cho m oi titong quan tieu ch uan. So phat
trien tiep theo cua cac quy trin h nay dddc tao ra bdi H erm an
Wold dan td i thuat toan NIPALS va m ot phien ban tong quat
sfla doi Cua th u at toan PLS tap tru n g vao viec tim ra cac bien
tiem an. Trong nhflng nam 1980, H erm an Wold da de xuat “m o
h in h thiet ke cO ban m em ” cua m in h dfldi dang PLS-SEM nhfl
la m ot giai phap thay the cho CB-SEM, CB-SEM dddc danh gia
la m o h inh cflng vi nhflng gia dinh khat khe hon nhieu ve phan
phoi dfl lieu va kich cfl m au (vi du: Falk & Miller, 1992). Mac du
ca hai phfldng phap tiep can da dfloc phat trien cung thdi diem,
CB-SEM da trd nen pho bien rong rai h o n do tinh san sang sdm
ho n thong qua phan mem LISREL ke tfl cuoi nhflng nam 1970.
PLS-SEM bat dau phat trien cho den khi ra m at phan m em PLS-
G raph cua W ynne C hin vao gifla nhflng nam 1990. Vdi viec
p h a th a n h SmartPLS 2 vao nam 2005, viec sfl dung PLS-SEM da
tang len theo cap so nhan.

Hieu dddc cac nguyen t i c va nhung phat trien gan day ve do

lflding. Cac bien khai niem the hien y tfldng hoac y nghi rong
34 Nhung v§n <J§ nSng cao trong PLS-SEM

rai ve cac khai niem trd u tdpng m a cac n h a nghien ch u thiet

- la p va-de xuat-de-do -lddng-trong-nghien ch u cua h p -q u a cac
khai niem nghien chu. D da tren dinh ng h ia khai niem , cac m o
h in h do ld d n g the h ien each do ld d n g khai niem th o n g qua
m o t tap bien quan sat. Cac m o hin h do ld d n g co th e p h an anh
ket qua, sh d u n g cac bien quan sat ket q u a (th e la, p h an an h
ket qua), hoac m ang tin h nguyen n h an , sh d u n g cac bien quan
sat ph an anh nguyen n h an hoac tong hpp. Trong khi cac khai
niem nghien chu diipc do bang cac b ien q u an sat nh an qua
co khoan sai so thi day khong phai la triid n g hpp xay ra cho
cac bien quan sat tong hpp, xac din h to an bo khai niem . Theo
truyen thong, cac bien quan sat tong h p p da dtipc xem nhii la
m o t p h d d n g tien de ket h p p m ot so bien de dai dien cho m ot
so doi tPpng m hi co y nghia diipc xac d in h bang viec Ida chon
cac bien quan sat. Tuy n hien, cac nghien chu gan day cho rang
cac cac bien quan sat tong hpp co the d d p c sh dung de do bat
ky loai thupc tin h nao m a khai ni$m tro n g tarn de cap den, bao
gom thai do, n h an thhc va y dinh hanh vi. Tat ca cac do lddng
cua cac bien khai niem la xap xi hoac la cac dai dien cho cac
bien khai niem , doc lap th each chung d d p c b at nguon. Do do,
khong phan biet chat ldpng m a m ot bien khai niem ve m at ly
thuyet ddpc n h an m an h va ddpc xac d in h ro rang ve m at hoat
dong, va sd k hat khe bao gom qua trin h phat trien m o h in h
do lddng, bdi vi cac bien tiem an xuat p h a t td d d lieu von da
k h o n g h o a n hao, bat ky ph ep do n a o tro n g cac m o h in h cau
tru e ta o ra chi dai d ien cho cac bien tie m an.

• T im hieu sd khac biet co ban gifla PLS-SEM va CB-SEM va

hieu khi nao sd dung tdng phddng phap. PLS-SEM nhan
m an h sd dd bao trong khi dong thdi d ap dng n h u cau ve dd
lieu va dac diem ky thuat cua cac m oi q u a n he. PLS-SEM toi
da hoa p h d o n g sai ddpc giai thich cua cac bien so tiem an noi
sin h bang each ddc ldpng cac moi quan h e m o hin h rieng phan
tro n g m ot chuoi hoi quy cua cac hoi quy b in h p h d o n g nhd
n h at b inh thddng. N gdpc lai, CB-SEM d d c ld p n g cac tham so
m o h in h sao cho sd khac biet gida hiep p h d o n g sai ddc ldpng
va m a tran phdP ng sai m au ddpc giam th ieu . M ot sd khac biet
khai niem rat quan tro n g gida PLS-SEM v a CB-SEM lien quan
Chipong 1: Tong quan ve sir phat trien g§n day va dang noi len trong PLS-SEM 35

den each moi p h d d n g phap xd ly cac bien tiem an trong m o

hinh. CB-SEM xem xet cac khai niem nghien cdu n h d la cac
yeu to thong th d d n g , tro n g khi PLS-SEM theo m ot goc nhin
mo h in h hon h o p sd dung trong so tong hop cua cac bien quan
sat de bieu dien cac khai niem . Q uan tro n g hdn, cac dai dien
tao ra bdi PLS-SEM khong diiOc gia din h la giong het vdi cac
khai niem m a ch u n g do ld d n g - giong n h d trd d n g hop do
liidng khai niem cua CB-SEM. PLS-SEM khong bi han che bdi
cac van de nhan dang, ngay ca khi m o h in h trd n en phde tap
- m ot tinh huong thiidng khong cho phep sd dung CB-SEM.
H on nda, PLS-SEM co the xd ly tot hon cac m o h in h do liidng
nguyen nhan. Cac nha nghien cdu nen xem xet hai phiiong
phap tiep can SEM la bo sung nhau va ap dung ky thuat SEM
nao phu hop n h at vdi m uc dich nghien cdu, dac diem dd lieu
va thiet lap mo h in h cua ho.

CAU HOI ON TAP

1. Ai da phat trien thuat toan PLS-SEM tong quat va y dinh dang
sau sd phat trien cua nd la gi£*
2. Sd khac biet gida PLS-R va PLS-SEM la gi?

3. Khac biet gida cac mo hinh nhan to chung va m o hinh tone hop
la gi?
4. Sd khac n h a u gida cac m o h in h do lddng ket qua va m o hinh do
lddng nguyen nh an la gi?

5. PLSc la gi va y dinh dang sau sd phat trien cua no la gi?

AU HOI TU DUY PHAN BIEN

1. Tai sao PLS-SEM la p h d d n g phap du tien hon CB-SEM de dd
bao?

2. Hay binh luan ve tuyen bo sau: “Cac bien quan sat trong cac m o
hinh do lddng nguyen nhan thi khong co sai so.”
3 . Sd khac nhau gida cac bien quan sat nhan qua va tong hop la gi?

4. N hdng loti ich cua viec sd dung PLSc qua CB-SEM va PLS-SEM
la gi?
 36 Nhung v§n cle ndng cao trong PLS-SEM

THUAT N G tf

Attribute Thuoc tinh

Average variance extracted P hiiong sai trich tru n g b inh
(AVE)
Causal indicators Cac bien quan sat n h an qua
CB-SEM SEM dda tren hiep p h d o n g sai
C om m on factor m odel M o hin h nhan to chung
C om m on variance PhdOng sai chung
Com posite indicators Cac bien quan sat tong h<?p
Com posite m odel approach Cach tiep can m o hin h tong hop
C onceptual variables Cac bien khai niem
C onsistent PLS (PLSc) PLS n hat qu an
Construct definition D inh nghla khai niem nghien cdu
Constructs Cac khai n iem nghien cdu
Covariance-based structural M o h in h p h d o n g trin h cau true
equation m odeling (CB-SEM) d da tren hiep p h d o n g sai
Effect indicators Bien q u an sat ket qua
Error variance Sai so p h d o n g sai
Factor (score) indete Do bat dinh (so) n h an to
Factor-based SE SEM dda tren nhan to
Focal object Doi tdOng tap trung
Formative m easurem ent m odel M o hinh do liidng nguyen nhan
Indicators Cac bien q u an sat
Items Cac bien do liidng
La
Latent variable Bien tiem an
CjZtent variable scores_______ Diem so bien tiem an
M anifest variables Cac bien q u a n sat/bieu thi
M easurement m ode Mo hinh do liid n g
MIMIC m odel Mo hinh M IM IC
Multiple indicators and multiple M o hinh da bien quan sat va da
causes (MIMIC) model nguyen n h a n
Partial least squares algorithm Thuat toan b in h p h d o n g toi thieu
rieng p h an
Chuong 1: Tong quan ve su phat trien gin <J3y va dang noi len trong PLS-SEM 37

Partial least squares path M o hin h ditdng dan binh phiidng

m odeling toi thieu rieng phan
Partial least squares regression H oi quy b in h phiiong toi thieu
(PLS-R) rieng phan
Partial least squares structural M o h in h phiiong trin h cau true
equation m odeling binh phiidng toi thieu rieng
phan
PLS algorithm T h u it toan PLS
PLSc PLSc o
PLS-R PLS-R O '
PLS-SEM PLS - SEM
Principal com ponents regression H oi quy cac thanh phan chinh
Reflective indicators Cac bien quan sat ket qua
Reflective measurement model M o h in h do liidng ket qua
Specific variance Phiiong sai dac thu
Theoretical m odel M o hinh ly thuyet
Unique variance Phildng sai don nhat/rieng

TAI LIEU DOC THEM

Bollen, K. A. (2011). Evaluating effect, composite, and causal indicators
in structura I t * ion models. M IS Quarterly, 35, 359-372.
Bollen, K. A., & Diam antopoulos, A. (2017). In defense o f causal-
formative indicators: A m inority report. Psychological Methods.
Advance online publication, doi: 10.1037/met0000056

lair, J. E, Hult, G. T. M., Ringle, C. M ., & Sarstedt, M. (2017). A primer

on partial least squares structural equation modeling (PLS-SEM)
(2nd ed.). Thousand Oaks, CA: Sage.

Hair, J. F., Hult, G. T. M., Ringle, C. M., Sarstedt, M., & Thiele, K. O.
(2017). M irror, m irror on the wall. A comparative evaluation o f
com ponent-based structural equation m odeling m ethods. Journal
o f the Academy o f M arketing Science. Advance online publication,
doi: 10.1007/sll747-017-0517-x
Rigdon, E. E. (2012). Rethinking partial least squares path m odeling: In
praise o f simple m ethods. Long Range Planning, 45, 341-358.
38 Nhung v£n d§ nang cao trong PLS-SEM

Rigdon, E. E. (2016). C hoosing PLS path m odeling as analytical m ethod

in E uropean .m anagem ent research:.. A .rea list perspective. Long
Range Planning, 34, 598-605.

Sarstedt, M., Hair, J. F„ Ringle, C. M., Thiele, K. O., & Gudergan, S. P.

(2016). M easurem ent issues w ith PLS an d CB-SEM: W here the bias
lies! Journal o f Business Research, 69, 3998-4010.

x o °

< f

c f

r >
 CHUONG2

MO HINH
NANG CAO

CAC KET QUA HOC TAP

*
1. Hi£u dddc tinh htfu ich cua cac mo hinh thanh phan thur bac
f u r u i '■

2. 03nh gia dung die loai hinh khac nhau cua HCMs va hieu
cSchsd dung chung trong PLS-SEM.
3. Hieu each chay phan tich HCM bang phan mem SmartPLS
va each dien g iii ket qua. *
4. Hieu cac khai n'bm coibSn ve phan tich phi tuyen trong ngtir

5 .Danh
5. Danh gia dung each sd dung phan mem SmartPLS de tide
lddng tac dong bac hai.
iui

TONG QUAN CHL/ONG

Vdi sd phde tap ngay cang tang cua cac ly thuyet va cac m o hinh nhan
qua trong khoa hoc xa hoi, cac nha nghien cdu d a sd dung ngay cang
nhieu cac m o hinh th an h phan th d bac trong m o h in h p h d d n g trin h cau
tru e binh phddng toi thieu rieng phan cua ho (PLS-SEM) (vi du: Ringle,
Sarstedt, & Straub, 2012). Cac m o hinh thanh phan th d bac de cap den
m ot khai niem dddc do lddng ct nhieu m dc do khai quat trong mot m o
h in h dddng dan PLS. C hinh xac hdn, cac m o h in h thanh phan thd bac
dai dien cho m ot khai niem tong quat hdn, dude d o d m dc do khai quat
cao hdn, dong thdi bao gom mot so thanh phan con, ma thanh phan con
nay bao gom nhieu dac diem cu the hon cua bien khai niem dddc bieu

39
40 Nhung v ln de n3ng cao trong PLS-SEM

dien bbi khai niem nghien ciiu nay. Md h in h th a n h phan thii bac cho
phep giam so lddng m oi quan he m o hinh cau tru e, lam cho m o hinh
diidng dan PLS trd nen tinh gon hon, dong th d i tang noi dung thong
tin diipc bao phu bdi cac khai niem nhat dinh (vi du: Jonson, Rosen, &
Chang, 2011). Trong chiidng nay, chung toi m d ta ban chat cua cac m d
h in h th anh phan th d b a c va thao luan each phat trien va danh gia chung
trong ngtf canh PLS-SEM.

Ngoai ra, chiidng nay de cap den m ot van de m d hinh hoa nang cao
khac co lien quan den cac m oi quan he phi tuyen. Cac khai niem tieu
chuan cu n g cd cac m oi quan he n h an qua trong cac m o hinh dUo'ng d in
PLS ngu y rang cac khai niem anh hiidng den m pt khai niem khac theo
moi quan he tuyen tinh. Tuy nhien, trong m ot sd triidng hdp, gia dinh
cho cac moi quan he do co the la phi tuyen. Khi m oi quan he gida hai
khai niem nghien cdu la phi tuyen, m dc do tac d p n g gida hai khai niem
khong chi p hu thuoc vao m dc do thay doi tro n g khai niem ngoai sinh
m a con can cd tren gia tri cua no. Khi phan tich cac tac dong phi tuyen
(Wold, 1982), cac nha nghien cdu phai dda ra gia dinh ve ban chat cua
tac dong. M ac du co the co nhieu loai tac dong khac nhau, n h d n g tac
dpng dang bac hai pho bien nhat. C hddng nay de cap den loai tac dong
nay trong PLS-SEM va m o ta viec thde hien va dan h gia cua n o bang
each sd d u n g Sm artPLS.^

CAC MO H IN H THANH PHAN TH l) BAC

Thuat ngtf va Dong ldc thuc day

Cuon sach cd ban ve PLS-SEM (Hair, Hult, Ringle, & Sarstedt, 2017) giai
quyet cac khai niem bac nhat, xem xet khai niem didbi m ot ldp khai quat
duy nhat. Tuy nhien, trong mot sd trd d n g hdp, cac khai niem m a cac
nha nghien cdu m uon kiem tra kha phde tap va cu n g cd the dddc van
hanh 6 m dc khai quat cao hdn. Thiet lap cac m o h in h bac cao hoac m o
hinh thanh phan thd bac (HCMs), vi chung th d d n g dddc de cap trong
ngd canh cua PLS-SEM (Lohmoller, 1989), hau h e t lien quan den viec
kiem dinh cac khai niem bac hai chda hai ldp k hai niem . Vi du, sd hai
long cd the ddpc do d hai m dc khai quat. Mot H C M tiep theo do lddng
sd hai long se bao gom m ot khai niem hai long ch u n g cung vdi m ot sd
khai niem con nam bat cac th u o c tinh cu the hdn v e sd hai long, chang
han n h d sd hai long vdi gia ca, sd hai long vdi chat ld d n g dich vu, sd hai
long vdi nhan vien va sd hai long vdi khong gian d ich vu. N hdng thuoc
 Chiwng 2: Mo hinh nang cao 41

tinh bac thap h d n cu the nay co the tao thanh khai niem sd hai long trd u
tddng hdn, n h d trong T d lieu 2.1.

Tu1lieu 2.1 . th t f b $ c
C ac khai niem Khai niem
bpc thap - b$c thu 1nh§t b§c ca o - bac th u hai

So* h a i lo n g v&i gia

Sqr h ai I6ng voi ch a t

hronig d ic h vu

S ir h<ii long voi nhan

S y h ai lo n g voi Khong
gifln d jc h vu

Thay vi m o hinh hoa cac thuoc tinh cua sd hai long n h d cac dong
cd cua sd hai long tong the va cac khai niem m uc tieu khac (vi du: long
trung thanh cua khach h&ng) trong m o hinh dddng dan PLS phdc tap
hdn la m o h inh dda tren m pt ldp khai niem ddn le, m o h inh bac cao lien
quan den viec lap sd do dong thdi cac khai ni&m bac thap hdn (LOCs)
va mot khai niem bac cao (HOC). Ve m at ly thuyet, qua trin h nay co
the dddc m d rong tdi bat ky s6 lddng nao cho ra ldp m o hinh thii ba, th d
td hoac thii nam , nhdng cac nha nghien ciiu thddng han che m o hinh
cua hQ thanh hai lop (tile la cac m o hinh bac hai). T d lieu 2.2 cho thay
mQt vi du thii hai ve m ot khai niem thii bac lien quan den d an h tieng
cua cong ty. Khac vdi vi du trddc, khai niem bac thap LOCs d o yeu thich
va nang ldc kha nang the hien cu the hdn ve khai niem bac cao H O C
danh tieng cua cong ty. Do la, m oi q u an he td H O C den LOCs.

Co n h ieu ly do de dda H CM s vao PLS-SEM (vi du: Edw ards, 2001;

Johnson va cong sd, 2011; Polites, Roberts, & Thatcher, 2012). M ot ly
do lien quan den sd trao doi trung thdc thong tin [bandw idth-fidelity
tradeoff), hoac y tddng cho rang cac khai niem dddc m d rong h d n la cac
yeu to dd bao tot hdn ve cac tieu chi trai dai tren nhieu m ien khai niem
42 Nhdng van de nang cao trong PLS-SEM

va theo cac khoang thdi gian. T bc la, neu m uc dich la de dd bao cac
h a n h vi ddpc xac dinh m ot each rong rai,. th i cac_khai niem da hddng
bac cao co the chiifng m inh co gia tri. Mot ly do khac la de vdot qua sd
nguy bien ldn, xay ra khi m ot hien tddng duy n hat dbpc danh gia rieng
biet dddi vo boc cua hai hoac nh ieu bien vdi cac ten goi khac nhau.

T u q u a n diem thuc te, H C M s cho phep cac n h a nghien cuu giam sd

luong cac moi quan he trong m d hinh cau true, lam cho m d hinh dubng
dan PLS trd nen kiem dung va de hieu hon. T u ligu 2.3 m inh hoa khia
canh nay. N hu co the thay trong md hinh phuc tap, co chin moi quan
he md hinh cau true lien ket cac bien tiem an ngoai sinh { Y , Y2 va T3)
vdi cac bien tiem an noi sinh (k4, Y' va Y6). Bang each the hien vao m ot
H O C, sd lu o n g he sd dudng dan co the d u g c giam xuong con sau, tao
ra mot m d hinh dd phan tich hon xet ve sd luong cac moi quan he md
hinh cau true. Trong tnrdng hop nay, HOC du g c g ia dinh la lam trung
gian hoan toan cac tac dong cua tung LOC len cac b ien tiem an noi sinh
(de biet them chi tiet ve trung gian, xem H air va cong su, 2017). Sir
phuc tap giam di cua m d hinh nay la do su tieu ton cua kha nang du bao
doi vdi cac bien tiem an noi sinh m a HOC tien d o an (arc la YA, Ys va Y6
trong T u lieu 2.3). Ly do khac vdi md hinh tac dong true tiep trong dd
tit ca cac bien tiem an ngoai sinh d u bao mot bien tiem an noi sinh (Tu
lieu 2.3. bang ben trai), trong HCM, chi mot minh H O C d u bao cac bien
tiem an noi sinh (T u lieu 2.3, bang ben phai).
 Chirang 2: Mo hinh nang cao 43

,..T
M6 hinh thanh phan thtf bac HCMs va m6 hinh
Tu" lieu 2.3
phdc tap
M o hinh phuc tap M o hinh thanh phan th ii bac HCM s

Cuoi cimg, cac HCM cung chdng m inh co gia tri neu cac bien quan
sat nguyen nhan trong m o hinh do lddng cua bien tiem an the hien m dc
do cong tuyen cao. Day la m o t van de quan trpng bdi vi nd co tac dong
den cac ddc ldpng ve trpng so va y nghia thong ke cua chung. C hinh
xac h«n, do cong tuyen cao co the dan den trong so sai lech va dau cua
chung co the bi dao ngdpc. H dn nda, sd cong tuyen lam tang cac sai
so chuan va do do lam giam kha nang chdng m in h ran g cac trpng so
d dc lddng la khac biet dang ke so vdi 0 (H air va cong sd, 2017). Cac
HCM cho phep xd ly cac van de ve cong tuyen bang each cung cap m o t
phddng tien de tai sap xep cac m o hin h do lddng. Vdi ly thuyet do lddng
ho trd bddc nay, cac nha nghien cdu co the chia nhd tap hop cac bien
quan sat va thiet lap cac khai niem rieng biet theo cau true bac cao hdn.
Vi du nh d , xem xet m ot khai niem do lddng nguyen nhan vdi bon bien
quan sat (x,-x4), trong dd x, va x2 cung n h d x3 va x4 co tildng quan cao.
Neu khai niem co y nghia, cac nha nghien cdu co the ph&n chia khai
niem nguyen nhan th an h hai khai niem con, moi khai niem dddc do
b& ngcacbien quan sat p h i cong tuyen (vi du: x va x3len m o t khai niem
va x , va x. len khai niem kia).

Nhdng can nhac ve mat khai niem

Viec thiet lap cac HCM yeu cau cac nha nghien cdu phat trien va sd
dung dinh nghia hoat hoa thich hop cua bien so khai niem dang ddpc
44 Nhung v£n dS nang cao trong PLS-SEM

xem xet. D inh nghia hoat hoa cho phep khai niem hoa m ot y tddng
khai quat de no dai dien cho linh vdc cua cac pham vi co the do lifdng.
quan sat co the ddpc nghien cdu (xem G hdO ngl). D inh nghla hoat hoa
h d d n g dan viec xac din h cac LOC co lien qu an, tro n g dd m oi th an h
phan de cap den m ot phan td dac biet (hoac th an h phan) dddc lien ket
vdi H O C va m oi th an h phan cd m ot bo bien qu an sat cd the dddc xac
dinh bdi tin h khac biet cua phan td m o ta LOC. D ong thdi, sd khac biet
dac trd n g nay cu n g phai dd lien q u an de chi n h d n g LOC nao q u an trong
doi vdi nghien cd u cu the m di dd d c dda vao m ot HCM . D inh nghia
hoat hoa, vdi cac th an h phan dac trd n g cua no, cd the khac nhau td
nghien cdu nay den nghien cdu khac vi m ot khai niem ly thuyet khong
phai la ban than no ddpc xac d in h la da h ddng hoac don hddng. Thay
vao do, m ot khai niem cd the d d d c xac dinh theo m ot trong hai each,
dai dien cho cac m dc khai quat khac nhau ve ly thuyet (Bollen, 2011).
Ve m at khai niem , cac HCM cd the ddpc thiet lap theo phdo n g
phap tiep can td dddi len (quy nap) hoac td tren xuong (suy dien) (vi
du: Johnson va cong sd, 2011). Trong phddng phap tiep can td dddi
len, m ot so bien tiem an ddpc ket h o p th an h m ot khai niem don le, trd u
td p n g hon. Ngdpc lai, trong phddng phap tiep can td tren xuong, m ot
khai niem trd u tdpng h o n ddpc din h nghia de bao gom m ot so thanh
phan, n h d trd d n g hop trong vi du ve sd hai long cua khach hang ddpc
mo ta d tren (T d lieu 2.1). Mac du th d d n g xuyen ddpc sd dung trong
nghien cdu thde nghiem trong m ot no Idc de giam do phde tap cua mo
hinh, chung toi khong khuyen ban chi tom tat thong tin trong m ot khai
niem trd u td p n g hon. Viec thiet lap m ot H C M trong PLS-SEM luon lien
qu an den viec m at thong tin do cac tac dong gian tiep gida cac H O C va
(cac) khai niem tieu chi thong qua H O C m di ddpc thiet lap thay the
cac tac dong trdc tiep gida cac H O C va (cac) khai niem tieu chi nay.
Tuy nhien, PLS-SEM cho phep ddc ldpng tat ca cac m oi quan he trong
m ot m ang lddi danh nghia rong h o n cua khai niem nghien cdu (nghia
la khong cd H O C ) m a khong lam mat thong tin. D o dd, khi ly thuyet
cau true ho trp sd bao gom m ot so ldpng ldn cac khai niem nghien cdu,
m at khac, cd the ddpc tom tat trong m ot H O C , viec chdng m inh sd xem
xet chung cua chung dd d i dang m ot HCM tren co sd phan tich m ot mo
hinh la khong du. Thay vao dd, cac HCM bat nguon td tren xuong cung
cap cho cac nha nghien cdu n h d n g hieu biet bo sung ve tac d ong cua cac
thanh phan khac nhau ddpc gan vao tro n g m ot khai niem nghien cdu
cu the. O day, y d jn h cua n h a nghien cdu la cu the hoa tac d ong cua cac
 Chucmg 2: Mo hinh nang cao 45

thanh phan nhii vay len cac khai niem nghien cbu khac trong m o hinh
th o n g qua H O C .
Ngoai viec sb dung ly thuyet de xac dinh tieu chi thu n h a n de chon
cac LOC ph u hpp, ban chat cua cac m oi quan he giba cac LOC va H O C
p h ai diipc lam ro. Mot H O C la m ot khai niem chung dbpc bieu dien
(trong mo h in h ket qua) hoac dbpc cau thanh (trong mo hinh nguyen
nhan) bbi cac thanh phan cu the cua no (tuc la cac LOC). N eu H O C
la ket qua, khai niem chung dbpc the hien trong mot so LOC cu the
hon. Loai m o h inh nay cung dbpc goi la m o hinh lan truyen - spread
m odel (Lohmoller, 1989). Neu H O C la nguyen nhan, no la sb ket hpp
cua m ot so LOC cu the dai dien cho cac thanh phan cu the hdn tao
th anh khai niem chung (Becker, Klein, & Wetzels, 2012; Edwards, 2001;
Wetzels, Odekerken-Schroder, & van O ppen, 2009). Loai m o hinh nay
cung dbpc goi la m o hinh thu thap - collect model. HCM trong T b
lieu 2.1 co m oi quan he hin h thanh tb cac LOC den H O C , dai dien cho
m oi dong gop tbdng doi cua LOC de hinh th an h nen HOC. Tuy nhien,
n eu hoat dong khac nhau, cac m oi quan he nay cung co the dbpc m o
hinh hoa theo hbbng ngbpc lai vbi cac LOC phan anh H O C. Mot dac
diem ky thuat dang nguyen nhan la phu hpp khi dinh nghla hoat dong
cua bien khai niem gpi y rang m ot sb thay doi trong gia trj cua LOC, vi
du, do thay doi trong danh gia cua ngbdi tra lbi ve dac diem dbpc ndm
bat bbi LOC thay doi gia trj cua HOC. T b d n g tb n h b cac bien quan sat
trong cac m o h inh do lbbng nguyen nhan, LOC khong can phai, nhbng
cb the, tbdng quan khi chung khong dai dien cho cac bieu hien cu the
cua HOC.

Thay vao do, dac diem cua mo hinh ket qua la thich hdp khi co mot
khai niem khai quat, tong quat h d n ma khai niem nay giai thich cac
moi tbdng quan giba cac LOC nhb trong T b lieu 2.2. Do db, can cb moi
tbdng quan dang ke giba cac LOC - tb d n g tb vbi cac m o hinh do lbbng
ket qua — dbpc gia djnh la do H O C tao ra. Nghia la, H O C la nguyen
nhan gia djnh giai thich cho m oi tb d n g quan giba cac LOC.

Ngoai mbi quan he giba H O C va cac LOC, cac HCM cb the dupe
m b ta tren cd sb dac diem ky thuat cua cac mb hinh do lb b n g cua cac
LOC (tbc la ket qua hoac nguyen nhan). Ket qua la, cb bon dang chinh
cua cac HCM dbpc phat trien (Tb lieu 2.4) dbpc thao luan tro n g cac tai
lieu hien nay (Jarvis, M ackenzie, & Podsakoff, 2003; Wetzels va cong sb,
2009). HCM ket qua - ket qua (D ang 1; T b lieu 2.4) cho biet mbi quan
46 NhGng van <Jd nSng cao trong PLS-SEM

he ph an anh ket qua gifla H O C va cac LOC, sau do tat ca deu dddc do
lddng theo ket qua. d dang HCM nay, cac L O C co m oi tdcfng quan cao
va H O C dai dien cho nguyen nhan gia trong viec giai thich nhflng moi
td d n g qu an nay. Lohm oller (1989) goi dang m d hinh H C M nay la m o
hinh nhan to chung thii bac - (hierarchical com m on fa c to r m odel),
trong dd H O C chung dai dien cho nhan to chung cua m ot so nhan td
cu the (tflc la cac LOC). Nguon: Dieu ch in h tfl H in h Bl trong Ringle,
Sarstedt, & Straub (2012).

■C ac dqtng md hinh tti^ h pihair ^

Nguon: Dieu chinh tic Hinh B l trang Ringle, Sarstedt, & Straub (2012)

Do dd, dang m d hinh HCM nay la phu hop n h a t neu m uc tieu cua
nghien cflu la thiet lap cac nhan td chung cua m o t sd LOC co lien quan
n h d n g khac biet ro rang (Becker va cong sd, 2012). Viec sfl dung cac
HCM ket qua - ket qua da tflng gay ra nhflng tra n h luan dang ke, vdi
 Chirong 2: Mo hinh nang cao 47

cac nha phe binh cho rang cac m o hinh nhii vay khong ton tai (hoac vo
nghia) vi cac do liidng ket qua nen mang tinh dcfn hddng va nen diiqc
hoan doi lan nhau ve m at khai niem , dieu nay m au thuan vdi quan diem
p h an biet da h d d ng ve m at ban chat (Lee & C adogan, 2013).

Nhflng ngddi de xddng phan doi cho rang quan diem chi trich nay
la thieu sot ve m at khai niem , ket luan rang cac n h a nghien ciiu n en xem
xet kieu m o hinh nay nhii la m ot Ida chon hoat dong chinh thong cho
cac khai niem nghien ciiu d a h d d n g (Temme & D iam antopoulos, 2016).
Dac biet, cac nha nghien ciiu co the sii dung cac HCM Ket qua - Ket
qua de do m ot khai niem nhieu lan tai cac th d i diem khac nhau. Trong
m ot thiet lap nhii vay, cac LOC dai dien cho cac phep do khac nhau cua
mot khai niem tai cac thdi diem khac nhau theo thdi gian (tiic la, cac
chuoi trin h td khac nhau cua m ot chuoi kiem dinh), m a H O C giai thich
dong thdi. Neu chuoi kiem dinh (tdc la cac LO C ) do liidng cung m ot
khai niem tai cac thdi diem khac nhau, chung co mdi tiicfng quan cao.
Lohm oller (1989) mo ta sd sap dat cua cac thanh phan nay nhii mot mo
hinh da chuoi (m ultiple battery m odel) va trin h bay m ot vi du m inh
hoa ve nang ldc chung cua hoc sinh. M di LOC dai dien cho m ot chuoi
kiem tra cac chi so bang ldi noi, sd va khong gian diiqc do tai cac thdi
diem khac nhau (vi du: cac bai kiem tra d dau, gifla va cuoi nam hoc)
cho cung m ot ldp (tdc la cung m ot ca nhan). Tuy nhien, m o h in h da
chuoi diid n g n h ii chiia ddqc ap d u n g trong cac nghien cdu PLS-SEM
cho den nay.

Trong mo hinh HCM ket qua - nguyen nhan (Dang II; Td lieu
2.4), H O C dai dien cho m ot khai niem nghien cdu tdng quat hon cua
cac LOC d d q c do lddng theo ket qua. Cac LOC cu the khong nhat thiet
phai chia se m ot nguyen nhan pho bien ma la h in h th an h H O C chung.
Barroso va Picon (2012) d d a ra m ot vi du ve m ot H C M ket qua - nguyen
n h an trong phan tich da hddng cua ho ve chi phi chuyen doi cam nhan
(xem C hd o n g 1). H o xac dinh m ot bo gom sau chieu h d d n g / khia canh
(chi phi ton that lqi ich, chi phi ton that ca nhan, chi phi rui ro kinh te,
chi phi danh gia, chi phi thiet lap va chi phi ton that tien te) dai dien cho
cac LOC cua chi phi chuyen doi cam nhan H O C tdng quat hon. “Mot
sd thay doi trong mot chieu hddng nao do khong nhat thiet ngu y m ot
sd sda doi tro n g m ot chieu hddng khac. Noi each khac, chung k hong
nhat thiet phai dong thay doi; thay vao do, mdi chieu h d d n g co the khac
nhau m ot each doc lap vdi nhflng h d d n g khac” (B arroso & Picon 2012,
tran g 532). Vi ly do nay, khong giong n h d m ot so nghien cdu trddc day,
48 Nhung v in de nang cao trong PLS-SEM

Barroso va Picon (2012) de xuat rang chi phi chuyen doi cam nhan la
m ot khai-niem Ibn-hctndbd& the hien-nhb m bt-thanh ph an cua cac LOC
khac nh au cua no (xem Becker va c<mg sb [2012] - cac vi du ve H CM
ket qua - nguyen nhan). Trong m ot m ang lbdfi cac m oi q u an he co the
lien q u an ld n h d n , H O C hoan toan tru n g gian cho cac m oi qu an he cua
LOC vdi cac bien tiem an noi sin h trong m o h in h d b b n g d a n PLS. Vi du,
trong HCM b T b lieu 2.3, H O C tru n g gian m ot each ho an toan cho m oi
q u an he giba LOC, va ba bien nghien cbu noi sin h Y4, Ys va Y6.
M ot dang khac la HCM nguyen nhan - ket qua (Dang IIJrTd lieu
2.4). HCM nguyen nhan - ket qua bao gom m ot H O C tong quat hdn
giai thich cac LOC dbdc do lbbng theo nguyen nhan. M uc tieu cua dan g
HCM nay la trich xuat m ot phan chung cua m ot so LOC d b d c dd lbbng
thed nguyen nhan da dbdc thiet lap de the hien cu n g m ot noi dung theo
ly thuyet. Tuy nhien, moi LOC xay d bng tren m gt tap h d p cac bien quan
sat khac nhau. Bang each sb dung m ot so LO C d b d c do theo nguyen
n h an , cac nha nghien cbu co the khac phuc van de rang m ot khai niem
doc lap dbdc do vbi cac bien qu an sat nguyen n h an kho co the bao phu
m ien khai niem m ot each day du. Viec sb dung LO C tbdng tb n h b n g
dbdc ph an biet do lbbng theo nguyen n h an khac nh au dai dien cho
H O C cu n g cap do bao p h u rQng h d n cua m ien khai niem nghien cbu
(Becker va cong sb, 2012). Mot vi du d ien h in h cu a dang HCM nay la
hieu qua tong the cua doanh nghiep, tro n g do m ot sb LOC db d c do
lbdng th e o nguyen n h a n the h ien cac dac d iem lien q u an den h ieu qua.
H O C dai dien cho m ot phan chung cua cac LOC (tbc la, hieu qua tong
th e cua cong ty trong HCM; Jarvis va cong sb, 2003; Petter, Straub, &
Rai, 2007). Ngoai ra, HCM nguyen nhan - ket q u a co the phuc vu nhb
m ot m b h in h da chuoi khi dbdc giai thich trong bbi canh HCM ket qua
- ket qua. Trong trbbng hdp do, cac LOC dai dien cho cung m ot khai
niem nghien cbu da dbdc do lbbng theo nguyen nh an bang cac bien
quan sat giong nhau va cho cung cac quan sat tai cac thbi diem khac
nhau. Trong dang m b hinh da chuoi nay, cac m bi quan he tb H O C tbi
cac LOC se co do lbn tbdng tb vi chung dai d ien cho m ot khai niem
dbdc dd tau cac thbi diem khac nhau.

Cuoi cung, HCM nguyen nhan - nguyen n h a n (Dang IV; T b lieu

2.4) xac d in h sb dbng gop tbdng doi cua cac L O C db d c dd lbbng then
nguyen n h a n chd HOC khai q uat hdn. D ang H C M nay chbng to hbu
ich de cau tru e m ot khai niem nghien cbu n g u y en nh an phbc tap vbi
 Chirang 2: Mo hinh nang cao 49

nhieu bien quan sat thanh m ot vai khai niem con, trong triidng hdp cac
nha nghien ciiu da diia ra m ot so khia canh cu the theo m ot khai niem
tong quat hdn. Mot lan nPa, hieu qua cong ty se dai dien cho m ot khai
niem ma xet ve ban chat no diipc do bang each sit dung dang H C M nay.
Trong khi dang HCM nguyen nh an - ket qua se bao gom cac bien quan
sat khac n h au ve hieu qua tong the nhii cac LOC, thi dang HCM nguyen
n h an - nguyen nhan bao gom cac LOC dai dien cho cac khia canh khac
nhau ve hieu qua nhii hieu qua cua cac hoat dong hoac phong ban khac
n h au (vi du: hieu qua ve R & D, hieu qua ve nhan sp, hieu qua ve ban
hang) xac d in h hieu qua tong the cua doanh nghiep (tPc la H O C ; Jarvis
va cong sp, 2003; Petter va cong sp, 2007) nhiing dieu do khdng nhat
thiet phai tiiong quan. Mot vi du khac cua m ot HCM nguyen n h an -
nguyen nhan la hon hop tiep thi (m arketing mix) diipc xac d in h befi bon
P (tPc la, g ia ca /price, san pham /product, chieu thi/prom otion va kenh
phan phoi/place).

Mac du m ot so dang m o hinh bien tiem an thP bac it co kha nang

diipc sP dung, tat ca cac dang nay co the diipc tlm thay trong nghien cPu
thpc nghiem . Ringle va cong sp (2012) bao cao rang trong so 109 nghien
cPu PLS-SEM trong tap chi hang dau ve he thong thong tin quan ly - M IS
Quarterly, 15 nghien cPu (23%) da bao cao 25 m o hinh HCM . Trong so
25 mo hinh nay, m o hinh HCM dang II ket qua - nguyen n h a n (13 m o
hinh; 52%) va HCM dang IV nguyen nhan - nguyen nhan (6 m o hinh,
24%) dope sP dung thodng xuyen nhat trong cac Png dung thpc nghiem
nay. Do do, cac nha nghien cPu dOPng nhii chu yeu ap dung H CM khi
mo hinh hoa cac moi quan he nguyen nhan giPa cac LOC va H O C . H on
nPa, trong h au het cac nghien cPu (13 nghien cPu, 87%) H O C diipc gan
vao m ot m ang liicri cac m oi quan he giPa cac bien tiem an phuc vu nhP
khai niem nguyen nhan (5 nghien cPu; 33%), ket qua (3 nghien cPu;
20%), hoac ca hai (5 nghien cPu; 33 %). Do do, viec thao luan khong
nen gidi han 6 cac HCM nhP cac khai niem nghien cPu rieng biet, con
dope goi la cac HCM dPng doc lap (stand-alone HCMs), nhOng cung
nen xem xet Png dung tiem nang cua chung trong m ang lOPi cac bien
tiem an dope gan vao trong mo hinh cau true (Becker va cong sp, 2012).
Hon nPa, cac vi du cua cac HCM d o p e thao luan trong phan nay chi
m inh hoa cho cac khai niem toong tp, khi dope dinh nghia khac nhau
trong cac m ang danh nghia ly thuyet khac nhau, co the can phai dope
mo hinh hoa tren mot dang HCM khac vdi m o ta 6 day.
50 Nhflng van de nang cao trong PLS-SEM

Nhflng can nhac ve mat ky thuat

Tong quan

PLS-SEM cho phep cac dac diem ky thuat va d o t ld d n g tat ca cac

dang HCM , n h d trinh bay trong T d lieu 2.4. Moi dang H O C cu the yeu
cau xem xet can than khi xac djnh va ddc lddng m o hinh, tuy nhien, vi
PLS-SEM yeu cau mdi khai niem nghien cdu trong m o hinh dddng dan
PLS cd it nhat m ot bien quan sat trong m o h inh do ld d n g cua no. Sd can
thiet nay khong chi danh cho cac LOC ma con cho H O C , do la m ot dai
dien khai quat cua cac LOC. N h d vay, ban chat cua m ot H O C khac vdi
ban chat SEM dda tren hiep phddng sai (CB-SEM), trong do H O C sd
dung cac bien tiem an khac de do ld d n g va khong co bien quan sat nao
trong m o h in h do lddng cua no. Vi ly do nay, cac H O C doi khi dddc goi
la cac bien khong cd thde (phantom variables) trong CB-SEM.

De xd ly van de do lddng cua cac HCM trong PLS-SEM, cac nha

nghien cdu cd the d d a tren hai p h d d n g p h ap tiep can chinh cho dac
diem ky thuat cua ho trong m o hinh dddng dan PLS (Becker va cong
sd, 2012; Ringle va co ngsd, 2012; W etzels va cong sd, 2009): (1) phdong
phap tiep can cac bien quan sat lap di lap lai va (2) each tiep can hai giai
doan. Sau khi gidi thieu hai each tiep can nay, chung toi thao luan m ot
so can nhac ky thuat bo sung va cac de xuat khac.

Phu'amg phap tiep can cac bien quan sat tai lap
Trong phd d ng phap tiep can bien quan sat lap di lap lai (tai lap), tat
ca cac bien quan sat cua cac LOC dd d c gan cho m o hinh do lddng cua
cac H O C (Lohmoller, 1989; Wold, 1982). Trong cac vi du m o hinh bac
2 trong T d lieu 2.4, p h ddng phap tiep can bien quan sat lap lai sd dung
cac bien quan sat x x den *9cua cac LOC de thiet lap m o hinh do lddng
cua HOC n h d trong Td lieu 2.5. D o do, cac bien quan sat dddc sd dung
hai lan: m ot lan cho cac LOC (he so tai / trong so sd cap) va m ot lan nfla
cho H O C (he so tai / trong so th d cap). Cach tiep can nay co the de dang
dddc m d rang den cac HCM bac th d ba hoac thd td (vi du: W etzels va
congsd, 2009), neu no co y nghla ve m at ly thuyet v a khai niem . Td lieu
2.5 cho thay each thiet lap bon dang HCM bang each sd dung p h d d n g
phap tiep can bien quan sat tai lap.

N hd dd d c m o ta trong C h d d n g 1, PLS-SEM d d a tren hai che do

(M ode) de ddc ldong do lddng cac khai niem nghien cdu, lien quan den
each p h d d n g phap ddc lddng trong so cua cac bien q u an sat, bieu thi sd
 Chu'o’ng 2: Mo hinh nang cao 51

Cac dang HCM Phtf0l»gPhap

TU lieu 2.5
quan s£t ' _________ ‘ v::’ ■
'
D ang 1: K et quit - K et qua

D ang III: Nguyen nhan - Ket qua ian - Nguyen nhan

Nguon: Dieu chinh tu Hinh B1 trong Ringle, Sarstedt, & Straub (2012).

dong gbp cua moi bien quan sat doi vbi bien tong hpp. M ode A tbong
bng vbi trong so tbong quan bat nguon tif m oi tb d n g quan hai bien so
giba moi bien quan sat va khai niem nghien cbu. M ode B tbong bng
vbi trong so hoi quy, tieu chuan trong phan tich hoi quy binh phbong
toi thieu thong thbbng. Trong so hoi quy dba vao tinh toan khong chi
la moi tbdng quan giba m*i bien quan sat va khai niem nghien cbu ma
con la m oi tbong quan giba cac bien quan sat. M ode A thbbng dbpc ket
hpp vbi bbc lbpng cac khai niem do lbbng theo ket qua, trong khi Mode
B dbpc sb dung de bbc lbpng cac khai niem do lbbng theo nguyen nhan.
T b o n g bng, cac HCM ket qua - ket qua va HCM ket qua - nguyen nhan
db p c chi d jnh vbi cac bien quan sat lap lai thong th b b n g d ba tren M ode
52 Nhung van de nang cao trong PLS-SEM

A, trong khi cac H C M nguyen n h an - ket qua va HCM nguyen nhan -

n guyen nhan thiffmg xfl tiling fldc lflpng M ode B.

lh a y vi tuan theo dng dung nhflng che do fldc lflpng nay, cac n h a
nghien cflu cung co th e sfl dung M ode A de fldc lflpng cac khai niem
do lfldng theo nguyen n h an va M ode B de fldc lflpng cac khai niem do
lfldng theo ket qua (xem C hflong 1). Lohm oller (1989) ung ho quan
diem rong ho n ve viec thiet lap va fldc lflpng cac H CM bang sfl phan
biet 16 each khac nhau de xac dinh va fldc lflpng cac H C M vfli ci.c khai
niem do lfldng theo ket qua, tuy thuoc vao floe lflpng m o hin h do lfldng
cac LOC va H O C (M ode A hoac Mode B), m oi quan he gifla cac LOC
va H O C (m o h in h tap h p p hoac mo h in h phan tan) va sP do trp n g so
m o h in h ben tro n g (so do trp n g sd dfldng dan vd i trp n g sd n h a n td).
Tac gia bao cao hai n h o m sap xep cho thay sfl h o i tu sdm (p rem a tu re
convergence), co nghia la thuat toan PLS-SEM hoi tu nhflng ket qua phu
thuoc vao gia tri bat dau va khong toi flu. Cac n h o m sap xep nay la (1)
l/d c lflpng M ode A cho cac LOC va H O C, cac m oi quan he di tfl cac
LOC den H O C , va sfl d u n g so dd trpng s6 dfldng d an; va (2) Lfdc lflpng
M ode A cho cac LOC, fldc lflpng M ode B cho H O C , cac moi quan he
tfl H O C den cac LOC, va sfl dung so dd trpng sd dfldng dan. Becker va
cong sfl (2012) chflng m inh cac ket qua nay dfla tren ng h ien cflu m d
phong phflc tap va cho thay rang so dd trpng sd d fld n g dan tao ra cac
th am sd sai chech Idn n h a t khi sfl dung fldc lflpng m d h in h chuan (tflc
la M ode A cho cac khai niem do lfldng theo ket qua) trong m ot HCM
ket qua - nguyen nhan. 1 rong cac thiet lap nay, cac so dd trpng sd nhan
td va trpng so trpng tam - nhflng so dd gan day m a Lohm oller khong
xem xet trong viec danh gia cac dang HCM - thflc hien tot hon nhieu ve
do chinh xac cua thong sd. Tuy nhien, nghien cflu tflong lai thi can thiet
de cung cap nhflng hieu biet chflng m inh sfl lfla chon thich h p p cua
so do trpng sd ben trong (trpng tam, nhan td, h o ac dfldng dan) khi sfl
dung cac HCM trong PLS-SEM. Trong boi canh nay, chung toi de nghi
sfl dung so dd trpng sd nh an td nhfl la m ot giai phap thda m an gifla cac
phflong an trpng tam va dfldng dan khi m d h in h dfldng dan PLS b ao
gom m p t hoac n hieu HCM .

Tflong tfl nhfl vay, sfl hieu biet ro rang bi bo so t khi di chech khoi
thflc tiin tieu chuan cua viec sfl dung fldc lflpng M ode A cho cac khai
niem dflpc chi dinh theo ket qua va Mode B cho cac khai niem dflpc chi
dinh theo nguyen n h an trong m ot HCM. Trong k h i Becker va cong sfl
(2012) khuyen nghi sfl dung M ode B de fldc lflpng H O C dflpc chi dinh
 Chirang 2: Mo hinh n§ng cao 53

trong m ot HCM dang ket qua - nguyen nhan, van chda ro lieu de xuat
nay co khai quat cho cac dang HCM khac hay khong. Do do, chung toi
khuyen ban nen trd lai ve thde hanh tieu chuan cho den khi nghien cdu
tiep theo chi d in h cac tinh huong phu hop cho viec sd dung Mode A va
M ode B.

Cach tiep can cac bien quan sat lap di lap lai dam bao sd chu y ky
thuat bo sung tro n g ba khia canh. Dau tien, cac moi quan he gida H O C
va cac LOC bi sai lech dang ke neu cac m o hinh do lddng cda cac LOC
thay doi m an h ve so ld d n g bien quan sat (Becker va cong sd, 2012).
Hay xem xet, vi du, m ot HCM vdi hai LOC, m ot vdi tarn bien quan
sat (LOC,) va m ot vdi hai bien quan sat (LOC2). Khi sd dung phddng
phap bien quan sat lap lai, tat ca 10 bien quan sat ddqc gan dong thdi
cho HOC. Vi chung chia se mot sd lddng ldn cac bien quan sat d HCM ,
mdi quan he m anh me hon gida H O C va LOC, dddc ky vong. H dn
nfla, nang ldc d d bao cua cac LOC la m ot phan chdc nang cua do tin
cay, phu thuoc vao sd lddng bien quan sat (D iam antopoulos, Sarstedt,
Fuchs, Kaiser, & W ilczynski, 2012). C ang n hieu bien quan sat dan den
do tin cay cang cao h d n va do dd, nang ldc d d bao m anh h d n (Peter &
Churchill, 1986), dac biet khi m b hin h do lddng la mb h in h ket qua. De
tran h bat ky sd sai lech ve m o hinh nao trong cac ddc lddng tham sd, cac
nha nghien cdu phai dam bao rang sd lddng bien quan sat la bang nhau
(hoac it nhat la cd the so sanh) gida cac LOC. Neu dieu nay la khong kha
thi td goc do ly thuyet do lddng, cac nha nghien cdu nen danh gia mot
each co he thong, hieu qua cua viec loai bo bien quan sat tren cac mdi
quan he gida H O C va cac LOC. Dieu nay dac biet ap dung khi cac LOC
dddc do lddng theo nguyen nhan vi cac m b hinh do lddng nguyen nhan
thddng dan den sd khac biet ve do phde tap cua m b h in h d o lddng, tuy
thubc vao do rong cua m ien khai niem . Ldu y rang neu cac bien quan
s&t cua bait Icy LOC nao bj loai bo, nha nghien cdu phai dam bao rang
cac bien quan sat tddng td cdng bi loai bo khoi HOC. N eu khong, sd sai
lech se ddqc d d a vao cac ddc lddng m b hinh.

Thd hai, cac nha nghien cdu nen danh gia khong chi cac m o hinh
do ld d n g cua cac LOC m a con m b h in h do lddng cua H O C . Tuy nhien,
khac vdi cac khai niem nghien cdu khac trong m o hin h d d d n g dan PLS,
viec danh gia H O C khong lien quan den cac m di quan he gida H O C va
cac bien qu an sat cua no m a la m di quan he gida H O C va cac LOC cua
no. Trong khi cac mdi quan he nay ddqc sap xep dddi dang he so dddng
dan trong ph an tich PLS-SEM, td quan diem mb hinh, chung tdong
 54 Nhung van d§ nang cao trong PLS-SEM

ling vdi he so tai (triidng h o p H C M ket qua - ket qua va H C M nguyen

nhan - ket quaJH oac trong so (trong triidng Hpp'HCM ket qua - nguyen
nhan hoac H C M nguyen nh an - nguyen nhan) va can phai diipc suy
d iin n h ii vay.

HOC Ket qua - nguyen nhan (Dang II) HOC Nguyen nhan - nguyen nhan
HCiM nhir la Bien tien an noi sinh
I
HCM nhir la Bien tien in noi sinh

Nguon: Chuyen the tv Hinh B2 trong Ringle, Sarstedt, & Straub (2012).

Trong cac H C M ket qua - ket qua va HCM nguyen nhan - ket
qua, m oi quan he gifla HOC va cac LOC la dau vao cho viec liefe lflpng
p hiiong sai tru n g binh dflpc trich xuat (AVE), do tin cay tong hpp va
do tin cay C ronbach’s Alpha. Gia tri p h a n biet gifla H O C va cac LOC
khong phai la m oi quan tam khi p h a n dfl ve m at khai niem va thflc
n g h iim dflpc m ong dpi, danh gia gia tri phan biet gifla cac yeu to mo
hinh nay la vo nghia. Tuy nhien, gia tri phan biet ca n phai dflpc thiet lap
gifla cac LOC tro n g m ot HCM cu the. Doi vdi loai d a n h gia nay, sfl dung
tieu chi ty le di biet-don nhat (heterotrait-monotrait) cfla cac m oi tflong
quan (HTM T) (Henseler, Ringle, 8c Sarstedt, 2015). Viec kiem tra g ia tr i
phan biet gifla H O C va cac khai niem nghien cflu k h ac trong m o hin h
co the gay tran h cai. Trong triidng h p p ban m uon kiem tra gia tri phan
biet cua H O C , ban can tinh toan tieu chi H TM T th e o each thfl cong
(xem H enseler va cong sfl, 2015) bang each sfl d u n g m oi tflong quan
khai niem nghien cflu gifla cac LOC nhfl m oi tflong quan m onotrait-
heterom ethod va he so tai cheo bien quan sat ben ng o ai H C M vdi cac
 Chu’cmg 2. Mo hinh nang cao 55

LOC nhii la mdi tddng quan heterotrait-heterom ethod. Cuoi cung,

trong cac HCM ket qua - nguyen nhan, hoac nguyen nhan - nguyen
n h an , cac nha nghien ciiu cung can phai danh gia sd cong tuyen cung
n h ii y nghla v a sd lien quan cua cac mdi quan he gida cac LO C va H O C.

Thil ba, viec sd dung cac HCM dang ket qua - nguyen nhan va
nguyen nhan - nguyen nhan gap vain de khi H O C co m o t bien tiem an
tac dong trddc. Nhii diipc m inh hoa trong Tii lieu 2.6, sd sap dat d o doi
hoi rang hau het cac p h d o n g sai cua H O C diipc giai thich bdi cac LOC
cua no, m ang lai gia tri R2 (gan den) 1,0 (R2 1,0). Ket qua la, bat ky he so
diidng dan nao khac (tiic la khac vdi LOC) cho cac m di quan be chi tdi
H O C (tiic la p, trong T 0 lieu 2.6) se rat nhd (co th e la 0) va khong dang
ke (Ringle va cong sd, 2012).

De xd ly cac nhom m o hinh n h d vay, Becker va cdng sd (2012) de

xua’t mo hinh hoa tac dong cua (cac) bien tiem an tac dong khong trd c
tiep len HCM n hdng thong qua cac LOC n h d m inh hoa trong T d lieu
2.7. Tong tac dong cua bien tiem an tien le ( Ts) len H O C bang tac dong
gian tiep cua no thong qua cac LOC (tiic la: p2- p ? + p }- pg + p 4- p5) cong
vdi tac dong trdc tiep cua bien tiem an tien de len H O C (tdc la p,). D d
la, cac nha nghien cdu can phai xac djnh them cac m di quan he gida
khai niem nghien cdu tien de nay va cac LOC va phan tich tdng tac dong
cua no len HOC. Phan tich nay, con ddpc goi la phan tich tac dong tdng
cua cac dang thu thap HCM, nen ddpc ap dung th d d n g xuyen de tran h
sai lam Loai II trong viec ddc ldpng tac dong trdc tiep gida bien tiem an
tien de va H O Q j^ tr o n g T d lieu 2.6).

Phan tich cac tac dpng tong cua dang HCM thu
Tu’ lieu 2.7 th|p bang each suf dung phuttng phap ti§'p can
bi£n quan sat t£i lap
1

1
56 Nhung v£n de nang cao trong PLS-SEM

Cach tiep can hai giai doan (Two-stage approach)

Cach tiep can hai giai doan cho cac HCM la m ot giai phap thay
the de giai quyet cac van de xay ra khi H O C co cac H C M ket qua -
nguyen n h an va nguyen n h an - nguyen nhan d d d c giai thich bdi (cac)
bien tiem an tac dong khong phai la cac LOC cu a no trong m ang lddi
m oi quan he rong h o n cua mo hin h dddng dan PLS (Ringle va cong sd,
2012). P h d o n g phap nay d d a tren nguyen t&c tdong td nhii each tiep can
hai giai doan trong phan tich bien dieu tiet trong PLS-SEM (H enseler &
C hin, 2010) d cho no ket hop phiidng phap tiep can bien quan sat lap lai
vdi phan tich diem so bien tiem an. Trong giai do an dau, phiidng phap
tiep can bien quan sat lap di lap lai — nhii diidc the h ien trong Tli lieu
2.6 — diidc sd dung de th u diidc diem so cua cac LOC. Cac diem so nay
dd d c liiu lai tro n g tap dd lieu diidi dang cac bien b 6 sung de phan tich
sau h o n trong giai doan th d hai. Sau do, trong giai doan th d hai, diem
so cua cac LOC phuc vu nhii cac bien h ien thi trong m o hin h do liidng
cua H O C (Tii lieu 2.8).

Td lieu 2.8

Nguon: Dieu chinh tir Hinh B2 trong Ringle, Sarstedt, & Straub (2012).

Ve m at ky thuat, ngiidi ta ildc liidng m o hin h diidng dan PLS cua

giai doan dau tien bang each sd dung phdong phap tiep can bien quan
sat lap lai (Tii lieu 2.6) va them diem so bien tiem a n n h d bien m di vao
tap d d lieu. Sau do, ngddi ta sd dung tap d d lieu n ay va tao ra m o hinh
dddng dan PLS cua giai doan thd hai bang each sd dung cac diem bien
tiem an cua giai doan dau tien n h d cac bien quan sat trong m o hin h do
 Chuong 2: Mo hinh nang cao 57

liidng cua H O C (Tii lieu 2.8). Mac du each tiep can ban dau diidc de
xuat de xd ly sai lam Loai II trong dang HCM ket qua - nguyen nhan va
nguyen nhan - nguyen nhan, each tiep can hai giai doan co the diidc
ap dung chung cho tat ca cac dang HCM (co nghia la cung co the dung
cho HCM ket qua - ket qua va nguyen nhan - ket qua). Cuoi cimg, vi
cac diem so bien tiem an cua giai doan dau trd thanh bien quan sat cua
cac m o h inh do liidng cua cac HOC, viec danh gia cd the d d a tren cac
tieu chi dddc th iet lap tot cho danh gia m o hinh do lddng ket qua va do
lddng nguyen nhan (xem Hair va cong sd, 2017).

Nhtfng de xuat va quy tac kinh nghiem

Bang chdng mo phong ve ldi the tddng doi cua cac bien quan sat
lap lai va phddng phap tiep can hai giai doan de ddc ld d n g HCM la
khan hiem va chi xem xet dang HCM ket qua - nguyen nh an . Ket qua
cong bo cua Becker va cong sd (2012) goi y rang phddng phap tiep can
bien quan sat lap di lap lai th d d n g dddc du tien h d n each tiep can hai
giai doan. Vdi viec gidi thieu phan tich tong tac dong cua cac kieu thu
thap HCM , phdd ng phap tiep can bien quan sat lap di lap lai cung cd the
dddc sd dung trong cac thiet lap ndi cac khai niem nghien c d u khac cac
LOC dd doan H O C 6 cac dang HCM ket qua - nguyen nhan va nguyen
nhan - nguyen nhan. Tuy nhien, each tiep can hai giai doan chdng td cd
gia trj neu nha nghien cdu tim each danh gia ban chat cua khai niem bac
cao hon bang each sd dung phan tich bo td (tetrad) khang dinh (CTA-
PLS; xem C hddng 3). Sd dung diem so bien tiem an cua cac LOC td giai
doan 1 lam dau vao cho dac diem m d h inh do lddng d giai do an 2, CTA-
PLS danh gia cau tru e hiep phddng sai cua md hinh do ld d n g H O C de
kiem d in h xem lieu HCM co tddng dng vdi m d h in h th u th a p hay m d
h in h lan to a hay khong. Danh gia n h d vay la k h o n g k h a th i vdi phddng
phap tiep can bien quan sat lap di lap lai vi CTA-PLS — khi dddc thde
hien trong cac chddng trinh phan m em n h d SmartPLS 3 — chi xem xet
cac hiep phd d n g sai trong cac m o hinh do lddng chd khong phai trong
sd cac bien tiem an. Dda tren cac ket qua CTA-PLS, cac n h a nghien cdu
cd dddc n h d n g hieu biet bo sung de xac dinh HCM va cd th e ddc lddng
lai md h in h b ln g each sd dung p h dong phap tiep can bien quan sat lap
di lap lai. T d lieu 2.9 tom tat cac quy tac kinh nghiem ve ph an tich HCM
trong PLS-SEM.
58 Nhdng van d§ ndng cao trong PLS-SEM

Til lieu 2.9

*

• Kiem tra va m o ta ro rang cac co sd ly thuyet va khai niem cua dang

HCM .

• Thong thiidng, sii d u n g phiiong phap tiep can bien quan sat tai lap
de m o h in h hoa cac H C M tro n g m o hinh diidng d an PLS. Dam bao
ran g cac LOC co so liio n g b ien q u an sat b a n g n h a u . N eu khong kha
thi, danh gia tac dong cua viec xoa bien quan sat len m oi quan he
gifla H O C va cac LOC.

• Sii d u n g phiiong phap tiep can h ai giai d o an k h i m uc d ich la danh

gia m oi qu an he giufa H O C va cac LOC bang each sii dung CTA-
PLS. Sii d u n g diem so bien tiem an til giai doaij 1 lam dau vao cho
dac tin h m o h in h do liidng d giai doan 2, v& chay CTA-PLS tren
m o h inh.

• Trong trildng hop H O C cua H C M k£t qua - nguyen nhan hoac

nguyen n h an - nguyen nhan co m ot khai n iem tac dong tien de
khac cac LOC, ap dung phan tich tdc dong to n g cua cac dang thu
thap H C M bang each sii dung phiiong phap tiep can bien quan sat
tai lap. .........................................................

• Thong thilOng, sii dung so do trong so n h an to de Udc lilong tham

so. Khi H O C d o lilOng ket qua diloc ilOc lilong bang M ode B, hay
sii dung so do trong so diidng dan.
• Dam bao rin g tat ca cac khai niem nghien ciiu trong HCM deu
d ap ling ta t ca cac tieu chi danh gia m o h in h d o liidng tieu chuan.
Khdc vdi cac khai niem nghien ciiu khac trong m o h in h diidng d an
PLS, viec dan h gia H O C khong quan tarn den cac m oi q u an he giufa
H O C va cac bien quan sat cua no m a quan tam den mdi quan he
gida H O C va cac LOC cua no. Cac mdi quan h e nay tiiong ling vdi
he so tai (trong trildng hop cac HCM ket qua - ket qua va nguyen
n h an - ket qua) hoac trong so (trong trildng h o p cac HCM . ket
qua - nguyen nhan, nguyen nhan - nguyen n h an ) va can diioc dien
giai tiiong ting. Gia tri phan biet khong can p h ai diioc thiet lap giuta
H O C va cac LOC. Tuy nhien, cac LOC nen th e hien gia tri phan
biet vdi t2t ca cac khai niem nghien ciiu khac tro n g m b hinh diidng
dan PLS. Xem xet kiem tra gia tri phan biet giufa H O C va cac khai
niem nghien ciiu khac trong mb hinh.
 Cbirang 2: Mo hinh nang cao 59

Tinh huong nghien cu ll minh hoa

Dfla tren mo h inh nghien cflu tinh huong va dfl lieu dflpc trinh bay
trong Chflcmg 1, chung toi phac thao each tao ra m ot HCM cho khai
niem danh tieng cong ty. Mo hinh danh tieng cua cong ty tap tru n g vao
cac khai niem COM P va LIKE dai dien cho hai chieu hfldng rieng biet
cua d an h tien g co n g ty (Tfl lieu 2.2). Thay vi mo hinh hoa tac dong rieng
biet cua cac khai niem tac dong (tflc la ATTR, C SO R, PERF va QUAL)
len COMP va LIKE cung nhfl anh hfldng cfla chung len cac bien ket
qua rieng biet (tflc la C U SA va CUSL), hai khai niem nghien cflu nay co
the dflpc xfl ly dflPi dang cac khai niem con cua khai niem danh tieng
cong ty tong quat hdn. Bang each thiet lap m ot khai niem nghien cflu
bac hai vdi COMP va LIKE dong vai tro la cac LOC, m d hin h dflpng
dan PLS trd nen tinh gon hdn. Tfl goc do ly thuyet do lfldng, COMP va
LIKE co the dflpc coi la nhflng sfl p h an an h cfla danh tieng cong ty (vi
du: Schwaiger, 2004), do do n g u y viec sfl dung dang HCM ket qua - ket
qua vi moi LOC dflpc do theo phan an h ket qua. Tuy nhien, nhflng
ngfldi khac lap lu an rang COM P va LIKE xac d in h d an h tieng cua cong
ty (vi du: Eberl, 2010), do dd suy ra dac diem ky thuat d an g HCM ket
qua - nguyen nhan. Vi chung ta chi xfl ly hai LOC, kiem din h CTA-PLS
doi hoi it nhat bon LOC (xem C hfldng 3 de biet them chi tiet), khong
the cung cap them moi quan tam thflc nghiem lien quan den hfldng cfla
m oi quan he gifla cac LOC va H O C. Sfl tflcfng quan cfla cac LOC nen
cao va phan anh thflc chat m ot HCM ket qua - ket qua (vi du: > 0,5).

Noi each khac, cac he so dflpc fldc lflpng cho cac moi quan he gifla
HOC va cac LOC, dflpc hieu la he so tai, rat co the se dan den cac gia tri
dfldi ngfldng mong m uon la 0,7. Ngflpc lai, doi vdi cac HCM ket qua -
nguyen nhan, moi tflcrng quan rat cao gifla cac LOC tao ra he so phong
dai phflflng sai (VIF) ldn hon 5 la dieu khong m ong m uon. Khi sfl dung
HCM ket qua - ket qua, COMP va LIKE co tddng quan 0,631, gia tri nay
ho trp cho dac diem ky thuat m o hinh HCM dang nay. D ong thdi, khi
gia dinh la m o t H C M ket qua - nguyen nhan, tfldng q u an gifla COM P
va LIKE la 0,633, tao ra gia tri VIF la bang 1 / (1 - 0,63 32) = 1,669. Do
dd, ca hai dang HCM deu co the dflpc chap nhan theo thflc nghiem . Vi
ly do nay, trong cac phan sau day, doi vdi muc dich m inh hoa, chung toi
trin h bay va phan tich ca hai each tiep can rieng biet.
60 Nhirng van dd nang cao trong PLS-SEM

HCM ket qua - ket qua

D e thiet-lap-HGM ket qua - ket qua-eua d a n h tie n g cong ty, chung
ta dda tren p h d d n g phap tiep can bien quan sat tai lap. Cac khai niem
COM P va LIKE dai dien cho cac LOC cua khai niem danh tieng cong
ty tong quat (REPU), dd d c do bang sau bien quan sat com p_l, comp_2,
comp_3, like_l, like_2 va like_3. N ghia la, tat ca cac bien quan sat cua cac
LOC ddpc do theo ket qua dong thdi dddc gan cho m o hinh do lddng
ket q ua cua H O C (Td lieu 2.10). Xet ve dac diem m o hin h cau true, b 6n
khai niem tac dong ATTR, CSOR, QUAL va PERF quan he trdc tiep den
REPU (thay vi COM P va LIKE). T ddng td, REPU hien co lien quan trdc
tiep den hai bien CUSA va CUSL.

D e tao ra HCM n h d dddc the hien trong T d lieu 2.10, hay di chuyen
den SmartPLS Project Explorer va nhan chuot phai vao Extended
m odel tro n g cong trinh nghien cdu danh tieng co n g ty. Trong m enu m d
ra, hay chon Copy. Bay gid nhap chuot phai m ot lan n d a va chon tuy
chon Paste trong m enu. M ot hop thoai m d ra ch o phep ban nhap ten
cho ban sao cua m o hin h mdi dd d c them vao. Sd dung ten td no giai
thich cho dang mo h in h chang h an n h d HCM (reflective-reflective).
Sau khi nhan n u t OK, m o hin h m di, la ban sao cua m o h in h danh tieng
cua cong ty, xuat hien tro n g Project Explorer va m o h in h se td dong
m d trong cda sd m d hin h hoa. Thay vi sd dung tuy chpn Copy va Paste,
ban cung co the sd dung tuy chpn Duplicate. Tiep theo, di chuyen den
cda sd md h inh hoa va chpn bien tiem an COMP. N hap chuot phai vao
COM P se m d m ot m enu bao gom tuy chpn D elete. Chpn tuy chpn nay
va tiep tuc bang each xoa khai niem LIKE.

Sd dung cac tuy chpn m enu Edit -> Add Varent V ariable..., Edit
-> Rename, va Edit Add C onnection ... hoac b a n g each sd dung cac
tuy chpn tiiOng dng trong thanh cong cu, nhap ba k h ai niem nghien cdu
m di vao cda so md hin h gan nhan COMP, LIKE va REPU. Tiep theo,
keo va tha cac bien td khung nhin Indicators d p h ia ddbi ben trai cua
m an hin h len cac bien tiem an trong cda sd m d h in h hoa. D am bao rang
ban chpn bo loc Show All Indicators (co m au x am ) trong Indicators
de cho phep gan lap lai com p_l comp_2, comp_3, like_ l, like_2 va like_3
vao khai niem REPU. M d h in h cuoi cung se giong n h ii m d h in h diidng
dan PLS diiptc the hien trong T d lieu 2.10. M ot each de dang va nhanh
ho n de them m ot bien tiem an vdi cac bien quan sa t cda no vao cda sd
m d h in h hoa la bang each nhan phim C ontrol va c h p n cac bidn quan sat
cua m ot khai niem nhat dinh (vi du: com p_l, comp_2, comp_3). Sau do,
 Chuwig 2: Mo hinh nang cao 61

nhap chuot trai vao cac bien quan sat da chon, gifl nut chuot va di chuyen
tat ca cac bien quan sat da chon cung m ot luc den cfla so m o hin h hoa.
Sau khi ban da tha cac bien quan sat tren m ot vung trong trong cfla so
m o hinh hoa, m ot bien tiem an xuat hien vdi cac bien quan sat da chon
trfldc dd dflpc gan. Bay gib di chuyen bien tiem an den vi tri b an m uon,
sfl dung tuy chon Rename, xuat hidn sau khi nhap chuot phai vao bien
tiem an va sfl d u n g tuy chon Connect trong th an h cong cu de tao m oi
quan he vdi cac bien tiem an khac trong PLS m o h in h dfldng dan. Thay
vi tao m di m o h in h H CM cho vi du d a n h tieng congty, ban c u n g c o the
tai cong trin h nghien cflu Corporate Reputation - Advanced PLS-SEM
Book tfl wvvw.pls-sem.com. Sau khi sfl dung tuy chon Im port -> Im port
Project From Backup File, ban co the m d m o hinh HCM (reflective-
reflective) trong cong trinh nghien cflu Corporate Reputation. Sau dd,
m o hinh dfldng dan PLS hien thi nhfl Tfl lieu 2.10 xuat hien trong cfla
so mo hinh hoa.

De fldc lflpng mo h in h dfldng dan PLS, hay nhap vao Calculate ->
PLS Algorithm trong the SmartPLS. Ngoai ra, ban co the nhap chuot
trai vao bieu tflpng banh xe vdi nh an Calculate trong th a n h cong cu.
C hay thuat toan PLS-SEM bang each sfl dung so do trp n g so nhan td
Factor va cai dat m ac dinh cho sd 12n lap toi da va tieu chi dflng. D am
bao chon tuy chon Mean Replacement trong th e Missing Values, va
khong chon vectd trpng sd trong the Weighting. Sau khi nh ap vao nut
Start Calculation, bao cao ket qua SmartPLS se tfl dong m d ra, cung
cap tuy chon After Calculation: Open Full Report da dflpc chon trong
hop thoai bat dau cua thuat toan PLS-SEM.
Trong bfldc tiep theo, can phai danh gia cac m o hinh do lfldng (H air
va cong sfl, 2017). Ket qua danh gia cua cac m d hinh do lfldng nguyen
nhan va ket qua chi khac nhau m ot chut so vdi cac ket qua dflpc trin h
bay trong ChflOng 4 va 5 trong Hair va cong sfl (2017). Dieu nay cung
co hieu lflc doi vdi cac khai niem LIKE va COM P, bay gid hoat dong vdi
vai trd la cac LOC. Cu the, do lfldng cua LIKE m ang lai mflc do thda
dang ve gia tri hoi tu (AVE = 0,747) va do tin cay nhat quan noi tai (do
tin cay tong h pp = 0,899; C ronbach’s alpha = 0,831). Tflong tfl, cac do
lfldng cfla COMP the hien gia tri hoi tu (AVE = 0,691) va do tin cay nhat
quan noi tai (do tin cay tong hpp = 0,870; C ronbach’s alpha - 0,776).
Tuy nhien, trpng tam chinh cfla chung ta la HOC m di thiet lap, m d hinh
do lfldng cfla nd dflpc mo ta bbi cac m oi quan he dfldng dan cua REPU
vdi hai LOC la LIKE va COMP. Co the thay trong Tfl lieu 2.10, cac he
so tai cfla HCM la 0,914 doi vdi LIKE va 0,891 doi vdi COMP. Vi cac h e
62 Nhdng vdn <3§ n&ng cao trong PLS-SEM
 Chu'o’ng 2: Mo hinh nang cao 63

so tai nay tddng dng ve m at ky thuat vdi cac he so d ddng d in trong m o

hinh d d d n g dan PLS, SmartPLS khong trinh bay so lieu thong ke m o
hinh do liidng cho HCM . Tuy nhien, co the de dang tinh toan so lieu
thong ke co lien quan m o t each thu cong. AVE la tru n g binh cua he so
tai b inh p h d d n g c u a H CM ; do do, co the thu dddc gia tri AVE la (0.9142
+ 0,8912)/2 = 0,815, dieu nay cho thay gia tri hoi tu.

TiiOng tii, ban co the tinh toan do tin cay tong hop theo each th u
cong bang each sd dung cong thiic sau:
M .<<>
B
M \
B
. 1=1 y

trong do I. tddng triing cho he so tai LOC i cua m o t H O C cu the dddc

do bang M cac LOC (i = 1, ..., M ), e. la sai so do liidng cua LOC i, va
var (e;)b ieu thi p h d d n g sai cua sai so do, dddc djn h nghla la 1 - ZCNhap
hai gia tri he so tai se diiqc:
_____________ (0,9144 0,891)2______________
P c ~ (0,914+ 0,891)2+(1 - 0,9142) + (1 - 0,8912)

J§
= -------------------------------- = 0,890
3,258 + 0,206 + 0,164
TiiOng tii, Cronbach’s alpha (chuan hoa) tro n g PLS-SEM diiqc
cho bdi:
, ,, M .r
C ro n b ac h sa= — ;------- ; —
(1 + (M - 1 ) . r )
trong do 7 dai dien cho mdi tiiOng quan tru n g b in h cua cac LOC. Vi
chung ta chi phan tich hai LOC (tdc la M = 2), td d n g quan cua chung
(tdc la 0,631) la mdi tiiOng quan trung binh cda chung va ban thu duqc
ket qua sau cho Cronbach’s alpha:
2 x 0 ,6 3 1 1,262
C ro n b a c h ’s a = ------------------------------ = -------------- = 0,774
(1 + ( 2 - 1) x 0,631) 1,631
N hdng ket qua do tin cay nhat quan noi tai nay cao hdn ngiidng khuyen
cao thddng dung la 0,7 (Hair va cong sd, 2017).
64 Nhung v ln de nang cao trong PLS-SEM

T rong buoc tiep theo, ban co the danh gia gia tri phan biet giua cac
L O C COM P va LIK E bang each sir dung tieu chi HTM T (H enseler
va cong svr, 2015). T u lieu 2.11 cho thay cac gia tri H TM T cung voi
khoang tin cay bootstrap cai thien va hieu chinh - sai lech 95%. Cac ket
qua cho thay gia tri phan biet giua C O M P va L IK E vi gia trj H TM T la
0,780 thap hon gia trj n g u o n g (chap nhan) la 0,85. Hem nira, khoang tin
cay bootstrap tu o n g u n g khong bao gom gia trj 1 (nghia la gia trj H TM T
thap hon dang ke so voi 1). N goai ra, chung ta thiet lap gia trj phan biet
g iu a cac LOC va khai niem do luong theo ket qua CUSL cung n h u voi
khai niem don bien C U SA. Tuy nhien, chung ta khong the thiet lap gia
trj phan b iet g iu a C O M P v a LIK E va H O C R E P U cua chung cung mot
luc. Tuy nhien, ket qua nay duoc m ong dpi, khi m o hinh do lucmg HOC
lap lai cac bien quan sat cua cac LOC cua no.

Vi cac ket qua dua ra su ho tra cho do tin cay va gia trj cua cac do
luong, bay g ib ban co the tiep tuc danh gia m o hinh cau true. T at ca cac
moi quan he mo hinh cau true deu co y nghia (p < 0,01). C ac khai niem
tac dong QUAL (0,444) va P E R F (0,228) co anh hubng m anh nhat len
REPU, trong khi CSOR (0,130) va ATTR (0,142) it anh huong hon (T u
lieu 2.10). Ban than R E P U co tac dong manh doi vbi CUSA (0,527),
ma bien nay lai co quan he manh voi CUSL (0,527). M oi quan he true
tiep giua R E P U va CUSL co phan yeu hon (0,308). G ia trj R 2 cua tat ca
cac bien tiem an noi sinh (tire la 0,715 cho R E P U , 0,277 cho CUSA va
0,543 cho C U SL) la tu o n g doi cao khi dua so lu o n g khai niem tac dong
vao tinh toan. 10*

T U lie u 2.11
Hi c h i H T M T
>
^G O M P CUSA CUSL L IK E REPU
COMP
0,465
CUSA
[0,351; 0,5711
0,532 0,755
CUSL
[0,412:0,643) [0,688:0,818]
0,780 0,577 0,737
LIKE
[0,696; 0,858] [0,487; 0,655] [0.655; 0,816]
1,100 0,562 0,685 1,073
REPU
[1,068; 1,143) [0,472; 0,645] [0,600; 0,766] [1,048; 1,105]

lU u y : Cac g ia tri trong d a u ngoac dai dien cho k h o a n g tin cay sai chech dUOc h ieu
ch in h vd cai th ien 95% cua cac g ia tri H T M T th u dUOc b a n g each chay bootstra p p in g
vdi 5000 m a u tro n g Sm artP L S.
 Chu’cmg 2: Md hinh nang cao 65

Md hinh HCM ket qua - nguyen nhan

La inAt Ida chon thay the cho HCM ket qua - ket qua, ngfldi ta co
th e xem xet ran g hai khia canh C O M P va LIKE tao th an h khai niem
REPU tong quat. Vi ly do nay, chung toi cung m inh hoa cho flng dung
HCM ket qua - nguyen nhan ap dung len mo hinh danh tieng cong ty.
Tu' lieu 2.12 cho thay ket qua m o h in h dfldng dan PLS tfl nhflng can
nhac ly thuyet / khai niem nay.
Tuy nhien, m o h in h dflpc the hien trong Tfl lieu 2.12 the hidn m dt
thiet lap HCM ket qua - nguyen nhan khong the chap nhan dtidc trong
PLS-SEM. Vi cac LOC nhfl COMP va LIKE tflong flng giai thich bdi (gan
nhfl) 100% phfldng sai cfla H O C REPU (R2 1,00), khong cd phiidng sai
con lai nao ma cac khai niem tien de nhfl ATTR, CSOR, PERF va QUAL
co the giai thich. Do do cac moi quan he cfla bon khai niem vdi REPU
trong mo hinh dfldng dan PLS co cac he sd fldc lfldng dfldng d in gan
bang khong (Tfl lieu 2.12). Ket luan rang cac m di quan he nay la sd
khong (va khong co y nghia thong ke) se la qua sdm do cac fldc lfldng
nay la ket qua cfla viec thiet lap H O C REPU ket qua - nguyen nhan.
Tuy nhien, m ot bflc tranh khac xuat hien khi phan tich tong tac
dong cfla m ot kieu thu thap HCM dflpc xac dinh. De dfla thflc hien
phan tich nay, ban se can phai them cac mdi quan he gian tiep xa hdn tfl
A T T R , CSOR, PERF vaQ U AL qua COMP va LIKE vao REPU, nhfl trong
Tfl lieu 2.13. Tiep theo, chay thuat toan PLS-SEM bang each nhap vao
Calculate -> PLS Algorithm trong the m enu. Sfl dung cai dat mac dinh
ngoai trfl so dd trpng sd, nen dat thanh Factor. Sfl dung bao cao ket qua
m d ra sau khi chay thuat toan PLS-SEM, bay gid ban cd the d an h gia cac
m d h in h do lfldng ket qua va nguyen nhan. Ket qua cfla d an h gia nay rat
giong vdi ket qua dflpc bao cao trong H air va cong sfl (2017). Tuy nhien,
sfl cong tuyen tiem an gifla cac LOC cfla HCM ket qua - nguyen nhan
d i i hoi sfl chfl y can trpng. De kiem tra cac van de cpng tuyen, hay vao
Quality Criteria -» Collinearity Statistics (VIF). Cd the thay rang cac
gia tri VIF cfla COMP (2,719) va LIKE (2,296) la dfldi ngfldng 5 dang ke,
dan den sfl flng ho rang sfl cong tuyen khong phai la m ot van de quan
trpng trong tinh huong nay. Cung thay rang gia tri VIF cfla QUAL tfldng
doi cao (4,202) nhflng van thap hdn ngfldng. Tiep theo, chuyen den
Final Results -> Total Effects, m d bang dflpc hien thi trong Tfl lieu 2.14.
Cot cd nhan REPU hien thi cho m oi hang cac tac dong (1) cfla cac LOC
COM P va LIKE len H O C REPU va (2) cfla bien REPU trong m o hinh
dfldng dan PLS (tflc la ATTR, CSOR, PERFva QUAL). Ban thay rang hai
LOC co tac d o n g cao tflong tfl len REPU (0,522 cho C O M P va 0,583 cho
LIK E ) va do do co sfl lien quan ngang nhau de h in h thanh H O C .
66 Nhung v§n d§ nang cao trong PLS-SEM
Chu’o’ng 2: M6 hinh nSng cao 67
68 Nhflng van de nSng cao trong PLS-SEM

Tfl lieu 2.14 I-Ket qua SmartPLS cho Tong tac dong

Trong so cac bien tien de / tien to cua REPU, QU AL co tac dong m anh
nhat (0,444), tiep theo la PERF (0,227), ATTR (0,142) va CSOR (0,132).

nh On g m o i q u an he phi tuyen

Trong PLS-SEM, cac m oi quan he gifla cac khai n iem nghien cflu co the
6 nhieu dang khac nhau. Trong khi cac moi quan he tuyen tinh (linear
relationship) co the dflqc bieu dien bang cac dfldng thang (vdi cac dfldng
doc dflctng hoac am) khi ve cac gia tri cua bien tiem an trong m ot bieu
do phan tan, cac moi quan he phi tuyen (nonlinear relationship) bao
gom tat ca cac quan h? khong phai la dfldng th an g m a la cac dfldng
cong. Cac m oi quan he p h i tuyen xay ra kha th fld n g xuyen nhflng nhin
chung kho xac d in h m ot flu tien ve n en tang ly thuyet. H on nfla, cac moi
quan he tuyen tin h thfldng xap xi cac m oi quan h e phi tuyen d u to t de
cac m di qu an he tuyen tin h xuat hien thda dang.

M ot vi du dien hinh ve m di quan he phi tuyen la sfl khuech tan

cua cac san pham m di nhfl m ot ham theo thdi gian (vi du: Bass, 1969).
Tflong tfl, m di quan he gifla cac hoat dong tiep thi (vi du: chi tieu quang
cao, hoat dong quang cao) va doanh sd ban s in p h am thfldng theo m ot
m di quan he phi tuyen. Cu the, chung ta thfldng-m ong doi m di quan he
tich cflc gifla hoat dong tiep thi va doanh sd ban hang. Tuy nhien, viec
tang cfldng hoat dong tiep thi lien quan den tac d o n g tich cflc nhflng
giam dan ve doanh sd (Hay & M orris, 1991), co n g h ia la sau moi m ot
don vi hoat dong tiep thi dong gop it hon vao viec tang doanh thu.
Tflong tfl, m ot vi du khac la gia dinh pho bien ve m di quan he tuyen
tinh tich cflc gifla sfl hai long va long tru n g th a n h , no co kha nang phi
tuyen (Eisenbeiss, CorneliSen, Backhaus, & H oyer, 2014). Vdi mflc do
hai long cao hon, tai m ot sd diem, do m anh cua ta c dong tich cflc den
 Chu’o’ng 2: Mo hinh nang cao 69

long trung th anh thay vao do giam di; cu the la, tac dong van duong
nhiing do ldn giam.

Viec thiet lap cac anh h d d n g phi tuyen doi hoi phai co lap luan ly
thuyet can th an . Ngoai ra, phan tich so bo co the ho trq cac n h a nghien
ciiu trong viec xac dinh cac moi quan he phi tuyen bang each ve cac gia
tri cua hai bien vdi nhau. Trong phan tich PLS-SEM, dieu nay se lien
quan den viec so sanh diem so cua hai bien tiem an trong m o t do thi
phan tan bang phan m em bang tin h nhii M icrosoft Excel hoac phan
m em thong ke nhii IBM SPSS (xem Sarstedt & M ooi, 2014).

Cach tiep can ban dau de xd ly cac moi quan he phi tuyen dUa tren
cac phep bien doi d d lieu cua m qt hoac ca hai bien gida m oi quan he
d d o c m ong dqi hoac q u an sat. M ot kieu dien hinh la phep bien doi ham
log, ap dung m ot logarit co so 10 cho moi quan sat. Theo din h nghla
logarit khong the ddoc tinh toan cho cac gia tri am, viec nay xay ra d
diem sd bien tiem an tieu chuan, cac nha nghien cdu can chuyen doi
ban dau cac bien quan sat cua cac m o h in h do lddng cua bien tiem an -
thay vi ban th a n cac bien tiem an. N eu phep bien doi khong tuyen tinh
hoa m ot each day du mdi quan he, chung ta phai tinh den m ot each ro
rang mdi qu an he phi tuyen trong m d hin h dddng dan PLS.

De nam bat tot hon khai niem phi tuyen va each kiem tra cac quan
he phi tuyen trong cac m d hinh d ddng dan PLS (Wold, 1982), hay xem
xet dai dien ch in h thde (ham da thiic) sau day cua m ot tac dong phi
tuyen (ldu y rang chung ta loai trd hang sd va sai sd de d o n gian hoa
m inh hoa): Y, = p lY 1 + p2Y* + ... + ps Y N. Trong phdong trin h nay, Y2
dai dien bien tiem an noi sinh va Y la bien tiem an ngoai sinh, trong
khi p, la he sd d ddng dan cua mdi quan he tuyen tinh gida Y, va Y,. Tac
dong phi tuyen cua mdi quan he du q c them bdi p j f * vao phan tuyen
tin h Pj Yj, theo do N la dai dien cho bac da thde, N la sd m u cao nhat xay
ra trong da thde. Neu N bang 1, chung ta gia d in h chi cd m o t m di quan
he tuyen tin h (tuc la Y2 = p, Y,), trong khi cac gia tri sd nguyen doi vdi
N ldn hon 1 xac dinh cac dang ham phi tuyen khac nhau. Vi du, doi vdi
N = 2, mdi quan he theo mot dang ham m u bac hai: Y2 = p, Y, + p 2T,2.
Doi vdi N = 3, mdi quan he la ham m u bac ba: Y2 = p, Y, + p2Yt2 + p 3Ys3.
Bac da thde xac dinh dang cua ham phi tuyen, trong dd cd khoang cac
diem quay N - 1 noi do ddc cua d u d n g cong thay doi dau. Vi du, khi
N = 2, ket qua ham bac hai co m ot diem ngoat (vi du: dau am cua do ddc
thay doi th an h dau duong n h d d d o c hien th i tro n g hop tre n cung ben
70 Nhung v ln 6& nang cao trong PLS-SEM

trai cua Tit lieu 2.15); khi N - 3, ham bac ba tiep theo cho thay hai diem
ngoat (vi du: vdi n h d n g thay doi dau cua do doc til am sang diidng, va
m ot lan n da thanh dau am nhii dddc hien thi trong hop tren cung ben
trai cua Til lieu 2.16).
Til lieu 2.15 trinh bay cac dang thay the cua cac tac d o n g bac hai
vdi cac gia tri khac nhau cua p, va p r He so p, cho biet m oi quan he
tuyen tin h tong the gida T, va Y2 la d ddng hay am . Tac dong phi tuyen
p 2 xac d inh hddng cua d d d n g cong. Vi du, cac d ddng chaim trong hai
bieu do hang dau cua T d lieu 2.15 cho thay ra n g m o i qu an he tdng the
(tdc la tac dong tuyen tin h ) gida y va Y2 la dddng. Them m gt sd hang
bac hai d ddng tao ra m ot d d d n g cong hinh loi hoac hin h chd u (dddng
lien net trong hop tren cung ben trai trong T d lieu 2.15). N gddc lai, viec
them m ot sd hang bac hai am se tao ra m ot dddng cong lorn hoac m ot
d ddng cong h in h chd u ngddc (dddng lien n£t 6 o tren cung ben phai
trong T d lieu 2.15). Hai dd thi dddi cung trong T d lieu 2.15 th e h ie n c a c

r. r,

Td lieu 2.16 cho thay cac dang tac dong bac 3 vdi cac gia tri khac
nhau cua p,, p 2 va py So vdi cac m o hin h tac d o n g bac 2 trong T d lieu
2.15, cac d d d n g cong the hien cac tac dong bac 3 c o hai diem ngoat (tdc
la thay doi cac sddn ddc cua chung). Vi du trong h o p tren cung ben trai
cua T d lieu 2.16 vdi cac gia trj p ; va p, d d dng n h d n g m ot gia tri am cho
p}, ham bac 3 loi luc dau nhdng thay doi thanh h a m lorn.
 Chucmg 2: Mo hinh ndng cao 71

Tu' lieu 2.16

p ! > 0 v k p 2 > 0 ; vkp3 < 0 p , < 0 vk P i > 0; v k p 3 > 0

vd:Y1 z , 2 l Y1 + Y ? - Y ? vd:Y2 = -2lYl + V,2 + V,3

Yi

G hi chu: D iidng diet doan = tu ye n tinh; d iid n g lam ch a m

bac 3

TU lieu 2.15 va 2.16 hien thi cac hinl g dac tra n g cua cac ham
phi tuyen cho m ot loat cac gia tri Y) v& ien, tro n g p h a n tich
hoi quy chuan, cac gia tri th d d n g d apc n •' pham vi nhd hon chi
bao gom vai m ien nhat dinh cua ham tong the. Vi du, xem xet m ot tac
dong bac 3 d apc hien thi trong hop ben trai cua T a lieu 2.16 va chi cho
phan gida cua ham roi vao pham vi gia trj co lien quan de phan tich (vi
du: khoang chdng td diem ngoat dau tien ben tay trai den diem ngoat
th d hai 6 phia ben tay phai cua ham so). Sau do, thay vi gia dinh m ot
ham bac 3, ta se chi tap trung vao mot h am tang nhe, gan n h d la tuyen
tin h (vdi do d oc ddcfng). Ben canh viec xem xet cac ham phi tuyen co
th e khac nhau, pham vi gia tri co lien quan dat ra m ot kho khan bo sung
trong viec dda ra tien gia dinh ve h in h dan g lien quan cua moi quan he
gida hai bien tiem an. Tuy nhien, trong PLS-SEM, diem so bien tiem an
ddpc chuan hoa, dieu nay lam phde tap kiem duyet pham vi gia tri.
'loi chung, cac tac dong bac 2 ddOc gia thuyet hoa da dai dien cho
lot nhiem vu rat thach thde. Tuy nhien, ly thuyet hoa sd ton tai cua
m ot tac dong.bac 3 tham chi con kho khan hon. Vi ly do nay, xem xet
cac tac dong bac 2 don gian hon th d b n g la trong tarn chinh khi mb hin h
hoa va kiem dinh cac tac dong phi tuyen trd khi ly thuyet cung cap bang
ch d n g thuyet phuc cho cac mdi quan he phi tuyen phde tap hdn. Trong
cac mb h inh khoa hoc xa hoi, cac tac dong bac 2 xay ra pho bien hon
cac loai tac dong phi tuyen khac. Do do, chung toi tap trung vao loai tac
dong phi tuyen nay trong phan con lai cua chdong nay.
72 NhCmg van d e nang ca o trong PLS-SEM

Mo hinh hoa tac dong bac hai trong PLS-SEM

De hieu cac tac dong bac hai dddc thdc hien n h d the nao trong cac
mo h inh d d d n g dan PLS, hay xem xet phiidng trin h bac hai 7, = p {Y (
+ P2Yl2. Mo h in h nay d d d c xac dinh theo each tuyen tin h vi cac he so
diidng dan va p 2 lien ket 7, va 7 2 vdi Y2 la tuyen tinh (vi du: Sarstedt
& Mooi, 2014). Tuy nhien, tac dong phi tuyen (bac hai) diidc gidi thieu
vdi bien T,2, dieu nay doi hoi sd tiiong tac cua 7 ( vdi chinh nd (tdc la
7 * 7 (). Chinh xac hdn, mot tac dong bac hai cd the dddc hin h thanh
nhii m ot triidng hdp dac biet cua m ot mo h in h dieu tiet (H air va cong
sd, 2017) trong dd 7, td dieu tiet m oi quan he gida 7, va Y2 (Tii lieu
2.17). N ghia la, moi quan he tuyen tinh gida 7 va 7 , thay doi ve do ldn
tuy thuoc vao gia tri cua 7,.

D ong thuaui vdi Rigdon, Ringle va Sarstedt (2010), td d n g td n h d

cac tac dong td d n g tac trong phan tich dieu tiet chuan, chiing ta cd
the td dieu tiet (self - m oderator) m o hinh bang each sd dung m ot so
hang tu'dng tac (interaction term). N hdng khac v d i phan tich dieu tiet
chuan, trong dd so hang tddng tac bieu dien td d n g tac gida 7, va m ot so
bien dieu tiet khac, trong td tddng tac, so han g b a c hai the hien td d n g
tac cua 7 vdi chinh nd - n h d dddc bieu dien tru n g p h d d n g trin h sau:
Y 2 = (/9j + p 2Y lj • 7j = p 2Y } + p2Y,2. Neu tac dong b a c hai la dddng, thi
cddng dd cua tac dong 7, len 7 2 tang cho gia tri 7 2 cao hdn. N gddc lai,
neu tac dong bac hai la sd am, gia tri cao h d n tro n g 7 , ngu y m ot tac
dong thap h d n cua 7 len 7 . Td lieu 2.18 m inh h o a khai niem td td d n g
tac vdi sd hang bac hai (nghia la 7 ,2 vdi mdi quan h e p2 den 7 2) n h d la
bien tiem an bo sung trong m d hinh dddng dan PLS, bao gom tich cua
7, vdi ch inh nd, ngoai m di quan he tuyen tinh p t d 7 den Yr T ddng td
n h d vay, trung trd d n g hdp cua m ot mdi quan he b a c ba, ngddi ta se bo
sung them khai niem 7 ,3vdi m di quan he p3 den 7 .
 Chuong 2: Mo hinh nang cao 73

....... •; ■■-.-■“■ p p
Td tddng tac trong PLS-SEM

De tao so hang bac hai Y;2 trong m o hinh dudng d a n PLS, cac
nguyen tdc tiiong tu ap dung nhu trong phan tich dieu tiet chuan. Theo
do, cac nha nghien cdu co the dUa tren cac p h u o n g phap tiep can theo
giao trinh co ban (Hair va cong sU, 2017): p h u o n g phap tiep can tich
bien quan sat, p h u o n g phap tiep can trUc giao va phuong phap tiep can
hai giai doan.

PhUong phap tiep can tich bidn quan sat la each tiep can tieu
chuan de tao so hang tUOng tac trong cac phan tich dUa tren hoi quy
va cung co cac tinh nang noi bat tro n g PLS-SEM. Cach tiep can tich
bien quan sat de thiet lap tac dong bac hai trong PLS-SEM bao gom
viec nhan moi bien quan sat chuan h o a cua bien tiem an ngoai sinh Y
vdi chinh no va tat ca cac bien quan sat khac trong cung m o t mo hinh
do ludng cua bien tiem an. N hflng cai goi la tich bien quan sat nay trd
th anh cac bien quan sat cua sd hang bac hai. P han td dau tien tro n g T u
lieu 2.19 m in h hoa each tiep can tich bien quan sat cho m o h in h trUdc
cua chung toi, gia dinh rang Y, duoc do bang hai bien qu an sat (x, va
x 2). TUOng dng, sd hang bac hai cd ba tich bien quan sat, dat ten la x ,* ,,
£,-£2>va x 2-x2. NhU trong phan tich bien dieu tiet tieu chuan, phuong
phap nay chi ap dung khi khai niem ngoai sinh Y duoc do ludng theo
ket qua. Hdn nda, vi phUdng phap tiep can nay yeu cau cac bien quan
sat cua khai niem ngoai sinh duoc tai sd dung trong md h in h do ludng
cua sd hang bac hai, each tiep can tich bien quan sat chac chan se cho ra
sU cong tuyen trong md h in h d u d n g d in , nhUng dieu nay cd the d u o c
xd ly de dang bang each sd dung each tiep can tn.ic giao (v u o n g goc).
Cach tiep can trUc giao la m ot phan m d rong cua each tiep can
tich bien quan sat va cung cap m ot giai phap cho van de cong tuyen cua
phuong phap tich bien quan sat. De bat dau, n h u trong p h u o n g phap
74 Nhung van fle nang cao trong PLS-SEM

tich bien quan sat, each tiep can trUc giao tao ra tat ca cac tich bien quan
sat cua so hang bac hai, x ^ x ^ x^-x^ va x 2-x2tro n g vi du cua chung ta. Sau
<36, no se hoi quy td n g tich bien quan s i t len tat c a cac bien quan sat cua
khai niem ngoai sinh Y . Trong m di m o hinh hoi quy, tich bien quan sat
(vi du: *,-*,) dai dien cho bien p h u thuoc, trong khi tat ca cac bien quan
sat cua khai niem ngoai sinh (6 day, x, va x 2) hoat dong nhii cac bien
doc lap:

= M i + b2X2 + en
x , x 2 = b3-Xj + b - x 2 + eu

■x. = “br-x,
X2 ~2
+ b,-x^
5 "1 '
+ e„
" 2 ' “ 22

Khi kiem tra cac ket qua, chung ta khong quan tam den cac he so
hoi quy b] den b6 nhiing de y den cac phan dif chuan h o a en , e{2, va en ,
m a cac he so nay trd thanh cac bien quan sat cua so hang bac hai (Tii
lieu 2.19). Phan tich nay dam bao rang cac bi£n qu an sat cua so hang
bac hai khong chia se bat ky phiiong sai nao vdi bat ky bien quan sat nao
cua khai niem ngoai sinh Tr Noi each khac, so hang bac hai la trdc giao
vdi Tj, loai trii bat ky van de cong tuyen giQa cac khai niem nghien cdu
ddqc sd d u n g d e thiet lap cac tac d o n g tuyen tinh va bac hai. VI each tiep
can trdc giao dda tren p h d d n g phap tich bien quan sat, no cung chi ap
dung khi khai niem ngoai sinh 7 ddpc do lddng theo ket qua.

Cach tiep can hai giai doan (cac tac dong p h i tuyen) ket hpp quy
trinh sau: ( 1) l/d c ldpng m o hinh m a khong co so hang tdpng tac va thu
ddpc diem so bien tiem an, va (2) sd dung diem so bien tiem an n h d la
cac bien quan sat cu a cac bien tiem an trong m o h in h phi tuyen (LVS
trong T d lieu 2.19). U day, tich so p h an td chon loc cua cac diem so bien
tiem an chuan hoa cua bien ngoai sinh Y-, vdi chinh no dong vai trd n h d
m ot bien quan sat cua sd hang bac hai. l_fu diem cu a phdP ng phap nay
la no co the ddpc ap dung cho ca hai bien tiem an ngoai sinh dd p c do
lddng theo ket qua va do lddng theo nguyen nhan.

De xac d inh phdcfng phap tiep can nao nen d d p c d u tien de tao
so hang bac hai, chung ta dpa tren H enseler va C hin (2010), n h d n g tac
gia nay da chay m ot nghien cdu m d phong m d ro n g ve phan tich bien
dieu tiet. Ho so sanh cac phdcm g phap tiep can ve d o nhay th o n g ke cua
chung, do chinh xac ddc ldpng diem va do chinh xac d d doan. C hung
toi cung dpa vao cac khuyen nghi cua H air va cong sd (2017, Chdcmg
7). Noi chung, each tiep can trp c giao chdng to cd gia tri kh i m uc tieu
 Chirong 2: Mo hinh nang cao 75

£• m
Cac htfdng tiep c^n d i dat dtfgrc so h?ng bac 2
trong PLS-SEM ; : '

! LVS(Y,)xLVS(Yi) f-

<b
f LVS(Y,)

cua phan tich la giam thieu sai so ifdc lupng cua tac dong phi tuyen hoac
de toi da hoa dii bao. Nhufng khi m uc tieu la de xdc dinh y nghia thong
ke cua tac dpng phi tuyen, phufong phap tiep can hai giai doan ditoc
76 NhOng v ln de nang cao trong PLS-SEM

khuyen khich. N hin chung, p h d o n g phap hai giai doan la p h ddng phap
linh hoat n h at vi no co the dddc sd dung bait ke bien ngoai sinh ddqc
do "thecTleet "qua hay nguyennhanT L Jo c!o,~cacHTiep can h af g iafd o a n
th d d n g diioc Ou tien.

Cuoi cung, td d n g td nhii sii tiidng tac trong phan tich bien dieu
tiet (H air va cong sii, 2017), dieu quan trong can liiu y Ja cac gia trj ddc
lddng cua p l va p 2 bieu thi do m an h cua m oi qu an he gida y va Y2 khi
Y, co m ot gia tri bang 0. Tuy nhien, tro n g nhieu m o hin h thiet lap, so
k h ong phai k hong phai la so tren thang diem cua Y . Neu day la trildng
hop, viec giai thich tac dong bac hai trd n en co van de. Day la ly do tai
sao cac bien quan sat cua bien doc lap Y can phai diiqc lav trung binh
hoac diiqc chuan hoa. Vi PLS-SEM sd dung duf lieu diiqc chuan hoa,
chung toi tap trung vao tuy chon nay. Tieu ch uan hoa dat d d q c bang
each trd gia tri trung b inh cua bien td moi qu an sat va chia ket qua cho
sai so chuan cua bien (Sarstedt & Mooi, 2014). Sau khi tieu chuan hoa,
gia tri tru n g bin h cua Y( dai dien cho diem tham chieu tro n g m o hin h
bac hai, tao dieu kien cho viec giai thich cac tac dong.

Danh gia cac tac dong phi tuyen

Cac tieu chuan d an h gia m o hinh do lddng va m o hinh cau true cung
ap dung cho cac m o hinh phi tuyen (de biet them chi tiet, xem Hair va
c o n g s d ,2017). Bien tiem an ngoai sinh do theo k e t qua Y, phai dap dng
tat ca cac tieu chi lien q u an ve do tin cay n h at q u a n noi tai, gia tri hoi tu
va gia trj ph an biet. T ddng td, tat ca cac tieu chi d a n h gia m o h in h do
lddng nguyen n h an phai d d q c xem xet td o n g d n g . Tuy nhien, doi vdi
so han g bac hai, khong co yeu cau n h d vay vi bien nay chi phuc vu n h d
m o t phdp do p h u trd, ddqc thiet ke de mo hin h h o a tac d o n g bac hai.
Do do, td d n g td nhd so hang tdong tac trong p h a n tich bien dieu tiet,
m o h in h do lddng cua so hang bac hai khong can phai ddqc danh gia
(H air va cong sd, 2017). Trong each tiep cam hai giai doan, cac tieu chi
danh gia chi ap dung cho giai doan dau tien n h d n g khong ap dung cho
giai doan th d hai vi c lc tieu chi nay chi lien quan d en cac khai niem do
lddng don bien.

Sau khi phan tich cac m o hin h do lddng, bdcfc tiep theo yeu cau
phan tich y nghla cua tac dong bac hai. Thu tuc kiem dinh bao g6m viec
tao phan phoi cua tham so bang each sd dung th u tu c bootstrapping (vi
 Chirang 2: Mo hinh nang cao 77

du: vdi 5000 m au sd dung tuy chon khong thay doi dau; de biet them chi
tiet, xem H air va cdng sii, 2017). Neu khoang tin cay 95% cua tac dong
phi tuyen khong bao gom gia trj bang 0, thi tac dong phi tuyen co sii
khac biet co y nghia thong ke so vdi 0 d m dc y nghla 5%. Trong trifdng
hop do, chung ta ket luan rang bien quan tarn cd tac dong phi tuyen cd
y nghia thong ke. Ngoai ra, chung ta cd the kiem tra gia tri p hoac gia tri
t tdOng dng de danh gia lieu tac dong phi tuyen cd y nghia hay khong.

Khi phan tich y nghia thong ke cua cac tac dong, d day, chung la
phai ghi nh d rang y nghia khong ngu y sd lien quan. D ac biet vdhE kh
th ddc m au ldn hon, tac dong rat nhd cd the trd nen cd y nghia. N hdng
dieu nay khong ngu y rang nhdng tac dong nay cd lien quan. Do dd,
bddc tiep theo la danh gia m dc do cua tac dong phi tuyen bang m dc do
tac d o n g /2. M dc do tac dong/ 2 danh gia sd thay doi trong gia tri R: cua
bien tiem an noi sinh y khi so hang bac hai ngoai sinh Y 2 d d o c loai bo
khoi md hinh. Gia tri m dc do tac dong P ddoc tinh n h d sau:

R2 —ft2
j. ^ n w h in h \>di tdc dpng bac hai mo hinh khong cd tdc ddng bdc hai

1 —R2m d hinh vdi tdc dong bdc hai

Do dd, cddng do tac d o n g / 2cho biet tac dong bac hai ddng gop
bao nhieu vao sd giai thich cua bien tiem an noi sinh. Cac h d d n g dan
chung de danh gia f 1 g d i y rin g c a c gia tri 0,02,0,15 va 0,35 dai dien cho
cac m dc tac dong nhd, tru n g binh va ldn (Cohen, 1988). Tuy nhien, khi
danh gia cac lap luan de ap dung cho tac dong dieu tiet, Aguinis, Beaty,
Boik va Pierce (2005) chi ra rin g cddng do tac dong tru n g binh trong
cac kiem din h dieu tiet chi la 0,009. Trong boi canh dd, Kenny (2015) de
xuat rin g 0,005, 0,01 va 0,025 tao thanh cac tieu chuan thde te hon cho
m dc do cac dong nhd, trung binh va ldn, tddng dng. N h d n g dng cung
chi ra rin g ngay ca n h dng gia tri nay la lac quan cho bai danh gia cua
Aguinis va cong sd (2005).

Ldu y m ot lan nda rang tac dong tuyen tinh th d d n g xuyen hon
khong cung cap m ot xap xi hop ly cua cac tac dong bac hai. Ngoai ra,
cac can nhac tieu chuan lien quan den do nhay thong ke phai ddoc ton
trong (xem them Hair va congsd, 2017). D odd, trd khi do lech chuan la
cdc doan, cac nha nghien cdu nen can than kiem tra xem lieu viec bao
gom mot so hang bac hai cd them vao phan tich cac tac dong tuyen tinh
hay khong.
 78 Nhdng vdn d§ ndng cao trong PLS-SEM

Giai thich ket qua

De giaT fh id T cac Tac'ttpng'Tiac hai,_hay xem xet cac ket qua ddc
lddng cua m o h inh tuyen tinh sau day vdi cac he so (chUa chuan hoa):
Y2= 0,2 • Y r l/d c ld d n g cua p, = 0,2 cho biet tac dong thay doi m ot ddn
vj trong 7, len 7 . C hinh xac hdn, m oi ddn vi bo sung cua V lam tang
Y2 len 0,2 ddn vj. Bay gid, neu chung ta xem xet them m ot tac dong bac
hai, cong th d c co the thay doi th an h Y2 - 0,6 • 7 ; - 0,15 • Y 2. He so p,
cho so hang Y , tuyen tinh thay doi td 0,2 trong m o h in h tuyen tinh ddn
gian th anh 0,6 trong m o hinh bac hai. Tuy nhien, ket qua nay khong cd
nghia la tac d o n g tuyen tinh tan g gap ba lan doi vdi m ot d d n vi thay doi
trong ^ vi hai ddc ld d n g cua p ; k h o n g th e so sa n h trd c tiep dddc. Cu
the h d n , d d c ld d n g p ; = 0,6 tro n g m o hinh bac h a i k h ong ngu y ran g
sd gia tan g cua m ot d d n vi Y: thay doi 7 , bang 0,6 , bdi vi m o t so hang
7, khac to n tai tro n g p h d d n g trin h (nghia la sd h a n g bac hai Y 2). Tac
dong cua ca hai sd hang phai dddc xem xet dong th d i de suy ra m ot each
ch inh xac mdi quan he gida Y, len du: khi chen cac gia tri 0, 1 va
2 cho 7 ; vao m d hinh bac hai (nghia la Y2 = 0,6 • Y, - 0,15 • 7 *), chung
ta thu dddc cac k "

Nhd cd the thay, thay doi m ot ddn vi cua 7, t d 0 den 1 doi hoi sd
thay doi cua 7, bang 0,45 ddn vi, trong khi thay doi m ot ddn vj td 1 den
2 doi hoi sd thay doi nhd hdn dang ke cua Y2 b in g 0,15 ddn vL D o dd,
trai ngddc vdi mdi quan he tuyen tinh, do ddc va d o dd tac dong cua sd
thay doi m ot ddn vi phu thuoc vao m dc do cua 7,. Vi ly do nay, ve ham
bac hai n h d trong T d lieu 2.15 rat htiu ich cho viec trin h bay va dien
giai ket qua.

Viec giai thich tac dong bac hai la tddng td k h i xem xet cac d d lieu
chuan hda va cac ddc lddng tham sd, n h d la pho bien trong cac phan
tich PLS-SEM. Trong tinh huong nay, cac he so c h i ra tac dong cua cac
thay doi trong cac gia tri cua bie’n tiem an lien q u a n den dd lech chuan,
thay vi cac gia tri cua nd n h d vay. C hinh xac hdn, xem xet p h d d n g trin h
bac hai d tren, sd 0 dai dien cho gia tri trung b in h cua 7 n h d diem
tham chieu cho phan tich. Neu m dc 7, tang (hoac giam ) m o t ddn vi dd
lech chuan, tac dong tuyen tinh cua 7 ( len 7 2 thay doi theo do ldn cua
 Chirong 2: Mo hinh nang cao 79

so hang bac hai p2. Vi du, hay xem vi du trddc, trong do tac dong tuyen
tin h bang 0,60 va tac dong bac hai p 2co gia tri la -0,15. Neu gia tri cua
bien nghien cdu ngoai sinh tang m ot don vi do lech chuan, ta se ky
vong moi quan he tuyen tinh gifla Y va Y2 giam 0,15 ddn vi, do do tac
dong d m dc cao hdn Y t bang 0,60 - 0,15 = 0,45. Day la n hdng can nhac
trong trd d n g hdp cac yeu to khac khong doi, co nghia la n h d n g thay doi
noi tren dddc dd bao trong khi moi th d khac trong m o hinh d ddng dan
PLS vim khong doi.

Td lieu 2.20 tom tat cac quy tac de phan tich tac dong phi tuyen
trong PLS-SEM.

H H p H B H Quy tac kinh nghiem de phan trch cac t

m i l l j g U phi tuyen trong PtS-SEN^^
• Thiet lap mong ddi cua ban ve cac moi quan he phi tuyen m ot
d u tien dda tren cac can nhac ly thuyet hoac phan tich trddc cac
moi quan he gida cac bien tiem an bang cac bieu do phan tan -
scatterplots. Hay nhd rang cac tac dong tuyen tinh thddng cung cap
cac tac dong phi tuyen hdp ly gain dung.

• De tao ra phi tuyen, sd dung phdong phap hai giai doan khi khai
niem ngoai sinh do lddng theo nguyen nhan hoac khi m uc dich la
de xac dinh m dc y nghla cua m ot tac dong phi tuyen. Ngoai ra, sd
dung each tiep can trdc giao, dac biet la khi m uc dich la de giam
thieu do lech ddc lddng cua tac dong bac hai hoac de toi da hoa dd
bao. Doc lap vdi cac khia canh nay, each tiep can hai giai doan rat
linh hoat va th d d n g dddc du tien cho viec tao so hang tddng tac.

• Tao so hang bac hai hoac bat ky tac dong phi tuyen nao khac dda
tren dd lieu dddc chuan hoa.
• Cac bien ngoai sinh va noi sinh phai dddc danh gia ve do tin cay va
• gia tri theo cac quy trin h danh gia tieu chuan doi vdi cac do lddng
ket qua va nguyen nhan. Tuy nhien, dieu nay khong ap dung cho
p h i tuyen, dieu m a d d a vao mo h in h do lddng p h u trd dddc tao ra
bang each tai sd dung cac bien qu an sat cua k h ai niem ngoai sinh.

• D anh gia xem cac tac dong phi tuyen cd y nghia hay khong bang each
sd dung cac khoang tin cay bootstrap 95%. Ddi vdi cac tac dong phi
tuyen cd y nghia thong ke, danh gia sd lien quan cua chung dda tren
cddng do tac dong/ 2 cua chung. Ket qua cao hdn 0,005, 0,01 va 0,025
tao thanh cac tac dong / 2 nhd, trung binh va ldn.
80 NhOng v in de nang cao trong PLS-SEM

Minh hoa tinh huong nghien c u Ij

De m in h hoa cho viec ddc liidng cac tac dong phi tuyen, hay xem
xet m o h inh danh tieng m d rong cua cong ty m o t lan nuta. N hii da chi ra
trd d c day, ly thuyet gpi y rang tac dong cua sii hai long cua khach hang
(CUSA) len long trung thanh cua khach hang (C U SL) la phi tuyen (vi
du: Eisenbeiss va cong sii, 2014). Vi ly do nay, chung toi gidi thieu tac
dong bac hai cua CUSA vao m o hin h gia dinh ra n g m ot tac dong diidng
cua CUSA len CUSL giam di khi m dc do CUSA tang len. Phan m em
SmartPLS cung cap tuy chon de tii dong bao h a m m ot so hang bac hai.
N han chuot phai vao khai niem muc tieu CUSL va chon tuy chon Add
Q uadratic Effect (Tii lieu 2.21).

Titv ch<>n the Add Quadratic :!

o Oelete Delete

E3 Rename F2

& Add M o d eratin g Effect ...

jtf Add Q u ad ratic Effect...

Change M o d e o f M easurem ent M odel Aft*Q

V Show In d icato rs of Selected C onstructs Ait* C

V Hide In d ic a to rs Of Selected C onstructs AK*X

©* Align In d ic a to rs Top Alt*W

gO Align In d icato rs Left Alt* A
Align In d icato rs Bottom Ait* 5
_o
©s Align In d ic a to rs Right Ah*D

_Align Selected Elements T op

Trong m an hinh sau, chi dinh CUSA lam Bien doc Jap. Trong the
Calculation Method, ban cd the chon gida cac p h ep tinh nhii tich bien
quan sat, trdc giao hoa va phiiong phap hai giai doan de tao so hang
bac hai. Trong nghien cdu tinh huong m inh h o a, moi quan tam chinh
cua chung ta la danh gia m dc y nghia cua tac d o n g phi tuyen (thiidng la
triidng hop trong cac dng dung PLS-SEM), do d o , each tiep can hai giai
doan la thich hpp. C hon tuy chon Two-Stage cu n g n h d Standardized
va Automatic trong Advanced Settings (Td lieu 2.22).
 Chu’o’ng 2: Mo hinh nang cao 81

Khi ban nhan OK, SmartPLS se bao gom so hang phi tuyen co nhan
la Quadratic Effect 1 trong cda so m o hinh hoa. Neu m uon, b an co the
nhap chuot phai vao khai niem nay va chon tuy chpn R enam e de chon
m ot ten khac (vi du: CUSA2). M au sac khac nhau cua no cung chi ra
rang bien nghien ciiu nay la mot so hang bac hai. Nhap chuot phai vao
so hang bac hai va chon tuy chpn thanh Show In d icato rs o f Selected
C o n stru cts. Bien quan sat CUSA diidc tao trong giai doan th ii hai cua
each tiep can hai giai doan sau do se xuat hien trong cua so m o hinh
hoa. Bay gid ban co the tien hanh phan tich bang each chay thuat toan
PLS-SEM (su' dung so do trong so diidng dan va thay the gia tri trung
binh cho cac gia tri bi thieu). Tii lieu 2.23 cho thay cac ket qua tronggcila
so m o hinh hoa. Phiidng trinh sau m inh hoa cac ket < m oi quan
he gida CUSA va CUSL: CUSL = 0,467-CUSA -0,046-'

Tii lieu 2.22 Sm aiPLS

Square Effect
— Basic Settings
Dependent Variable

Independent Variable

Calculation Method iu ct Indicator

i Stage
logonalization

— Advanced
Product Term Generate C j Unstandardized
O Mean Centered
'&) Standardized

Vi CUSA la m ot khai niem don bien, ban khong the ap dung tieu chi
danh gia m o hinh do liidng va do do, trdc tiep tap tru n g vao y nghia cua
so hang bac hai C U SA \ co moi quan he bang -0,046 vdi CUSA. De danh
gia lieu sd hang bac hai nay cd y nghia thong ke hay khong, hay chay
thu tuc bootstrapping bang each vao Calculate -> Bootstrapping trong
m enu SmartPLS hoac vao cda so M odeling va nhap vao bieu tddng
Calculate, tiep theo la Bootstrapping (ldu y rang dau tien b an can phai
quay lai cda so Modeling trddc khi bieu tiidng Calculate xuat hien).
Gid lai tat ca cac cai dat cho viec xd ly gia tri khuyet va thuat toan PLS-
SEM n h d trong ddc ldpng mo hinh ban dau va chpn tuy chpn No Sign
Changes, 5000 mau bootstrap va tuy chpn Basic Bootstrapping. Trong
82 Nhung vdn de nang cao trong PLS-SEM

cai dat nang cao, chon kiem dinh Bias-Corrected and Accelerated
(BCa) Bootstrap, tw o-tailed va m dc -y nghia 0^)5 (T d lieu 2.24).-Tiep
theo, nhap vao Start Calculation. Sau khi chay th u tuc, SmartPLS hien
thi cac ket qua bootstrapping cho cac m o h in h do lddng va m o hin h cau
true tro n g cda so m o hinh hoa. Sd dung hop C alculation Results d phia
dddi ben trai cua m an hinh, ban cd the chon xem SmartPLS hien thi gia
tri t hoac gia tri p tro n g cda so m o hin h hoa hay khong.

C hung ta th u ddqc gia tri p la 0,055 va co the k e t luan rang so hang

bac hai CUSA2 khong cd y nghia d m d c y nghia 5%. Ldu y rang cac ket
qua ddqc trin h bay d day se cd kha nang khac v d i ket qua cua ban va
se thay doi lai khi chay lai bootstrapping vi thu tu c nay ddqc xay ddng
tren cac m au ddqc rut ngau nhien. Bang each tru y cap bao cao ket qua
bootstrapping, ban cd the truy cap tdng quan chi tiet h o n ve ket qua.
Bang dddi Final Results -» Path Coefficients cung cap tdng quan ve ket
qua, bao gom sai sd chuan, gia tri trung binh cua bootstrap, gia tri t va
gia tri p. Bao cao ket qua bootstrapping cung c u n g cap khoang tin cay
bootstrap. N hap vao the Confidence Intervals B ias Corrected trong
bao cao ket qua bootstrapping cho thay cac k h o an g tin cay ddqc tao
ra td p h d o n g phap BCa (Td lieu 2.25). Ddi vdi m o i quan he CL7SA2->
 Chirong 2: Mo hinh nang cao 83

CUSL, ban thay rang so 0 rOi vao khoang tin cay ddcfc dieu chinh sai
chech 95% vdi gidi han diidi -0,097 va gidi han tren 0,004. Do dd, cung
di(a tren khoang tin cay ddpc dieu chinh sai chech 95%, ban cd the ket
luan rang CUSA khong cd tac dong bac hai cd y nghia thong ke len
CUSL.

Tai thdi diem nay, m ot mdi quan he bac hai se bi bac bo va mdi
quan he tuyen tinh cd y nghia gida CUSA va CUSL se ddpc gid lai. Tuy
nhien, vdi m uc dich m inh hoa, chung ta tiep tuc phan tich mac du sd
hang bac hai khong cd y nghia. ,- 0 '
Viec giai thich tac dong bac hai cua CUSA sd dung tac dong tuyen
tin h cd y nghia td CUSA den CUSL vdi gia tri 0,467 la diem tham chieu
(Td lieu 2.25). De tang CUSA theo m ot d o n vi do lech chuan, ban them
tac dong bac hai vao gia tri tuyen tinh. Vi sd hang bac hai la sd am,
ngddi ta ky vong mdi quan he gida CUSA va CUSL trd nen yeu hon (tdc
la 0,467 - 0,046) khi CUSA tang theo mot don vi do lech chuan. Ngdpc
lai, khi CUSA giam m ot don vi do lech chuan, ngddi ta hy vong mdi
quan he gida CUSA va CUSL trd nen m anh hon (tdc la 0,467 + 0,046).
Q ua dd, ban cd ddqc tac dong bac hai mong dqi gida CUSA va CUSL
tdong td n h d ddpc the hien trong Td lieu 2.26.

T d lieu 2 .2 4 ring trong SmartPLS

0 Setup j 0 Partial Least Squares (? ) Missing Values W eighting

— Basic S e ttin g s ------

Subsamples 5000

( 3 Do Parallel Processing

Sign Changes # No Sign Changes

(D Construct Level Changes
0 Individual Changes

A m o u n t o f Results ® Basic B ootstrapping

O Complete Bootstrapping

— A dvanced S e ttin g s -

Confidence Interval M ethod o Percentile Bootstrap

0 Studentized Bootstrap
® Bias-Corrected and Accelerated (BCa) Bootstrap
0 OavisionHinkley's Double Bootstrap
O Shi's Double Bootstrap

Test Type C .‘ One Tailed % Two Tailed

Significance Level o.os
84 Nhung vin d§ ndng cao trong PLS-SEM

Td lieu 2 .2 5 K h o a n e titi f !

I flg "Extended m odel.iphm j 0 Bootstrapping (Run *1) S3 1SPSS

P a th Co e ff ic ie n ts

jj?) Mean, STDEV. T-Values, P Values Confidence Intervals Confidence Intervals Bias Corrected i Samples

Original SampL. Sample Mean (... Bias 23% 973% i

ATTR -> COMP 0.086 0.057 -0.029 -0178 0161
ATTR -> UWE 0167 0106 -0.061 •0368 0347
COMP •> CUSA 0146 0146 0.001 0.014 0382
COMP -> CUSL 0.008 0.009 0.001 -0.090 0118
CSOR •> COMP 0.059 0.061 0.002 -0.039 0.168
CSOR -> LJk€ 0.178 0178 •0.000 0.069 0389
CUSA •> CUSL 0.467 0.468 0.001 0382 0356
CUSA2 •> CUSL -0.046 -0.047 -0.001 -OJ097 -0D04
UKE -> CUSA 0436 0.436 -0.000 0315 0348
LIKE -> CUSL 0348 0347 -0.001 0.451
PERF •> COMP 0.295 0.300 0.005 0162 0.432
PERF •> LIKE 0117 0121 0.004 0368
QUAL •> COMP 0.430 0.430 -0.000 0303 0355
QUAL •> UKE 0380 0387 0.007 0.280 0328

Tuy nhien, do bootstrapping xay ddng dda tre n viec tao dd lieu
ngau nhien, ket qua cda ban co the hoi khac so vdi n h dng ket qua ddoc
trin h bay d day. Do do, vi cac ket qua kiem dinh co y nghla ddoc tao ra
trong m inh hoa nay rat gan vdi xac suat sai so 5% ma chung ta da chi
dinh la cap dd tdi han, ket qua m a ban tao ra co th e tiet lo rin g tac dong
bac hai cda CUSA va CUSL la co y nghla. Do dd, ch u n g toi khuyen ban
nen lap lai phan tich bootstrapping de dat dddc sii td tin trong cac ket
qua dddc tao ra. Phan tich bo sung giup hieu ro m o i qu an he phi tuyen
co y nghia hay khdng lien qu an den viec kiem tra sd lien q u an cda tac
dong bac hai. C udng do nhd cda he so (tdc la -0,046) trong vi du nay
da cho biet - du cd y nghia hay khong - rang so h a n g bac hai khong lien
quan cu the. Do do, bddc cuoi cung cua danh g ia ket qua giai quyet
m dc do tac d o n g / cua so hangjbac haL.Bang.cach tru y cap vao Quality
Criteria -> f Square trong bao cao ket qua thuat toan SmartPLS, ban
thay rang m dc do tac dong / cua so hang bac hai c d gia tri la 0,0 1. Gia
tri nay thap h d n gidi han dddi 0,02, theo C ohen (1988), d m dc toi thieu
dai dien cho m ot m dc do tac dong nhd. Giai thich t d do hdn cda Kenny
(2015), ngddi da de xuat rang cac gia tri ldn h d n 0,005, 0,01 va 0,025
thiet lap cac m dc dp tac dpng nhd, trung binh va Ion, chd thay m dc do
anh hddng td nhd den tru n g binh.
 Chu'ong 2: Mo hinh nang cao 85

Mol quan h£ bSc 2 giufa Su1hai long va Ibng

Tu1lieu 2.26
trung thanh cua khich h^ng

Tom lai, sd khong co y nghla cua tac dong bac hai va m dc do

tac dong / 2 thap cua no gpi y rang ban nen xem xet m ot m oi quan
he tuyen tin h th ay vi m ot moi quan he phi tuyen. T rong khi ca hai
phiiong phap deu co ket qua r^t gio n g n h au (Tii lieu 2.26), m oi quan
he tuyen tinh it phde tap hon khi diia ra gia din h trd d c ve m oi quan
he, the hien tin h can doi, m o hinh hoa moi quan he trong PLS-SEM va
dien giai cac ket qua.

<c tinh hdu ich cua cac ma hinh thanh phan thd bac
. M ot m o hinh HCM bao gom m ot khai niem tong quat
(tdc la HOC), diiqc do d m dc trd u tiiong cao hon, dong
thdi bao gom m ot so thanh phan con (tdc la cac LOC), bao gom
nhieu dac diem cu the hon cua khai niem nghien cdu nay. Cac
HCM cho phep giam so ldqng cac m di quan he m o hinh cau
true, lam cho m o hinh d ddng dan PLS trd nen kiem dung hon,
dong thdi tang ldpng thong tin ddqc giai thich cua noi dung
ddqc bao phu bdi cac khai niem td o n g dng.

• Danh gia cac dang khac nhau cua HCM va hieu each sd dung
chung trong PLS-SEM. Viec thanh lap HCMs xay dung dda
tren nhdng can nhac ly thuyet/khai niem ddqc thiet lap can
than. Tren co sd do, cac n ha nghien cdu chon td bon dang chinh
cua HCM. Mdi dang trong so nay m o ta m oi quan he cu the gida
86 Nhflng van de nang cao trong PLS-SEM

H O C va cac LOC cang nhii m o hinh do liidng dddc sii dung

• de-van- hanh eae-khai-niem -nghien cdu d cap-do thap-hdn: ket
qua - ket qua, ket qua - nguyen nhan, nguyen nhan - ket qua va
nguyen nhan - nguyen nhan. N hin chung, H O C cua cac HCM
ket qua - ket qua va nguyen nhan - ket qua dai dien cho m ot
khai niem nghien ciiu tong quat hdn - tiiong til nhii cac m o
h in h do liidng ket qua — dong thdi giai th ich tat ca cac LOC
ph u thuoc. N gddc lai, H O C diidc hinh th a n h bdi cac LOC trong
cac dang H CM ket qua - nguyen nhan va n guyen nh an - nguyen
nh an , tiiong td n h d cac m o hin h do lddng nguyen nhan. Cac
p h d d n g phap tiep can bien quan sat lap di lap lai, phan tich tong
tac dong cita m ot dang th u thap HCM, va each tiep can hai giai
doan cho phep m o hin h hoa va ddc lddng cac HCM trong PLS-
SEM. Khi xac dinh va ddc lddng cac HCM trong PLS-SEM, cac
nha nghien cdu can xem xet cac khia canh khac, lien quan den
so bien quan sat tren LOC, sd do trpng so th u at toan PLS-SEM
va sd dung trong so M ode A va Mode B.

* H ieu each chay phan tich HCM bang phan mem SmartPLS
va each dien giai ket qua. Cac nha nghien cdu co the sd dung
SmartPLS de m o hinh bat ky dang nao tro n g so bon dang HCM
dd d c gidi thieu trong chddng nay. Khi p h an tich ket qua ddc
lddng cua HCM , cac nha nghien cdu can danh gia can than
khong chi cac m o hinh do ld d n g cua cac LOC ma con la m o
h in h do lddng cua H O C . K hac vdi cac k h a i niem nghien cdu
khac trong m o h in h d ddng dan PLS, viec d a n h gia H O C khong
qu an tam den cac m oi quan he gida H O C va cac bien quan sat
cua nd n h dng quan tam den m oi quan he g id a H O C va cac LOC
cua nd. Trong khi cac m oi quan he nay d d d c ve dddi dang he so
d d d n g dan tro n g phan tich PLS-SEM, td q u a n diem m o hinh
hoa, chung tddng dng vdi he so tai (trddng hdp H CM ket qua -
ket qua va nguyen nhan - ket qua) hoac tro n g sd (trd d n g hdp
HCM ket qua - nguyen nhan hoac nguyen n h a n - nguyen nhan)
va can dddc dien giai n h d vay.
* Hieu cac khai niem cd ban ve phan tich p h i tuyen trong bdi
canh PLS-SEM. Cac tac dong phi tuyen x ay ra khi m di quan he
gifla hai khai niem nghien cflu khong theo m o t dfldng thang m a
la m ot d ddng cong k h i ve cac gia tri cua bien tiem an trong m ot
bieu dd phan tan. Khi m di quan he gifla h a i khai niem la phi
 Chuwig 2: Mo hinh nang cao 87

tuyen thi ctidng do cua tac dong giua hai khai niem phu thuoc
vao cac gia tri va do lbn cua sti thay doi tro n g khai niem ngoai
sinh. Khi p h an tich cac tac dong phi tuyen, cac nha nghien cdu
phai diia ra m ot gia din h ve ban chat cua tac dong. Mac du co
nhieu loai tac dong khac nhau, nhting tac dong bac hai la pho
bien nhat.

• Danh gia each set dung phan mem SmartPLS de ddc lifting
tac dong bac hai. Phan tich tac dong bac hai trong PLS-SEM
yeu cau cac nha nghien cdu diia vao m ot so hang tiiong tac de
tinh sti td -td d n g tac cua khai niem ngoai sinh. C ach tiep can
tich bien quan sat, each tiep can trdc giao va p h d d n g phap tiep
can hai giai doan la ba phiiong phap h d u ich de m o hin h hoa so
hang tddng tac. Phddng phap tich bien quan sat va phiiong phap
tiep can trdc giao dtidc gidi han doi vdi cac thiet lap m a trong do
bien tiem an ngoai sinh dddc do lddng theo ket qua. C ach tiep
can hai giai doan co the dtipc sd d u n g khi cac do liidng nguyen
nhan va ket qua dddc bao gom . Cach tiep can trtic giao chdng to
co gia tri khi m uc tieu cua phan tich la giam thieu sai chech tide
liidng cua tac d o n g b ac h ai hoac de toi da hoa dti bao. Tuy nhien,
khi muc tieu la xac dinh y nghia thong ke cua tac dong bac hai,
each tiep can hai giai doan diidc tiu tien. Noi chung, each tiep
can hai giai doan la each tiep can linh hoat nhat va diioc tiu tien.
n y

c a u hoi On tap

1. HCM la gi? Mb ta tdng dang trong so bon dang khac nhau cua
dddc gidi thieu trong chtidng nay.
leu chi nao lien quan den danh gia cac dang HCM khac nhau?

Hau qua gi xay ra khi cac LOC cd so liidng bien quan sat khac
nhau dang ke?

4. Giai thich khai niem td - ttidng tac trong boi canh m o hinh hoa
cac tac dong bac hai.

5. Cach tiep can linh hoat nhat de tao ra m ot so hang bac hai lien
quan den m o hinh do liidng cua khai niem ngoai sinh la gi?

6. Lam the nao de ban danh gia cac ket qua cua m ot mo hinh p h i
tuyen?
88 Nhflng vdn de ndng cao trong PLS-SEM

CAU h 6 i tl / d u y p h a n b i e n

1. Moi khai niem nghien cflu co the dflpc do d cac m dc do trflu

tflpng khac nhau khong?

2. Khi nao ban sfl dung phflong phap tiep can hai giai doan HCM ?

3. Lam the nao de cac HCM giai quyet cac van de ve gia tri phan
biet tro n g PLS-SEM?

4. D da ra m ot vi d u ve m ot tac dong bac 3.

5. Giai thich y nghia nhflng he so dfldng dan trong m ot m o hinh

phi tuyen.

6. Thao luan cac phdong an phan tich tac d o n g p'hi tuyen trong
PLS-SEM.

THUAT NGCT

Bandwidth-fidelity tradeoff Sd can bang gifla ldpng thong tin

chuyen tai va do chinh xac
B o tto m -u p approach Cach tiep can tfl dd d i Jen
C ollect m odel Md hin h th u thap
Cubic effect N^C Tac dong bac 3
Direct effect $ Tac dong trdc tiep
Form ative-formative HCM H CM nguyen n h a n - nguyen nhan
Formative-reflective HCM H CM nguyen n h a n - ket qua
HCM Hierarchical com p o n en t m odel
Hierarchical com m on factor M d hinh nhan to chung thfl bac
model
Hierarchical com ponent Md hinh thanh p h a n thfl bac
m odel (HCM)
H igher-order construct Khai niem bac cao
(HOC)
H igher-order m odel Md hinh bac cao
HOC H igher-order construct
Indirect effect Tac dong gian tie p
Interaction term Sd hang td d n g tac
Jangle fallacy Nguy bien Idn
 Chuwig 2. Mo hinh nang cao 89

Linear effect Tac dong tuyen tinh

L inear relatio n ship Q uan he tuyen tinh
LOC Low er-order construct
Log tran sfo rm atio n C huyen doi sang ham log
L ow er-order c o n stru ct (LOC) Khai niem bac thap
M ediation Trung gian
M ediator m odel M o hinh bien trung gian
M o deration Sd dieu tiet . , - 0 '
M ultiple b attery m odel Mo hinh da chuoi
N o n lin ear effect Tac dong phi tuyen
N o n lin ear relatio n sh ip Q uan he phi tuyen
O rth o g o n alizin g approach Cach tiep can trdc giao (vuong goc)
Polynom ial H am da thdi
Polynom ial degree Bac da thde
Prem ature convergence Sd hoi tu sdm
P ro d u ct in d icato r approach Cach tiep can tich bien quan sat
P ro d u ct in d icato rs Tic h bien quan sat
Q u ad ratic effect ac dong bac 2
Tac
R eflective-form ative HCM H C M ket qua - nguyen nhan
Reflective-reflective HCM H C M ket qua - ket qua
R epeated in d ic a to rs approach Cach tiep can tich bien quan sat tai lap
for cho HCM
S econd-order co n stru c t Khai niem bac 2
S elf-interaction T d tdong tac
Spread m o d el Mo hinh lan toa
^ S ta n d - a lo n e H C M H C M doc lap (ddng m ot m inh)
Top-dow n ap p roach C ach tiep can td tren xuong
Total effect Tac dong tong
Total effects analysis o f Phan tich tac dong tong cua cac dang
collect-type HCM s HCM thu thap
Two-stage ap p ro ach C ach tiep can hai giai doan (cac tac
(n o n lin ea r effects) dong phi tuyen)
Tw o-stage ap p ro ach for C ach tiep can hai giai doan cho cac
HCM s HCM
90 Nhung van d§ nang cao trong PLS-SEM

TAI LIEU DOC THEM

Becker, J.-M., Klein, K/STWetzelsTNL 12012). H ierarchical latent variable
m odels in PLS-SEM: Guidelines for using reflective-form ative type
models. Long Range Planning, 45, 359-394.

Henseler, J., Fassott, G., Dijkstra, T. K., & W ilson, B. (2012). Analysing
quadratic effects o f form ative constructs by m eans of variance-based
structural equation m odelling. European Journal o f Information
Systems, 21, 99-112.

Rigdon, E. E., Ringle, C. M., & Sarstedt, M. (2010). Structural m odeling

o f heterogeneous data with partial least squares. In N. K. M alhotra
(Ed.), Review o f marketing research (pp. 255-296). A rm onk, NY: M.
E. Sharpe. ...
Ringle, C. M., Sarstedt, M., & Straub, D. W. (2012). A critical look at the
use o f PLS-SEM in M IS Quarterly. M IS Quarterly, 36, iii-xiv.

Wetzels, M., O dekerken-Schroder, G., & van O ppcn, C. (2009). Using

PLS path m odeling for assessing hierarchical construct m odels:
G uidelines and em pirical illustration. M IS Quarterly, 33, 177-195.

W ilson, B. (2010). Using PLS to investigate interaction effects betw een

higher order branding constructs. In V. Esposito Vinzi, W. W. C hin,
J. Henseler, & H. Wang (Eds.), Handbook o f partial least squares:
Concepts, methods and applications (pp. 621-652). New York, NY:
Si
 CHUONG 3

Danh gia
Md hinh nang cao

1. Biet each danh gia mo hinh do liidng vdi phan ti'ch bo td

khang djnh trong PLS-SEM (CTA-PLS).
2. Hieu each chay CTA-PLS trong SmartPLS va each dien giai
ket qua.
■ ~
3. Danh gia phan tich bieu do tam quan trong - hieu suat
(IPMA).
4. Hieu each thufc hien IPMA bSng phan mem SmartPLS.
, ,

TONG

cd ban trong bat kv phan tich mo hinh phiidng trin h

phan nhd nhat nao (PLS-SEM) la tran h sd thieu chinh
hinh do lddng (vi du: xac dinh mo h in h do liidng la ket qua
i no la mo hinh nguyen nhan). Bat ky sii that bai n ao trong viec
cac mo hinh do liidng m ot each chinh xac d eu cd th e dan den
cac Ode liidng khong chuan xac ve cac tham so. De giai quyet van de
d ac diem ky thuat m o h in h do liidng thich hop, chiidng nay gidi thieu
phan tich bo td khang d inh (confirmatory tetrad analysis) cho PLS-SEM
(CTA-PLS). H inh thanh tren ly thuyet do lddng du tien va dda tren ly
thuyet va logic (tham khao m o hinh do lddng ddpc m d ta trong ChdOng
1), CTA-PLS cung cap nen tang de danh gia th d c nghiem lieu dd lieu
cd hd trp dac diem ky thuat m d hinh do lddng nguyen n h a n hay dac
diem mo h in h do lddng ket qua hay khong. Nghien cdu gan day keu
92 Nhung vdn <J§ n&ng cao trong PLS-SEM

goi svi dung thdo'ng xuyen CTA-PLS de tran h viec xac dinh m o hin h
do liidng thieu ch in h xac noi chung (vi du: Bollen & D iam antolqulos,
2017) va xac dinh PLS-SEM noi rieng (vi du: H air, Sarstedt, Ringle, &
M ena, 2012).

M ot m oi quan tarn khac khi ap dung m o h in h hoa d d d n g dan PLS

la each trin h bay tot nhat cac ket qua cua m o h in h cau true. Mot each
tiep can tuyet vdi de giai thich va dien giai cac ket qua p h an tich la danh
gia bieu do tarn quan trong - hieu suat (IPM A ). C h u de th d hai chung
toi de cap tro n g chiidng nay m d rong bao cao ket qua chuan ve cac ddc
liidng he so diidng dan, R2, va cac tham so tiidng tii, ddng thdi bo sung
phan giai thich trd c quan dda tren cac gia tri tru n g binh cua cac diem
sd bien tiem an. C hinh xac hdn, IPMA tiidng p h a n ytti tac dong tong,
dai dien cho sii anh hiidng cua cac khai niem ng h ien cdu nguyen nhan
len m ot khai niem nghien cdu m uc tieu cu the, v d i diem sd bien tiem
an tru n g b in h cua chdng, qua do chi ra cddng d o dd bao cua cac khai
niem tac dong ddng triidc. Sd ph an tich nay ch o phep nhan dang cac
khai niem nguyen nh an (dd bao tac dong) cd tarn quan tro n g td d n g doi
cao (thap) tro n g v iec d d bao khai niem m uc tieu (tdc la: cac khai niem
cd tac dong tong m an h [yeu]), n h d n g cung cd h ie u suat tddng doi thap
(cao) (tdc la: diem so bien tidm an tru n g binh th a p [cao]). N hdng phat
hien nay dac biet hdu ich cho viec d iln giai 6 b a t ky cho nao de nhan
m anh nhdng nd ldc de cai thien hi£u suat cua m o t khai niem m uc tieu
chinh ddoc ddc lddng tro n g m o hinh dddng dan PLS.

PHAN TICH B0 TLf KHANG D jNH

Phudng phap
Xac dinh sai m o hinh do lddng la m ot mdi de d o a den tin h gia tri cua
ket qua SEM (vi du: Jarvis, MacKenzie, & Podsakoff, 2003). Vi du, m o
h in h hoa cac bien tiem an phan a n h ket qua (tdc la: vdi m ot m o hinh do
lddng ket qua) khi khai niem hoa md hinh do ld d n g , trong khi cac bien
qu an sat cd dac diem nguyen nhan (tdc la: m o h in h do lddng nguyen
nhan) cd the dan den ket qua sai lpch. Ly do la cac bien quan sat m ang
tin h nguyen nhan khong nhat thiet phai td d n g q u a n va th d d n g khong
cd sd td d n g quan cao. Ngoai ra, cac bien quan s a t nguyen nhan tao ra
cac he sd tai ngoai thap h d n khi dddc the hien tro n g md hinh do lddng
ket qua. Vi cac bien quan sat cd he sd tai ngoai th a p hdn (< 0,40) nen
bi loai bo khoi cac khai niem tiem an do lddng p h a n anh ket qua (xem
 Chirong 3: Danh gia Mo hinh nang cao 93

Hair, Hult, Ringle, & Sarstedt, 2017, C hiidng 4), viec xac dinh khong
dung mo h in h do lddng chang han nhii xem no la mo hinh ket qua
trong khi no p h ai la mo hinh nguyen nhan dan den viec xoa cac bien
quan sat m a chung thde sd can diipc gid lai. Bait ky no ldc nao de sang
loc cac mo h in h do lddng dda tren cac m o hinh tddng quan giufa cac
bien quan sat co the co n hdng hau qua bat lpi cho gia tri noi dung cua
khai niem nghien cdu (vi du: D iam antopoulos & Siguaw, 2006) vi du
nh d viec loai trd m ot bien quan sat (khong chinh xac) th d d n g co tic
dong dang ke len ket qua m o hinh do lddng va cau true.

Cac phdong tien chinh de quyet dinh lieu m ot m o hinh do lddng la

nguyen nhan hay ket qua ddpc xac dinh thong qua luan diem ly thuyet.
Cac h d d n g dan n h d quy tac quyet dinh cua Jarvis va cong sd (2003) (xem
them H airva cac congsd, 2017, C h d d n g 2 ;v a C h d d n g 1 trong cuon sach
nay) to ra hdu ich (Td lieu 3.1). T ddng td, Bollen va D iam antopoulos
(2017) cung cap sd hddng dan h d u ich:

Cac th d nghiem lien quan den td duy la m ot cdng cu m ang tinh khai
niem de xac d in h ban chat cua m ot bien quan sat. M ot n h a nghien
cdu nen hin h dung m ot sd thay ddi trong bien quan sat va hoi lieu
sd thay ddi nay co thay ddi gia tri cua bien tiem an hay khong. Neu
co, day la bang chdng ly thuyet ung ho cac bien quan sat ddpc xac
dinh la n h an qua hay nguyen nhan. Ngoai ra, nha nghien cdu nen
tddng td p n g xem sp thay doi trong bien tiem an va td hoi lieu dieu
nay co kha nang thay ddi gia tri cua (cac) bien quan sat hay khong.
Neu cd, dieu nay m inh chdng rang cac bien quan sat la phan anh
ket qua. Cac thd nghiem n h d vay co the cung cap sd ung ho m ang
tin h khai niem cho m ot loai bien quan sat ddi vdi loai khac.
Trong khi vice m inh chung mo hinh do lirdng ve m at ly thuyet
va khai niem la rat huu ich de tranh sai lam khi xac dinh m o hinh do
lirdng (B ollen & Diamantopoulos, 2016), cung cd cac bien phap thirc
nghiem co the giup cac nha nghien cuu hieu de xac dinh m ot mb hinh
do lirdng thuoc loai ket qua hoac nguyen nhan. Cu the, B ollen va Ting
(1993, 2000) da gidi thieu phan tich bo tu k h a n g d in h , p h u o n g p h a p ma
Gudergan, Ringle, Wende va Will (2008) da ap dung no vao phan mem
PLS-SEM . Phan tich bo tirkhang djnh trong PLS-SEM (CTA-PLS)
cho phep cac nha nghien cdu danh gia tren thirc nghiem xem xet lieu
dac diem mo hinh do lirdng dupe lira chon dira tren nen tang ca sd ly
thuyet co d u p e ho trp bdi dir lieu nghien ciru hay khong (R igdon, 2005).
Chinh xac hon, CTA-PLS co the xac nhan su phu hop cua dac diem ky
94 Nhirng vdn de nang cao trong PLS-SEM

thuat m o hinh do luong ket qua. M at khac, khi phuong phap nay khong
xac nhan-su phu hqp-Gua-md-hinh-do lu o n g ket-qua, no dua ra su ung ho
cho dac diem mo hinh do ludng nguyen nhan. N ghia la, phuong phap
nay cho phdp xac djnh huong cua cac mdi quan he m o hinh do ludng
hoac cung cap su ung ho cho m ot su lira chpn thay the khac. Tuy nhien,
su chuyen doi che do do ludng cua cac bien tiem in (tir dang ket qua
sang dang nguyen nhan va ngugc lai) neu chi d u a tren ket qua CTA-
PLS se khong cd y nghia trir khi su logic ve ly thuyet nen hoac khai
niem cung cap sir ung ho cho su thay doi nay.

T u lieu 3.1
I Cac quy tac quy£t dinh xem x£t>m$t khit
cifa dupe do ludng nguyen nhio hayft^t q s
Md hinh Nguyen nhan Md hinh Ket qua
l.H udng quan he nhan qua Hudng cua quan he Hudng cua quan he
cua khai niem nghien ciiu nhan qua la tit cac bien nhan qua la til bien
duoc do liidng ngu y bdi quan sat di den bien nghien ciiu den cac
dinh nghia khai niem nghien ciiiu— bien quan sat cua nd.
Cac bien quan sat (items) Cac bien quan sat xac Cac bien quan sat
(a) xac dinh cac dac tinh djnh cac dac tinh cua la bieu thi cua bien
hoac (b) co bieu thi cho bien bien nghien ciiu. nghien ciiu.
nghien ciiu khong? V

Nhiing thay doi trong cac Nhiing thay doi trong Nhiing thay doi trong
bien quan sat co cac bien quan sat se bien quan sat khong
nhiing thay doi tro gay ra nhiing thay doi gay ra sii thay doi trong
nghien ciiu khong? trong bien nghien ciiu. bien nghien ciiu.
Nhiing thay doi trong bien Nhiing thay doi trong Nhiing thay doi trong
nghien ctiu cd gay ra nhiing bien nghien ciiu khong bien nghien ciiu gay ra
thay doi trong cac bien quan gay ra thay doi trong nhiing thay doi trong
sat khong?_______________ cac bien quan sat. cac bien quan sat.
2. Kha nang hoan doi cua cacCac bien quan sat Cac bien quan sat nen
bien quan sat. khong can phai hoan dupe hoan doi cho
doi cho nhau. nhau.
Cac bien quan sat cd cung Cac bien quan sat Cac bien quan sat phai
noi dung hoac noi dung cd khong can cd noi dung cd noi dung tUdng tU;
tUOng tu khong? Cac bien tUOng tU; cac bien quan cac bien quan sat nen
quan sat cd chia se chung sat khong can chia se chia se mot chu de
mot chu de khong? mot chu de chung. chung.
Khi bo bdt mot trong cac Viec bo bien quan sat Viec bo mot bien quan
bien quan sat cd lam thay cd the thay doi mien sat khong dupe thay
doi mien khai niem cua bien khai niem cua bien doi mien khai niem
nghien ciiu hay khong? nghien cdu. cua bien nghien cUu.
 Chuong 3: Danh gia Mo hinh nang cao 95

ngh* n
Mo hinh Nguyen nhan Mo hinh K2t qua
3. Hiep phiiong sai gifla cac Khong can thiet cho Cac bien quan sat diipc
bien quan sat cac bien quan sat thay ky vong se thay doi
d6i cung vdi nhau. cung nhau.
Lieu mot sii thay d6i cua Khong can thiet.
mot trong cac bien quan sat A
co lien quan den nhflng thay
doi trong cac bien quan sat
khac khong?
4. Mang Ifldi moi quan he Mang liidi moi quan he Mang liidi moi quan•xP°
cua cac bien quan sat trong cho cac bien quan sat he cho cac bien quan
bien nghien cdu. co the khac nhau. sat khdng nen khac
biet nhau.
Cac bien quan sat co diipc ky Cac bien quan sat Cac bien quan sat phai
vong co cung cac tac ddng va khong can ph&i co cd cung cac tac dong va
cac he qua khong? cung cac tac dpng va cac he qua.
cac he q u ^

CTA-PLS xay diing tren khai niem cac b o t i i - t e tr a d ( t ) , m o ta m oi

quan he gida cac cap hiep phiidng sai. De h ifu ro h o n ve tetrad la gi,
hay xem xet bien tiem Sn diipc do liicmg theo ket qua V vdi bon bien
quan sat x, den x4. Doi v 6i bien nghien cdu nay, chung ta th u diipc sau
hiep phUOng sai (c>) gifia tat ca cac cap co the cua bon bien quan sat, nhii
trong Tii lieiiT ^?

Mot tetrad la sii khac biet trong tich cua m ot cap hiep phiidng sai
va tich cua m ot cap hiep phiiong sai khac. Sau hiep phiiong sai cua bon
bien quan sat dan den sau cap duy n h at tao th an h 3 bo td tetrad:

7 |2 3 4 CT1 2 ’ ° 3 4 C71 3 ’ C r24

7 1342 - ° ’l3 ' C r42 ~ O ’,14 '3 2 va

1423 = cr,14 ' O ’, 2 43

Trong cac m o hinh do liidng ket qua, moi tetrad diipc m ong doi
co gia tri bang khong, va do do, de bien mat di — cac t e tr a d t r i f t tie u
(vanishing tetrad) nay diioc sd dung lam dau vao cho cac phan tich. Ly
do la, phu thuoc vao m o hinh lay m au theo mien (vi du: N unnally &
96 Nhung v in de nang cao trong PLS-SEM

Bernstein, 1994), cac bien quan sat ket qua dai dien cho m ot khai niem
hoac dacd ien ftoflT H a nhauT

TU lieu 3.2

Xi x2 x3 X4

X,

*2 °12

013 d jl
X3

X4 0 |4 034

&
D o do, sii khac biet gida cac cap hiep p h d d n g sai cua cac bien quan
sat dai dien cho khai niem theo each td d n g ta n en bang 0, diia ra m o
hinh lay m au theo m ien co gia tri do diiqc gia d in h bang m o hin h do
liidng ket qua. N eu chi co m ot gia tri tetrad trong m o t m o h in h do liidng
khac biet co y nghia thong ke so vdi so 0 (nghia la no k hong triet tieu),
ngiidi ta phai bac bo dac diem m o hin h do liid n g ket qua va thay vao
do, gia dinh dac diem cua m o hin h nguyen n h a n diiqc thay the. Noi
each khac, CTA-PLS la m ot kiem dinh thong ke xem xet gia thuyet
Hn: r = 0 (nghia 1& tetrad bang 0 va triet tieu) va gia thuyet thay the
Hy r ^ 0 (nghia la tetrad khong bang 0). Nghia la, CTA-PLS ban dau gia
d in h m ot dac diem do liidng ket qua. Tuy nhien, m ot kiem d in h thong
ke cd y nghia cung cap sd ung ho cho H v cho thay rang mo hin h do
liidng la p h an anh nguyen nhan.

Bollen va Ting (2000) cung cap m ot so vi d u so m inh hoa tinh

hflu ich cua CTA trong bdi canh SEM dda tren h iep phddng sai (CB-
SEM). Mac du cac quy trin h khac nhau, viec ap d u n g he thong CTA de
dan h gia cac m o h in h do ld d n g tro n g PLS-SEM tufdng td n h d CB-SEM
(Bollen & Ting, 2000). M qt yeu cau can thiet cho bat ky CTA-PLS nao
la cac bien quan sat co lien quan - it nhat la d m o t m dc do nao do. Neu
chung khong td d n g quan, tat ca cac tetrad theo d in h nghia se bang 0,
dieu nay se lam cho CTA - PLS vo nghia. Do dd, m ot CTA-PLS trddc
tien yeu cau kiem dinh xem lieu it nhat m ot so b ien quan sat cua m o
hinh do lddng td d n g quan cd y nghia thong ke h ay khong, vi du, bang
each sd dung phan m em IBM SPSS Statistics (de b iet them chi tiet, xem
 Chu'ong 3: Danh gia Mo hinh nSng cao 97

Sarstedt & Mooi, 2014). Neu yeu cau nay diioc dap dng, CTA-PLS bao
gom cac bd d c chinh tiep theo sau day:

1. H inh th anh va tinh toan tat ca cac tetrad cho m o h in h do liidng

cua m ot bien tiem an.

2. Xac d inh va loai bo cac tetrad dii thiia.

3. Thde hien kiem dm h y nghia thong ke de xem m oi tetrad cd

triet tieu khong.

4. D anh gia xem lieu cac tetrad khong dii thiia cua m o hinh do
liidng cd triet tieu hay khong.

CJ Bode 1, tat ca cac tetrad cua cac m o hinh do 10dng b ien tiem an
ddoc tinh toan. M ot can nhac quan trong la viec xay diing tetrad ddi
hoi it nhat bon bien quan sat cho m oi m o hinh do lddng. N eu khong,
CTA-PLS se khong tao ra ket qua cho khai niem nghien cdu. Bollen &
Ting (2000) va G udergan va cong sd (2008) cung cap m ot so ldi khuyen
bo sung ve each xd ly cac tin h huong d d d i bon bien quan sat (tdc la hai
va ba cho m oi mo h in h do lddng), n h d n g trong PLS-SEM, chung toi
khuyen ban n en ap dung C TA chi tren cac m o h in h do liidng cd it nhat
4 bien quan sat. Noi chung, se cd [ml / ((m - 4)!4!)]-3 tetrad cho cac mo
hinh do lddng vdi m bien quan sat (Bollen & Ting, 2000). Vi du, m ot
m o h in h do lddng vdi nam , tam, hoac m d d ib ie n quan sat tao ra 15,210,
hoac 630 tetrad.

Bddc 2 cua C M -PLS xac dm h va loai bo cac tetrad thda. Sd dd

thda ton tai bat cd khi nao m ot tetrad cd the ddpc dai dien bdi hai tetrad
khac. Vi du, hay xem xet ba tetrad dddc trich td T d lieu 3.2 (tdc la, r 1234,
Tn42’ t h 23^- tetrad cd the ddqc bieu dien bang sd ket h * p dai so cua
hai tetrad khac (Bollen & Ting, 1993). Do dd, m ot tetrad la thda va cd
the ddqc loai b i (vi du: r |423), de lai hai tetrad khong thda cho phan tich
lai (vi du: r |2}4, r )342).

Trong khi hai bddc dau tien cua CTA-PLS x d ly vdi viec tao va chon
cac tetrad (khong thda) cho moi m o h in h do lddng, cac Bddc 3 va 4 giai
quyet viec kiem dinh y nghla thong ke cua chung. Trong Bddc 3, CTA-
PLS dda tren bootstrapping de kiem dinh xem cac tetrad khong thda
cd khac biet cd y nghia thong ke so vdi 0 hay khong (xem H air va cong
sd, 2017, ChdPng 5, de biet them thong tin ve thu tuc bootstrapping va
cac khuyen nghi cho cac cai dat thuat toan). Phan tich m o t so ldqng ldn
cac tetrad khong dd thda bang each sd dung bootstrapping ddi hoi phai
98 Nh&ng van d i nang cao trong PLS-SEM

chay m ot so liidng ldn cac kiem dinh. Khong nhflng phan tich phflc tap
nhfl vay cang trd nen ton thdi gian. ma con co mo_t van de nghiem trong
hon lien quan den each tiep can nay, dflpc goi la sfl phong dai alpha
(con dflpc goi la van de da kiem dinh). Ih u a t ngfl nay de cap den thflc
te la ban cang tien hanh nhieu kiem dinh d m ot m dc y nghia nhat dinh,
ban cang cd nhieu kha nang nhan dflpc ket qua cd y nghia trong khi dieu
nay khong nhfl vay (tflc la sai lam Loai I). So lflpng tetrad khong thfla
ca n g ld n tro n g m ot m d hinh do lfldng cu the, kha nan g bac bo gia thuyet
H q cang cao (nghia la gia tri cua m ot tetrad khac biet dang ke so vdi 0) se
chi xay ra m ot each ngau nhien. Vi ly do nay, tro n g Bfldc 4, CTA-PLS ap
dung hieu ch inh B onferroni de dieu chinh sfl p h o n g dai alpha. Theo ket
qua cua viec hieu chinh Bonferroni, cac kiem d jn h tetrad trong m oi m d
h in h do lfldng khong gia dinh mflc y nghia cua alpha (thfldng la 10%)
nhflng alpha dflpc chia cho so lflpng cac tetrad k h o n g thfla. Vi du, vdi
10 tetrad khong thfla trong m ot m d h in h do lfldng, m oi kiem dinh se gia
din h m ot mflc y nghia la 0,05 / 1 0 = 0,005. Bfldc 4 tin h toan cac khoang
tin cay dflpc dieu chinh sai chech va dflpc dieu ch in h theo Bonferroni
cua cac tetrad khong thfla cho m ot mflc sai so dflpc xac dinh trfldc.

M ot tetrad khong thfla cd sfl khac biet d a n g ke so vdi so 0 neu

khoang tin cay dflpc dieu ’chinh sai lech va diipc dieu chinh theo
B onferroni cua nd khong bao gom sd 0. Tflc la, n eu m ot m d hinh do
lfldng chi cd cac tetrad triet tieu, chung ta khong the bac bo gia thuyet
khong (null) va gia d in h do la m ot dac diem m d h in h do lfldng ket qua.
Mat khac, neu chi cd m ot trong cac tetrad khong thfla cua m d hinh do
lfldng khac biet dang ke so vdi 0 (nghia la, khong phai tat ca cac tetrad
deu la tetrad triet tieu), ta nen xem xet m ot dac d iem m d hinh do lfldng
nguyen nhan. Trong m oi triidng hpp, dieu quan trp n g can lflu y la neu
CTA-PLS khong ung ho m o t m d hinh do ket qua, cac dieu chinh trong
m d h in h phai phfl hpp vdi cac can nhac ly thuyet / khai niem va khong
chi dfla tren cac ket qua thflc nghiem .

Tfl lieu 3.3 cho thay m ot vi du ve hai bien tie m an dflpc do lfldng
theo ket qua y, va Y1 vdi bon va nam bien q u a n sat tflcrng flng. Md
hinh do lfldng cua T, vdi b on bien quan sat yeu c a u phan tich hai tetrad
khong thfla, tro ng khi m d hin h do lfldng cua Y2 vd i nam bien quan sat
yeu cau xem xet nam tetrad khong thfla.

Tfl lieu 3.4 cho thay cac tetrad (triet tieu) va dfla ra m ot vi dp ve
gia tri cua chung cung nhfl ket qua bootstrapping cho 5000 m au (sai
 Chucmg 3: Danh gia Mo hinh nang cao 99

so chuan bootstrap, gia tri t va gia tri p cua moi tetrad don). Cac gia tri
tetrad r 1235 va r ,352 khac nhau dang ke so vdi 0. Tuy nhien, nhiing ket qua
nay khong tin h den vain de da kiem dinh. Vi ly do nay, cot cuoi cung
trong Tii lieu 3.4 cho thay 90% diiqc dieu chinh sai chech va khoang tin
cay bootstrap diipc dieu chinh theo phdong phap Bonferroni.

Mot tetrad co the khac biet co y nghla thong ke so vdi 0 khi diipc
phan tich trong sii co lap nhiing khi tinh toan cho vain de da kiem dinh,
nd cd the khong cd y nghia diia tren khoang tin cay bootstrap diipc
dieu chinh theo B onferroni va hieu chinh sai chech 90%. Trong vi du
diipc the hien trong Tii lieu 3.4, chung ta thay rang gia tri hai tetrad cua
7 (tdc la r (,35 va r lJ52) cd y nghia vi so 0 khong roi vao khoang tin cay
bootstrap diipc dieu chinh theo Bonferroni va hieu chinh sai chech 90%
(tdc la hai gia trj tetrad nay khong triet tieu). Gia tri cua tetrad t )23. cao
h o n dang ke so vdi 0, trong khi gia tri cua tetrad r thap hon dang ke
so v d i 0. D o dd, cac ket q u a CTA-PLS cho thay rang 7 2 n en diipc xac
din h la bien nghien cdu nguyen nhan. Liiu y m ot lan nda rang ket qua
nay khong cd nghia la chung ta chuyen doi m ot each may m dc th an h
dac diem do ld d n g nguyen nhan, vi bait ky thay doi nao ve do ld d n g phai
dope chdng m inh bang cac xem xet mang tinh ly thuyet. Trai ngiipc vdi
7 2, tat ca cac tetrad cua md hinh do lddng cua bien y khong khac biet
dang ke so vdi 0, dieu nay chdng m inh thde nghiem dac diem m d hin h
do liidng ket qua.
100 NhCrng vln dd nang cao trong PLS-SEM

Gia tri
tetrad
Sai JcTdAwSt
bootstrap
p f §
m
0,194 0,150 1,298 0,194 [-0,099; 0,488]

is -0,115 0,182 0,632 0,527 [-0,469; 0,245]

■” ■ Gia tri
% tetrad bootstrap
*W ’ *
|£ffP
:Air• /;>-•; 0,159 0,139 1,140 0,254 [-0 ,1 6 5 ^ ,4 8 3 ]
; 1234.
0,223 0,145 1,538 0,124*; H-0,116; 0,558]
T }243 -

rms 0,483 0,142 3,408 0,001 ^ 0 ,1 5 1 ; 0,811 ]

"T
1352
-0,346 0,121 2,856 0,004 [-0,626; -0,062]
“T -0,089 0,136 0,656 0,512 [-0,404; 0,230]
13.45

3Cl = 90% cac khoang tin cay bootstrap dirge dieu chinh sai chech va dugc
dieu chinh theo Bonferroni

Cuoi cung, chung toi m uon trin h bay hai liiu y: D au tien va triidc
het, CTA-PLS khong phai la giai phap cho m oi van de va ket qua cua
no khong lam cho cac nha nghien ciiu suy nghl ky liidng ve dac diem
ky thuat cua cac mo hin h do liidng. Thay doi hiid n g cua cac m o h in h
do liidng chi diia vao ket qua CTA-PLS la khong hpp ly. Thay vao do,
cac ket qua CTA-PLS nen diipc hieu la dau vao d e xac nhan lap luan ly
thuyet va khai niem co ban hoac phe binh nhiing danh gia lai can than
dac diem m o h inh do liidng theo logic ly thuyet va khai niem. Thil hai,
ket qua CTA-PLS khong cho phep diia ra ket luan ve gia tri noi dung cua
cac m o hinh do liidng. Tile la, cac ket qua cua CTA-PLS khong cho biet
lieu cac phan cua pham vi bien nghien ciiu co dtipc cac nam bat bdi cac
bien quan sat hay khong. Khia canh nay cua m ot p h an tich thiic nghiem
doi hoi sii suy luan can than dpa tren ly thuyet va logic, co le can them
nghien ciiu d in h tin h va co kha nang vdi sii h o trd cua cac chuyen gia
am hieu ve m ien khai niem.

Cac nguyen tac ca ban ve each tien hanh CTA-PLS diiPc the hien
trong Tii lieu 3.5. Trong nghien ciiu m inh hoa sau day, chung toi trinh
bay each ap dung phan tich cho vi du m o hinh d a n h tieng cdng ty bang
each sd dung phan m em SmartPLS.
 Chircmg 3: Danh gia Mo hinh nang cao 101

Minh hoa tinh huong nghien cull - Phan tich bo tfl khang djnh
Phiiong phap CTA-PLS dope thpc hien trong phan m em SmartPLS
3, tao dieu kien thuan lpi cho viec sfl dung no m ot each d o n gian. Lflu
y rang SmartPLS cung cho phep chay CTA-PLS tren ca hai bien tiem an
do lfldng ket qua va nguyen nhan. Trong ca hai triidng hpp, gia thuyet
khong (null) gia dinh mo hinh do lfldng ket qua khi tien hanh kiem
d in h thong ke. Ket qua CTA-PLS flng ho hoac thach thflc m o hin h do
lfldng ket qua hay nguyen nhan da chon. Tuy nhien, vi nhflng ly do
dflpc neu d tren , CTA-PLS trong Sm artPLS chi bao gom cac bien tiem
an co bon bien quan sat hoac nhieu hon.

M inh hoa cua chung ta ve CTA-PLS d p a tren m o h in h danh tieng

cong ty, co ba bien tiem an vdi it n hat bon bien quan sat cho m oi bien
tiem an: CSOR, PERFva QUAL. De bat dau phan tich, chung ta can phai
kiem tra xem cac bien quan sat cho moi m d hin h do cd m di tflong quan
rat gan bang 0. Trong triidng hpp dd, tetrad se theo dinh nghia bang 0 va
phan tich trd n en vo nghia. De cd dflpc cac m di tflong quan gifla cac bien
quan sat, hay dieu hfldng den Corporate Reputation - Advanced PLS-
SEM Book trong Project Explorer va nhaip dup chuot vao Extended
m odel. Tiep theo, chuyen sang Calculate -» PLS A lgorithm tren m an
hinh SmartPLS.

Ngoai ra, ban cd the nhap chuot trai vao bieu tflpng banh xe vdi
n h an Calculate tro n g cdng cu. M ot hop thoai m d ra tii dd ban cd the
chon PLS Algorithm. Gifl lai cac thiet lap mac dinh (tflc la, so dd trpng
sd dfldng d in ) nhfl dflpc md ta trong Hair va cong sfl (2017) va nhap
vao Start Calculation. Sau khi hoan thanh cac tinh toan, bao cao ket
qua se tfl d o n g m d ra . K hi ban nhap vao Indicator Data (Correlations)
trong Base Data d phia dfldi ben phai cua bao cao, SmartPLS se hien
'"thi tat ca cac m di tflong quan cua bien quan sat. Bay gid ban cd the
kiem tra xem cac tfltfng quan gifla cac bien quan sat cua cac khai niem
CSOR, PERFva QUAL cd khac hoan toan vdi 0 hay khong. De cd each
nhin tong quan tot hon, ban cung cd the sao chep va dan cac mdi tflong
quan vao ph an m em bang tinh nhfl M icrosoft Excel bang each nhap vao
Copy to Clipboard (tflc la, Excel Format) phia tren bang ket qua. Ket
qua cho thay cac bien quan sat cd mdi tflong quan nhd n hat la 0,406 doi
vdi CSOR, 0,261 doi vdi PERFva 0,321 doi vdi QUAL (Tfl lieu 3.6). Cac
gia tri ro rang khac biet vdi sd 0, vi vay chung ta cd the tiep tu c thflc hien
CTA-PLS. Neu nghi ngd, ban cung cd the chay m ot phan tich tflong
quan rieng b iet bang each sfl dung cac chflOng trin h phan m em thong
102 Nhung van d i nang cao trong PLS-SEM

ke nhii IBM SPSS Statistics, Stata hoac Statistica. Bay gid chung ta co the
tiep tu c v d i CTA-PLS..

Td lieu 3.5

• Thiet lap phiiong thde do lddng (tdc la ket qua hoac nguyen nhan)
tren co sd ly thuyet va khai niem. De lam nhii vay, cung xem xet cac
quy tac quyet djnh ve dinh tinh, chang han n h ii quy tac thd 4 cua
Jarvis va cong sd (2003), de h o trp viec Ida chon che do dda tren ly
luan ly thuyet. Sd dung CTA-PLS de chdng m in h thde nghiem cac
can nhac theo ly thuyet cua ban.
• N eu co the, xem xet cac bien nghien cdu co it n h a t bon bien quan sat
trong CTA-PLS. Neu khong va chi khi hoan to an can thiet, dda theo
Bollen va Ting (2000) va G udergan va cong sd (2008). Cac tac gia
nay cung cap m ot so ldi khuyen bd sung ve ca c h x d ly cac tinh hudng
diidi bon bien quan sat (tdc la hai va ba cho moi m o hinh do lddng).

• Kiem tra m di tiiong quan cua cac bien quan sat cho moi m d hinh
do lddng. N hdng m di tdong quan nay phai co y nghia thong ke.
Neu khong, tetrad xap xi bang 0 va CTA-PLS trd nen vo nghia.

• Tao tat ca cac tetrad tren moi m d h in h do lddng va loai trd cac tetrad
dd thda. Phan tich xem lieu cac tetrad khong th iia cd khac biet dang
ke so vdi sd 0 hay khong (nghia la, lieu chdng cd triet tieu khong).

• Vi kiem d inh y nghia thong ke tin h toan cho vai tetrad khong thda
tren m oi phep do lddng cung m ot luc, hay sti dung hieu chinh
B onferroni de giai thich cho van de da kiem dinh.

• De kiem d inh m dc y nghia thong ke, chay CTA-PLS cho m ot sd

ldpng ldn cac m iu con bootstrapping (vi du: 5000) va tao khoang
tin cay bootstrap ddpc dieu chinh sai chech 90% va chinh theo
B onferroni cho m oi tetrad khong thda.

• Neu khoang tin cay cua m ot tetrad trong m ot m d h in h do lddng

khong bao gom sd 0, bac bo m d h in h do lddng ket qua va gia djn h
m o t m d h in h do lddng nguyen nhan. N gdpc lai, neu tat ca cac
khoang tin cay bao gom 0, ket qua CTA-PLS theo kinh nghiem
cung cap su ung ho cho dac diem m d h in h do liid n g ket qua.

• Ket qua CTA-PLS dda ra hddng d in ve dac diem ky thuat m d hin h do

lddng dung, nhdng quyet dinh cuoi cung cua phiiong thde do lddng
luon ddpc xay ddng dda tren cac can nhac ve ly thuyet va khai niem.
 Chimng 3: Danh gia Mo hinh nang cao 103

De chay CTA-PLS, nhap vao Calculate ■» Confirm atory Tetrad

Analysis (CTA). Trong hop thoai m d ra (Tii lieu 3.7), dat thong so
5,000 subsamples cho thu tuc bootstrapping va chon kiem dinh 2 duoi
(Two Tailed). Chung toi thiidng de xuait chon Significance Level la 0,1
trong m ot CTA - PLS. N hap chon Start Calculation 6 phia dudi ben
phai hop thoai CTA-PLS de chay phan tich.

Tii lieu 3.6 Moi tocrng quan gitfe

aorJL

Sau khi tinh toan hoan tat, bao cao ket qua se tii dong m d ra. Trong
Final Results, ban co the nhap vao CSOR, PERF va QUAL. Doi vdi
m oi bien tiem an, m ot bang ket qua xuait hien tiidng tii nhii bang dUcfc
hien thi trong Tu lieu 3.8. Liiu y rang ket qua cua ban se hoi khac so
vdi ket qua trong Tu lieu 3.8 do CTA-PLS xay diing diia tren thu tuc
bootstrapping thu ngiu nhien cac m au con duqc sd dung de phan tich.
 104 Nh&ng van de n§ng cao trong PLS-SEM

Cot O riginal Sample (O) cho thay ket qua cua cac tetrad khong
-thda. V i.du,.doi vdi b ien iiem ..an USQR,.m o .h in h do liidng cua.no co
nam bien quan sat, phan tich bao gom nam tetrad khong thiia r I234, t 1243,
r 1235, r 13S2, r |345. So liipng tetrad diipc xem xet cho m oi bien tiem an trong
phan tich theo cap so nhan tang phu thuoc vao so bien quan sat cho
m oi m o h in h do liidng. Trong khi m d hinh do vdi so liidng toi thieu
bon bien quan sat chi bao gom hai tetrad khong thda, co nam tetrad cho
nam bien quan sat (CSOR va PERF) va 20 tetrad cho m o h in h do li
cua QUAL vdi tarn bien quan sat.

Tii lieu 3 .7

— Basic Se ttin gs

Subsam ples

(v| Do Parallel Processing

Test Type © O n e Tailed ® Two Tailed

Significance Level i o

Cac khoan diipc dieu chinh sai lech va diipc dieu chinh
theo Bonfern cac tetrad khong thda co khac biet dang ke so
Low adj. va C l Up adj. trong bao cao ket qua cho thay cac gidi
vdi 0. C l Lov
han tren va dd d i cua khoang tin cay dope dieu ch in h theo Bonferroni
va dieu chinh sai chech 90%. Neu 0 rdi vao k hoang tin cay, tetrad khong
khac biet co y nghla thong ke so vdi so 0, dieu nay ngu y rang no la m ot
S * tetrad triet tieu. NgOpc lai, neu 0 khong rcti vao k h o an g tin cay dope dieu
chinh sai chech va diiqc dieu chinh theo B onferroni, thi tetrad khong
th d a cd sd khac biet dang ke so vdi 0. Tinh huong nay xay ra khi cac dau
cua Cl Low adj. va Cl Up adj. ca hai deu am hoac ca hai deu diidng.
Day la triidng hop doi vdi tetrad 4, 6 va 121 cua khai niem QUAL. Cac
con so xuat phat td danh sach day du cua cac te tra d phat sinh (vi du, 210
tetrad cho m o h inh do liidng cua QUAL vdi 8 bien quan sat). D anh sach
nay trd nen giam dang ke sau khi xac dinh cac tetra d khong thda trong
Bode 2 cua CTA-PLS. Tuy nhien, SmartPLS gid cac so triidc dd de chi
ra nhdng tetrad (khong thda) da dope chpn cho phan tich cuoi cung.
Tii lieu 3.8

Chu’o’ng 3: Danh gia Mo hinh nang cao

g jt L E ^ E ^ J jg j

105
10: paflperf3.paf4perf 5
106 NhOng vSn de nang cao trong PLS-SEM

m O'
I I I fe t& 3 p s d Ol Ol OV 00 oi r r r rn
■ 00 •p" Ol 't f O
o- O O' j n —
IT) V© f—-s
O
vO
o - 1 1
o O o o ” O 0 0 1
'
■ ^
v© '/O v :/ . tooc pv Ol 0
«n
m
rn
in
d O' Ol Ol m ol
l i f e vp o vr, rn v© Oi
m j p d
© o ' o o" o ' O O 01" 0 9
■ 1 I f f 1 i

00 00 00
000^
W ' 00 oo 00 00 oo oo 00 oo oo
0 O
■ '■
k ? J
o
00
o
oo
O
00
0
00^ 00
0
°°«
o
oo
O
°0
o
oo 00 8
oT
B l 01 of oT 01 01 01 oT oT oT d"
oT

m in
m n in
H i IS fi^ E E LO
o
in
o
•n
o o o
in
O
n
o o
m
o 0
0
O
0 8
■ <D ©^ o o o o o
§ ? .3 f a § O o ' © o ' 0* o" cT
o
o '
<=5.
o cT 0' i o '
< cK J
V •.
m o VO «n o ov oi
00
-rT
d
m i
d
Ov
00 >n O' O. 3 •*r O n
ro O' o Ol Ol n .
©~ O O O
o ' o" o ' 1 o*' o o" o CH-
. 1 ’
d OV VO vO 00 O ! Ov
m m 00 rr Ov OO
o ^r ^r ^r rn vO Ol d O.
3 8 o Ol m o o m
O ' Oi*
C
o"
■
©"
1 © o"
i o" o" o ' o ' o 1
O ' <

Ol cn Ol o m rn O 8
o o o o O O O
m ;' s ;- '^ O, O
©^ o o o 8 8 o . O
L
: s
'■
: © o ' © O
cT
f o" O o O' O
1 O

in O' m ■^r o m 00 : OO
00
Ov
04 i
s
o
m
o
Ov
Oi
oo §
rn 0■<=t d
0'
o o o
k
o o
00
o O 0 0
• '
■ t \ "V
d m
•a a
3
s r
^
m
VO
o
Ov
o
O'. s oo
O'
m
Ov
«n in 0
0
tT
O,
Ov
v©
in rn flO V© ° \ s Ov,
d ' O '
£ .5 ^ rn - /o f of 1-^ o '' o
y - ir ; ;0< . ' - CO';i
?** ' '
A -r
in
5
t r M ?■•■••

£ l; •
Ill O'
CO

o '
Ov
m

5
m
rt-

o"
Ol

o"
m
*T

C>
o>
m

o ” o '
Ov
O'

o '
d

o '
O'
m
o '
O'
m

o '
d

o '"

m B & m
VO m v© v©
oc 0 Ov o •^r r- o o- rn 00
S i i vn 01 o- ^r OO VO m v©
- . or m d
Ol cn Ol^
i l l o" oT ©' o '
i o" cf o ' o <P
o ” 0" O'

< "I ^ OV
lOl
O'!
Ol
ro
00
Tf
VO
^r
m
■^r
00
VC
VC
v© 0
01
o
rn
V©
^r
O'
OJ^
O'
rn
O'
00
0
Ol
0
Is
m
' i l l
©' o ' ©' o ’
■ o '
O i,
o" <o o ' o ' 1
' 0' o '
| P |

1
l i 1 3
:> V
•*
■ ■ '3 - :
m
‘° i s
;? 3
■P .
5 IE # r
. <xi "1
W m

. o 4 o f
5 $ 1 1

: 5". : rr
l lc ri
; CP!. a 'f l ; %
'' r n ' • in '-
p i r?
■ • %
M
-
:D '
.. P
O
,'
|v '<%
f "i& i?
C3
a-
■3 ' ■
P
§ :
«3 1
q-
5 1 '
c5..
P
CT ';
mS :
p
'CT '•
''I
■ 5 .
t& iV .
- g .;-
a*

5
•q * §•

5
io o d . in ".
■ rn v. '• i •' r
CO m 5 MV : -rc3 -•^r c3 ■ E m
■« a - P3 . r& , - S
p . w sk P q ,:c r ;

iflPlli
CO . p m
cr . zr

if
• zr & ■■ ;£ ? >
3 f S - . ■S i , -
< 2* ■■ & : ■

*S
»-
& ^ S ig

JV
! § £ ¥ ''•■-•3
.V
M ■

•_',Cvi?v
..
2
O' *
:>;^ 5
&
P ^ ‘s |
.
i
; O :
J z im #
■ p'
lw ? 5

■S ■ :f,
§•
$ m

.:••mr ^ 5
Chu’o’ng 3: Danh gia Mo hinh nang cao 107
108 Nhung v ln de nang cao trong PLS-SEM

K hoang tin cay cua tetrad 4 cua QUAL co can d udi 0,079 va can tren cua
0,880 va do.do khong bao.gbm 0. Tuong.tU,. tetrad 121 co gia tri diidng
dang ke. Vdi can dudi -0,687 va can tren -0,004, khoang tin cay cua
tetrad 6 cung khong bao gom 0 (nghia la nd co gia tri am ldn). N hiing
ket qua nay cho thay m o hin h do ludng cua QUAL thuc sii la m o hinh
nguyen nhan, cung cap ung ho cho dac diem ky thuat goc cua no.

Tu' lieu 3 .9 a A b p ” sqUan s a t c 0 a c i k b i& n n "hic‘n a 5 u tro n g

Trach nhietn x a hoi cua cong ty (CSOR)

csor_l [Cong ty] cU xd theo each cd y thde xa hoi.
csor_2 [Cong ty] th in g than trong viec cung cap thong tin cho
cong chung.
csor_3 [Cong ty] cd thai do cong bang doi vdi cac doi thu
canh tranh.
csor_4 [Cong ty] quan tarn den viec bao ve m di trUdng.
csor_ 5 [Cong ty] khong chi quan tarn ve lai nhuan.
H ieu qua h o a t dong (PERF)
p e r fj [Cong ty] la m ot cong ty dupe quan ly rat tot.
Perf - 2 [Cong ty] la m o t cong ty cd kinh te on dinh.
perf_3 Rui ro kinh doanh cua [cong ty] la khiem ton nhat so
vdi cac doi th u canh tranh.
perf_4 [Cong ty] cd tiem nang tang trUdng.
perf_5 [Cdng ty] cd tarn nh in ro rang ve tUOng lai cua cong ty.
Chat lifting (QUAL)
qual_l San pham /dich vu duoc cung cap bdi [cdng ty] cd chat
lu dng cao.
qual_2 [Cdng ty] la m ot nha doi m di chd k hong bat chudc vdi
y i ton trong [nganh].
sU
qual_3 San pham /djch vu cua [cdng ty] cd gia tot.
qual_4 Dich vu [cdng ty] tot.
qual_5 Cac quan tarn cua khach hang d u o c danh gia cao tai
[congty].
qu a ljS [Cdng ty] la doi tac dang tin cay doi vdi khach hang.
qual_7 [Cdng ty] la m ot cdng ty dang tin cay.
qual_8 Toi rat ton trong [cdng ty].
 ChU’O’ng 3: Danh gia Mo hinh nang cao 109

Ngddc lai, tat ca cac tetrad khong thda cua CSOR va PERF triet tieu
vi tat ca cac khoang tin cay cua cac tetrad deu co so 0. Do do, chung ta
gia dinh rang cac m o hinh do lddng cua cac bien ngien cdu nay la theo
dang ket qua, doi nghich cac gia din h ve m at ly thuyet va khai niem.
Theo cac ket qua nay, chung ta tiep theo kiem tra cac nen tang cua cac
dac diem m o hinh do lddng ap d u n g cac quy tac quyet d in h dinh tinh
cua Jarvis va cong sd (2003) (Td lieu 3.1). Kiem tra cac bien quan sat
cua khai niem (Td lieu 3.9) cho thay rang day la n h d n g n guon kha doc
lap tao th anh bien nghien cdu hdn la phan anh ket qua cua no. Noi each
khac, sd thay doi trong bien nghien cdu khong nhat th iet ngu y sd thay
doi dong th d i cua tat ca cac bien quan sat. Do do, chung ta gid nguyen
dac diem kv thuat do lddng ban dau, hdn nda trong thde te la cac nghien
cdu khac n h au da sd dung dac diem ky thuat nay trong nghien cdu
trd d c (vi du: Eberl, 2010; Schwaiger, Sarstedt & Taylor, 2010). Hay nhd
rang cac ket qua CTA-PLS khong nen ddpc ap dung m ot each may m oc,
dac biet khi cdc bien quan sat nguyen nhan co the bieu hien moi tddng
quan manh me, tuy nhien, khong nen qua ldn (xem C hddng 1).

Tom lai, cac ket qua CTA-PLS chi ra rang m o hinh do lddng cua
QUAL nen dddc xac dm h nhd la m o hinh nguyen nhan tro n g khi cac
m o hinh do lddng cua CSOR va PERF phai dddc xac dinh la m o hinh ket
qua. Trong khi n h d n g phat hien thde nghiem nay phu hdp vdi cac can
nhac m ang tinh ly thuyet va khai niem lien quan den QUAL, chung ddi
nghich vdi dac diem ky th u at ban dau doi vdi cac bien CSOR va PERF.
Tuy nhien, vi quyet dinh each xac din h cac mo hinh do lddng chu yeu
d d a tren cac can nhac ly thuyet va khai niem , cac ket qua CTA-PLS nen
dddc xem la nguyen nhan de kiem tra lai, n h d n g khong n h at th iet phai
xac dinh cac dac diem mo hinh do lddng cua cac bien nghien cdu nay.
Sd kiem tra lai cac nen tang khai niem cho hai bien nghien cd u nay - sd
dung cac quy tac quyet dmh dinh tinh cua Jarvis va co n g sd (2003) - tuy
nhien, ung ho dac diem mo hinh do lddng nguyen nhan ban dau doi vdi
CSOR va PERF.

PHAN TICH BIEU DO TAM QUAN TRONG - HIEU SUAT

Tong quan
Phan tich bieu do tarn quan trong - hieu suat (IPM A; con d d d c goi la
ma tran tam quan trong - hieu suat, bieu do tac dong - hieu suat
hoac phan tich bieu do du tien) m b rong bao cao ket qua PLS-SEM
110 Nhung v§n d§ nang cao trong PLS-SEM

tieu chuan cua cac ddc ldong he so ddo'ng dan va cac tham so khac bang
each them m ot quy trinh nham xem xet cac gia tri trung binh diem so
bien tiem an (vi du: Fornell, Johnson, A nderson, Cha, & Bryant, 1996;
Martilla & James, 1977; Ringle & Sarstedt, 2016; Slack, 1994). C hinh xac
h d n , IPM A tiid ng phan vdi tac dong tong, dai dien cho tam quan trong
cua cac bien nghien cdu tac dong (doc lap) trong viec dii bao m ot bien
nghien cdu m uc tieu cu the, vdi cac diem so bien tiem an tru n g binh cua
chung bieu thi hieu suat cua chung. Sd hop ly la viec xac dinh cac bien
nghien cdu tac dong cd tam quan trong tiidng doi cao trong viec d d bao
bien nghien cdu m uc tieu (tdc la, cd tac dong tong m anh), n hdng co
hieu suat td d n g doi thap (tdc la, diem so bien tiem an tru n g binh thap)
de viec cai thien co the dddc thde hien. Cac m o ta sau day dda tren
h d dng dan cua Ringle va Sarstedt (2016) ve p h d d n g phap.

De m inh hoa khai niem ve m o t IPMA, hay xem xet m o hinh d ddng
dan trong T d lieu 3.10 vdi bon bien nghien cdu tii Y t den Y . Trong m o
hin h d d d n g dan nay, Y4 dai dien bien m uc tieu cuoi cung, dddc d d bao
trdc tiep bdi Y t, Y , va Yy H dn nda, Y; va Y , cd tac dong gian tiep len 7^
thong qua Y r Viec them cac tac dong trdc tiep va gian tiep cua cac bien
nghien cdu tac dong tao ra tong tac dong cua chung len Y , dai dien cho
sd quan trong trong IPMA. N gddc lai, cac diem so bien tiem an trung
binh cua cac bien nghien cdu nay bieu thi hieu suat cua chdng, theo
nghia la cac gia tri cao cho biet hieu suat ldn hon.

IPM A ket h o p hai khia canh nay qua m ot do thi bang each lien ket
cac tac dong tong (chda chuan hoa) tren true x vdi cac diem so bien
tiem an, dddc dao dong tren pham vi td 0 den 100, tren true y. Ket qua
la m ot bieu do n h d trong T d lieu 3.11. De giai thich, chung ta tap trung
vao cac bien ng hien cdu d khu vdc phia dd d i ben phai cua bieu do tam
quan tro n g - hieu suat. Cac bien nghien cdu nay cd tam quan trong cao
doi vdi bien nghien cdu m uc tieu m a chdng cd th e cd tac dong m anh
n h d n g cho thay hieu suat thaip. Do do, cd tiem n a n g dac biet cao de cai
thien hieu suat cua cac bien nghien cdu ddoc d a t trong pham vi nay.
Ngddc lai, du tien cai thien hieu suat cua cac b ien nghien cdu co tam
quan trong thaip h o n lien quan tdi cac bien nghien cd u khac it ro ret hdn.
Trong Td lieu 3.10, Y dac biet quan trong de giai th ich bien nghien cdu
m uc tieu Y4. C hinh xac hdn, trong trdbng hop cac yeu td khac khong
doi (tdc la, khi m oi th d khac van la hang so), m ot d d n vi tang len trong
hieu suat cua Y] lam tang hieu suat cua Y4 bang gia trj tac dong tong cua
Y len Y4 la 0,84. Vi hieu suat cua Y; tddng doi thap, cd nhieu cho de cai
 Chirong 3: Danh gia Mo hinh nang cao 111

thien, lam cho khia canh phu thuoc cua bien ng hien ciiu nay dac biet
phu hop vdi cac hoat d o n g quan ly hoac chinh sach. Trong khi vi du gidi
thieu nay cho thay m ot IPM A d cap do bien nghien cdu, phan tich cung
co the ddqc chay tren cap do bien quan sat. Trong trddng hop nay, cac
diem dd lieu rieng le trong bieu dd tam quan trong - hieu suat diiqc lay
td cac gia tri tru n g binh cua bien quan sat va tac dong tong cua chung
len m ot bien nghien cdu m uc tieu cu the.

Cac nghien cdu PLS-SEM ban dau cho thay ket qua IPM A cung
cap n h d n g h ieu biet q u a n trong ve vai trd cua bien nghien c d u tien de
va sd lien qu an cua chdng doi vdi cac hoat dong q u an ly (vi du: Hock,
Ringle, & Sarstedt, 2010; K ristensen, M artensen & G ronholdt, 2000;
M artensen & G ronholdt, 2010). IPM A cung dac biet hdu ich khi tddng
phan vdi ket qua PLS-SEM td phan tich da n h o m (C hddng 4), n h d
m ot so ng h ien cdu m inh h o a (Rigdon, Ringle, Sarstedt, & G udergan,
2011; Schloderer, Sarstedt, & Ringle, 2014; Volckner, Sattler, H ennig-
T hurau, & Ringle, 2010).

Ngtton: Dieu chinh tu Ringle & Sarstedt (20/6).

IPMA dda tren thu tuc gom 5 bddc ddqc trinh bay trong T d lieu 3.12
(Ringle & Sarstedt, 2016). Bddc dau tien lien quan den viec kiem tra xem
lieu cac yeu cau de tien hanh phan tich co ddqc dap dng hay khong (Bddc
112 Nhung v in d i n^ng cao trong PLS-SEM

1). Phan tich nay tien hanh vdi viec tinh toan cac gia tri hieu suat cua bien
ticm-an- (Bribe 2)-va ca€-gia-tr-i tim q u a n -tro n g c u a chung (Bcidfc 3). Ti£u.
chi bieu do tam quan trong - hieu suat cho m dt bien nghien cdu m uc tieu
dddc chon dita tren cac ket qua nay (Bode 4). Cuoi cung, IPMA cd the
dddc m d rong d cap do bien quan sat de co difqc th in g tin cu the h o n ve
cac hoat dong quan ly hoac chinh sach cd the hieu qua nhat (Biidc 5). Cac
phan sau giai thich tiing bifdc chi tiet hdn.

0.5 0.6
Tam quan trong
y C / __________________
T V Nguon: Dieu chinh tie Ringle & Sarstedt (2016).

Cac budc cua phan tich bieu do tam quan trong -

TU lieu 3 .1 2
hiiu sulft

Bud’C 1 Kiem tra cac yeu cau

Bu’d’c 2 Tinh toan cac aid tri hieu suat

Bu’d’c 3 . Tinh, toan cac gia tri tdrtt quan, trong

Budc 4 Tao biiu dd tam quan trong - hiau suit

Bu’d’c 5 i Md rong IPMA o’ cap d$ b iin quan sat

Nguon: Dieu chinh tic Ringle & Sarstedt (2016)-
 Chuccng 3: Danh gia Mo hinh nang cao 113

Thde hien IPMA mot each he thong

Butfc 1: Kiem tra cac yeu cdu

Cac dng dung IPMA phai dap dng ba yeu cau. Dau tien, viec thay
doi ty le cac diem so bien tiem an tren pham vi td 0 den 100 yeu cau tat
ca cac bien quan sat trong m o h in h ddcfng dan PLS co thang do dinh
ldqng (tham so) hoac it nhat la m ot thang do khoang each deu nhau.
Trong m ot thang do deu nhau, cac khoang each ddqc phan bo theo cac
don vi bang nhau, thddng la trddng h o p khi sd dung thang do thd td.
Tuy nhien, de cho m ot thang do th d td ddqc deu nhau, thang do can
phai dd q c can bang, co nghla la co m ot so ldqng bang n h a u cac m dc
tich cdc (khang dinh) va tieu cdc (phu dinh). Do do, cac th an g do Ida
chon b a tb u o c (tdc la thang do k h o n g co m dc tru n g lap), noi ch in h xac
la khong can bang nhau. Vi du, th an g diem Likert 5 diem vdi hai m dc
p h u dinh (h o an toan khong dong y va khong d o n g y), m o t tuy chon
tru n g lap va hai hai m dc khang din h (dong y va hoan toan dong y) co
the ddqc coi la m ot thang do khoang each deu nhau. Tuy nhien, cac bien
quan sat ddqc do tren thang do danh nghia khong the d d q c sd dung
trong IPMA. T rong m ot thang do danh nghia, viec gan cac so cho cac
dac tinh doi tdqng nhat dinh co the hoan doi, bieu d iln bat ky sd thay
doi thang do tuy y nao. Thd hai, viec m a hoa tat ca cac bien quan sat phai
co cung m ot khu ynh hddng do lddng. Gia tri thap tren thang do phai
dai dien cho ket qua phu din h hoac thap, va gia tri cao phai dai dien cho
ket qua k h ln g djnh hoac cao. Neu yeu cau nay khong ddqc dap dng,
chung ta khong the ket luan rang diem so bien tiem an cao h o n the hien
hieu suat tot hon. Trong trddng hop nay, viec ma hoa bien quan sat can
phai ddqc thay doi bang each dao ngdqc thang do (vi du: tren thang 5
diem, 5 trd thanh I va 1 trd th an h 5, 4 trd thanh 2 va 2 trd th an h 4, va 3
khong thay doi). 'Ihd ba, bat ke m o hin h do lddng nguyen nhan hay ket
qua, cac ddc ldqng trong so ngoai phai la so ddong. Neu trong so ngoai
la so am, thi diem so bien tiem an se khong nam trong pham vi td 0 den
100, nhdng se nam trong khoang td -5 den 95 ch&ng han. Ldu y rang co
cac ly do khac nhau doi vdi cac trong so ngoai am (khong m ong m uon).
Neu trong so ngoai la am va co y nghia thong ke, nha nghien cdu nen
kiem tra bien quan sat va thang do cua no. Co the cd mot khuynh h ddng
khac so vdi cac bien quan sat khac trong m o hinh do lddng, ddi hoi phai
dao ngdqc th ang do. Trong trd d n g h q p trong so ngoai khong cd y nghia
thong ke (cd dau am ), n h a nghien c d u cd the xem xet loai b o cac bien
quan sat dd. Cuoi cung, trong so ngoai am cd the la ket qua cua sd da
114 Nhung v in ell nang cao trong PLS-SEM

cong tuyen cao. Vi du, cac gia tri he so phong dai phiidng sai (VIF) la 5
va cao h d n cho thay m ot van de lien quan den kh'a nang da cong tuyen
(H air va cong sd, 2017). Trong trd b n g hop nay, nha nghien ciiu cung co
the can nhac viec loai bo bdt cac bien quan sat. Tuy nhien, viec loai bo
cac bien quan sat khoi cac m o hinh do liidng lien quan den m ot so can
nhac bo sung nhii diidc giai thich bdi Hair va cqng sii (2017, C hiidng
5) chi tiet hdn.

Mac dii khong phai la m ot yeu cau chinh thde de chay IPM A, cafe
n h a nghien cdu nen xem xet can than cac thiet lap mo h in h diidng dan
PLS m a thien ve sd dung IPMA d cap do bien quan sat. Khi m ot bien
tiem an co m dc do liu tien cao cho m ot bien m uc tieu cy the diidc xac
dinh, no dac biet thuan ldi de phan tich sau hon m o h in h do liidng cua
bien nghien cdu tien de tac dong nay d cap bien quan sat. D anh gia
nay dac biet h d u ich khi m o hinh do liidng dddc Xac dinh la m o hinh
nguyen nh an (xem C hiidng 1) - nhii triidng hyp tro n g mo hinh m inh
hoa cua chung ta trong Tii lieu 3.10 (tdc la, cac b ien quan satx , d e n x l2).
Trong triidng hop nay, cac bien quan sat m o ta cac k hia canh hinh thanh
bien nghien cdu tiidng dng, trong khi trong so cua chung bieu thi tam
qu an tro n g cua tdng khia canh tro n g m oi quan he nay. Do dd, cac khia
canh co ban cua cac bien quan sat cd trong so cao can diidc chu y hdn
de xac dinh cac hoat dong quan ly hoac chinh sach nham cai thien hieu
suat cua bien ng hien cdu m uc tieu (de dng dung, xem th em H ock va
cong sii, 2010). Liiu y rang IPM A cd the diidc ap d u n g tre n bat ky loai
md h in h diidng dan PLS nao, bat ke m d h in h do liidng cua bien tiem
an th uoc do liidng nguyen nhan hay ket qua. IPM A xay d d n g tren cac
trong so ngoai - diidc giai thich chi tiet h d n trong cac p h an tiep theo —
va PLS-SEM luon cung cap cac lidc liidng trong sd ngoai, cung nhii khi
m d h in h do liidng diidc xac dinh la ket qua.
Budc 2: Tmh toan cac gia trj ve hieu suit
Dut lieu bien quan sat xac dinh diem sd bien tiem an va do dd cho
biet hieu suat cua chung. Tiidng td, khi tien hanh m o t IPM A d cap do
bien quan sat, gia tri tru n g binh cua m ot bien quan sa t the hien hieu suat
trung b inh cua nd. Khi tinh toan gia trj trung binh tren bien nghien ciiu
hay tren bien quan sat, dieu quan trong can n h d la cac bien quan sat cd
the diidc do tren cac thang do khac nhau. Vi du, m ot vai bien quan sat cd
the sd dung thang do vdi cac gia tri tii 1 den 5, tro n g khi cac bien quan
sat khac sd dung thang do vdi cac gia tri td 1 den 7 hoac td 1 den 9. De
tao thuan ldi cho viec dien giai va so sanh cac m dc hieu suat, IPMA do
 ChLfang 3: Danh gia Mo hinh nang cao 115

lddng lai diem so bien quan sat tren mot khoang tit 0 den 100, vdi 0 dai
dien cho hieu suat thap nhat va 100 dai dien cho hieu suat cao nhat. Vi
hau het cac nha nghien cdu deu quen thuoc vdi viec giai thich cac gia tri
p h an tram , loai thang do hieu suat nay rat de hieu. Viec do liidng lai cua
mot quan sat j ddi vdi bien quan sat i tien hanh thong qua cdng thde:

^rescaled £ [ * ;,] - m in[xj iqo

m ax f.rj - min[jr(]

trong dd x. la quan sat thd i trong m o hinh diidng dan PLS, E[.] bieu thi
diem sd thde te cua bien quan sat i cua ngiidi tra ldi j va min[.] va max[.j
the hien gia tri nhd nhat va ldn nhat cua bien quan sat. Dieu quan trong
can liiu f la cac gia trj nhd nhat va ldn nhat chi cac gia trj kha di tren
m ot thang do nhat dinh (vi du: 1 va 5 tren thang do tii 1 den 5) va khong
phai gia tri nhd nhat va ldn nhat cua cac phan hoi thde te khi lay mau (vi
du: 2 va 4) tren thang do td I den 5). Vi du, theo cdng thde nay, gia tri 4
tren thang do td 1 den 5 se trd thanh ( 4 - ^ / (5-1) • 100 = 75. Tat ca cac
diem dd lieu ddqc sd dung de d«c ldqng m o hin h dddng dan PLS ddqc
bien ddi lai theo each nay.
T d lieu 3.13 cho thay m ot doan trich cua d d lieu bien quan sat goc
(n = 300) ddqc sd dung de ddc ldqng m o hinh m au td T d lieu 3.10. Tat
ca cac bien quan sat dddc do lddng tren thang do td 1 den 5.

T d lieu 3.14 cho thay dd lieu bien quan sat td T d lieu 3.13, ddqc
thay doi lai th ang diem trong pham vi td 0 den 100, la dau vao cho viec
tin h to an diem so bien tiem an da thay doi lai. Ngoai ra, gia tri tru n g
b inh cua cac bien quan sat da thay ddi lai th an g diem the hien gia tri
hieu suat cda chung (vi du: 79 cho bien quan sat x va 77,5 bien quan
sat x 2), cac gia tri nay sau dd d d q c sd dung cho IPxMA d cap do bien
q u an sat.

Bait ke m o hinh do lddng cua bien tiem an la ket qua (tdc la V ) hay
nguyen nh an (tdc la T,. Y2 va K3), diem sd bien tiem an ddqc thay ddi
thang diem la mot sd ket hop tuyen tinh cua d d lieu bien quan sat da
thay ddi thang diem va trong sd ngoai ddqc thay ddi each tinh. Do do,
IPMA cd the ddqc thde hien cho cac khai niem nghien cdu ket qua va
nguyen nhan nhdng c h isd dung cac trong sd ngoai trong ca hai trddng
hop. Mac du IPMA cho phep sd dung ca hai loai m o hin h do lddng
(cung nhd cac khai niem nghien cdu don bien), no dac biet hdu ich
tro n g cac tinh huong khi bien nghien cdu m uc tieu noi sinh cd m b hinh
 116
Nhung v§n de nang cao trong PLS-SEM

Nguon: Dieu chinh tic Ringle & Sarstedt (2016).

Tif lieu 3.14 BlIfiffiniifraT
--------
• „
r V
Truong hop X2 *3 *4 As xi 9 *10 'lr ll A'l2 *.3 XI4 15 XI6

1 100 25 0 100 100 25* 75 50 50 50 75 100 0 25 75 50

2 75 100 25 50 75 0 0 25 100 75 100 50 50 100 75 50

Chirong 3: Banh gia Mo hinh nang cao

L. J
3 75 0 50 75 100 50 100 100 50 75 100 50 100 50 0
r
4 0 50 25 0 0 75 0 50 25 25 75 100 0 75 75

5 50 25 0 50 100 25 25 50 100 k o 25 50 0 50 75 100

/
... ... \ ... ...
•s.

299 25 75 50 75 75 25 25 0 75 50 O ) 75 25 100 75 50
'.■- V. (N
300 25 75 50 0 0 0 75 50 100 25 50 100 75 25 0 100
.
Gid tri ♦ / I
79,0 77,5 59,5 33,5 66,0 87,5 59,5 53,0 59,5 91,5 85,5 90,0 59,5 79,0 88,5 79,0
trung hinh . J

117
Nguon: Dieu chinh tic Ringle & Sarstedt (2016).
 118 Nhdng vbn de nang cao trong PLS-SEM

do lddng ket qua trong khi cac bien nghien cdu ngoai sinh co cac m o
hin h do ld o n g n g u y e n n fiahT T roriglinh 'hudng nay, cac cai tien hieu
suat cua cac bien quan sat trong cac mo hinh do lddng nguyen nhan
dan den sd gia tang hieu suat cua cac bien nghien cd u ngoai sinh, do do
chuyen thanh nang cao hieu suat cua bien nghien ciiu m uc tieu (xem
Hock va cong sd, 2010).

De co d d d c trong so da thay doi each tinh, trd b c het chung ta phai

tinh tro n g so chda chuan hoa bang each chia td n g trong so da chuan
hoa cho do lech chuan cua bien quan sat tdong dng. M ac dii tro n g so
ngoai da ch u an h o a bat nguon td ddc ldong m o hin h d d d n g dan PLS
chuan, viec ddc ldong do lech chuan cua tdng bien q u a n sat d d a tren dd
lieu bien q u an sat ban dau. Vi du, neu x, cd trong so chuan hoa la 0,2 va
do lech chuan la 1,619, k e t qua tro n g so chda chuan h o a la 0,164. T d lie u
3.15 cho thay cac trong so bien qu an sat (chuan hoa va chda chuan hoa)
cung vdi do lech chuan cua bien qu an sat lien q u a n den m o h in h m au
cu a ch u n g ta trong T d lieu 3.10.

C uoi cung, chung ta thay doi lai trong so ngoai chda chuan h o a de
tong cua ch u n g bang m ot cho m oi m o hinh do lddng. Vdi m uc dich nay,
chung ta can chia trong so chda chuan hoa cua td n g bien quan sat (vi
du: 0,1824 doi vdi bien quan sat X() cho tong tro n g sd chda chuan hoa
cua tat ca bien quan sat thuoc cung m ot mo hin h d o lddng. Ddi vdi y ,
tdng cua tat ca cac trong sd quan sat khong chuan h o a la 0,124 + 0,168 +
0,191 + 0,406 = 0,889. Do do, ddi vdi bien quan sat x , chung ta thu ddoc
trong sd ngoai chda chuan hoa va d d o c thay ddi lai la 0,399 sau khi chia
0,24 cho 0,889. Cot cuoi cung trong T d lieu 3.15 cho thay ket qua cua
trong so ngoai ddoc thay ddi each tinh.

Trong b d d c tiep theo, IPM A sd dung dd lieu b ien quan sat da thay
ddi each tin h (Td lieu 3.14) va trong sd ngoai d d o c thay ddi each tinh
(T d lieu 3.15) de tin h toan diem sd bien tiem an d d o c thay ddi bang
each ket hop tuyen tinh d o n gian. Vi du: diem d d lieu dau tien trong
vecto diem sd cua y, la:

100 . 0,139 + 25 . 0,189 + 0 * 0,215 +100 • 0,457 = 64,3

T d lieu 3.16 cho thay ket qua diem sd bien tie m an cim g vdi gia tri
tru n g binh cua chung. Trong vi du cua chung ta, Y co gia tri tru n g binh
(tdc la hieu suat) la 53,7, Y2 la 85,6, T3 la 61,8 va Y4 la 78,1. N h d n g ket
qua nay d o n g vai trd n h d dau vao cho m uc dan h gia h ieu suat trong bieu
do tam qu an trong - hieu suat.
 ChLfang 3: Danh gia Mo hinh nang cao 119

Tilth toan cac trong sd ngc»ai chiia chu an hoa va

Tii lieu 3.15 "
tinh lai cdc trong sd ngodi
Bien Bien Trong Do lech Trong sd Trong so
tiem quan sd ngoai chuan cua ngoai chda ngoai d d d c
an sat chuan hoa bien quan sat chuan hoa tinh lai

*» 0,2 1,619 0,124 0,139

*2 0,3 1,789 0,168 0,189

v
*3 0,4 2,099 0,191 A ©

*< 0,5 1,231 0,406 \ 0,457

*5 0,1 2,099 0 ,0 4 ft? ^ 0,114

*6 0,1 1,101 0,091 0,218

v
2
*7 0,4 3,357 0,119 0,285
0,4 ■3,160 0,383

*9 0,1 0,057 0,136

*10 0,1 0,057 0,137

V
*3
*11 0,4 )X ^ 2 7 0 0,176 0,422

*.2 w 2,653 0,151 0,361

*13 2,164 0,139 0,186

XU 0,3 1,874 0,160 0,215

y
o
o'
CO

1,413 0,212 0,286

0,3 1,291 0,232 0,313
-------- r

Tinh toan cac gia trj ve tam quan trong

Tam quan trong cua m ot bien nghien ciiu ve mat giai thich m ot
bien nghien ciiu lien ket trdc tiep hoac gian tiep k h ac (bie'n ng h ien ciiu
dich) tro n g mo h in h cau tru e co nguon goc tit tong tac dong cua m oi
quan he gida hai bien nghien ciiu nay. Tdng tac dong la tdng cua cac tac
dong triic tiep va tat ca cac tac dong gian tiep trong mo hinh cau true
(Hair va cong sii, 2017). Vi du, de xac dinh tdng tac dong cua T, len T,
(Tii lieu 3.10), chung ta phai xem xet tac d o n g triic tiep cua mdi quan
he gida hai bien nghien cdu (0,50) va ba tac dong gian tiep thong qua
Y2\ a Yy td o n g dng:
120 NhGng vPn de n§ng cao trong PLS-SEM

Y, -» Y2 Y4 = 0,50 . 0,50 = 0,25

Y, -> Y2 Y3 -> Y4 = 0,50 • 0,25 • 0,25 = 0,03125

Y, -> Y3 Y4 = 0,25 . 0,25 = 0,0625

Tu’ lieu 3 .1 6

Cac diem so bien tiem an diidc quy doi lai

r, ^3 - V
I 64,3 76,3 63,3 42,5
2 57,6 75,4 18, 67,9
3 55,5 82,0 76,5 45,1
4 14,8 47,0 26,9 63,5
5 34,5 37,7 43,3 63,5
...
299 62,7 62,4 22,9 63,3
300 28,4 69,4 47,1 50,6
Gid tri trung binh 53,1 85,6 61,8 78,1
Nguon: Dieu chinh tir Ringle & Sarstedt (2016).

Viec cong cac tac dong gian tiep rieng le se tao ra tong tac dong
gian tiep cua Y, len Y4khoang 0,34. Do do, tong tac dong cua Y len Y4 la
0,84 (= 0,50 + 0,34). Tac dong tong nay the hien tam quan trong cua Y
trong viec d d bao bien m uc tieu Y4. D ieu quan tro n g can liiu y la IPM A
dda tren cac tac dong chda chuan hoa de tao dieu kien cho m ot sd giai
thich trong trddng hpp cac yeu to khac khong doi (tdc la, khi m oi th d
khac la hang so) tac dong cua cac bien d d bao len b ien m uc tieu. C hinh
xac hon, bang each dda tren cac tac dong chda chuan hoa, chung ta co
the ket luan rang sd gia tang hieu suat cua m ot b ie n dd bao n h a t dinh
se lam tang hieu suat cua bien m uc tieu theo do ld n tong tac dpng chda
chuan hoa cua no.
T d lieu 3.17 tom tat tat ca cac tong tac dpng lien quan den bien m uc
tieu Y4 cua chung ta. Ldu y rang Y3 khong cd tac d o n g gian tiep len Y4.
Do dd, tong tac dong cua nd bang vdi tac dong trd c tiep la 0,25. Tai th d i
diem nay, sau khi tinh toan cac gia tri tam quan tro n g va hieu suat, tat
ca thong tin can thiet de ve bieu do tam quan tro n g - hieu suat san sang
de dung.
 ChLfO’ng 3: Danh gia Mo hinh nang cao 121

I Tdng tac dong, tac dong trtjfc tiep, tac d$ng gian
j ti^jp trong IPMA ; ,

Khai niem Tdc dong Tdc dong Tdng tdc

tdc dong triic tiep len Y4 gian tiep len Y4 ddng len Y4

0,50 0,34 0,84

0,50 0,06 0,56 ^

0,25 — A )
Nguon: Dieu chinh tu Ringle & Sarstedt (2016).
Ghi chu: Tat cd cdc tdc dong deu la tdc dong chUa chuan hoa.

B i/d c 4 : Tao bieu do Tam quan trong - Hieu suat

IPMA tap trung vao m ot bien nghien cdu m uc tieu quan tam trong
mo hinh diidng dan PLS. Do do, bddc dau tien trong viec tao ra m ot
bieu do tam quan trong - hieu suat doi hoi phai chon bien m uc tieu.
Trong vi du cua chung ta 6 Tii lieu 3.10, Y4 dai dien cho m ot bien m uc
tieu nhii vay. De tao bieu do tam quan trong - hieu suat cua Y4, chung
ta can sii dung gia tri tam quan trong va hieu suat cua cac bien tien de
tac dong cua Y, (tdc la Y , Y2 Y3). T d lie u 3.18 tom tat cac gia tri cua tam
quan trong va hieu suat cua bieu do nay, nhii da thu thap bdi cac bddc
IP M A trd d c do.

Sii phan tan thong tin diidc trin h bay trong Tii lieu 3.18 cho phep
chung ta tao ra mot bieu do tam quan trong - hieu suat nhii trong Tii
lieu 3.1.1. True x dai dien cho tam quan trong cua Yt, Y2 va Y3de dii bao
bien m u c tieu Y4, trong khi true y m o ta hieu suat cu a Y,, Y2 va Y} ve
diem so bien doi tiem an diidc thay doi do ldn trung binh cua chung.
De co djnh hiidng tot hdn, cac nha nghien cdu cung co the ve them hai
diidng trong bieu do tam quan trong - hieu suat; gia tri tam qu an trong
tru n g b inh (nghia la diidng th an g d u n g ) vagia trj hieu suat tru n g binh
(vi du, diidng ngang) cua cac bien nghien cdu diidc hien thi (Tii lieu
3.19). Doi vdi vi du cua chung ta, Yt, Y2 va Y}co tam quan trong tru n g
binh la 0,55 va hieu suat trung binh la 67,0 (Td lieu 3.18). Hai dddng
bo sung nay phan chia bieu do tam quan trong - hieu suat th an h bon
vung vdi cac gia tri tam quan trong va hieu suat dddi va tre n m dc tru n g
binh. Noi chung, khi phan tich bieu do tam quan trong - hieu suat, bien
nghien cdu d vung ben phai phia dddi (tdc la tam quan trong tre n tru n g
122 Nhung van d§ nang cao Irong PLS-SEM

binh va hieu suat dudi trung binh) th e hien ca hoi ldn nhat de nhan
sd cai thien. Tiep theo la phia tren ben phai, phia diidi ben trai va cuoi
cung, cac kh u vdc phia tren ben trai.

Dtf liiu cua bieu do Tam quan trong > Hi?u

suat doi vdi bien nghien cuU Ys
Tam quan trong H ieu suat

r. 0,84 53.7
0,56 85,6

^3 0,25 r 6 j,s
Gia tri tru n g binh 0,55 67,0

f r , . .

Bieu do Tim quan trong - iat cua bien

Tu" lieu 3.19
nghien cull muc tieu Y dS die j chinh

Nguon: Dieu chinh tic Ringle & Sarstedt (2016).

Qua do, bieu do tam quan trong - hieu suat cua b ien nghien cdu Y4cung
cap h udng dan cho viec liu tien cac hoat dong q u a n ly hoac chinh sach
cd cam qu an trong cao doi vdi khia canh phu th u o c m uc tieu da chon,
nh iin g yeu cau cai th ien hieu suat.
Trong vi du cua chung ta, bieu do tam quan tro n g - hieu suat cua
bien nghien cd u Y4(T ii lieu 3.19) cho thay Y( cd h ieu su at tiiong doi
 Chipong 3: Danh gia Mo hinh nang cao 123

thap la 53,7. So vdi cac bien nghien cdu khac, hieu suat cua Y la dddi
m dc trung binh. Mat khac, vdi tong tac dong la 0,84, tam quan trong
cua bien nghien cdu nay la dac biet cao. Do do, sd gia tang m ot ddn
vi ve hieu suat cua Y, td 53,7 len 54,7 se lam tang hieu suat cua Y4 len
0,84 diem td 78,10 len 78,94. Do dd, khi cac n h a quan ly n ham vao viec
tang hieu suat cua bien nghien cdu m uc tieu Y4, du tien hang dau cua
ho la cai thien hieu suat cua cac khia canh dddc nam bat bdi Y , vi bien
nghien cdu nay co tam quan trong cao nhat (tren tru n g binh), n h d n g
hieu suat td d n g doi thap (dddi m dc tru n g binh). Cac khia canh lien
qu an den bien nghien cdu Y2 va Y3 theo du tien th d hai va th d ba.

Bifdc 5: Md rong IPMA d cap do bien quan sat

IPM A khong gidi han d cap do khai niem nghien cdu. C hung ta
cung co the tien hanh m ot IPMA d cap bien quan sat de xac dinh cac
linh vdc cai tien cd lien q u an va cu the hdn. C h in h xac h d n , chung ta co
th e giai thich cac tro n g so ngoai d d d c thay doi each tinh khi tam quan
tro n g tddng doi cua m ot bien quan sat dddc so vdi tro n g so cua cac bien
qu an sat khac trong m ot m o h in h do lddng cu the.

Cac gia tri tam quan trong dddc lay tii tong tac dong cua bien quan
sat len bien nghidn cdu m uc tieu, la ket qua cua viec nhan cac trong so
ngoai dddc thay doi each tinh cua cac bien quan sat cua cac bien tac
dong vdi tong tac dong chda chuan hoa len bien nghien cdu m uc tieu.
Vi du, lien quan den cac bien quan sat cua Y(, chung ta se nhan cac
tro n g so ngoai d d d c thay doi each tin h cua x t tdi (tdc la 0,139, 0,189,
0,215,0,457; T d lieu 3.15) vdi tdng tac dong ch d a chuan hoa cua Y len
Y4 trong m d h inh cau true (tdc la 0,84). Phan tich nay cho ra cac gia tri
tam quan tro n g cua x l den x 4, td d n g dng la 0,117, 0,159,0,181 va 0,384.
Cac gia tri hieu suat dddc tinh td gia tri tru n g binh cua cac bien quan
sat cua dd lieu da dddc thay ddi each tinh (tdc la 79, 77,5, 59,5 va 33,5;
T d lieu 3.14). Vdi dd lieu nay cho tat ca cac bien quan sat cua Y,, Y2 va
Y , chung ta co the tao ra m ot bieu dd tam q u an trong - hieu suat n h d
tro n g T d lieu 3.20.

N hdng ket qua nay cho thay rang bien quan sat x 4 nen dddc du tien
cao nhat de cai thien, vi no co tam quan trong td d n g ddi cad n h at n h d n g
hieu suat th ap nhat. Tang m ot ddn vj ve hieu suat cua x4 lam tang hieu
suat cua Y4 th eo gia tri tam quan trong cua x , la 0,384 (trong trd d n g
h d p cac yeu td khac khong ddi). Cac bien quan sat x s, x }, x 7 va x 2 theo
th d td du tien thd hai den th d nam . Cac bien quan sat khac dddc trin h
124 Nhung v in de nang cao trong PLS-SEM

bay trong Tit lieu 3.20 it lien quan hdn doi vdi viec c&i th ien hieu suat
cua Y4. Thay vi tao m ot bieu do tam quan trong - hieu suat cho tat ca cac
bien quan sat (nhii trong T il lien 3.20), nd th d d n g hii'u ich h o n trong cac
dng dung de tao ra cac bieu do tam quan trqng - hieu suat rieng biet cho
cac bien quan sat cua m oi bien nghien ciiu (tile la, 3 bieu do tam quan
trong - hieu suat d cap do bien quan sat cho cac khai niem nghien ciiu
Ylt K2 va Y} trong vi du cua chung ta).

ghg - Hieu suat cua

Tu lieu 3.20 Y<, Y., va Y, Jen L
muc tieu Y .
ioo ■

90 A x lO * x12
A *x6
x ll
80 ■ xl

•3
lA 70
♦ X5
■ 2 (V
3
<Si-
o 60 A x9
if
50 •

40 -
V IS x4
30
0.0 0.2 0 .3 0 .4

Tam quan trong

Nguon: Dieu chinh tic Ringle & Sarstedt (2016).

Tii lie
ieuu ;3.2
5.21 tong ket cac quy tac kinh nghiem khi thiic hien IPMA.

TU lieu 3.21 Cac quy tic kinh nghiem cho phan tich IPMA

• C hi sii dung cac bien quan sat diioc do tren th an g do dinh liidng
hoac th ang do khoang each bang nhau (th iid n g la thang do thil tii
hoac cao hdn).

Tat ca m a hoa cua tiing bien quan sat phai c d cung hiidng ty le
(vi du, gia tri thap tren thang do phai dai d ie n cho ket qua tieu
ciic hoac thap va gia tri cao phai la ket qua tich ciic hoac cao. Neu
khong chung ta khong the ket luan rang gia tri tiem an cao h d n the
hien hieu suat tot hdn).
 Chucmg 3: Danh gia Mo hinh nang cao 125

• Bat ke mo h in h do lddng ddoc xac dinh la nguyen nhan hay ket

qua, cac ddc ldOng trong so ngoai phai co dau giong nhau (vi du:
dau diiong) de dam bao kha nang dien giai ket qua dau ra.

• Doi vdi kha nang dien giai ket qua dau ra, tong tac dong tro n g m o
h inh cau true se co cung dau (vi du: dau ddong).

• Chon m ot bien nghien cdu m uc tieu va chay IPMA nham vao

chinh no hoac tat ca cac bien nghien ciiu tac dong cua no trong mo
h inh cau true.

• Sd dung tong tac dong chda chuan hoa va diem so bien tiem an
trung b inh tren thang do ty le til 0 den 100 de tao bieu do tam quan
trong - hieu suat.

• Tap tru n g vao nhdng bien nghien cdu tac dong co hieu suat tiiong
doi thap va tam quan trong tilOng doi cao (tdc la, tong tac dong
ldn) len bien nghien cdu muc tieu. Viec giai thich cac ket qua trong
dieu kien cac yeu to khac khong doi cua cac ket qua gia dinh rang
viec tang hieu suat cua m ot bien nghien cdu tac dong len 1 diem
lam tang hieu suat cua bien nghien cdu m uc tieu ddqc ch o n bang
do ldn cua tong tac dong.

Minh hoa tinh hudng nghien ciiti - Phan tich bieu do Tam quan
trong - Hieu suat
Minh hoa cua chung ta ve IPM A diia tren vi du ve m o h in h danh
tieng cua cong ty. Til lieu 3.22 cho thay m o hin h va ket qua PLS-SEM
khi sd dung dd lieu thde nghiem va phan mem SmartPLS 3.

Cac ket qua td bootstrapping vdi 5000 m au bang each sd dung tuy
chon khong thay doi dau cho thay hau het cac he so diidng dan deu co
y nghla thong ke (p <0,01). Chi co cac moi quan he dddng d an CSOR
COMP, ATTR -» COMP va COMP -> CUSL la khong co y nghla. Bay
gid chung ta m d rong phan tich va chay IPM A bang each th d e hien
theo quy trinh ddoc mo ta trong T d lieu 3.12. Dau tien, chung ta kiem
tra cac yeu cau de thiic hien IPM A (Bddc 1). Sau khi xem xet bang cau
hoi, chung ta thay rang dd lieu quan sat da ddoc do lddng tren thang do
Likert 7 diem vdi cac m dc doi xdng can bang va trung lap (xem ChdOng
1). Cac bien quan sat diipc do tren thang do khoang each deu nhau. Hon
nda, doi vdi tat ca cac bien, gia tri cao h o n the hien ket qua tot hon (xem
126 NhOng van de ndng cao trong PLS-SEM
 Chuong 3: Danh gia Mo hinh nang cao 127

C hiidng 1). Do do, chung ta khong can dao ngddc thang do bat ky bien
quan sat nao. De co them thong tin ve dd lieu, nhap dup vao Corporate
reputation data [344 records] trong Project Explorer (Td lieu 3.23).
Che do xem duf lieu m d ra (Td lieu 3.24), cung cap them thong tin ve
tap d d lieu (vi du: diem d an h dau gia tri khuyet) va m ot so thong ke
m o ta.Tiep theo, chung ta kiem tra cac d au cu a he so tai ngoai. Sau
khi chay thuat toan PLS-SEM, xem Results Report (Td lieu 3.25) trong
SmartPLS, no cung hien thi trong so cua tat ca cac bien quan sat. Tat c£
cac trong so ngoai deu dddng, ngoai trd qual_4, co trong so la -0,005.
Tuy nhien, trong so bien quan sat nay cdc ky gan bang 0 va khong co y
nghia, do do tac dong cua trong so am den viec thay doi thang do lddng
la cdc ky han che. Do do, chung ta gid lai bien quan sat nay va tiep tuc
phan tich.

T u lieu 3 .2 3 SiTiartPlS Project Expl

Project Explorer

f, n Corporate Reputation
a li**1 Corporate Reputation - A d v a n ce d PLS-SEM B o o k
Extended m odel
H C M (reflective-form ative)
H C M (reflective-reflective)
^ Inadm issible H C M (ref lect'rve-formative)
^ S im ple m odel
^ Corporate reputation data case w ise de le tion [336 records]
fjj^ C o rp o ra te reputation data [344 record s]

N eu trong so (am) ldn hdn (vi du: -0,10), chung ta se phai xda bien
quan sat nay d e phan tich IPMA. Tuy nhien, m ot bddc n h d vay phai
phu h dp vdi ly thuyet dd lddng trong dd cac bien quan sat cdn lai van
bad tru m day du pham vi khai niem nghien cdu (xem H air va cdng sd,
2017, C hddng 5).

Sau dd chung ta chay IPMA bang each nh ap vao C alculate ->

Importance-Performance Map Analysis (IPM A) trong th an h trinh
don. Ngoai ra, ban co the nhap chuot trai vao bieu tddng banh xe
Calculate tren thanh cong cu va chon tu y chon tddng dng trong hop
com bo m d ra. Trong hop thoai sau day (Td lieu 3.26), xac dinh bien
nghien cdu m uc tieu CUSL va chon tuy chon AH Predecessors o f the
Selected Target Construct. Quan trong nhat, chung ta can xac dinh gia
128 Nhung v in <Je nang cao trong PLS-SEM

tri nho n h at va ldn n h a t cua mdi bien quan sat can thiet cho viec thay
doi tinh toan cua-duUieu-thanh^thang diem tii (Lden 100. N hii TU.lieu
3.26, SmartPLS td dong doc cac gia tri nho n h at va ldn nhat nay tii d d
lieu. Tuy nhien, n e u ngUdi tra ldi khong sd dung toan b o thang do (vi
du: gia tri n h d nhat th d e te la 2 thay vi 1), SmartPLS khong the tinh lai
dd lieu nh d ban m ong m uon. Neu trd d n g hop nay xay ra, hay thay doi
gia tri nhd nhat theo each thu cdng bang each nhap vao o tdong dng
trong cot M in hoac M ax va nhap gia tri nhd nhat hoac ldn n hat cda
thang do. N goai ra, ban cd the dong th d i xac d in h gia tri nhd n h a t va
ld n nhat cu a tat ca cac b ien quan sat. De lam n h d vay, nh ap cac gia tri
tdong dng ben canh Min: /Max: d cuoi hop thoai va nhap vao Apply to
All. Trong vi du cua chung ta, tat ca ngddi tra ldi da sd dung day dd cac
thang do bien quan sat n h d ddoc chi ra tro n g cdt M in va M ax cua che
do xem d d lieu SmartPLS (T d lieu 3.24). Do dd, chung ta duy tri cai dat
mac dinh (tdc la: so dd trong sd dddng dan) va tiep tuc bang each nhap
vao Start Calculation.

■ in iB im w i s
C o rp o ra te rep u tatio n data.txt 23 ! '

Delimiter Semicolon Encoding: UTF-8

Value Quote Character None Sample size: 344
Number Format: USIagU&Q£2e!3) Indicators: 41
Missing Value Marker ;9£ Missing Values: 11

M ed ia n Min Max S tandard O cviaticn E xceti Kurtcsis Sfcevtnesf

2X00 1.000 4000 1.003 •0513 0.747

2.000 1X100 2X100 0.482 .1.684 •0571

S jOOO 1000 7000 1.433 •0324 •0 264

6.030 1.000 7X100 1375 0.616 0166

6.000 1.000 7.000 1.458 -0 188 •0677

5000 1000 7000 1347 -0J99 -0AQ5

4.000 1000 7.000 1348 0501 -0312

5X100 1000 7000 1371 -0.941 •0325

5.000 1.000 7000 1313 0268 -0.792

6.000 1.000 7000 1.744 0.040 0351

6.000 1.000 7000 1355 0930 -1301

6X)00 1X100 7.000 1.174 0.777 -0.768

4.000 1.000 7000 1469 -0 383 -0.042

3X100 1000 7000 1.651 0 .5 6 4 0.497

4000 1000 7.000 1.478 -0.468 0.061

3.000 1000 7.000 1462 0 .4 6 3 0197

 Chuang 3: Danh gia Mo hinh nang cao 129

Tu”lieu 3.25 Bao cao ket qua SmartPLS 'v tj

Outer Weights
eotip m i rm m I * SPEWS? m m
wm m m 0.414
attr_2 0.201
0.658
cotnp_1>§ft*& 0.469
0.365 -A .
0.372
csor_1 0.306
m m m m 0.037 ♦
m &m m 0.406
csor_4 0.080
csor 5 0.416
c u s a tiM ^ 1.000
cusf_1 0.369
cusL2 ; 0.420
cust.3 m 0.365
Hke_1 m 0.419
llke_2 0.374
Hke_3 0.363
perf_4;lf i 0.468
perf_2^V ' 0.177
perf_3;; 0.194
perf_4 0.340
perf_5 0.199
qual_1 0.202
quat 2 0.041
<?ual_3 0.106
quaL4 -0.005
qual_5 0.160
quat_6 0.398
qual_7 0.229
qual_8 0.190

Gid day, SmartPLS tu dong tinh gia tri hieu suat va tam quan trong
(Btidc 2 va 3) va tao ra bieu do tam quan trong - hieu suat (Bddc 4). Sau
khi hoan th an h cac tinh toan, SmartPLS md bao cao ket qua. Ban dau,
no cho thay ket qua cua cac he so du’dng dan chuan hoa. Vao Q uality
Criteria Importance-Performance Map [CUSL] (constructs,
unstandardized effects), bao cao bao gom bieu do tam quan trong -
130 Nhu>ng van de nang cao trong PLS-SEM

hieu suat nhii diipc hien thi trong Tii lieu 3.27. C ac bieu dien do hoa
"cua tam quan tfphg-T neiT suat"siJdung tongT acclpngchiia chuan hoa,
m a ban co th e truy cap bang each vao Final Results -» Total Effects
va nhap vao the C onstructs, unstandardized. Vao Final Results
Performance/Index, ban co the tru y cap cac gia tri hieu suat da thay
doi each tin h cua cac bien tiem an va bien bieu thi (tdc la bien quan sat)
bang each nhap vao cac the LV Performances va M V Performances.

lg co gi ngac nhien, chung ta thay rang hai trong ba bien tac

lie tiep, sii hai long cua khach hang (C U SA) va sii yeu thich
co tam quan trpng dac biet cao doi vdi CUSL (Tii lieu 3.27).
Juan trpng hd n, chat liipng cam nhan (QUAL) co tam quan trpng ldn
nhat doi vdi CUSL so vdi ba khai niem ngoai sinh khac la A TTR , CSOR
va PERF. Do do, cac hoat dong quan ly hoac chinh sach liu tien cai thien
CUSA va LIKE, cd the dat diipc bang each tap tru n g vao bien ngoai sinh
QUAL. Viec giai quyet QUAL dac biet hiia hen vi bien nghien cdu nay
cd m ot m o h in h do liidng nguyen nhan, trpng so bien quan sat cua no
cho biet khia c p ih nao la quan trpng nhat de cai th ien chat liipng cam
nhan cua ngiidi tra ldi.
 Chirong 3: Danh gia Mo hinh nang cao 131

Dieu quan trong can lPu y la bieu dien do hoa cua cac ket qua IPM A
khac voi m in h hoa ket qua PLS-SEM tro n g SmartPLS. Thay vi hien thi
gia tri R2 cua bien tiem an noi sinh trong m o h in h diidng dan PLS (Tii
lieu 3.22), ket qua IPM A cho thay gia tri hieu suat cua m oi bien tiem a n
(Tii lieu 3.28); Thay vi hien thi h e so tai hoac trong so ngoai chuan hoa
(T ii lieu 3.22), ket qua IPM A cho thay trpng so ngoai khong chuan h o a
va diipc thay ddi each tinh cua cac m o hin h do liidng, b a t ke chung cd
dang nguyen nhan hay ket qua (TP lieu 3.28).

De co them thong tin cu the ve each tang hieu suat cua cac bien
nghien cPu, dac biet la cua QUAL, cac phan tich sau day tap trung vao
cap do bien qu an sat (Bpdc 5). LPu y m ot lan nPa rang mot IPM A d cap
do bien quan sat la cd the bat ke thong so ky th u at m o h in h do lP dng
cua bien tac dong. Tuy nhien, m ot phan tich b ien quan sat lien quan
dac biet h P u ich trong cac thiet lap m o h in h do lpdng nguyen nhan.
Khi di den Quality Criteria -> Im portance-Perform ance Map [CUSL]
(indicators, unstandardized effects) trong bao cao ket qua, SmartPLS
cho thay bieu do tam quan trong - hieu suat cua cac bien quan sat nhP
d p p c hien thi trong TP lieu 3.29. Tuy nhien, viec hien thi tat ca cac bien
q u an sat nay khong cu n g cap hPdng dan hP u ich de xac d inh cac net dac
trPng can dppc giai quyet de cai thien long tru n g th an h cua khach hang.
NhOng v&n d e n§ng ca o trong PLS-SEM
 Chirong 3: Danh gia Mo hinh nang cao 133

Vi ly do nay, chung ta chuyen sd chu y cua chung ta ve cac bien nghien

cdu dddc do liidng theo nguyen nhan trong m o hinh diidng dan PLS vi
chung dai dien cho cac net dac triing tiidng doi cu the co th e dddc giai
quyet de cai thien hieu suat. Trong sd cac bien nghien cdu nay, QUAL
ro rang co tam quan trong ldn n hat tro n g khi cho thay hieu suat tru n g
binh. Do dd, ch ung ta tap trung vao IPM A cua cac bien quan sat cua
QU AL

Trong bao cao ket qua tai Q uality Criteria Importance-

Performance Map [CUSL] (indicators, unstandardized effects),
SmartPLS cho thay cac tong tac dong chda chuan h o a cua bien quan
sat trong the Indicator Total Effects for [CUSL] va cac gia tri hieu suat
trong the Indicator Performances for [CUSL]. T d lidu 3.30 cho thay
cac ket qua nay ddi vdi cac bien quan sat cua QUAL, cho phep tao ra
bieu dd tam q u an trong - hieu suat n h d trong T d lieu 3.31. IPM A cho
thay rang bien quan sat qual_6 (tdc la: [Cdng ty] la ddi tac dang tin cay
cho khach hang), cd tam quan trong dac biet cao trong viec cai thien
long tru n g th an h cua khach hang. T d d n g td, qua\J7 (tdc la: [Cdng ty] la
m ot cdng ty dang tin cay) cd tam quan trong td o n g ddi cao n h d n g cung
cho thay hieu suat dac biet thap. N h d n g ket qua nay cho thay rang cac
cdng ty nen n ham vao viec nh an m anh do tin cay va sd dang tin cay cua
ho khi nham den m uc tieu nang cao long tru n g th an h cua khach hang.

Bieu
TU lieu 3.29

tmportance-Performance Map

&
<& CUSL to

0.00 0.05 0.10 0.15 0.20 02f 0.30 035 0-40 o.45 0.50 0.55 0.50
Total Effects

<s aftrj # *<tr_2 aftr_3 com pj comp_2 c<wp_3 csorj ►c so r_2 a c to r 3 « esc* 4 « cw t 5
• 4 • §*..3 r perlj • P*f_2 perfJJ petf_4 < per(_S qy^j cu# 2
A q u a i j ♦ quaM • q u aij r tfx ijg * t. qual.8
134 Nhung v in de n3ng cao trong PLS-SEM

Cac ket qua cua IPM A cung cap nen tang thiic nghiem cho m ot
c u o c lh a o Iu lii t6t Ron ve each cai thien hieu suat cua m 6t bien m uc
tieu. Dac biet, IPM A xac dinh cac bien quan sat cu the co tam quan
trpng cao n h iin g hieu suat tiiong doi thap, sau do trd th an h cac khu vdc
trpng tam de quan ly cac hoat dong nham dat diipc cac cai thien hieu
suat. Cac phan tich dinh tinh va din h ldpng tiep theo co the bao gom
cac yeu to quan trpng bo sung n h ii chi phi va thdi gian, va ket qua la, co
the doi hoi phai ilu tien sda doi cac hoat dong. VI du: m ot khu vdc co
m dc do du tien cao ve cai thien hieu suat, nhii IPM A tiet 16, co the roi
vao hamg liu tien thap hon, neu viec cai thien hieu suat cua no rat ton
kern va can nhieu thPi gian. Theo ket qua cac phan tich nay, bam co diipc
bdc tran h toan dien ve cac han h dong hieu qua cho cac cd hoi cai thien
hieu suat tong the cho bien nghien c d u quan tam ch in h cua ban (vi du:
long tru n g th anh cua khach hang).

M M nali<
|i r ^ f K T M c T « dpi Vi ca c b ie n q u a n sat cu a b ie n n g h ie n ctiu

Bien quan sat cua

T&m quan trong Hieu suat
QUAL
qual_ 1 0,051 67,539

qual_2 0,010 56,202

q u a l_ 3 r ^ ^ 0,025 68,023

qual_4 -0,001 56,880

qual_5 0,039 66,860

qual_6 0,091 65,407

qual_7 0,052 56,638

qual_8 0,047 63,953

Gia tri trung binh 0,039 62,688

 Chifcmg 3: Danh gia Mo hinh nang cao 135

in trong - hieu suat cho cac 111

sm

70 •

68 ' ♦ qual 8
< q u a l / qUaL1
66 '
♦ qual_6
64 • ♦ qual_8
■**
- g 62 •
3 60 •
4).

* 58<
56 -
qual_4
♦ qual_2 ♦ qual 7 o °
54 ■

52 -
AS"
0.00 0.05 0.10
T im q u a n t r o n y i /

TOM TAT
> v
Biet each danh lining thde cua m ot m o hinh do liidng
vdi phan tich ang djnh trong PLS-SEM (CTA-PLS).
CTA - PLS each tiep can de danh gia thiic nghiem m o
h inh do liidng cua bien tiem an (tdc la, nguyen nh an hoac ket
qua). Phiidng phap nay yeu cau, theo ly tiidng, co it nhat bon
bien quan sat cho moi m o h in h do lddng. Trong trd d n g hop cua
m ot m o hinh do lddng ket qua, tat ca cac bo td khong th d a cua
n o la 0 va do do triet tieu. Tuy nhien, n eu it n hat m ot tro n g so
cac bo td tetrad khong th d a khac biet d an g ke so vdi 0, ta nen
xem xet viec bac bo m o hinh do lddng ket qua va thay vao do,
gia dinh m ot dac diem do lddng nguyen nhan. Khi dan h gia
phiiong thde cua m pt m o hinh do lddng, ban nen chu yeu d d a
vao cac can nhac ve m at ly thuyet va khai niem cung vdi cac
bang chdng thUc nghiem do CTA-PLS cung cap.
• Hieu each chay CTA-PLS trong SmartPLS va each dien giai
ket qua. Viec thde thi CTA-PLS bang each sd dung phan m em
SmartPLS yeu cau chon m ot so ldong ldn cac m au bootstrap
(thiidng la 5000). Dtia tren ket qua bootstrapping, SmartPLS tra
ve cac khoang tin cay ddoc dieu chinh sai chech va diioc dieu
ch in h theo B onferroni cho cac bo td k hong th d a cua tat ca cac
136 Nhung van de nang cao trong PLS-SEM

mo hin h do lddng v6i bon bien quan sat hoac nhieu hon. D da
“tfeim hurtg~k£t: qu&Trayrban cd~the“quyet dinh lieu mpt~bp~td
tetrad k hong th d a co y nghia thong ke hay khong (tdc la, lieu
no co triet tieu hay khong). Qua do, SmartPLS cung cap m ot
phdd n g tien d d n gian kiem dinh thdc nghiem phddng thdc cua
mo h inh do lddng.
• Hieu ro gia tri cua phan tich bieu do tam quan trong - hieu
suat (IPMA). Muc dich cua IPM A la mof rong bao cao ket qua
PLS-SEM tieu chuan ve ddc lddng he so d d d n g dan bang each
them mot tham so nham xem xet gia tri tru n g bin h cua cac diem
so bien tiem an. C hinh xac hon, IPM A td o n g phan vdi tong tac
dong, dai dien cho tam quan trpng cua bien tdc dong trong viec
d in h h in h m ot bien nghien cdu m uc tieu n h a t dinh, vdi diem so
bien tiem an tru n g binh cua chung bieu thi hieu suat cua chung.
Muc tieu la xac din h n hdng bien tac d o n g cd tam quan trpng
td d n g doi cao doi vdi bien nghien cdu m uc tieu (tdc la: n hdng
bien cd tong tac dong ldn) n h d n g cd h ieu suat tddng doi thap
(tdc la: diem so bien tiem an trung bin h thap).
• Hieu each thdc hien IPMA b&ng phan m em SmartPLS. Trddc
khi chay m ot IPM A trong SmartPLS, cac nha nghien cdu can
kiem tra m ot so yeu cau, lien quan den th an g do bien quan sat
va trpng so bien quan sat thu ddpc td p h an tich PLS-SEM tieu
chuan. IPM A n h d vay yeu cau chon m ot bien m uc tieu va xac
dinh cac gia tri nhd nhat va ldn nhat cua cac thang do bien quan
sat. SmartPLS tao ra cac bieu dd tam quan trp n g - hieu suat theo
cap do bien ng h ien cdu cung n h d cap do b ie n q u an sat.

I. Bo td (tetrad) la gi va ban se tinh nd n h d th e nao?

ia CTA-PLS la gi?

3. Lam the nao de ban quyet dinh lieu m d t md hinh do lddng

thuoc dang'do lddng ket qua hay nguyen n h a n dda tren cac ket
qua CTA-PLS?
4. Muc dich cua IPM A la gi?

5. Khi nhin vao bieu dd tam quan trpng - hieu suat, viec giai thich ket
qua trong trddng hpp cac yeu td khac k hong doi co nghia la gi?
 Chuong 3: Danh gia Mo hinh nang cao 137

CAU HOI TO DUY PHAN BIEN

1. Tai sao mo hinh do liidng thieu chinh xac la mot van de quan trong?

2. Sd dung tieu chi quyet dinh dinh tinh cua Jarvis va cong sii
(2003) de thiet ke mot each tiep can do liidng ket qua va nguyen
nh an de do liidng sii hai long cua khach hang vo'i n h a cung cap
djch vu dien thoai di dpng cua ho.

3. N eu cac ket qua CTA-PLS khong khdp vdi dac diem m o h in h

do liidng da gia dinh, lam the nao de cuoi cung ban quyet d in h
dan g do liidng nao se sd dung?

4. Lam th e nao ban co the rut ra mot each cd he th d n g cac ham y

thiic tien td cac ket qua IPM A? Khi nao va lam the nao ban se
bao gom cac ldp bien quan sat?

5. Dat ten cho m ot phiiong dien quan tam bo sung (th d ba) cua
m ot IPMA ma ban se xem xet. Tai sao va lam the nao nhiing can
nhac bo sung nay cd th£ thay d 6i liu tien cac h o a t dong triidc
dd cua ban?

THUAT NCLf o V
Alpha inflation: Phong dai Alpha
Bias-corrected and Bonferroni Khoang tin cay dieu chinh theo
- adjusted confidence Bonferronni va hieu chinh sai
intervals chech
Confirmatory tetrad analysis in Phan tich bo td khang dinh trong
PLS-SEM (CTA-PLS) PLS-SEM (CTA - PLS)
Direct effect Tac dong trd c tiep
Equidistant scale Thang do khoang each deu
Forced-choice scale T hang do bat buoc Ida chpn
Formative measurement model Mo hinh do lddng nguyen nhan
Impact-performance map Bieu do tac dong - hieu suat
Importance Tam quan trong
Importance-performance map Phan tich bieu do tam quan trong
analysis (IPMA) - hieu suat
Importance-performance matrix M a tran tam quan trong - hieu suat
Indirect effect Tac dong gian tiep
IPMA Phan tich IPMA
138 Nhung v in <J6 n§ng cao trong PLS-SEM

M easurement m odel Xac djnh sai m o h in h do liidng

m isspecification _ ______
Metric scale Thang do din h ldOng
M ultiple testing problem Van de da kiem dinh
N om in al scale Thang do dan h nghia
Nonredundant tetrads Cac bo td tetrad khong thda
Ordinal scale Thang do th d td / th d bac
Performance H ieu suat
Priority map analysis Phan tich bieu do d u tien
Reflective m easurem ent m odel Mo hin h do lddng ket q u a ^ ^
Rescaling Thang do quy doi lai
Tetrad (x) Bo td .V U

Total effect Tac dong tong

Vanishing tetrad * &
Bo td triet tieu

TAI LIEU DOC THEM

JO
Bollen, K. A., & Ting, K.-F. (2000). A tetrad test for causal indicators.
Psychological Methods, 5, 3-22.

Gudergan, S. P., Ringle, C. M., Wende, S., 8cWill, A. (2008). C onfirm atory
tetrad analysis in PLS path m odeling. Journal o f Business Research,
6 1 ,1238-1249.
Hock, C., Ringle, & M ., 8c Sarstedt, M. (2010). M anagem ent of
m ultipurpose stadium s: Im portance and perform ance m easurem ent
o f service interfaces. International Journal o f Services Technology
and M anagement, 14,188-207.

Mikulic, J., Prebezac, D., & Dabic, M. (2016). Im portance-perform ance

analysis: C om m on m isuse o f a popular technique. International
Journal o f M arket Research, 58, 775-778.

Ringle, C. M ., 8c Sarstedt, M. (2016). G ain more insight from your PLS-

SEM results: The im portance-perform ance m ap analysis. Industrial
M anagement & Data Systems, 116, 1865-1886.
Schloderer, M. P., Sarstedt, M„ 8c Ringle, C. M. (2014). The relevance
o f reputation in the nonprofit sector: The m oderating effect
o f sociodem ographic characteristics. International Journal o f
Nonprofit and Voluntary Sector Marketing, 19,110-126.
 CHL/ONG 4

Md hinh hoa
khong dong nhat
quan sat dUdc

1.. Hieu ro tam quan trong cCia yjj

nh^t trong PLS-SEM.
2. Gi&i tfakK si/ khac bi§t gida t | | l
. di/gc'WI(h6ng quan sat ducte
1 Pli|u khcii nlfm bat bi£n mo hh
r\o trong PLS-SEM.
■’ ‘?«L ' ‘ ^
4. Hieu phan tich da nhom va thu

TONG QUAN CHUONG

Cac moi quan he trong cac mo h inh diidng dan PLS ngu y rang cac bien
tiem an ngoai sinh giai thich cac bien tiem an noi sinh m a khong co bat
ky 4nh h d d n g he thong nao cua cac bien khac. Tuy nhien, trong nhieu
triidng hop. gia dinh nay khong ton tai. Vi du: ngiidi tra ldi co the khong
dong nhat trong cam nhan va danh gia bien tielm an, tao ra sii khac biet
dang ke trong he so diidng dan tren hai hoac nhieu nhom ngiidi tra ldi
(vi du: nhom nd so vdi nhom nam ). N han thay rang cac cau true dut
lieu khong dong nhat thddng co mat, cac nha nghien cdu ngay cang
quan tam den viec xac dinh va hieu nhdng khac biet n h d vay. Trong
thdc te, viec khong xem xet tinh khong dong nhat co the la moi de dpa
tdi gia tri cua cac ket qua mo hinh phdong trinh cau true binh phddng
toi thieu rieng phan (PLS-SEM) (Becker, Rai, Ringle, & Volckner, 2013;
Hair, Sarstedt, Ringle, & Mena, 2012).
139
140 Nhdng van d i nang cao trong PLS-SEM

C hung ta tim hieu ve cac khai niem khac nhau cho phep cac nha
nghien cdu m o h in h hoa d d lieu k h o n g dong n hat nhd m ot giai phap.
C hd d n g nay dau tien cung cap m ot cai nhin tong quan ve tinh khong
dong nhat quan sat ddpc va khong quan sat dddc, cho thay viec bo qua
cac cau true d d lieu khong dong nhat co the gay ra cac ket qua sai lech
n h d the nao. Tiep theo, chung ta thao luan ve sd bat bien do lddng, day
la m oi quan tam hang dau trddc khi so sanh cac nhom d d lieu. B ing
each thiet lap sd bat bii’n do lddng, cac nha nghien cd u co the td tin rang
sd khac biet nhom trong cac ket qua ddc ldpng m o hin h khong den td
sd khac biet ve noi dung hoac y nghla cua cac bien tiem an trong cac
nh o m cung n h d khong td thang do. B e danh gia sp bat bien do lddng
tro n g boi canh PLS-SEM, cac nha nghien cdu co the sd d ung thu tuc bat
bien do lddng cua m o hinh tdng hpp (M ICOM ). Tiep theo, chung toi
gidi thieu cac dang phan tich da nh o m khac n h au ddpc sd dung de so
sanh cac tham so (thddng la cac he sd dddng d an ) gida hai hoac n hieu
nh o m d d lieu. P han tich da nhom PLS-SEM th d d n g ddpc ap dung khi
cac nha nghien cdu m uon kham pha n h d n g khac biet cd the ddpc tru y
nguyen trd lai cac dac diem quan sat n h d gidi tin h hoac quoc gia xuat
xd. Trong tin h h u ong nay, gia dinh rang cd m o t bien dieu tiet p h an loai
(vi du: gidi tinh) anh h ddng d en cac m oi quan he trong m o hinh d ddng
dan PLS. Tuy nhien, trong khi d d lieu phan loai ch o phep de dang xac
dinh dac diem cua nhom , cac dd lieu khac cung cd the cung cap m ot ca
sd de xac dinh cac nhom . D o dd, m uc dich cua p h a n tich da nhom la tiet
16 tac dong cua bien phan loai nay. C hddng nay ket thuc bang m ot m inh
hoa ve danh gia tinh bat bien mo hin h do lddng b&ng each sd dung thu
tuc M ICOM va phan tich da nhom PLS.

TIN H KHONG DONG NHAT QUAN SAT DLfOC VA KHONG

DONG NHAT KHONG QUAN SAT DUOC
C ac d n g dung cua PLS-SEM thddng phan tich to a n bo dd lieu, ngu y
gia thiet rang d d lieu goc td mot tong the dong nhat. Gia dinh cac dac
diem d d lieu td d ng doi dong nhat nay thddng k h o n g thde te. Cac ca
n h an (vi du: trong hanh vi cua ho), cac cong ty (vi du: trong cau true
cua ho) hoac m oi trd d n g (vi du: sd bien dong) th d d n g khac nhau va viec
tong hdp dd lieu gida cac quan sat co kha nang ta o ra ket qua gay hieu
nham . Viec khong xem xet tinh khong dong nhat n h d vay co the la m ot
m oi de doa tdi gia tri cua cac ket qua PLS-SEM, vi no co the dan den
ket luan khong chinh xac (Sarstedt & Ringle, 2010). Becker, Rai, Ringle
 Chuong 4: Mo hinh hoa khong dong nhdt quan sat dugc 141

va Volckner (2013) giai quyet van de nay m ot each chi tiet va cung cap
cac vi du ve cac tinh huong co he so diidng dan diiOng va am khac nhau
dang ke, cho thay gia tri khong y nghla gain bang 0 d cap dC( lieu tong
hop. Ket luan rang khong co m oi quan he nao to n tai gida cac khai niem
nghien cdu se khong co gia trj va gay hieu nham.

Mo h inh diiqc the hien trong T d lieu 4.1, trong do sd hai long cua
khach hang vdi san pham ( YJ phu thuoc vao hai nhan td - sd hai long
vdi chat ldpng ( 7j) va sd hai long vdi gia ( Y2) - m inh hoa cac van de xuat
phat td that bai trong xd ly khong ddng nhat bdi canh cua PLS-SEM. Gia
sd cd hai phan nhom cd kich thddc m au tdong td nhau. N hom 1 bao
gom khach hang nam , trong khi N hom 2 bao gom khach hang nd. Ca
hai nhom deu khac nhau ve y th d e gia cua ho n h d d d p c chi ra bdi cac
he sd d d dng dan td n g phan n h o m cu the khac n h a u /Q h in h xac hon,
anh hddng cua chat ldpng cam nhan (Kj) den sd hai long cua khach
h a n g (L p la m an h hon nhieu trong sd nam gidi ( p f 1= 0,50; chi sd trong
ngoac do n bieu thi nhom ) so vdi nd gidi ( p ^ - 0,10). N gdpc lai, gia ca
cam nhan (72) cd a n h hddng m anh h o n d e n s d hai long cua khach hang
(T3) trong sd n d gidi (p2U) = 0,35) so vdi nam gidi (p2(l) = 0,25). Trong vi
du nay, tin h khong ddng nhat phan anh ket qua nhom khach hang (nd)
cd sd nhay cam ve gia nhieu hon n h d n g nhay cam ve chat ldpng kem
hon, doi lap vdi nhom kia (nam gidi). Td quan diem ky thuat, cd m ot
bien dieu tiet phan loai gender chia tap dd lieu thanh hai nhom khach
hang va do dd yeu cau ddc ldpng hai m d hinh rieng biet, n h d dupe chi
ra trong T d lieu 4.1. Q uan trong hon, neu chiing ta khong nhan ra sd
khong ddng nhat gida cac nhom va phan tich m d hinh bang each sd
dung toan bo dd lieu, cac ddc ldpng he sd dddng dan c u n g cap mot
bdc tranh khong day du ve cac m di quan he md hinh. Do la, ca hai ddc
ldpng se bang khoang 0,30 khi sd dung toan bo dd iieu, do dd dan cac
nha nghien cdu ket luan rang gia ca va chat ldpng la quan trong n h d
nhau doi vdi nam va nd, mac d u khong phai vay. TdPng td n h d vay, sd
khac biet nhom lien quan trong cac ddc ldpng m d hinh cd the huy bo
lan nhau, tao ra cac anh hddng khong dang ke khi phan tich dd lieu d
cap tdng hpp (vi du, xem Sarstedt, Schwaiger, & Ringle, 2009). Do dd,
dieu quan trong la phai xac dinh, danh gia va, neu cd, xd ly tin h khong
ddng nhat trong dd lieu.
Sd khong ddng nhat trong d d lieu co the quan sat ddpc hoac khong
quan sat ddpc. Khi sd khac biet gida hai hoac nhieu nhom dd lieu lien
quan den cac dac diem co the quan sat ddpc, ch ln g han n h d gicfi tinh,
142 Nhdng vdn d§ nang cao trong PLS-SEM

tuoi tac hoac quoc gia xuat xd, diiofc goi la khong dong nhat quan sat
diiqc (observed heterogeneity). Cac n h a nghien ciiu co the sd dung cac
dac diem co th e quan sat nay de phan chia d d lieu th an h cac nhom quan
sat rieng biet va thde hien phan tich PLS-SEM nh o m cu the, nhii m inh
hoa trong Tii lieu 4.1 lien quan den gidi tin h cua khach hang. Ngdqc lai,
tin h khong dong nhat khong quan sat diiqc (unobserved heterogeneity)
ngu y rang sd khac biet gida hai hoac nhieu nhom d d lieu khong xuat
hien m ot du tien td m ot dac tin h cd the quan sat cu the hoac ket hop
cua m ot so dac tin h n h d n g trd nen ro rang trong n h d n g khac biet trong
he so dddng dan cau true.

= 0,25

Nflr (50% dii lieu)

pi = 0,30

Trong no ldc phat hien tinh khong dong nhat khdng quan sat ddpc
lien quan den bien noi sinh va ngoai sinh, cac n h a nghien cdu da sd
dung th d d n g xuyen cac ky thuat phan tich cum (cluster analysis), chang
 Chiming 4: M6 hinh hoa khong d6ng n h lt quan sat dime 143

han n h d k-means (Hair, Black, Babin, & A nderson, 2010; Sarstedt &
M ooi, 2014) ve dd lieu chi bao hoac diem so bien tiem an bat nguon td
phan tich trddc do cua toan bo tap dd lieu. Phan vung ma phan tich nay
tao ra sau do d d p c sd dung lam dau vao cho cac ddc ldpng PLS-SEM
theo nhom cu the. Cach tiep can nay de ap dung n h d n g thieu sot ve m at
khai niem bdi vi cac ky thuat phan cum truyen thong bo qua cac m oi
quan he m o hinh dddng dan m a cac n h a nghien c d u da xac dinh trddc
khi phan tich. N hdng chinh xac la n hdng m oi quan he nay co kha
chiu trach nhiem cho m ot so khac biet nhom . Do do, khong cd
ngac n h ien khi nghien cdu trddc day da chi ra rang cac
phan cum truyen thong thdc hien rat kern trong viec
biet nhom trong PLS-SEM (vi du: Sarstedt & Ringle, 2010).

N han thay nhdng han che cua phddng phap tiep can tuan td, nghien
cdu phddng phap luan trong PLS-SEM da de xuat m ot so phddng phap
cu the de xac dinh va xd ly khong dong nhat khong quan sat dddc, thddng
dddc goi la ky thuat ldp tiem an (latent class techniques). N hdng ky thuat
nay co truyen thong lau dcfi trong nghien cdu SEM (CB-SEM) dda tren
hiep phddng sai (vi du, Jedidi, Jagpal, & DeSarbo, 1997; Masyn, 2013;
M uthen, 1989) va da dddc chung m inh rat hdu ich trong viec xac dinh
tinh khong dong nhat khong quan sat ddpc va phan vung dd lieu mot
each phu hop. Phan vung ket qua sau do co the ddpc phan tich cho n hdng
khac biet co y nghia bang each sd dung cac phddng phap phan tich da
nhom . Ngoai ra, cac ky thuat ldp tiem an co the xac dinh rang tinh khong
dong nhat khong quan sat ddpc khong anh hddng den ket qua, hd trd
phan tich mot mo hinh duy nhat dda tren dd lieu cap tong hdp.

Cac ky thuat khac nhau de xac dinh khong dong nhat khong quan
sat ddpc trong PLS-SEM bao gom cac thuat toan tddng dong (Ringle,
Sarstedt, & Schlittgen, 2014; Ringle, Sarstedt, Schlittgen, & Taylor,
2013), binh phddng toi thieu trpng so (Schlittgen, Ringle, Sarstedt, &
Becker), 2016) va mot so phddng phap tiep can khac (vi du: Becker, Rai,
Ringle, & Volckner, 2013; Esposito Vinzi, Trinchera, Squillacciotti, &
Tenenhaus, 2008). Xem them Hair, Sarstedt, M atthew s va Ringle (2016).
Trong C hdong 5, chung toi gidi thieu hai thu tuc ldp tiem an, binh
phddng toi thieu rieng phan hdn hpp hdu han (FIM IX-PLS) va phan
doan dinh hddng dd bao trong PLS-SEM (PLS-POS), cho phep xac
dinh va xd ly tinh khong dong nhat khong quan sat ddpc trong mo h in h
d d d n g d in PLS.
144 Nhflng v&n de nang cao trong PLS-SEM

KIEM Dj N H BAT BIEN m 6 HINH DO Ll/CJNC

Mot m oi quan tam chinh tra d e khi so sanh cac ddc laq n g tham so
nhom cu the ddi vdi cac khac biet cd y nghla bang each sfl dung phan
tich da n h o m la dam bao bat bien do lad n g (m easurem ent invariance),
cung con ddpc goi la tactng d aong do ladng (measurem ent equivalence).
Bang each thiet lap sp bat bien do lddng, cac n h a nghien cflu cd the tp
tin rang sii khac biet nhom trong a d c laqng m o hinh khong phai la ket
qua tfl noi dung va / hoac y nghla rieng biet cfla cac bien tiem an gifla
cac nhom . Cac bien doi trong m di quan he cau tru e gifla cac bien tiem
an cd the xuat phat tfl cac y nghla khac nhau m a cac nh o m tra ldi quy
cho cac hien tap n g dang d a p c do ladng, th a y vi sp khac biet thpc sp
trong cac m di qu an he cau true. Ly do cho sp k h ac biet n h a vay co the
xuat phat tP, vi du, (a) ng ad i tra ldi bao gom cac gia tri van hoa khac
nhau, ngadi giai th ich m ot bien phap nhat d in h theo each th d e khac
nh au ve khai niem ; (b) gidi tinh, dan toe hoac cac dac diem ca nhan
khac doi hoi phai tra ldi cac tai lieu chinh thde th e o cac each khac nhau
m ot each cd he thong; va (c) nhP ng ngadi tra ld i sP dung cac tuy chon
cd san tren th ang diem khac nhau (vi du: xu h d d n g chon hoac khong
chon cac m Pc do /th ai cpc danh gia). Hult va cong sp (2008, tr. 1028)
m o ta nhflng mdi quan tam nay va ket luan ra n g “that bai trong viec
thiet lap tao n g daong dfl lieu la nguon tiem nan g cua sai sd do lad n g ”
(tflc la, sp khac biet gifla nhflng gi dapc dp d jn h do d ap c va thpc sp do
dapc). Khi sp bat bien do lad n g khong co mat, n o cd the lam giam sflc
m an h cua cac kiem dinh th o n g ke, anh hddng d en do ch in h xac cua cac
bo ddc ldpng va cung cap cac ket qua sai lech. T dm lai, khi sp bat bien do
la d n g k hong d a p c bieu thi, bat ky ket luan n ao v e m di quan he m d hin h
>van de. Do dd, cac so sanh da nhom yeu cau thiet lap sp bat bien
tng de dam bao tinh gia tri cua cac ket q u a va ket luan.

ac nha nghien cdu da de xuat nhieu phdcfng phap khac nhau de

danh gia sp bat bien do lddng ddi vdi CB-SEM. P han tich nhan td k h ln g
d jnh da nhom dpa tren cac h d d n g dan cua S teenkam p va B aum gartner
(1998) va Vandenberg va Lance (2000) la p h d o n g phap pho bien nhat
de danh gia bat bien. Tuy nhien, cac ky thuat do lddng bat thddng ddpc
thiet lap tot de danh gia cac m o hinh nhan td chung (com mon factor
models) cua CB-SEM va cac phan m d rong co lien quan den cac m d
h in h do lddng nguyen nhan (D iam antopoulos & Papadopoulos, 2010)
 Chircmg 4: Mo hinh hoa khong dong nhat quan sat du-pc 145

khong the de dang chuyen sang cac m o hinh tong hop cua PLS-SEM.
Vi ly do nay, Henseler, Ringle va Sarstedt (2016) da phat trien thu tuc
bat bien do lddng cua md hinh hon h op (MICOM). Thu tuc M ICOM
ddoc xay d d n g dda tren diem so cua cac bien tiem an. Trong PLS-SEM,
cac bien tiem an nay duoc bieu dien la cac hop td (ghep lai), tdc la sd ket
hop tuyen tinh cua cac bien quan sat va trong so bien quan sat ddoc ddc
ldqng bdi thuat toan PLS-SEM. D o do, trong m o ta thu tuc M ICOM ,
chung ta noi ve sd tong hop khi de cap den cac thde the (diem so) ma
thuat toan PLS-SEM sd dung de bieu dien cac bien tiem an ddpc xac
djnh bdi nha nghien cdu.

Thu tuc M ICOM bao gom ba bddc: (1) b it bien cau hinh, (2) bat bien
thanh phan, va (3) tdong dddng tri trung binh tong hop va p h d d n g sai. Ba
bddc cd lien quan theo cap bac, nhd ddpc hien thi trong Td lieu 4.2. Dieu
nay cd nghia la sd bat bien cau hinh la dieu kidn tien quyet cho sd bait bien
thanh phan, m ot lan nda la dieu kien tien quyet de danh gia m ot each cd
y nghia sd tdong dddng tri trung binh tong hop va phdong sai.

Neu sd bat bien cau hinh (Bddc 1) va sd bat bien thanh phan (Bddc
2) ddoc thiet lap, sd bat bien do lddng tiin g phan (partial measurement
invariance) dddc xac nhan. Khi bat bien do lddng tdng phan ddpc xac
nh an cho tat c i cac bien tiem an trong m o hinh dddng d in PLS, cac
nha nghien cdu cd the so sanh cac he so dddng dan bang cong cu phan
tich da nhom . Neu sp bat bien do lddng tdng ph an ddpc thiet lap va,
ngoai ra, cac th an h phan tong hpp cd gia tri tru n g binh va phdong
sai bang nhau tren cac nhom , sd bat bien do lddng toan phan (full
measurement invariance) ddpc xac nhan. Td goc do m o h in h do lddng,
cac n ha nghien cdu cd the gop dd lieu cua cac nhom khac nhau, neu
chung den td cac nhom rieng biet ban dau va ddpc hddng loi td sd gia
tang ve sdc m anh thong ke. Tuy nhien, sd bat bien do lddng toan phan
khong ngu y rang khong cd sd khac biet trong cac m d hinh cau true (tdc
la he sd d d d n g d in ). Sau nay can ddpc kiem dinh bang p h d o n g phap
phan tich da nhom . Chi khi phan tich da nhom chi ra rang cac m d hinh
cau tru e cung la bat bien (tdc la, khong cd sd khac biet dang ke trong he
sd dddng dan), cac nha nghien cdu cd the gop d d lieu va tap tru n g vao
phan tich cap do tong the. Trong cac phan sau, chung ta se thao luan chi
tiet hon td n g bddc (doi vdi dng d ung M ICOM ve dd lieu td nam quoc
gia, xem Schlagel & Sarstedt, 2016).
146 NhCmg v§n cl& nang cao trong PLS-SEM
 Chiwng 4: Mo hinh hoa khong dong n h it quan sat du’o'c 147

Budc 1: Bat bien cau hinh

Bddc 1 giai quyet viec thiet lap sd bat bien cau hinh de dam bao rang
moi bie’n tiem an trong m o hinh diidng d in PLS ddpc xac dinh la nhd
nhau cho tat ca cac nhom . Bien thien cau hinh ton tai khi cac khai niem
nghien cdu dupe tham so hoa va ddc ldpng bang nhau gida cac nhom .
Mot danh gia dinh tinh ban dau ve dac diem cua bien tiem an tren tat ca
cac nhom phai dam bao rang ba yeu cau sau day ddpc dap dng:
Bien quan sdt dong nhat tren mo hinh do liidng. M oi m o hi:
lddng phai sd dung td d n g td so bien quan sat va thang do gida cac
nhom . Kiem tra chinh xac xem cac bie'n quan sat tddng td ap dung
cho tat ca cac nhom co ve kha d o n gian hay khong. Tuy nhien, khi tien
hanh khao sat bang cac ngon ngd khac nhau, viec ap dung cac thde tiln
nghien cdu th d e nghiem tot (vi du: dich thuat va dich ngdpc lai) la dieu
quan trong nhat de thiet lap sd tddng dddng bien quan sat. Trong boi
canh nay, m ot danh gia ve m at va / hoac gia tri chuyen gia co the giup
xac m inh lieu cac nha nghien cdu co sd dung bo bien quan sat n h d nhau
tren cac nhom hay khong.
X d ly d d lieu ddng nhat. Viec xd ly d d lieu bien quan sat phai giong
het nh au tren tat ca cac nhom , bao gom m a hoa (vi du: m a hoa bien gia),
m a hoa dao ngdpc va cac dang khac cua viec ma hoa lai cung n h d xd ly
dd lieu (vi du: xd ly gia tri khuyen hoac chuan hoa). Cac gia tri ngoai lai
can ddpc phat hien va xd ly tddng td.

Thiet lap thuat toan dong nhat hoac tieu chi toi du hoa. Phddng
sai dda tren phddng phap ddc ldpng m o hinh n h d PLS gom co nhieu
bien the vdi cac ham m uc tieu va cac thiet lap thuat toan khac nhau
(vi dy, Ida chpn trong so tai ngoai ban dau va so do tro n g so m o hinh
ben trong; Hair, Ringle, & Sarstedt, 2011; Henseler, Ringle, & Sinkovics,
2009). Cac nha nghien cdu phai dam bao rang sd khac biet trong ddc
ldpng mo hinh nhom cu the khong phai la ket qua cua cac thiet lap
thuat to an khac nhau.

Sd bat bien cau hinh la dieu kien can thiet n hdng chda du de rut
ra cac ket luan cd gia tri td cac phan tich da nhom . Cac nha nghien cdu
cung phai dam bao rang sd khac biet trong he so dddng d in khong phai
la ket qua td sd khac biet trong each th d e bien tiem an ddpc hin h thanh
tren cac nhom . Bddc tiep theo, sd bat bien thanh phan, tap trung vao
khia canh nay.
148 NhOng van <J§ nang cao trong PLS-SEM

Butic 2: Bat bien thanh phan

Su bat bien thanh phan ton tai khi diem so to n g hop la giong nhau
gifla cac nhom , m ac du co the co sd khac biet ve trong so nh o m cu the
diipc sd dung de tinh toan diem so. Bddc 2 cua thu tuc M IC O M ap
dung m ot kiem dinh thong ke de danh gia lieu cac diem so tong hop
co khac nhau dang ke gifla cac nhom hay khong. Vdi m uc dich nay, thu
tuc M IC O M kiem tra c, la tflong quan gifla diem so tong hop Y(,)va Y^':

c = cor (Y05, Y*25)

Sd bat bien thanh phan yeu cau c bang 1. Ve m at ky thuat, thu tuc
kiem dinh gia thuyet k hong (H 0) rang c la 1. De th iet lap sfl bat bien
th an h phan, chung ta khong dflpc bac bo gia thuyet H 0 nay. Tflc la, neu
kiem dinh m ang lai gia tri p ldn hon 0,05 (trong trfldng hop chung ta
gia d in h m dc y nghia la 5%), chung ta cd the gia d in h sfl bat bien thanh
phan. Ngflflc lai, neu chung ta bac bo gia thuyet k h o n g (H 0), chung ta
khong the thiet lap bat bien th an h phan cho khai niem nghien cdu cu
the dang dflpc xem xet. Lflu y rang kiem din h nay k h ong hoat dong doi
vdi khai niem do n bien bdi vi m oi quan he ben ng o ai (don) cua chung
luon luon la 1. Do dd, diem so bien tiem an Y05 v a Y<2) cua k hai niem
d o n bien giong het nhau m ot gia trj c bang I.

De kiem dinh gia thuyet, th u tuc M ICOM d d a tren khai niem hoan
vi (Fisher, 1935). Tflong tfl nhfl bootstrapping, cac phep kiem dinh
hoan vi tao ra m ot phan phoi tham chieu tfl dfl lieu thde te. Tuy nhien,
thay vi lay m au cac quan sat tfl tap dfl lieu ban dau vd i sfl thay the - vi no
la trfldng hp p trong bootstrapping - kiem dinh h o a n vi ngau nhien trao
doi cac quan sat gifla cac n h o m nhieu lan. Kiem d in h hoan vi thddng
dflpc sfl dung trong vo so boi canh vi no cung cap p h d d n g phap tiep can
hidu q u a de kiem din h phi tham so, cung nhfl k h i cd m au n h d (vi du:
Ernst, 2004; G ood, 2000).
D anh gia bat bien thanh ph an trong M IC O M tuan theo phddng
ph ap 5 bddc, tro n g dd Tfl lieu 4.3 m in h hoa:
1. l/d c ldpng nhom cu the cu a cac m d hin h d iid n g dan PLS de thu
dflpc diem so tong hpp cfla nhom 1 vdi nm quan sat va nhom 2
vdi h (2) quan sat. Lflu y rang trong vi du c u a hai nhom tong cfla
n(1) va n(2) bang n, la tong so quan sat trong toan bo tap dfl lieu.

2. Tinh toan cac tddng quan c gifla diem sd t6 n g h p p cua nhom 1

va 2.
150 Nhdng van de nang cao trong PLS-SEM

3. H oan vi ngau nh ien cua dd lieu, co nghia la cac quan sat ddqc
gan ngau nhien cho c£c nhom . Theo quan diem chinh thdc, V 1'
qu an sat ddqc ru t ngau nhien ma khong can thay the tit tap
hop dii lieu tong hop va gan cho nhom 1. So quan sat con lai
«(y diiqc gan cho nhom 2. Cac nha nghien cdu nen sd dung toi
thieu 1000 hoan vi.

4. Doi vdi m oi phep h o an vi chay tin h toan u cu a td o n g q u an cu

gida diem so tong hop cua nhom 1 va n h o m 2.

5. Doi vdi kiem dinh gia thuyet, quy trin h sap xep cac ket qua
tdong quan cua cac hoan vi cu theo th d td giam dan. Gia tri
ngddng la 95% (hoac trong vi du cua chung ta, so th d td 950
cua 1000 hoan vi trong dan h sach dddc sap xep) va gia tri tdong
quan tdong dng. N eu c nhd h o n gia tri do, n o nam ngoai hoan vi
95% dda tren khoang tin cay va khac biet cd y nghia so vdi 1 (d
m dc p < 0,05), dieu nay doi hoi sd bac bo gia thuyet khong (H 0)
rang c bang 1. Do do, diem so tong h o p la khac biet cd y nghia,
va sd bat bien th an h p h an chda ddqc thiet lap. Trong tinh huong
ngdOc lai, neu c khong rOi vao duoi 5%, n h d n g vao khoang tin
cay 95%, diem so tong hop khong khac b iet cd y nghia, dieu nay
chdng m inh sd bat bien thanh phan. D da tre n ket qua hoan vi,
nd cung cd the tra ve gia tri p cua c. Trong phan tich nay, gia tri
p dai dien ty le ph an tram cua cac ket qud cu tdong quan hoan
vi cd gia tri thap h o n c. M ot gia tri p tre n 0,05 cho thay rang
c khong khac biet cd y nghia so vdi 1, cd nghia la sd bat bien
th anh ph an d a dddc thiet lap.

Khi sd bat bien cau hinh ton tai va sd bat b ien thanh phan ddqc
thiet lap cho tat c& cac bien tiem an trong m o h in h d ddng dan PLS, cd
sd bat bien do ld dng td n g phan. Trong trddng h o p nay, cac n h a nghien
cdu cd the so sanh cac he sd d d dng dan bang p h d d n g phap phan tich da
nhom . Neu bat bien do lddng tdng phan khong th e thiet lap cho bien
tiem an, thi cac so sanh nhom cu the bang each s d dung phan tich da
nhom tren bat ky m oi quan he nao lien quan d e n bien tiem an nay la
khong kha thi. Trong trd d n g hop nay, cac nha n g h ien cdu cd the tranh
chay phan tich da nhom c£ thay va phan tich td n g nhom rieng biet m a
khong co gang so sanh ket qua doi vdi hai nhom . N goai ra, ngddi ta cd
the loai bo cac bien tiem an khong dat ddqc sd b a t bien th an h phan, ly
thuyet ddqc cung cap ho trd bddc nay. Quan tro n g hon, khi thdc hien
bat ky thay doi nao doi vdi m o hinh, M ICOM can phai ddqc chay lai.
 Chifcng 4: Mo hinh hoa khong dong nhat quan sbt dugc 151

Bddc 3: Su1can being gitfa tri trung binh tong hop va phddng sai

Neu ket qua cua Bddc 2 ho trd phep bat bien do liidng, viec danh gia
p h ai tiep tuc vdi d a n h gia can bang ve cac gia tri tru n g b in h tong h o p va
cac phiidng sai.

Bddc cuoi cung cua thu tuc M ICOM dau tien yeu cau ddc liidng
m o h in h dii on g d in PLS bang each sii dung dut lieu gop (tdc la tong
hdp). Sau do chung ta kiem tra xem cac gia tri tru n g bin h va phiidng
sai gida cac diem so tong hop cua nhom dau tien va diem so tong h d p
cua nhom th d hai co khac nhau ve trung binh va phddng sai cua chung
hay khong. Liiu y rang, khac vdi Biidc 2, chung ta khong sd dung diem
so tong hdp diidc tinh bang tru n g binh rieng biet, n h d n g thay vi lidc
lddng PLS-SEM nhom cu the, chung ta sd dung toan bo tap dd lieu va
gan diem cho mdi nhom sau do. De phan tich sd td d n g d d d n g cua cac
gia tri tru n g b inh, gia thuyet khong la:

O day, “pooled/gop chung” chi ra rang diem so tdng hdp cua m ot nhom
nhat dinh bat nguon td viec phan tich dd lieu gop chung.

Phan tich sd td d n g dddng cua cac phddng sai yeu cau xac din h
logarit cua ty sb phddng sai cua diem sd to n g h d p cua ca hai nhom . N eu
logarit cua ty le nay m an g tinh thong k ek h o n g k h a cO , chung ta k e t luan
rang cac phd d n g sai la bang nhau gida cac nhom . Gia thuyet khong (H 0)
td d n g dng la:

H0 : log(var( Y ^ led) / var(Y™ed)) =

lo g (v a r()^ ,J) - lo g (v a r(r^ w )) = 0

Viec kiem dinh hai gia thuyet nay tuan theo each tiep can td d n g
td nh d tro n g Bddc 2. MICOM hoan doi (nghla la sap xep lai) th an h
vien n h o m quan sat nhieu lan va tao ra sd phan phoi th d e nghiem ve sd
khac biet ve gia tri tru n g binh va logarit cua p h d d n g sai. Sd bat bien d o
ld d n g toan p h an dd d c thiet lap khi khong cd sd khac biet cd y nghla ve
gia tri tru n g binh va (logarit) phddng sai gida cac nhom . Day la trd b n g
h d p neu h o an vi dda tren khoang tin cay cua sd khac biet ve gia tri
tru n g binh va (logarit) phddng sai bao gom sd k hac biet ban dau ve gia
152 Nhflng van de ndrig cao trong PLS-SEM

tri tru n g binh va phiiong sai nhfl ddc ldpng m o hin h ban dau (tdc la,
khong co hoan vi); Ngdpc lairn e u e flsd k h a c -b ie t co y nghia, chung ta
phai thiia nhan rang sii bat bien do liidng toan phan khong the dflpc
thiet lap.

Tom lai, so sanh cac m oi quan he m o h in h theo nhom cu the doi

vdi cac khac biet cd y nghia bang each sii dung p h an tich da nhom yeu
cau thiet lap bat bien cau h in h (Biidc 1) va bat bien th an h phan (Biidc
2). N eu hai biidc nay ho trp bat bien do liidng, thi sii bat bien do liidng
tdng p h an diiOc thiet lap. Vi trong tam thiidng la ve cac m oi quan he
m o h in h cau true, chung ta cd the chay phan tich da nhom ve cac he so
d iid n g dan bang each sd dung m ot trong cac phiidng phap diipc m o ta
tro n g cac ph an tiep theo. Tuy nhien, m ot ngoai le tro n g vSn de nay la so
san h cac so hang tiiong tac, diipc tao ra tro n g qua trin h phan tich bien
dieu tiet (xem H air va cong sii, 2017). Mo h in h d o liidng cua so hang
tiio n g tac dai dien cho m ot phep do phu trp ket h p p cac moi lien he gida
bien dieu tiet va khai niem nghien cdu ngoai sinh tro n g m o h in h diidng
dan. Tuy n h ien , dac tinh nay dien ta bat ky danh gia m o h in h do liidng
nao ve so hang tiiong tac va so sanh cd lien quan gifla cac n h o m khac
nhau ve dan h gia bat bien la vo nghla. Q uan tro n g h o n , n eu bat bien do
liidng tdng phan khong the diipc thiet lap cho m ot bien tiem an, so sanh
n h o m cu the bang each sfl dung phan tich da n h o m ve bat ky m di quan
he lien qu an d en bien tiem an nay la khong kha thi.

N hac lai, cac n h a nghien cdu cd the phan tich cac nh o m m ot each
rieng biet, neu bat bien do liidng td n g phan khong bieu thi, khong diia
ra bat ky ket luan nao ve cac tac dong lien q u an d en n h o m hoac thay
ddi m d h in h bang each loai bo cac bien tiem an k h o n g dat dflpc sfl bat
bien th a n h p h a n . C uoi cung, neu bat bien do liid n g tiin g p h a n dflpc xac
n h a n va cac h p p tfl cd gia tri trung b in h va p h iio n g sai bang nhau gifla
cac nhom , thi bat bien do liidng toan p h an dflpc xac nhan, ho trp them
p h an tich dfl lieu gop, khi dflpc xem xet ap dung.

Phan tich da nhom

He sd diidng dan dflpc tao ra tfl cac m au khac nhau hau nhfl luon
khac nhau ve sd ldpng, nhflng cau hoi dat ra la lieu sfl khac biet cd y
nghla thong ke hay khong. Phan tich da nhom giu p tra ldi cau hoi nay.
Ve m at ky thu at, phan tich da nh o m kiem dinh gia th u y et H 0 rang he so
diidng dan gifla hai nhom (vi du, p0) trong nhom 1 v a p (2) trong nhom 2)
 ChLfong 4: Mo hinh hoa khdng ddng nhat quan sat dime 153

khong khac biet dang ke (vi du, p (l) - p (2)), dong nghia vdi sii khac biet
tuyet doi gifla cac he so diidng dan la 0 (tflc la, H 0: |p(l) - p U)| = 0). Gia
thuyet thay the tiiong flng H, la h£ so diidng dan khac nhau (tflc la, H ():
p (n ^ p(2) n ( ^ j c ^ c j 1 ]< j 1 a C > |p d ) _ p « | > q )

Nghien cflu da de xuat m ot sd phiiong phap tiep can phan tich da

nhom dfloc m inh h o a trong Tii lieu 4.4 (Sarstedt, Henseler, & Ringle,
2011). Khi so sanh hai nhom dfl lieu, cac nha nghien cflu can phai phan
biet gifla kiem dinh tham so va mot sd phiiong phap tiep can phi tham sd.

Tfl lieu 4.4 Cac each thflc phar iom tront

C ic each thirc.p n
trong P I
tftf

Hai nhbm
$ N hieu hon
hai nh6m

Kiem dinh Kiem dinh phi Kiem dinh Om nibus ve

tham so tham $6 | | su khac biet nhom (OTG)
I____
PLS-MGA Hoan vi

Kiem djnh tham so

Kiem djnh tham sd (Keil va cong sii, 2000) la phiidng phap dau tien
dfloc ap dung trong cac nghien cflu PLS-SEM va da dfloc ap dung rong
rat do de thflc hien. Cach tiep can nay la m ot phien ban sfla doi cua
hai kiem d inh mau doc lap tieu chuan f-test, dfla tren cac sai sd chuan
bat nguon tfl thu tuc bootstrapping. Cflng nhfl kiem dinh tieu chuan
f-test, phiiong phap tham sd cd hai phien ban (Sarstedt & Mooi, 2014),
tuy thuoc vao viec cac phflOng sai tdng the co the dfloc gia dinh la bang
nhau (hom oscedastic) hay khong bang nhau (heteroscedastic). Dang
thfl hai co the dfloc kiem djnh bang phiiong phap Levene’s test nhii
dfloc m d ta tro n g vi du Sarstedt va Mooi (2014).

Neu phiiong sai bang nhau, phep kiem djnh thong ke (tflc la gia tri
f thflc nghiem ) dfloc tinh nhfl sau:
154 Nhdng van de nang cao trong PLS-SEM

t= P (,)- P (2)
i_ u p - i l . (I))2+ (« (2> - d 2 _ . s e ( p m f _ n ~ r _ 1_
V (*"> + n (2) - 2) W J + („ (1) + * (2) - 2) ) V « (,) « ( 2 )

Trong cong th d e nay, p {[) (pi2)) m o ta he so d ddng dan dddc so sanh

trong n h o m 1 (nhom 2), trong do n(,) (n(2)) la so quan sat trong nhom
1 (nhom 2). Cu6i cung, se(pw ) (se(p{2))) m o ta cac sai so chuan cua cac
ddc ld d n g th am so cua nhom 1 (n h o m 2), co the th u diidc thong qua
bootstrapping. De loai bo gia thuyet H fl cac he so d d d n g dan bang nhau,
gia tri t th d e nghiem phai ldn hdn gia tri tdi han tii phan phoi t vdi
rt(i> + n <D_ 2 bac tu do.

N gddc lai, neu cac phdd n g sai khong bang nhau, cac nha nghien
cdu phai sd dung m ot phien ban sda doi cua kiem djn h t W elch-
Satterthwaite (Satterthwaite, 1946; Welch, 1947), thong ke kiem djnh cd
d a n g n h d sau:
,0 )
t =
(n (l) - 1 ) -1 ) (2)\2
se(p(C ,(2) se(pw )

Thong ke kiem d jnh nay cung tiem can phan phoi t nhdng vdi bac td
do (df) sau day:

^ -se(p"'f + S !£ ^ ± > -se (p'«f

n
•-2

Nghien cdu trUdc day cho thay p h d d n g phap tham so kha td do va cd

the m ac sai lam Loai I (Sarstedt, Henseler, & Ringle, 2011). H dn nda, td
quan diem khai niem , p h d d n g phap tham so la kho hieu (m d ho) vi no
dda tren cac gia dinh phan phoi, khong phu hdp vdi tin h chat phi tham
so cda PLS-SEM. Trong boi canh nay, cac nha nghien cdu d a de xuat
mpt so thay the phi tham so de phan tich da nhom.
 Chucmg 4: Mo hinh hoa khong dong nhit quan sat du’o'c 155

KIEM D IN H PHI THAM SO

PLS-MGA
Henseler va cong sii (2009) da de xuat p h ddng phap phi tham so
PLS-MGA xay d d n g tren ket qua bootstrapping cua td n g nhom dii lieu.
Doi vdi m ot m oi quan he cu the tro n g m o hinh d iid n g dan PI.S, each
tiep can cua chung so sanh tiing ddc ld d n g bootstrap cua m ot nhom vdi
tat ca cac tide ld d n g bootstrap khac cua cung m ot tham so trong nhom
khac. Bang each dem so lan xuat hien trong dd tide ld d n g bootstrap cua
nh o m th d n h at ldn h a n n h o m th d hai, p h d d n g phap tiep can xuat phat
td gia tri p vdi kiem dinh 1 duoi (one-tailed). Cach tiep can nay cd the
dddc m o ta dddi dang chinh thde n h d sau:

B j j

Doi vdi cac tham so tong the thde (30) va (3(2) cua m di nhom dd lieu,
phddng phap nay xac djnh xac suat cd dieu kien p(p<0 ^ p (2) I P(l) < f t 2))
m a ddc lddng p (1) cua he so dddng dan nhat dinh trong nhom 1 ldn
h d n hoac bang he so p (2) d d d c ddc lddng cua no tro n g nh o m 2, theo dd
p h an tren trong dau ngoac ddn danh dau nhom tddng dng. Bang each
sd dung PLS-MGA, nha nghien cd u m uon dam bao rang xac suat cd
dieu kien nay thap hdn m dc « dddc chi dinh trd d c khi ket luan rang
Pl>) ldn hdn /?(2). Trong phddng trinh tren, B la so lan chay bootstrap,
p (P (s J lL ..', B) vh p '-p ( j = 1,..., B) 1& c*c k£t qua bootstrap nhom
cu the cua he so dddng dan, p w va p [2) bieu thi gia tri trung binh cua
cac m au bootstrap nhom cu the, va 0 (tdc la theta trong chd Hy Lap)
la ham bddc ddn vi, cd gia tri la 1 neu doi so vddt qua 0, khac 0. Ldu y
rang theo Henseler va cong su (2009), tat ca cac ket qua bootstrap n hom
cu the can phai dddc hieu chinh cho ddng bdi sd khac biet gida ddc
lddng tham so nhom cu the ban d au va gia tri tru n g b in h tren cac gia tri
bootstrap nhom cu the.

De m inh hoa nguyen ly lam viec cua PLS-MGA, hay xem xet m ot
m o h in h dd n gian vdi hai khai niem nghien cd u (bien), d d d c ddc lddng
trong hai nh o m cho he so dddng dan p (1) = 0,336 va p(2) = 0,501. C hdng
ta m uon kiem djnh gia thuyet rang he so dddng dan trong nhom dau
 156 Nhdng van <Je nang cao trong PLS-SEM

tien ldn ho n so vdi nhom th d hai. De kiem d in h gia thuyet nay, chung
ta rut ra 10 m au bootstrap cho p (n va p <2} va do i chieu cac ddc ldong
nhii trong TO lieu 4.5. C ot dau tien the hien 10 m au bootstrap cua he
so diidng dan trong n h o m dau tien, trong khi hang dau tien bieu dien
10 m au bootstrap trong nhom th d hai. Bay gid chung ta so sanh tdng
lidc ldong bootstrap cua p (1) (vi du, tide ldOng cua m au bootstrap dau
tien 0,357) vdi mdi ddc ldOng bootstrap cua p (2) (tdc la 0,494, 0,423,...,
0,538) va dem so IdOng tru d n g hop p(,) ^ Pu> ddoc chi ra bdi dau X trong
T d lieu 4.5. C hia sd nay (tdc la, 11) cho tdng sd lan so sanh (tdc la 100)
m ang lai gia tri p, la 0,11 trong trd d n g hop cua chung ta. Vi vay, chung
ta k h o n g the ket luan rang he so diidng dan p tro n g nhom dau tien la ldn
hon dang ke so vdi nhom th d hai.

PLS-MGA lien quan den m ot sd liiong ldn so sanh cac ddc ldong
bootstrap. Doi vdi phan tich ban dau ve sU khac biet nhom , chung toi
de xuat rut ra 500 m au bootstrap, sd dung cai d at thuat toan (vi du:
tuy chon khong thay doi dau) n h d ddoc de xuat trong H air va cong sd
(2017). Doi vdi phan tich cuoi cung, sd dung 5000 m lu bootstrap, ket
qua la cd 25.000.000 so sanh cho m oi tham sd.

Cuoi cung, dieu quan trong la nhan m anh ra n g each tiep can PLS-
M GA chi cho phep kiem dinh cac gia thuyet th e o m ot phia. C u the,
phan m em SmartPLS luon kiem dinh gia thuyet rang p<l) ldn hctn p<2).
T ro n g trdo’ng h o p b an m u o n kiem d in h h d d n g n g iio c lai (td c la
pi2) > p in ^ cgn j£y ^ <ji q Ua gia tr ip de th u ddoc gia tr ip thifc
te cho gia thuyet cua chung ta. Viec sd dung p h iio n g phap PLS-MGA
de kiem dinh cac gia thuyet hai phia la khong th e khong cd gidi han
vi ph an phoi dOa tren bootstrap khong nhat th ie t p hai doi xdng. Dac
diem nay rd rang han che kha nang dng dung cua nd khi cac nha nghien
cdu th d d n g can den cac kiem dinh hai duoi, m ac du cac mdi quan tam
thiidng xuycn (vi du: C ho & Abe, 2012).

ci
K iiem dinh hoan vi
Kiem d inh hoan vi ban dau ddoc phat trie n bdi C hin (2003) va
ddoc chdng m inh them bdi C hin va Dibbern (2010). N h d ten goi cua
nd va tdong tii vdi vai trd cua nd trong Bddc 2 cua th u tuc M1COM,
kiem dinh hoan vi trao doi n g lu nhien cac quan sat gida cac nhom dd
lieu va ddc ldong lai md hinh cho moi hoan vi. T in h toan sd khac biet
gida cac he sd dddng dan theo nhom cu the tren m oi hoan vi cho phep
kiem d jnh xem chdng cd khac nhau ve tdng the h a y khong.
 C ac m a u B o o tstrap p to
'C X

Chu'o’ng 4: Mo hinh hoa khdng c!6ng nhat quan sat dugc

0,357 0,226 '< (,3 1 8 0,281 0,372 0,318 0,296 0,308 0,415 0,272

0,494

0,423
"bc 0,324 X X
a* 1
2
44 0,591
V> y
§ 0,698
CQ
3
0,291 X X X X X
B
u ( c £ x
0,509
o
0,400 X

0,526
$> •
0,538

G hi chu: D a u X bieu th i trUdng hdp p (1) > p'X2)

157
158 Nhung van de nang cao trong PLS-SEM

Kiem d in h hoan vi tuan theo quy trin h gom sau bfldc:

“1. T hflchien fldc lfldng nhom ' c u th e cua cac mo hinh dfldng dan
PLS de co dflpc cac tide liiong he so diidng dan cho nhom 1 va 2.

2. Tinh toan sii khac biet gifla cac fldc lfldng he so diidng d in theo
nh o m cu the: d = p (1)- p (2).
3. Tao hoan vi ngau nhien cua dfl lieu, cd nghia la cac quan sat
diictc gan ngau nhien cho cac nhom . Theo quan diem chinh
thflc, n(1) quan sat diiqc ru t ra m a khong can thay the tii tap
hq p dfl lieu tong h o p va gan cho nhom 1. So quan sat cdn lai
n (2) diiqc gan cho nhom 2. Cac nha ng h ien cflu n$n sfl d u n g sd
liiong toi thieu 1000 hoan vi.
4. Tien h an h fldc liiong nhom cu the cua cac m o h in h diidng d in
PLS cho m di lan chay ho an vi u. Vi du, vdi 1000 lan hoan vi,
chung ta thu diiqc 1000 fldc liiong m d h in h cho n h o m 1 (tflc la,
) va cho nhom 2 (tflc la, p ^ ).

5. T inh toan sfl khac biet trong cac fldc lflqng he sd dfldng d i n khi
chay hoan vi cu the: du = p ^ 1- p ^ .

6. Tao khoang tin cay dfla tren hoan vi hai du o i 95%. C u the hon,
sap xep gia tri ket qua du theo thfl tfl tang dan va xac djn h hai
gia tri tach biet (1) gia tri cao n hat tfl 97,5% tdi thap nhat 2,5%
(can dfldi) va (2) gia tri cao nhat tfl 2,5% tdi thap nhat 97,5%
(can tren). Neu chenh lech ban dau d cua tide lflqng he sd dfldng
d in cu the theo nhom khong roi vao k hoang tin cay, thi nd cd
y nghia thong ke. Ngoai ra, dfla tren ket q u a hoan vi, SmartPLS
cung cap gia tri p cua sfl khac biet d. Gia tri p n h d h d n 0,05 cho
thay rang sfl khac biet d gifla cac he sd dfldng d in nhom cu the
khong roi vao khoang tin cay dfla tren h o an vi 95% va do dd cd
y n g h ia thong ke.

Kiem dinh hoan vi noi chung da dflpc chflng m inh la hoat dong rat
tot tren m ot loat cac dieu kien (vi du: Ernst, 2004; Good, 2000). Q uan
trong nhat, kiem dinh hoan vi cd the kiem soat dang tin cay doi vdi sai
lam Loai I khi viec gan cac quan sat xay ra n g iu nhien, nhfl trfldng hqp
trong flng d u n g kiem dinh hoan vi trong PLS-SEM. Tfldng flng, nghien
cflu trfldc day chi ra rang kiem dinh hoan vi thflc hien m ot each than
trong hd n so vdi kiem din h tham so ve sfl khac b iet cd y n ghia (Sarstedt,
Henseler, & Ringle, 2011). Trong boi canh nay va vi dac diem cua kiem
 Churnig 4: Mo hinh hda khong cJdng nhat quan sdt du’P’c 159

din h phi tham so, chung toi thddng khuyen ban nen sd dung kiem dinh
hoan vi. Tuy nhien, sii khac biet ldn ve kich th d d c m au theo nhom cu
the co nhdng hau qua bat Ipi cho viec thde hi£n kiem din h hoan vi.
Do do, khi m au cua m ot nhom ldn hon gap ddi kich thifdc cua nhom
khac, cac n h a nghien cdu nen diia tren phdong phap PLS-MGA (trong
triidng hop kiem din h gia thuyet m ot phia) hoac phiiong phap tham
so. N h d m ot p h d o ng phap thay the, cac nha nghien cdu cd the rut ngau
n h ien m ot m au khac tii n h o m ldn sao cho cd kich thddc tiiong diidng
vdi nhom nhd hon va so sanh hai m au bang each sd dung kiem djn h
hoan vj.

So sanh nhieu hdn hai nh<5m

Cac phdong phap chuan de so sanh da nhom trong PLS-SEM cd diem
chung la chdng kiem dinh sii khac biet trong cac tham sd gida hai nhom .
Tuy nhien, cac nha nghien cdu th d d n g xuyen gap phai cac tin h huong
ma ho m uon so sanh m ot tham sd (vi du: he sd diidng dan) tren nhieu
hon hai nhom . Trong PLS-SEM, do p h d e tap cua phan tich nay — ddoc
xac d inh bdi sd lddng so sanh he sd dddng dan nhom cu the — tang
theo do phde tap cua md h in h d d d n g dan (nghia la sd ld d n g m di quan
he) va sd nhom . Vi du, m ot m d h in h cd 10 mdi quan he d ddng dan trong
md hinh cau trdc va nam nhom yeu cau 100 so sanh cac he sd dddng
dan theo nhom cu the. Neu khong cd sd khac biet cd y nghla gida cac
nhom , kha nang tim thay it n h at m ot sd khac biet cd y nghia la ldn
hon 95% khi tien h an h 100 so sanh. P han tich p h d e tap n h d vay khong
n hdng ton thdi gian, m a con cd m ot van de nghiem trong hdn lien quan
den each tiep can nay, ddoc goi la phong dai alpha (alpha in fla tio n )
(con dddc goi la van de da kiem dinh-m ultiple testing problem). Dieu
nay de cap den thde te la ban cang tien hanh nhieu thd nghiem d m dc
y nghia nhat dinh, ban cang cd nhieu kha nang nhan ddoc ket qua cd y
nghia khi dieu nay khong dung (tdc la sai lam Loai I). Vi du: gia sd m dc
y nghia 5% va thde hien tat ca cac so sanh theo cap n h d ddoc chi dinh
d tren, xac suat chung cua sai lam Loai I (cung ddoc goi la fam ilywise
error rate- FWER) khong phai la 5% m a la 99,41% (Sarstedt & Mooi,
2014). Noi each khac, khi tien hanh tat ca cac so sanh nhom theo cap,
FWER n h an h chdng vdot qua m dc sai lam Loai I cd the chap n h an ddoc
(tdc la chap nhan d dong tin h gia), vi du 5%.
Trong bdi canh nay, khi so sanh nhieu hon hai nhom , cd hai van
de chd y dam bao. Dau tien, chdng ta can phai kiem djnh xem he sd
160 Nhung van d§ nang cao trong PLS-SEM

d d d n g d an co khac vdi he sd khac trong it nhat m ot trong cac nhom

(tile la, kiem d inh cho sd khac biet tong tfi£)~Hay khong. Vdi m uc dich
nay, chung toi de nghi sd dung hoan vi dda tren phan tich phdo n g sai
(ANOVA), dieu ma duy tri FWER, khong phu th u o c vao cac gia dinh
phan phoi va the hien do nhay thong ke d m dc cd the chap nhan diidc.
Trong triidng h op chung a tim thay m ot sii khac biet dang ke, biidc th d
hai la d a n h gia va xac din h gida cac n h o m he so d iid n g d an khac (tdc la
so sanh tdng cap) bang each sd dung cac quy trin h chuan (vi du: kiem
d inh hoan vi). Tuy nhien, ket qua nen ddpc dieu ch in h de giai thich cho
van de da kiem dinh - m ultiple testing problem .

Trong phan sau, chung toi se giai thich chi tiet h o n hai biidc nay.
N hii m ot liiu y than trong, dieu quan trong la hay n h d rang ban cang
tim kiem nhieu, cang cd nhieu kha nang la ban se tim thay m ot cai gi
dd. Vi ly do nay, ban nen tap trung vao viec phat hien sd khac biet chac
chan gida cac nhom trong PLS-SEM.

Kiem dinh sukhac biet tong the

Cach tiep can tieu chuan de kiem dinh ve sd khac biet tong the
theo m o t tham so tren nhieu nhom lien quan den viec chay kiem dinh
F trong ANOVA. Tuy nhien, theo phdo n g dien dac diem tham so cua
kiem din h va cac tin h chat tiem can cua nd khi sd dung cac ket qua
bootstrapping, so sanh cac nhom khac nhau bang phdong phap kiem
dinh F ANOVA khong cd y nghia trong boi canh PLS-SEM.
N hd m ot bien phap khac phuc, Sarstedt, H enseler va Ringle (2011)
da phat trien kiem dinh om nibus ve sd khac b iet nhom (om nibus test
o f group differences - OTG), ket hop thu tuc b o o tstrap p in g vdi kiem dinh
hoan vi de bat chddc kiem dinh F tong the. C ach tiep can nay duy tri
m dc sai lam Loai I theo quy dinh cua n h a ng h ien cdu (tdc la FWER)
v i cung cap m dc do thong ke cd the chap nhan d d p c trong khi khong
phu thuoc vao cac gia din h phan phoi. C ach tiep c a n OTG bao gom cac
bddc sau:

1. Chay th u tuc bootstrapping tren m di n h o m rieng biet de lay cac

phan phoi tham chieu cua cac tham so n h o m cu the.

2. T inh ty le phdong sai. Day la ty le phdOng sai ddpc giai thich bdi
bien nhom va phdong sai tong the:
 Chworig 4: Md hinh hoa khong ddng nhat quan sat dU’O’c 161

K - B - ( l / / C - t ) £ ( p < n - p )2
f - __________________ uJ-------------------

i=i j-\

Trong p h fld n g trin h n a y p j'* la d d c ld d n g th a m so tit m a u b o o ts tr a p th ii

j (j = 1 , B) tro n g n h o m th ii i (i = 1 , K ), p (,) la t r u n g b in h c u a cac
d d c liidng th a m so b o o tstra p cu a n h o m i, va p la g ia tr i t r u n g b in h
to n g q u a t cua cac tide liid n g th a m so b o o ts tra p tre n ta t ca cac n h o m .

3. Tao ra ca c h o a n vi c u a cac d a c liid n g th a m s o p * 0 g ifla cac

n h o m , rie n g b ie t c h o m o i m a u b o o ts tra p . Q u a t r i n h b a t d a u
v d i m a u b o o ts tra p d au tie n b a n g ea ch tra o d o i n g a u n h ie n cac
d d c ld d n g th a m so p,01 gifla cac n h o m . T i£p th e o , d ie u tflo n g
tii c u n g d u a c th d c h ie n c h o m a u th fl h a i va c a c m a u b o o ts tr a p
co n lai c h o d e n k h i ta t ca cac d d c lfld n g th a m s o d a d d d c h o a n
vi c h o ta t ca cac m a u b o o ts tr a p B. Q u a tr in h n a y ta o ra to n g
so (ic!)®’1 h o a n vi k h ac n h a u , ra t k h o d e h o a n t h a n h t i n h to a n
- vi d u , doi vdi B = 5000 m l u b o o ts tra p va K = 3 n h o m , yeu
cau (S!)4" 9 = 9,508 x 1O3809 h o a n vi. D o do, e a c h tie p can O T G
th a m gia m o t each n g a u n h ie n , vi d u , 5000 h o a n v i v a tin h ti so
p h fld n g s a i ch o m o i p h e p ch ay h o a n vi / (Z = 1 ,..., U ).

4. T in h gia tri p b an g each sfl d u n g h a m b fld c H e a v is id e H . H a m

b fldc H eaviside la m o t h a m k h o n g lien tu c co g ia t r i la 0 c h o doi
so am va 1 ch o do i so dfld n g . T ro n g b o i c a n h cfla p h fld n g p h a p
O TG , no tra ve 1 khi FR - m a n g lai gia tri d fld n g v a 0 d o i v d i g ia
tr i Sm:

c£>
N eu each tiep can O T G ch o th a y m o t ta c d o n g co y n g h ia , c h fln g ta c o
th e ket lu a n rang it n h a t h e so d fld n g d a n cu a m d t n h o m k h a c b ie t d a n g
ke so v*i cac n h o m k h ac . Tuy n h ie n , d ie u nay k h o n g c o n g h ia la ta t ca
cac he so d fld n g dan cfla cac n h o m k h a c n h a u d a n g ke. D e d a n h gia cac
n h o m n a o k hac n h a u , chflng ta can ap d u n g k ie m d in h so s a n h cap.

C ach tiep can O T G chfla d floc b ao go m tro n g S m a rtP L S . T u y

nhien, m a R cua O T G cd san tro n g p h a n tai x u o n g ta i h ttp .://w w w .p ls-
sem .com /.
162 Nhung van <3§ nang cao trong PLS-SEM

So sanh th eo cap

Cach tiep can OTG cho phep xac dm h cac h e so diidng dan cho
m ot so sanh cap doi bang each sd dung kiem d in h h o an vi chang han. Vi
cac kiem d inh nhii vay chi so sanh hai nhom cung m o t luc, b an can chay
kiem dinh nay nhieu lan cho den khi ban thde hien tat ca cac so sanh co
lien quan. Vi du, neu ban m uon so sanh mot m oi quan he diidng dan cu
the p tren 5 nhom , ban phai tien h an h 10 so sanh.

Trong vi d u nay, Tii lieu 4.6 cho thay each b ao cao ket qua so sanh
cua he so diidng dan cu the tren nhieu nhom (vi du: bang each bao cao
sii khac biet va y nghia cua chung trong moi o).

NhU da chi ra 6 tren, m oi quan tam chinh cu a viec tien hanh so

sanh cap la phong dai alpha, theo do xac suat tong the cua sai lam Loai
I (tdc la FW ER) tang theo cap so n h an vdi so lan so sanh. M ot each
tiep can tieu ch uan de kiem soat FW ER dd la h ie u c h in h B o n ferro n i
(Bonferroni correction). Thay vi sd dung m ot m dc y nghia cu the trong
quyet dinh kiem dinh, phep hieu chinh Bonferroni kiem dinh tdng gia
thuyet rieng le tai m dc y nghia alpha / m. Vi du, trong trd d n g hop cd 5
nhom , se cd m = 10 so sanh, tao ra m d c m dc y nghia la 0,05 / 10 = 0,005
thay vi 0,05. M ot each tiep can pho bien khac la th u tu c S idak (Sidak
procedure), sd dung 1 - (1 - alpha)l/m trong moi kiem dinh gia thuyet. 6
day, trong vi du ve nam nh d m (tdc la 10 so sanh), ngddi ta se sd dung
m dc y nghia 1 - (1 - 0,05)1/10 = 0,005116 thay vi 0,05.

Nhom 1 N hdm 2 Nhdm 3 N hom 4 N hdm 5

N hdm 1

^ N hdm 2 p® -p®

N hdm 3 p<i) , p P > p » -p®

N hdm 4 p<» - p ® p® -p® p<3) -p®

p(l ) . p ® p(3) , p ( 5) p( 4) _ p(S)

N hdm 5 p® -p®

Ghi chu: p, dai dien cho he so dudng dan cu the; trong md hinh cdu true; so sanh qua 5
nhom; cdc so trong ngoqc don cho biet nhdm md he so cu the da thu dupe.
 Chuong 4: Mo hinh hoa khong ddng nhat quan s it dime 163

Ca hai each tiep can deu chong lai sd gia tang FW ER khi thde hien
n h ieu so sanh tren m ot so n h o m dd lieu. Tuy nhien, viec duy tri FWER
dan den viec hao hut sdc m anh thong ke. Vi du, vdi m = 10, doi vdi m ot
so sanh don 1£ ddpc coi la dang ke, vdi hieu chinh Bonferroni, sd tac
dong can ldn h o n 43% so vdi m dc alpha 5%.

Quy tic ktnh nghiem de danh gia bat bien va

Tii lieu 4 .7
phan tfcH da nhdm v-'"' . ■
Danh gia bat bien

• Chay m ot phan tich da nhom yeu cau bat bien do lddng rieng phan,
ddpc dda ra khi sd bat bien cau h in h va bat bien th an h phan cd hieu
ldc. Neu sd bat bien do lddng rieng phan ddpc xac nhan va cac h p p
td cd gia tri trung bin h va p h d o n g sai b§ng n h au gida cac nhom , thi
cd sd bat bien do lddng toan phan, ho trp phan tich dd lieu gop.
• De dan h gia sd bat bien cau h inh, kiem tra xem cac m o h in h do
lddng cd sd dung cung cac bien quan sat gida cac nhom ve so ldpng,
noi dung va ma hoa hay khong. Bat k y d a n g x d ly d d lieu nao (vi du:
gia tri ngoai lai, cac gia tri khuyet) phai xay ra giong nhau cho cac
nhom . Ddc ldpng m o hinh theo nhom cu the phai dda tren cung
n h d n g cai dat thuat toan.

• Sd dung 1000 hoan vi (hoac nhieu hon) de danh gia sd bat bien
thanh phan va tddng d dong tri tru n g binh tong hpp va phdong sai
tro n g cac Bddc 2 va 3 cua thu tuc MICOM.

• Sd bait bien t hanh phan ddpc thiet lap khi cac tdPng quan ban dau
gida diem so tong hpp cua nhom 1 va 2 (tdc gia tri c) ldn hon 5%-
phan vi cua phan phoi thde nghiem cua cu. Trong nghien cdu kham
pha va khi cd m au nhd, viec quay trd lai 1% -phan vi la hpp ly hon.

• D e kiem d in h sd can bang tri tru n g binh tong h p p va p h d o n g sai

g ida cac nhom , kiem tra xem sd khac biet ve cac trj tru n g bin h va
phdong sai cd y nghia thong ke hay khong.

Phan tich da nhom

• Sd dung phep kiem dinh hoan vi de phan tich da nhom bang each
sd dung 1000 hoan vi (hoac nhieu hon).

• Khi cd m au cua m ot nhom ldn hdn gap doi kich thddc cua nhom
khac, lay ngau nhien m ot mau khac td nhom ldn sao cho cd kich
thddc tddng d uong vdi nhom nhd hon va so sanh hai m au bang
164 Nh&ng v in de n§ng cao trong PLS-SEM

phep kiem dinh hoan vj. N goai ra, sii dung phiiong phap PLS-MGA
(trorig'frllcing hOp’laerrrdrhH ^ tTiuyet m ot chieu) hoac phiiong
phap tham so.

• Khi sii dung phiiong phap PLS-MGA, sii dung 500 m au bootstrap
cho danh gia ban dau va 5000 m au bootstrap de danh gia cuoi cung
ve sii khac biet nhom . Sii dung cai dat thuat toan bootstrapping
chuan (vi du: tuy chpn khong thay doi dau).

• Khi sii dung phiiong phap tham so, kiem din h sii khac
p hiiong sai nh o m . Trong triid n g hop cac phiio n g sai
ke, ap dung kiem din h t W elch-Satterthwaite.

» De phan tich nhieu h o n hai nhom , hay sii d u n g phiiong phap OTG.

Do do, cac nha nghien ciiu thpdng thich th u tpc Sidak h o n la hieu
chinh B onferroni vi each tiep can nay dat diipc m iic do thong ke co the
chap nhan diipc. N ghien ciiu da dOa ra m ot loat cac phiiong phap thay
the (va phiic tap hon) de duy tri FWER (de biet chi tiet, xem H ochberg
& T am hane, 2011).

T o lieu 4.7 tom tat cac quy tac kinh nghiem doi vdi sp ket h p p phan
tich da nh o m va d an h gia b i t bien tro n g PLS-SEM.

TINH HUONG NG HIEN Ctfu M IN H HOA - DANH GIA BAT

BIEN VA PHAN TICH DA NHOM
De m inh hoa viec sP dung phan tich da n h o m c u n g vdi viec d an h gia sp
bat bien m o hinh do liidng bang thu tuc M IC O M , chung ta sP dung m o
h in h m d rong (Extended m odel) trong cong tr in h nghien cPu D anh
tieng cong ty tro n g Corporate Reputation- Advanced PLS-SEM Book.
Thay vi phan tich tap d p lieu tong hpp vdi 344 q u a n sat, chung ta q u an
tam d en viec p h an tich lieu tac dong trong m o h in h co khac b iet dang
ke doi vdi khach hang co goi d ien thoai di dong tra triidc so vdi khach
h an g theo h p p dong hay khong. N ghien cPu th u th a p dfl lieu tren kh ach
h an g theo h p p d o n g - contract plans (servicetype = 1; n n) = 219) so vdi
nhflng ngPdi cd goi tra triidc - prepaid plan (servicetype = 2; n (2) = 125),
va chung ta se so sanh hai nhom dd.

Khi tham gia vao phan tich da nhom , c h u n g ta can dam bao rang
so ldpng cac quan sat trong moi nhom dap Png cac quy tac k inh nghiem
doi vdi cac yeu cau kich thudc m au toi thieu. Vi so m ui ten toi da chi
 Chupong 4: Md hinh hoa khdng dong nhit quan sat dupe 165

vao m ot bien tiem an la tam , chung ta can it n hat 8 10 - 80 quan sat cho
moi nhom , theo quy tac 10 lan. Theo cac khuyen nghi khat khe hon tit
phan tich do nhay (Hair va cong sii, 2017), 54 quan sat tren moi nhom
la can thiet de tim ra gia tri J?2khoang 0,25 d m dc y nghla 5% va do nhay
80%. Do do, kich thudc m iu theo nhom cu the co the diipc coi la du ldn.

Trong bu d c dau tien, chung ta phai dinh nghia n h o m bien m a

SmartPLS sit d u n g de phan chia tap dd lieu. De lam n h u vay, bam
dup vao tap d d lieu Corporate Reputation Data [344 records] trong
Corporate Reputation - Advanced PLS-SEM Book. SmartPLS se m d
m ot tab mdi (Tii lieu 4.8), cung cap thong tin ve tap dii lieu va dinh
dang cua nd (data view).

Tii lieu 4.8

C orporate reputation data.txt S3

D elim iter Encoding: UTF-8

Value Quote Character mple size: 344
Number Format Indicators: 41
Missing Value Marker Missing Values: 11

Indicators Raw File

Indicator Max
serviceprovi 4.000
servicetype 2.000
com pel 7.000
co/np_2 7.000
c o m p .3 7.000
7.000
likej 7.000
like.3 7.000

<b
c u s lj 7.000
cusl_2 7000
cusl 3 7.000

Khi md cda so dd lieu (data view), ba tuy chpn xuat hien trong
m enu phia tren cda so md hinh: Add Data Group, Generate Data
Groups, va Clear Data Groups. De xac dinh m ot nhom bien m di, kich
vao Generate Data Groups va m ot hop thoai m di se m d ra (Tii lieu 4.9).
Ben canh N am e prefix, ban cd the chi dinh m ot tien to m a SmartPLS
sd dung trong viec dat ten cua cac nhom trong bao cao ket qua. Trong
vi du cua chung ta, chung ta duy tri tien to mac dinh Group_. Ben dudi
166 NhCmg v ln <J§ ning cao trong PLS-SEM

Group Colum n, chung ta co the chi djn h m ot hoac nhiefu n h o m bien.

-Khi n h a p eh uot trai-vae-h^p hefl eanh-Greup C olum n 0, d an h sach-tat
ca cac bien trong tap dii lieu se xuat hien. D ang sau m oi bien, SmartPLS
se hien thi so liidng cac gia tri duy nhat. C uon xuong va dinh vi bien
servicetype, co hai gia tri duy nhat. N hap chuot trai vao servicetype va
dong hop thoai bang each nhap ch u o t trai vao OK.

Q uay lai che do xem dO lieu, SmartPLS se hien thi tab bo sung
co n h an Data Groups, n h an nay cung cap th o n g tin ve cac kich thddc
m au theo n h o m cu the. Bay gid ban co the dong che do xem dd lieu va
quay lai cda so m o hin h hien thi m o hin h m d rong ve d an h tieng cong
ty. Trddc khi tiep tuc, hay dam bao rang tap d d lieu ch in h xac (tdc la
Corporate Reputation D ata [344 records]) dddc kich hoat. N eu p hong
ch d tddng d n g khong xuat h ie n bang m au xanh la cay, hay n h ap chuot
phai vao Corporate Reputation Data [344 records] va nhap vao Select
Active Data File.

Generate Data Groups

Name prefbc GROUP_

— Group Columns
Group column 0:

Group column 1:

Group column 2:

— Prune groups
Total: 2
mum cases 80

De chay quy trin h M ICO M , ban can tim den Calculate

Permutation, ban co the tim thay d tren cung cua m an hinh SmartPLS.
Ngoai ra, b an co the nhap chuot trai vao bieu tddng banh xe vdi nh an
Calculate tro n g th an h cong cu. M ot m enu vdi cac tuy chon phan tich
thay th e se m d ra td do ban co the chon chay thu tuc Permutation.
Trong hop thoai tiep theo (T d lieu 4.10), trddc tien chung ta can
chi dinh cac n h d m can dddc so sanh. De lam n h d vay, hay nhap vao
 Chircmg 4: Mo hinh hoa khong ddng nhat quan sat dirac 167

m enu ke ben Group A va chpn GROUP_servicetype (1.0). Tiep theo,

nh ap vao m enu ke ben Group B va chon GROUP_servicetype (2.0). C hi
dinh it nhat 1000 hoan vi va kiem djnh hai duoi (two-tailed). Bdi vi loai
h in h dich vu diipc biet la anh h d d n g den each khach hang lien quan
den trai nghiem , chung ta gia dinh m dc y nghia la 0,05. Tiep theo, nhap
vao Start Calculation. Luc nay, SmartPLS ddc liipng mo hinh cho tiing
nhP m rieng biet, bang each sii dung cung chung cac cai dat thuat toan.
Cac m o h inh dPdng dan PLS cung n h d viec xd ly d d lieu ddpc sd dung
trong ca hai nhom la giong het nhau, do la m ot yeu cau can th iet cho
viec thiet lap sd bat bien cau h in h trong Bddc 1 cua th u tuc M ICOM .
H d n nda, vi cac ddc ldpng mo hinh nhom cu the cua chung ta cung
d d a tren cac thiet lap thuat toan giong nhau, sd bat bien cau h in h (Bddc
1 tro n g M ICOM ) ddpc thiet lap. Cac phan tich sau day gidi quyet cac
Bddc 2 va 3 cua thu tuc MICOM. Trong bao cao ket qua m d ra, di den
Quality Criteria -> MICOM. Tab dau tien co n h a n Step 2 (xem T d lieu
4.11) cho thay ket qua danh gia sd bat bien th an h phan.

11/ lieu 4.10

0 Setup 0 Partial Least Squares^ ©
© Missing
M i Values $31 Weighting

: ________
p—
|<iROUP_stivicetype(1.0)
. t : ______________________________

| CROUP_servicetype{j.0)

« MOO

0 One Tailed (§> Two Tailed

icance level 0.05

o Parallel Processing

Ldu y rang vi hoan vi la mot qua trinh ngau nhien, ket qua cua ban
se hdi khac so vdi nhdng gi ddpc trin h bay d day. T ddng td, ket qua se
hdi khac m oi khi ban chay lai th u tuc M ICOM .
C ot 5% cho thay ty le 5% phan vi cua phan phoi thdc nghiem cu.
So sanh cac tddng quan c gida diem so tong h o p cua nhom th d nhat
va nh o m th d hai (cot Original Correlations) vdi 5% phan vi cho thay
ld p n g phan vi luon nhd hdn (hoac bang) tddng quan c cho tat ca cac
khai niem nghien cdu. Ket qua nay cung ddpc ho trd bdi cac gia trj p
cao hon 0,05, cho thay mdi tddng quan khong thap hdn dang ke so vdi
 168 Nhdng vSn <3& nang cao trong PLS-SEM '

1. Vi du, gia trj td d n g quan ban dau cua A T T R la 0,993. Ket qua nay
.n am trong.khoang. tin .ciy jd d a Jre n h o an vi tdcfng dng vdi can dddi
0,948; theo do, gia tri p cua A T TR la 0,708, n h ii diiqfc hien thi trong cot
P e rm u ta tio n p-V alues, lcfn hcfn dang ke so vdi 0,05 (Tii lieu 4.11). Do
dd, tiiong quan ban dau cua A TTR khong khac biet dang ke so vdi 1,
ho trd ket luan rang sd bat bien thanh phan da diiqc thiet lap cho khai
niem nghien cdu nay. T ddng td, chung ta chdng m inh rang sd bat bien
th an h ph an da ddqc th iet lap cho tat ca cac khai niem da bien trong m o
h inh. Ldu y rang chung ta khong the chay B ode 2 cho khai niem don
bien CUSA vi m oi q u an h e ben ngoai cua nd la 1 theo thliet iet ke.n
ke. D o dd,
chung ta cd the bo qua gia tri p cua CUSA (T d lieu 4.11), khon;lg hoan
to an bang 0 trong vi du nay vi gia tri can dddi la I ?9999'i 999998 do
lam tron cac ket qua hoan vi nhat dinh. Tom lai, cac ket qua td Bddc 2
ho trq tin h bat bien do lddng tdng phan. D o dd, chung ta co the so sanh
cac he so dddng dan ddqc chuan hoa gida cac n h o m bang phdong phap
phan tich da nhom vdi sq tu tin.

Tu' lieu 4.11

on Permutation Mean 50%; Permutation p-Values

0.982 0948 0.708

0999 0997 0128
0961 0907 0827

m 1.000 0330
0.999 0998 0806
1.000 0999 0853
0.970 0933 0.630
0969 0940 0155

De kiem tra xem lieu phdp bat bien do ld d n g toan phan cd the ap
dung ddpc hay khong, hay nh ap vao tab S tep 3 tro n g bao cao ket qua
(xem T d lieu 4.12).
Chircmg 4: Mo hinh hoa khong dong nhat quan s£t dime 169
170 Nhung v4n de nang cao trong PLS-SEM

Hai cot dau tien trong T u lieu 4.12 cho thay sd khac biet tru n g binh
gifla diem .so Ldng_h.<2p_jJie.Q jc:et.qua tii .sfl d d c ld p n g m o h in h ban dau
va theo quy trin h hoan vi tflong flng (lflu y rang do h an che ve khong
gian, anh chup m an hinh khong hien thi toan bo n h a n cot). Hai cot tiep
theo cho thay can dfldi (2,5%) va can tren (97,5%) cua khoang tin cay
95% cua sfl khac biet tru n g binh diem so. N hfl chung ta co the thay, m oi
khoang tin cay bao gom sfl khac biet ban dau tro n g cac gia trj tru n g
binh, chi ra rang khong co sfl khac biet dang ke tro n g cac gia tri tru n g
binh cua cac bien tiem an tren hai nhom . Vi du, doi vdi A T T R , sfl khac
biet ban dau ve cac gia tri tru n g binh cua cac diem so bien tiem an la
-0,064, nam trong khoang tin cay tflong flng vdi can dfldi -0,217 va can
tren 0,232. Ket qua trong cot Perm utation p-Values ho trp h o n nfla
p h at hien nay cho A T T R va m pi khai niem ng h ien cflu k h ac tro n g m o
h in h dfldng dan PLS vi tat ca cac gia tri p deu ldn h o n 0,05. Cac cot tiep
theo hien thi cac ket qua tflong tfl cho cac phflOng sai tdng hop. M ot lan
nfla, tat ca cac k hoang tin cay bao gom gia tri b an dau va tat ca cac gia
tri p deu ldn h o n 0,05 m ot each ro ret. Do do ch u n g ta ket luan rang tat
ca cac gia tri tru n g binh tong hqp va phflOng sai la bang n h a u , dfla ra
sfl h6 trp cho bat bien do lfldng to an phan. Tfl lieu 4.13 tom tat ket qua
ph an tich M IC O M nay. Trong sfl can nhac cac ket q u a nay, chung ta tiep
tuc bang each kiem tra ket qua cua phan tich da n h o m . Sfl danh gia cua
chung ta tap tru n g vao kiem dinh ho an vi, co the tru y cap bang each vao
Final Results -> Path Coefficients trong dau ra h o a n vi SmartPLS.
tap&siaEasgai
MICOM Step 1 V
Bat bien cau hinh dfloc thiet lap? Co
MICOM Step 2
5% phan vi cua Bat bien thanh
Thanh Tfldng quan P
phan phoi thflc phan dflpc
phan c value
nghiem cu thiet lap?
ATTR 0,993 0,945 0,673 Co
COMP 0,998 0,997 0,114 Co
CSOR 0,985 0,906 0,828 Cd
CUSA 1,000 1,000 0,066 Cd
CUSL 1,000 0,998 0,813 Co
LIKE 1,000 0,999 0,648 Cd
PERF 0,979 0,933 0,598 Co
QUAL 0,961 0,940 0,248 Co
 Chicong 4: Mo hinh hoa khong d6ng nh§t quan sat du’o’c 171

MICOM Step 3
Su khac biet
Thanh gia tri trung Khoang tin cay P Tri trung binh
phan binh tong 95% value bang nhau?
hop (= 0)
ATTR -0,064 [-0,225; 0,221] 0,601 C6
COMP 0,060 [-0,238; 0,236] 0,611 C6
CSOR 0,072 [-0,216; 0,223] 0,539 C6
CUSA -0,097 [-0,226; 0,214] 0,420 c T
CUSL -0,060 [-0,229; 0,219] 0,587 (q a
LIKE -0,054 [-0,227; 0,229] 0,653 Co
PERF -0,023 [-0,222; 0,221] 0,897 ( C6
QUAL -0,010 [-0,229; 0,244] 0,949 C6
ATTR 0,190 [-0,297; 0,254] 0,691 C6
COMP 0,135 [-0,298; 0,266] 0,639 Co
CSOR -0,106 [-0,294; 0,262] 0,581 C6
CUSA 0,230 [-0,374; 0,351] 0,613 Co
CUSL -0,070 [-0,340; 0,358] 0,721 Co
LIKE 0,012 4-0,258; 0,248] 0,747 Co
PERF 0,038 [-0,306; 0,293] 0,927 Co
QUAL 0,154 [-0,309:0,271] 0,981 Co

Kilm dinh hoan vj trong SmartPLS S

P a th C oefficient*

V r V w C o p y t o C lip
M a tr a j

J P a th C o t f f io e .- . P a th C o e f fic ie ... P a th C o tffia t... P a th C o e f fic ie .- 25% 975% P e r m u ta tio n p - V a lu e s

i A H R -> COM P 0263 04M9 0214 •0.002 -0223 0225 03331

A T T R -> U tt 0245 0233 0212 -0014 -0283 0237 0.421

a C O M P -> CUSA 0214 0208 0206 -0004 -0270 0265 0.475

C O M P *> CUSI 0233 •0X158 0292 04)01 -0231 0238 0210

C S O R •> C O M P -04)26 04)91 •0218 04)07 -0207 0238 0307

CSO R -> UKE 0.096 0299 •0203 04)03 -0241 0236 0421

CUSA •> C U S l 0 j605 0.445 0260 04)01 •0276 0262 0066

U K E*> CUSA 0365 0.476 •0211 04)01 -0250 0247 0345

U K E -> CUSl 0285 0.406 -0221 0.001 *0231 0231 0.063

PERF -> C O M P 0210 0305 -04)95 0002 -0270 0266 0.496

PERF •> UKE •04)02 0272 -0274 0001 -0289 0285 0244

Q U A l •> C O M P 0535 0.418 0217 04)01 -0282 0274 0.418

Q UAL •> UKE 0501 0338 02(5 04)15 -0242 0265 0226
172 Nhdng v§n de n§ng cao trong PLS-SEM

Hai cot dau tien trong Til lieu 4.14 cho thay h e so dddng d in ban
d au tro n g nhom 1 va.n h o m 2 td d n g dng, theo d o la sd khac.b iet cua
chung trong tap dii lieu goc ban dau va phep kiem dinh hoan v| tiiong
dng (ldu y rang do h an che ve khong gian, anh chup m an h in h khong
h ien th i toan bo nhan cot). N hii co the thay, hau h e t m oi quan he mo
h in h cau true khong khac nhau giufa hai nhom .

Cac ngoai le duy nhat la m oi quan he gida C O M P va CUSL, CUSA va

CUSL, cung nhii LIKE va CUSL, khac biet co y nghia 6 m dc 10%. C hinh
xac hdn, tac ddng gida COM P va CUSL la khac biet co y nghia (p < 0,10)
gida cac khach hang theo goi hop dong (p r,) = 0,133) va khach hang theo
goi tra trddc (p(2) = -0,058). T ddng td, moi quan he gida CUSA va CUSL
la khac biet co y nghia (p < 0,10) gida cac khach hang theo goi hdp dong
(p(l) = 0,605) so vdi n hdng khach hang co goi tra trdcfc (p(2} = 0,445). Cuoi
cung, tac dong cua LIKE len CUSL la khac biet co y nghia (p <0,10) gida
cac khach hang theo goi h d p dong (p(l) = 0,185) so vdi n h d n g ngddi cd goi
tra trddc (p(2) =0,406).

D e phan tich sau h d n cac tac dong nhdm cu the, chdng ta co the
chay p h d d n g phap phan tich da nhdm khac tren d d lieu. D e lam n h d
vay, hay vao Calculate -> M ulti-G roup Analysis (M GA), ban cd the tim
thay d tren dau m an h in h SmartPLS. Ngoai ra, ban cd the nhap chuot
trai vao bieu td d ng banh xe tren th an h cong cu v a nhap vao tuy chon
Multi-Group Analysis (MGA).
|p l
Td lieu 4.15

O Setup < 5 Partial le a rt Squares Bootstrapping (? ) Missing Values ijjt Weighting

— - Croups A
3 Select AU
® GROUP_servicetype(lt)) 125 cases
0 GROUP_servicetype(2t)) 219 cases

O Select All
O GROUP_serwcetype(l.Q) 125 cases
fy l GROUP_servicetypc(2X)) 219 cases
 Chimng 4: Mo hinh hoa khong (Jong nhat quan sat du>pc 173

Trong hpp thoai sau (Til lieu 4.15), chung ta lai can phai xac dinh
cac nhom dupe so sanh. De lam dieu nay, danh dau vao o ben canh
GROUP_servicetype (1.0) dupi Groups A. Tiep theo, chpn GROUP_
servicetype (2.0) dupi Groups B. Trong cac tab khac, chung ta co the
chi din h cac thiet lap lien quan den thuat toan PLS-SEM, th u tuc
bootstrapping, xil ly gia tri khuyet, va trpng so bien. Sii dung cac cai dat
tieu chuan cho tat ca cac tuy chpn n h u dupe m o ta trong vi d u cua Hair
va congsU (2017).

Trong bao cao ket qua m P ra, di den Final Results -» Path
Coefficients va chpn tab PLS-MGA de truy cap ket qua cua phuong
phap phan tich da nhom nay (Tu lieu 4.16).

Tu' lieu 4 .1 6

Vi PLS-MGA dai dien cho m ot kiem dinh m ot duoi (one-tailed),

cac gia tri p nay cho biet lieu he so dudng dan trong nhom dau tien co
lPn hdn dang ke so vPi nhom thd hai hay khong. Day la trUPng h pp cua
cac m oi quan he COMP CUSL va CUSA -> CUSL. Bang each lay gia
tri 1-p, chung ta cung co the danh gia lieu co sp khac biet dang ke theo
huong khac hay khong.

Ket qua cho thay day la trudng h p p cua m oi quan he LIKE CUSL
vdi gia trip la 1 - 0,971 = 0,029. LUu y rang each tiep can PLS-MGA dUa
tren thu tuc bootstrapping, do la m ot qua trinh ngau nhien. D o do, ket
qua kiem din h cua ban se khac vdi ket qua dupe trinh bay P day, nhUng
sU khac biet khong dang ke. Cuoi cung, Tu lieu 4.17 trinh bay ket qua
cua cac kiem dinh tham so, ban co the truy cap bang each nhap vao tab
tUdng Png trong bao cao ket qua cua Multi-Group Analysis (MGA).
 174 Nhung van <J4 nang cao trong PLS-SEM

Chung toi hien thi ket qua cua kiem din h tham so vi thu tuc M IC O M
e k r t h a y rang cac-phddng sai khong-khac'Tihau gifla cac nhom . Tuy
nhien, ket qua kiem dinh t W elch-Satterthwaite, ban co the tim thay
trong tab tiio n g ting cua bao cao ket qua, th i rat giong nhau.

Tu' lieu 4.17

Path Coefficients

V K k h -S a tttrttn w iit T ...j C « V < ° O p b M r d :

_________ i________
Path CoefTKiefltS'diff ( ( GROUP.setvi«typefl.O} - 6ROUP_,etvicetype{2-0} D t-Vilut(G ROuP_s«V'-- p - V a M G R O U R j o v i. . .

Trong viec so sanh, chung t6i thay ket qua p h a n tich da nhom PLS
khong tao ra ket qua n h au gifla cac phiio n g phap, nhii trong Tii
lieu 4.18. M ac du cj khuyen ban nen sd dung phep kiem dinh
hoan vi, nhQng hi tiep can da phiiong p h ap n h ii vay cung cap

Tii lieu 4 .1 8 Ket q u a d a nhom I P tl- g if fa c a c p h iid n g p h a p

Kiem dinh PLS- Kiem dinh Kiem dinh t
He sd dudng ddn
hoan vi MGA tham sd Welch-Satterthwaite

& A T TR -> COMP

A T TR -> LIKE

COM P -> CUSA

COMP -> CUSL X X X X

CSOR -> COMP

CSOR -> LIKE

CUSA -> CUSL X X X X

 Chwc/ng 4: Mo hinh hoa khong ddng n h it quan sdt dugc 175

Td lieu 4 .1 8 K£t q u a da n h o m PLS gitfa c a c p h u o n g p h a p

Kiem dinh PLS- Kiem dinh Kiem dinh t
He so dudng ddn
hoan vi MGA tham so Welch-Sat ter thwaite
LIKE -> CUSA

LIKE -> CUSL X X X X

PERF -> COMP

PERF -> LIKE

QUAL -> COMP

QUAL -> LIKE
.
Ghi chii: dau X chi sU khac biet cd y nghia (p < 0,10) cita he s6 dudng ddn gidta
cdc nhom.

TOM TAT

Hieu ro tam quan trong cua viec xem xet tinh khong dong
nhat trong PLS-SEM. Cac dng dung cua PLS-SEM th d d n g dda
tren gia d in h rang d d lieu dai dien cho m ot to n g the dong nhat.
Trong nhdng trddng hdp n h d vay, viec ddc ldong m ot mo hinh
dddng dan PLS bao triim duy nhat dai dien cho tat ca cac quan
sat. Tuy nhien, trong nhieu dng dung tro n g the gidi thde, gia
dinh ve tinh dong nhat cua mau la khong thde te bdi vi ngddi
tra ldi co kha nang khong dong nhat trong nhan thde va danh
gia cua ho ve h ien tddng tiem an. Trong khi viec xem xet tinh
khong dong nhat la day hda hen td goc do thde te va ly thuyet
de tim hieu ve sd khac biet gida cac nhom ngddi tra ldi, thddng
la can thiet, nhdng cung day thd thach, de thu ddoc ket qua co

<b gia tri. Vi du, khi co sd khac biet dang ke gida cac he so dddng
dan gida cac nhom , mot phan tich theo m dc dd lieu gop cd the
huy bo cac tac dong nhom cu the. Ket qua phan tich n h d vay cd
the se gay hieu nham nghiem trong va dda ra cac khuyen nghi
sai lam cho cac nha quan ly va cac nha nghien cdu. Do dd, cac
nha nghien cdu ddpc khuyen nen xem xet van de cau true dd
lieu khong ddng nhat trong cac nd ldc lap m o hinh cua ho.

Giai thich sd khac biet gida tinh khong ddng nhat quan sat
ddpc va khong quan sat ddpc. Sd khong ddng nhat cd the quan
Nhung van de nang cao trong PLS-SEM

sat ddpc trong sd khac biet gida hai hoac nhieu nhom d d lieu
licit q u an derrcac dac d iem -quan-satd ddc, c h in g han n h d gidi
tinh, tuoi tac hoac quoc gia xuat xd. Cac nha nghien cd u cd the
sd dung cac dac diem co the quan sat nay de phan chia d d lieu
th an h cac nhdm quan sat rieng biet va thde hien ph an tich PLS-
SEM theo nhdm cu the. Thd hai, tinh khong dong nhat khong
the quan sat ddoc d cho no khong p h u thuoc vao m ot dac diem
quan sat cu the hoac ket hpp cua m ot so dac diem . De xac dinh
va xd ly tinh khong dong nhat khong quan sat ddpc, cac nha
nghien cdu cd the d d a ra m ot loat cac ky th u at ldp tiem in.

H ieu khai niem bat bien md hinh do ld d n g va sd danh gia cua

no trong PLS-SEM. Cac so sanh nhdm la co gia tri neu sd bat
bien do lddng da ddpc thiet lap. Q ua do, cac n h a nghien cdu
dam bao rang sd khac biet nhdm trong cac ddc ldpng m d hinh
khong phai la ket qua td noi dung va / hoac y nghla rieng biet
cua cac bien tiem an gida cac nhdm . Khi sd bat bien do lddng
k h o n g ddpc chdng m inh, bat ky ket luan nao ve m oi quan he
mo h in h deu cd van de. Thu tuc bat bien d o lddng cua cac hpp
td (M ICOM ) la m ot cdng cu h d u ich tro n g PLS-SEM. Thu tuc
bao gom ba bddc kiem din h cac khia canh khac nhau cua phep
bat bien do lddng: (1) bat bien cau h in h (td c la tham so va each
ddc ldpng n h d nhau), (2) bat bien th an h p h a n (tdc la trong so
bien qu an sat bang nhau), va (3) gia trj tru n g bin h tdng hpp va
p h d p n g sai bang nhau.

H ieu phan tich da nhom va thde hien n o trong PLS-SEM.

Phan tich da nh d m cho phep kiem din h s d khac biet gida cac
he sd dddng dan nhdm cu the co y nghia thong ke hay khong.
Cac n h a nghien cdu da de xuat cac p h d d n g phap tiep can khac
nh au cho phan tich da nhdm , vdi kiem d in h t dda tren phdPng
phap tham sd ddpc ap dung th d d n g xuyen nhat. Cac giai phap
thay the n h d kiem dinh hoan vi va PLS-M GA khong php thuoc
vao cac gia dinh phan phoi. Trong viec so san h , kiem dinh hoan
vi co dac tinh thong ke dac biet thuan ldi, va chung toi de xuat
phddng phap nay khi phan tich sd khac b iet ve tham sd gida hai
nhdm . Trong trddng hdp cd nhieu h d n h a i nhdm , viec sd dung
OTG la phu hdp.
 Chirang 4: Mo hinh hoa Khong d6ng nhat quan sat du'o'c 177

CAU HOI ON TAP

1. Su khac biet giUa tinh khong dong nhat quan sat dUdc va khong
dong nhat khong quan sat dUdc la gi?

2. Giai thich ba bUcrc cua thu tuc M ICOM .

3. Tai sao ban nen chay phan tich da nhom trong PLS-SEM?

4. Cac phuo ng phap phan tich da nhom nao co san trong PLS-
SEM? Thao luan ve Uu va nhUdc diem cua chung.

. , ,
CAU HOI T l/ DUY PHAN BIEN
. . f QX
1. Tai sao viec xem xet tinh khong dong nhat tlrong
t khi
phan tich cac m o hinh d ubng dan PLS?

2. N han xet binh luan ve tuyen bo sau: SU bat bien do ludng khong
phai la m ot van de trong PLS-SEM vi p h uong ph ap tap tru n g
vao dU bao va k ham pha.

3. Cac ham y cho phan tichfgw hi tien hanh bat bien do

ludng toan phan la gi?

4. Giai thich sU khac biet ve khai niem giUa phan tich da nhom va
phan tich cac tac dong tUdng tac.

THUAT NGO &

Alpha inflation. Phong dai Alpha
Bonferroni correction TUdng quan Bonferroni
Bootstrap samples M au Bootstrap
Bootstrapping Thu tuc Bootstrapping
Cluster analysis Phan tich cum
Common factor model M o hinh nh an to chung
Compositional invariance Bat bien thanh phan
Confidence interval Khoang tin cay
Configural invariance Bat bien cau hinh
Equality o f com posite mean TUdng dUdng tri tru n g binh tong
values and variances h(?p va phUdng sai
Familywise error rate Ty le sai so chung n h o n hat
178 Nhdng vdn d§ ndng cao trong PLS-SEM

FIMIX-PLS xem Finite m ixture partial least

~ squares
Finite m ixture partial least Binh phddng toi thieu tiing phan
squares (FIMIX-PLS) hon hpp h d u han
Full m easurem ent invariance Bat bien do lddng toan phan
Latent class techniques Cac ky thuat IPp tiem an
Measurement equivalence TdcJng ddong do lddng
Measurement invariance Bat bien do lddng
Measurement invariance o f Thd tuc bat bien do lddng cua m o
com posite m odels (MICOM) h in h to n g hop
procedure
MICOM xem M easurement invariance o f
composite models
M ultigroup analysis Phan tich da n h o m
Multiple testing problem Van de da kiem dinh
Observed heterogeneity Khong dong n h at quan sat ddpc
Omnibus test o f group Kiem dinh O m nibus ve sd khac
differences (OTG) o V biet nhom
OTG xem Omnibus test o f group
differences
Parametric test £ Kiem dinh tham so
Partial m easurem ent invariance Bat bien do lddng td n g phan
Permutation Hoan vi
Permutation test Kiem din h hoan vi
PLS-MGA Phan tich da nhom trong PLS-
SEM (M ulti-G roup Analysis)
PLS-POS xem Prediction-oriented
segmentation in PLS-SEM
Prediction-oriented Phan doan dinh hddng d d bao
segm entation in PLS-SEM trong PLS-SEM
(PLS-POS)
Sidak procedure Thu tuc Sidak
Unobserved heterogeneity K hong dong n h a t khong quan sat
dddc
 Chirang 4: M6 hinh hoa khong Pong nhat quan sat dirge 179

TAI LIEU DOC THEM

Becker, J.-M., Rai, A., Ringle, C. M., 8c Volckner, F. (2013). Discovering
unobserved heterogeneity in structural equation m odels to avert
validity threats. M IS Quarterly, 37, 665-694.

C hin, W. W., 8c Dibbern, J. (2010). A perm utation based procedure for

m ulti-group PLS analysis: Results of tests of differences on sim ulated
data and a cross cultural analysis of the sourcing of inform ation
system services between G erm any and the USA. In V. Esposito Vinzi,
W. W. Chin, J. Henseler, 8c H. W ang (Eds.), Handbook o f partial
least squares: Concepts, methods and applications in marketing and
related fields (pp. 171-193). Berlin, G erm any: Springer.

D iam antopoulos, A., 8c Papadopoulos, N. (2010). Assessing the

crossnational invariance o f formative m easures: Guidelines for
international business researchers. Journal o f International Business
Studies, 61, 1203-1218.

H ahn, C„ Johnson, M. D„ H errm ann, A., 8c Huber, F. (2002). C apturing

custom er heterogeneity using a finite m ixture PLS approach.
Scbmalenbach Business Review, 54, 243-269.

Henseler, J., Ringle, C. M., 8c Sarstedt, M. (2016). Testing m easurem ent

invariance o f composites using partial least squares. International
M arketing Review, 33, 405-431.

Sarstedt, M„ Henseler, J., 8c Ringle, C. M. (2011). M ulti-group analysis

in partial least squares (PLS) path m odeling: Alternative m ethods
and em pirical results. Advances in International Marketing, 2 2 ,195-
218:

Schlagel, C., 8c Sarstedt, M. (2016). Assessing the m easurem ent

jp * invariance of the four-dim ensional cultural intelligence scale across
countries: A composite model approach. European Management
Journal, 34, 633-649.
 <?
«v
&
<o<?
<b
 CHISONG 5

Md hinh hoa
khong dong nhcit
khong quan sat d

CAC KET QUA HOC TAP

1. H t|u dac phtrcfng ph^p tiep can FI

v& xCr ly sp khong d<5rtg nhi
2. Hjeu each sCrdung FfMlX-PLSvi PL
phan t(ch Idp ti£m an,
3. Hieu rq each thirc hl£n phan tic h fjj
4,. Biet cich sCr dung SmartPLS de chfj

T6NG QUAN CHl/GNG

Nhii da m o ta tro ng Chiiong 4, cac nha nghien ciiu thddng sii dung cac
dac diem co the quan sat de phan chia dd lieu th an h cac n h o m va ddc
lilong cac m o h in h rieng biet, til do tin h toan sii khong dong nhat quan
sat ddpc (observed heterogeneity). Tuy nhien, cac nguon goc cua sd
khong dong nhat trong dii lieu kho co the diipc biet trddc m ot each day
du. Do do, cac tin h huong phat sinh trong do sii khac biet lien quan den
sii khong dong nhat khong quan sat diipc (unobserved heterogeneity)
ngan can viec thu diipc cac ket qua chinh xac khi phan tich theo cap do
d d lieu tong hp p che giau tac dong nhom cu the. Viec khong xem xet
tin h khong dong nhat co the han che gia tri cua cac ket qua m o hinh
phdpng trin h cau true binh phiiong toi thieu rieng phan (PLS-SEM).
Vi the, viec xac dinh va - neu can - xd ly tinh khong dong nhat khong
quan sat diipc la rat quan trong khi sd dung PLS-SEM. Ngiipc lai, neu

181
182 Nhung van <Jd nang cao trong PLS-SEM

sii khong dong nhat khong quan sat ddoc khong anh h d d n g dang ke
den k eL q u a ,x a cu ah a .n g lu en .x ik x .ith e.p h a Ji. t!ch..dCi lieu tren m dc do
tong hop, day la thuan ldi cho viec tong quat hoa cac ket qua cua chung.

Trong chiidng nay, chung ta tim hieu ve hai phiidng phap, hon hdp
h d u han PLS (FIM IX-PLS) va phan doan dinh h d d n g d d bao trong PLS
(PLS- POS), cho phep cac nha nghien cdu xac dinh va xd ly sd khong
dong nhat k h o n g quan sat dddc tro n g cac m o h in h d d d n g dan PLS. Dda
tren m o ta nguyen tac hoat dong cd ban cua p h d o n g phap, chdng toi giai
thich each chung co the dddc sd dung cung nhau de xac din h so ld d n g
p h an d o an tiem an de gid lai dd lieu va ddc lddng cac m o h in h dddng
dan phan doan cu the. C hdng toi cung giai thich each tai tao cac phan
doan tiem an bang cac bien co the quan sat dddc va co y nghia ve m at
quan ly de bien ket qua ban dau thanh sd hieu biet co the hanh dong.
C h d d n g nay ket thuc vdi m ot dng dung thde te cu a cac p h ddng phap
dda ra dda tren vi du ve dan h tieng cong ty va p h an m em SmartPLS.

Kham pha tinh khong dong nhat

khong quan sat dddc trong cac mo hinh dddng dan PLS
Cach thong thddng de xd ly cac cau trdc d d lieu khong dong nhat
dda tren cac phd ong phap phan tich cum tieu chuan, chang han n h d
phan cum k-m eans (k-means clustering) (Sarstedt & M ooi, 2014). Trong
boi canh cua PLS-SEM, chdng ta co the sd dung p h a n cum k-m eans tren
dd lieu bien quan sat hoac diem so bien tiem an th u dddc td phan tich
ban dau cua tap dd lieu tong the de tao th an h cac nhom d d lieu. Phan
vung m a p h an tich nay tao ra sau do co the dddc s d dung lam dau vao
cho cac ddc ld d n g PLS-SEM theo nhom cu the. Viec nay de ap dung,
n h d n g m ot n h d d c diem ldn cua each tiep can n h u vay la cac phdong
phap nay khong tinh den cac moi quan he m o hinh do ld d n g va cau trdc
khi dddc xac d in h bdi n h a nghien cdu. Tuy n h ien , chinh xac la n hdng
moi quan he nay, co kha nang chiu trach nhiem c h o m ot so khac biet
nhom . D ong thdi, nghien cdu trd d c day da chi ra ra n g cac phdd n g phap
phan cum truyen thong thde hien rat kern lien q u a n den viec xac dinh
sd khac biet n h o m (Sarstedt & Ringle, 2010). Thda n h a n n h d n g han che
nay, n ghien cdu p h d d n g phap luan tro n g PLS-SEM da de xuat vo so cac
p h d d n g phap cu the de xac dinh va xd ly tin h k h o n g dong nhat khong
quan sat dd d c, th d d n g ddoc goi la ky thuat ldp tiem an (latent class
techniques) hoac ky thuat phan doan dda tren h oi dap (response-based
segmentation techniques). N hdng ky thuat nay da c h d n g m inh rat h d u
 Chu'cng 5: Mo hinh hoa khong <J6ng nhat khong quan sat dugc 183

ich de xac djnh tinh khong dong nhat khong quan sat diiqc va phan
vung dd lieu m ot each phu hqp. Ngoai ra, ky thuat lop tiem an co the xac
d in h rang tinh khong dong nhat khong quan sat diiqc khong anh hfldng
d en cac ket qua, ho trq phan tich m ot m o hin h duy nhat d p a tren m dc
dfl lieu tong hqp. Nghien cflu d a dfla ra vo so cac ky thuat khac nhau,
vi du, thuat toan phat sinh (Ringle, Sarstedt, & Schlittgen, 2014; Ringle,
Sarstedt, Schlittgen, & Taylor, 2013), binh phiiong n h d nhat (Schlittgen,
Ringle, Sarstedt, & Becker, 2015) va cac each tiep can khac de xac dinh
tinh khong dong nhat khong quan sat diiqc bang each sfl dung PLS-
SEM (de xem xet, xem Hair, Sarstedt, M atthew s & Ringle, 2016).

Ky thuat ldp tiem an dau tien va noi bat nhat la b in h phflOng toi
th ie u h o n hop hflu h an (FIM IX-PLS; Hahn, /ohnson, H errm ann, &
Huber, 2002; Sarstedt, Becker, Ringle, & Schwaiger, 2011). D ung nhfl
ten goi cua no, phflOng phap tiep can dfla tren khai ni£m m o hinh hon
h q p hflu han, gia djnh rang tong the la m ot hon h q p cu a cac h am m at do
n h o m cu the. M uc dich cua FIMIX-PLS la p h an tach cac p h an phoi hon
h q p to n g th e va cac th o n g so flcfc lflqng (vi du: h e so dfldng dan) cua
m oi n h im trong m ot k h u n g hoi q u y (tflc la, hoi quy hon hq p ; Becker,
Ringle, Sarstedt, & Volckner, 2015; Wedel & Kamakura, 2000). Tfl lieu
5.1 cho thay mot vi du ve m ot phan phoi hon hqp m a FIMIX-PLS nham
den de phan tach thanh cac phan phoi phan doan cu the.

Nguon: Phong theo Hair, Sarstedt, Matthews, & Ringle (2016).

184 Nhflng van d§ nang cao trong PLS-SEM

FIMIX-PLS theo hai bfldc. Trong bfldc dau tien, th u at toan PLS-
SEM dluln^flixc,chaiyir.en tapiifliieu_diydujde.c6 dfloc diem so cua tat
ca cac bien tiem an trong m o hinh. Nhflng diem so bien tiem an nay sau
do dong vai tro nhfl dau v&o cho m qt loat cac phan tich hoi quy hon hop
tro n g bfldc thfl hai. Phep hoi quy hon hqp cho phep phan loai xac suat
dong thdi cac quan sat th an h cac nhom va fldc lflqng cac he so dfldng
dan theo n h o m cu the. Trong khi nha nghien cflu can xac din h ro so
lflqng phan do an, FIMIX-PLS cung cap m ot loat cac bien phap thong ke
cung cap chi bao ve so lflqng phan doan co kha nang lam cho dfl lieu bi
anh hfldng. Day la m ot lqi the ro rang cua FIMIX-PLS vi tren thflc te cac
ky thuat ldp tiem an khac dfla tren PLS-SEM khong cung cap thong tin
chi tiet ve k h ia canh nay. Cac n g h ien cflu m o p h o n g chi ra rang FIM IX-
PLS bieu 16 dang tin cay sfl ton tai cua tinh khong d o n g n hat trong cac
m o h inh dfldng dan PLS va cho biet so lflqng phan doan thich hqp de
gifl lai tfl dfl lieu (Sarstedt, Becker, va cong sfl, 2011). Tuy n hien, cung
luc do, FIMIX-PLS ro rang bi h an che ve m at xac din h ch in h xac cau
tru e phan doan co ban, khi ddqc xac dinh bdi cac he so dfldng dan cua
m ot nhom cu the (Ringle va cong sfl, 2013; Ringle va cong sfl, 2014),
dac biet la khi m o hinh dfldng dan co ca thang d o lfldng nguyen n h an
(Becker, Rai, Ringle, & Volckner, 2013). Hem nfla, FIMIX-PLS chi co
kha nang nam bat ddqc tin h khong dong n h at tro n g cac m oi quan he
m o h inh cau trdc va khong the giai thich tinh khong dong n h at tro n g
cac m o h in h do lfldng, dieu nay han che tinh hflu d u n g cua no doi vdi
cac thiet lap nghien cflu thflc nghiem .

De khac phuc nhflng han che cua FIMIX-PLS, cac nghien cflu da
de xuat m ot loat cac Ifla chon thay the khac. M ot each tiep can day hfla
hen la phan doan theo dinh hfldng dfl bdo trong PLS-SEM (PLS-
POS) cua Becker, Rai, Ringle va V olckner (2013), m ot each tiep can co
ban khac vdi FIMIX-PLS. Thay vi xac djnh tin h k hong dong nhat d cap
do phan phoi (Tfl lieu 5.1), PLS-POS dan dan p h an bo lai cac quan sat
tfl phan doan nay sang phan do an khac vdi viec ap dung tieu chi m uc
tieu, la toi da hoa phflOng sai dd q c giai thich do giai phap phan doan
cung cap. C u the hon, PLS-POS tin h toan k hoang each cua td n g quan
sat den ph an doan rieng cua no cung nhfl cac p h a n doan khac de quyet
d in h th anh vien nhom cua no. Khi m ot q u an sat co khoang each ngan
n h at den phan doan rieng cua no, no van nam tro n g phan doan hien tai.
Neu khong, phflOng phap gan (gan lai) cac quan sat cho p h an doan thay
the ma nd the hien khoang each ngan nhat. Thuat to a n lap di lap lai quy
 Chirang 5: Mo hinh hoa khong dong nh&t khdng quan sat dupe 185

trin h nay cho tat ca cac quan sat trong tap dii lieu, do do cai thien m ot
each he thong giai phap phan doan hddng den tieu chi m uc tieu (nghia
la toi da hoa ldpng phiidng sai diicfc giai thich tren cac phan doan).
Nghien cdu m o phong cua Becker va cong sii (2013) cho thay PLS-POS
xac djnh m ot each dang tin cay cac cau true phan doan va hoat dong tot
hdn cac ky thuat phan doan thay the. Tuy nhien, PLS-POS khong diia ra
khuyen nghj cu the ve so liidng phan doan can xem xet trong phan tich.

Doi lap vdi boi canh nay, va bcfi vi viec thiic hien chung cua cac
phiidng phap tiep can nay trong SmartPLS 3, cac nha nghien cdu dddc
h d d n g ldi td viec sd dung ket hdp FIMIX-PLS va PLS-POS. Bddc dau
tien lien q u an d en viec chay FIMIX-PLS de xac dinh s6 ld d n g phan
doan thich ho p de gid lai td dd lieu. Giai phap FIMIX-PLS sau do phuc
vu n h d la m ot diem khdi dau de chay PLS-POS sau dd. PLS-POS cai
thien giai phap FIMIX-PLS va cung xac dinh tinh khong dong n hat
tro n g cac m o h in h do lddng. x Q >
Tuy nhien, vi cac phan doan diiqc xac dinh la — theo d in h nghia -
tiem an, cac ddc lddng mo hinh chi cung cap chi d an gidi han ve cac m di
quan he m ong doi trong thde tien. D o dd, bddc cuoi cung, lien quan den
viec tai tao cac ph an doan tiem an bang cac bien cd the qu an sat va cd y
nghia ve m at quan tri de chuyen do i cac ket qua ban dau thanh sd hieu
b ie tc o th e h an h dpng. T d lieu 5.2 to m ta t cac b d d c trong d n g dung ket
hdp cd he thong cua FIMIX-PLS va PLS-POS. Xay dpng theo Hair va
cong sd (2016) va Sarstedt, Ringle, va G udergan (2016), trong cac phan
sau chung ta se th ao luan tdng biidc m ot each chi tiet hdn.

Quy trinh man

Td lieu 5.2
FIMIX-PLS va

7.
Giai doan 1
Siat doe "'•;i Chay thu tyc FIMIX-PLS

Giai doan 2 fi
186 Nhung v ln de nang cao trong PLS-SEM

Butfc 1: Chay thu tuc FIMIX-PLS

Viec chay thu tuc FIMIX-PLS yeu cau nha nghien cdu thdc hien m ot so
Ida chon lien quan den cac thiet lap thuat toan. Q uy trinh ride liiong m o
h in h trong FIMIX-PLS tuan theo nguyen tac co kha nang xay ra, trong
do khang d inh rang tat ca cac bang chdng trong m ot m au co lien quan
den cac th am so m o hin h diiqc chiia trong ham kha nang (likelihood
function). H am kha nang nay diiqc toi da hoa bang each sd dung th u a t
toan toi d a h o a ky v ong (exp ecta tio n -m a xim iza tio n -E M ). Thuat toan
EM luan phien gida thdc hien bddc m ong dqi (E) va baric toi da (M).
Baric E tao ra m ot ham ve ky vong cua log-likelihood, ddqc danh gia
bang each sd dqng aric laqng hien tai cua cac tham so. Baric M tinh
toan cac tham so bang each toi da hoa log-likelihood vong daqc xac
dinh trong baric E. Cac baric E va M d aq c ap d u n g lien tuc cho den khi
ket qua on djnh. 6 n dinh dat d aq c khi khong co sii chi thien dang ke ve
gia tri (log) likelihood td m o t lan lap tdi baric tiep theo. Gia tri ngdring
1-10'5 d a q c khuyen nghj lam tieu chi ddng de d am bao rang th u at toan
hqi tu ri m dc do hqp ly thap cua cac thay doi lap lai tro n g cac gia tri
log-likelihood. K hi tieu chi d d n g d a q c dat rat thap, th u at toan FIMIX-
PLS co the khong hoi tu trong m ot khoang thrii gian hqp ly. Do do, nha
nghien cdu cu n g can xac d in h so lan lap toi da sa u dri th u a t toan se tq
dong cham ddt. Viec chi d inh so laqng toi da 5000 vong lap tharing dam
bao sq can bang hqp ly gida viec dam bao thrii gian chay tin h toan cri the
chap nhan d aq c va thu d a q c ket qua du chinh xac.

Sd dung th u at toan EM de aric laqng m o h in h thi cri sdc hap dan

bdi vi no ra t hi^u qua va luon hoi tu vdi m ot so phan doan d aq c xac
dinh trad e. Sq hqi tu co the xay ra ri m dc toi a u cu e bo, tuy nhien, dieu
nay co nghia la giai phap nay chi toi du so vdi che giai phap tricing tq
nhring khong phai tren toan bo (Steinley,2003). D e d ie u tra sq xuat hien
co the cri cua m ot toi du cue bo, cac nha nghien c d u nen chay FIMIX-
PLS nhieu lan. Viec khrii tao trong FIMIX-PLS x ay ra ngau nhien, cri
nghia la m oi khi FIMIX-PLS ddqc khrii tao, th u a t toan sd dung cac gia
tri bat dau d aq c chon ngau nhien khac nhau d e ride lriqng tham so.
Nrii chung, ket qua cua nhieu phep tinh FIM IX-PLS se rat giong nhau.
N hdng neu ket qua khong giong nhau, thi m o t set toi du cue bo da xay
ra va giai phap khong phu hqp vdi cac giai phap k h ac nen ddqc loai bo.
Phu h q p vdi ket qua nghien cdu mo phong ky th u a t (Sarstedt, Becker,
va cong sd, 2011), chung toi de xuat sd dung 10 lan lap lai p h ddng phap
FIMIX-PLS va chon giai phap cri gia tri log-likelihood tot nhat. Mot he
 Chirong 5: Mo hinh hoa khong ddng nhat khong quan sat du-oc 187

qua khac cua ban chat ngau nhien cua thuat toan la viec danh so cac
phan doan khong d d p cx a c dinh. Tdc la, ket q u a doi vdi m o t phan doan
co the xuat h ien tro n g m o t p h an d o an khac khi FIMIX-PLS diipc chay
lai. D ac diem nay th d d n g diipc goi la chuyen do i n h a n (label switching)
(McLachlan & Peel, 2000).

Mot xem xet quan trong h d n nda khi chay FIMIX-PLS lien quan
den viec xd ly cac gia tri khuyet. Kessel, Ringle, va Sarstedt (2010) d a
chi ra rang chi 5% gia tri khuyet trong m ot bien co the gay ra cac vSn
de nghiem trong tro n g phan tich FIMIX-PLS khi thay the chung bang
tru n g binh m au tong the cua cac gia tri hpp ly cua bien do (tdc la thay
the gia tri trung binh). Trong trd d n g hpp nay, viec tuy chon xd ly gia tri
khuyet tao ra m pt tap hpp cac diem so chung, dieu m a FIMIX-PLS xac
dinh la m ot phan doan dong nhat rieng biet. Ket qua la, so ldpng ph an
doan co the se diipc xac d in h q u a m dc va cac q u a n sat thpc sp thuoc ve
cac phan do an k hac se bi buoc vao phan d o an gia tao nay. Do do, thay
th e gia tri tru n g binh k h o n g diipc sd dung tro n g boi canh FIMIX-PLS,
ngay ca khi chi co rat it gia tri khuyet trong tap d d lieu. Thay vao do, cac
nha nghien cdu n en loai bo tat ca cac trd d n g hpp bao gom cac gia tri
khuyet trong bat ky bien quan sat nao diipc sd dung trong m o hinh khoi
phep phan tich (tdc la xoa toan bo - casewise deletion). Trong khi each
tiep can nay cu n g co van de, dac biet khi cac gia tri bi thieu m ot each
ngau nhien (Sarstedt & M ooi, 2014), no tranh diipc viec tao ra m ot phan
doan gia tao nhii trd d n g hpp thay the gia tri trung binh va cac phiiong
phap quy doi khac^phang han nhii quy doi EM va quy doi hoi quy.

Cuoi cung, thuat toan FIMIX-PLS can diipc chay cho so liipng phan
doan xen ke, bat dau bang giai phap m ot-phan doan. Vi so liipng phan
doan khong diipc xac d in h trddc, cac nha nghien cdu phai so sanh cac
giai phap vdi so ldpng phan doan khac nhau ve tin h day d u va kha nang
d iln giai th o n g ke cua chung. Pham vi so liipng phan doan co the p h u
thuoc vao sii tiiong tac gida kich thiidc mau va yeu cau kich thiidc m au
toi thieu de ddc ldpng m ot each dang tin cay m o hinh da cho. Vi du, khi
phan tich m pt tap dd lieu vdi 200 quan sat va yeu cau kich thiidc m au
phan doan toi thieu la 50, th i khong h p p ly de chay FIMIX-PLS vdi hPn
4 phan doan. D o dd, bat buoc phai xem xet cac yeu cau kich thddc m au
toi thieu cua m d hinh cu the (vi du, tro n g Hair, H ult, Ringle, & Sarstedt,
2017), trddc khi xac dinh m ot loat cac giai phap phan doan de xem xet
trong phan tich FIMIX-PLS. Sd ldpng phan doan toi da ly thuyet can
xem xet ddpc cho bdi sd nguyen ldn nhat, bang each chia cd m au n cho
188 Nhung v ln de ndng cao trong PLS-SEM

cd m au toi thieu. Tuy nhien, vi khong chac rang cac quan sat diiqc phan
b6 dong deu-tren-eae phan-doan, dac b ie tla khi-gidi han tren cao, can
nhac so liiqng phan doan thap h o n th d d n g diiqc d u tien.

Budc 2: Xac djnh so ludng phan doan

M ot thach thiic co b an vcfi viec ap dung FIMIX-PLS la xac din h so n hom

de gid lai tii dii lieu. Viec xac din h m ot so n h o m phu h o p la rat quan
trong, vi nhieu quyet d in h quan tri d d a tren ket qua nay. N h d Becker va
cong sd (2015, trang 644) Idu y r in g “so ldpng p h an doan khong chinh
xac dan den viec p h an d o an ddbi m iic hoac qua m iic co the dan d en cac
quyet d inh quan tri khong chinh xac lien quan den, vi du: n h am m uc
tieu khach hang, dinh vi san p ham hoac xac d in h m arketing h o n h o p toi
du”. FIM IX- PLS cho phep cac nha nghien cdu tin h toan tieu chi thong
tin dda tren kha nang xay ra (likelihood-based inform ation criteria) (con
ddqc goi la tieu chi Ida chon m o h in h - m odel selection criteria), cung
cap m o t dau hieu cho thay co bao nh ieu phan d o a n gid lai td dd lieu.
Tieu chi th o n g tin dong th d i tin h den sd phu h o p (tdc la kha nang xay
ra likelihood) cua m ot m o h in h va so ldqng th am so ddqc sd dung de
dat duqc sd p h u hpp do. D o do, tieu chi th o n g tin bieu thi m ot ham
phat likelihood. Do la, likelihood am cong vdi th d i han phat, dieu nay
tang len theo so ldpng cac p h an doan (Sarstedt, 2008). Gia tri cua m ot
tieu chi th o n g tin nhat d in h cang n h d th i giai phap phan doan cang tot.
Vi du noi bat ve tieu chi thong tin bao gom Tieu chi th on g tin Akaike
(Akaike’s Inform ation Criterion-AIC; Akaike, 1973), AIC ddpc sda
doi vdi he so 3 (M odified AIC w ith factor 3 - A IC 3; Bozdogan, 1994),
AIC nhat quan (C onsistent AIC - CAIC; B ozdogan, 1987) va Tieu chi
th on g tin Bayesian (Bayesian Inform ation Criterion-BIC; Schwarz,
1978) ). Ve viec trin h bay chinh thiic cua cac tieu ch i nay, xem them vi
du Sarstedt, Becker, va cdng sd (2011).

Tieu chi thong tin khong ddqc do lddng trong pham vi gia tri nhat
d in h (vi du: tii 0 den 1). lh a y vao do, cac tieu c h i co the cd gia tri d
hang 100 hoac hang 1000, tuy thuoc vao diem b a t dau cua thuat toan
FIMIX-PLS, ddqc dat ngau nhien. Tuy nhien q u an trong la m oi gia tri
cua tieu chi cd the ddpc so sanh gida cac giai p h ap k h a c nhau vdi cac so
p h an doan khac nhau. D o dd, n h a nghien cdu can kiem tra m ot so giai
ph ap vdi so lddng ph an doan xen ke va chon m o h in h giam thieu tieu
chi thong tin cu the.
 Choang 5: Mo hinh hoa khong dong nhat khong quan sat dope 189

Sarstedt, Becker, va cong sii (2011) da danh gia hieu qua cua cac tieu
chi thong tin khac nhau trong FIMIX-PLS tren m ot loat cac hop thanh dd
lieu va mo hinh. Ket qua cua chung cho thay rang cac nha nghien cdu nen
cung nhau xem xet AIC3 va CAIC. Bat cd khi nao hai tieu chi nay chi ra
cung mot so ldpng phan doan, ket qua nay co the cho biet so ldpng phan
doam thich hpp. AIC vdi he so 4 (AIC4; Bozdogan, 1994) va BIC noi chung
hoat dong tot, trong khi cac tieu chi khac the hien xu hddng danh gia
qua cao ro ret. Dieu nay dac biet cd gia tri doi vdi AIC, thddng xac dinh
so ldpng phan doan chinh xac theo ba hoac nhieu phan doan. Van con
cac tieu chi khac, chang han nhd Chieu dai mo ta toi thieu 5 (M inim um
Description Length 5-MDLs; Liang, Jaszak, & Coleman, 1992), cho thay
xu h ddng danh gia thap ro ret. Cac nha nghien cdu cd the sd dung thong
tin nay de xac djnh m ot so ldpng nhat dinh cac phan doan hpp ly. Vi du,
khi AIC chi ra m ot giai phap 5 phan doan, viec gid lai m ot so ldpng nho
cac phan doan dddng nhd ddpc bao dam . T d lieu 5.3 cung cap tong quan

(
ve cac tieu chi thong tin ddpc Ida chpn vai neu bat hieu suat cua chung
trong boi canh FIMIX-PLS.
y .
Tieu chi thong tin khong phai la “vien dan bac” de xac djn h so
ldpng phan doan phu hpp nhat trong FIMIX-PLS, vi cac tieu chi n h d
AIC, va BIC khong cung cap dau hieu cho thay each tach biet cac phan
doan dd. Vi ly do nay, cac nha nghien cdu nen xem xet viec sd dung bo
sung cac bien phap dda tren entropy (do m at tra t td), chang han n h d
th o n g ke en tro p y chuan (norm ed e n tro p y statistic- EN; Ramaswamy,
DeSarbo, Reibstein, & Robinson, 1993). EN sd dung xac suat thanh
vien phan doan cua cac quan sat de cho biet phan vung cd dang tin cay
hay khong. Cang cd nhieu quan sat the hien xac suat thanh vien phan
doan cao, thi chi n h an h phan doan se cang ro rang hon. EN nam trong
khoang td 0 den 1, vdi cac gia tri cao hon cho biet phan vung chat ldpng
tot hdn. Nghien cdu trd d c day cung cap bang chdng rang cac gia tri EN
tre n 0,50 cho phep phan loai ro rang dd lieu vao so ldpng p h an d o an da
xac djnh trddc (v id u: Ringle, Sarstedt, & Mooi, 2010).

Khi quyet dinh so ldpng phan doan can gid lai, dieu dac biet quan
trpng can n h d la th u at toan EM luon hoi tu vdi sd phan doan ddpc xac
dinh trddc. Tuy nhien, ket qua cd the la FIMIX-PLS buoc m ot tap hpp
nhd dd lieu vao cac phan doan khong lien quan, don gian b d i vi nha
n ghien cdu da chpn sd ldpng phan doan qua cao. Cac phan do an khong
lien quan nay chi chiem m ot phan khong ddng nhat trong tap d d lieu
tdng the va th d d n g qua nhd de dam bao tinh gia tri cac ket qua nhom cu
190 NhOng van de nang cao trong PLS-SEM

the (Rigdon, Ringle, & Sarstedt, 2010). Do do, ngoai cac tieu chi thong
tin va E N ,-n h a ngh ie ir c du-can-xem -xet can-than cac kich thiidc phan
doan do FIMIX-PLS tao ra. N eu phan tich tao ra m ot phan doan khong
lien quan qua nhd de dam bao phan tich cd gia tri, nha nghien cdu nen
xem xet giam so lutctng phan doan hoac giam ph an do an nay va tap
trung vao ph an tich va giai thich cac phan doan ldn h d n khac.

Viet
T ieu chi H ieu su at tro n g FIM IX-PLS
ta t
AIC Tieu chi thong tin • Hieu suat yeu
Akaike • Xu the rat m an h theo h ddng danh
A kaike’s Information gia qua cao sd ld p n g phan doan
Criterion • Cd the ddpspsit dung de xac dinh
gidi han tren cua cac giai phap phan
do an hpp ly

a ic 3 Tieu chi thong tin • C dng bang vdi h ieu suat tot
Akaike ddpc sda • Cd xu hddng d a n h gia qua cao sd
doi vdi he so 3 ldpng phan d o an
M odified Akaike’s • H oat dong tot k h i ket hpp vdi CAIC
I n fo r m a tio n ^ va BIC
Criterion With
Factor 3
Tieu chi thong tin • H ieu suat tot
‘ "X Akaike ddpc sda • Cd xu hddng d a n h gia qua cao va
doi vdi he sd 4 qua thap sd ld p n g phan doan
M odified A kaike’s
Information
Criterion With
Factor 4
BIC T ieu chi thong tin • Hieu suat tot
Bayesian • Cd xu hddng d a n h gia qua thap sd
Bayesian ldpng phan d o a n
Information • Nen ddpc xem x et cung vdi AIC3
Criterion
i
 Chirang 5: Mo hinh hoa khong ddng nhat khong quan sdt dope 191

Tieu chi thong tin duWc li/a chon va cua

Tu”lieu 5.3

V iet
Tieu chi H ieu su at tro n g FIM IX-PLS
ta t
CAIC Tieu chi thong tin • Hieu suat tot
Akaike nhat quan • C d xu h d dng danh gia qua thap so
Consistent A kaikes ldpng phan doan
Information • Nen ddpc xem xet cung vdi AIC,
Criterion
M DLS Chieu dai m o ta toi • Hieu suat yeu
thieu 5 • Xu the rat m anh theo h ddng danh
M inim um gia qua thap so ldpng phan doan
Description Length • Cd the ddpc sd dung de xac dinh
5 gidi han dddi cua cac giai phap phan
doan hpp ly
Nguon: Phong theo Hair, Sarstedt, Matthews, & Ringle (2016, trang 70).

Cuoi cung, dieu quan trong can ldu y la each tiep can thuan tuy
theo hudng dd lieu chi cung cap h d d n g dan set bo ve so ldpng phan
doan can chpn. P hdong phap giai quyet van de bang kinh nghiem n h u
tieu chi thong tin va EN co nhdng han che, bdi vi chung nhay cam vdi
cac dac diem cua m o hinh va dd lieu.

Vi du, ket qua nghien cdu cua Becker va cong sd (2015) cho rang
ngay ca m dc do cong tuyen thap trong m o h in h cau true cd the cd hau
qua xau doi vdi hieu suat cua tieu chi thong tin. FIMIX-PLS la m ot cong
cu kham pha va can du p e xd ly n h d vay. Do dd, bat ky quyet dinh nao
lien quan den so lupng phan doan can xem xet th d e te (vi du: Sarstedt,
Schwaiger & Ringle, 2009). Vi du, cac nha nghien cdu cd the cd m ot
kien thde tien nghiem hoac m ot ly thuyet cd the ho trp ly luan cua ho.
T d o n g td n h d vay, so ldpng phan doan phai dii n h d de dam bao tinh tiet
kiem va de xd ly, n h d n g mdi phan doan cung phai du ldn de dam bao lpi
th e chu y (Sarstedt & Mooi, 2014).

Bddc 3: C h ay th u tu c PLS-POS
Viec chay th u tuc PLS-POS yeu cau nha nghien cdu thde hien m ot so
Ida chpn ve cai dat thuat toan. Trong phan tiep theo, chung toi giai thich
192 Nhung v in fl§ nang cao trong PLS-SEM

ngan gon cac bddc chinh cua phan tich PLS-POS — nhii dupe the hien
trong--TU-lieu-5;4 — va-xflTy-eec ltfa ehen--khae-nhau de thdo-hien. De
biet chi tiet ve thuat toan PLS-POS, hay xem Becker, Rai, Ringle va
V olckner (2013).

K hi bat dau thuat toan PLS-POS, Ida chpn d a u tien can th d c hien
la so lUpng ph an doan can gid lai tii dfl lieu. Nhfl d a neu d tren, cac nha
nghien cflu nen ru t ra ket qua FIMIX-PLS cung vdi nhflng can nhac
thflc te de dfla ra quyet dinh nay. Sfl dung so p h a n doan lam dau vao,
PLS-POS sau d o gan ngau nhien tflng quan sat c h o m ot tro n g cac phan
doan va tiep tuc vdi phan tich. Ngoai ra, PLS-POS cd the sfl dung phan
vung cuoi cung do phan tich HMIX-PLS lam dau vao. Tflc la, m oi quan
sat dflpc gan cho phan doan noi cd xac suat th a n h vien cao n hat nhfl
dflpc chi ra bdi FIMIX-PLS. Phan vung nay sau d o dflpc sfl dung lam
phan vung khdi dau cua PLS-POS.
 Chirang 5: Md hinh hoa khong ddng nhat khong quan sat dime 193

Trong giai doan khdi tao PLS-POS, cac nha nghien cdu co tuy chon
chay m ot tien ph an d o an (presegmentation). Neu dieu nay diipc chon,
PLS-POS sd dung phan vung khdi dong va sau do gan lai tat ca cac quan
sat cung luc cho phan doan gan nhat cua chdng trong vong dau tien —
trddc khi gan lai cac quan sat tdng cai m ot khi chay cac lan lap tiep theo
cua thuat toan PLS-POS (Biidc 2 va 3). Viec gan lai tat ca cac quan sat
cung m ot ldc trong vong dau tien khong dam bao rang tat ca cac thay
doi deu gop phan cai thien tieu chi toi liu hoa. Tuy nhien, tuy chon nay
cd the tiet kiem thdi gian vi no thiidng doi hoi sp cai thien dang ke phan
vung khdi dong. Viec tien phan doan chi chdng m inh cd gia tri khi PLS-
POS khdi tao phan vung khdi dong mot each ngau nhien. Trong tn ld n g
hpp nay, cac nha nghien cdu nen chay PLS-POS nhieu lan vi chat liipng
cua giai phap cuoi cung phu thuoc vao phan vung khdi dong. T ddng
tp nhii FIMIX- PLS, hay chon giai phap tot nhat - lien quan den gia tri
cuoi cung cua tieu chi toi ilu hoa - trong 10 lan lap lai de tran h hoi tu d
m ot m dc toi du cue bo. Tuy nhien, khi sd dung giai phap FIMIX-PLS
lam phan vung khdi dong, chung toi khuyen ban khong nen sd dung tuy
chon tien phan doan vi biidc nay ddi hoi nhdng thay doi dang ke trong
giai phap FIMIX-PLS. Khi giai phap FIMIX-PLS da diipc hoi tu trong
giai phap toi d u toan cue hoac cue bo, cac thay doi di kem vdi viec tien
phan doan se bop meo phan tich FIMIX-PLS.

Ngoai ra, cac nha nghien cdu can phai quyet dinh ch ieu sau tim
kiem (search depth), xac dinh sd ldpng quan sat dng cd toi da cho viec
phan bo lai phan doan (xem Bddc 4). Neu chieu sau tim kiem cd gia tri,
vi du la 10, PLS-POS chi phan tich 10 dng cd dau tien trong danh sach
sap xep cac quan sat de phan bo lai. Thong thddng, cac nha nghien cdu
nen phan tich tat ca cac quan sat dng cd va do dd, sd dung chieu sau tim
kiem bang vdi so ldpng quan sat. Tuy nhien, viec giam chieu sau tim
kiem cd the thuan lpi trong cac cai dat phan doan phde tap (nghia la
m pt sd ldpng ldn cac quan sat va / hoac sd phan doan cao) de giam thdi
gian tin h toan, cd the trd nen qua m dc. Tuy nhien, ddi vdi cac tinh toan
PLS-POS cuoi cung, chung toi khuyen ban nen sd dung chieu sau tim
kiem bang sd quan sat trong tap dd lieu day du.

Mot quyet dinh khac trong viec khdi tao thuat toan PLS-POS lien
quan den viec Ida chon tieu chi to i d u hoa (optim ization criterion).
Phu hp p vdi tinh chat d d bao cua thuat toan PLS-SEM (Lohmoller,
1989; Wold, 1982, 1985), muc tieu cua PLS-POS la hinh thanh cac
nhdm quan sat nham toi da hoa dd bao cua cac bien tiem an noi sinh
194 Nhung vdn d l nang cao trong PLS-SEM

trong fldc lflpng m o hin h dfldng dan cua chung. Vdi m uc dich nay, m ot
tieu c h f tdi flu hoa thich-hpp la-toi da-hda-tdng-eua phfldng-sai-dflpe-giai
thich cua bien tiem an noi sinh (tdc la, gia tri R2 cua chung) tren tat ca
cac n h o m (phan cum de dfl bao toi da; A nderberg, 1973). Ngoai ra, cac
nha nghien cd u cd the sd dung tong trong so cua tong gia tri R2 cua moi
nhom lam tieu chi toi flu hoa, theo do kich thfldc phan do an tflong doi
cua cac nh o m khac nhau ddng vai tro nhfl trong so. Ket qua la, cac gia
tri R2 cua cac ph an doan ldn hon ddng gop m anh m e h o n vao viec tinh
toan tieu chi toi flu hoa so vdi cac phan doan nhd hon. C uoi cung, cac
nha nghien cdu cd the tap tru n g tieu chi toi flu hoa chi tren m pt bien
tiem an noi sinh quan trpng dac biet trong mo h in h dfldng dan PLS.
Vi du ve cac khai niem nghien cdu m uc tieu chinh 1& sfl hai long cua
khach hang va long tru n g th a n h cua khach h an g trong cac m o hin h hai
long cua khach hang (vi du: Fornell, Johnson, A nderson, Cha, & Bryant,
1996) va dfl dinh sd dung trong cac m o hinh chap nhan cong nghe (vi
du: A l-G ahtani, H ubona, & W ang, 2007). Tuy n h ie n , chung toi thfldng
khuyen ban nen xem xet tat ca cac bien tiem an noi sinh, sd dung tong
trpng so cua tong gia tri R2 cua m oi nhom lam tieu chi toi flu hoa.

Trong Bfldc 2 va 3, PLS-POS fldc lflpng cac m o hinh dfldng dan

PLS theo nhom cu the va sd dung cac ket qua lam dau vao de tinh toan
tieu chi toi flu hoa. U Bfldc 4, phiidng phap xac d in h m ot tap cac quan
sat dng cd de gan lai td phan doan hien tai cua chung cho m ot phan
doan thay the de cai thien tieu chi toi flu hoa. Theo Squillacciotti (2005,
2010), PLS-POS sd dung bien phap khoang each de gan lai cac quan
sat: Khoang each cua m ot quan sat tdi m ot ph an doan cang gan, kha
nang tien doan cua quan sat nay lien quan den p h an doan do cang cao.
Phiidng phap khoang each dfla tren phan dfl m o hinh cau true, dd la
sfl k h ac biet gida diem so dfl bao va thflc te cua (cac) bien tiem an noi
sinh. C hinh xac h d n , p h iidng phap khoang each s d dung sai sd dfl b ao
(prediction errors) cua cac quan sat, bang vdi b in h phiidng phan dfl m o
h in h cau true. Liiu y rang phan dfl dflpc binh p h iid n g m a khac biet ldn
h d n gida diem so dflpc dfl bao va thflc te ddng g o p m anh h d n vao sai sd
dfl bao h d n la n hdng khac biet nhd. Tdng binh phiidng phan dfl (tdc la
cac sai sd dfl bao) dai dien cho gia tri khoang each.
Trong khi diem sd bien tiem an noi sinh thflc te la ket qua td phan
tich PLS-SEM nhom cu the, cac diem sd dfl b a o can phai dflpc tinh
rieng. D ieu nay dflpc thflc hien bang each nhan cac diem sd bien tiem
an ngoai sinh cua cac quan sat vdi fldc lfldng he sd dfldng dan tfldng
 Chuong 5: Mo hinh hoa khong ddng nhat khong quan sat fluo'c 195

ilng va tinh tong cua cac tich nay. De m inh hpa viec tinh toan diem so
bien tiem an ddpc dd bao va sai so d d bao, hay xem xet vi du ve hai phan
doan co kich thddc bang nhau va ddc tin h he so d d dng dan cu the theo
ph an doan cua chung nhii diipc hien thi trong TP lieu 5.5. H dn nda, hay
xem xet mot quan sat thuoc phan doan 1 vdi diem so bien thien tiem
an (chuan) la 0,1 doi vdi bien y , 0,6 doi vPi bien Y2 va 0,5 doi voi bien
Yy De xac d inh m dc do quan sat nay dp bao len bien muc tieu Y} trong
phan doan rieng cua no, PLS-POS nhan diem so bien tiem an cua quan
sat doi vPi bien V vPi he so dddng dan p, (tdc la 0,1 • 0,7 = 0,07) cung
nh d diem bien tiem an cua quan sat doi vdi bien V, vdi he so dddng dan
p 2 (nghia la 0,6 • 0,2 = 0,12). Cong gop hai tich nay tao ra bien tiem an
dd bao cua quan sat doi vdi bien Y} la 0,07 + 0,12 = 0,19. Diem so bien
tiem an thde te cua Yy doi vdi quan sat nay la 0,5, sinh ra m ot sai so dd
bao la (0,19-0,5)2 = 0,0961. T dong dng, sai so d d bao cua quan sat nay
cho phan doan 2 ddpc sinh ra bdi [(0,1 • 0,2 + 0,6 • 0,7) -0,5]2 = 0,0036.
C hung ta thay rang quan sat co sai so d d bao cua no trong phan doan 1
cao hon so vdi phan doan 2. Noi each khac, quan sat cd kha nang d d bao
cao h o n trong phan doan 2 vi sai sd dd bao n h o h o n so vdi phan doan
1. Tong binh phdong phan dd (hoac cac sai so dd bao) dai dien cho cac
gia tri khoang each cuoi cung cua mdi quan sat doi vdi phan doan hien
tai cua no va cac phep gan phan doan thay the cd the co.

Dda tren cac gia tri khoang each cua cac quan sat, PLS-POS tao
ra mot d an h sach cac quan sat dng cd cho viec gan lai d Bddc 4 cua
th u at toan (Td lieu 5.4). Neu m ot quan sat cd khoang each ngan n h at
den phan doan rieng cua nd, quan sat nay k hong trd th an h m ot dng cd
cho viec gan lai. Ngdpc lai, neu quan sat gan vdi phan doan khac h d n
la phan doan rieng cua no (nhd ddpc giai thich doi vdi vi du trong T d
lieu 5.5), no phu hpp hon vdi phan doan dd ve kha nang dd bao, do do
196 NhCmg van de nang cao trong PLS-SEM

co kha nang cai thien tieu chi toi du hoa sau khi gan lai. Neu m ot quan
sat co khoang each ngan hon tdi nhieu phan d o an khac, phan doan co
khoang each ngan n h at se diipc xem xet trong danh sach. Sau do, cac
quan sat dng cd ddpc sap xep theo thd td khoang each giam dan. Neu
danh sach cac dng cd cho viec gan lai la trong rdng, thuat toan tien tdi
Biidc 7 va ddng lai vdi tinh toan ket qua cuoi cung (T p lieu 5.4).

Trong Bode 5, PLS-POS gan lai quan sat d au tien trong danh sach
cac dng cd vao phan doan thay the m a no cd k hoang each ngan nhat.
M ot thach thde chinh cua phddng phap nay la khong xac dinh dope
nhiem vu gan dd lieu vi khong biet ket qua PLS cu the theo nhdm se
thay doi n h d the nao sau khi quan sat dope gan lai cho m ot nhdm khac.
Vi ly do nay, thuat toan PLS-POS chi trao doi m o t quan sat tai m ot thdi
diem. Ddi vdi cac n h d m d d lieu da thay ddi, phdom g phap nay cap nhat
ket qua PLS-SEM phan doan cu the va tieu chi td i du hoa. Neu tieu chi
tdi du hoa ddpc cai thien, thuat toan duy tri viec gan lai va tien hanh
sang Bddc 6. Mat khac, neu tieu chi tdi du hoa k h o n g cai thien, thuat
toan se khong gan lai quan sat. Sau dd, d in h gia tddng td tiep tuc vdi
quan sat dn g cd tiep theo. Ket qua la, thuat toan PLS-POS chi gan lai
m ot quan sat n eu bddc nay cai th ien tieu chi tdi du hoa cu a giai phap
phan doan mdi.

Trong Bddc 6, PLS-POS kiem tra xem da d at den sd Ian lap tdi
da chda. Neu chda, thuat toan tiep tuc vdi lan lap tiep theo bat dau
vdi Bddc 2. Mac khac, th u at toan tiep tuc vdi B ddc 7, bddc cuoi cung,
tinh toan ket qua PLS-POS cua giai phap phan d o a n bao gom m d hinh
d ddng d in PLS cd the theo nhdm . Sd lan lap tdi da phai du cao de cd
ddpc giai phap gan vdi m dc tdi d u toan cue. C h u n g tdi khuyen ban nen
sd dung sd lan lap tdi da b in g sd lan lap lai cao ho n : gap 1000 lan hoac
gap hai lan sd quan sat.

& to m lai, PLS-POS tinh toan sd khong ddng nhat trong m d hinh
, Png d in PLS bang each sd dung do lddng k h o an g each tao dieu kien
cho viec gan lai cac quan sat trong nd ldc cai th ie n tieu chi tdi du hoa
th eo d jn h h d d n g d d bao. P hd o n g phap nay th d d n g ap d u n g cho tat ca
cac m d h in h d d d n g d in PLS bat ke p h d o n g th u 'c m d h in h do lddng
(de biet them chi tiet, xem Becker, Rai, Ringle, & Volckner, 2013), phan
phoi dfl lieu, hoac do phde tap cua m d hinh cau triic. Giong n h d thuat
toan EM tro n g FIMIX-PLS, PLS-POS ddi mat vdi van de m a nd co the
d in den m ot giai phap tdi du cue bo do sfl dung p h d d n g phap leo ddi
(Hill-climbitig). Do do, neu khong xay ddng tren p h a n vung khdi dong
 Chirnng 5: Mo hinh hoa khong dong nhit khong quan sat dirge 197

FIMIX-PLS, m ot dng dung lap lai cua PLS-POS (vi du, 10 lain tren so
phan doan diipc xac dinh triiPc) vdi cac phan vung khdi dong khac
nhau diipc khuyen khich.

Bi/dc 4: G iai th ich cau true phan doan tiem an

Sau khi hoan th an h phan tich, PLS-POS cung cap cho ngiidi dung
tdng th anh vien nhom quan sat cung vdi cac fldc liipng m o hinh phan
doan cu the cho cac m o hinh do liidng va cau true. Vi cac phan doan
diipc xac dinh la - theo din h nghia - tiem an, cac fldc ldpng m o hinh
khong giai thich nhdng gi tao nen cac phan doan tiem an nay. Viec
chuyen cac ket qua ban dau thanh sd hieu biet cd the hanh dong yeu cau
nha nghien cdu giai thich cac phan doan theo cac bien e<3 th e quan sat
va cd y nghia ve mat quan tri. De lam n h d vay, cac n h a nghien cdu can
chuyen cau tru e phan doan tiem an sang cau tru e quan sat diipc bang
viec p h an vung d d lieu bang each sd dung m ot hoac nhieu (cac) bien
giai thich. Phan vung ket qua cua d d lieu p h an ldn phai td d n g dng vdi
nhom dd p c PLS-POS tao ra. Phan tich n h d vay cung d d p c goi la p h a n
tich hau n g h iem (ex p o st analysis).

Phan tich hau nghiem cd the lien quan den viec chay nhi phan
(trong trd d n g h pp hai phan doan tiem an) hoac da thdc (trong triidng
hpp cd ba hoac n hieu phan doan tiem an) hdi quy logistic vdi lien ket
phan doan n h d bien phu thuoc va bien giai thich tiem nang n h d bien doc
lap (vi du: Money, Hillenbrand, Henseler, & Da Camara, 2012; W ilden
& Gudergan, 2015). So sanh sd chenh lech log cho phep xac din h cac
bien doc lap cd tac dong m anh nhat den cau true phan doan tiem an. Sd
dung cac bang cheo, cd the tddng phan phan pho'i tan so cua bien lien
ket phan doan PLS-POS vdi m ot hoac nhieu bien giai thich. T d lieu 5.6
trin h b&y m ot vi du ve bang cheo cho 400 quan sat, hai phan doan PLS-
POS va bien giai thich gidi tinh gender. Tan suat cao cho thay phan ldn
ngddi tra ldi tro n g phan doan PLS-POS 1 la nd. Ngdpc lai, phan doan
PLS-POS 2 chu yeu bao gom nhdng ngddi tra ldi la nam gidi. C hinh xac
hdn, bien gidi tinh gender giai thich 188 + 177 = 365 cua 400 phep gan
phan doan PLS-POS, la mot chong cheo cua 365/400 = 91,25%. N gddc
lai, neu mdi o cd 100 ngddi tra ldi, gidi tinh se khong phu h d p de giai
thich ve phan doan PLS-POS. Noi chung, sd phan bo cang ngang nhau
gida cac o, thi bien giai thich it hdu ich hdn va ngddc lai.

Cac nha nghien cdu cung da ru t ra cac thuat toan xay d d n g cay
(tree-building) nhd Chi-squared A utom atic Interaction D etector
198 Nhdng v in d§ nang cao trong PLS-SEM

(CHAID) hoac cay phan loai va hoi quy (CART) de xac d in h cac bien
giai- thich-phu-hqp-(vi-du; Ringle-va cdngsd,.2QL0; Sarstedt & Ringle,
2010). Cac ky th uat nay co the d u q c sii dung de d d bao cac gia tri cua
m ot bien ph an loai (trong trUdng hop cua chung ta la cac lien ket phan
do an tiem an cua quan sat) tif m ot hoac nhieu bien d d bao lien tuc va /
hoac bien ph an loai. Ca hai ky thuat CHAID va CART se xay ddng cac
cay, trong do m oi giao diem (khong gidi han - nonterm inal) xac dinh
m ot dieu kien phan chia, de tao ra m ot p h an loai toi Uu. Sau do, cac giao
diem ket qua cd the duqc sd dung de xac din h cac ket hop cua cac bien
giai thich m a the hien m ot sd chdng cheo ldn vdi phan doan ldp tiem an.

Td lieu 5.6 HHHf U bang H H rm & iab) pSBlll 1

Fem ale M ale Sum
PLS-POS segm ent 1 188 (47,00%) 14(3,50%) 202 (50,50%)
PLS-POS segm ent 2 21 (5,25%) 177 (44,25%) 198(49,50%)
Sum 209 (52,25%) 191 (47,75%) 400(100,00%)

Q uan trong hon, de chay th an h cong m ot ph an tich hau nghiem ,

cac nha nghien cdu phai cd kha n an g xem xet m ot loat cac dac diem
quan sat cd the phuc vu nhd la dau vao cd the. Viec kiem tra qua it dac
diem quan sat han che kha nang tai tao phan vung PLS-POS cua nha
nghien cdu. Vdi y nghl n&y, cac nha nghien cdu n en dan h gia lieu m ot
bien giai thich d o n le hay m ot tap cac bien cd y ng h la ly thuyet ve viec
lam sang td sd khac biet cd the cd trong cac he so d d d n g dan gida cac
phan doan da xac dinh. Do dd, d an h gia vai trd giai thich cua cac bien
cd the, de phan tich ldp tiem an cd the ddqc thde h ien hoan chinh hdn,
nen ddqc xem xet trong giai doan thiet ke nghien cdu khi th u thap cac
bien m o ta hoac cac bien khac cd the quan trong.

Tuy nhien, viec tai tao phan vung PLS-POS van la m ot nhiem vu rat
kho khan, vi cac dac diem quan sat thddng khong p h u hdp vdi cau true
ph an doan tiem an. Trong bdi canh nay, sd chdng cheo 60% gida phan
vung PLS-POS va m ot phan vung dd q c tao ra bdi (cac) bien giai thich
cd the dddc coi la thda dang.

Butic 5: L/dc lddng ca c mo hinh phan doan cu th e

Khi nha nghien cdu da xac dinh m ot hoac n h ie u bien giai thich
p hu h dp vdi phan vung PLS-POS, bddc cuoi cung la ddc lddng cac md
h inh phan doan cu the nhd ddqc chi dinh bdi (cac) bien giai thich. Khi
 Chirong 5: Mo hinh hoa khong Pong nhat khong quan sat Pupc 199

lam nhii vay, nha nghien ciiu phai dam bao rang tait ca cac do liidng m o
hinh dap dng cac tieu chuan chat liipng chung nhii diipc ghi trong vi
du H air va cong sd (2017). Nhufng phan tich nay h o an thanh phan tich
Idp tiem an. Tuy nhien, cac phan tich sau hon co the lien quan den viec
kiem dinh sii khac biet bang so gida cac he sd diidng dan cu the theo
phan do an cung cd khac biet dang ke bang p h d d n g phap phan tich da
nhdm hay khong, bao gom kiem dinh bat bien d o liidng nhii diipc m d
ta trong ChdPng 4. T d lieu 5.7 to m ta t cac quy tac phan tich ld p tiem an
trong PLS-SEM.

l i t lieu 5 .7
Cat d a t th u a t todn FIM IX-PLS:
• Sd dung tieu chuan diing cua I-10 s va sd ldpng tdi da 5000 vong
lap.
« Sd dung giai phap tot nhat trong 10 lan lap lai de tran h hoi tu b m dc
tdi du cue bo (dia phiiong).
• Khi cac bien quan sat cd gia tri khuyet, hay sd dung chdc nang xoa
toan bo (casewise delection). K hong sd dung thay the gia tri tru n g
b inh hoac bat ky phddng phap tinh nao khac.
• De xac dinh pham vi sd phan doan hpp ly, hay sd dung m ot phan
doan lam gidi han dddi va sd nguyen ldn nhat khi chia kich thddc
m au cho kich thiidc m iu phan doan tdi thieu lam gidi han tren.
Xac d inh so htdng p h a n doan can g ift lai:
• T ieu chi thong tin: Neu AIC} va CAIC chi ra cung m ot sd ldpng
phan doan, hay chon giai phap nay. Ngoai ra, cung xem xet AIC3
va BIC. Cung xem xet sd phan doan nhd dd p c chi ra bdi AIC4 va
BIC. Noi chung, hay chon it phan doan hon dd p c chi ra bdi AIC va
nhieu phan doan hon so vdi chi djnh cua MDL5.
• EN tot nhat nen cao hon 0,5.
• Dam bao rang kich thddc phan doan dap dng cac yeu cau kich
th ddc m au tdi thieu. Neu khong dap dng, hay giam sd ldpng phan
doan hoac loai bo cac phan doan khong lien quan va tap trung vao
cac phan do an con lai.
• C an nhac thde te trong viec tin h toan. Neu co the, hay de thong tin
va ly thuyet du tien ddng gop vao cac phan doan thde sd. Dam bao
rang giai phap cd lien quan den quan trj.
200 Nhdng van nang cao trong PLS-SEM

• •rse jg sn jB g

Tu1lieu 5.7 Qui tac kinh nghietn doi vdi phan tich ldp tiem
C ai d a t th u d l toan PLS-POS:
• Sii dung cac ket qua FIMIX-PLS de xac djn h so lddng phan doan
can gid lai tii dii lieu. Xem xet lay giai phap FIMIX-PLS lam phan
vung k h a i dong cho PLS-POS.
• Khong sd dung tuy chon tien phan doan khi sd dung giai phap
FIM IX-PLS lam phan vung khdi dong cho PLS-POS.
• Khi sd d u n g viec gan ngJu nhien lam phan vung khdi dong, hay sd
dung giai phap tot n h a t trong 10 lan lap lai de tran h hqi ty d m dc
Hq i tu
tdi du cu e bo.
• Chon so lan lap cao tdi da m a thuat toan PLS-POS se thdc hien.
C o n sd nay phai bang vdi sd liiong cao hdn sau day: gap 1000 lan
hoac hai lan sd lan quan sat.
• Chieu sau tim kiem phai b ln g sd quan sat.
• Xem xet sd dung tdng trong sd cua tat ca cac gia tri R2 cua cac bien
nghien cd u lam tieu chi tdi tiu hoa.
P h a n tich ha u nghiem:
• Phan vung dd lieu b&ng each sd dung m ot bien giai thich, hoac ket
h ap m ot sd bien giai thich, tao ra m ot nhom d d lieu m a phan ldn
td d n g d ng vdi d d lieu ddqc tao ra bdi PLS-POS.
• Sd chdng cheo 60% gida phan vung PLS-POS va m ot phan vung
ddqc tao bdi (cac) bien giai thich dddc coi la thda dang. Xem xet
viec chay phan tich da nhom PLS sau khi th iet lap (tdng phan)
phep bat bien do lddng.

TINH HUONG NG HIEN CUD M IN H HQA - PHAN TICH LOfP

EM AN

Bddc 1: Chay thu tuc FIMIX-PLS

M inh hoa cua chung tdi ve phan tich ldp tiem an d d d c lay m ot phan td
nghien cd u tin h huong cua M atthews, Sarstedt, H a ir va Ringle (2016) ve
viec sd dung HM1X-PLS, cung d d a tren md h in h m d rong ve danh tieng
cua cong ty. Bdi vi cac van de co the phat sinh khi s d dung phep quy ket
trung binh de xd ly dd lieu khuyet, chung tdi da x o a tat ca cac quan sat
cd gia tri khuyet trong giai doan kiem tra dd lieu (td c la: xoa casewise)
trd d c khi ph an tich thdc te. Sau khi xoa bo, 336 q u a n sat con gid lai.
 Chucrng 5: Mo hinh hoa khong dong nhat khong quan sdt du’gc 201

Tap dfl lieu da cat giam nay da dddc dfla vao SmartPLS Project
vcfj ten Corporate Reputation - Advanced PLS-SEM Book, chung toi
sfl dung cho tat ca cac m inh hoa nghien cflu tinh huong trong cuon
sach nay (xem C hddng 1) va co tai www.pls-sem .com . D ieu h d d n g den
cfla so SmartPLS Project d phia ben tay trai. N h ap chuot phai vao tap
dfl lieu Corporate reputation data casewise deletion [336 records] va
chon Select Active Data File tfl trin h ddn (Tfl lieu 5.8). T iep theo, bam
dup vao Extended model, va m o hin h dddng dan PLS m d rpng cho vi
d u danh tieng cua cong ty se m d ra trong cfla so m o h inh h o a SmartPLS.
Lflu y rang ban co dddc ket qua td d n g tfl khi sfl dung tap dfl lieu day
du vdi 344 quan sat va Ida chon Casewise D eletion trong the M issing
Values trong hop thoai bat dau khi chay thuat toan ph an doan. Tuy
nhien, de m uc dich m inh hoa tot hdn, chung toi sfl dung Corporate
reputation data reduced vdi 336 quan sat.

Tit lieu 5.8 lyta chon t|pixjaf lieu da id H

Project Explorer ea § § £ ?

•a Q Corporate Reputation - Advanced PLS-SEM Book

J l , Extended m odel
H C M (reflective-formative)
H C M (reflective-reflective)
Inadm issible H C M (reflective-formative)
Sim ple m odel
| Corporate reputation data case wise dele Won [336 records]
(O Rename F2

% C opy Ctrl+C

tf Paste Ctrl+V

Duplicate Ctrl+D

0 Delete Delete

Select A ctiv e Data File

Q Create New Project

Create New Path M odel

pi"] Im port Project from Backup File

In<
pi"] Im port Projects from a Folder

p l"l Im port Data File

pl"| Im port Sam ple Projects

202 NhCmg vSn d l nSng cao trong PLS-SEM

De bat dau phan tich FIMIX-PLS, hay nhap vao Calculate Finite
Mixture-(FIMIX) Segm entation trong th an h .trin h ddn. Ngoai-ra, ban
co th£ nhap chuot trai vao bieu tiidng banh xe tren th an h cong cu va
chon tuy chpn tiid n g dng tro n g hop ket h d p m d ra. Sau khi chpn tuy
chpn FIMIX-PLS, hop thoai tro n g Tii lieu 5.9 x uat hien. Doi vdi phan
tich ban dau, bat dau vdi giai phap m ot phan doan va sd dung cai dat
m ac d in h cho tieu chi ddng (l»10's), so vong lap toi da (5000) va so
lan lap lai (10). Hop thoai cd them hai tab de xac dinh cac thiet lap
th u at toan PLS-SEM chuan va viec x d ly cac gia tri khuyet. Sd dung
cai dat m ac d inh cho thuat toan PLS-SEM. Cuoi cung, nhap vao Start
Calculation.

Number of Segments ;\

Maximum Iterations

Stop Criterion

— Advanced Settings
" IP

O Use Unstandardized Latent Variable! Scores?

0 Estimate Regression Intercept?

— R epetitions-----------------------
FIMIX-PLS can be executed several times and selects the solution with the best Lnl value to avoid local
optima.

Number of Repetitions 10
_

sa u khi hoi tu, SmartPLS 3 m d ra m ot tab bao cao m di, cho thay ket
qua phan tich FIMIX-PLS. Triidc khi phan tich c h i tiet ket qua, chung
ta can chay lai FIMIX-PLS cho cac giai phap phan d o an cao hctn (vi du:
td 2 den 5 ph an doan). De xac din h gidi han tren c u a day cac giai phap
p h an doan can xem xet, hay kiem tra cac yeu cau kich th d d c m au toi
th ieu nh ii dd d c quy dinh trong H air va cong sd (2017). Vdi so liidng
toi da 8 dau m ui ten chi vao bat ky khai niem n g h ien cdu nao trong
m o h inh (cac bien quan sat nguyen n h an cua Q U A L) va gia dinh m dc y
nghia 5%, cung n h d R2 toi thieu bang 0,25, chdng t a can 54 quan sat de
tide liidng tin cay m d hinh. S6 nguyen ldn nhat td viec chia kich thddc
m au (tdc la 336) chd kich thiidc m au toi thieu (tdc la 54) tao ra gidi han
 Chiro’ng 5: Mo hinh hoa khong ddng nhat khong quan sat dirge 203

tren ly thuyet la 6,22 = 6. Tuy nhien, do do phdc tap cua m o hinh, chung
ta chi xem xet m ot den nam giai phap p h an doan, dac biet la do sd p h an
bo d o n g deu cac qu an sat - can thiet de dap dng cac yeu cau kich thddc
m au toi thieu - rat kho xay ra. Do do, chung ta chay lai FIMIX-PLS cho
hai den nam phan doan, sd dung cac cai dat thuat toan d tren va lUu bao
cao ket qua cho moi lan chay.

Bi/dc 2: Xac djnh so lddng phan doan

De xac dinh so lddng phan doan can gid lai td dd lieu, chung ta can kiem
tra cac chi so phu hdp m a ban co the tim thay trong Q uality Criteria ->
Fit Indices trong moi bao cao ket qua FIMIX-PLS. De tao thuan ldi cho
viec so sanh gida cac giai phap so lddng phan doan khac nhau, ban co
the xuat cac gia tri sang chddng trinh bang tinh n h d M icrosoft Excel.
De lam nhd vay, hay nhap vao Export to clipboard: CSV trong tab Fit
Indices va dan cac gia tri vao tep Excel. Td lieu 5 .10 c u n g c a p tong quan
ve cac gia tri log-likelihood (L n L ),c a c tie u c h ith o n g tin va EN cho giai
ph ap 1 d en 5 p h an doan.

Tu" lieu 5.10 Chi bao phD hq

S S o hidng p h a n doan
Tieu chi 1 3 4 5
s j v
LnL -1406,86 £ 1353,83 -1319,79 -1290,62 -1261,28
A IC 2777,66 2745,58 2723,20 2700,57
2 8 4 7 ,2 V
a ic 3 2 8 ,4 - 2812,66 2798,58 2794,24 2789,57
a ic 4 2881,72 2847,66 2851,58 2865,24 2878,57
^2912,61 2911,26 2947,89 2994,25 3040,29
b' K
C AIC 2929,61 2946,26 3000,89 3065,25 3129,29
m dl5 3308,18 3725,65 4181,11 4646,31 5111,18
EN n/a 0,43 0,64 0,67 0,70

Hai khia canh dang ghi nhd: Thd nhat, cac gia tri cua ban se khac
vdi cac gia tri ddoc m o ta trong T d lieu 5.10, vi FIMIX-PLS khdi tao
ngau nhien va do do tao ra cac ket qua khac nhau m oi khi no ddoc chay.
Tuy nhien, ket luan khong nen khac biet ve mat so ldong cac phan doan
can gid lai. Thd hai, hay nhd rang doi vdi moi phep do phu hop, giai
phap toi du la so p h an doan co gia tri thap n h at (xem so in dam trong
T d lieu 5.10), ngoai trd EN, trong do gia tri cao h d n cho biet sd phan
tach cac phan doan tot hon.
204 NhOng van de nang cao trong PLS-SEM

Ih a t khong may, AIC3 va CAIC khong chi ra cung m ot so liidng

p h a n doanTva<-a-AIC3va-Bi€-eSng4ch6ngv-A4G-tho thay m ot giai phap
nam phan doan, cho thay rang so liidng chinh xac co the thap hdn dieu
nay. Mat khac, CAIC va dac biet la M DLS, ca hai ket qua trong giai phap
m ot phan doan, cho thay rang hai hoac nhieu phan doan can diidc xem
xet. Phan tich sau h d n cho thay hai tieu chi k h ac co h ie u suat tot trong
viec phat hien m ot so phan doan thich hdp, AIC, va BIC, ca hai chi ra
hai phan doan, do do cung cap h o trd ban dau cho giai phap nay. Tuy
n h ien , giai p h ap hai phan doan the hien gia tri EN diidi 0,50, cho thay
rang hai phan doan khong diidc tach biet tot.

doan lien q u ari

Ktch cd p h a n doan lien
S d lddng P han P han Phan Phan P han
p ha n doan doan 1 doan 2 doan % doan 4 doan 5
2
3
0,515
0,520
0,485
0,412 0,068
)
4 0,526 0,354 0,085 0,035
5 0,477 0,282 0,106 0,093 0,043

Trong bao cao ket qua, vao F inal Results -» Segm ent Sizes,
SmartPLS 3 cho thay cac kich thiidc phan doan tiid n g doi cua giai phap
FIMIX-PLS doi vdi m ot so phan d o an nhat dinh diidc xac dinh triidc.
Kiem tra kich thiidc phan doan tiidng doi tren cac giai phap FIMIX-PLS
cho cac so ph an doan khac nhau, nhii diidc tom ta t trong Tii lieu 5.11,
cho thay r&ng viec chon nhieu h d n hai phan doan k h o n g hop ly. Liiu y
rang kich thiidc cac phan doan nay khong phai la ket qua cua viec phan
cum m anh m e cac quan sat diia tren xac suat toi da c u a thanh vien phan
doan m a tii cac hoi quy binh phddng nho nhat co tro n g so sd dung cac
xac suat nay lam dau vao. Vi du: doi vdi giai phap ba phan doan, tach
kich cd phan doan thanh phan doan 1: 52% (52% c u a 336 = 175), phan
doan 2: 41,2% (41,2% cua 336 - 138) va phan d o a n 3: chi 6,8% (6,8%
cua 336 = 23). Co the thay, vdi 23 q u an sat, phan d o an 3 q u a n h d de
phan tich PLS-SEM phan doan cu the. Cach giai quyet van de nay la
xem xet giam p h a n doan th d ba vi n d qua n h d de d a m bao tin h gia tri
phan tich, va thay vao do tap tru n g vao phan tich v a dien giai hai phan
doan ldn hdn.
 Chuang 5: Mo hinh hda khong ddng nhat khong quan sat dugc 205

Bao cao ket qua cua SmartPLS 3 tra ve cac gia tri R2 cua giai phap
FIMIX-PLS trong Quality Criteria -> R Square. Tii lieu 5.12 cho thay
cac gia tri R 2 phan doan cu the cung vdi cac gia tri R2 co tro n g so la ket
qua cua viec nhan cac gia tri phan doan cu the vdi cac kich thiidc phan
doan tiidng doi. Gia tri R2 trong phan doan 1 cao hdn dang ke so vdi tap
dQ lieu day dii, trong khi cac gia tri trong phan doan 2 nhd h d n . Chung
ta cung thay rang gia tri R2 trung binh cd trong so cua giai phap hai phan
doan FIMIX-PLS chi cao hdn m ot chut so vdi gia tri cua to an bo dtit
lieu. N hiing ket qua nay chdng m inh mdi quan tam cua chung ta ve giai
phap phan doan FIMIX-PLS: C hung ta cd the cd m ot m iu cd kha nang
ddng nhat. M at khac, khi kiem tra cac he sd diidng dan p h an doan cu
the FIMIX-PLS tro ng Final Results -> Path Coefficients, ch u n g ta thay
rang m ot sd khac biet cd the tiidng doi (kich thiidc cua chung) va quan
tam doi vdi viec h in h thanh va dien giai cac nhom d d lieu khac nhau. Vi
du, cac cddng do cua he sd diidng d in chuan hoa td CUSA den CUSL
khac nhau hd n 0,15 diem trong giai phap hai phan doan.

;ia trj FIMIX-F’LS doi vdi Gi;ai uh&o 2 -

■ a ^ doan » \ v y . '
Cac gia tri^E ^J Cac gia tri
R 2 trung binh
Toan bo FIMIX-PLS FIMIX-PLS co trong so
tap dd lieu Phan doan 1 Phan doan 2 FIMIX-PLS
COM P 0,629 0,812 0,519 0,670
CUSA 0,293 0,530 0,164 0,352
CUSL 0,5 0,807 0,389 0,604
LIKE 0,557 0,814 0361 0,594

N hin chung, ket hdp n h dng ket qua nay cho thay tinh khong ddng
nhat cd the khong pho bien trong nghien cdu cua chdng ta. Cac chi sd
phu hop m ang lai ket qua m d ho ve sd ldong phan doan de g id lai td dd
lieu trong khi kich th d d c phan doan td o n g doi chi cho phep giai phap
2-phan doan. Viec kiem tra cac gia tri R2 phan doan cu the h d trd ket
luan nay vi gia tri R2 trung binh cd trong sd chi cao hdn m ot chut so vdi
gia tri R2 dddc tao ra tren tap dd lieu day du. Chi cd he sd d d d n g d in cu
the cua phan doan hda hen m ot sd khac biet nhdng tddng doi nhd. Do
dd, viec sd dung cac phan doan khac nhau de phan tich sau h d n cd the
khong cd ldi. Tuy nhien vdi m uc dich m inh hoa de phac thao each PLS-
POS cd the dddc ap dung, chung ta xem xet chi tiet h d n ve giai phap
2-phan doan bang each sd dung PLS-POS.
 206 Nhung van <Jd nang cao trong PLS-SEM

BUtfc 3: Chay thu tuc PLS-POS

Trong bddc’tiep theo,"chung ta chay thutuc~PLS-POS bang each nhap

vao Calculate Prediction-O riented Segm entation (POS) trong
thanh m enu. Ngoai ra, ban co the nhap chuot trai vao bieu tdpng banh
xe tren th an h cong cu va chpn tuy chon tiidng ufng trong hop ket hpp
m d ra. Sau do, hop thoai trong TP lieu 5.13 xuat hien. Doi vdi so ldpng
cua Groups, chon 2 — n h d diipc bieu thi bdi phan tich FIMIX-PLS
triidc dd. Vi tap hpp dui lieu cd 336 quan sat sau khi xoa bo hang loat
va 2-336 = 672, thap hon 1000, hay chpn gia tri m ac din h la 1000 cho
M axim um Num ber o f Iterations. Search D epth phai bang sd quan sat,
la 336. N eu gia tri m ac din h la 1000 khong thay doi, SmartPLS tp dong
ddng khi 336 da dat diipc vi khong the tim kiem viipt qua so ldpng quan
sat tdi da trong tap dd lieu. Trong Advanced Settings, chung ta chpn
FIMIX- PLS Segm entation thay vi tuy chpn Random A ssignm ent
doi vdi Initial Separation trong PLS-POS va k hong chpn ap dung tuy
chpn Pre-Segm entation. H on nuta, chung ta k h o n g m uon tap tru n g
vao m ot bien tiem an noi sinh d u y n h at tro n g m d h in h dan h tieng cua
cong ty n h d CUSL. Thay vao do, chung ta m u o n toi da hoa giai thich
nhom cu the cua tat ca cac bien tiem an npi sinh cung m ot luc. Do dd,
chung ta chpn Sum o f all Construct W eighted R-Squares cd chdc nang
O ptim ization Criterion. Cuoi cung, nhap vao Start Calculation.

S etup P a r tia l Least Squares Finite Mature (FJMDQ Segmentation (J) Missing Values

— Basic S e ttin g s -------

O ,
$ K
Groups

M a x im u m iterations
T

‘ jq q q
f ||

jppjl

Search Depth j 1000

— Advanced S e ttin g s — -
Initial Separation 0 R a n d o m A ssig n m e n t <§) FIM IX S e g m e n ta tio n

P re -Se g m e n ta tion |p] Use Pre -S e g m e n ta tio n in th e First Rou n d

O p tim iza tio n Criterion 0 S u m of all C o n stru ct R-Squares

0 S u m of Ta rg e t C o n s tru c t R -S q u are
( § ) S u m of all C o n stru ct W e ig h te d R-Squares
0 S u m o f Ta rg e t C o n stru ct W e ig h te d R -S q u a re
 Chuong 5: Mo hinh hoa khong dbng nhat khong quan sat du’o'c 207

Sau khi hoi tu, SmartPLS 3 m d m ot tab Report mdi, hien thi ket
qua phan tich PLS-POS. Trade khi p h an tich chi tiet ket qua, chung
ta can chay lai PLS-POS de thu diiqc 10 giai phap. Trong Report ket
qua PLS-POS, hay chuyen den Q uality Criteria -> Change in Objective
Criterion va chon giai phap tot nhat (tdc la ket qua gia tri m uc tieu cao
nhat). Doi vdi giai phap nay, chung ta bat dau phan tich ket qua vdi gia
tri R2 — bang each dieu hiidng den to Q uality Criteria -» R Square —
va kich thddc phan doan (tiidng doi) bang each di tdi Final Results -5>
Segm ent Sizes trong Report. Tii lieu 5.14 tom hit ket qua PLS-POS cua
giai phap 2 phan doan. Trong khi phan doan 1 (kich th d d c p h an doan
tiidng doi: 56,8%) tra ve gia tri R2cao hdn dang ke, ket qua doi vdi phan
d o an 2 (kich thddc phan doan td d n g doi: 43,2%) d m dc td d n g td v a c a o
h d n so vdi ket qua doi vdi tap dd lieu day du. Q u a n trong hdn, gia tri R2
tru n g binh co trong so cua giai phap PLS-POS cho hai phan doan cao
h d n dang ke so vdi gia tri R2 cua tap d d lieu day du. Ket qua nay la rat
m ong m uon cho mot giai phap phan doan.

Td lieu 5.15 cho thay cac he so diidng dan cu the theo phan doan
td phan tich PLS-POS. Chung ta thay sii khac biet dang ke trong cac
ildc lddng he sd diidng d in gida hai nhom . Vi d u , trong khi COM P co
tac dong diidng ro ret len GUSL trong phan doan th d nhat (0,242), tac
dong tiidng tii la am trong phan doan th d hai (-0,302). Ngiidc lai, trong
tap d d lie u day du, he sd diidng dan tddng dng gan bang khong (0,011).

TU lieu 5.14 ia p phdn doan PLS-POS

Cac gia tri R 2 Cac gia tri R 2

trung binh
Toan bo PLS-POS PLS-POS co trong so
tap dd lieu Phan doan 1 Phan doan 2 PLS-POS

CO M P 0,629 0,770 0,618 0,704

CUSA 0,293 0,521 0,421 0,478

J
CUSL 0,562 0,706 0,555 0,641

LIK E 0,557 0,631 0,657 0,642

208 Nhung v irt de nang cao trong PLS-SEM

T lflifu 5 .1 5 | J g
fl'ts -I 'o s * " t(iin m ''*U ” 2 f>i,Sn
Toan bo Phan Phan Phan doan 1-
tap dd lieu doan 1 doan 2 Phan doan 2*
A TTR * COMP 0,085 0,206 -0,115 0,321
A T T R * LIKE 0,159 0,144 0,113 0,031
C O M P * CUSA 0,135 0,673 -0,438 1,112
C OM P * CUSL 0,011 0,242 -0,302 0,544
CSOR * C O M P 0,057 -0,129 0,194 -0,323
CSOR * LIKE 0,190 0,267 0,168 0,099
CUSA * CUSL 0,510 0,586 0,1 0,405
LIK E * CUSA 0,445 0,069 0,808 -0,739
L IK E * CUSL 0,334 0,096 0,747 -0,651
PERF * C OM P 0,295 0 ,3 9 7 \! $ 1 7 5 0,223
PERF * LIKE 0,116 0,267 -0,114 0,381
QUAL * CO M P 0,431 0,455 0,588 -0,133
QUAL * LIKE 0,378 t 0,223 0,692 -0,470
* T hay v i sU k h a c biet tu y e t d o i d itd c s it d u n g th u d n g x u y e n , chung ta s d d u n g hieu so
s it k h a c biet cac h e so d iid n g d d n cu the cua p h a n doan I tr it cho p h a n d o a n 2. Ket
q u a la, sit k h a c biet m a n g d a u dU dng chi ra rang he so d itd n g d d n cua p h a n d o a n I
ldn h d n p h a n d o a n 2 v a ngddc lai.

Butic 4: G iai th ich cau true phan doan tiem an

De co the giai thich cac phan doan dddi goc do cac bien quan sat va co y
nghia thdc te, chung ta chay tiep phan tich hau nghiem de xac dinh m ot
hoac nhieu bien giai thich phu hop vdi phan v u n g PLS-POS theo each
tot nhat co the. De bat dau, dau tien chung ta p h a i chuyen phan vung
n h d diiqc chi ra bdi PLS-POS tdi tap dd lieu goc. De lam n h d vay, hay
m d tap d d lieu g6c vdi 336 quan sat trong M icrosoft Excel hoac m ot so
ph an m em bang tinh khac. Tiep theo, quay lai bao cao Results SmartPLS
cua PLS-POS va hien th i Final Results * Group A ssignm ent. Sd dung
tuy chon Copy to Clipboard -> Excel Format d goc tren ben phai cua
giao dien ket qua va dan gan nhom vao trang tin h M icrosoft Excel da sd
dung trddc dd (cot trong tiep theo d phia ben phai d d lieu cua ban). Xoa
cot vdi th o n g tin “unit” va doi ten bien “group” th a n h “PLS-POS group”.
Ket qua se giong nhd T d lieu 5.16. Ldu tap d d lieu vdi cai ten trd c quan
 Chiccmg 5: Mo hinh hoa khong dbng nhlt khong quan Seit (Ju’p’c 209

(vi du: “C orporate reputation data 336 pls-pos”) d dinh dang d d lieu
(csv) gia tri dddc tach nhau bang dau phay.

Bay gid chung ta co the sd dung tap dd lieu nay de so sanh phan
vung PLS-POS vdi cac phan doan diidc chi ra bdi cac bien quan sat khac
trong tap dd lieu. That khongm ay, tap d d lieu danh tieng cua cong ty chi
co hai bien n h d vay, cho biet nha cung cap dich vu cua tdng ngddi tra
ldi (nha cung cap dich vu 1-4) va loai dich vu m a ngddi tra ldi sd dung
(tra trddc hoac h o p dong). D o do, co hoi tai tao phan vung PLS-POS
day du la td o n g doi thap. Chung toi dac biet khuyen ban nen phat trien
bang cau hoi cua m inh, bao gom nhieu bien giai thich tiem nang nhat
co the de tan g co hoi tai tao day du cac ph an doan tiem an trong phan
tich hau nghiem . / V

ptf^Th6ii
Spi
PLS-POS
cusa sw itch_l switch_2 sw itch_4
groups
5 3 1 ( 2 2
7 5 ,J 4 4 1
6 4 2 3 2
6 3 4 2 2
6 3 1 5 4 2
6 3 2 4 4 2

7 1 3 5 4 1

1 < 1 5 3 2
5 5 4 1

6 5 3 5 4 2

3 5 5 5 4 1
4 3 4 3 2 1
4 2 2 3 1 1
5 4 3 4 3 1
5 4 4 1 3 1

N hd m dt m inh hoa, trong dng dung nay, chung toi sd dung bang
cheo (crosstabs) don gian de so sanh phan vung PLS-POS vdi phan
vung ddOc tao ra bdi cac bien serviceprovider va servicetype. Vdi m uc
210 Nhdng van de nang cao trong PLS-SEM

dich nay, chung toi khuyen ban n en sii dung m ot chiidng trin h thong
ke nhii-IBM SPSS Statistics-, -rmHep-duPhfu-mcf-rpng (vi du, C orporate
reputation data 336 pls-pos.csv) va sii dung tuy chon phan tich bang
cheo cho cac bien serviceprovider va nhom PLS-POS (xem Sarstedt &
Mooi, 2014, de m inh hoa cac bang cheo sii dung IBM SPSS Statistics).
Tii lieu 5.17 trin h bay ket qua phan tich nay.

viirig PLS-POS va Nha cu m g

Tfl lieu 5 .1 7
f i S I t
Nha cung cap Cac phan doan PLS-POS
dich vu Tong
(Service Provider) 1 2

I 81 (24,11%) 47 (13,99%) 128 (38,10%)

2 71 (21,13%) 52(15,48% ) 123 (36,61%)

3 22 (6,55%) 21 (6,25%) 43 (12,80%)

4 17 (5,06%) A a 0 7 ,4 4 % ) 42 (12,50%)

Tdng 191 (56,85%) 145 (43,15%) 336

So sanh so liipng c6t, chung ta thay rang ket q u a tot nhat dat diipc
khi gan ngiidi tra ldi sii dung nha cung cap dich vu 1, 2 hoac 3 cho phan
doan PLS-POS 1 va nhflng ngiidi sfl dung nha cu n g cap dich vu 4 cho
phan doan PLS-POS 2. Bang each sfl dung viec gop nhom , (81 + 71 +
22 + 25) / 1336 = 59,2% so ngiidi diipc hoi phu h p p vdi phan vung PLS-
POS. Tuy n h ie n , vdi n h a cung cap d jch vu 4 p h a n d o an PLS-POS 2 chi
co 42 q u a n sat. Vi m ot m au co 42 q u an sat la k h o n g dfl de ddc liipng mo
h inh d iid n g d in PLS cho p h an doan 2 m ot each d an g tin cay, chung ta
cung gan ngfldi dung cfla n h a cung cap djch vu 3 ch o p h an doan 2 bdi vi
n ha cung cap dich vu 3 gan nhfl cung so liipng ngiidi tra ldi thuoc phan
doan PLS-POS 1 va 2. Do dd, chung ta xem xet n h fln g ngiidi tra ldi sfl
dung nha cung cap dich vu 1 hoac 2 la phu hpp vdi phan d o an tiem an
d au tien, trong k h i ngiidi dung n h a cung cap djch vu 3 va 4 la phu hpp
vdi p h an do an tiem an thd hai. Viec gop nhdm nay cung h p p ly tfl quan
diem noi dung vi cac n h a cung cap djch vu 1 va 2 dai d ien cho hai n h a
cung cap vien th o n g ldn da quoc gia trong khi cac n h a cung cap dich vu
3 va 4 la cac n h a cung cap djch vu vien thong nhd hon chi hoat dong d
m ot so quoc gia.
 Chuang 5: Mo hinh hoa khong dong nhit khong quan sat <Ju>gc 211

Bang each sd dung viec gop nhom , (81 + 71 + 21 + 25) / 336 =

59,2% sd ngObi dope hoi phu hop vdi phan vung PLS-POS. Mac du
chong cheo chi thap hon m ot chut so vdi gia tri ngildng 60%, ket qua
nay khong phai la rat thda dang. Ben canh sd chong cheo thap, cac
nh o m k£t qua rat bat binh d in g ve do ldn cua chung, trong khi FIMIX-
PLS va PLS-POS cung cap cac nhom co do ldn gan n h d bang nhau.
Ngoai ra, khi doi chieu bien servicetype - m a chung ta da sd dung trong
PLS-MGA tro n g C hdong 4 - vdi phan vung PLS-POS, ket qua tham
chi con toi te ho n vdi sd chong cheo chi la 52,7%. Vi khong cd bien giai
thich them nao cd san trong tip dd lieu, chung toi tiep tuc m inh hoa, sd
dung serviceprovider lam bien giai thich. Ldu y rang m o ta ro rang ve cac
phan doan da ddpc xac dinh nhin chung la kha kho khan.

Bddc 5: I ldc lddng cac mo hinh phan doan cy the

Bddc tiep theo trong phan tich hau nghiem la tinh toan ket qua phan
doan cu the. Vdi muc dich nay, chung ta can din h nghia serviceprovider
nh d m ot bien nhom . Tuy nhien, serviceprovider cd bon gia tri duy nhat,
m a chung ta can phai tu lai thanh hai nhom . Trong trd b n g hpp cua
chung ta, nha cung cap dich vu cd gia tri 1 va 2 tdPng dng vdi phan
doan PLS-POS dau tien, trong khi cac gia tri 3 va 4 tdPng dng vdi phan
doan PLS-POS thd hai. Trong che do Data View SmartPLS, tuy chpn
Add Data Groups trong th an h trin h don cho phep chung ta xac dinh
hai nhom nay. N hap vao n u t Add D ata Groups se m d h o p thoai trong
dd chung ta co the xac dinh cac bien nhom m di. Trong Group Name,
chi dinh ten bien nhom m di (vi du: serviceprovider 1 + 2) va xac dinh
cac gia tri theo Group Terms, n h d dope hien thi trong phan tren cua
To lieu 5.18.

Sau khi nhap vao OK, SmartPLS se tao m ot nhom bao gom cac
quan sat do, trong dd nha cung cap dich vu nhd hpn 3 (tdc la, 1 va 2).
Ih e o each toong to, chung ta tao bien nhom m di (vi du: serviceprovider
3 + 4) vdi tat ca cac quan sat trong do nha cung cap dich vu ldn hon 2
(tdc la 3 va 4), n h o dope trinh bay trong phan dodi cua T o lieu 5.18.

Sau khi da xac dinh cac nhom , chung ta cd the Ode lOPng rieng m o
hinh d o d n g dan PLS cho td n g nhom dd lieu. De thpc hien dieu nay, hay
chay th u at toan PLS-SEM bang each vao Calculate -» PLS Algorithm
trong thanh trin h ddn. Ngoai ra, ban co the vao cda sd M odeling, nhap
chuot trai vao bieu topng banh xe tren thanh cong cu va chpn tuy chpn
tUdng dng trong hop ket hpp m d ra. Hop thoai m d ra cung cap cac
212 Nhung v in de nang cao trong PLS-SEM

tuy chon chuan de chay thuat toan PLS-SEM, nhflng co m ot tab m di

co n h a n -Data Gr-uops.--Gifl--lai-cac thiet-lap-m ac dinh cho thuat toan
PLS-SEM, cac gia tri khuyet, trong so va bam vao tab Data Gruops.
Nhfl dflpc the hien trong Tfl lieu 5.19, m ot m enu se m d ra, tfl do ban co
th e chon m ot hoac n h ieu n h d m de diipc p h an tic h rieng. Kiem tra cac
h o p ben canh serviceprovider 1 + 2 v a serviceprovider 3 + 4 de chon hai
nhdm nay cho cac fldc liipng m o hinh diidng d an PLS theo nhdm cu
th e, sau do la Start Calculation.

Configure Data Group

Group Name i serviceprovider 1+2

— ' Group Terms -----------------

Link multiple terms: © AND © OR

serviceprovider

Add Term

Configure Data Group

Group Name servicepi

— Group Terms
Link multiple terms: © A M

is higher than

Setup f|1 Weighting Data Groups

Run algorithm for the following selected data groups:

E Select All
m serviceprovider 1+2 251 cases
13 serviceprovider 3+4 85 cases
 Chirang 5: Mo hinh hoa khong ddng nhlt khong quan s£t dupe 213

Sau khi hoan thanh cac phan tich cua tap hpp dd lieu tong hpp
va td n g nhom , SmartPLS se m b bao cao ket qua. Ban dau, SmartPLS
hien thi ket q ua phan tich dd lieu tong hpp, nhP ng ban co the de dang
ldd t qua cac nhom khac bang each nhap vao trin h don keo xuong ben
canh D ata Groups hoac vao nut Next b en dpdi bang ket qua. C hung
ta cung can chay rieng bootstrapping cho moi nhom bang each vao
Calculate -> B ootstrapping hoac sd dung bieu tppng banh xe trong cda
so M odeling. Sd dung cai dat mac dinh, nhP ng chac chan rang ban lai
chpn tat ca cac nhom trong tab Data Groups.

TP lieu 5.20 cung cap tong quan ve d d lieu tong h p p va ket qua
n h o m cu the, bao gom tat ca cac tieu chi dan h gia m o h in h do lpdng
nguyen nhan va ket qua nhP dppc ghi trong vi du cua Hair va c6ng sd
(2017). Ket qua danh gia m o hinh do lpdng ho trp dp tin cay va tinh gia
tri cua cac bien phap.

So sanh cac Pdc lPpng he so dP dng d in cua hai nhom n h a cung cap
dich vu cho th ay sP khac biet ro rang trong cac tac dong m o h in h cau
tru e. Vi du, tro n g khi tac dong cua CO M P len CUSL k hong co y nghia
tro n g nhom nha cung cap dich vu dau tien (tdc la, cac nha cung cap
v iln th o n g ldn), nd m ang y nghia am tro n g nhom th d hai (td c la cac nha
cung cap v iln thong nhd). T dong tp, tac dong cua cac so hang doi vdi
COMP va LIKE thay doi dang ke gida cac nhom . Vi du, tro n g nhom dau
tien, CSOR cd anh hddng dang ke den LIKE, nhP ng khong tac dong len
COMP, dieu ngppc lai xay ra doi vdi nhom thd hai. Cac p h a n tich sau
hon cd the lien quan den viec kiem dinh xem lieu n h d n g khac biet nay
trong he so dPdng d in cd dang ke hay khong bang each sd dung phan
tich da nhom (C hpong 4).

E E U E ! Ket q u a cu oi cu n g 1‘hSrr doan Nh& <

Mau Nha cung cap Nha cung cap
ban dau dich vu 1 va 2 dich vu 3 va 4
N 336 251 85
Kich cd phan doan 100,0% 74,7% 25,3%
tPPng doi
Path Path Path
A T T R -> COM P 0,085 0,047 0,275'"
A T T R -> LIKE 0,159" 0,176" 0,001
COM P -» CUSA 0,135" 0,143' 0,191
214 NhCmg v in <J§ ning cao trong PLS-SEM

otis: i r u n g ^ d i c h vu
CO M P * CUSL 0,011 0,097 -0,128'
CSOR * C O M P 0,057 0,011 0,215"
CSOR * LIKE 0,190"' 0,192'" 0,105
CUSA * CUSL 0,510"' 0,461'*' 0,611"'
L IK E * CUSA 0,445"' 0,501'" 0,235"
LIKE * CUSL 0,334"* 0,300"' 0,414"'
PERF * C O M P 0,295"' 0,357'" 0,162
PERF * LIK E 0,116' 0,129 -0,017
QUAL * C O M P 0,431'” 0,454'" 0,281"
QUAL * LIK E 0,378"' 0,360"’ 0,664'"
Cac do ltfcrng ket qua (Reflective measures)
Gia tri h oi tu + +
(C onvergent validity)
(AVE)
Do tin cay (Reliability) +
(do tin cay tong hop C *
-com posite reliability,
he so Cronbach’s a)
Gia tri phan biet + +
D iscrim inant validity
(H T M T ^ J J (
Cac do ltfdng nguyen nhan (Form ative measures)
Gia tri h oi tu + + +
(Convergent validity)
Cong tuyen + + +

t nghia va sii lien + + +

qu an cua cac bien
quan sat (Significance
and relevance o f the
indicators)
Gia tri R2 (R2value)
CO M P 0,629 0,646 0,654
CUSA 0,293 0,365 0,141
CUSL 0,562 0,557 0,625
 Chirong 5: Mo hinh hoa khong d<5ng nhat khong quan sat du’P'c 215

;r ? > ~ - a- .•,-r - ;<y

Tu" lieu 5 .2 0 jo t cung I'han do^n N ha cung cap djch vy

LIKE 0,557 0,570 0,525
| Gia trj jR2 co trong sd (W eighted R1 values)
COMP 0,629 0,648
CUSA 0,293 0,308
CUSL 0,562 0,574
LIKE 0,557 0,559
***p < 0,01; **p < 0,05; *p <0,10.
tieu chi danh gia md hinh do ludng thda man / khong thda man theo Hair va
+ /- =
congstf (2017).

Tuy nhien, so sanh cac he so dddng dan tii hai ni a cung cap
dich vu vdi cac nhom tii phan tich PLS-POS cho ket qua khong
ngang hang vdi nhau. Vi du, mdi q u an he qu n hdng tdong ddi
yeu 0,43 gida COMP va CUSA trong p ha cung cap dich vu
1 th i cao ho n dang ke trong giai phap ,673). Dieu tdong td
cung xay ra ddi vdi hau het cac m di quan he m d hin h cau true khac, cho
thay ket qua PLS-POS khong the diiqc tao ra day dii bang each sd d ung
bien n h a cung cap dich vu. Ket qua nay khong gay ngac nhien khi cd
sd chdng cheo gidi han gifla cac phan vung PLS-POS va nha cung cap
dich vu.
De danh gia them giai phap phan doan, chung ta tinh toan gia tri R2
co trong sd la tdng gia tri R2 cua phan doan cu the, ddqc tinh theo kich
thddc phan doan tdOng ddi. Vi du, ddi vdi COMP trong phan nhdm n h a
cung cap dich vy, R2 trong sd bang 0,747 • 0,646 + 0,253 • 0,654 = 0,648.
Tiep theo, chung ta co the so sanh ket qua cac gia tri R2 cd trong sd
(ddqc tao ra bdi gop nhdm dd lieu) vdi cac ket qua gia tri R2 tdng the
td phan tich cap dd lieu tdng hop. Phan tich nay cho thay rang viec gop
nhdm bang each sd dung cac bien nha cung cap djch vu lam tang kha
nang dd bao trong m au cua md hinh so vdi phan tich cap do tdng hop.

TOM TAT
Hieu cac phddng phap tiep can FIMIX-PLS va PLS-POS de
xac dinh v& xd ly tinh khong ddng nhat khong quan sat dddc.
F1MIX- PLS la each tiep can ldp tiem an dau tien va noi bat
nhat trong PLS-SEM. D da tren khai niem hoi quy hdn hop,
FIMIX-PLS ddc ldqng ddng thdi he sd dddng dan cua tdng
 216 Nhflng v ln de nang cao trong PLS-SEM

th an h vien nhom quan sat doi vdi m ot so nhom dddc xac dinh
trddc-. FIM IX-PLS-diflng to dae-biet hflu-ich cho viec xac d jn h
so lddng phan doan de trich xuat til dut lieu nhilng khong xac
din h dd dc cau true phan doan cd ban va khong quan tam den
tin h khong dong nhat trong cac m o h in h do liidng. PLS-POS,
p h d d n g phap tiep can leo doi dan dan p h an bo cac quan sat
gifla cac phan doan vdi m uc dich toi da h o a tong the phiidng sai
diidc giai thich tren tat ca cac phan doan, vildt qua nhflng han
che nay. Cach tiep can nay da dddc diia ra de xac dinh m ot each
dang tin cay cac h e so dfldng dan cu the theo tdng doan nhflng
khong dfla ra m ot chi dan ro rang ve so lfldng phan doan can
xem xet trong phan tich.

H ieu each sfl dung FIMIX-PLS va PLS-POS cung nhau trong

phan tich ldp tiem an. Theo cac diem m a n h va diem yeu cua
td n g p h dd n g phap, phan tich ldp tiem an trong PLS-SEM dau
tien dfla tren FIMIX-PLS de xac dinh so lfldng phan doan thich
hd p de gifl lai td dfl lieu. Giai phap FIM IX-PLS ddng vai tro nhfl
m ot giai phap khdi dong de chay PLS-POS sau dd. PLS- POS
cai thien giai phap FIMIX-PLS va cho ph ep xem xet tin h khong
dong n h at trong cac m d h in h do lfldng cu a cac bien tiem an.
M ot ph an tich hau nghiem nham m uc d ich xac din h cac bien
giai thich cd the dddc sfl dung de xac d in h phan vung dfl lieu
th an h cac nhom (quan sat dddc), phan ld n tddng flng vdi cau
true ph an doan tiem an do PLS-POS tao ra. Bfldc cuoi cung lien
quan den viec kiem dinh cac m o hinh dfldng d an phan d o m cu
th e dfla tren cac phan doan dddc xac din h b d i cac bien giai thich
da xac d inh triidc dd.

H ieu each thflc hien phan tich FIMIX-PLS bang SmartPLS.

M ot can nhac quan trong triidc khi chay FIM IX-PLS lien quan

<b den viec xfl ly cac gia tri khuyet. Trong n h fln g sai lech dang ke
se xay ra, tran h sfl dung thay the gia tri tru n g binh. lh a y vao dd,
hay xoa tat ca cac quan sat cd gia tri khuyet bang each sfl dung
xoa casewise. Viec chay quy trin h FIM IX-PLS chinh nd doi hdi
phai thflc hien m ot sd lfla chpn lien quan d e n cac thiet lap thuat
toan. Khi chay FIMIX-PLS, hay sfl dung tie u chi ddng la 10'5,
so lfldng tdi da 5000 vong lai va 10 lan lap lai de tran h hoi tu d
mflc tdi flu cue bo. De xac dinh pham vi sd phan doan hdp ly,
hay sfl dung mot phan doan lam gidi han dfldi va sd nguyen ldn
 Chu'o'ng 5: Mo hinh hoa khong ddng nhat khong quan sat dugc 217

nhat khi chia kich thiidc m au cho kich thiidc m au phan doan
toi thieu lam gidi han tren. Gidi han tren nay cd the diidc dieu
chinh, tuy thuoc vao do phde tap cua m o hinh va kich thddc
mau. Sau khi chay FIMIX-PLS, ket qua bao gom, vi du, phan
vung cuoi cung, kich thddc phan doan tddng doi, ket qua tieu
chi ldu gid phan doan va ket qua PLS-SEM phan doan cu the.

• Biet each sd d u n g Sm artPLS de chay ph an tic h PLS-POS.

Chay PLS-POS trong SmartPLS yeu cau dda ra m ot so quyet
dinh. Cac nha nghien cdu can phai xac dinh so lddng phan
doan de gid lai td dd lieu. Q uyet dinh nay nen dda tr£n ket qua
FIMIX-PLS, hoac bang each xac dinh so phan doan phu hdp vdi
tieu chi thong tin va phan loai hoac bang each sd dung giai phap
FIMIX-PLS lam phan vung khdi dong. Khi sd d u n g khdi tao
ngau nhien, cac nha nghien cdu nen ch$y thuat toan PLS-POS
10 lan de tranh hoi tu tai toi d u cue bo. Trong qua trinh khdi
tao ngau nhien, cac nha nghien cdu cd tuy chon chay th u tuc
tien phan doan de cai thien giai phap PLS-POS. PLS-POS cung
yeu cau cac nha nghien cdu xac dinh chieu sau tim kiem , td d n g
d ddng vdi so ldong quan sat. C uoi cung, cac nha nghien cdu cd
the chon trong so cac tieu chi toi du hoa khac nhau. Sd dung
tieu chuan toi du hoa cua R2 cd trong so cua tat ca cac bien noi
sinh trong mo hinh dddng dan PLS thddng la m ot each tiep can
tot. Sau khi ch^y thuat toan, PLS-POS tra ve, trong so n h d n g thd
khac, phep gan sau cung cac quan sat cho cac phan doan, kich
thddc phan doan, ket qua tieu chi toi du hoa va ket qua PLS-
SEM phan doan cu the.

ia phan tich ldp tiem an la gi?

2. FIM IX-PLS la gi? D iem m a n h va diem yeu cua p h d d n g phap

la gi?

3. PLS-POS la gi, va phddng thde nay hoat dong n h d th e nao?

4. Tai sao ban nen sd dung FIMIX-PLS va PLS-POS cung nhau

trong phan tich ldp tiem an?

5. Muc dich cua phan tich hau nghiem la gi va tai sao lai quan trong?
 218 Nhung wan de nang cao trong PLS-SEM

CAU HOI TU DUY PHAN BIEN

1. M ac dich cua tieu c h fh o n d g n (e n tro p y )1 a gi, va tai sao sii can
nhac cua no lai quan trong khi quyet dinh so liicrng phan doan?
2. T rong trP dng hop nao chung ta se phu nhan sii hien dien cua
tin h k h ong dong n h at dang ke va tiep tuc vdi p h an tich cap dp
tong the?

3. Tai sao viec thu thap du cac bien giai thich tiem nang trong giai
do an thiet ke nghien cdu lai quan trong doi vdi viec chay th an h
cong phan tich hau nghiem ?
4. Phan tich mot each nghiem tuc tuyen bo sau: “Bdi vi ket qua tii
ph an tich hau nghiem se khong bao gid d ap dng day du cau true
phan doan tiem an diipc tao ra bdi cac phiicing phap ldp tiem an,
nd n h at thiet se cd hau qua bait lpi cho tinh gia tri cua cac ket qua”.

THUAT NGU

AIC: A kaikes Inform ation i th o n g tin Akaike

Criterion
AIC}: M odified Akaikes Tieu chi th o n g tin Akaike diipc sda
Inform ation C riterion W ith doi vdi h e so 3
Factor 3
Akaike’s Inform ation Criterion xem AIC
(AIC)
sian Inform:ation Criterion
Bayesian Tieu chi th o n g tin Bayesian
5
B!C xem Bayesian Inform ation
C riterion (BIC)
e f t : C onsistent AIC Tieu chi th o n g tin Akaike nhat
quan
Cluster analysis Phan tich cum
Clustering Phan cum
Consistent AIC (CAIC) xem CAIC
EN: N orm ed entropy statistic Thong ke entropy chuan
Ex post analysis P han tich hau nghiem
 Chu’o’ng 5: Mo hinh hoa khong d6ng nhit khong quan s^t dime 219

Expectation m axim ization (EM) Thuat toan toi da hoa ky vong

algorithm
FIMIX-PLS: Finite m ixture Binh p h dong toi thieu rieng ph an
partial least squares hon hop hiiu han
Finite mixture partial least xem FIMIX-PLS
squares (FIMIX-PLS)
Inform ation criteria Tieu chi th ong tin
k-m eans clustering Phan cum k-m eans
Label switching
Latent class techniques
Chuyen doi nh an
Ky thuat ldp tiem an o°
M odel selection criteria Tieu chi Ida chon m o hinh
Modified AIC with factor 3 (AIC3) xem AIC3
Norm ed entropy statistic (EN) xem EN
Observed heterogeneity Tinh kh ig dong nhat quan sat

O ptim ization criterion li toi da hoa

Prediction errors Sai so dii bao
Prediction-oriented segmentation Phan doan dinh hiiofng dii bao
in PLS-SEM (PLS-POS) trong PLS-SEM
Presegmentation Tien phan doan
Response-based segm entation Ky thuat phan doan diia theo ph an
techniques hoi
Search depth Chieu sau tim kiem
Unobserved heterogeneity Tinh khong dong nhat khong quan
sat diioc

<3 I LIEU DOC THEM

Becker, J.-M., Rai, A., Ringle, C. M., & Volckner, F. (2013). Discovering
unobserved heterogeneity in structural equation m odels to avert
validity threats. MIS Quarterly, 37,665-694.

H ahn, C., Johnson, M. D., H errm ann, A.. & Huber, F. (2002). C apturing
custom er heterogeneity using a finite m ixture PLS approach.
Schmalenbach Business Review, 54, 243-269.
220 NhOng v3n de nSng cao trong PLS-SEM

Hair, f. F., Sarstedt, M., M atthews, L., & Ringle, C. M. (2016). Identifying
an d treating unobserved heterogeneity w ith FIMIX-PLS: Part I -
M ethod. European Business Review, 2 8 ,63-76.

M atthews, L., Sarstedt, M., Hair, J. F., & Ringle, C. M . (2016). Identifying
and treating unobserved heterogeneity with FIMIX-PLS: Part II-A
case study. European Business Review, 28, 208-224.

Sarstedt, M., Becker, J.-M., Ringle, C. M., & Schwaiger, M. (2011).

Uncovering and treating unobserved heterogeneity with FIMIX-
PLS: W h ich m odel selection criterion provides an appropriate
nu m b er o f segm ents? Schmalenbach Business Review, 63, 34-62.

Wedel, M., & Kam akura, W. (2000). M arket segmentation: Conceptual

and methodological foundations (2nd ed.). Norwell, MA: Kluwer
Academic.

,c f

/V
rC " '
Giai thich thuat nguf

AIC: xem Akaikes Information Criterion.

AIC3: xem Modified A IC with factor 3.
Akaike’s Inform ation Criterion (AIC) (Tieu chi thong tin
xem Information criteria.
Alpha inflation (Phong dai Alpha): de cap den thpc te la ban cang tien
h an h nhieu kiem d inh d m dc y nghla nhat dinh, ban cang co n hieu kha
nang nhan diipc ket q ua co y nghia khi dieu nay khong xay ra (tdc la sai
lam Loai I).
Attribute (Thuoc tinh): m ot phan td cua dinh nghia bien nghien cdu.
N o dinh nghia kieu chung cua thuoc tin h m a bien nghien cdu trong tam
de cap, chang han nhii thai do/quan diem.
Average variance extracted (Phddng sai trich): la m dc d o m a bien
nghien cdu giai thich sp bien thien cua cac bien quan sat no.
Bandwidth-fidelity tradeoff (Trao doi chinh xac thong tin): la van de
nan giai th p c te do td sp can bang gida viec sd dung cac bien phap do
lddng se bao phu phan ldn cac bien the trong m ot dac diem (do lddng
cap pham vi) hoac cac bien phap do lddng se danh gia m ot vai dac diem
cu the (do lddng cap khia canh) chinh xac hdn.
Bayesian Inform ation Criterion (BIC): xem Information criteria.
Bias-corrected and Bonferroni-adjusted confidence intervals
(Khoang tin cay Bonferroni-adjusted v i Bias-corrected): m ot dang
khoang tin cay ddpc sd dung de kiem dinh y nghia cua nh ieu bo td
(tetrad) ddpc xem xet theo m o h inh do lddng.
BIC: xem Bayesian Information Criterion.
Bonferroni correction (Dieu chinh Bonferroni): m ot p h d d n g phap
ddpc sd dung de chong lai sp gia tang sai so chung (familywise error
rate) khi thpc hien nhieu so sanh tren vai nhom dd lieu.
Bootstrap sam ples (Cac mau Bootstrap): so ldpng m au d d p c ru t ra
trong thd tuc bootstrapping. Thong thddng, 5000 m au hoac nhieu hon
ddpc khuyen khich.

221
222 Nhung van de nSng cao trong PLS-SEM

Bootstrapping: la m ot ky thuat lay m au lai de ru t ra m o t so lddng ldn

c a c m l u t d cac.dd lieu_ban-dau-(vdisd thay_the).va ddc lddng cac m o
hinh cho m oi mau. No dddc sd dung de xac din h cac sai so tieu chuan
cua cac d dc liidng he so de danh gia y nghia thong ke cac he so m a
k hong can d d a tren n h d n g gia dinh ve phan phoi.
Bottom -up approach (Cach tiep can td dddi len): m ot each de thiet
lap m ot H C M trong do m ot so bien tiem an (LOCs) dd d c ket hdp thanh
m ot bien nghien cdu ddn le, trd u td d n g hdn (H O C ).
CAIC: xem Consistent AIC.
Causal indicators (Bien quan sat nguyen nhan): la m ot dang bien
quan sat dung trong m o hin h do lddng nguyen nhan. Cac bien nghien
cdu dddc do bang cac bien quan sat nguyen nhan co mpt khoan sai so,
dieu nay ngu y rang cac bien quan sat khong hoan toan h in h th an h bien
nghien cdu.
CB-SEM (M o hinh CB-SEM): xem Covariance-based structural
equation modeling.
Cluster analysis (Phan tich cum): la m ot ldp cac p h ddng phap phan
vung m ot tap cac doi td d n g vdi m uc dich la th u d d d c sd td d n g dong cao
ben trong cac nhom dddc hin h th an h va khac biet cao g id a cac nhom .
Clustering (Phan cum): xem Cluster analysis.
Collect m odel (Mo hinh tap hqp): m ot dang H C M trong dd H O C la sd
ket hop cua m ot so LOC cu the dai dien cho cac th a n h phan cu the hdn
tao th an h khai niem chung. Moi quan he gida H O C va LOC la phan anh
nguyen nh an .
Com m on factor m odel (M o hinh nhan to chung): gia d in h rang m oi
bien qu an sat tro n g m o t tap qu an sat do lddn g la m ot ham tuyen tinh
cua m ot hoac nhieu nhan to chung. Phan tich n h a n to kham pha (EFA),
phan tich nh an to khang dinh (CFA), va CB-SEM la 3 dang chinh cua
cac phan tich d d a tren m o hinh nh an to chung.
Com m on variance (Phddng sai chung): phdd n g sai m a m ot bien quan
sat chia se vdi cac bien quan sat khac trong m o h in h do lddng cua m ot
bien nghien cdu.
Com posite indicators (Bien quan sat tong hdp): la dang bien quan sat
dung trong m o hinh do lddng nguyen nhan. Cac b ien nghien cdu dddc
do bang cac bien quan sat tong hdp khong cd khoan sai so, dieu nay ngu
y rang cac bien quan sat hinh thanh hoan toan bien nghien cdu.
 Giai thich thuat ngu> 223

Com posite model approach (Phddng phap tiep can m o hinh tong
hdp): m ot each tiep can de ddc lddng cac dai dien bien nghien cdu.
M uc tieu cua no la tinh tong phdong sai trong cac bien quan sat h d n la
giai thich moi td d n g quan gida cac bien quan sat.
C om positional invariance (Tinh bat bien thanh phan): ton tai khi cac
trong so bien quan sat bang nhau gida cac nhom .
Conceptual variables (Bien so khai niem): y td d n g khai quat hoac suy
n g h i ve cac khai niem trd u td d n g ma cac nha nghien cdu thiet lap va de
xuat de do lddng trong nghien cdu cua ho.
Confidence interval (Khoang tin cay): cung cap gidi han d d d i va tren
cua cac gia tri trong do tham so tong the thde sd se rdi vao x&c suat nhat
dinh (vi du: 95%). Trong PLS-SEM, viec xay ddng khoang tin cay dda
tren cac sai so chuan bootstrapping.
Configural invariance (Tinh bat bien cau hinh): xuat hien khi khai niem
nghien cdu ddoc tham so hoa va ddc lddng bang nhau gida cac nhom.
Confirmatory tetrad analysis in PLS - SEM (CTA-PLS) (Phan tich
bo td khang dinh trong PLS-SEM (CTA-PLS)): cho phep kiem dinh
thong ke xem mo h inh do lddng la nguyen nhan hay ket qua.
Consistent AIC (CAIC) (A lC n h atq u an ): xem Information criteria.
Consistent PLS (PLSc): la m ot bien the cua thuat toan PLS-SEM tieu
chuan, dda ra cac d dc ld d n g m o hinh thich hdp m a khong lam suy giam
moi tddng quan gida cac cap bien tiem an, bang each mo p h dng ket qua
CB-SEM.
Construct definition (D inh nghia khai niem nghien cdu): each cu the
trong do m ot bien so khai niem dddc do trong m ot nghien cdu cu the
va co the khac vdi m ot nghien cdu khac.
Constructs (khai niem nghien cufu): do lddng cac khai niem trdu tddng,
phde tap va khong the quan sat trdc tiep. Khai niem nghien cdu dddc the
hien trong mo hinh dddng dan bang hinh tron hoac hinh bau due.
Covariance-based structural equation m odeling CB-SEM (Mo hinh
phddng trinh cau true dda tren hiep phddng sai): dddc sd dung de
kh§ng dinh (hoac bac bo) ly thuyet. Viec nay dddc thde hien bang each
xac djnh m ot m o hinh de xuat theo ly thuyet, co the ddc lddng ma tran
hiep phddng sai cho m ot tap dd lieu mau.
CTA-PLS: xem Confirmatory tetrad analysis in PLS - SEM (CTA-PLS).
224 Nhdng van de nang cao trong PLS-SEM

Cubic effect (Tac dong bac 3): m oi quan he phi tuyen dai dien bdi m ot
da th d e bac 3; xem them .fdc ddng phi tuyen vk-da thde.
D irect effect (Anh hddng/tac d ong trdc tiep): la m oi quan he trdc tiep
gida 2 bien tiem an trong m o h in h d d d n g dan PLS.
Effect indicators (Bien quan sat anh hddng): xem Reflective indicators.
EN: xem Norm ed entropy statistic.
Equality o f com posite mean values and variances (Sd can bang cua
gia trj trung binh tong hdp va phddng sai): la cac yeu cau cuoi cdng
de thiet lap day du bat bien do lddng. , - 0 '
Equidistance scale (Thang do khoang each deu): thang do tro n g do cac
k h oang dddc ph an bo theo don vi bang nhau. Vi du, th an g do 5-diem
Likert vdi hai hang p h u din h (hoan toan khong d o n g y va khong dong
y), m pt tuy chon tru n g lap va hai hang khang d in h (dong y va hoan toan
dong y) co the d d p c coi la m ot th an g do khoang each deu nhau.
Error variance (P h d ong sai sai so): la m ot phan cua p h d d n g sai rieng
cua bien q u an sat, dd p c gia djn h la ngau n h ien va khong d an g tin cay
(td c la, sai so do ld d n g va lay m id)*
Ex post analysis (Phan tich hau nghiem ): n ham xac d in h m ot hoac
nhieu bien giai thich phu h p p vdi ket qua phan doan Idp tiem an theo
each tot nhat co the de tao thuan ldi cho phan tich da nhdm .
Expectation m axim ization (EM) algorithm (Thuat toan tdi da hoa ky
vong): thuat toan lap lai de tim ddc ldpng cd k h a nang tdi da cua cac
tham sd trong m ot m d hin h thong ke. Thuat toan luan phien gida th d e
hien bddc m ong dpi (E) va bddc tdi d a (M). Bddc E tao ra m ot ham ky
vong cua log-likelihood, ddpc danh gia bang each sd dung ddc ldpng
hien tai cua cac tham sd. Bddc M tin h toan cac th am sd bang each tdi da
h o a ky vong log-likelihood d d p c tim thay trong b d d c E. Cac b d d c E va
M ddpc ap d u n g lie n tuc cho d en khi ket q u a on dinh.
Factor (score) indeterm inacy (D o bat dinh (sd) nhan td): cd nghia la
cd m ot sd ldpng vo han cac tap h p p cac nhan td p h u h p p vdi cac yeu cau
cua m ot m d h in h nhan td chung n hat dinh.
Factor-based SEM (SEM dda tren nhan td): xem Covariance-based
structural equation modeling.
Familywise error rate (Ty le sai sd chung nhd nhat): xac suat tao ra
m ot hoac n h ieu sai lam Loai I khi th d e hien nh ieu so san h tren nhieu
nhdm dd lieu.
 Giai thich thuat ngu 225

FIMIX-PLS: xem Finite mixture partial least squares (FIMIX-PLS).

Finite mixture partial least squares (FIMIX-PLS) (Binh phddng toi
thieu tdng phan hdn hdp hflu han): la phddng phap tiep can lcfp tiem
an cho ph ep xac din h va xd ly tin h khong d d n g nhat k h ong quan sat
dddc trong m o hinh dddng dan PLS. Phddng phap ap dung hoi qui hon
hdp de ddc lddng ddng thdi tham sd nhom cu the va xac suat cua cac
quan sat cua cac nhanh phan doan.
Focal object (D oi tddng tieu diem): m ot phan td cua din h nghia bien
nghien cdu. No de cap den thde the m a m ot thuoc tinh d d d c ap dung.
Forced-choice scale (Thang do Ida chon bat buoc): thang do khong co
hang tru n g lap.
Formative measurement m odel (M d hinh do lddng nguyen nhan): la
dang thiet lap m o h in h do lddng m a tro n g d o chieu cua cac m ui ten di
td bien quan sat tdi bien nghien cdu.
Formative- formative HCM (Mo hinh thanh phan thd bac Nguyen
nhan - nguyen nhan): cd cac mo h in h do lddng nguyen nhan cua tat ca
cac LOC trong HCM va m di quan he dddng dan nguyen nhan gida cac
LOC va H O C (tdc la cac LOC h inh thanh nen HOC).
Form ative- reflective HCM (M o h in h thanh phan thd bac Nguyen
nhan - ket qua): cd cac m o hinh do lddng nguyen n h a n cu a tat ca
cac LOC tro n g HCM v i moi quan he dddng dan k£t qua td H O C tdi
cac LOC.
Full measurement invariance (D o lddng bait bien toan phan): dddc
xac nhan khi (1) bat bien cau hinh, (2) bat bien thanh phan, va (3) sd
can bang eua gia tri trung binh tdng hop va p h ddng sai dddc chdng
m inh.
HCM: xem Hierarchical component model.
Hierarchical common factor m odel (Mo hinh nhan to chung thd
bac): xem reflective-reflective HCM.
Hierarchical component model (HCM) (Md hinh thanh phan thd
b a c ): la m ot cau true bac cao (th d d n g la bac hai) trong dd chda nhieu ldp
khai niem nghien cdu va cd m dc trd u td o n g cao. 1ICM bao gom th an h
ph an bac cao trdu tddng hon (higher-order constructions - H O C), lien
quan tdi hai hoac nhieu th an h phan bac thap (lower-order constructions
- LOCs) theo h ddng ket qua hoac nguyen nhan.
226 Nhdng v in d§ nang cao trong PLS-SEM

Higher-order construct (H O C) (Khai niem nghien ciiu bac cao): la

m ot khai niem nghienxflu tdng-quat dai d ien cho tat ca LOC phu thuoc
trong m ot HCM .
Higher-order m odel (M o hinh bac cao): xem Hierarchical component
model (HCM).
HOC: xem Higher-order construct.
Impact-performance map (Bieu do tac d ong-hieu suat):
dien do hoa ve ket qua bieu do tac dong- hieu suat; xem
performance map analysis.
Importance (Tam quan trpng): la m ot thuat ngfl tro n g IPM A. No
td d n g dfldng vdi tac dong tong chfla chuan hoa cua vai bien tiem an len
bien muc tieu. r fO -
Im portance-perform ance map analysis (IPM A) (Phan tich bieu do
tam quan trong-hieu suat): m d rong bao cao ket qua PLS-SEM tieu
chuan ve fldc lfldng he so dfldng dan bang each th em m o t chieu/phfldng
dien vao cac phan tich de xem xet gia tri trung bin h cua diem so bien
tiem an. Cu the, IPM A td d n g phan vdi tac dong tong cua m o hin h cau
true len m ot khai niem nghien cflu muc tieu cu th e vdi trung binh diem
so bien tiem an cua tien khai niem nghien cflu nay.
Im portance-perform ance m atrix (Ma tran tam quan trong-hieu
suat): xem Importance-performance map analysis (IPM A).
Indicators (Cac bien quan sat): la nhflng quan sat do lfldng trdc tiep
(dfl lieu tho), thfldng dddc goi la bien do lfldng (item s) hoac bien bieu
thi (m anifest), dddc trinh bay trong m d hinh dfldng dan bang cac hinh
chfl nhat.
Indirect effect (Tac d ong gian tiep): the hien m di quan he gifla hai bien
tiem an thong qua m ot bien nghien cflu thfl ba (vi du bien trung gian)
trong md h inh dfldng dan PLS. Neu p i la m di q u an he gifla bien tiem an
va bien tru n g gian, p 2 la m di quan he gifla bien tru n g gian va bien tiem
an noi sinh, tac dong gian tiep la tich cua p, va pr
Inform ation criteria (Tieu chi thong tin): cac d o lfldng thong ke ve
chat lfldng tddng doi cua m ot giai phap phan d o a n nhat dinh tddng
phan sfl ph u hdp (tflc la, likelihood) cua m ot giai p h a p va so lfldng tham
sd dddc sfl dung de dat dddc sfl phu hdp do, ta n g theo sd lfldng phan
doan. Gia tri cua m ot tieu chi thong tin nhat dinh cang n h d thi giai phap
phan doan cang tot. Nghien cflu da de xuait m ot lo at cac tieu chi thong
 Giai thich thuat ngd 227

tin nhii Tieu chi thong tin cua Akaike (AIC), AIC diidc sda doi vdi he
so 3 (AIC3), AIC n h a t quan (CAIC) va Tieu ch uan th o n g tin Bayesian
(BIC), the hien cac xu hudfng khac nhau va danh gia qua m dc so lddng
phan doan phu hdp.
Interaction term (So hang/dai liidng tiidng tac): la m ot bien phu trd
dddc dda vao m o hinh dddng dan de tinh toan tac dong bac hai, khi
xem xet sd td tddng tac; xem them self-interaction.
IPMA: xem Importance-performance map analysis(IPM A).
<5"
Items (Cac bien quan sat/bien do liidng): xem Indicators. cr
Jangle fallacy (Sai lam ldn): m o ta suy luan rang hai do lddng (vi du:
thang do) vdi cac ten goi khac nhau do lddng cac khai niem nghien cdu
khac nhau.
k-means clustering (Phan cum k-means): m ot nhdm cac thuat toan
phan cum khong th d bac hoat dong bang each phan tach cac quan sat
vao m ot so nhdm dddc xac dinh trddc va sau dd gan lai cac quan sat
cho den khi vai m uc tieu so lien quan den phan biet cum dddc dap dng.
Label switching (Chuyen doi nhan): xay ra khi nhan cua m ot phan
doan cu the thay doi td m ot lan chay FIMIX-PLS sang bd d c tiep theo.
Laten class techniques (Cac ky thuat ldp tiem an): la m ot ldp each tiep
can nham tao thuan ldi cho viec kham pha va x d ly van de khong dong
nh at khong quan sat dddc. Cac each tiep can khac nhau da dddc de
xuat, vi du n h d hon h d p h d u han, thuat toan goc, h o ac cac p h d d n g th d e
“le o d o i” (hill-climbing) cho PLS-SEM.
Latent variable (Bien tiem an): xem Constructs.
Latent variable scores (D iem so bien tiem an): cac gia tri dddc tin h de
bieu thi cac bien tiem an.
Linear effect (Tac dong tuyen tinh): bieu thi bang dddng th^ng khi
d d d c ve d d d i dang b ieu do.
Linear relationship (M oi quan he tuyen tinh): xem Linear effect.
LOC: xem Lower-order construct.
Log transformation (Chuyen doi log): m ot dang chuyen doi dd lieu de
tin h toan cac moi quan he phi tuyen, ap dung ham logarit c d s o 10 cho
m oi quan sat.
Lower-order construct (LOC) (Thanh phan bac thap): la cap ben dd d i
cua H O C trong m ot HCM.
228 Nhirng v ln de nang cao trong PLS-SEM

Manifest variables (Cac bien hien thi): xem Indicators.

Measurement equivalence(Tuiangdu'cfngdotiiPng): xem M easurement
invariance.
M easurement invariance (Bat bien do lcfotng): de cap den sit so sanh
phan tra lcti cua tap bien do liidng gida cac nhom . Trong so n h d n g th d
khac, bat bien do lddng ngu y rang tac dong cua bien dieu tiet phan loai
bi h an che tdi he so d ddng dan va khong dinh dan g tdi sd khac biet gida
cac n h o m cd lien quan trong m d hin h do lddng.
M easurement invariance o f com posite hiodels (M ICOM ) procedure
(Quy trinh bat bien do lddng cua md hinh tdng h op MICOM): la m ot
chudi cac kiem dinh de d an h gia tinh bat bien cu a cdc phep do lddng
(khai niem nghien cdu) thong qua nhieu nhom dO lieu. Q uy trin h gom
cd 3 bddc de kiem dinh cac khia canh khac nhau cu a b it bien do lddng:
(1) bat bien cau hin h (tdc la n h d nhau ve tham so va each ddc ldpng),
(2) bat bien th an h phan (tdc la n h d nhau v i tro n g so bien quan sat), va
(3) tdong dddng ve phddng sai va gia tri tru n g b in h tong hdp.
Measurement m od el (Mo h in h do lddng): la m o t th a n h p h an cua m o
hinh d ddng dan, cd chda cac bien quan sat va m o i quan he cua chung
vdi cac khai niem nghien cdu.
M easurement model m isspecification (Sai lech m o hinh do lddng):
m o ta viec sd dung m o h in h do lddng ket qua k h i nd la nguyen nhan
hoac dung m d hin h do lddng nguyen nhan khi n d la ket qua. Sai lech
m d h inh do lddng thddng dan den ket qua vo gia tri va cac ket luan
nham .
M ediation (Trung gian): dai dien tinh hudng ma trong dd bien trung
gian, tro n g m ot chdng m dc nao dd, th u hut tac d o n g cua m ot bien tiem
an len m ot bien tiem an noi sinh tro n g md hinh d iid n g d an PLS.
Mediator m odel (Mo hinh bien trung gian): xem Mediation.
Metric scale (Thang do dang so/din h ldpng): m o t dang thang do cd
hang sd ddn vi trong dd khoang each gida cac d iem thang do la bang
nhau (thang do khoang), do dd cho phep dien dat s d khac biet tuyet doi
cua cac diem thang do. Trong trd b n g hpp thang d o bo sung cd diem 0
tuyet doi, ty le gida cac diem thang do cd the d d p c dien giai (thang do
ty le).
MICOM: xem M easurement invariance o f composite tnodels (M ICOM )
procedure.
 Giai thich thuat ngu 229

MIMIC m odel (Mo hinh MIMIC): xem Multiple indicators a n d multiple

causes model.
Model selection criteria (Tieu chi Idachon mo hinh): xem Information
criteria.
M oderation (Sd dieu tiet): xay ra khi tac dong cua bien tien an len
bien tiem an noi sinh phu thuoc vao gia tri cua bien thii ba, goi la bien
dieu tiet.
Modified AIC with factor 3 (AIC3) (AIC sda doi vdi he so
Information criteria.
Multigroup analysis (Phan tich da nhom ): kiem dm
tham so (hau het he so dddng dan) co khac nhau m ot each co y nghia
gifla 2 nhom hay khong. Nghien ciiu de xuat m ot chudi each tiep can
phan tich da bien, dda theo thii tuc bootstrapping hoac h o an vi.
Multiple battery model (Mo hinh da chudi): m ot dang H C M trong
do cac LOC do lddng cung m ot bien nghien ciiu tai cac th d i diem khac
nhau bang each sit dung dd lieu tii cung m ot nhom ngiidi tra ldi.
Multiple indicators and multiple causes (MIMIC) m od el (M o hinh
da bien quan sat va da nguyen nhan: m ot dang m o hin h phiiOng trinh
cau true ket hop ca bien quan sat nguyen nhan va bien quan sat ket qua
de do liidng cac bien tiem an.
Multiple testing problem (Van de da kiem djnh): xay ra khi sai lam
Loai I cua m ot loat cac kiem dinh tang theo cap so nhan.
Normal scale (Thang do danh nghia): la m ot thang do liidng trong do
cac con s6 diiqc gan dung de xac din h va phan loai cac doi tiipng (vi du
con ngiidi,**cdng ty, san pham ).
Nonlinear effect (Tac dong phi tuyen): khong diiqc bieu dien b&ng m ot
diidng thang khi ve tren bieu do nhiing b&ng m ot diidng cong.
Nonlinear relationship (Moi quan he phi tuyen): xem N onlinear effect.
Nonredundant tetrads (Cac bo tii khong dii thda): cac bo td diiqc xem
xet de kiem d inh y nghia trong CTA-PLS.
Normed entropy statistic (EN) (Thong ke entropy chuan): m ot do
liidng thong ke sd dung xac suat cua cac quan sat cua th an h vien phan
doan nhd la m ot dau hieu cho thay each phan doan tot n h d the nao. Gia
tri tren 0,50 cho phep phan loai dd lieu ro rang thanh cac phan doan.
230 Nhdng van de nang cao trong PLS-SEM

Observed heterogenity (K hong dong nhat quan sat ddoc): xay ra khi
nguyen n h an e u a s d khong-dong-nhat-ddpcbiet-den va co the d d d c truy
lai cac dac diem co the quan sat n h d nhan khau hoc (vi du gidi tinh,
tuoi, th u nhap).
O m nibus test o f group differences (OTG) (Kiem dinh O m nibus ve sd
khac biet nhdm): la p hdo n g phap phan tich da n h d m d d a vao PLS-SEM
de so sanh ket qua tham so cua 2 hoac nhieu nhdm cung m ot luc. Nd
td o n g d d o n g vdi kiem dinh F trong hoi quy hoac p h an tich ANOVA.
O ptim ization criterion (Tieu ch i toi du hoa): m ot do lddng nhat
d inh cd gia tri xac din h chat ld o n g cua m ot tap h o p cac th d n g so dddc
kiem djnh.
Ordinal scale (Thang do thd bac): la m ot thang do lddng trong dd cac
con so dd q c gan chi ra vi tri td d n g doi cua doi tdpng trong m ot day trat
td sap xep.
O rthogonalizing approach (Cach tiep can trdc giao): la m ot each tiep
can de md h in h hoa so hang phi tuyen (vi du: bac hai) vdi cac bien quan
sat trdc giao.
OTG: xem Om nibus test o f group differences (OTG).
Parametric test (Kiem dinh tham so): m ot bien th e da nhdm , dai dien
cho m ot phien ban sda doi cua kiem djnh t hai m au doc lap khi phdong
sai tong th e cua cac nhdm ddqc coi la bang nhau (hom oscedastic). Neu
khong (tdc la, cac p h dong sai la khong bang n h au hoac khong dong
nhat), phep kiem dinh tham so sd dung mot phien ban sda doi cua phep
kiem dinh Welch-Satterthwaite.
Partial least squares algorithm (Thuat toan binh phdong toi thieu
rieng phan): cho phep ddc ld p n g cac m o hinh d d d n g dan vdi cac bien
tiem 3n.
Partial least squares path m odeling (Mo hinh dddng d in binh
phdong toi thieu rieng phan): xem Partial least squares structural
equation modeling.
Partial least squares regression (PLS-R) (Hoi quy binh phdong toi
thieu rieng phan): m ot p h dong phap ddpc thiet ke de giam thieu van de
da cong tuyen tro n g cac m o hin h hoi quy. Nd sd d u n g p h an tich thanh
phan chinh trich xuat cac tong hpp tuyen tin h cu a cac bien doc lap va
diem so td d n g dng cua chung, xem xet moi quan h e gida cac bien doc
lap va phu thuoc, va toi da hoa viec giai thich bien phu thuoc.
 Giai thich thuat ngu 231

Partial least squares structural equation m odeling (PLS-SEM) (M o

hinh phddng trinh cau true binh phddng toi thieu rieng phan): la
m ot phd o n g phap dda tren hpp td de doc ldpng m o hinh phdong trin h
cau true. M uc dich la de to i da h o a phdong sai d d p c giai thich cua cac
bien tiem an noi sinh.
Partial measurement invariance (Bat bien do lddng rieng phan):
d d p c xac nh an chi khi (1) bat bien cau h in h va (2) bat bien thanh ph an
d d p c m in h chdng.
Permormance (Hieu suat): la m ot thuat ngb trong IPM A m ieu ta gia tri
trung b in h cua diem so (va thay doi ty le) chda chuan hoa cua m ot bien
tiem an hoac m ot bien quan sat.
Permutation (Hoan vi): tao ra sd phan bo tham Ikhao cu a vai tap th am
so td db lieu thbc te bang each trao doi ngau nh ien cac quan sat giba cac
nhom nhieu lan.
Permutation test (Kiem dinh hoan vi): la m ot bien the cua phan tich
da nhom trong PLS-SEM. No hoan doi ngau nhien cac quan sat giba cac
nhom va bdc ldpng lai m o hinh de lay ddpc thong ke kiem dinh ve cac
sd khac biet nhom .
PLS algorithm (Thuat toan PLS): xem Partial least squares algorithm.
PLSc: xem Consistent PLS.
PLS-MGA: la m ot ky thuat phan tich da nhom d d a tren bootstrap.
PLS-POS: xem Prediction-oriented segmentation approach in PLS-SEM
(PLS-POS).
PLS-R: xem Partial least squares regression.
PLS-SEM: xem Partial least squares structural equation modeling.
Polynom ial (Da thbc): m ot bieu thde toan hoc gom cd m ot tong cac
so hang, theo dd moi so hang bao gom mot bien ddpc nang len m ot luy
thda va nhan vdi m ot he so.
Polynomial degree (Bac da thbc): xac dinh so ldpng cac so hang ddpc
tong hpp trong mot da thbc; xem them Polynomial.
Prediction error (Sai so db bao): dbpc sd dung trong PLS-POS de gan
lai cac quan sat tb phan doan den phan doan kia. C hung tdong dng vdi
bin h phd o n g sd khac biet cua quan sat giba diem so dd doan va thde te
cua (cac) bien tiem an noi sinh.
232 Nhung van <Je nang cao trong PLS-SEM

Prediction-oriented segm entation in PLS-SEM (PLS-POS) (Phan

doand inh h iiofng du bao trong RLS-5EM ):-la-phddng phap p h a n d o a n
dda tren khoang each doi vdi PLS-SEM.
Premature convergence (H oi tu sdm ): xay ra khi th u at toan PLS-SEM
hoi tu n h d n g ket qua p h u thuoc vao gia tri kh d i dau va do do khong
toi du.
Presegem entation (Tien phan doan): m ot tuy chon cho PLS-POS,
trong vong dau tien, gan tat ca cac quan sat cung m o t luc cho phan doan
gan n h a t cua chung. Sau do, trong cac lan lap tiep th eo , PLS-POS chi gan
lai m 6 t quan sat cho m di lan lap.
Principal com ponents regression (Hdi quy thanh phan chinh): thdc
hien phan tich th an h phan chinh tren cac bien d o c lap va cac th an h
phan chinh d d q c sd dung n h d la cac b ien .d d .b ao /giai thich cho bien
phu thuoc. N o tap tru n g vao viec giam quy m o cu a cac bien doc lap m a
khong tinh den m oi quan he gifla cac bien doc lap va p h u thuoc.
Priority map analysis (Phan tich sd dd du tien): xem Importance-
performance map analysis (IPMA).
Product indicator approach (Phddng phap tich bien quan sat): la
each tiep can de m o hin h h o a s6 hang phi tuyen (vi du: bac hai). No lien
quan den viec n h a n tat ca cac bien quan sat cua b ien tiem an ngoai sinh
de thiet lap m o hinh do lddng cua so hang phi tuyen (vi du: bac hai).
Cach tiep can nay chi ap dung cho cac bien tiem a n ngoai sinh ddqc do
lddng ket qua.
Product indicators (Tich bien quan sat): cac bien quan sat cua m ot so
hang td d n g tac, ddqc tao ra bdi phep nhan cua m oi bien quan sat cua
bien nghien cdu vdi m oi bien quan sat cua bien dieu tiet; xem them
product indicator approach.
Quadratic effect (Tac dong bac hai): ddqc bieu d ie n bang dddng cong
qu an he phi tuyen ddqc dac trd n g bdi da th d c bac 2; xem them nonlinear
effect va polynomial.
Reflective indicators (Bien quan sat ket qua): cac b ien quan sat cua mo
hinh do lUdng ket qua.
Reflective m easurem ent m odel (Mo hinh do ld d n g ket qua): m ot dang
thiet lap m o hinh do lddng trong dd chieu m ui ten d i td bien nghien cdu
den cac bien quan sat, bieu thi gia din h rang b ie n nghien cdu gay ra
p h ep do (ch in h xac hdn, sd bien thien dong th d i) c u a cac bien q u an sat.
 Giai thich thuat ngu 233

Cac bien quan sat trong cac m o hin h do ldong ket qua cung ddpc goi la
cac bien quan sat tac dong.
Reflective -form ative HCM (Mo hinh thanh phan thd bac Ket qua-
nguyen nhan): co cac m o hinh do lddng ket qua cua tat ca cac LOC trong
HCM va m oi quan he dddng dan nguyen nhan td cac LOC tdi HOC.
Reflective -reflective HCM (Mo hinh thanh phan thd bac Ket qua
- ket qua): cd cac m o hinh do lddng ket qua cua tat ca cac LOC trong
HCM va m oi quan he dddng dan ket qua td HOC tdi cac LOC; cung con
dd p c goi la mo hinh nhan to chung thanh phan thvt bac.
Repeated indicators approach for HCM (Phdong phap bien quan sat
tai lap cho HCM): la m ot dang thiet lap mo hinh do lddng trong HCM
ma sd dung cac bien quan sat cda cac LOC n h d cac bien quan sat cua
H O C de tao ra m ot HCM trong PLS-SEM.
Rescaling (Thay doi thang do): thay doi cac gia trj cua thang do cua
bien so de phu hpp vdi pham vi ddpc xac dinh trddc (vi du: 0 den 100).
Response-based segm entation techniques (Ky thuat phan doan dda
tren phan hoi): xem Latent class techniques.
Search depth (Chieu sau tim kiem): m ot tham so trong PLS-POS xac
dinh so ldpng quan sat toi da ddpc xem xet de gan lai cho m ot phan
doan khac.
Second-order construct (Khai niem nghien ciiu bac hai): xem
Hierarchical component model.
Self-interaction (Td tdong tac): xay ra khi tac dong cua m ot bien tiem
an ngoai sinh len bic'n tiem an noi sinh phu thuoc vao gia tri cua bien
tiem an ngoai sinh.
Sidak procedure (Thu tuc Sidak): m ot phdong phap ddpc sd dung de
chong lai sd gia tang sai so chung nho nhat (family- wise e rro r rate) khi
th d e hien nhieu so sanh tren nhieu nhom dd lieu.
Specific variance (Phdong sai dac thu): la m ot phan cua phdong sai
rieng cua bien quan sat, ddpc gia dinh la co he thong va dang tin cay.
Spread m odel (Mo hinh lan truyen): m ot dang HCM , trong do H O C
ddpc bieu hien trong m ot so LOC cu the hon. Moi quan he gida H O C
va cac LOC la ket qua.
Stand-alone HCM (HCM doc lap): m ot HCM khong d d p c gan vao
trong m ot m ang ldcfi bien nghien cdu to ldn hon.
234 Nhirng van <J§ ndng cao trong PLS-SEM

Tetrad (t) (Bp tii): la sU khac nh au ve tich cua m ot cap hiep phiidng sai
va tich cua m o t cap h ie p .p h p p n g sa ik h a c . Trong m o h in h do liidng ket
qua, sU khac nh au nay diipc gia din h bang 0 hoac gan bang 0; do la bo
td dope ky vong de triet tieu. So k h o n g triet tieu bo tii tro n g m o h in h do
liidng bien tiem an d O a ra c a c n g h in g d v e d a c tin h k e t qua cua no, g d iy
m ot dac tin h nguyen nhan.
Ih eoretical m odel (Mo hinh ty thuyet): m ot tap h p p cac p h ddng trin h
vdi cac bien ch in h thiic hoa m ot ly thuyet.
Top-down approach (Cach tiep can tii tren xuong): m ot each de thiet
lap m ot H CM trong do m ot bien nghien ciiu trd u td p n g h d n (HO C)
diipc d inh nghia bad gom m ot so th a n h phan (cac LOC).
Total effect (Tac dong tong): la tong cua tac dong triic tiep va (cac) tac
dong gian tiep lien quan gida m ot bien tiem an va m o t bien tiem an noi
sinh tro n g m o h in h diidng d in PLS.
Total effects analysis o f collect-type HCMs (Phan tich tac dong tong
cua cac HCM dang tap hpp): lien ket bidn nghien ciiu tien de cua m ot
H O C theo dang HCM ket qua-nguyen nhan hoac nguyen nhan-nguyen
nhan vdi cac LOC. Tac dong cua bien nghien ciiu tien de len H O C bang
vdi tac dong trd c tiep cua no cong vdi tdng cac tac d o n g gian tiep thong
qua cac LOC.
Two-stage approach (nonlinear effects) (Cach tiep can hai giai doan
(cac tac dong phi tuyen): m ot each tiep can de m d h in h hoa so hang phi
tuyen (vi du bac hai). Cach tiep can nay cd the d p p c sd dung cho bat ky
dang bien nghien cdu ngoai sinh nao cho du nd ddcfc do ld d n g theo ket
qua, nguyen n h an , hoac dai dien m ot khai niem ng h ien cdu d o n bien.
Two-stage approach for HCMs (Cach tiep can h a i giai doan cho cac
HCM): cho phep ddc ldpng m di quan he dudng dan gida m ot bien
nghien cdu tien de cua m ot H O C trong m ot H C M k e t qua-nguyen nhan
hoac w ot H C M nguyen nhan - nguyen nhan.
Unique variance (Phiidng sai rieng): p h d d n g sai cua bien quan sat
khong chia se vdi cac bien quan sat khac trong m o hin h do lddng cua
bien nghien cdu. Nd gom cd p h d d n g sai dac thu va p h d d n g sai sai sd.
Unobserved heterogeneity (K hong ddng nhat k h o n g quan sat diidc):
xay ra khi nguon cua cau true dd lieu khong ddng n h a t khong dupe biet
den (m ot each day du).
Vanishing tetrad: xem Tetrad.
Tai lieu tham khao

Aguinis, H., Beaty, J. C., Boik, R. J., & Pierce, C- A. (2005)- Effect size
and power in assessing moderating effects of categorical variables using
multiple regression: A 30-year review. Journal o f Applied Psychology, 90,
94-107.
Akaike, H. (1973). Information theory and an extension of the maximum
likelihood principle. In B. N. Petrov & F. Csaki (Eds.), Second international
symposium on information theory (pp. 267-281). Budapest, Hungary:
Academiai Kiado.
Al-Gahtani, S. S., Hubona, G. S., 8c Wang, J. (2007). Information technology
(IT) in Saudi Arabia: Culture and the acceptance and use of IT.
Information & Management, 44, 681 -691.
Albers, S. (2010). PLS and success factor studies in marketing. In V. Esposito
Vinzi, W. W. Chin, J. Henseler, & H. Wang (Eds.), Handbook of partial
least squares: Concepts, methods and applications in marketing and related
fields (pp. 409—425). Berlin, Heidelberg: Springer.
Anderberg, M. R. (1973). Cluster analysis for applications. New York, NY:
Academic Press.
Antonakis, J., Bendahan, S., Jacquart, P., 8c Lalive, R. (2010). On making
causal claims: A review and recommendations. Leadership Quarterly, 21,
1086-1120.
Astrachan, C. B„ Patel, V. K., & Wanzenried, G. (2014). A comparative study of
CB-SEM and PLS-SEM for theory development in family firm research.
Journal o f Family Business Strategy, 5,116-128.
Atinc, G., Simmering, M. J., & Kroll, M. J. (2012). Control variable use and
reporting in macro and micro management research. Organizational
Research Methods, 15,57-74.
Babin, B. J., Hair, J. F., 8c Boles, J. S. (2008). Publishing research in marketing
journals using structural equation modelling. Journal of Marketing
Theory and Practice, 16, 279-286.
Barroso, C., 8c Picon, A. (2012). Multi-dimensional analysis of perceived
switching costs. Industrial Marketing Management, 41, 531-543.
Bass, F. M. (1969). A new product growth for model consumer durables.
Management Science, 15, 215-227.
Bearden, W. O., Netemeyer, R. G., 8c Haws, K. L. (2011). Handbook of marketing
scales: Multi-item measures of marketing and consumer behavior research.
Thousand Oaks, CA: Sage.

235
236 Nhdng van de nang cao trong PLS-SEM

Becker, J.-M., Klein, K., 8c Wetzels, M. (2012). Hierarchical latent variable

models in PLS-SEM: Guidelines for using reflective-formative type
models. Long Range Planning, 45, 359-394.
Becker, J.-M., Rai, A., & Rigdon, E. E. (2013). Predictive validity and
formative measurement in structural equation modeling: Embracing
practical relevance. In Proceedings o f the 34th International Conference
on Information Systems.
Becker, J.-M., Rai, A., Ringle, C. M„ 8c Volckner, F. (2013). Discovering
unobserved heterogeneity in structural equation models to avert validity
threats. MIS Quarterly, 37, 665-694.
Becker, J.-M., Ringle, C. M., Sarstedt, M„ 8c Volckner, F. (2015). How
collinearity affects mixture regression results. Marketing Letters, 2 6 ,643-
659.
Bentlet, P. M., 8c Huang, W. (2014). On components, latent variables, PLS and
simple methods: Reactions to Rigdon’s rethinking of PLS. Long Range
Planning, 4 7 ,136-145.
Binz Astrachan, C., Patel, V. K., 8c Wanzenried, G. (2014). A comparative
study of CB-SEM and PLS-SEM for theory development in family firm
research. Journal o f Family Business Strategy, 5 , 116-128.
Bollen, K. A. (1989). Structural equations with latent variables. New York, NY:
Wiley.
Bollen, K. A. (2002). Latent variables in psychology and the social sciences.
Annual Review o f Psychology, 5 3 ,605-634.
Bollen, K. A. (2011). Evaluating effect, composite, and causal indicators in
structural equation models. MIS Quarterly, 35, 359-372.
Bollen, K. A., & Bauldry, S. (2011). Three Cs in measurement models: Causal
indicators, composite indicators, and covariates. Psychological Methods,
16,265-284
Bollen, K. A., & Davis, W. R. (2009). Causal indicator models: Identification,
estimation, and testing. Structural Equation Modeling, 16, 498-522.
Bollen, K. A., 8c Diamantopoulos, A. (2017). In defense of causal-formative
indicators: A minority report. Psychological Methods. Advance online
publication, doi: 10.1037/met0000056
Bollen, K. A., 8c Ting, K.-F. (1993). Confirmatory tetrad analysis. In P. V.
Marsden (Ed.), Sociological methodology (pp. 147-175). Washington,
DC: American Sociological Association.
Bollen, K. A., 8c Ting, K.-F. (2000). A tetrad test for causal indicators.
Psychological Methods, 5, 3-22.
Borsboom, D., Mellenbergh, G. J., 8c van Heerden, J. (2003). The theoretical
status of latent variables. Psychological Review, 110, 203-219.
 Tai lieu tham khao 237

Bozdogan, H. (1987). Model selection and Akaike’s Information Criterion

(AIC): The general theory and its analytical extensions. Psychometrika,
52, 345-370.
Bozdogan, H. (1994). Mixture-model cluster analysis using model selection
criteria in a new information measure of complexity. In H. Bozdogan
(Ed.), Proceedings of the First US/Japan conference on frontiers o f statistical
modelling: An information approach (Vol. 2, pp. 69-113). Boston, MA:
Kluwer Academic.
Bruner, G. C., James, K. E., 8c Hensel, P. J. (2001). Marketing scales handbook:
A compilation of multi-item measures. Oxford, UK: Butterworth-
Heinemann.
Campbell, D. T., 8c Stanley, J. C. (1966). Experimental and quasi- experimental
designs for research. Boston, MA: Houghton Mifflin. ♦
Carrion, G. C., Henseler, J., Ringle, C. M„ 8c Roldan, J. L. (2016). Prediction-
oriented modeling in business research by means of PLS path modeling:
Introduction to a JBR special section. Journal of Business Research, 69,
4545-4551.
Cassel, C., Hackl, P., 8c Westiund, A. H. (1999). Robustness of partial least
squares method for estimating latent variable quality structures. Journal
of Applied Statistics, 26, 435-446.
Chin, W. W. (1994). PLS Graph [Computer software). Calgary, Canada:
University of Calgary. A ......................................................
Chin, W. W. (1998). The partial least squares approach to structural equation
modeling. In G. A. Marcoulides (Ed.), Modern methods fo r business
research (pp. 295-358). Mahwah, NJ: Lawrence Erlbaum.
Chin, W. W. (2003). A permutation procedure for multigroup comparison of
PLS models. In M. Vilares, M. Tenenhaus, P. Colho, V. Esposito Vinzi,
8c A. Morineau (Eds.), PLS and related methods: Proceedings of the
International Symposium PI.S’03 (pp. 33-43). Lisbon, Portugal: Decisia.
Chin, W. W. (2010). How to write up and report PLS analyses. In Esposito
V ife jw . W. Chin, J. Henseler, 8c H. Wang (Eds.), Handbook of partial
t squares: Concepts, methods and applications in marketing and related
ids (pp. 655-690). Berlin, Heidelberg: Springer,
in, W. W„ 8c Dibbern, J. (2010). A permutation based procedure for multi­
group PLS analysis: Results of tests of differences on simulated data and
a cross cultural analysis of the sourcing of information system services
between Germany and the USA. In V. Esposito Vinzi, W.W. Chin, J.
Henseler, 8c H. Wang (Eds.), Handbook of partial least squares: Concepts,
methods and applications in marketing and related fields (pp. 171-193).
Berlin, Germany: Springer.
Cho, H. C , 8c Abe, S. (2012). Is two-tailed testing for directional research
hypotheses tests legitimate? Journal of Business Research, 66, 1261-1266.
238 NhOng v in de nang cao trong PLS-SEM

Cliff, N. (1983). Some cautions concerning the application of causal modeling

methods. Multivariate Behavioral Research, 18, 115-126.
Cohen, J. (1988). Statistical power analysisfor the behavioral sciences. Mahwah,
NJ: Lawrence Erlbaum.
DeVellis, R. F. (2011). Scale development. Thousand Oaks, CA: Sage.
Diamantopoulos, A. (1994). Modelling with LISREL: A guide for the
uninitiated. Journal of Marketing Management, 10, 105-136.
Diamantopoulos, A. (2005). The C-OAR-SE procedure for scale development
in marketing: A comment. International Journal o f Research in Marketing,
22, 1-9.
Diamantopoulos, A. (2006). The error term in formative measurement
models: Interpretation and modeling implications. Journal of Modelling
in Management, 1,7-17.
Diamantopoulos, A., 8< Papadopoulos, N. (2010). Assessing the crossnational
invariance of formative measures: Guidelines for international business
researchers. Journal of International Business Studies, 6 1 ,1203-1218.
Diamantopoulos, A., & Riefler, P. (2011). Using formative measures in
international marketing models: (daptionary tale using consumer
animosity as an example. Advances in International Marketing, 1 0 ,11 -30.
Diamantopoulos, A., Riefler, P., & Roth, K. P. (2008). Advancing formative
measurement models. Journal of Business Research, 6 1 ,1203-1218.
Diamantopoulos, A., Sarstedt, M„ Fuchs, C„ Kaiser, S., 8< Wilczynski, P.
(2012). Guidelines for choosing between multi-item and single-item
scales for construct measurement: A predictive validity perspective.
Journal o f the Academy o f Marketing Science, 40, 434-449.
Diamantopoulos, A., & Siguaw, J. A. (2006). Formative vs. reflective indicators
in measure development: Does the choice of indicators matter? British
Journal o f Management, 13, 263-282.
Diamantopoulos, A., & Winklhofer, H. M. (2001). Index construction with
formative indicators: An alternative to scale development. Journal of
Marketing Research, 38, 269-277.
Dibbern, J., & Chin, W. W. (2005). Multi-group comparison: Testing a PLS
model on the sourcing of application software services across Germany
and the USA using a permutation based algorithm. In F. W. Bliemel,
A. Eggert, G. Fassott, & J. Henseler (Eds.), Handbuch PLS-Pfadmod-
ellierung. Methode, Anwendung, Praxisbeispiele (pp. 135-160). Stuttgart,
Germany: Schaffer-Poeschel.
Dijkstra, T. K. (2014). PLS’ janus face—Response to Professor Rigdon’s
“Rethinking partial least squares modeling: In praise of simple methods.”
Long Range Planning, 47, 146-153.
 Tai lieu tham khao 239

Dijkstra, T. K„ & Henseler, J. (2011). Linear indices in nonlinear structural

equation models: Best fitting proper indices and other composites.
Quality & Quantity, 45, 1505-1518.
Dijkstra, T. K., & Henseler, J. (2015a). Consistent and asymptotically normal
PLS estimators for linear structural equations. Computational Statistics
& Data Analysis, 81, 10-23.
Dijkstra, T. K., 8c Henseler, J. (2015b). Consistent partial least squares path
modeling. MIS Quarterly, 39,297-316.
Dijkstra, T. K., 8c Schmermelleh-Engel, K. (2014). Consistent partial least
squares for nonlinear structural equation models. Psychometrika, 79,
585-604.
Do Valle, P. O., 8c Assaker, G. (2016). Using partial least squares structural
equation modeling in tourism research: A review of past research and
recommendations for future applications. Journal of Travel Research, 55,
695-708.
Eberl, M. (2010). An application of PLS in multi-group analysis: The need for
differentiated corporate-level marketing in the mobile communications
industry. In V. Esposito Vinzi, W. W. Chin, J. Henseler, 8c H. Wang (Eds.),
Handbook o f partial least squares: Concepts, methods and applications in
marketing and related fields (pp. 487-514). Berlin, Germany: Springer.
Eberl, M„ 8c Schwaiger, M. (2005). Corporate reputation: Disentangling the
effects on financial performance. European Journal o f Marketing, 39, 838-
854.
Edwards, J. R. (2001). Multidimensional constructs in organizational behavior
research: An integrative analytical framework. Organizational Research
Methods, 4, 144-192.
Eisenbeiss,M.,Cornelifien,M.,Backhaus, K.,8cHoyer, W. D. (2014). Nonlinear
and asymmetric returns on customer satisfaction: Do they vary across
situations and consumers? Journal of the Academy o f Marketing Science,
42c 242-263.
Ernst, M. D. (2004). Permutation methods: A basis for exact inference.
Statistical Science, 19,676-685.
Esposito Vinzi, V., Trinchera, L, Squillacciotti, S., 8c Tenenhaus, M. (2008).
REBUS-PLS: A response-based procedure for detecting unit segments in
PLS path modelling. Applied Stochastic Models in Business and Industry,
24, 439-458.
Falk, R. F., 8c Miller, N. B. (1992). A primer for soft modeling. Akron, OH:
University of Akron Press.
Finn, A. (2012). Customer delight: Distinct construct or zone of nonlinear
response to customer satisfaction? Journal of Service Research, 15, 99-
110.
240 Nhung v&n de nang cao trong PLS-SEM

Fisher, R. A. (1935). Ihe design of experiments. New York, NY: Hafner.

Fornell, C. G., 6c Bookstein, F. L. (1982). Two structural equation models:
LISREL and PLS applied to consumer exit-voice theory. Journal of
Marketing Research, 19,440—452.
Fornell, C. G., Johnson, M. D., Anderson, E. W., Cha, J., 8c Bryant, B. K.
(1996). The American Customer Satisfaction Index: Nature, purpose,
and findings. Journal o f Marketing, 60, 7-18
Fu, J.-R. (2006). VisualPLS—An enhanced GUI fo r LVPLS (PLS 1.8 PC)
(version 1.04) [Computer software]. Retrieved from http://www2 .kuas.
edu.tw/prof/fred/vpls
Gilliam, D. A., 8c Voss, K. (2013). A proposed procedure for cc uct
definition in marketing. European Journal o f Marketing, 47, 5-26.
Good, P. (2000). Permutation tests: A practical guide to resampling methods for
testing hypotheses. New York, NY: Springer.
Grace, J. B., 8c Bollen, K. A. (2007). Representing general theoretical
concepts in structural equation models: The role of composite variables.
Environmental and Ecological Statistics, 1 5 ,191-213.
Gudergan, S. P, Ringle, C. M., Wende, S., & Will, A. (2008). Confirmatory
tetrad analysis in PLS path modeling. Journal o f Business Research, 61,
1238-1249.
Guide, V. D. R., 8c Ketokivi, M. (2015). Notes from the editors: Redefining
some methodological criteria for the journal. Journal of Operations
Management, 37, v-viii.
Guttman, L. (1955). The determinacy of factor score matrices with implications
for five other basic problems of common-factor theory. British Journal o f
Statistical Psychology, 8 , 65-81.
Hahn, C„ Johnson, M. D„ Herrmann, A., 8c Huber, F. (2002). Capturing
custojmgr heterogeneity using a finite m ixture PLS approach.
Schmalenbach Business Review, 54, 243-269.
Hair, J. F., Black, W. C., Babin, B. J., 8c Anderson, R. E. (2010). Multivariate
data analysis. Englewood Cliffs, N): Prentice Hall.
Hair, J. F., Hollingsworth, C. L., Randolph, A. B„ 8c Chong, A. Y. L. (2017). An
updated and expanded assessment of PLS-SEM in information systems
research. Industrial Management & Data Systems. Advance online
publication, doi: 10.1108/IMDS-04-2016-0130
Hair, J. F., Hult, G. T. M., Ringle, C. M., 8c Sarstedt, M . (2017). A primer on
partial least squares structural equation modeling (PLS-SEM) (2nd ed.).
Thousand Oaks, CA: Sage.
Hair, J. F., Ringle, C. M., 8c Sarstedt, M. (2011). PLS-SEM: Indeed a silver
bullet. Journal of Marketing Iheory and Practice, 1 9 ,139-151.
 Tai lieu tham khao 241

Hair, J. F„ Ringle, C. M., 8c Sarstedt. M. (2013). Partial least squares structural

equation modeling: Rigorous applications, better results and higher
acceptance. Long Range Planning, 46, 1-12.
Hair, J. F, Sarstedt, M., Matthews, L„ 8c Ringle, C. M. (2016). Identifying and
treating unobserved heterogeneity with FIMIX-PLS: Part I-method.
European Business Review, 28, 63-76.
Hair, J. F, Sarstedt, M., Pieper, T., 8c Ringle, C. M. (2012). The use of partial
least squares structural equation modeling in strategic management
research: A review of past practices and recommendations for future
applications. Long Range Planning, 45, 320-340.
Hair, J. F„ Sarstedt, M., Ringle, C. M., 8c Mena, J. A. (2012). An assessment of
the use of partial least squares structural equation modeling in marketing
research. Journal of the Academy of Marketing Science, 40,414-433.
Hair, J. F„ Hult, G. T. M„ Ringle, C. M„ Sarstedt, M„ & Thiele, K. O. (2017).
Mirror, mirror on the wall. A comparative evaluation of component-
based structural equation modeling methods. Journal o f the Academy
of Marketing Science. Advance online publication, doi: 10.1007/sl 1747-
017-0517-x
Hay, D. A., 8c Morris, D. J. (1991). Industrial economics and organization:
Theory and evidence. New’ York, NY: Oxford University Press.
Helm, S., Eggert, A., & Garnefeld, I. (2010). Modelling the impact of corporate
reputation on customer satisfaction and loyalty using PLS. In V. Esposito
Vinzi, W. W. Chin, J. Henseler, 8c H. Wang (Eds.), Handbook of partial
least squares: Concepts, methods and applications in marketingand related
fields (pp. 515-534). Berlin, Heidelberg: Springer.
Henseler, J., 8c Chin, W. W. (2010). A comparison of approaches for the
analysis of interaction effects between latent variables using partial least
squares path modeling. Structural Equation Modeling, 17,82-109.
Henseler, }., 8cDijkstra, T. (2016) .ADAHCO 2 [Compu ter software J. Ret rieved
from http://w'ww.compositemodeling.com
Henseler, J., Dijkstra, T. K., Sarstedt, M„ Ringle, C. M., Diamantopoulos, A.,
Straub, D. W„... Calantone, R. J. (2014). Common beliefs and reality
about partial least squares: Comments on Ronkko 8c Evermann (2013).
Organizational Research Methods, 17, 182-209.
Henseler, J., Fassott, G., Dijkstra, T. K., 8c Wilson, B. (2012). Analysing
quadratic effects of formative constructs by means of variance-based
structural equation modelling. European Journal of Information Systems,
21,99-112.
Henseler, J., Hubona, G., 8c Ray, P. A. (2016). Using PLS path modeling in
new technology research: Updated guidelines. Industrial Management &
Data Systems, 116, 2-20.
 242 NhCmg v ln <J6 n§ng cao trong PLS-SEM

Heriseler, J., Ringle, C. M., & Sarstedt, M. (2012). Using partial least squares
path-modeling in international advertising-research: Basic concepts and
recent issues. In S. Okazaki (Ed.), Handbook of research in international
advertising (pp. 252-276). Cheltenham, UK: Edward Elgar.
Henseler,)., Ringle, C. M„ 8c Sarstedt, M. (2015). A new criterion for assessing
discriminant validity in variance-based structural equation modeling.
Journal o f the Academy o f Marketing Science, 43, 115-135.
Henseler, J., Ringle, C. M„ 8c Sarstedt, M. (2016). Testing measurement
invariance of composites using partial least squares. International
Marketing Review, 33, 405-431.
Henseler, J., Ringle, C. M., 8c Sinkovics, R. R. (2009). The usetof\>artial
least squares path modeling in international marketing. Advances in
International Marketing, 20, 277-320.
Hochberg, Y., &Tamhane, A. C. (2011). Multiple comparisonprocedures.
arisoni New
York, NY: Wiley.
Hock, C., Ringle, C. M., & Sarstedt, M. (2010). Management of multipurpose
stadiums: Importance and performance measurement of service
interfaces. International Journal of Services Technology and Management,
1 4 ,188-207.
Howell, R. D., Breivik, E., 8c Wilcox, J. B. (2007). Reconsidering formative
measurement. Psychological Methods, 12,205-218.
Howell, R. D., Breivik, E., & Wilcox, f. B. (2013). Formative measurement: A
critical perspective. Data Base fo r Advances in Information Systems, 44,
44-55.
Hsu, H. M., ChangcT. C., 8c Lai, T. W. (2016). Physicians’ perspectives
of adopting computer-assisted navigation in orthopedic surgery.
International Journal of Medical Informatics, 94, 207-214.
Hubona, G. (2015). PLS-GUI [Computer software). Retrieved from http://
www.pls-gui.com
Hult, G. T. M„ Ketchen, D. J., Griffith, D. A., Finnegan, C. A., Gonzalez-
Padron, T., Harmancioglu, N Cavusgil, S. T. (2008). Data equivalence
in cross-cultural international business research: Assessment and

<9 guidelines. Journal of International Business Studies, 39, 1027-1044.

arvis, C. B MacKenzie, S. B., 8c Podsakoff, P. M. (2003). A critical review
far
of construct indicators and measurement model misspecification in
marketing and consumer research. Journal of Consumer Research, 30,
199-218.
Jedidi, K„ Jagpal, H. S., 8c DeSarbo, W. S. (1997). Finite-mixture structural
equation models for response-based segmentation and unobserved
heterogeneity. Marketing Science, 16, 39-59.
Jisha, P. R., 8c Thomas, I. (2016). Quality of life and infertility: Influence of
gender, years of marital life, resilience, and anxiety. Psychological Studies,
6 1 ,159-169.
 Tai lieu tham khao 243

Johnson, R. E., Rosen, C. C„ & Chang, C.-H. (2011). To aggregate or not

to aggregate: Steps for developing and validating higher-order multi­
dimensional constructs. Journal of Business Psychology, 26, 241-248.
Jones M. A., Mothersbaugh, D. L„ & Beatty, S. E. (2000). Switching barriers
and repurchase intentions in services. Journal of Retailing, 76, 259-274.
Joreskog K. G. (1969). A general approach to confirmatory maximum
likelihood factor analysis. Psychometrika, 34, 183-202.
Joreskog, K. G. (1973). A general method for estimating a linear structural
equation system. In A. S. Goldberger & O. D. Duncan (Eds.), Structural
equation models in the social sciences (pp. 85-112). London, UK:
Academic Press.
Joreskog, K. G. (1978). Structural analysis of covariance and correlation
matrices. Psychometrika, 43, 443-477.
Joreskog, K. G., & Wold, H. (1982). The ML and PLS techniques for modeling
with latent variables: Historical and comparative aspects. In H. Wold &
K. G. Joreskog (Eds.), Systems under indirect observation, Part I (pp. 263-
270). Amsterdam, Netherlands: North-HoilancL
Kansky R„ Kidd, M., & Knight, A. T. (2016). A wildlife tolerance model
and case study for understanding human wildlife conflicts. Biological
Conservation, 201,137-145.
Kaufmann, L., & Gaeckler, J. (2015). A structure review of partial least squares
in supply chain management research. Journal of Purchasing & Supply
Management, 21, 259-272.
Keil, M„ Saarinen, T„ Tan, B. C. Y„ Tuunainen, V., Wassenaar, A., & Wei, K.-
K. (2000). A cross-cultural study on escalation of commitment behavior
in software projects. MIS Quarterly, 24, 299-325.
Kenny, D. A. (2015). Moderation. Retrieved from htte://davidakenny.net/ cm/
moderation.htm
Kessel, E, Ringle, C. M., & Sarstedt, M. (2010). On the impact of missing values
on model selection in FIMIX-PLS. Proceedings o f the 2010 INFORMS
Marketing Science Conference, Cologne, Germany.
Kline, R. B. (2015). Principles and practice o f structural equation modeling (4th
ed.). New York, NY: Guilford Press.
Kristensen, K., Martensen, A., &Gronholdt, L. (2000). Customer satisfaction
measurement at Post Denmark: Results of application of the European
customer satisfaction index methodology. Total Quality Management,
11, 1007-1015.
Kwofie, T. E., Adinyira, E., & Fugar, F. (2015). Nature of communication
ineffectiveness inherent in the procurement systems on mass housing
projects. Journal of Construction Engineering, 914520.
Law, K. S., & Wong, C.-S. (1999). Multidimensional constructs in structural
equation analysis: An illustration using the job perception and job
satisfaction constructs. Journal of Management, 25, 143-160.
244 Nhung v in de nang cao trong PLS-SEM

Law, K. S., Wong, C.-S., & Mobley, W. H. (1998). Toward a taxonomy of

muhidimensionzhcottStrvtcterThe-Acudenty-of-Management Review, 23,
741-755.
Lee, L., PetteqS., Fayard, D., & Robinson, S. (2011). On the use of partial least
squares path modeling in accounting research. International Journal of
Accounting Information Systems, 12, 305-328.
Lee, N., & Cadogan, J. W. (2013). Problems with formative and higher-order
reflective variables. Journal of Business Research, 66, 242-247.
Liang, Z., Jaszak, R. J., & Coleman, R. E. (1992). Parameter estimation of
finite mixtures using the EM algorithm and information criteria with
applications to medical image processing. IEEE Transactions on Nuclear
Science, 3 9 ,1126-1133.
Liu, J., Zhao, X., & Yan, P. (2016). Risk paths in international construction
projects: Case study from Chinese contractors. Journal o f Construction
Engineering and Management, 142, http://dx.doi.org/10.1061/(ASCE)
C0.1943-7862.0001116
Lohmoller, J.-B. (1984). Das programmsystem LVPLS fur pfadmodelle
mit latenten variablen [The program system LVPLS for path models
with latent variables]. ZA-Information/Zentralarchiv fu r Empirische
Sozialforschung, 14, 44-51.
Lohmoller, J.-B. (1987). LVPLS 1.8. Cologne, Germany: Zentralarchiv fiir
Empirische Sozialforschung.
Lohmoller, J.-B. (1989). Latent variable path modeling with partial least
squares. Heidelberg, Germany: Physica.
MacCallum, R. C„ & Browne, M. W. (1993). The use of causal indicators
in covariance structure models: Some practical issues. Psychological
Bulletin, 114, 533-541.
MacCallum, R. C., Browne, M. W„ & Cai, L. (2007). Factor analysis models as
approximations. In R. Cudeck& R. C. MacCallum (Eds.), Factor analysis
at 100: Historical developments and future directions (pp. 153-175).
Mahwah, NJ: Lawrence Erlbaum.
MacKenzie, S. B., Podsakoff, P. M., & Podsakoff, N. P. (2011). Construct
measurement and validation procedures in MIS and behavioral research:
Integrating new and existing techniques. MIS Quarterly, 35, 293-334.
Marcoulides, G. A., & Chin, W. W. (2013). You write but others read: Common
methodological misunderstandings in PLS and related methods. In H.
Abdi, W. W. Chin, V. Esposito Vinzi, G. Russolillo, & L. Trinchera (Eds.),
New perspectives in partial least squares and related methods (pp. 31-64).
New York, NY: Springer.
Martensen, A., & Gronholdt, L. (2010), Measuring and managing brand
equity: A study with focus on product and service quality in banking.
International Journal o f Quality and Service Sciences, 2, 300-316.
 T&i lieu tham khao 245

Martilla, J. A., 8c James, J. C. (1977). Importance-performance analysis.

Journal o f Marketing, 41, 77-79.
Masyn, K. E. (2013). Latent class analysis ad finite mixture modeling. In P.
Nathan 8c T. Little (Eds.), Ihe Oxford handbook of quantitative methods
(pp. 551-611). New York, NY: Oxford University Press.
Mateos-Aparicio, G. (2011). Partial least squares (PLS) methods: Origins,
evoution, and application to social sciences. Communications in statistics
- Theory and Methods, 40, 2305-2317.
Matthews, L., Sarstedt, M., Hair, J. E, 8c Ringle, C. M. (2016). Identifying and
treating unobserved heterogeneity with FIMIX-PLS: Part II - A case
study. European Business Review, 2 8 ,208-224.
McLachlan, G. J., 8c Peel, D. (2000). Finite mixture mod k, NY:
Wiley.
Meyer, J. P., & Allen, N. J. (1991). A three-component conceptualization of
organizational commitment. Human Resource Management Review, 1,
61-89.
Michell, ). (2013). Constructs, inferences il measurement. New
Ideas in Psychology, 31, 13-21.
Minkman, E., Rutten, M. M., & van der San (2017). Acceptance
of mobile technology for citizen science in water resource management.
Journal o f Irrigation and Drainage Engineering. Advance online
publication, doi: 10.1061/(ASCE)IR. 1943-4774.0001043
Monecke, A., 8c Leisch, F. (2012). semPLS: Structural equation modeling
using partial least squares. Journal of Statistical Software, 48. Retrieved
from http://www.jstatsoft.org
Money, K. G., Hillenbrand, C., Henseler, J., 8c Da Camara, N. (2012). Exploring
unanticipated consequences of strategy amongst stakeholder segments:
The case of a European revenue service. Long Range Planning, 45, 395-
42jU
Muthen, B. O. (1989). Latent variable modeling in heterogeneous populations.
Psychometrika, 54, 557-585.
1, C. (2016). The use of partial least squares structural equation modelling
(PLS-SEM) in management accounting research: Directions for future
theory development. Journal of Accounting Literature, 37,19-35.
Nunnally, J. C., & Bernstein, I. (1994). Psychometric theory. New York, NY:
McGraw-Hill.
Pai, H. C. (2016). Development and validation of the simulation learning
effectiveness scale for nursing students. Journal of Clinical Nursing, 25,
3373-3381.
Pedrosa, R. B. D. S., Rodrigues, R. C. M., Oliveira, H. C., 8c Alexandre, N.
M. C. (2016). Construct validity of the Brazilian version of the self­
246 Nhdng van de nang cao trong PLS-SEM

efficacy for appropriate medication adherence scale. Journal of Nursing

Measurement, 24, 18E-31E.
Peng, D. X., 8c Lai, F. (2012). Using partial least squares in operations
management research: A practical guideline and summary of past
research. Journal of Operations Management, 3 0 ,467-480.
Peter, J.-P., 8c Churchill, G. A. (1986). Relationships among research design
choices and psychometric properties of rating scales: A meta analysis.
Journal o f Marketing Research, 2 3 ,1-10.
Petter, S., Straub, D„ 8c Rai, A. (2007). Specifying formative constructs in
information systems research. MIS Quarterly, 31,623-656.
Polites, G. L., Roberts, N., 8c Thatcher, J. B. (2012). Conceptualizing models
using multidimensional constructs: A review and guidelines for their
use. European Journal o f Information Systems, 21, 22-48.
Raithel, S., 8c Schwaiger, M. (2014). The effects of corporate reputation
perceptions of the general public on shareholder value. Strategic
Management Journal, 36, 945-956.
Raithel, S., Wilczynski, R, Schloderer, M. R, 8c Schwaiger, M. (2010). The
value-relevance of corporate reputation during the financial crisis.
Journal o f Product and Brand Management, 19, 389-400.
Ramaswamy, V., DeSarbo, W. S., Reibstein. D. f., 8c Robinson, W. T. (1993).
An empirical pooling approach for estimating marketing mix elasticities
with PIMS data. Marketing Science, 1 2 ,103-124.
Reinartz, W„ Haenlein, M., 8c Henseler, J. (2009). An empirical comparison of
the efficacy of covariance-based and variance-based SEM. International
Journal o f Research in Marketing, 26, 332-344.
Richter, N. F, Sinkovics, R. R., Ringle, C. M., 8c Schlagel, C. (2016). A critical
look at the use of SEM in international business research. International
Marketing Review, 33, 376-404.
Rigdon, E. E. (2005). Structural equation modeling: Nontraditional
alternatives. In B. Everitt 8c D. Howell (Eds.), Encyclopedia o f statistics in
behavioral science (pp. 1934-1941). New York, NY: Wiley.
Rigdon, E. E. (2012). Rethinking partial least squares path modeling: In praise
of simple methods. Long Range Planning, 45, 341-358.
Rigdon, E. E. (2014). Rethinking partial least squares path modeling: Breaking
chains and forging ahead. Long Range Planning, 4 7 ,161-167.
Rigdon, E. E. (2016). Choosing PLS path modeling as analytical method
in European management research: A realist perspective. European
Management Journal, 34, 598-605.
Rigdon, E. E., Becker, J.-M., Rai, A., Ringle, C. M., Diamantopoulos, A.,
Karahanna, E., . . . Dijkstra, T. (2014). Conflating antecedents and
formative indicators: A comment on Aguirre-Urreta and Marakas.
Information Systems Research, 25, 780-784.
 T&i lieu tham khao 247

Rigdon, E. E„ Preacher, K. J., Lee, N., Howell, R. D., Franke, G. R., &
Borsboom, D. (2011). Overcoming measurement dogma: A response to
Rossiter. European Journal of Marketing, 45,1589-1600.
Rigdon, E. E., Ringle, C. M., & Sarstedt, M. (2010). Structural modeling of
heterogeneous data with partial least squares. In N. K. Malhotra (Ed.),
Review of marketing research (pp. 255-296). Armonk, NY: M. E. Sharpe.
Rigdon, E. E., Ringle, C. M., Sarstedt, M., & Gudergan, S. P. (2011). Assessing
heterogeneity in customer satisfaction studies: Across industry
similarities and within industry differences. Advances in International
Marketing, 22, 169-194.
Ringle, C. M., & Sarstedt, M. (2016). Gain more insight from your PLS-
SEM results: The importance-performance map analysis. Industrial
Management & Data Systems, 116,1865-1886.
Ringle, C. M., Sarstedt, M., 8cMooi, E. A. (2010). Response-based segmentation
using finite mixture partial least squares: Theoretical foundations and
an application to American customer satisfaction index data. Annals o f
Information Systems, 8 , 19-49.
Ringle, C. M„ Sarstedt, M., 8< Schlittgen, R. (2014). Genetic algorithm
segmentation in partial least squares structural equation modeling. OR
Spectrum, 36, 251-276.
Ringle, C. M„ Sarstedt, M„ Schlittgen, R., & Taylor, C. R. (2013). PLS path
modeling and evolutionary segmentation. Journal of Business Research,
6 6 ,1318-1324.
Ringle, C. M., Sarstedt, M., 8c Straub, D. YV. (2012). A critical look at the use of
PLS-SEM in MIS Quarterly. MIS Quarterly, 36, iii-xiv.
Ringle, C. M., Wende, S., & Becker, J.-M. (2015). SmartPLS 3 (Computer
software). Retrieved from http://www.smartpls.com
Ringle, C. M., Wende, S., & Will, A. (2005). SmartPLS 2 [Computer software).
Retrieved from http://www.smartpls.com
Robins, J. (2012). Partial least squares (Editorial). Long Range Planning, 45,
309-311.
Robins, J. (2014). Partial least squares revisited. Long Range Planning, 47,131.
Roldan, J. L., & Sanchez-Franco, M. ). (2012). Variance-based structural
equation modeling: Guidelines for using partial least squares in
information systems research. In M. Mora, O. Gelman, A. L. Steenkamp,
& M. Raisinghani (Eds.), Research methodologies, innovations and
philosophies in software systems engineering and information systems (pp.
193-221). Hershey, PA: IGI Global.
Ronkko, M„ & Evermann, J. (2013). A critical examination of common beliefs
about partial least squares path modeling. Organizational Research
Methods, 16, 425-448.
248 Nhu>ng vbn d§ nang cao trong PLS-SEM

Rbnkko, M., McIntosh, C. N., Antonakis, J., & Edwards, J. R. (2016). Partial
•least-squares-path-mcdelingrTTme-forstrme-serious second thoughts.
Journal o f Operations Management, 47-48,9-27.
Rossitei, J. R. (2002). The C-OAR-SE procedure for scale development in
marketing. International Journal o f Research in Marketing, 19,305-335.
Rossiter, J. R. (2011). Measurement for the social sciences. Ihe C-OAR-SE
method and why it must replace psychometrics. Berlin, Heidelberg:
Springer.
Salzberger, T., Sarstedt, M„ & Diamantopoulos, A. (2016). Measurement in
the social sciences: Where C-OAR-SE delivers and where it does, ribt.
European Journal o f Marketing, 5 0 ,1942-1952.
Sanchez, G., Trinchera, L., 8c Russolillo, G. (2015). R Package PLSPM: Toolsfor
Partial Least Squares Path Modeling (PLS-PM) (version 0.4.7) [Computer
software]. Retrieved from http://cran.r-project.org/ web/packages/plspm
Sarstedt, M. (2008). A review of recent approaches for capturing heterogeneity
in partial least squares path modelling. Journal o f Modelling in
Management, 3 , 140-161.
Sarstedt, M., Becker, J.-M., Ringle, C. M.,&Schwaiger, M. (2011). Uncovering
and treating unobserved heterogeneity with FIMIX-PLS: Which model
selection criterion provides an appropriate num ber of segments?
Schmalenbach Business Review, 63,34-62.
Sarstedt, M„ Diamantopoulos, A., 8c Salzberger, T (2016). Should we use
single items? Better not. Journal o f Business Research, 69, 3199-3203.
Sarstedt, M., Diamantopoulos, A., Salzberger, T„ 8c Baumgartner, P. (2016).
Selecting single items to measure doubly-concrete constructs: A
cautionary tale. Journal o f Business Research, 69, 3159-3167.
Sarstedt, M., Ringle, C. M., 8c Hair, J. F. (2017). Partial least squares. In C.
Homburg, M. Klarmann, 8c A. Vomberg (Eds.), Handbook o f marketing
research. Berlin, Heidelberg: Springer.
Sarstedt, M„ Hair, J. K, Ringle, C. M„ Thiele, K. O., 8c Gudergan, S. P. (2016).
Measurement issues with PLS and CB-SEM: W here the bias lies! Journal
o f Business Research, 69, 3998-4010.
stedt, M., Henseler, J., 8c Ringle, C. M. (2011). Multi-group analysis in
partial least squares (PLS) path modeling: Alternative methods and
empirical results. Advances in International Marketing, 2 2 ,195-218.
Sarstedt, M., 8c Mooi, E. A. (2014). A concise guide to market research. The
process, data, and methods using IBM SPSS statistics (2nd ed.). Berlin,
Germany: Springer.
Sarstedt, M„ 8c Ringle, C. M. (2010). Treating unobserved heterogeneity in
PLS path modelling: A comparison of FIMIX-PLS with different data
analysis strategies. Journal o f Applied Statistics, 37, 1299-1318.
 Tai li£u tham khao 249

Sarstedt, ML, Ringle, C M., Henseler, J., & Hair, J. F. (2014). On the
emancipation of PLS-SEM. A commentary on Rigdon (2012). Long
Range Planning, 47, 154-160.
Sarstedt, M., Ringle, C. M., Smith, D., Reams, R., & Hair, J. F. (2014). Partial
least squares structural equation modeling (PLS-SEM): A useful tool for
family business researchers. Journal of Family Business Strategy, 5, 105-
115.
Sarstedt, M., & Schloderer. M. R (2010). Developing a measurement approach
for reputation of non-profit organizations. International Journal of
Nonprofit & Voluntary Sector Marketing, 15, 276-299.
Sarstedt, M., Schwaiger, M., & Ringle, C. M. (2009). Do we fully understand
the critical success factors of customer satisfaction with industrial goods?
Extending Festge and Schwaiger’s model to account for unobserved
heterogeneity. Journal of Business Market Management, 3 , 185-206.
Sarstedt, M„ Wilczynski, P., & Melewar, T. (2013). Measuring reputation in
global markets: A comparison of reputation measures’ convergent and
criterion validities. Journal of World Business, 48, 329-339.
Satterthwaite, F. E. (1946). An approximate distribution of estimates of
variance components. Biometrics Bulletin, 10-114.
Schlagel , C., & Sarstedt, M. (2016). Assessing the measurement invariance
of the four-dimensional cultural intelligence scale across countries: A
composite model approach. European Management Journal, 34,633-649.
Schlittgen, R., Ringle, C. M ., Sarstedt, M., & Becker, J.-M. (2016). Segmentation
of PLS path models by iterative reweighted regressions. Journal of
Business Research, 69,4583-4592.
Schloderer, M. P., Sarstedt, M., & Ringle, C. M. (2014). The relevance
of reputation in the nonprofit sector: The moderating effect of
sociodemographic characteristics. International Journal o f Nonprofit and
Voluntary Sector Marketing, 19,110-126.
SchneeweiR, H. (1991). Models with latent variables: LISREL versus PLS.
Statistica Neerlandica, 45, 145-157.
Schonemann, P. H., & Wang, M.-M. (1972). Some new results on factor
indeterminacy. Psychometrika, 37, 61-91.
Schwaiger, M. (2004). Components and parameters of corporate reputation:
An empirical study. Schmalenbach Business Review, 56,46-71.
Schwaiger, M., Raithel, S., & Schloderer, M. P. (2009). Recognition or
rejection: How a company’s reputation influences stakeholder behavior.
In J. Klewes 8c R. Wreschniok (Eds.), Reputation capital: Building and
maintaining trust in the 21st century (pp. 39-55). Berlin, Heidelberg:
Springer.
Schwaiger, M., Sarstedt, M., & Taylor, C. R. (2010). Art for the sake of the
corporation: Audi, BMW Group, DaimlerChrysler, Montblanc, Siemens,
250 Nhirng van de nang cao trong PLS-SEM

and Volkswagen help explore the effect of sponsorship on corporate

reputations. Journal of Advertising Research, 50, 77-90.
Schwarz, G. (1978). Estimating the dimensions of a model. Annals of Statistics,
6, 461-464.
Shmueli, G. (2010). To explain or to predict? Statistical Science, 2 5 ,289-310.
Slack, N. (1994). The importance-performance matrix as a determinant of
improvement priority. International Journal o f Operations and Production
Management, 44, 59-75.
Spearman, C. (1927). The abilities of man. New York, NY: Macmillan.
Squillacciotti, S. (2005). Prediction-oriented classification in PLS path
modeling. In T. Aluja, J. Casanovas, V. Esposito Vinzi, & M. Tenehaus
(Eds.), PLS and marketing: Proceedings of the 4th International Symposium
on PLS and related methods (pp. 499-506). Paris, France: DECISIA.
Squillacciotti, S. (2010). Prediction-oriented classification in PLS path
modeling. In V. Esposito Vinzi, W. W. Chin, J. Henseler, tk H. Wang (Eds.),
Handbook o f partial least squares: Concepts, methods and applications in
marketing and relatedfields (pp. 219-233). Berlin, Germany: Springer.
Steenkamp, J. B. E. M., & Baumgartner, H. (1998). Assessing measurement
invariance in cross national consumer research. Journal o f Consumer
Research, 25, 78-107.
Steinley, D. (2003). Local optima in k-means clustering: What you don’t know
may hurt you. Psychological Methods, 8, 294-304.
Temme, D., & Diamantopoulos, A. (2016). Higher-order models with
reflective indicators: A rejoinder to a recent call for their abandonment.
Journal of Modelling in Management, 1 1 ,180-188.
Temme, D., Kreis, H„ &Hildebrandt, L. (2010). A comparison of current PLS
path modeling software: Features, ease-of-use, and performance. In V.
Esposito Vinzi, W. W. C hin,). Henseler, & H. Wang (Eds.), Handbook of
partial least squares: Concepts, methods and applications (pp. 737-756).
Heidelberg, Germany: Springer.
Tenenhaus, M„ Esposito Vinzi, V., Chatelin, Y.-M., & Lauro, C. (2005). PLS
path modeling. Computational Statistics &Data Analysis, 48, 159-205.
Test&Go. (2006). SPAD-PLS version 6.0.0 [Computer software], Paris, France.
Thurstone, L. L. (1947). Multiple factor analysis. Chicago, JL: University of
Chicago Press.
Valencia, J. L., & Diaz-Llanos, F. J. (2003). Regresion PLS en las ciencias
experimentales [PLS regression in the experimental sciences]. Madrid,
Spain: Editorial Complutense.
Vandenberg, R. J., & Lance, C. E. (2000). A review and synthesis of the
measurement invariance literature: Suggestions, practices, and
recommendations for organizational research. Organizational Research
Methods, 3,4-70.
 Tai li?u tham khao 251

Volckner, E, Sattler, H„ Hennig-Thurau, T., & Ringle, C. M. (2010). The role

of parent brand quality for service brand extension success. Journal of
Service Research, 13, 359-361.
Wedel, M„ & Kamakura, W. (2000). Market segmentation. Conceptual and
methodologicalfoundations (2nd ed.). Norwell, MA: Kluwer Academic.
Welch, B. L. (1947). The generalization of “student’s” problem when several
different population variances are involved. Biometrika, 34, 28-35.
Wetzels, M„ Odekerken-Schroder, G., 8c van Oppen, C. (2009). Using PLS
path modeling for assessing hierarchical construct models: Guidelines
and empirical illustration. MIS Quarterly, 33, 177-195.
Wickens, M. R. (1972). A note on the use of proxy variables. Econometrica,
4 0,759-761. v \ J
Wilden, R„ 8c Gudergan, S. (2015). The impact of dynamic capabilities on
operational marketing and technological capabilities: Investigating the
role of environmental turbulence. Journal of the Academy of Marketing
Science, 4 3 ,181-199.
Willaby, H. W„ Costa, D. S. J., Burns, B. D., MacCann, C., 8c Roberts, R. D.
(2015). Testing complex models with small sample sizes: A historical
overview and empirical demonstration of what partial least squares (PLS)
can offer differential psychology. Personality and Individual Differences,
84, 73-78.
Wold, H. O. A. (1966). Estimation of principal components and related
methods by iterative least squares. In P. R. Krishnaiah (Ed.), Multivariate
analysis (pp. 391-420). New York, NY: Academic Press.
Wold, H. O. A. (1973). Nonlinear iterative partial least squares (NIPALS)
modeling: Some current developments. In P. R. Krishnaiah (Ed.),
Multivariate analysis III (pp. 383-407). New York, NY: Academic Press.
Wold, H. O. A. (1980). Model construction and evaluation when theoretical
knowledge is scarce: Theory and application of partial least squares. In
J. Kmenta 8c J. B. Ramsey (Eds.), Evaluation of econometric models (pp.
47-74). New'York, NY: Academic Press.
Wold, H. O. A. (1982). Soft modeling: The basic design and some extensions.
In K. G. Joreskog 8c H. Wold (Eds.), Systems under indirect observations:
Part II (pp. 1-54). Amsterdam, Netherlands: North- Holland.
Wold, H. O. A. (1985). Partial least squares. In S. Kotz 8c N. L. Johnson (Eds.),
Encyclopedia of statistical sciences (pp. 581-591). New York, NY: John
Wiley.
Zhang, Y., 8c Schwaiger, M. (2009). An empirical research of corporate
reputation in China. Communicative Business, 1, 80-104.
 Nhung van de nang cao
trong md hinh phUdn^p trinh
cau true binh phuttng toi thieu rieng phan

B ien dich: TS. Nguyen Q uang Anh

C h iu trach n h ie m x u a t ban:
Cr
G iam doc ♦
PH AN N G O C CH INH A ;

C hiu trach n h iem n o i d ung:

T ong bien tap
PH A N NG O C CHINH

B ie n tap:
Nguyen Thj P h uon g Thir

T rinh bay, bio, bien tap k y th u a t:

Tran Thj Bao Ngoc

S u a ban in:
Pham Dang Binh

In 1.000 cuon, kho (16 x 24) cm, tai Cong ty CP in bao bi va XNK Tong hop
IBis Hoang Dieu, Phuong 13, Quan 4, TP. H o Chi M inh
So xac n h an DKXB: 4664-2018/C X B IPH /1-112/TC
So QDXB: 264/QD-NXBTC cap ngay 14 th an g 12 nam 2018
ISBN: 978-604-79-2003-7. In xong va nop liiu ch ieu Q uy I nam 2019