7 - MCAAD - ANOVA Factoriala
7 - MCAAD - ANOVA Factoriala
7 - MCAAD - ANOVA Factoriala
Scopul
• Scopul acestei tehnici de analiză este să testeze dacă diferenţele
dintre mediile grupurilor sunt semnificative atunci când avem
două sau mai multe variabile independente.
0
Cu exercitii Fara exercitii
Fara Dieta
0
Cu exercitii Fara exercitii
• Indivizii ce NU țin dietă:
-0.5
creșterea cu 2 sau 1 kg dacă
-1
fac exerciții, respectiv dacă nu
-1.5
fac
-2
-2.5
Kilograme scăzute
-2
30
20 no barb.
one barb.
10
0
no alch. one alch
30
no barb.
20 one barb.
10
0
no alch. one alch.
20 no barb.
one barb.
10
0
no alch. one alch
• Factori:
– Sexul (B, F)
– Rasa (albi, negri, alta)
• Interacțiune posibilă:
– Gen* rasa
Nu putem folosi mai multe ANOVA?
• Din cauza ca lucram de mai multe ori cu acelasi set de date exista
riscul de a creste pe alpha (de a face erori de rangul I)
• Acest risc este controlabil: Bonferroni correction
• Sau putem combina cele 2 (3, 4, …) variabile independente si
construi o singura variabila cu k1*k2*… clase (ex: femeie alba
ocupata, ...)
• DAR – motivul cel mai important pentru care folosim ANOVA
factorial este posibilitatea de a studia interacțiunile.
Factorial ANOVA in SPSS
Comanda in SPSS
UNIANOVA
prestg80 BY sex race
/METHOD = SSTYPE(3)
/INTERCEPT = INCLUDE
/POSTHOC = race ( BTUKEY )
/PLOT = PROFILE( race*sex )
/PRINT = DESCRIPTIVE HOMOGENEITY
/CRITERIA = ALPHA(.05)
/DESIGN = sex race sex*race .
Descriptive Statistics
• Efectele principale:
– SEX: sig=0,566
– RASA sig=0,000
+ INTERACTIUNEA
Respondent's Sex * Race of Respondent
• Barbati • Femei
– albi – albe
– negri – negrese
– alta – alta
R's Occupational Prestige Score (1980)
a,b,c
Tukey B
Subset
Race of Respondent N 1 2
Black 174 37.22
Other 42 40.67 40.67
White 1202 43.84
Means for groups in homogeneous subsets are displayed.
Based on Type III Sum of Squares
The error term is Mean Square(Error) = 165.071.
a. Uses Harmonic Mean Sample Size = 98.721.
b. The group sizes are unequal. The harmonic mean of
the group sizes is used. Type I error levels are not
guaranteed.
c. Alpha = .05.
• Diferente semnificative intre albi si negri – ceilalti sunt grup tranzitoriu (nu sunt
semnificativ diferiti nici de albi, nici de negri)
Avem interacțiune!
Estimated Marginal Means of R's Occupational Prestige Score (1980)
46
44
42
Estimated Marginal Means
40
38
Respondent's Sex
36
Male
34 Female
White Black Other
Race of Respondent
Respondent's Sex * Race of Respondent
a. Exact statistic
b.
Design: Intercept+ANXIETY+TENSION+ANXIETY * TENSION
Within Subjects Design: TRIAL
Mauchly's Test of Sphericityb
Measure: MEASURE_1
a
Epsilon
Approx. Greenhous
Within Subjects Effect Mauchly's W Chi-Square df Sig. e-Geisser Huynh-Feldt Lower-bound
TRIAL .187 11.254 5 .049 .536 .902 .333
Tests the null hypothesis that the error covariance matrix of the orthonormalized transformed dependent variables is
proportional to an identity matrix.
a. May be used to adjust the degrees of freedom for the averaged tests of significance. Corrected tests are displayed in the
Tests of Within-Subjects Effects table.
b.
Design: Intercept+ANXIETY+TENSION+ANXIETY * TENSION
Within Subjects Design: TRIAL
Tests of Within-Subjects Effects
Measure: MEASURE_1
Type III Sum
Source of Squares df Mean Square F Sig.
TRIAL Sphericity Assumed 991.500 3 330.500 152.051 .000
Greenhouse-Geisser 991.500 1.608 616.432 152.051 .000
Huynh-Feldt 991.500 2.707 366.284 152.051 .000
Lower-bound 991.500 1.000 991.500 152.051 .000
TRIAL * ANXIETY Sphericity Assumed 8.417 3 2.806 1.291 .300
Greenhouse-Geisser 8.417 1.608 5.233 1.291 .300
Huynh-Feldt 8.417 2.707 3.109 1.291 .301
Lower-bound 8.417 1.000 8.417 1.291 .289
TRIAL * TENSION Sphericity Assumed 12.167 3 4.056 1.866 .162
Greenhouse-Geisser 12.167 1.608 7.564 1.866 .197
Huynh-Feldt 12.167 2.707 4.495 1.866 .169
Lower-bound 12.167 1.000 12.167 1.866 .209
TRIAL * ANXIETY Sphericity Assumed 12.750 3 4.250 1.955 .148
* TENSION Greenhouse-Geisser 12.750 1.608 7.927 1.955 .185
Huynh-Feldt 12.750 2.707 4.710 1.955 .155
Lower-bound
12.750 1.000 12.750 1.955 .200
Measure: MEASURE_1
Type III Sum
Source TRIAL of Squares df Mean Square F Sig.
TRIAL Linear 984.150 1 984.150 247.845 .000
Quadratic 6.750 1 6.750 3.411 .102
Cubic .600 1 .600 1.051 .335
TRIAL * ANXIETY Linear 1.667 1 1.667 .420 .535
Quadratic 3.000 1 3.000 1.516 .253
Cubic 3.750 1 3.750 6.569 .033
TRIAL * TENSION Linear 10.417 1 10.417 2.623 .144
Quadratic .083 1 .083 .042 .843
Cubic 1.667 1 1.667 2.920 .126
TRIAL * ANXIETY Linear 9.600 1 9.600 2.418 .159
* TENSION Quadratic .333 1 .333 .168 .692
Cubic 2.817 1 2.817 4.934 .057
Error(TRIAL) Linear 31.767 8 3.971
Quadratic 15.833 8 1.979
Cubic 4.567 8 .571