Sampling
Sampling
Sampling
SAMPLING
CHAPTER PREVIEW
Proportionate stratified randorn sampling has mar stratum while underepresents the other strata. ln this example, when he randonly draws 500
easily applied in military studies where r a n k t e n aPpatons. 5 readily and males and s00 females, he is overepresenting a female stratum and underrepresenting
male stratum.
project, Likewise, such sampling is considered important in the siusdw of scbol yems
t ane
15/05/2024
350 Test,
Meanarememtr ad Research AMbod b Schandural Scence
Sangtig 391
Disproportionate stratified random sampling has both advantages and duavatagns,
major an are indicated asfolie
Agven hereunder area sampling has many research utilioies or applications.
proportionate t e i h e d random sampling is comparatively less itime-cotining t a Lage scale suveys of political, eligious and social belhaviout are easily conducted by
s nol
because here the imvestigator edhe
making proportionate reorent inaccessible, area
2. In disproportlonate stratifed rondom 1peclic indivihuals are unobtainable or ane
h lesieatoc s able t0 EVe weidl . sampling be
he particular goups of elements that are not noreented as freguently in he Populaian a
Compared with other elements Gi) Public opinion polls are casily and snoothly conducted using an area sampling plan.
Area sampling has some advantapes and disadvantages. The important advanags are
nemajordisadvantages of dispropontionatestratified random sampling are gven below Fiven below:
some other trata anpng certain stratum of the population is ovetrcpheenled and arv When larger geographical areas are to be covered, it is easier to tse ziea
one set of elementsorrnesamples draw. Ihezugi bonmay
r htd have the ig of the indivicuais inhbiting a pln ama He imply draw some geographical
the sample. Such type of sample may not bea true ections randomly, and, subseguenty, he interviews all the tamilies or persons living in the
representative. anoomiy drawn sections, In this sense, arca sampling is easier than the other two precedin6
2. Disproportionate stratified random sampling assurnes that the investiga khows t
onginal population. This automatically meants thatthis method of sarnnli 2. Inarei sampling respondents can readily be substiuted for other respondents within the
original population, asituation where the investigator has no idea about the comporition of ame random section Ths is pernitted because cters of elenents are sapled and npt
the individuals.
a.3. In this method of sampling the investigator is requircd to clasiby the po_ulation ir 3, Area sampling saves both time and money. The investigator can concentrate his eforts
diferen substrata and from each substratum, he takes more or less, equal number of caes
drawing asample from such aclasified set, he may actually misclassifycertain eletents, thati comnared tnthe h hot nhabi molin The iniestieator need vot travel preat
hemay put element Ainto stratumXwhereas Amight belongto stratun Z distances to interview specific individuals residing at random points in a cetain geographical
efcf
in social science reearches. disproporionale ameknore ppuar an
BUse of ease and convenience
Tegion.
proportionate stratified random sampling. Aea lar cluster) sampling pessSses the tratolexiy ha mataea
Sampins
AREA (OR CLUSTER) SAMPLING stages, For example, aftar drawing several sections randomly in the first stage, the investigator
Ares lorcl
may decide o stratilyeach ofthe fint-stage units and selet sections from sorne strata to constitute
babilit
method has its eien in the B d a n
oling, Such san he second stage units. The imetigatoe may use a third type of sampling plan in the third or
determine the effect of various kinds of fertilizers, sol treatments and a variety of planting 5. Sill another advantage of area sampling is that the respondents can teadily be
methods on crop yield, mostly used this method of sampling. In social sciences apnlication d rubstiuted for other respondents within the sarne random section This further increases the
area sampling has been etensive in survey reseach and fieldresearch. degree of flexibilly in the area sampling
eiahbourhooe
itiesaresmpins eneraly geoyaphcaldrvsionsateoy c o m m o Area sampling has some disadvantages too. The important disadvantages are given belor
Samping
called sample. The investigator or interviewer proceds to interview all elements of the randonmly a r h l e ennintoh mehodeofonhahility amplin
drawn areas or clusters, That s the reason why this method of sampling is alsco known as 2. In arca sampling there is no correct way to ensure thateach sampling unit included in an
Cluster sampling arEa sample willbe of equal sze. In fact, here the researcher has litle contol overthe size ofeach
Suppose the investigator wants to asess he eoole of Tamil Nedu t cluster. This introduces bias into the samples.
amsn tnil N individuals included in one clster
o him and shen dide it into various setions according to a nunbar c ttlcal and den chters mav be that an individual is interviewed in
horizontal grid lines drawn acros the total area, He will then number each section fromni 1o N, N one cluter and next moming he travels to another area that falls within another randomly
being equal to the total number of sections, With the help of the table of random nunbus, he wil selected section.
draw a specfied number of sections to constitute the sample that he will finally tudy. The Despite these disadvantages, area sampling lis a common and popularmethod of sampling in
behavioural researches.
ivestigator will, then, interview ail persons or members offamilies living in those seciions. lf any
o
drawn t etle ww acroing to receity QUOTA SAMPLING
kurther subdivision and selection of samples can be done to different stages. Thit s called Quota sampling is one of the important types of non-probability sampling methods which is
multi-stage sampling aprarently similar to stratiied radom sampling, In quola sanpling the investigator recognizes
15/05/
s54 Tets, Moaaurvents and Besarch Atethote i
tbateurat Sieno Samping 355
naccidental sampling the probability of ampling error is high. Thettioe, the vala
and reliability of this method are badly affected mal soclal telations among inindlviduals.
ueo or ooanrg n mpp hidden' population which is not
and socioloÝvantages, it will not be an exapgeration to saythat in my ptycholo doreinble aswhn hedealswahdur addcts. hardened ctiminals, prostutes. and so on
reearches this method of sampling is frequently used uch case, snowball sarnpling is Used. Here the researcheridentie be obrains till
SYSTEMATIC SAMPLING hin. he obtains the name anid addreses of otnbldor'snowbal. Coleman et
Systematic sampling sanother mehod of non-orobsbiliy sampling plar iough the (Hla67edeuball nln uccosiulyin astudy of difusion of medical hom
e5somwhat misleading in the sense that all probability sanod are
soaould
physicians. The purpose was odetermine how physicians came to noldfsion
included u n d e r u eto th, t often sounds that ycs such as drugs and related supples Snowaiansclioues through which
infornation s reealed. tt rnay anewer the questions: Do the physicians read o e
Systematic sampling may be delined a drawin or selecting every rth Person from
Predetemined list of elements or individual Selecting every 5th rol nuinber in a clas ol il njoumals or hear about it in some medical eminar r the he ingi How
udents will constitute systematic sampling. Likewise, drawing every 8th nare omaelepho ror convention, whom do they contac a m e w t And sp on.
n e aimate sampling we payateno tpl El b ota tesarrh apolications in relatively smail business and
Such sarmpling s mOre
rial orpanisations where N s expected not to exceed 100. among
element selected is arandom one) andat the simeine, possesses some non-probabilay r specitic segerte
such as excluding ali persons between every nth element chosen. enient to the studies of social change and diffusson of inlormation
Systernatic sampling has some advantages and disadyantages. The importan advantagef ocial organizations.
The important advantuges are
mentioned below: Snowbal sampling has some advantages and disadvantages.
btaining a smple of elenm
Sytemabc sampings relativelYa quick menoa o
the investie liinates 1. Snowball sampling, which is primarily a sociometric groupsampling technique, has proved
method of samp
yimportant and is helpful in studying small informal social and ls impact upon formal
otherwise taken in different methods of sampling. anizational structure.
2Systematic sarnpling makes it very easy to check whether every nth number or name veals cortsnunicatlon patterm in comimunity organization
been selected, In case, thete occurs any ernor in counting, that is, i the invetigator selects 2. Snowball ampins kiny can also be studied with the help of such
Numoer instead of 5th number, his sample will not be hity poW
senousyhainet mpling technique.
a tble ofandom mmr tordin the camnle rfixed ouctafrom cach stratum cf t 3. Themethod of snowball sampling ls anenable to various scientitic sampling procedures
arious stages such as use of random
population in orderto have proportional representation. numtnoleasdeeribed
sniethese advarntages, snowbalsan below.
Despite these advantages, there are some limitations of sysematic sampling cumbersome ot hen N is large or sy
indicated overleaf: Snowbal sampling becomes
1. Systematic sämpling igneres all persons betwcen every nth element chocen, Obviout statistical
2. This mathod of sampling does not allow the researcher to use probability hey are
then, itis not a probability samplio8 plan. ethods, ln fact, the elerments included in sample ate not randomly drawn and
noling error increases ifthe list is arranged in a partic
order, say. the list increases or decreascs with respect to some trait such as age, educatio pendent on the subjective choices of the originally selected respondents. This introduces
income, caste, etc. In such asltuation abias will be introduced in the sample, because mo t R e whall samoliog is used in the study of diffusionof some specie
individuals are likely to be drawn from one group, particularly from that when
group thewhich
lis onand social changes
enimos eenty on lhe group. Such type of bias may also resut
SATURATION SAMPLING AND DENSE SAMPLING
Despite these limitations, systematic sampling is frequently used in psychological ant of sampllog techniques which are used less
Colenan (1959) häs cmpnaedline Saluration sampling is defined a
sociological researches because this method of sampling posesses the trait of both probabilly
and non-probability sampling Mrawing alt elements or individuals havíng characteristics of interest to the investigator. Drawing
SNOWBALL SAMPLING all pysicians having at least the age of 45 (Urom asmall community, woula be e st
Mmpling Dense smpng whe esecber olects 50% or more from the
Snowball sampling, which is a non-probabilily sampling method, is basicaly indirecty
sociometric. nis defined as having all the persons in agroup or onganisation identifying thel iokesamaioriy of indviduas haviog specified tralts or characteristics which are
friends whco in turn identily their friends and associates until the researcher observes that interest to him, it is called dense sampling For example, if the esearcher selects 500 to 600
behaviour is usually used
udents from a population of 1,000 students, it will constitute dense sampling
15/05/2024
356 Tet, loaerees and Bengrch
hod in bastounal Scaencs
Sanpl 357
The advantaee
methods of samuration sampling and dense sampling is hat the are cneu
h t h e population does not eceed 1,000. in caeit exC
methods of sarmoline dilfererce between the population value and the sampie value.
of a sample taken
The srlationship
ecumbersome and inconvenient.
DOUBLE SAMPLING apopulation will not ordinarily be equal to the rnean o T h e larger the size ofa
Double: en thesize of the samgle and the era, as elationship between the sze o
mnle n stis name implies is defined as drawing asample cof indivdals fom an eand tbn etrk demtraied in Fieure 141. When the size of etion to those of
having N-10p00 sgator randomly draws asample of 0 ma pop helorm and the statistics of the sample distribution becomec
a e u e d Hut a very lge
dividuals, he again randomly draws
further study, This is called double sumoling nlation. Hence, the error is reduced and Iheretore a aohlerns, which soetmes
becatie it tends
that result
UPpore the reseurcher is studying the atitude of newly married couples towards o
plarnning through mall questionnaire. For thi. he malled 1.000 questionnaiis lo one thou
aing in dilerent localities, He finds that only 40% that is 4o0) qutionnaires
questionnaire lo eet t ranooTy a sample of tu ai angl
tnowledge towards the differen s of fan
planning. This is called double sampling
oouble sampling to be an effective part of the research, tis essentialat the samS
plan tollowed from the beginning should maximire the representativeness of the indi
ently. Ths method has the disadvantage of taking much irt and labo
Se of sangla
esror
REQUISITES OF AGOOD SAMPLING METHOD F 144 elationahp betveen e siac ofsanple and he
Any sampling method to be good and sclentilically sound must possess at least the
lolwing propertie:
COMMON ADVANTAGES OF SAMPLING METHODS nt andeasy, In any
eenatveness of the rample
2 the purpoe of any sound sampling procedure is 1o make amping mTee l l t r s
The first equirement nethod is that the selected satrples mat ahavioural research the whale popairo esofthesample dawn from the population
representativeof the population. What is meant by representativeness ofthe samp In the wo Popuationbaod conckdon is peneralized, Apat from serving n
of Kerlinger (1973: 119),"Ordinarily, representative mcans to betypical of apopulation, thatis Unction of generalization, sarnpling has some additional advantages, ple land not the
e anetso the popalaton. In teserch a rpreseal emeans th 1, Accuracy is increased When one is entrusted with exrns can easily
question, Truly speaking, an investicator cannot be cerain about the representativenes o SFulation, naturally, a Fea vou scan alvp be analyzedwith great care. Not
sample unless the entire population is tested. But this is ordinarily not feasibie. Hence, he mu at a lower cost when the number of
ly this. the services of trained personnel can be utilized that
Chses to be processed is low. The overall resuts of all thiss the accuracy of the investigaton
atisty hime tohewie. Aample can be regarded as being reprevenative of he populatig
as is in the populationy l sex is the levat charareritie oodtion of at Iincreased.
When the data are to be collected from a limited number
of cases rather
particular university afd if the population has a male female ratio of 60:40, a representatiy C o s 3 e u u l eentailk areduced cost 1o the investigator. Not only
with only a
Sanple would be one where the ratio of males and fernales is also 60:40. For ensurin sthe tine taken in processing of the data is also reduced. When one is dealing
calculation
the chances of accidental errot in statistical
a suitable and approariae definition of the obuenaticns which ae to contitute samne imited umber of data,
frarmed. An observation must be defined in the light of the definiöon of population othervige t8 so minimised. frcom a sample rather than from the
3. Speed is increased When the data are collected
result will be a biased srnple. Suppose that the population is defined as all those fen ales who ars the investigator can do his work speedily without doing any injustice to thethe
Investigation. This is one of the vital considerations specially when the infomation about
cbservation has been defined differently from the population, Obviously, this definition o ample is reeded urgently
cbservation will automatically excude those who refuse to admit that they are widows and 'SAMPLING DISTRIBUTION
to the research
oling is that the sarnples In any sampling situation, three types of distributions are of common interest
said to be adequate when it is al suflicient sie.There ts no tiid nule treandine the nvesigator: he sample distribution, the population distributlon,disributiont
and the sampins
sample. Buta largersample, in general, is prelered because it tends to reduce the er.ot, which is The meaning of and distinction between these three types of
through an illustration.
15/05/2024
Sonpnur 359
358 Ts, Afeaurement and Bnurch hod be hoa Scicee the
Supao tahle and sends to vary, naturaily,
Suppose in a small town the total number of graduate parents is J Owin t luther if the natue of the population is such hal t no a eod to difler. Generaly, a
mene n nowing the mean fertility rate anong thee reforo, we an kiccessive samples taken from the sane populaton nooutatian and the accus
examine only a random mnf
n n d data my bn aranged Nndom sample provides a poo
amle, As the size of the samnlaion, For
fquency distributlon and the deiired statisticsmy be calculated This lyntributioc FProamation depena l ditrution come nearer to ted by the frequency
omele when e re ol the samole is 100, the distributo a he fom of the
rown as the sample distribution becase it shows the disribution e cal ondfion a very clear
an nhomiy rom the population. Now, f win e t d aane them in f
institutionw is usually smooth enough to give
polygon
successive sampis and
dstribution, ths distrlbutionis termed as the tlation disribution. in realty,e lons and population distribution luctuate less than others in
paramcters of a population are not ordinarily known, Therelore, an esthese he There are certain statistics which
mling forn a nomat d n o e s of
POgertes of the population is made on the basis of the lom ano erme'ta im hence, the sarmpling errorsthn other measuresof cenan because they
distribution will alo be nermal thatiadandad)deviation of the fotilty rae ol t abiih Hane thre oare more reliable than olner bence, tend to fluctuate
300 graduate parents are 3.89 and 1.02 respectivey, we assume that the men the stande population paamts
approximate more closely the coesponding the
deviation of the population distribution are also 3.89 and 1.02 ess from sarmple to sample and thus produce least sam which are o
espee ortant distributions
importa
enhs example To0 sampes con et 100 cifferent In most sampling situatlons Ihere e e anolicatlon of the sampts
Concen to the reader in ordet o a i i o n distribution and the s a P e mean
distibutions, that ls,100 diferent means and 100diferent standard dedation we goup
100 means and the 100 standard deviations into frequenicy cdisributios YWe will he pxnlaied with an exampler Suppoe, w nd Seel factory,
samping Csribtion of mean and the sampling
aichsobained if al otelieenee score of workers (in the age range of 2 0 a 30, 000,Owing to time
samples of the given size are aken from the pooulatlon. Further suppose that thetotal number ofworker in mnle of 300 and administer a
and money consideratiots,w e o r on the test by 300 workers expit
SAMPLING ERROR
refers to the process of drawing a lm andardized tet on l dtesnple distribution. Thus, the sample dst
descriptive statistics. Hence, sanple
Se a mutve described by any
nlinr procedure used, the sample mu obyened distribution which can be sampling situalion.
represent, as far as possible, all charactLritcs of the population that are related lo the proble distribution is the concrete distribution in any on the whoie population, that is
under study. However, ifwe decide to administe the tet wn convert this set into frequenicy
Astatistical meaure calculated froma sample ts and the co
knownasalen e on 30,000 workers, weshal be e t i n diariution, Ordinarily, the form and he
imtruii iomntordto a8an obainedmeasure or derivcmeaure disibution, ths w hten are not known. However, we can inter them on
mean calculated from a sample of a population is called "statistlc mean and the nlue of mea can be interred
odihtion the sanple distribution is nomal,the itsarnple distribution to the
ascribing any irGgularities in
based directly upon the population is known as parameter mean anpi3 eror i t population distribution is also nomal,
orot b difernce
hen' antet and an etinale that paraneter which is deriv ampling fluctuations. of the sample
population distribution on the basis
from a sample. The inlerences drawn regarding lo the above exarnple we do not have the
A sampling problen is said to exist whenever the conclusions obtained on the basis of distribution may not be correo i c h wnid be obtained again in the second of
nieal in bind samle of 100 from the population, f we were to take, say, 100 sanpiestheof s0o eac
standard deviations. If ail 1
histher the sanpling ero, the poprer the statistica inlerence. The tern "sampling "or does ngt would be petting 100 different means, 100 varying wotd have an
sugzest any mistake in the process of sampling itself rather it merely sugsest the chancg 100 standard deviations are grouped in trequeiy d b a i o n As this is not
vamtons wch are nevtable wmen anumoer of randomy seected simpies a n romtho exact savpling distribution of men ana uibuion whlch may be thought ol as a
characteristics fom the sample characteristicr no longer romains an exact process i has bean done, sampling distibuonsa l nf a gven statistic were computea m en
oberved that a sample statistic does not ordinarily agree with that cbtained lc the second O h t n o u l eof even size actually taken from the popoaio
possit distribution of the majority of
third or fourth sarrpleor with the population pararneter, The amount of flucluation ta: the sampla
Variallangtwofactors: Satisticians have been able to deternine the sampling
errot
the normal population. When the:
comnonly used statistics in the samples taken from investigator to delermine the re
distribution of a statistic is known, it is possible for the
e popaaon
o0 Size of the sample 15/05/2024
300 Tent, Measem au
Benih atbod r Bbur i
Iequency wih whlch andort anpln
dilerent sumnle yales Jed to occu dAer k
dion in which the statistics have some of them asunedor bypodicular s
eiaive fres
o yhe investigator is able to lnow the probabiliy owing lho al
gure henpling fluctuations or chante,
Ultimately.
reanding the valar ofae
15
oe1o judge the soundnesola hunhek
inapooulaion SOCIAL SCIENTIFIC RESEARCH
15/05/2024 151