PurposiveSampling1
TedPalys SimonFraserUniversity
Tosayyouwillengageinpurposivesamplingsignifiesthatyouseesamplingasaseriesof strategicchoicesaboutwithwhom,whereandhowtodoyourresearch.Twothingsareimplicit inthatstatement.Firstisthatthewaythatyousamplehastobetiedtoyourobjectives. Secondisanimplicationthatfollowsfromthefirst,i.e.,thatthereisnoonebestsampling strategybecausewhichisbestwilldependonthecontextinwhichyouareworkingandthe natureofyourresearchobjective(s). Purposivesamplingisvirtuallysynonymouswithqualitativeresearch.However,becausethere aremanyobjectivesthatqualitativeresearchersmighthave,thelistofpurposivestrategies thatyoumightfollowisvirtuallyendless,andanygivenlistwillreflectonlytherangeof situationstheauthorofthatlisthasconsidered. Andyet,certainlytherearesomeobjectivesandintereststhatcharacterizequalitative research.Foronething,qualitativeresearchersarelessofteninterestedinaskingaboutcentral tendencyinalargergroup(e.g.,Whatdomostpeopleinthispopulationthinkaboutan issue?),andmuchmoreinterestedincasestudyanalysiswhyparticularpeople(orgroups) feelparticularways,theprocessesbywhichtheseattitudesareconstructedandtherolethey playindynamicprocesseswithintheorganizationorgroup.Embeddedinthisistheideathat whoapersonisandwherethatpersonislocatedwithinagroupisimportant,unlikeother formsofresearchwherepeopleareviewedasessentiallyinterchangeable.Research participantsarenotalwayscreatedequalonewellplacedarticulateinformantwilloften advanceyourresearchfarbetterthananyrandomlychosensampleoffiftyandthewaywe sampleneedstotakethatintoaccount. Thegeneralthemehereisthatthebiggestquestionanyresearcherneedstoaskhimorherself iswhatexactlyitisthats/hewantstoaccomplishandwhats/hewantstoknow,andthe appropriatesamplingstrategywillfollowfromthat.Someexamplesofthekindsofpurposive alternativesavailableinclude: StakeholderSampling:Particularlyusefulinthecontextofevaluationresearchand policyanalysis,thisstrategyinvolvesidentifyingwhothemajorstakeholdersarewho areinvolvedindesigning,giving,receiving,oradministeringtheprogrammeorservice beingevaluated,andwhomightotherwisebeaffectedbyit. ExtremeorDeviantCaseSampling:Sometimesextremecasesareofinterestbecause
FromLisaM.Given(Ed.)(2008).TheSageEncyclopediaofQualitativeResearchMethods.Sage:ThousandOaks, CA,Vol.2,pp.697698.
1
theyrepresentthepurestormostclearcutinstanceofaphenomenonweare interestedin.Forexample,ifwewereinterestedinstudyingmanagementstyles,it mightbemostinterestingtostudyanorganizationthatdidexceptionallywelland/or anotherthathadhighexpectationsbutdidexceptionallypoorly.
TypicalCaseSampling:Sometimesweareinterestedincasessimplybecausetheyare notunusualinanyway.Forexample,yearsagoHowardBeckerandsomeofhis colleagueswereinterestedinstudyinghowmedicalstudentsweresocializedintothe profession.TheydidtheirresearchattheUniversityofKansasMedicalSchoolnota highlyprestigiousmedicalschoolsuchasHarvardorJohnsHopkinsexactlybecause therewasnothingunusualaboutitand,forthatreason,wasprobablysomewhattypical ofthemedicalschoolexperience. ParadigmaticCaseSampling:Acaseisparadigmaticwhenitisconsideredthe exemplarforacertainclass.Forexample,ifonewantedtostudythemanagementof professionalsportsteams,theparadigmaticcaseinhockeyofasuccessfulfranchise wouldbetheMontrealCanadiens;forbaseballitwouldbetheNewYorkYankees. MaximumVariationSampling:Searchingforcasesorindividualswhocoverthe spectrumofpositionsandperspectivesinrelationtothephenomenononeisstudying, andwouldincludebothofthepreviouscategories,i.e.,bothextremeandtypicalcases plusanyotherpositionsthatcanbeidentified. CriterionSampling:Thisinvolvessearchingforcasesorindividualswhomeetacertain criterion,e.g.,thattheyhaveacertaindiseaseorhavehadaparticularlifeexperience. Forexample,acolleagueofmineisdoingresearchwithmenwhohavebeenclientsof sexworkers;thiswouldbeconsideredcriterionsampling. TheoryguidedSampling:Researcherswhoarefollowingamoredeductiveortheory testingapproachwouldbeinterestedinfindingindividualsorcasesthatembody theoreticalconstructs.Asthiscouldbeconsideredaparticulartypeofcriterion sampling,italsoillustratestheoverlapsthatcanexistbetweenthesecategories(e.g., theorybasedsamplingmightalsoleadtheresearchertolookforparticularlyintenseor extremecases). Criticalcasesampling:Heretheresearchermightbelookingforadecisivecasethat wouldhelpmakeadecisionaboutwhichofseveraldifferentexplanationsismost plausible,orisonethatisidentifiedbyexpertsasbeingaparticularlyusefulsitebecause ofthegeneralizationsitallows,e.g.,recentfindingsthatlifeexistsatthebottomofthe oceanwherethereisnosunlight,bittercold,andimmensepressure,suggeststhatlife canexistalmostanywhere. DisconfirmingorNegativeCaseSampling:Withthisstrategytheresearcherislookingto extendhisorheranalysisbylookingforcasesthatwilldisconfirmit,bothtotesttheory
andsimplybecauseitisoftenfromourfailuresthatwelearnthemost.Thegeneral principlehereis,Ifyouthinkyourresultsarenotgeneralizableortheexistenceofa particularkindofcasewillundermineallthatyouknowtobetrueabouta phenomenon,thenlookforthatkindofcase. ExpertSampling:Heretheresearcherislookingforindividualswhohaveparticular expertisethatismostlikelytobeabletoadvancetheresearchersinterestsand potentiallyopennewdoors.
Thesedonotexhaustthepossibilitiesbutillustratesomeofthestrategiclensesthroughwhich purposivesamplingcanbeconsidered.Thegeneralprinciple,however,remains,Thinkofthe personorplaceorsituationthathasthelargestpotentialforadvancingyourunderstandingand lookthere.
