Upload
lamminh
View
215
Download
2
Embed Size (px)
Citation preview
MethodsWorkshopfortheMinistryofHealthandLong-TermCare
AssessingResearchProtocols:MixedMethodsResearchBy:YelenaPetrosyan,MD
Mixedmethods:Definition
Mixedmethodsconsistsofthecollectionoranalysisofbothquantitativeandqualitativedatainasinglestudyinwhichthedataarecollectedconcurrentlyorsequentially,aregivenapriority,andinvolvetheintegrationofdataatoneormorestagesintheprocessofresearch.Themajormixedmethodsdesigns
1. Theconvergentparalleldesign
Theconvergentparalleldesign(convergent/triangulationdesign)occurswhentheresearchersuseconcurrenttimingtoimplementthequantitativeandqualitativestudiesduringthesamephaseoftheresearchprocess.Thetwomethodsinthisdesignhaveanequalprioritysothatbothplayanequallyimportantroleinaddressingtheresearchproblem.Thisdesignkeepsthestudiesindependentduringthedatacollectionandanalysisandthenmixesormergestheresultsduringtheoverallinterpretation.Thepurposefortheconvergentdesign
• Totriangulatethemethodsbydirectlycomparingandcontrastingquantitativeresultswithqualitativefindingsforcorroborationandvalidationpurposes;
• Toillustratequantitativeresultswithqualitativefindingstodevelopamorecompleteunderstandingofaphenomenon.
Advantagesoftheconvergentdesign
• Itisanefficientdesign,inwhichbothtypesofdataarecollectedduringonephaseoftheresearchatroughlythesametime.
• Eachtypeofdatacanbecollectedandanalyzedseparatelyandindependently,usingthetechniquestraditionallyassociatedwitheachdatatype.
Challengesinusingtheconvergentdesign
• Mucheffortandexpertiseisrequired,particularlybecauseoftheconcurrentdatacollectionandthefactthatequalweightisusuallygiventoeachdatatype.Thiscanbeaddressedbyformingaresearchteamthatincludesmemberswhohavequantitativeandqualitativeexpertise,orbytrainingsingleresearchersinbothquantitativeandqualitativeresearch.
MethodsWorkshopfortheMinistryofHealthandLong-TermCare
• Researchersneedtoconsidertheconsequencesofhavingdifferentsamplesanddifferentsamplesizeswhenmergingthetwodatasets.Differentsamplesizesmayoccurbecausethequantitativeandqualitativedataareusuallycollectedfordifferentpurposes.
• Itcanbechallengingtomergetwosetsofverydifferentdatainameaningfulway;• Itcanbechallengingifthequantitativeandqualitativeresultsdonotagree.
Contradictionsmayprovidenewinsightsintothetopic,butthesedifferencescanbedifficulttoresolveandmayrequirethecollectionofadditionaldata.
2. Theexplanatorysequentialdesign
Theexplanatorysequentialdesignusessequentialtiming.Thisdesignstartswiththecollectionandanalysisofquantitativedata,whichhasthepriorityforaddressingthestudy’squestions.Thisfirstphaseisfollowedbythesubsequentcollectionandanalysisofqualitativedata.Thesecond,qualitativephaseofthestudyisdesignedsothatitfollowsfromtheresultsofthefirst,quantitativephase.Finally,theresearcherinterpretstowhatextentandinwhatwaysthequalitativeresultshelptoexplaintheinitialquantitativeresults.Thepurposefortheexplanatorydesign
• Touseaqualitativestrandtoexplaininitialquantitativeresults;• Tousequantitativeresultsaboutparticipantcharacteristicstoguidepurposeful
samplingforaqualitativephase.Advantagesoftheexplanatorydesign
• Thisdesignappealstoquantitativeresearchers,becauseitoftenbeginswithastrongquantitativeorientation.
• Itstwo-phasestructuremakesitstraightforwardtoimplement,becausetheresearcherconductsthetwomethodsinseparatephasesandcollectsonlyonetypeofdataatatime.Therefore,singleresearcherscanconductthisdesign.
• Thisdesignlendsitselftoemergentapproacheswherethesecondphasecanbedesignedbasedonwhatislearnedfromtheinitialquantitativephase.
Challengesinusingtheexplanatorydesign
• Thisdesignrequiresalengthyamountoftimeforimplementingthetwophases;• Itcanbedifficulttosecureinstitutionalreviewboard(IRB)approvalforthisdesign,
becausetheresearchercannotspecifyhowparticipantswillbeselectedforthesecondphaseuntiltheinitialfindingsareobtained.
• Theresearchermustdecidewhichquantitativeresultsneedtobefurtherexplained.• Theresearchermustdecidewhotosampleinthesecondphaseandwhatcriteriato
useforparticipantselection.
MethodsWorkshopfortheMinistryofHealthandLong-TermCare
3. Theexploratorysequentialdesign
Theexploratorysequentialdesignalsousessequentialtiming.Incontrasttotheexplanatorydesign,theexploratorydesignbeginswithandprioritizesthecollectionandanalysisofqualitativedatainthefirstphase.Buildingfromthequalitativeresults,theresearcherconductsasecond,quantitativephasetotestorgeneralizetheinitialfindings.Finally,theresearcherinterpretshowthequantitativeresultsbuildontheinitialqualitativeresults.Purposefortheexploratorydesign
• Togeneralizequalitativefindingsbasedonafewindividualsfromthefirstphasetoalargersamplegatheredduringthesecondphase;
• Theresultsofthefirst,qualitativemethodcanhelpdeveloporinformthesecond,quantitativemethod;
• Todevelopandtestaninstrumentbecauseoneisnotavailable;• Toidentifyimportantvariablesforthequantitativestudywhenthevariablesare
unknown;• Toexploreaphenomenonindepthandmeasuretheprevalenceofitsdimensions.
Advantagesoftheexploratorydesign
• Separatephasesmaketheexploratorydesignstraightforwardtodescribe,implement,andreport.
• Theresearchercanproduceanewinstrumentasoneofthepotentialproductsoftheresearchprocess.
Challengesinusingtheexploratorydesign
• Thetwo-phaseapproachrequiresconsiderabletimetoimplement,potentiallyincludingtimetodevelopanewinstrument.
• ItisdifficulttospecifytheproceduresofthequantitativephasewhenapplyingforinitialIRBapprovalforthestudy;
• Researchersshouldconsiderusingasmallpurposefulsampleinthefirstphaseandalargesampleofdifferentparticipantsinthesecondphasetoavoidquestionsofbiasinthequantitativestudy.
• Proceduresshouldbeundertakentoensurethatthescoresdevelopedontheinstrumentarevalidandreliable.
4. Theembeddeddesign Theembeddeddesignoccurswhentheresearchercollectsandanalyzesbothquantitativeandqualitativedatawithinatraditionalquantitativeorqualitativedesign.Inanembeddeddesign,theresearchermayaddaqualitativestudywithinaquantitative
MethodsWorkshopfortheMinistryofHealthandLong-TermCare
design,suchasanexperiment,oraddaquantitativestudywithinaqualitativedesign,suchasacasestudy.Intheembeddeddesign,thesupplementalstudyisaddedtoenhancetheoveralldesigninsomeway.
Thepurposefortheembeddeddesign• Toincludequalitativedatatoanswerasecondaryresearchquestionwithinthe
predominantlyquantitativestudy;• Toimproverecruitmentprocedures;• Toexaminetheprocessofanintervention;• Toexplainreactionstoparticipationinanexperiment.
Advantagesoftheembeddeddesign• Thisdesigncanbeusedwhentheresearcherdoesnothavesufficienttimeor
resourcestocommittoextensivequantitativeandqualitativedatacollectionbecauseonedatatypeisgivenlessprioritythantheother.
• Bytheadditionofsupplementaldata,theresearcherisabletoimprovethelargerdesign.
• Researchersusinganembeddeddesigncankeepthetwosetsofresultsseparateintheirreportsorevenreporttheminseparatepapers.
Challengesinusingtheembeddeddesign
• Theresearcherneedstohaveexpertiseinthequantitativeorqualitativedesignusedinadditiontoexpertiseinmixedmethodsresearch.
• Theresearchermustspecifythepurposeofcollectingqualitative(orquantitative)dataaspartofalargerquantitative(orqualitative)study.
• Itcanbedifficulttointegratetheresultswhenthetwomethodsareusedtoanswerdifferentresearchquestions.
Samplingissues
• Somechallengesspecifictoconcurrentdesigns(i.e.,mergingquantitativeandqualitativeresearch)includehavingadequatesamplesizesforanalyses,usingcomparablesamples,andemployingaconsistentunitofanalysisacrossthedatabases.
• Forsequentialdesigns(i.e.,onephaseofqualitativeresearchbuildsonthequantitativephaseorviceversa),theissuesrelatetodecidingwhatresultsfromthefirstphasetouseinthefollow-upphase,choosingsamplesandestimatingreasonablesamplesizesforbothphases,andinterpretingresultsfrombothphases.
Analyticandinterpretiveissues
MethodsWorkshopfortheMinistryofHealthandLong-TermCare
Issuesariseduringdataanalysisandinterpretationwhenusingspecificdesigns.Whentheinvestigatormergesthedataduringaconcurrentdesign,thefindingsmayconflictorbecontradictory.Astrategyofresolvingdifferencesneedstobeconsidered,suchasgatheringmoredataorrevisitingthedatabases.Fordesignsinvolvingasequentialdesignwithonephasefollowingtheother,thekeyissuessurroundthe“pointofinterface”inwhichtheinvestigatorneedstodecidewhatresultsfromthefirstphasewillbethefocusofattentionforthefollow-updatacollection.Makinganinterpretationbasedonintegratedresultsmaybechallengingbecauseoftheunequalemphasisplacedoneachdatasetbytheinvestigatororteam,theaccuracyorvalidityofeachdataset,andwhetherphilosophiesrelatedtoquantitativeorqualitativeresearchcanorshouldbecombined.IntegrationofQualitativeandQuantitativedataMergingdata• Byreportingresultstogetherinadiscussionsectionofastudy,suchasreportingfirst
thequantitativestatisticalresultsfollowedbyqualitativequotesorthemesthatsupportorrefutethequantitativeresults.
• Theconversionortransformationofonedatatypeintotheothersothatbothcanbeanalyzedtogether:
o Quantitativedataarenumericallycodedandincludedwithquantitativedatainstatisticalanalysis.
o Quantitativedataaretransformedintonarrativeandincludedwithqualitativedatainthematicanalysis.
Connectingdata
o Thisintegrationinvolvesanalyzingonedataset(e.g.,aquantitativesurvey),andthenusingtheinformationtoinformthesubsequentdatacollection(e.g.,interviewquestions,identificationofparticipantstointerview).Inthisway,theintegrationoccursbyconnectingtheanalysisoftheresultsfromtheinitialstudyorphasewiththedatacollectionfromthesecondstudyorphase.
Embeddingdata
o Theresearchermayembedasupplementalqualitativedatawithinalargerquantitative(e.g.,experimental)designorembedaquantitativedatawithinalargerqualitative(e.g.,narrative)design.Theembeddednatureoccursatthedesignlevel,inthattheembeddedmethodisconductedspecificallytofitthecontextofthelargerquantitativeorqualitativedesignframework.
MethodsWorkshopfortheMinistryofHealthandLong-TermCare
AchecklistforreviewingNIHmixedmethodsapplication
Significance• Doestheapplicationmakeaconvincingcasethattheproblemisrelevant(e.g.,if
aimsareachieved,theworkwillimproveknowledgeorpractice)?• Cantheproblembebeststudiedthroughthemultipleperspectivesofmixed
methodsresearch?
Investigator(s)
• Dotheinvestigator(s)havetherequiredskillstoconductallproposedmethods(e.g.,investigator(s)havepriorpublicationsand/orgrantsrelatedtoproposedqualitative,quantitative,andmixedmethods;co-investigatorswithappropriateexpertiseareidentifiedtoleadeachmethodasneeded)?
• Isthereevidencethattheprojectleadershipiscommittedtomixedmethodsresearch(e.g.,eachcomponentofthestudyisaddressedsufficientlyandconsistentlythroughouttheapplication;therearereferencestocurrentrelevantliteratureonmixedmethods;investigatorshaveexperiencesinprofessionaldevelopmentinmixedmethods)?
• Hastheapproachtocollaborationbeendescribed(e.g.,frequencyofmeetingsbetweenleadersofdifferentcomponents,managementofdifferencesbetweenco-investigators)?
Innovation
• Doestheuseofmixedmethodsprovideaplatformforinnovativeinvestigationoftheresearchproblem(s)(e.g.,providesinsightsintomechanismsoforganizationalchangenotpossiblewithasinglemethod)?
• Isthecombinationofmethodsusedinnovative,orthewayinwhichtheyareintegratedinnovative?
Approach
• Isthereadescriptionofthephilosophyortheoryinformingtheresearchandthewaysthisphilosophyortheoryshapestheinvestigation?
• Havetheapplicantsofferedaconvincingexplanationofwhymixedmethodsresearchisneededtoaddressthestudyaimsandthevalueaddedbyusingthisapproach(e.g.,explainedhowalternativedesignswouldbeinappropriateorinadequate)?
• Isthereacleardescriptionofthefullstudydesign,includingwhereintegrationoccurs(e.g.,usingacomprehensivefigureormatrix)?
• Istheintegrationofthemethodswelldescribed,includingthetiming,techniques,andresponsibilitiesforintegration?
• Isthedesignappropriateforthestudyaims?
MethodsWorkshopfortheMinistryofHealthandLong-TermCare
• Arethemethodsconsistentwithestablishedstandardsofrigorforquantitativeandqualitativedatacollectionandanalysis(e.g.,sampling,samplesizeandanalysisplansarespecifiedforeachmethod,withappropriatecitations)?
• Willappropriatecomputersoftwarebeusedforeachanalyticcomponent,andifnot,isaconvincingrationaleprovided?
• Isthestudyfeasiblewithinitsproposedtimeframeandresources(e.g.,atimetableisprovidedthatallocatestimefordataintegration)?
Environment• Isthereevidencethattheinstitutionsupportsmixedmethodsresearch(e.g.,forums
formultidisciplinarycollaborations,facultywithfundingformixedmethodsresearch)?
MethodsWorkshopfortheMinistryofHealthandLong-TermCare
ReferencesCreswell,J.W.,&PlanoClark,V.L.(2007).Designingandconductingmixedmethods
research(Chapter4Choosingamixedmethodsdesign).ThousandOaks:SagePublication.Cohen,D.,&Crabtree,B.(2008).RobertWoodJohnson,qualitativeresearchguidelines
project.Retrievedfromhttp://www.qualres.org/.Creswell,J.W.,&PlanoClark,V.L.(2011).Designingandconductingmixedmethods
research(2nded.).ThousandOaks,CA:Sage.Frechtling,J.(January,2002).The2002userfriendlyhandbookforprojectevaluation.WashingtonD.C.:NationalScienceFoundation.Retrievedfromhttp://www.nsf.gov/pubs/2002/nsf02057/start.htm.LevinJ.S.,Glass,T.A.,Kushi,L.H.,Schuck,J.R.,Steele,L.,&Jonas,W.B.(1997).Quantitativemethodsinresearchcomplementaryandalternativemedicine,MedicalCare,
35,1079-1094.O’Cathain,A.(2010).Assessingthequalityofmixedmethodsresearch:Towardacomprehensiveframework.InA.Tashakkori&C.Teddlie(Eds.),SAGEhandbookofmixedmethodsinsocial&behavioralresearch(2nded.)(pp.531-555)ThousandOaks,CA:Sage.O’Cathain,A.,Murphy,E.,&Nicholl,J.(2008).Thequalityofmixedmethodsstudiesin
healthservicesresearch.JournalofHealthServicesResearchPolicy,13(2),92-98.Tashakkori,A.,&Creswell,J.W.(2007).Theneweraofmixedmethods.[Editorial].Journal
ofMixedMethodsResearch,1(1),3-7.Caracelli,V.J.,&Greene,J.C.(1993).Dataanalysisstrategiesformixed-methodevaluationdesigns.EducationalEvaluationandPolicyAnalysis,15,195-207.Greene,J.C.(2007).Mixedmethodsinsocialinquiry.SanFrancisco:JohnWiley&Sons.Morse,J.,&Niehaus,L.(2009).Mixedmethoddesign:Principlesandprocedures.WalnutCreek,CA:LeftCoastPress.O’Cathain,A.O.,Murphy,E.,&Nicholl,J.(2010).Threetechniquesforintegratingdatainmixedmethodsstudies.BMJ,341,c4587.Song,M.,Sandelowski,M.,&Happ,M.B.(2010).Currentpracticesandemergingtrendsinconductingmixedmethodsinterventionstudiesinthehealthsciences.In:A.Tashakkori&
C.Teddlie(Eds.),Handbookofmixedmethodsinsocial&behavioralresearch(2nded.,pp.
725–747).ThousandOaks,CA:Sage.NationalInstitutesofHealth.Bestpracticesformixedmethodsresearchinthehealth
sciences,2010.