Beruflich Dokumente
Kultur Dokumente
OverviewofWorkshop
DefinitionsandterminologyofMMResearch PhilosophicalAssumptions MixedMethods:NutsandBolts Break Planningyourmixedmethodsstudy
Sampling DataCollection
Resources
Creswell&PlanoClark(2011)Designingand conductingmixedmethodsresearch.Thousand Oaks,CA:SagePublications,Inc. Teddlie&Tashakkori(2009)FoundationsofMixed MethodsResearch:IntegratingQuantitativeand QualitativeApproachesintheSocialand BehavioralSciences.LosAngeles:Sage Publications,Inc.
ConsensusDefinitionofMMResearch
Mixedmethodsresearchisthetypeofresearchin whicharesearcherorteamofresearchers combineselementsofqualitativeandquantitative approaches(e.g.,useofqualitativeand quantitativeviewpoints,datacollection,analysis, inferencetechniques)forthepurposeofbreadth anddepthofunderstandingandcorroboration Johnsonetal.(2007).
Keyterminology
Qualitative(QUAL)componentandQuantitative (QUAN)componentareoftenreferredtoas strands
PhilosophicalIssues
QuantitativeTradition
Philosophicalunderpinnings:positivism/post positivism Deductivelogic Dataarerepresentednumerically Associatedterms:surveyresearch,probability sampling,experimentalandquasiexperimental designs,descriptiveandinferentialstatistics
QualitativeTradition
Philosophicalunderpinnings:constructivism Inductivelogic Dataarerepresentedtextuallyorpictorially Associatedterms:groundedtheory,ethnography, casestudies,purposivesampling,categoricalvs. contextualizingstrategies,trustworthiness, credibility
MixedMethodsTradition
Philosophicalunderpinnings:pragmatism Bothdeductiveandinductive Dataarerepresentedbothnumericallyand textually/pictorially Associatedterms:concurrent(parallel)and sequentialmixeddesigns,triangulation,data conversion,inferencequality
Philosophicaldebatesaboutmixed methods
Incompatibilitythesis fundamentaldifferences betweenQUANandQUALapproachesaresogreat thatmethodscannotbemixed Pragmatism whatisthebestwaytoanswera researchquestion;bothmethodsofferdifferent waysofansweringresearchquestions
MixedMethods:NutsandBolts
CharacteristicsofMMStudies
Involvescollectionandanalysisofqualitativeand quantitativedatainwaysthatare:
rigorous framedepistemologically/theoretically
Themethodsaremixedby
orderingthemsequentially mergingthem embeddingonestrandwithintheother
ResearchQuestionsthatCallforMM
Exploringthemeaningofaconstructor phenomenonfrommorethanoneperspective Explanationofanomalousfindingsorgetting behindthemechanismofactionofaneffect Theorydevelopmentfollowedby testing/extension Measuredevelopmentusinggroundedconcepts Augmentingevaluationstudieswithbetter understandingofinterventionimplementation
WaysThatDesignsVary
Levelofinteractionbetweenstrands Relativepriorityofstrands Timingorpacingofeachstrand Pointofinterface(atwhichpointintheresearch processarethestrandsmixed?):during interpretation,dataanalysis,datacollection? Researchstance,epistemology
TypologyofMixedMethodsDesigns
Convergentparallel Explanatorysequential Exploratorysequential Embedded Caveat:evolvingfieldwithevolvinglanguage
AdaptedfromCreswell&PlanoClark(2011)
ConvergentDesigns
QUANandQUALstrandsareconductedseparately yetconcurrentlyandmergedatthepointof interpretation Equalprioritygiventoeachstrand Usedtoformamorecompleteunderstandingofa topic,ortovalidateorcorroboratequantitative scales
ConvergentParallelDesign
QUAL DataCollection QUAL DataAnalysis
Compare orRelate
QUAN DataCollection
QUAN DataAnalysis
Convergent Parallel Design Example: Conceptual Adequacy of the Drug Attitude Inventory for Youth
Design Decisions
Choice of instrument Questionnaire Rating scale Sampling Convenience Representative
Design Decisions
Demographics Youth DAI Parent DAI Adherence Ratings Clinical Scales (CDRS,YMRS,CBCL)
Choice of method Interview Ethnography Focus group Sampling Purposive Convenience Setting
Metainference
Mixed Methods Question: Can prediction of youth attitudes toward psychotropic treatment be improved by knowledge about the factor structure of the DAI in youth and their subjective experiences experiences of treatment?
Townsend,Floersch,&Findling,2010
ExplanatorySequentialDesign
Methodsareimplementedsequentially,(QUAN QUAL) Usedwhenresearcherwishestousequalitative findingstohelpinterpretorcontextualize quantitativeresults
QUANData Collection and Analysis
Followup with
ExplanatorySequentialExample:ACT SocialNetworkStudy
QUANData Collection and Analysis QUALand QUANData Collection andAnalysis
Angell&Test,2002;Angell,2003
ExploratorySequentialDesign
Methodsareimplementedsequentially,(QUAL QUAN) TheQUALstrandisconsideredexploratory,tobe followedbyfurthertestingandverificationduring theQUANphase
Buildsto
Interpretation
ExploratorySequentialDesignExample:Measuring ProceduralJustice(PJ)inPoliceEncounters
Reviewof existing QUALstrand: instruments Consumer andliterature interviews ledtoresearch Analysisof question:Do discrete existingPJ encounters instruments usinggrounded capture dimensional featuresof analysis contacts betweenpolice andcitizens Interpretation: PJexperiences withmental area)contextualizedby illness?
negativeexpectationsandb) sensitivetosmallgesturesof humanity
QUANstrand Instrument development Cognitive interviewing Expertreview Surveyof consumersusing finalinstrument (PCES) EFAandRasch Analysis
Interpretation: PCESpredicted reactionstopoliceencounter (resistance,cooperation)
Watson,Angell,Vidalon,&Davis(2010)
EmbeddedDesign
ResearcherconductingeitheraQUALorQUAN studyembedsasmallerstrandoftheother method,asanenhancement Secondarystrandcanbeconcurrentorsequential
EmbeddedDesignExample:CTI EvaluationandFidelityStudy
RCTofCriticalTimeIntervention(CTI)forMenLeaving Prison.QUANDataCollection,n=220 Fidelity/ProcessStudy QUALdatacollection,n=24 DataCollection Decisions: Interviews FocusGroups Fieldnotes Recordabstraction Samplingcriteria DataAnalysis Decisions: Coding(open, selective,axial) ofinterviewsand documents/ Narrative analysis?
VariationsontheMMDesigns
Multiphaseformat Multilevelformat MonostrandConversion(nottrulymixed methods):conversionofQUALdatatoQUANor QUANdatatoQUAL,withoutadditionalstrands Transformativestance
Questions?
Break
PlanningYourMixedMethodsStudy
Sampling:GeneralConsiderations
Strategychosenshouldbeappropriatetoeach respectivestrand Balancebetweensaturationofphenomenonor theory(qualitativegoal)andrepresentativeness (quantitativegoal)
SamplingStrategies (TeddlieandTashakkori,2010)
Parallelmixedmethodssampling(paralleluseof probabilityandpurposivestrategies,either concurrentlyorwithatimelapse).
Onesamplemaybeasubsetoftheother Bothstudiesmayusesametotalsample
DataCollection
(Teddlie&Tashakkori,2009)
SelfReportTechniques
Interviews Questionnaires AttitudeScales Personalityinventories Projectiveinstruments
ObservationalMethods
Participantobservation,nonparticipantobservation
Sociometry
Socialnetworkanalysis
SecondaryDataAnalysis
Archivalanalysis Metaanalysis 1998)
MultipleModesofDataCollection(Tashakkori&Teddlie,
DataAnalysis
QuantitativeDataAnalysis
Descriptive
summarizingdata,lookingfortrendsandpatterns; means,frequencies,measuresofvariability
Inferential
hypothesistesting,inferencesaboutapopulation characteristic;significancetests(2,t,F),multiple regression,ANOVA,MANOVA,MANCOVA, hierarchicallinearmodeling,timeseries,eventhistory
QualitativeDataAnalysis
Oftenongoingduringdatacollection(e.g.,purposive sampling,modificationofinterviewquestions,etc.)
Groundedtheory Thematicanalysis Negativecaseanalysis
FRACTURINGVS.CONTEXTUALIZING
Categoricalstrategies:producecategoriesthatfacilitatecomparisons;e.g., constantcomparativemethod Contextualizingstrategies:interpretnarrativedatainthecontextofthe wholetext,focusingoninterconnectionsbetweenstatements,events,etc.; e.g.,phenomenology
SIMILARITYVS.CONTRAST
MixedMethodsDataAnalysis
(CreswellandPlanoClark,2011)
QUAN+QUAL=convergeresultsCONVERGENTDESIGN QUAN qual=explainresultsSEQUENTIALEXPLANATORY DESIGN QUAL quan=generalizefindingsSEQUENTIALEXPLORATORY DESIGN QUAN(+qual)=enhanceexperimentEMBEDDEDDESIGN TRANSFORMATIVEDESIGN usesatransformativetheoretical perspectivetoadvocateforsocialchange,addresssocialinjustice, orgivevoicetomarginalized/underrepresentedgroup. MULTIPHASEDESIGN aprogramofresearchthatinvolvesseveral studies;canhavecombinationsofsequentialandconcurrent designs
MixedMethodsDataAnalysis
(Creswell&PlanoClark,2011)
Convergentparallel:mergeddataanalysisforpurposesof comparingresults
CollectandanalyzeQUALandQUANdata Strandsareanalyzedindependently(couldbequalitizing/quantitizing strategiesalso) Howwillthetwostrandsbecompared? Howwilltheyberepresented?
Explanatory:connecteddataanalysistoexplainfindings
Collectandanalyzequantitativedata;derivesecondresearchquestion Designandconductqualitativeresearch Analyzequalitativedataforanswerstosecondaryresearchquestion Linkresultsfrombothstrands howdoqualitativeresultsexplain quantitativefindings?
Does the factor structure of the DAI in adults fit the youth data? If not, what is the factor structure of the DAI in youth? How well do DAI items correlate with one another? Do they measure a single construct or multiple constructs?
Univariate descriptives Bivariate correlations Structural Equation Modeling Factor analysis Parallel Analysis (SPSS, LISREL)
Are there elements of youth medication experience that the DAI does not capture?
In vivo codes Intermediate codes Superordinate codes Peer review Constant comparative approach (Atlas TI)
Compare
Converge
Can prediction of youth attitudes toward psychotropic treatment be improved by knowledge about the factor structure of the DAI in youth and their subjective experiences of treatment?
MixedMethodsDataAnalysis
(Creswell&PlanoClark,2011)
Exploratory:connecteddataanalysistogeneralize findings
Collectandanalyzequalitativedata;usequalitativedatatodesign quantitativecomponent Collectandanalyzequantitativedata Linkresultsfrombothstrands:howdoquantitativeresultsextend qualitativefindings?
Embeddeddesign:merged(concurrentdesign)or connected(sequentialdesign)analysis
Collectandanalyzeprimarydataset;decidehowembeddeddatawillbe usedandwheretheyshouldbeincorporatedintotheprimaryanalysis Analyzesecondarydatasetdictatedbywhereitisembeddedinthelarger design Howdotheembeddedfindingsintegratewiththeprimarystudyfindings?
EvaluatingMixedMethodsStudies
Five elements:
Foundational element
Quality of literature review and theory base
Construct validation
Validity of QUAN, QUAL, and mixed elements
Inferential consistency
Consistency of links between various strands of the study (see table on following slide)
Utilization/historical element
Whether and how the studys findings went on to be used in future work
Consequential element
Social acceptability and consequences of study findings
Construct Validation
(Dellinger & Leech, 2007; Leech, Dellinger, Brannagan, & Tanaka, 2010)
RepresentingMixedMethodsData
(Creswell&PlanoClark,2011)
Introduction explicit integration of both paradigms from the outset Literature review integration of inductive/deductive reasoning, why the literature needs this type of study Posing the research question what are the questions and why do they call for two paradigms? Methods present both methodologies, in their respective languages, integrated under the umbrella of the research question Results present results of both modes of data collection Discussion role of meta-inference
ExampleI
FindingsfromDAIStudy (Townsend,Floersch,&Findling,2010)
StudyFlowchart
QUAN Data Collection
Demographics Youth DAI Parent DAI Decision-Making Scales Adherence Ratings Clinical Scales (CDRS,YMRS,CBCL)
Univariate descriptives Bivariate correlations Structural Equation Modeling Factor analysis Parallel Analysis (SPSS, LISREL)
In vivo codes Intermediate codes Superordinate codes Peer review Constant comparative approach (Atlas TI)
Compare
Converge
Can prediction of youth attitudes toward psychotropic treatment be improved by knowledge about the factor structure of the DAI in youth and their subjective experiences of treatment?
ParallelDataReductionStrategies
@stigma {0-56}
&Desire_for_Normality {0-0}
&Crazy_Identity {0-0}
&Educating_Others {0-0}
StructuralEquationModelOne
DAIOriginalFactorStructure
RMSEA .061
[ideal = <.05 (Kaplan, 2000)]
X2
df 258 420.38
ExploratoryFactorAnalysis
FactorSelectionCriteria:
Maximumlikelihoodestimation Eigenvalue> 1.0 Minimumitemloading> .30 Retained4+items >4itemsbutdifferentiatedwellfromotherfactors Qualitativedataindicateretentionofitems/factorsisjustified
EFAOne
Twofactorswerenotinterpretedfurtherbecausetheywereeachcomprisedofonlyoneitem (items8and13)andhadvalues>1.0.(Heywoodcases)
EFATwo28items(SupplementedbyParallelAnalysis)
Twofactorsretained,accountingfor36.61%ofthevarianceinDAIscore. Factorlabels:
PositiveFeelingstowardMedication NegativeFeelingstowardMedication
QualitativeAnalysisMethods
Constantcomparativeapproach analysisofinvivocodes followedbyintermediatelevelcodingandsynthesisinto higherordersuperordinateconcepts.
@stigma {0-56}
&Desire_for_Normality {0-0}
&Crazy_Identity {0-0}
&Educating_Others {0-0}
SummaryofQualitativeThemes
DAIrelatedconcepts
Positivefeelings Negativefeelings Health/IllnessModel InternalLOC ExternalLOC RelapsePrevention
Conceptsnot representedintheDAI
Balancedresponsibility Ambivalence ChangeoverTime Adherenceifeffective Expectations Inclusionintreatment decisions Autonomy Stigma
SidebySideComparisonExample: ConclusionsSection
QuantitativeFindings
Insummarizingthefindingsoftheadolescent qualitativeandquantitativeanalyses,itis apparentthatthesevendimensionsfoundin theoriginalDAIemployedwithadultsare presentinteens conceptualizationsoftheir viewsofmedication.However,adolescents maythinkaboutthosedimensions,suchas sideeffectsandrelapseprevention, differentlythanadultsdo.Thesefindingsare supportedbyboththeEFAandqualitative results.EFAfindingsindicatethatsideeffects differentiatedintospecificcognitiveeffectsas wellasotherharmsforyouth.Adolescents didnotappeartoviewrelapsepreventionas theadultsdid,insteadevaluatingtheir perceivedneedformedicationinrelationto theirsymptomstabilityinformingtheir attitudestowardmedication.
QualitativeFindings
Thequalitativedatahighlightedthepresence ofeightadditionalthemesnotreflectedinthe originalfactorstructureoftheDAI,including importantthemessuchaspersonalautonomy, inclusioninmedicationdecisionmaking,the experienceofstigma,andlevelofpersonal responsibilityvs.externalcontrolover behavior.Thesethemeshighlightthat medicationexperiencesandtheformationof attitudestowardpharmacologicaltreatment foryoutharecomplexandmultiply determined,renderingitdifficulttorepresent adolescentattitudestowardmedicationfully withasingleinstrument.Thesefindingspoint tothepotentialvalueofcreating individualized,qualitativeassessmenttoolsto captureyouthexperienceswithmedication ratherthanrelyingonasinglequantitative measureorsetofsubscales.
Discussionsectionhighlights divergentresults
JointDataDisplayExample
HighlightsConvergenceofFactorAnalyticandThematicAnalysisResults
PositiveFeelings 2 goodoutweighsbad 21 thoughtsareclearer 26 happieronmeds 29 inbettercontrol QualitativeThemes Emotional Cognitive Physical NegativeFeelings 5R takeb/cofpressurefrom others 14R medicationisslowacting poison 19R Idratherbesickthan takingmedications QualitativeThemes Emotional Cognitive Physical FactorLoading .474 .789 .487 .598 Quotations Icaneithernottakeitandbeagrouch,ortakeitandbehappy.SoIwouldmuchrathertakeitandbehappy. Sittingdown,payingattentiontotheteacher,nottalking.JustpayingattentionanddoingwhatIneedtodo. Igainenergytowanttodomyworkandlisteningandfocusandtrytodomybestandgive110%. FactorLoadings .315 .714 .658 Quotations IdontlikesometimesIfeellikeoutofmybodyorjustlikenotmyself,moreanxioussometimes. WellIhadittoldmetheTopamaxorwhatevercausesshorttermmemoryloss,soIassumeitstheTopamaxorwhatever. Makemesick.Makemegetthebubblegutsandstuff.AndIwishthattherewasamedicinethatdidnthavesideeffects.
MergedDataDisplayExample
Countsofquotationsrepresentedgraphicallytodemonstrateconvergence withfactoranalyses positive andnegative factorssupportedbyallformsofQUANand QUALevaluation
Distribution of DAI-Related Quotations across Respondents
40 35 Number of Quotations 30 25 20 15 10 5 0 4 7 10 12 13 14 16 17 18 20 22 23 24 25 27 29 30 32 38 41 Respondent Positive Negative Illness Model ILOC ELOC Harm Relapse Prev.
Number of adolescent quotations linked to original DAI factors Positive Subjective Feeling 103 Negative Subjective Feeling 31
Health/Illness Model
Harm/ Toxicity 48
Relapse Prevention 34
21
ExampleII
ForensicAssertiveCommunityTreatment(FACT) Evaluation(Angell&Watson,inprogress) Keymixedmethodquestions:
HowisACTmodifiedinthecontextofprisonreentry? Howdoestheagencysrecoveryorientedmissionshape thetranslationofACTtoFACT? Howdoesengagementofconsumersoccurandtowhat extentisleverageinvolved? Whataretheuniquefeaturesofclientprovider relationshipsinFACT? Howdoesengagementrelatetoratesofsuccess (avoidanceofreincarceration)?
OverallDesign:ConvergentParallel
Sample=21adultswithmentalillnessleavingprisonandentering FACTprogram
PanelDesign
Client reported outcomes assessedat BL,1,3,6, 9,18mos staff reported outcomes monthly
Interpretation/ MetaInference
AnalysisStrategies
Sample=21adultswithmentalillnessleavingprisonandentering FACTprogram
QUAL Descriptive PrePost Comparisons Outcome differencesby risklevel
Interpretation/ MetaInference
QandA/TechnicalAssistance