Validity

Concept

Validityisthedegreeofvalidityofameasurement,thatis,thedegreetowhichameasurementtoolcanactuallymeasurethecharacteristictobemeasured,orsimplyreferstotheaccuracyandusefulnessofatestsex.Validityisthemostimportantconditionthatascientificmeasurementtoolmusthave.Insocialmeasurement,thevalidityofquestionnairesorscalesusedasmeasurementtoolsisrelativelyhigh.Discriminationvaliditymustclarifythepurposeandscopeofthemeasurement,considerthecontenttobemeasuredandanalyzeitsnatureandcharacteristics,checkwhetherthecontentofthemeasurementisconsistentwiththepurposeofthemeasurement,andthendeterminewhetherthemeasurementresultreflectsthedegreeofthecharacteristictobemeasured;

Nature

1.Validityisrelative

Thevalidityofanytestisforacertaingoal,orTestswillonlybeeffectiveiftheyareusedforpurposesandoccasionsthatareconsistentwiththeobjectivesofthetest.Therefore,whenevaluatingthevalidityofthetest,thepurposeandfunctionofthevaliditytestmustbeconsidered.

2.Validityhascontinuity

Thevalidityofatestisusuallyexpressedbyacorrelationcoefficient,whichonlydiffersindegree,ratherthan"all-in"Orthedifferenceof"nothing".Validityisbasedontestresults.

Function

1.Predictionerror

Theactualmeaningofthevaliditycoefficientisoftenexpressedbythedeterminingcoefficient,whichmeansthesquareofthecorrelationcoefficient,whichindicatesthatthetestpredictscorrectlyOrtheratioofthevarianceoftheexplainedeffectstandardtothetotalvariance.

2.Predictivescores

(Predictthescoresfromthepredictedscores)IfthetwovariablesXandYarelinearlyrelated,aslongastheregressionequationbetweenthetwoisdetermined,thenOnevariablecanbeestimatedfromanothervariable.

3.Predictionefficiencyindex

(Theerrorreducedcomparedtoblindprediction)

Type

Internalvalidity

Theconceptofinternalvalidity

Theinternalvalidityofastudyreferstothedegreeofclaritywithwhichacertainrelationshipexistsbetweentheindependentvariableandthedependentvariableofthestudy.Iftherelationshipbetweentheindependentvariableandthedependentvariableisnotaffectedbytheexistenceofothervariables,andthusbecomesambiguousorcomplicated,thenthisstudyhasinternalvalidity.Thequestionsitinvolvesare:

(1)Isthereacertainrelationshipbetweenthetwoormorevariablesstudied?

(2)Isittruethatthechangeoftheindependentvariablecausedthechangeofthedependentvariable?

Researchdesignshouldeffectivelycontrolandeliminatethevariousvariablesthatmaybeinvolved,sothatvariablesthatarenotrelatedtotheresearchobjectiveshavelittleornoimpactontheresearchresults.Therefore,betweentheresearchvariables(Forexample,therelationshipbetweentheindependentvariableandthedependentvariableisdefiniteandtrue,whichmeansthattheinternalvalidityofastudyishigh.

Factorsaffectinginternalvalidity

Therearemanyfactorsaffectinginternalvalidity.Insummary,therearefourmainfactors:history,researchsubjects,researchmethodsandprocedures,andstatisticalregressioneffects.Aspectfactors.

Conditionsofinternalvalidity

Theinternalvalidityofthestudywillnotbeformedautomatically.Theinternalvalidityisobtainedmainlythroughresearchdesign,carefulselectionofvariables,effectivecontrolofvariousvariables,ensuringthedefiniterelationshipbetweenresearchvariables,andeliminatingtheinfluenceofvariablesunrelatedtotheresearchobjectivesontheresearchresults.Thefourfactorsmentionedaboveshouldbecontrolledduringresearch.Themosteffectivewayistouserandomizationprocedures.Forvariousspecificsituationsofeducationalscientificresearch(suchasdifferentresearchactivities),therewillbecertaindifferencesinthetypes,numbers,andeffectsoffactorsthataffectinternalvalidity.Theyshouldbeanalyzed,estimated,identified,andadoptedaccordingtothespecificconditions.Measuresshouldbecontrolledoreliminatedtoimprovetheinternalvalidityoftheresearch.

Externalvalidity

Externalvalidity

TheexternalvalidityoftheresearchreferstothegeneralizationanduniversalapplicationoftheresearchresultstothepopulationfromwhichthesamplecomesandtootherThedegreeintheoverallpopulation,thatis,therepresentativenessanduniversalapplicabilityoftheresearchresultsandvariableconditions,timeandbackground.Externalvaliditycanbesubdividedintotwocategories:overallvalidityandecologicalvalidity.

Overallvalidityreferstothedegreeandabilityoftheresearchresultstoadapttothepopulationfromwhichtheresearchsamplecomes,orthegeneralsignificancetothepopulation.Tomaketheresearchresultsapplicabletothepopulation,asamplemustbeselectedrandomlyfromthepopulationsothatthesampleisrepresentativeofthepopulation.Iftheselectedsamplesarebiasedortoosmalltorepresentthepopulation,theresultswillbedifficulttosummarizethecharacteristicsofthepopulation.

Ecologicalvalidityreferstothedegreeandabilityofresearchresultsthatcanbegeneralizedandadaptedtootherresearchconditionsandscenarios.Tomaketheresearchresultsapplicabletootherresearchconditionsandscenarios(forexample,thedifferencesbetweenindependentanddependentvariables,researchprocedures,researchbackground,researchtime,andresearchers,etc.),researchconditionsandscenariosmustbespeciallydesignedtoensurethatotherTheconditionsandscenariosarerepresentative.

Itisgenerallybelievedthatinternalvalidityisanecessaryconditionforexternalvalidity,butitisnotasufficientcondition.Researchresultswithlowinternalvaliditycannotbesaidtobeuniversallymeaningfultootherscenarios;however,researchresultswithhighinternalvaliditymaynotnecessarilybegeneralizedtootheroverallsandcontexts.Theimportantsignificanceofeducationalscientificresearchistodiscovertheuniversallawsofeducationalactivitiesandguidethedevelopmentofeducationalwork.Therefore,itisveryimportanttoimprovetheexternalvalidityofresearchresults.Nomatterhowhightheinternalvalidityofastudyis,ifitsresultsareonlyadaptedtoaspecificscope,specificmeasurementtools,specificresearchprocedures,andspecificresearchconditions,then,fromtheperspectiveofobtaininggeneralknowledgeandrevealinguniversallaws,Itsvalueandsignificancearenotgreat.Therefore,theexternalvalidityandinternalvalidityofthestudyarenotinferiorinimportance.

Factorsinfluencingexternalvalidity

Factorsinfluencingexternalvaliditymainlyincludefouraspects:researchsubjects,variabledefinitionandtestingaspects,researchmethodsandprocedures,Fortheexperimenter.

Conditionsofexternalvalidity

Theabovefourfactorssometimesexistalone,andsometimestheyhaveseveralinfluencesatthesametime.Toimprovetheexternalvalidityoftheresearch,wemustpayattentiontoeliminatingandcontrollingtheabove-mentionedvariousinfluencingfactorsintheresearch.Therequirementofexternalvalidityisthattheresearchcanmeettheobjectiveconditionsandbeapplicabletoalargerpopulation.Amongthem,thekeypartistodoagoodjobofsampling.Samplingworkincludesnotonlythesamplingofsubjects,butalsotheselectionofrepresentativeresearchbackground(workplace,school,home,laboratory),researchtools,researchprocedures,andtime.Thecloserthesamplingbackgroundistotheactualscenario,thestrongertheavailability,applicability,andgeneralizationoftheresearchresults.Generallyspeaking,randomsampling,increasingthedegreeofsimulatingrealisticscenarios,usingavarietyofrelatedresearchmethods,andchangingresearchconditionstoseekuniversallymeaningfulconclusionsareimportantconditionsforobtainingexternalvalidityandimprovingtheapplicabilityofresearchresults.

Conceptionvalidity

Introductiontoconceptionvalidity

Theconceptionvalidityofresearchreferstotherationalityandscientificityoftheoreticalconceptionorhypothesis,anditstransformationintoTheappropriatenessandoperabilityoftheresearchobjectives.Itinvolvestheestablishmentofaresearchplanandatheoreticalconceptionofmeasurementindicators(oratheoreticalconceptionofobservationindicators)anditsoperationalization,thatis,thevalidityofthetheoreticalconceptionanditsconversion.Inorderfortheresearchtohaveahighconceptionvalidity,thetheoreticalconceptionoftheresearchmustfirstberigorouslystructuredandclearlystructuredtoformacertain"conceptionnetwork",andsecondly,astrictabstractandoperationaldefinitionoftheresearchcontent(suchastheresearchconceptionFeatures,giveclearoperationaldefinitions).

Factorsaffectingconceptvalidity

Therearethreemainfactorsaffectingconceptvalidity:researchconception,researchmethodsandprocedures,subjectiveexpectationsofexperimenters,andparticipantsSpeculationonresearch.

ConditionsforConceptualValidity

Forresearchtohavehigherconceptionvalidity,thefollowingfourconditionsshouldbemet:

(1)TheoreticalconceptionrequiresstructureRigorous,logical,andhierarchical,formingakindof"conceptionnetwork".Forexample,thedevelopmentofchildren'sself-awarenessisdividedintothreeaspects:self-knowledge,self-experienceandself-control,andself-knowledgeisfurthersubdividedintotheknowledgeofthephysicalself,theknowledgeofthepsychologicalselfandtheknowledgeofthesocialself.Suchatheoreticalconceptionismorerigorous,complete,andlayered,andformsa"network"foreasyunderstandingandresearch.

(2)Clearlyandaccuratelydefinetheenvironmentalconditionsandvariablesoftheresearch.Forexample,thesubjectof"StudyonDivergentThinkingofStudentsintheFifthGradeofPrimarySchool"requiresacleardefinitionoftheagegroup,physicalandintellectualdevelopment,learning,life,andsocialenvironmentofthesubjects,usingbothwrittenandlanguageformstobecompleteandaccurate.Statetheresearchvariables.

(3)Makeaccurateandstrictoperationaldefinitionsfortheresearchvariables,andselectcorrespondingandobjectiveobservationindicators.

(4)Avoidusingasinglemethodorasingleindicatortorepresentoranalyzemulti-dimensional,multi-level,andmulti-facetedthingsandactivities,andusemultiplemethodsandmultipleindicatorsasmuchaspossibletoanalyzefromdifferentperspectivesStudythesametheoreticalideas.

Educationalscienceresearchoftencontainscomplexandmulti-dimensionaltheoreticalideas.Howtoimprovethevalidityofresearchideasisaproblemthatneedstobepaidspecialattentiontowhenconductingresearchdesign,anditisalsoanimportantcontenttoimprovethelevelofeducationalscienceresearch..

Validityofstatisticalconclusions

Thevalidityofstatisticalconclusionsisaboutthevaliditytestofthedataanalysisandprocessingproceduresoftheresearch,orinotherwords,itisthedataanalysisproceduresandmethodsoftestingtheresultsoftheresearch.Anindicatorofeffectiveness.Thebasicquestionofthevalidityofstatisticalconclusionsistostudytheerror,thesourceofvariationandhowtoproperlyusethestatisticalsignificancetest.Itdoesnotinvolvethesourceofsystematicdeviation,buttheproblemofstudyingtheerrorvariationandhowtoproperlyusethestatisticalsignificancetest.Forexample,whenusingasmallsampleofresearchdata,becausethesamplecompositionandmeasurementdatahaverelativelylargefluctuationsandpoorstability,itisunreliabletorelyonstatisticalsignificancetomakeinferences.Inthiscase,poweranalysisshouldbeusedtoseehowmucheffectcanbetestedforacertainsamplerange,degreeofvariation,andalphalevel.Thisisthequestionconsideredforthevalidityofstatisticalconclusions.

Factorsaffectingstatisticalvalidity

Themainfactorsaffectingstatisticalvalidityare:lowstatisticalpower,violationoftheconditionsofuseofstatisticalmethods,andlowmeasurementreliability.

Conditionsforthevalidityofstatisticalconclusions

Thevalidityofthestatisticalconclusionsofthestudymainlydependsontwoconditions:oneisthequalityofthedata,andthevalidityofthedataanalysisprocedureisbasedonthedataOnthebasisofthequalityofdata,thevalidityofstatisticalconclusionscannotbediscussedintheresearchofpoordataquality;thesecondisthestatisticaltestmethod.Variousstatisticalmethodsusedindataanalysishavetheirownclearrequirementsforstatisticaltestconditions.Inthisstudy,thestatisticaltestconditionsarenotclearorviolated,whichwillsignificantlyreducethevalidityofthestatisticalconclusions.

Evaluationmethod

Contentvalidity

(content-relatedvalidity)

Whatiscontentvalidity

Contentvalidityreferstotheapplicabilityoftestquestionstorelevantcontentorbehavioralsampling,soastodeterminewhetherthetestisarepresentativesamplingofthedesiredbehavioraldomain.

Twoevaluationmethodsofcontentvalidity

1.Expertjudgmentmethod;2.Statisticalanalysismethod(raterreliability\duplicatereliability\halfreliability\retestmethod);3.Empiricalinferencemethod(experimentaltest)

Characteristicsofthreecontentvalidity

Contentvalidityisoftenconfusedwithfacevalidity.Thefacevalidityisdeterminedbythelayman'ssuperficialinspectionofthetest.Itdoesnotreflectwhatthetestactuallymeasures,butonlyreferstowhatthetestlookslikeonthesurfacetobemeasured;contentvalidityisdeterminedbyaqualifiedjudge(Experts)Establishedtoevaluatethetestindetailandsystematically.

Construct-relatedvalidity

(construct-relatedvalidity)

Onewhatisconstruct-relatedvalidity

ItmeansthatthetestcanmeasuretheoreticallyThedegreeoftheconceptionorcharacteristicofthetest,thatis,whethertheresultofthetestcanconfirmorexplainthehypothesis,termorconceptionofacertaintheory,andthedegreeofexplanation.

Methodsfortheestimationofthevalidityofthesecondconception

1.Analysisofthetestitself(usecontentvaliditytoverifythevalidityoftheconception)

2.BetweenthetestsMutualcomparison:Compatibilityvalidity(comparisonwiththesamematuretest),discriminationvalidity(comparisonwithsimilarordistinguishabletests),factoranalysismethod

3.EfficacystandardeffectProofofthedegreeofresearch

4.Confirmationbyexperimentalandobservationmethods

Validity-relatedvalidity

(criterion-relatedvalidity)

Efficacycriterion,thatis,thereferencestandardformeasuringthevalidityofthetest,referstothebehaviorofourinterestthatcanbedirectlyandindependentlymeasured.Thebehaviorweareinterestedinisthebehaviortobepredicted.Thisisageneralconcept,soitmustbedeterminedbyoperablemeasurementstohavepracticalsignificance.Therefore,itisnecessarytosubdividetheperformancestandardintotwolevels,oneisthe"conceptualperformancestandard"atthetheoreticallevel,andtheotheristhe"effectstandardmeasurement"attheoperationaldefinitionlevel.

Validitycriterionvalidityisalsocalledempiricalvalidity,whichreflectstheeffectivenessofthetestinpredictingthebehaviorofanindividualinacertainsituation.

Accordingtowhetherthebenchmarkdataisobtainedatthesametimeasthetestscores,itcanbedividedintosimultaneousvalidity(theconsistencyofthescoresofpeoplewithhighmoraleandlowmoraleinthemoraletest.)andpredictivevalidity..

Commonlyusedbenchmarksinclude

1.Academicachievement:suchasschoolperformance,academicqualifications,relatedawardsandhonors,teachers’assessmentofstudents’intelligence,etc.,whichareoftenusedasintelligencetestsEfficacycriteriacanalsobeusedascriteriaforcertainmultipleaptitudetestsandpersonalitytests;

2.Actualworkperformance:Itisthemostsatisfactorybenchmarkmeasurement,forgeneralintelligencetests,personalitytestsandsomeabilitiesEfficacycriteriaofthetendencytest;

3,specialtrainingperformance;

4,mentalillnessdiagnosis;

5,gradeevaluation:theobserverwantstomeasureaccordingtothetestApersonaljudgmentbasedontheperformanceofthesubjects’psychologicaltraits;

6.Comparisonofbenchmarkinggroups:thatis,tofindouttwogroupsthatdifferinbenchmarkingperformance,andcomparetheirThedifferenceintestscores;

7.Previouslyvalidtests:Thecorrelationbetweenanewtestandapreviouslyvalidtestisoftenusedasevidenceofvalidity.

Agoodperformancestandardmustmeetthefollowingconditions:

①Theperformancestandardmustmosteffectivelyreflectthegoalofthetest,thatis,theperformancestandardmeasurementitselfmustEffective;

②Efficacycriteriamusthavehighreliability,stabilityandreliability,andnotchangewithtimeandotherfactors;

③Efficacycriteriacanbemeasuredobjectively,withdataorgradeavailableToshow;

④Themethodofeffectstandardmeasurementissimple,time-savingandlabor-saving,economicalandpractical.

Evaluationmethodofvaliditycriterionvalidity

1.Correlationmethod:Validitycoefficientisthemostcommonlyusedvalidityindex,Especiallythestandardvalidity.ItisexpressedintermsofthePearsonproductdifferencecorrelationcoefficient,whichmainlyreflectsthecorrelationbetweentestscoresandefficacystandardmeasurements.Whenthetestscoreisacontinuousvariableandthecriteriondataisadichotomousvariable,thetwo-rowcorrelationformulaorthetwo-rowcorrelationformulacanbeusedtocalculatethevaliditycoefficient;whenthetestscoreisacontinuousvariableandthecriteriondataisagradeevaluation,useJassCalculationofPendorseriesrelatedformulas.

2.Differentiationmethod:Itisamethodtotestwhetherthetestscorecaneffectivelydistinguishthegroupdefinedbytheefficacystandard.Aftercalculatingthetvalue,youcanknowwhetherthedifferenceinscoresissignificant.Ifthedifferenceissignificant,itmeansthatthetestcaneffectivelydistinguishthegroupsdefinedbythevalidcriteria;otherwise,thetestisinvalid.Theoverlappercentagecanbecalculatedbycalculatingthepercentageofpeopleineachgroupwhosescoreexceeds(orbelow)theaverageoftheothergroup;inaddition,thepercentageofthecommonareaof​​thetwogroupscanalsobecalculated.Thegreatertheamountofoverlap,thesmallerthedifferencebetweenthetwogroupsofscores,thatis,theworsethevalidityofthetest.

3.Hitratemethod:Whenthetestisusedasthebasisforselection,theratioofthecorrectdecisionisusedasamethodofvalidity.Therearetwowaystocalculatethehitrate,oneistocalculatethetotalhitrate,andtheotheristocalculatethepositivehitrate.

4.Expectationtablemethod:Itisatwo-waytable,withpredictedscoresarrangedontheleftsideofthetable,andtheeffectstandardonthetopofthetable.Thelargerthepercentagenumbersonthediagonalfromthelowerlefttotheupperright,andthesmallertheotherpercentagenumbers,thehigherthevalidityofthetest;onthecontrary,themorescatteredthenumbers,thelowerthevalidity.

Identificationprocedure

Fromcontentvalidity,tocriterionvalidity,andthentoconstructvalidity,itcanberegardedasakindofprogression,thatis,constructvalidityneedstobemorethancriterionvalidityMoreinformation,criterionvalidityrequiresmoreinformationthancontentvalidity.

Validityhashigherrequirementsthanreliability.Reliabilityisanecessaryconditionforvalidity.Withoutreliabilitymeasurementtools,thereisnovalidity,buthighreliabilitymeasurementtoolsmaynothaveHighvalidity.

Influencingfactors

I.Factorsofthetestitself

Therepresentativenessoftestmaterials,testlength,testquestiontypes,difficulty,distinction,andarrangementmethodswillallbeAffectvalidity.

2.Interferencefactorsintheimplementationofthetest

1.Influencingfactorsofthemaintest

2.Influencingfactorsofthesubjects

3.Thenatureofthesamplegroup

Theheterogeneityofthesamegroup

Twointerferencevariables:Thenatureofthesamplegroupincludesage,gender,educationlevel,intelligence,motivation,interest,occupationandAnyrelatedcharacteristics,duetotheinfluenceofthesecharacteristics,makethetesthavedifferenttestabilitiesfordifferent.

Giselliproposedasetofmethodsonhowtofindinterferencevariables:

①Useregressionequationstoobtaineachperson’spredictedperformancescore,andcomparethescorewiththeactualperformancescoreIwanttocomparethescoresandgetthedifferencescoreD.IftheabsolutevalueofDislarge,itindicatesthattheremaybeinterferencevariablesinthetest.

②Accordingtotheanalysisofthecompositionofthesamplegroup,findthecontrolgroup,andcalculatethevalidityseparately,soastofindouttheinterferencevariablesliketheaboveexampleofthetaxidriver.

③Forthegrouptobetested,accordingtosomeeasilyvisibleinterferencevariables,itisdividedintotwosubgroupswithhighpredictabilityandlowpredictability.Forgroupswithhighpredictability,thevalidityofthetestobtainedwillbeimproved.

Four,thenatureoftheefficacystandard

Agoodefficacystandardmustmeetthefollowingconditions:

①Theefficacystandardmustmosteffectivelyreflectthetestobjectives,namelyEfficacystandardmeasurementitselfmustbeeffective;

②Efficacystandardmusthavehighreliability,stableandreliable,andnotchangewithtimeandotherfactors;

③Efficacystandardcanbemeasuredobjectively,Canbeexpressedbydataorgrade;

④Themethodofeffectstandardmeasurementissimple,time-savingandlabor-saving,economicalandpractical.

Testdimensions

1)Facevalidity.Referstothevolumestandardthatthetestshouldmeet,thatis,whetherasetoftestquestionsisappropriateonthesurface.Forexample,ifatestofreadingcomprehensionincludesdialectvocabularythatmanysubjectshavenotlearned,itcanbeconsideredthatthetestlacksfacevalidity.Facevalidityisaguaranteefactorfortestingthenormallevelofsubjects.

2)Contentvalidity.Referstowhetherasetoftestquestionshastestedthecontentthatshouldbetestedorwhetherthetestedcontentreflectstherequirementsofthetest,thatis,thedegreeofrepresentativenessandcoverageofthetest.Forexample,ifacertainsetofpronunciationskillstestquestionsonlyexaminecertainskillsthatarenecessaryforpronunciation,suchasthepronunciationofasinglephoneme,andnottheaccent,intonation,orthepronunciationofthephonemeinaword,thenthecontentofthetestiseffective.Thedegreeisverylow.

3)Constructvalidity.Referstothedegreetowhichtheitemsofasetoftestquestionsreflectthebasicaspectsofthetheoryonwhichthetestisbased.Forexample,basedonstructurallanguagetheory,itisbelievedthatsystematiclinguistichabitsareacquiredthroughsentencepatterns.Then,testitemsthatemphasizevocabularyandgrammaticalenvironmentlosetheircompilationvalidity.

4)Empiricalvalidity.Empiricalvalidityisameasureoftheeffectivenessofatest,whichisobtainedbycomparingatestwithoneormorestandardscales.Therearetwotypesofempiricalvalidity:oneisconcurrentvalidity,thatis,theresultofatestiscomparedwiththeresultofanothervalidtestatasimilartime,orthecoefficientobtainedbycomparingwiththeteacher'sappraisal;Thesecondispredictivevalidity,whichisacoefficientobtainedbycomparingtheresultofatestwiththelaterlanguageability,orwiththeteacher'slaterappraisalofstudents.

Generallyspeaking,whentestingthevalidityofatest,inadditiontoexaminingthecontentofthetestpaperaccordingtotherequirementsofthesyllabusandthetheoryofconceptvalidity,itisalsonecessarytousethemethodofcalculatingthecorrelationcoefficient.Thequantitativemethodistocalculatethecorrelationcoefficientbetweenthistestpaperandanothertestpaperthathasbeendeterminedtoaccuratelyreflectthesubject'slevel.Thehigherthecoefficient,thegreatertheeffectiveness.Thevalidityoftheclassroomtestshouldbebetween0.4-0.7,andthevalidityofthelargertestshouldbeabove0.7.

Validityanalysis

Therearemanymethodsforvalidityanalysis,andthemeasurementresultsreflectdifferentaspectsofvalidity.Themethodscommonlyusedtoanalyzethevalidityofquestionnairesmainlyincludethefollowing.

Singleitemandtotalsumcorrelationvalidityanalysis

FaceValidity.Alsocalledcontentvalidityorlogicalvalidity,itreferstothesuitabilitybetweenthecontentofthemeasurementandthemeasurementtarget.Itcanalsobesaidtorefertowhethertheselecteditemofthemeasurement"looks"inlinewiththepurposeandrequirementsofthemeasurement.Mainlybasedonthesubjectivejudgmentofthesurveydesigner.

Thismethodisusedtomeasurethecontentvalidityofthescale.Contentvalidityisalsocalledsurfacevalidityorlogicalvalidity,whichreferstowhetherthedesigneditemcanrepresentthecontentortopictobemeasured.Thecontentvalidityisoftenevaluatedbyacombinationoflogicalanalysisandstatisticalanalysis.Logicanalysisgenerallyinvolvestheresearcherorexpertjudgingwhethertheselecteditem"looks"inlinewiththepurposeandrequirementsofthemeasurement.Statisticalanalysismainlyadoptsthecorrelationanalysismethodofsingleitemandsumtoobtaintheevaluationresult,thatis,calculatethecorrelationcoefficientbetweenthescoreofeachitemandthetotalscoreoftheitem,andjudgewhetheritiseffectiveaccordingtowhetherthecorrelationissignificant.Ifthereareanti-intentionalitemsinthescale,theyshouldbeprocessedinreversebeforecalculatingthetotalscore.

CriterionValidityanalysis

CriterionValidity.Alsoknownascriterionvalidityorpredictivevalidity.Criteriavalidityreferstowhetherthedataobtainedbythescaleismeaningfulcomparedwiththevalues​​ofotherselectedvariables(criteriavariables).Accordingtothetimespan,thecriterionvaliditycanbedividedintosimultaneousvalidityandpredictivevalidity.Criterionvalidityanalysisisbasedonacertaintheorythathasbeendetermined,selectinganindicatorormeasurementtoolasthecriterion(efficacycriterion),andanalyzingtherelationshipbetweenthequestionnaireitemsandthecriterion.Differentvalues​​andcharacteristicsshowsignificantdifferences,whicharevaliditems.Themethodofevaluatingcriterionvalidityiscorrelationanalysisordifferencesignificancetest.Inthevalidityanalysisofquestionnaires,itisoftenverydifficulttochooseanappropriatecriterion,whichlimitstheapplicationofthismethodtoacertainextent.

ConstructValidityAnalysis

ConstructValidity.Itreferstothedegreeofcorrespondencebetweenacertainstructurereflectedinthemeasurementresultandthemeasuredvalue.Themethodusedinstructuralvalidityanalysisisfactoranalysis.Themostconcernedquestionis:Whatcharacteristicsdoesthescaleactuallymeasure?Whenevaluatingconstructvalidity,investigatorsshouldtrytoexplainthetheoreticalquestionof"whythescaleiseffective"andconsiderwhatinferencescanbedrawnfromthistheoreticalquestion.Structuralvalidityincludeshomogeneityvalidity,heterogeneityvalidityandsemanticlogicvalidity.Somescholarsbelievethatthemostidealmethodofvalidityanalysisistousefactoranalysistomeasurethestructuralvalidityofthescaleortheentirequestionnaire.Themainfunctionoffactoranalysisistoextractsomecommonfactorsfromallvariables(items)ofthescale.Eachcommonfactorishighlycorrelatedwithacertaingroupofspecificvariables.Thesecommonfactorsrepresentthebasicstructureofthescale.Throughfactoranalysis,wecanexaminewhetherthequestionnairecanmeasureacertainstructureassumedbytheresearcherwhendesigningthequestionnaire.Intheresultsoffactoranalysis,themainindicatorsusedtoevaluatethevalidityofthestructurearecumulativecontributionrate,degreeofcommonalityandfactorloading.Thecumulativecontributionratereflectsthecumulativeeffectivenessofthecommonfactoronthescaleorquestionnaire,thecommonalityreflectstheeffectivenessofthecommonfactortoexplaintheoriginalvariable,andthefactorloadreflectsthedegreeofcorrelationbetweentheoriginalvariableandacommonfactor.