Validation of questionnaires, with emphasis on the SDQ

Validationofquestionnaires,withemphasisontheStrengthsandDifficulties
Questionnaire(SDQ)
Introduction
Wewillfocusonvalidationofquestionnairesinthisteaching.Thisisbecauseseveralof
ushavetranslatedquestionnairesintoanotherlanguageorareplanningtodoso,and
weknowthatthereisaneedtoinvestigatewhetherthetranslatedversionofthe
instrumentworkswell.Thequestionis:Howdowedothisvalidationprocess?Wewill
usetheStrengthsandDifficultiesQuestionnaire,theSDQ,asanexample.Severalofus
haveusedthisquestionnairetoassessmentalproblemsinchildren.Ithasbeen
translatedintoalmost90languages,andisextensivelyused,includinginmanychildand
adolescentpsychiatricoutpatientdepartmentshereinNorway.
Then,whatisvalidation?LetmegiveyouLast´sdefinitionfrom“Adictionaryof
epidemiology”(Last2001):Validationis“Theprocessofestablishingthatamethodis
sound.”Thisshowsthatvalidationisabroadconceptinepidemiology.Thatthemethod
is“sound”raisesthenextquestion:“Soundforwhat?”Thatis,weneedtoconsiderthe
useofaninstrumenttobeabletotellifitiswellvalidated.AsKane(2006)statesit,one
oftheauthoritiesonvalidation:“Validationinvolvestheevaluationoftheproposed
interpretationsandusesofmeasurements…Itisnotthetestthatisvalidatedanditisnot
thetestscoresthatarevalidated.Itistheclaimsanddecisionsbasedonthetestresults
thatarevalidated…”Wewillreturntothisstatementlateron.Butnow,Ijustwantto
underlinethatvalidationoftheSDQiscloselylinkedtohowweuseit,andhowwe
interprettheresults.MosttimestheSDQisusedasascreeninginstrument–weuseitto
screenforpsychiatricproblemsinchildrenandadolescents.Thenitisveryimportantto
validateitasascreeninginstrumentandknowexactlywhichconclusionswecandraw
fromtheresultsweget.
Planfortheteaching
First,Iwillgiveyouanoverviewofthevalidationprocess.ThenIwillgiveyousome
detailsabouteachstep,beforewewillsummarizewhatwouldbeagoodstrategyfor
validationofaquestionnaireliketheSDQ.
Then,wewilltakealookathowtheSDQhasbeenvalidatedindifferent
countries,andparticularlyinEnglandbyGoodman,whereitwasmade.Wewillnotgo
throughalltheSDQvalidationstudiesavailable,astherearealotofthem.ButthoseI
havepickedwillatleastgiveyouanimpressionofthequalityofthevalidationofthis
questionnairearoundtheworld.
Overviewofthevalidationprocess
LetmejustmentionthatwhenIsearchedforrelevantpapersforourtheme,Icame
acrossapaperfromCIH.PeterChipimoandKnutFylkesnespublishedapaperin2010
onComparativevalidityofscreeninginstrumentsformentaldistressinZambia
(Chipimo&Fylkesnes2010).Actually,thisisoneofthebestvalidationstudiesIhave
comeacrosssofar.Irecommendyoutoreadit.
Whatarethemainingredientsofavalidationstudy?LetusthinkabouttheSDQ.
ItsoriginalversionisinEnglish,andletusimaginethatweconsidertranslatingitinto
Nepalese,becauseweneedascreeninginstrumentformentalproblemsamongchildren
andadolescentsinNepal,andasyetthereisnoauthorizedversionoftheSDQin
Nepalese.Letuslookattheprocess:
1
Firststage:Electionandtranslation/adaptationoftheinstrument
1.EvaluatetheusefulnessoftheSDQforourneed
2.Considertheaimofthetranslation/adaptationprocess
3.Thetranslation/adaptationprocessitself
Secondstage:Examiningthevalidityoftheinstrument
1.Samplesforvalidation
2.Differenttypesofvalidityandreliabilityandhowtheyaretested
a)Criterionvalidity
b)Contentvalidity
c)Constructvalidity
d)Internalconsistency
e)Reproducibility
f)Responsiveness
g)Floororceilingeffects
h)Interpretability
Atablesummarizingthevalidationprocess
Firststage:Electionandtranslation/adaptationoftheinstrument
1.EvaluatetheusefulnessoftheSDQforourneed
Inthisevaluationprocessweshouldanswerthefollowingquestions:
a)Exactlywhatdoweneedaninstrumentfor?
OfteninacountrylikeNepal,oneofthefirstinstrumentswewouldneedisaneffective
screeningtool,asprofessionalhealthworkersarescarceandresourcesingeneralare
few.Ifwecouldscreenformentalproblems,thenwecouldselectfortreatmentthose
childrenwhoaremostheavilyaffected.
b)IsthereanyinstrumentalreadytranslatedintoNepalesethatcanservethepurpose?
Ifthisisthecase,wewouldsavealotofworkandproblems,asthetranslationand
validationprocessisprettydemanding.Wewouldbewillingtocompromisesome
regardingthequalitiesoftheinstrument,ifitwasalreadytranslatedandtested.
c)Ifnot,whicharethe“candidateinstruments”availableinEnglishorinotherlanguages
thatmaybetranslated?
ThereareotherinstrumentsthantheSDQavailable,althoughnowadaysitdoesnot
seemlikeitForexample,moreextensivequestionnairesliketheverywellvalidated
ASEBAsystem(Achenbach,CBCL,TRF,YSR).Weshouldconsulttheliteratureaswellas
expertsinthefieldbeforewedecidedwhichquestionnairetouse.Butitshouldbenoted
thattheSDQseemstobeasgoodastheASEBAsystemtoidentifychildrenand
adolescentswithcommonpsychiatricdiagnoses.
d)Whichofthesecandidateinstrumentswillbethemostusefulforourpurpose?
Again,whatdoweneedtheinstrumentfor?IstheSDQreallythemostuseful
questionnaireavailableforourpurpose?TheSDQisuseful,butithasitsweaknesses.
ArethereotherinstrumentsthatcouldbemoreusefulnowinNepal?
2
e)Hastheoriginalversionofthisinstrumentbeenwellenoughvalidated?
Whenwehavefoundthemostpromisinginstrument,thenweneedtotakeacareful
lookatthevalidationprocess.Thatis,inourcase,hastheSDQbeenwellenough
validatedinEngland?Ifnot,weshouldbecarefultochooseit,becausewedon´tknow
enoughaboutit–wedon´tknowhowitreallyworks.
2.Considertheaimofthetranslation/adaptationprocess.
YouhavealreadybecomeawarethatIuse“translation/(slash)adaptation”.Thisis
becauseitisprettycommontotalkaboutadaptationofaninstrumentinsteadof
translation.Wecannotjusttranslateaninstrumentwordbywordintothenewlanguage
andthinkthatitwillwork.Weneedtoadaptittothenewlanguageandculture.This
takesmuchmorethansimpletranslation–wehavetoworkonituntilthevalidation
processshowsthatitworksthesamewayasitdoesintheoriginalversion.
Therefore,beforewemoveontothepracticalthings,wewillneedtotakealook
atsomeprerequisitesforusingaquestionnaireinadifferentculture.Weneedto
considertheaimofthetranslation.Forexample,theSDQwasdesignedinBritain.We
wantaNepaleseversionthatworksexactlythesamewayinNepalastheEnglish
versiondoesinEngland.
WhataretheprerequisitesforbeingabletousetheSDQinothercultures,suchas
theNepalese?Inthiscontextwetalkaboutequivalenceofthequestionnaireinthe
EnglishandNepalesecultures.IwilllistthefourlevelsofequivalencethatHuiand
Triandis(Hui&Triandis,1985)mention.WhenIhaveusedtheirterminology,Ineedto
informyouthatthereareanumberofdifferentconceptsusedaboutthedifferent
equivalencelevels,andthisis,ofcourse,confusing.ButIhopethatyouwillgraspthe
mainpoints.
1.Conceptual/functionalequivalence
Thisisthefirstandmostnecessaryrequirementforcross‐culturalcomparison.Then,
whatdoesitmean?“Aconstructthatcanbemeaningfullydiscussedinthecultures
concernedissaidtohavecross‐culturalconceptualequivalence.”Thismeansthata
constructexistsinbothcultures,andthatthisconstructisrelevant,andacceptableand
hassimilarmeaninginthetwocultures.
Theexampletheauthorsgiveistheconstruct“weight”.Weightlacksconceptual
equivalenceifyouintendtocompareabushel(skjeppe,36,4liter)oforangesandlove,
becauseweightisirrelevantasanattributeoflove.Inourexample,doespeerproblems
asaconstructexistintheNepaleseculture,andisitrelevantandacceptable?Doesit
havesimilarmeaningasintheBritishculture?
Conceptualequivalenceiscloselytiedwithfunctionalequivalence,whichin
psychologicalresearchhastodowiththesimilaritybetweenthegoalsofdifferent
behaviours.Twoactsofaggressionarefunctionallyequivalentacrosstwoculturesif
peopleofbothculturesemitsuchbehavioursincertainsituations,toachievecertain
purposes.Forexample,doSDQbehaviouralproblemssuchasdisobedience,stealing,
lyingandfightinghavethesamefunctionsintheBritishandtheNepaleseculture?
Orpeerproblemitemssuchas“rathersolitary,tendstoplayalone”,“hasatleastonegoodfriend”,
“generallylikedbyotherchildren”,“pickedonorbulliedbyotherchildren”and“getsonbetterwithadults
thanwithotherchildren”?
2.Equivalenceinconstructoperationalization
Operationalizationisthetransitionfromtheorytomeasurement.Ifaconstruct,suchas
3
childhoodpeerproblems,isoperationalizedinthesamewayinBritainandinNepal,
thentheinstrumentthatismadeisequivalentinconstructoperationalizationacross
thesetwocultures.Inaddition,theoperationalizationshouldbeequallymeaningfulin
thetwocultures.Letmementionanexampleoflackingequivalence:Operationalizing
aggressionintermsofverbalinsultswouldlackequivalencewhentheobjectiveisto
studyaggressivebehaviourbetweenmutepeopleandthegeneralpopulation.
Inourcase,theoperationalizationofpeerproblemsintheSDQshouldbeequally
meaningfulinBritainandinNepal.Thisalsomeansthatpeerproblemsas
operationalizedintheSDQispaidthesameamountofattentioninthetwocultures.
3.Itemequivalence
Thismoreconcreteandmicro‐levelofequivalencepresupposesthetwomentioned
typesofequivalence.Letusassumethataconstructhassimilarmeaningintwocultures,
andthatitisoperationalizedinsimilarways.Thenthenextconsiderationisthatthe
constructhastobemeasuredbythesameinstrument,inourcasetheSDQ.Onlyby
doingthiscanculturesbecomparednumerically.Thatis,onlythenisitpossibleto
comparelevelsofanxiety,andhyperactivitybetweenNepaleseandBritishchildren.On
theitemlevel,theinstrumentsusedinthedifferentcultureshavetobeidentical.For
example,eachofthe25itemsoftheSDQshouldmeanthesamethingtosubjectsin
BritainandNepal.Ifthisisnotthecase,thentheSDQineffectrepresentstwoseparate
tests,oneforeachculture.Ifthishappens,directcomparisonoftestscoresismisleading
andillegitimate.
MyfriendKamranSalayevinAzerbaijanexaminedthefactorstructureofthe
parentversionoftheSDQandfoundthatthepeerproblemsitem“pickedonorbullied
byotherchildren”loadedhighestontheconductproblemssubscale.Why?Probably
becausemostparentsthoughtthatachildispickedonorbulliedbecauseitbehaves
badlytowardsotherchildren.Thatis,itwasperceivedasabehaviouralproblemmore
thanapeerproblem.Myquestionis:DoestheAzeriversionoftheSDQshowconceptual,
operationalizationand/oritemequivalencewiththeEnglishversion?
4.Scalarequivalence
Scalarequivalenceisonlypresentifthe3othertypesofequivalencearepresentandifit
canbedemonstratedthattheconstructismeasuredonthesamemetric.Thismeans
thatanumericalvalueontheSDQscalereferstothesamedegree,intensity,or
magnitudeoftheconstructinbothBritainandinNepal.Thistypeofequivalenceisideal
forconcretecross‐culturalcomparison,butitisthemostdifficulttoachieve.
TorbjørnTorsheimandI(Sanneetal,2009)examinedtheparentandteacher
versionsoftheNorwegianSDQformeasurementinvariance.Firstofall,wehadtofind
outifitwaspossibletodoso,thattherewasscalarequivalence.Withsomereservations,
wefoundthatitwassafetocomparethetwoversions,thattherewasscalar
equivalence.However,congruentwithanearlierstudy,wefoundthatparentsand
teachersdifferedsubstantiallyintheirabilitytodiscriminateononePeerproblemitem,
namely“hasatleastonegoodfriend”.ThesamewasthecasefortheProsocialitem
“sharesreadilywithotherchildren”.Parentsansweredmorefavourablythanthe
teachers,probablybecauseofwhatwecallsocialdesirability.Theparentsanswered
morefavourablybecausetheydesiredtogivethemoresociallyacceptableanswer.They
wantedtobeabletocommunicatethattheirchildrenhadafriend,andthattheyshared
readilywithotherchildren.Buttakingthesetwoitemsintoaccount,wefoundit
sufficientlysafetocomparetheratingsofteachersandparents.Andwhenwedidthat,it
4
showedthatthemeanSDQscorewaslowerfortheteacherversioncomparedwiththe
parentversionforallfactors.KyrreBreivikandIarealsoworkingonapaperwherewe
comparethescoresoftheBritishandtheNorwegianSDQ.Butmypointhereisjustto
stressthatifwewanttocompareSDQ‐scoresbetweentwocountriesorcultures,then
wehavetohaveall4levelsofequivalencepresent.Ifnot,acomparisonofscoresdoes
notmakesense.
Aswefinishthispartaboutequivalence,letmejustaddthatacleardemarcation
ofthesefourtypesofequivalenceisnoteasy–thereisconsiderableoverlapbetween
them.
3and4:Howtodemonstrateitemandscalarequivalence?
Hui&Triandis(1985)aswellasothers(eg.Acquadroetal2008)arguefortheuseoftheitemresponse
theory(IRT)approachinthevalidationofaninstrument.Inthiswayitispossibletobypasstheproblem
ofselectingarelevantandunbiasedcriterionforjudginganinstrument.Wewillshortlycomebacktothis
aswetalkaboutthedifferentkindsofvalidity.ButletmejustsayherethatIRTusesitemparameters
derived“internally”,andinthiswayweavoidtheuseof“external”criteriasuchasa“goldstandard”.Item
characteristiccurves(ICC),whichrepresenttheprobabilitiesofrespondingtoaniteminacertain
specifiedmanneratdifferentlevelsofthelatenttraittobemeasured,areobtainedfromdifferentcultures.
Suchdifferencescanpointtothelackofequivalencebetweenthetwoculturesonaparticularitem.Onthe
otherhand,aninstrumentthathassimilarICCsacrosscultureshas,atleastinpart,demonstrateditsitem
equivalenceandscalarequivalence.
3.Thetranslation/adaptationprocessitself
ThenwehavecomesofarthatwewanttotranslatetheSDQfromEnglishintoNepalese.
Thisisaprettyresource‐demandinganddifficultpart,thatis,ifwewanttomakeagood
translation.Butathoroughtranslationwillpayoff,bothinclinicalpracticeandin
research.However,wehavetobeawarethataninstrumentwhichhasaveryaccurate
translationisnotfoolproof.Thereforewehavetotestitafterwards,tofindoutwhether
thetranslationworksasitshouldornot.
Thepossibilityofnon‐equivalenceonanabstractlevel,suchasconceptualequivalence,cannotbe
neglected.Differenceinsocialdesirabilitylevelsandmotivationtorespondontherespondents´part,and
inconsistenttestadministrationproceduresaresomeotherpotentialproblemsthatcanplagueastudy.
Kimberlin&Winterstein(2008)remindsusthatself‐reportsofbehaviourareparticularlysubjectto
problemswithsocialdesirabilitybiases.ThisisimportanttokeepinmindwhendealingwiththeS‐SDQ.
Severalauthorshavewrittenaboutthetranslationprocess.Ihaveusedareview
paperbyAcquadroetal(2008)tosummarizethemostimportantelements.Thismeans
thatIwillnotgointodetailhere,asitisanextensivetheme.Buteveryteamdecidingto
translateaninstrumentshouldspendconsiderabletimeplanningtheprocess.
Mostauthorsrecommendamultistepapproachinvolvingacentralizedreview
process.Thismeansthattherearesomekeypeopleplanning,leadingandreviewingthe
wholeprocess.However,Acquadroetalfoundthateachresearchgroupinreality
proposesitsownsequenceoftranslationeventsandweightseachstepdifferently.Some
havearguedforashorter,fasterandmoreinexpensivetranslationprocessinsome
situations(Mathiasetal1994),asmostexpertssuggestprettyresourcedemanding
procedures.However, most agree with Acquadroetal,arguingthatarigorousand
multistepprocedureleadstobettertranslations.Buttheyalsounderlinethatthereisno
empiricalevidenceinfavourofonespecificofthesethoroughmethods.Theygive
differentexamplesofproceduresthattheythinkareacceptableandrecommendable.Let
mesummarizeoneofthese,inordertoshowyouthemostimportantingredientsand
giveyouanexampleofhowitcanbedone:
5
1.Permissionfromtheauthor
Inourcase,weneedtocontactthecreatoroftheSDQ,RobertGoodman.Goodmanisa
goodman,butheisalsoaverybusyman.Authorizingtranslatedinstrumentsisagreat
dealofwork.LetmetellyouwhatweexperiencedinHerat,Afghanistan.TheSDQhas
beentranslatedintotwoofthemainlanguagesinAfghanistan,DariandPashto.Inthe
cityofHerat,mostpeoplespeakDari,butadifferentdialectthanwhattheyspeakinthe
capital,Kabul.ThismeansthattherearecertainwordsintheDariversionoftheSDQ
thatpeopleinHeratdon´tunderstand.However,HeratisclosetoIran,wheretheyspeak
Farsi,andtheHeratdialectofDariisclosertoFarsithanitistoDarispokeninKabul.
But,evenintheFarsiversiontherearewordsthatpeopledon´tunderstandinHerat.We
wouldliketousetheSDQinresearchinHerat,andthereforeIaskedGoodmanfor
permissiontotranslateitintoDari,Heratdialect.Buthedidnotgivethispermission,
becauseheseesitasaproblemiftheSDQistranslatedintovariousdialects.Also,the
workforhimduringthetranslationprocesswouldbetoomuchforhimtoaccept.So,
whatdowedonow?
2.Forwardtranslation
Atleasttwodifferenttranslationsdonebybilingualtranslatorsareneeded.Themother
tongueofthetranslatorsshouldbethetargetlanguage.Thatis,ifwewanttotranslate
theSDQfromEnglishtoNepalese,thenthetranslatorsshouldbefluentinboth
languagesandhaveNepaleseastheirmothertongue.
Ingeneral,completingaquestionnaireshouldnotrequirereadingskillsbeyond
thatofa12‐year‐old.IknowthatGoodmanhasmadesurethatthisisthecaseforthe
EnglishversionoftheSDQ.However,whentranslatingtheSDQintootherlanguages,we
havetomakesurethatthesameisthecaseinthenewculture.Thisisoftenevenmore
importantthaninEnglandandotherWesterncountries,astheeducationallevelina
countrylikeNepalislow,andwecanexpectlargepartsofthepopulationtohavea
limitedvocabularyincomparisontohighlyeducatedpeople.
3.Synthesisofthetranslations
Thetwotranslatorsshouldproduceonetranslationoutofthetwoversions.
4.Back‐translation
Twoback‐translationsareseenasaminimum,andthosedoingitshouldbeblindedto
theoriginalversion.Theyshouldhavethesourcelanguageastheirmothertongue.That
is,inourexample,twopersonswithEnglishastheirmothertongueshouldback‐
translatetheNepaleseversionoftheSDQintoEnglish.
Letmeaddthatithasbeenmuchdiscussedwhetherback‐translationis
necessary.Assomestate:Back‐translationis“nottheinfalliblequalitycontroltoolitis
purportedtobe”.However,althoughthisstephasitsweaksides,mostauthorities
recommendincludingitinthetranslationprocess.
TheNorwegianversionoftheSDQwasmadebyateamtranslatingitfrom
EnglishintoNorwegian,andbysomeothersdoingtheback‐translation.Butthenthe
adaptationprocessofthisfirstversionstopped,andthefirstback‐translationbecame
thefinalversion.Someyearsago,whenIstarteddoingfactoranalysisoftheNorwegian
SDQ,myfirstanalysisshowedthatsomethingwaswrongwiththefirstoftheconduct
problemsitems.InEnglishthisitemsays“Oftenhastempertantrumsorhottempers”.
Thishadbeentranslated“Harofteraserianfallellerdårlighumør”.Thisitemloaded
higheronEmotionalproblemsthanonConductproblems.Why?Ifwelookatthelast
6
partoftheitem,“hottempers”inEnglishand“dårlighumør”inNorwegian,the
Norwegianpartcanbeback‐translated“badmood.”Thatis,quiteafewNorwegian
respondentsperceivedthisasanemotionalproblemmorethanabehaviouralproblem.
Ifsomeonehaddoneasimpleexaminationofthefactorstructureinasmallsample,this
translationweaknesswouldhavebeendiscovered.ButwhenIfoundout,itwastoolate
tochangethetranslation,becauseithadalreadybeenusedintheBergenChildStudy.
Bytheway,thisitemevenintheoriginalversionisnotoptimal.Generally,items
shouldbeshort,clearandsimple,andshouldnotconsistoftwoquestionsatthesame
time.However,thisitemreads:“Oftenhastempertantrumsorhottempers”.Thatis,this
itemisnotasclearasitshouldhavebeen.
5.Expertcommittee
Thiscommitteeshouldbecomposedofmethodologists,healthprofessionals,language
professionals,andallthetranslatorsinvolvedintheprocess.Theoriginaldevelopersof
thequestionnaireshouldbeinclosecontactwiththecommittee.Thecommitteeshould
studytheprocesssofar,andagreeonaversionthatwecouldcallthepre‐finalversion.
Theyshouldaimatachievingthetwofirstlevelsofequivalenceatthispoint,thatis,
conceptualequivalenceandequivalenceinconstructoperationalization.
6.Testofthepre‐finalversion
Ideally,30‐40personsshouldbetested.Eachsubjectshouldcompletethequestionnaire
andbeinterviewedaboutthemeaningofeachitem.Thisstageprovidesarough
evaluationofcontentvalidity.Wewillreturntotheconceptofcontentvaliditysoon.
7.Submissionofdocumentationtothedevelopersorcoordinatingcommitteeforappraisal
oftheadaptationprocess
Thisisaprocesstoensurethatallstepshavebeenperformedandfullydocumented.
AsIsaid,thisisoneexampleofathoroughtranslationprocess.Therearemanywaysto
doit,butoneofthemainpointsinallproceduresisthateachstepinthetranslationis
thoroughlyreportedinwriting.Inthiswayitwillbepossibletogobackandidentify
weakspotsinthetranslationprocessatalaterstage.
Theauthorsemphasizethatthetranslationisonlythestartoftheadaptation
process.Asmentioned,thepre‐finalversionneedstobetested.Thatis,theexamination
ofthepsychometricpropertiesstarts.
References
AcquadroC,ConwayK,HareendranA,AaronsonN,fortheEuropeanRegulatory
IssuesandQualityofLifeAssessment(ERIQA)Group.Literaturereviewofmethodsto
translatehealth‐relatedqualityoflifequestionnairesforuseinmultinationalclinical
trials.ValueinHealth2008,3:509‐521.
ChipimoPJ,FylkesnesK.Comparativevalidityofscreeninginstrumentsfor
mentaldistressinZambia.ClinicalPractice&EpidemiologyinMentalHealth,2010,6:4‐
15.
EMGO+(InstituteforHealthandCareResearch).Questionnaires:selecting,
translatingandvalidating.01.01.2010.http://www.emgo.nl/kc/preparation/
research%20design/8%20Questionnaires%20selecting,%20translating%20and%20val
idating.html
7
FayersPM,MachinD.Qualityoflife.Theassessment,analysisandinterpretation
ofpatient‐reportedoutcomes.2nded.Wiley,2007.
HuiCH,TriandisHC.Measurementincross‐culturalpsychology.Areviewand
comparisonofstrategies.Journalofcross‐culturalpsychology1985,16:131‐152.
KaneMT(2006).Validation.InBrennanRL(Ed),EducationalMeasurement,4th
ed.Westport:PraegerPublishers,pp17‐64.
KimberlinCL,WintersteinAG.Validityandreliabilityofmeasurement
instrumentsusedinresearch.AmJHealth‐SystPharm2008,65:2276‐2284.
LastJM.Adictionaryofepidemiology,4thed.OxfordUniversityPress,2001.
MathiasSD,FiferSK,PatrickDL.Rapidtranslationofqualityoflifemeasuresfor
internationalclinicaltrials:avoidingerrorsintheminimalistapproach.QualityofLife
Research1994,3,403‐412.
SanneB,TorsheimT,HeiervangE,StormarkKM.TheStrengthsandDifficulties
QuestionnaireintheBergenChildStudy:Aconceptuallyandmethodicallymotivated
structuralanalysis.PsychologicalAssessment2009;21:352‐364.
SchmidtME,SteindorfK.Statisticalmethodsforthevalidationofquestionnaires.
Discrepancybetweentheoryandpractice.MethodsInfMed2006,4:409‐413.
TerweeCB,BotSDM,deBoerMR,etal.Qualitycriteriawereproposedfor
measurementpropertiesofhealthstatusquestionnaires.JournalofClinical
Epidemiology2007;60:34‐42.
8
9