Linking TOEFL iBT
™
Scores to
IELTS
®
Scores –
A Research Report
ETS
December 17, 2010
LinkingTOEFLiBT™ScorestoIELTS®Scores–AResearchReport
1
Abstract
TOEFL®testscoreshavebeenwidelyacceptedasevidenceofnonnativeEnglishspeakers’proficiencyinthe
useofEnglish.TheTOEFLtestmeasurestesttakers’abilitytocommunicateinEnglish‐mediumcollegesand
universities.
TheTOEFLiBT™testmeasuresacandidate’sabilitytocombinelis tening,reading,speakingandwritingskills
withauthenticandinnovativeintegratedtasks,makingitthemostcomprehensiveEnglish‐language
proficiencytest.Acceptedbyover7,500scoreusersinmorethan130countries,theTOEFLtestisthemost
widelyrecognizedEnglish‐languagetestintheworld.
Recently,ETSconductedscorecomparisonresearchbetweentheTOEFLtestand
analternativetest,IELTS
(InternationalEnglishLanguageTestingSystem).Fortheresearch,ETSobtainedasampleof1,153students
whohadbothIELTSandTOEFLscores.Usingequipercentilelinking,thecorresponding TOEFLscorethat
wouldpassthesamepercentageoftesttakersforeachIELTSscoreswasobtained.Thescore
comparison
resultsforeachsection(Listening,Speaking,Reading,andWriting)andthetotaltestshowedthatmostofthe
studentsscoredinthemiddletomid‐highscorerangesonbothtests.Limitationsoftheresearchand
suggestionsforfutureresearcharealsodiscussed.
LinkingTOEFLiBT™ScorestoIELTS®Scores–AResearchReport
2
LinkingTOEFLiBT™ScorestoIELTSScores
Introduction
TOEFL®testscoreshavebeenwidelyacceptedbyeducationalandotherinstitutionsthroughouttheworldas
evidenceofnonnativeEnglishspeakers’proficiencyin Englishlanguageuse,andtheTOEFLtesthasreceived
verypositivefeedbackfromscoreusersandinstructorsintheEnglishasaSecond Language(ESL)community.
Thereare
manyoccasionswhenscoresfromdifferenttestsorevenfromdifferentversionsofthesametest
needtobelinkedfordecisionmakingorotherpurposes.Forexample,thescoresontheSAT®IandACT®
tests,twowellknowncollegeadmissionstestsintheUnitedStates,havebeenlinked
toeachothertohelp
universityadmissionsofficesusethescorestoevaluateapplicants.Aresearchstudywascarriedoutbyboth
ETSandACTresearchstafftodevelopascorecompar isontableforthispurpose(Pommerich,Hanson,Harris,
&Sconing,2000.)Anotherexampleisthat,whentheTOEFLiBTtest,
wasdevelopedtobetakenviathe
Internetandtoreplacethecomputer‐adaptivetest(cBT)version,astudywasconductedtolinkTOEFLiBT
scorestoTOEFLcBTscores(Wang,Eign or,&Enright,2007.)
Recently,ETSconductedscorecomparisonresearchbetweentheTOEFLtestandanalternativetest,IELTS
(InternationalEnglishLanguageTestingSystem).AsabestpracticerequiredbytheGuidelinesforPracticeby
theInternationalLanguageTestingAssociation(ILTA,2007),theStandardsforEducationalandPsychological
Testing(AERA,APA,NCME,1999),andtheETSstandards(ETS,2002,p.45,),appropriatepsychometric
proceduresshouldalwaysbeusedtolink
scoresfromtwodifferentassessmentsifthescoresaretobe
compared.Tocomplywiththisstandard,ETSpsychometricstaffconductedaresearchstudytoexploreand
establishempiricalrelationshipsbetweenthescoresonthetwotests.Thisresearchwasdesignedandcarried
outtoanswerthefollowingtworesearch
questions:
1. WhatTOEFLiBTsectionscoresarecomparabletoIELTSsectionscores?
2. WhatTOEFLiBTtotalscoresarecomparabletoIELTStotalscores?
LinkingTOEFLiBT™ScorestoIELTS®Scores–AResearchReport
3
FindingthecorrespondingTOEFLiBTscoresforeachoftheIELTSscoreswasaccomplishedbywhatisknown
asscorelinkingorscorecomparison(calledscorecomparisonhereafterinthisreport)inthefieldof
educationalmeasurement.
Inthisreport,thesamplesandtheanalyseswillbedescribedintheMethod
section.ThesectionforResults
willpresentscorecomparisonsforthefourskillsectionsandthetotalscores.Thisisfollowedbythe
Discussionsectionthatconsiderstheinterpretationoftheresultsandtheirimplicationsintermsofthe
generalizationofthefindingsfromthisstudy,evaluatesthelimitationsof
theresearch,andmakes
suggestionsforfutureresearch.
LinkingTOEFLiBT™ScorestoIELTS®Scores–AResearchReport
4
Method
DataCollection
Inordertoacquireempiricalevidenceforlinkingthescoresonthetwotests,scoreswereobtainedfrom
studentswhohadtakenbothtests(Pommerich,2007).Thedata collectioneffortforthisresearchstartedin
2008andcontinuedinto2 009.Thefirstbatchofdata,obtainedin2008,
contained217studentswho
submittedtoETScopiesof theirTOEFLandIELTSscorereports.Inordertoincr easethesamplesizeandthe
stabilityoftheresults,ETSthencontactedalargenumberoftesttakersviae‐mailandencouragedthe
participantstosubmittheirlatestscoresonboth
teststoETS. Ove r1,000stude ntsrespondedandreported
theirscoresonthetwoteststoETS.Thetwobatchesofdatawerethencombinedforthestudy.
Beforeanyanalysesstarted,thefirsttaskwastocleanthedata,whichinvolvedremovingunusablerecords.
Recordsthathadmissingscores
ononeormoresectionsorthathadinvalidscoreswereremovedfrom
subsequentanalyses.Forexample,onereportedIELTSscoreof4.7wasidentifiedasaninvalidscoreandwas
eliminatedfromtheanalysisdataset
1
.Afterthisdatacleaningprocedure,1,1 53recordswereretainedfor
theanalysesdescribedinthenextsection.Closeto1,000recordsincludedtheparticipants’countryof
testinginformation,includingabout70countriesfromaroundtheworld.ThelargestgroupwasfromChina
includingHongKong(41%),followedbyUSA(6% ),
andJapan(5%).Theotherswereallbelow4%.
DataAnalysis
Tounderstandthestatisticalcharacteristicsofthescoresonthetwotestsfromthefinalsampleof1,153
studentsandtoestablish therelationshipsbetweenthescoresonthetwotests,thefollowingtypesof
analysiswereconducted.The
findingsfromtheanalysesarepresentedintheResultssection.
1.Descriptivestatisticswerecomputedtoevaluatesuchinformationastheaveragescoresoneachtestat
boththesectionandthetotalscorelevel,andhowmuchvariation(standarddeviation)wasobservedinthe
scores.Thecorrelationbetweenthescores
onthetwotestswasalsocomputedtoseetowhatextentthe
scoresonthetwotestswererelatedtoeachother.
1
IELTSscoreshave9bandsinone‐halfincrements(e.g.,3,3.5,4,4.5 )
LinkingTOEFLiBT™ScorestoIELTS®Scores–AResearchReport
5
2.Linkingscoresisbasicallyatransformationfromascoreononescaletoascoreontheotherscale
(Holland,2007)andcanbeaccomplished inavarietyofwaysdependingonthenatureofthedataandthe
purposeofthelinking(Kolen&Brennan,1995;Holland&Dorans,
2006).Tofindthecorrespondence
betweentheTOEFLiBTandIELTStestsforeachsectionandforthetotalscore,ETSpsychometricstaff
conductedextensiveanalysesbetweenMarchandAugust2009usingavarietyofstatisticalmethods
includingregression‐basedprediction,conditionalprobability‐basedscorematching,andequipercentile
linking.In
reviewingtheresultsandinconsiderationofthepurposeofthisresearch,wedeemeditmost
appropriatetousetheequipercentilelinkingmethod.Thisapproachhasbeenusedextensivelyinavarietyof
settings(e.g.thelinkingoftheSATtestwiththeACTtest) andallowedustoidentifythe
corresponding
TOEFLiBTscoreswhichwouldrestrict/passasimilarpercentageofthestudentsinthesampleaswouldthe
IELTSscores.Giventheneedforbothteststoscreenoutapplicantsinacomparablemanner,webelievethis
tobethemostdefensibleandstraightforwardapproach.
Results
DescriptiveStatisticsoftheSample
TOEFLiBTscoresrangefrom0to30foreachsectionand0to120forthetotaltest;thetotaltestscoreisthe
sumofthefoursections.IELTSreportsbothsectionandtotalscoresona9‐bandscaleinone‐half(0.5)
band
increments;thetotaltestscoreistheaverageofthefoursections.Thisgeneralinformationaboutthe
differentscorescalesofthetwotestscanhelpputthescoresoftheresearchsampleinperspective.For
example,becauseofthecompactIELTSscorescales,IELTSscaleswillshowmuch
smallerscorevariation
(standarddeviationvaluesinTable1)thantheTOEFLiBTscales.Also,forsomeIELTSscoresinTables3to7,
thecorrespondingTOEFLiBTscoreswillshowscorerangesduetothelargerTOEFLscales.
Table1givesthesamplesize,themeanscores,andthescore
variations(standarddeviation,scorerange)of
thetwotests.OnIELTS,theReadingandListeningmeanscoreswereboth6.8,andtheSpeakingandWriting
meanswere6.3and6.1,respectively.Thetotaltestmeanscorewas6.6.OntheTOEFLiBTscale,theReading
andListeningmeanscoreswere
bothcloseto21andtheSpeakingandWritingmeanscoreswere20and
21.6,respectively.Thetotalmeanscorewas83.6.
Toevaluatetheoverallperformanceofthestudentsinthisresearchsample,the2009testperformance
informationonthetwotestswasobtainedfromtherespectivewebsitesand
ispresentedinthecolum nfor
2009populationmeaninTable1.
LinkingTOEFLiBT™ScorestoIELTS®Scores–AResearchReport
6
Generallyspeaking,thestudentsinthissample,asagroup,scoredhigheronbothteststhantheirrespectiv e
populationsdidin2009.Inotherwords,thissampleappearedtobemoreablethanthegeneralpopulations
ofIELTSandTOEFLiBT testtakers,respectively.Thiscouldbetheoutcome of
selfselectionofthe
participantsinthatstudentshavinghighertestscoresmighthavebeenmorewillingtoreporttheirscores
thanthosehavinglowerscores.Asaresult,therewereveryfewlowscoresinthedatabutrelativelymore
scoresatthemiddletomid‐high levels. Ourlevel
ofconfidenceinthe resultsforthelowscorelevelswould
thenberelativelylowerthanthatforthemiddletomid‐highlevels.
LinkingTOEFLiBT™ScorestoIELTS®Scores–AResearchReport
7
Table1.MeansandStand a rdDeviations
Test N Mean
Standard
Deviation
Score
Range
2009
Population
Mean*
IELTSTotal 1,153 6.6 0.8 3.5‐9 5.8
IELTSListening 1,153 6.8 1.1 2.5‐9 6
IELTSSpeaking 1,153 6.3 0.9 1.5‐9 5.7
IELTSReading 1,153 6.8 1.1 2.5‐9 5.8
IELTSWriting 1,153 6.1 0.8 2‐9 5.4
TOEFLTotal 1,153 83.6 20.3 23‐119 79
TOEFLListening 1,153 20.9 7.6 1‐30 19.4
TOEFLSpeaking 1,153 20.0 3.8 5‐30 19.7
TOEFLReading 1,153 21.2 7.8 1‐30 19.9
TOEFLWriting 1,153 21.6 4.6 5‐30 20.5
*The2009populationmeanscoresbasedontheinformationprovidedattheTOEFLandIELTSwebsites.
Table2liststhecorrelationcoefficients betweenthescoresfromthesamemeasuresaswellasthescores
fromthetotaltests.Acorrelationcoefficientindicatestheextenttowhichthescoresonthetwotestsare
relatedtoeachother.Acorrelationcoefficientgoesfrom0fornorelation
atallto1foraperfectrelation.
Thecorrelationbetweentotalscoreswillalwaysbehigherthanthatbetweensectionscoresbecausetotal
scorescontainallthesectionscoreinformationandaremorereliable.Asexpected,thetotaltestscore
correlationis0.73,higherthananysectionscorecorrelation,which
variesfrom0.44forWritingto0.68for
Reading.Generallyspeaking,forthesco res tobelinked,moderatelyhighcorrelations(aro und0.7)wouldbe
verydesirable,butmoderatecorrelations(around 0.5to0.6)arenotatypical.
LinkingTOEFLiBT™ScorestoIELTS®Scores–AResearchReport
8
Table2.CorrelationsbetweenIELTSandTOEFLiBTSectionandTotalScores
Score Correlation
IELTSListening&TOEFLListening 0.63
IELTSSpeaking&TOEFLSpeaking 0.57
IELTSReading&TOEFLReading 0.68
IELTSWriting&TOEFLWriting 0.44
IELTSTotal&TOEFLTotal 0.73
Thenextsectionsprovidetheequipercentilelinking resultsforeachofthesectionsandthetotalscore.The
resultsarepresentedintabularform.Whilewewereabletoobtaincorrespondingscoresforthetwotestsat
eachoftheIELTSscorebands,thenumb er ofcasesperscoreband
varied,withgenerallyveryfewcasesin
theverylowestandhighestscorebands.Ingeneral,assamplesizesincrease,resultsbecomemorestable
andreliable.Consequently,inthisstudy,wehavemoreconfidenceintheresultswheretherearerelatively
largercasecounts(i.e.,5%ofthesample)perscore
bandandwerecommendcautionininterpretingthe
resultswherethecasecountsarerelativelysmall.Wehavenoted,byshadingtheareaoftherangeofscores
onthetables,wheresamplesizesperscorebandaregreaterthan5%ofthetota l.
ListeningSectionScoreComparisonResults
Table
3presentstheequipercentilelinkingmethodbasedscorecomparisonresultsonthetwoListening
sections.TheshadedareainTable3indicatesthatmoststudentshadscoresfrom5.5to8.5onIEL TSand7to
29onTOEFLiBTtest,withveryfewstudents(lessthan5%orso)at
theotherscorelevels.Aswasmentioned
earlier,duetothelargerscaleofTOEFLscoresthanIELTSscores,someIELTSListeningscoresarefoundto
correspondtoarangeofTOEFLscoresinsteadofasingleTOEFLscore.Forexample,anIELTSListeningscore
of5.5wouldcorrespondto
TOEFLiBTListeningscoresof7to11.Insummary,thetabulatedresultsindicate
thatanIELTSListeningscoreof6.0wouldcorrespondtoTOEFLListeningscoresof12to19;anIELTSListening
scoreof6.5wouldcorrespondtoTOEF LListeningscoresof20to23;anIELTSListeningscore
of7.0would
correspondtoTOEFLListeningscoresof24to26;andsoforth.Inallcases,theIELTS scoresandtheir
correspondingTOEFLscores(orscoreintervals)wouldpasssimilarpercentagesoftesttakers.
Table3.ListeningSectionScoreComparisons
LinkingTOEFLiBT™ScorestoIELTS®Scores–AResearchReport
9
IELTSScore
TOEFL
Score
9 30
8.5 29
8 28
7.5 27
7 24‐26
6.5 20‐23
6 12‐19
5.5 7‐11
5 4‐6
4.5 3
<=4 0‐2
SpeakingSectionScoreComparisonResults
Table4presentsthescorecomparisonresultsfortheSpeakingsections.Theinformationcanbeinterpreted
inthesamewayasdescribedaboveforListening.AsshownbytheshadedareainTable4,mostscoreswere
withintherangeof5.0to8.0on
theIELTStestand14to27ontheTOEFLiBTtest.Again,anumberofIELTS
scorescorrespondtoTOEFL scorerangesinsteadofsinglescoresduetoscaledifferencesbetweenthetwo
tests.Forexample,thetableshowsthatanIELTSscoreof5.0wouldcorrespondtoTOEFL
Speakingscoresof
14to15;anIELTSSpeakingscoreof5.5wouldcorrespondtoTOEFLSpeakingscoresof16to17;anIELTS
Speakingscoreof6.0wouldcorrespondtoTOEFLSpeakingscoresof18to19;anIELTSSpeakingscoreof6.5
wouldcorrespondtoTOEFLSpeakingscoreof
20to22;andsoforth.Inallcases,theIELTSscoresandtheir
correspondingTOEFLscores(orscoreintervals)wouldpasssimilarpercentagesoftesttakers.
LinkingTOEFLiBT™ScorestoIELTS®Scores–AResearchReport
10
Table4.SpeakingSectionScoreComparisons
IELTSScore TOEFLScore
9.0 30
8.5 28‐29
8.0 26‐27
7.5 24‐25
7.0 23
6.5 20‐22
6.0 18‐19
5.5 16‐17
5.0 14‐15
4.5 12‐13
<=4 0‐11
ReadingSectionScoreComparisonResults
Table5presentsthescorecomparisonresultsfortheReadingsections.Theinformationcanbeinterpretedin
thesamewayasdescribedaboveforListeningandSpeaking.AsshownbytheshadedareainTable5,most
scoreswerewithintherangeof5to8.5
ontheIELTStestand4to29ontheTOEFLiBTtest.Thetableshows
thatanIELTSReadingscoreof5.0wouldcorrespondtoTOEFLReadingscoresof4to7;anIELTSReading
scoreof5.5wouldcorrespondtoTOEF LReadingscoresof8to12;an
IELTSReadingscoreof6.0would
correspondtoTOEFLReadingscoresof13to18;anIELTSReadingscoreof6.5wouldcorrespondtoTOEFL
Readingscoresof19to23;anIELTSReadingscoreof7.0wouldcorrespondtoTOEFLReadingscoresof24to
26;andsoforth.Note
thataTOEFLReadingscoreof29wouldcorrespondtobothIELTSReadingscoresof8
and8.5.ThisoccurredbecausewewantedallIELTSbandscorepointstoberepresentedinthetable.Since
thehighestTOEFLiBTscore(30)hastomaptothehighestIELTSbandscore
(9)andpercentiledata suggest
thatanTOEFLiBTscoreof29mapstoanIELTSbandscoreof8,thenithastobethecasethatanTOEFLiBT
scoreof29alsomapsto8.5,i.e.,thereisnoTOEFLiBTscorepointbetween29and30to
maptotheIELTS
bandscoreof8.5.Inallcases,theIELTSscoresandtheircorrespondingTOEFLscores(orscoreintervals)
wouldpasssimilarpercentagesoftesttakers.
LinkingTOEFLiBT™ScorestoIELTS®Scores–AResearchReport
11
Table5.ReadingSectionScoreComparisons
IELTSScore TOEFLScore
9 30
8.5 29
8 29
7.5 27‐28
7 24‐26
6.5 19‐23
6 13‐18
5.5 8‐12
5 4‐7
4.5 3
<=4 0‐2
WritingSectionScoreComparisonResults
Table6presentsthescorecomparisonresultsfortheWritingsections.Theinformationcanbeinterpretedin
thesamewayasdescribedaboveforListening,ReadingandSpeaking.AsshownbytheshadedareainTable
6,mostscoreswerewithintherangeof5
to7onIELTSand 14to28ontheTOEFLiBTtest.Thetableshows
thatanIELTSWritingscoreof5.0wouldcorrespondtoTOEFLWritingscoresof14to17;anIELTSWriting
score5.5wouldcorrespondtoTOEFLWritingscoresof18to20;anIELTSWriting
scoreof6.0would
correspondtoTOEFLWritingscoresof21to23;anIELTSWritingscoreof6.5wouldcorrespondtoTOEFL
Writingscoresof24to26;andsoforth.Inallcases,theIELTSscoresandtheircorrespondingTOEFLscores
(orscoreintervals)wouldpasssimilarpercentages
oftesttakers.
LinkingTOEFLiBT™ScorestoIELTS®Scores–AResearchReport
12
Table6.WritingSectionScoreComparisons
IELTSScore TOEFLScore
9.0 30
8.5 30
8.0 30
7.5 29
7.0 27‐28
6.5 24‐26
6.0 21‐23
5.5 18‐20
5.0 14‐17
4.5 12‐13
<=4.0 0‐11
LinkingTOEFLiBT™ScorestoIELTS®Scores–AResearchReport
13
TotalScoreComparisonResults
Table7presentsthecomparisonresultsforthetotalscoresonthetwotests.Thecomparisonoftotalscores
canbeinterpretedinthesamewayasdescribedaboveforthesectionscores.Asshownbytheshadedarea
Table7,mostscoreswerewithintherange
of5.5and7.5ontheIELTStestand46and109onTOEFLiBTtest.
Lessthan5%oftesttakersateachscoreorscoreintervalreceivedscoresoutsideofthisrange.Becauseof
themuchlargerTOEFLtotalscorescale(0to120)thanIELTS(0to
9),theTOEFLscorerangecorrespondingto
anIELTSscoretendstobeverybroad.Forexample,anIELTSscoreof5.5wouldcorrespondtoTOEFLscores
of46to59;anIELTSscoreof6.0wouldcorrespondtoTOEFLscoresof60to78;anIELTSscoreof6.5would
correspondtoTOEFLscoresof79to93;anIELTSscoreof7.0wouldcorrespondtoTOEFLscoresof94to101;
andsoforth.Inallcases,theIELTSscoresandtheircorrespondingTOEFLscores(orscoreintervals)would
passsimilarpercentagesoftesttakers.
Table7.TotalScore
Comparisons
IELTSScore TOEFLScore
9 118‐120
8.5 115‐117
8 110‐114
7.5 102‐109
7 94‐101
6.5 79‐93
6 60‐78
5.5 46‐59
5 35‐45
4.5 32‐34
0‐4 0‐31
LinkingTOEFLiBT™ScorestoIELTS®Scores–AResearchReport
14
Discussion
Tobetterunderstandthe scorecomparis onresultsonboththesectionandthetotaltestlevels,itis
importanttoevaluatetheresults inthecontextofthechara cterist icsofthetests,theresearchsample
employed,theintendeduseoftheresults,andthelimitationsofthestudy.Itisalso
worthwhiletothink
aboutwhatfurtherresearchcanbeconductedinthefuturetoenhancethequalityofthescorecomparisons.
EvaluationoftheScoreComparisonResultsandLimitations
Therearetwoconsiderations regardingtheinterpretationoftheresults.Thefirstisthecharacteristicsofthe
twotests.Itis
recognizedthattheTOEFLandIELTStests,whiletheyhavethesamenumberofsections
measuringsimilarskillsandunderlyingconstructs,werebuiltfromdifferentframeworksanddifferentte st
blueprints.Forexample,theTOEFLiBTtestisthelatestversionoftheTOEFLtestandisbuiltonthe
frameworkof
communicativecompetencybasedoncurrentsecondlanguageacquisitiontheo ryandpractice.
TheTOEFLiBTtestcontainsnumerousinnovativetasktypesthatcanonlybepresentedonacomputer‐based
testandtapintoawiderarrayofEnglishproficiencyskillsthanapaper‐basedtest.Asaresult,TOEFLiBT
scoresdonotmeanexactlythesamethingasIELTSscores,andmoderatecorrelations
2
betweenthescores,
asseeninTable2,canbeexpected.Aspreviouslynoted,forthescorestobelinked,moderatelyhigh
correlations(around0.7)wouldbeverydesirable,butmoderatecorrelations(around0.5to0.6)arenot
atypical.
Thesecondconsiderationisthesample.Asin manyotherstudies
ofthisnature,aself‐selectedgroupof
participantsvo lunteeredtoreporttheirIELTSandTOEFLscorestoETS,whichwouldlimitgeneralizingthe
resultstosamplesthatdiffersignificantlyfromthisresearchsample.Inparticular,thenumbersof
participantswithrela tivelylowandveryhighscores weresmall.Futuresamples
thatincludemorecaseswith
lowerandhigherscoreswouldberecommendedifanunderstan dingof themappingofscoresatthesescore
levelsisdesired.Whilethepresentcasecountsaresufficienttoevaluatethemappinginthemiddle ofthe
scorescales,cautionshouldbeexercisedwhencomparing
thelowerandhigherscores.Asdescribed
previously,thetablesincludeshadingtoindicatethosescorelevelswherewehaverelativelymore
confidenceinthescorecomparability.
2
Correlationcanbeusedasanindexthatisusedinevaluatingthepotentialforlinkingthescoresfromdifferenttests
thatareintendedtomeasuresimilarconstructsbutareconstructedaccordingtodifferentframeworks.
LinkingTOEFLiBT™ScorestoIELTS®Scores–AResearchReport
15
FutureResearch
Thescorecomparisonresultsfromthisresearchweregroundedintheempiricalinformationthatwas
obtainedfromthestudentswhotookbothIELTSandTOEFLiBTtests,andwerebasedonanappropriate
psychometricprocedurethatusedscoresfromthetwoteststoestablishtherelationshipsbetween thetwo
setsofscores.Asinanyotherscorecomparisonresearch,thecurrentresearchandthefindingsforthescore
comparisonbetweentheIELTSandTOEFLiBTtestswillbenefitfrommoreresearchinthefuture,especial ly
fromfutureresearchstudiesbasedonmorecontrolledsamplesintermsofsamplesize,
sampleselection,
datacollection,andstudentbackgroundinformation suchasnationality,nativelanguage,yearsoflearning
English,etc.Withalargerandbettercontrolledsample,thedatacollectionqualitywillbeimprovedanda
sufficientnumberofstudentsacrossallscoreswillbeprovidedtoenhancethescorecomparisonresults.
LinkingTOEFLiBT™ScorestoIELTS®Scores–AResearchReport
16
Reference
AmericanEducationalResearchAssociation,AmericanPsychologicalAssociation,&NationalCouncilon
MeasurementinEducation.(1999).Standardsforeducationalandpsychologicaltesting.Washington,DC:
AmericanEducationalResearchAssociation.
ETS.(2002).ETSStandardsforqualityandfairness.Princeton,NJ:ETS.
Holland,P.W.(2007).Aframeworkandhistoryforscorelinking.In
N.J.Dorans,M.Pommerich,&P.W.
Holland(Eds.),Linkingandaligningscoresandscales(pp.5‐30).NewYork,NY:Springer.
Holland,P.W.,&N.J.(2006).Linkingandequating.InR.L.Brennan(Ed.),Educationalmeasurement(4the
ed.,pp.187‐220).Westport,CT:PraegerPublishers.
InternationalLanguage
TestingAssociation.(June2007).ILTAGuidelineforPractice.RetrievedMay12,2010
from />Kolen,M.J.,&Brennan,R.L.(1995).Testequatingmethodsandpractices.NewYork,NY:Springer.
Pommerich,M.,Hanson,B.A.,Harris,D.J.,&Sconing,J.A.(2000).Issuesincreatingandreporting
concordanceresultsbasedonequipercentilemethods(ACTResearchRep.No.2000‐1).IowaCity,IA:ACT.
Pommerich,M.(2007).Concordance:TheGood,theBad,andtheUgly.InN.J.Dorans,M.Pommerich,&P.
W.Holland(Eds.),Linkingandaligningscoresandscales(pp.199‐216).NewYork,NY:Springer.
Wang,L.,Eignor,D.R.,&Enright,M.K.(2008).AFinalAnalysis.InC.A.Chapelle,M.K.Enright,&J.Jamieson
(Eds.),BuildingaValidityArgumentforanewTestofEnglishasaForeign Language(pp.259‐318).NewYork,
NY:Routledge.
Copyright © 2010 by Educational Testing Service. All rights reserved. The ETS logo, LISTENING. LEARNING. LEADING. and TOEFL are registered
trademarks of Educational Testing Service (ETS). TOEFL iBT is a trademark of ETS. IELTS is a registered trademark jointly owned by the British Council, IDP
Education PTY Ltd, and University of Cambridge Local Examinations Syndicate.