Carretta & Ree 1995 Near Identity of Cognitive Structures in Sex and Ethnic Groups (PUBLIC VERSION)

7/30/2019 Carretta & Ree 1995 Near Identity of Cognitive Structures in Sex and Ethnic Groups (PUBLIC VERSION)

1/18

AL/HR-TP-1997-0006 UNITEDSTATESAIRFORCE RESEARCHLABORATORY

NEARIDENTITY OFCOGNIT IVESTRUCTUREIN SEX AND ETHNIC GROUPS

ThomasR.CarrettaTrainingEffectivenessBranch

WarfighterTrainingResearchDivision

Malcolm J.Ree ForceDevelopmentBranchCognitionandPerformanceDivision

HUMAN EFFECT IVENESSD IRECTORATE7909LindberghDriveBrooksAFBTX78235-5352

August1998

Approvedforp u b l i c release; i s t r i b u t i o n unlimited.A IR FORCEMATERIELCOMMAND A IR FORCERESEARCHLABORATORYHUMANEFFECTIVENESS D IRECTORATE WARFIGHTER TRAIN ING RESEARCHDIVISION 6001SouthPowerRoad,Building55 8MesaAZ5206-0904


2/18

NOTICES

Thiseportspublishednhenterestofscientificandechnicalinformationexchangeanddoesnotconstituteapprovalordisapprovalofitsideasorfindings.

UsingGovernmentdrawings,specifications,orotherdatancludedinhisocumentornyurposetherhanovernment-relatedprocurementdoesnotinanywayobligatetheUSGovernment.hefactthattheGovernmentformulatedorsuppliedthedrawings,specifications,orotherdata,oesnotlicensetheholderoranyotherpersonorcorpo-ration,orconveyanyrightsorpermissiontomanufacture,use,orsellanypatentedinventionthat mayrelatetothem.

ThefficefPublicffairsaseviewedhisaper,ndtsreleasableto theNationalTechnicalInformationService,whereitwillbeavailable to thegeneralpublic,includingforeignnationals.

Thispaperhasbeenreviewedandisapprovedfor publication.

THOMASR.CARRETTAEE H .ANDREWS Project Scientist echnicalDirectorLYNN A.CARROLL ,Colonel,USAF Chief,WarfighterTrainingResearchDivision

Pleasenotify AFRL/HEOP,2509KennedyCircle,Building125,Rm158,BrooksAFB TX 78235-5118,ifyouraddresschanges,or ifyo uno longerwanttoreceiveou rtechnicalreports.ou may writeorcallthe STINFOOfficeatDSN 240-3877,or Commercial(210)[email protected].


3/18

REPORTDOCUMENTATIONPAGE Form ApprovedOMBNo.0704-0188Publicreportingburdenfor thiscollectionof informationisestimatedtoaverage hourp er response,including the timefor reviewing instructions,searchingexistingdatasources,gathenand maintainingthe dataneeded,and completingand reviewing the collectionofinformation.end comments regardingthisburdenestimateor any otheraspectofmis collectiono'information, includingsuggestions for reducing thisburden,toWashingtonHeadquarters Services.DireetorateforInformationOperationsandReports.21 5JeffereonDavisHignway.S1204.Arlington.VA 22202-4302. and tothe OfficeofManagementand Budget.PaperworkReductionProject(0704-0188).Washington.DC 205031.GENCYUSE ONLY(Leaveblank) 2 .EPORTDATE

August99 8 3.EPORTTYPEANDDATESCOVEREDInterim-July1996toSeptember19964.ITLEANDSUBTITLE 5.UNDINGNUMBERS NearIdentity ofCognitiveStructureinSexan dEthnicGroups6.UTHOR(S)ThomasR.CarrettaMalcolmJ.Re e

PE -2205FPR -1123TA -BlW U -017 .ERFORMINGORGANIZATIONNAME(S)ANDADDRESS(ES)Air ForceResearch LaboratoryHumanEffectivenessDirectorateWarfighterTrainingResearch Division,TrainingEffectivenessBranch 7909LindberghDriveBrooksAFB TX 78235-5352

8.ERFORMINGORGANIZATION

9.PONSORING/MONITORINGAGENCYNAME(S)AND ADDRESS(ES)A irForceResearchLaboratoryHumanEffectivenessDirectorateWarfighterTrainingResearch Division 6001SouthPowerRoad,Bldg55 8 MesaAZ5206-0904

10.PONSORING/MONITORINGAL/HR-TP-1997-0006

11 .UPPLEMENTARYNOTESAir ForceResearchLaboratoryTechnicalMonitor:rThomasR.Carretta,510)36-3622 12a .ISTRIBUTION/AVAILABILITY STATEMENTApprovedfor publicrelease;distributionunlimited.

12b .ISTRIBUTIONCODE

13 .BSTRACT(Maximum20 0words)ComparisonsofaptitudefactorstructuresweremadeinlargesamplesofyoungAmericanswho tookamultipleaptitude testbattery.Thefactormodelthatwa susedhadbeenstatisticallyconfirmed.Itincludedhierarchicalgandfivelower-orderfactrepresentingVerbal.Math,Spatial,TechnicalKnowledge,andPerceptualSpeed.The modelshowedgoodfitforboththesexes(maleandfemale)andethnicgroup White,Black,Hispanic,AsianAmerican,an dNativeAmerican)comparisons.Th eproportionsof totalandcommonvarianceaccountedfor bygan dthefivelower-orderfactorsweresimilarformen andwomenandforal lfiveethnicgroups.onfirmatoryfactortechniquesthatimposedstatisticalconstraintstestedif theloadinof thetestswerethesameforbothsexesandfor Whitesersuseachoftheother ethnicgroups. lthoughmanyof the teststhedifferencesintheloadingswerestatisticallysignificant,mostdifferencesweresmallinmagnitude(lessthan.05).The mnotabledifferencesoccurredforatestofaviationknowledge.Itha dalowerloadingformen thanforwomenonbothhierarchicalgandonthelower-orderTechnicalKnowledgefactor.Italsoha dalowerloadingforWhitesthanforBlacksanAsianAmericansonthelower-orderTechnicalKnowledgefactor.Correlationsbetweenfactorloadingsforthesexgroupsaforallpairsofethnicgroupswereveryhigh,approachingr=1.0.Regressionsbetweenpairsofgroupsindicatedthattherewnomeandifferenceinloadingsbetweenmalesandfemalesandamongtheethnicgroups.Thesefindings,alongwithpreviouresearch,presentaconsistentpictureofnearidentityofstructureofintellectforsexandethnicgroups.

14.UBJECTTERMS Academicaptitude;FOQT;irForceOfficerQualifyingTest:rmedServicesVocationalAptitudeBattery;SVAB;viationknowledge;thnicgroups;ntellect;Mathability;Perceptualspeed;exes;patialability;echnicalknowledge;erbalability17.SECURITYCLASSIFICATIONOFREPORT

Unclassified18.SECURITYCLASSIFICATIONOF THISPAGE

Unclassified19.SECURITYCLASSIFICATION OF ABSTRACT

UnclassifiedNSN7540-01-280-5500

15 .UMBEROF PAGES1 9

16.RICECODE 20 .IMITATIONFABSTRACT

UL StandardForm298(Rev2-89)rescribed by ANSIStdZ-39-18298-102 COMPUTERGENERATED


4/18

CONTENTS Page

SUMMAR YINTRODUCTION

StudiesofMean DifferencesStudiesofFactorialSimilarity

METHODParticipantsMeasuresProcedures

RESULTS- : : : . ': . . . . .DISCUSSIONREFERENCESFIGURES

FigureNo. 1tructuralModelof theAir ForceOfficerQualifyingTest

TABLESTableNo. Page

1ercentageofTotalVarianceAccountedfo r by FactorsWithintheGroups2actorLoadingsby Se xandEthnicityfo r theHierarchicalg3actorLoadingsby SexandEthnicityfortheResidualizedLower-OrderFactorsin


5/18

PREFACE

ThiseffortwasconductedunderWorkUnit1123-B1-01,PilotSelectionandClassificationSupport,whichisdedicated toresearchintotheselectionandclassificationofU.S.Air Forceaircrew personnel.W ewouldliketothankPaulJ.Riouxforhistechnicalassistanceinpreparing thedatafiles,andDrJosephL.Weeksfor hiscomments.Thispaperhasbeenpublished previouslyasCarretta,

T.R,& Ree,M.J.(1995),Nearidentityofcognitivestructureinsexandethnicgroups.Personalityand IndividualDifferences,7P,149-155.

SendwrittencorrespondencetoAFRL/HEAB,7909LindberghDrive,BrooksAFB,TX 78235-5352.endelectronic mailtoThomas.Carretta(S)williams.afmil.

IV


6/18

NEAR IDENTITY OFCOGNITIVE STRUCTURE IN SEX AND ETHNICGROUPS

SUMMARYComparisonsofaptitude factorstructuresweremadein largesamplesofyoungAmericans

whotookamultipleaptitudetestbattery.Thefactormodelthatwas usedhadbeenstatisticallyconfirmed.Itincludedhierarchicalgan dfivelower-orderfactorsrepresentingVerbal,Math,Spatial,TechnicalKnowledge,and PerceptualSpeed.The modelshowedgoodfitfo rboththesexes(malean dfemale)an dethnicgroup(White,Black,Hispanic,Asian-American,and Native-American)comparisons.Theproportionsof totaland commonvarianceaccounted fo rby gan dth efivelower-orderfactorsweresimilarfo rmenan dwomenan dfor al lfiveethnicgroups.Confirmatoryfactortechniquesthatimposedstatisticalconstraints testedif th eloadingsof th etestswereth esamefo rbothsexesan dfo rWhitesversuseachofth eotherethnicgroups.Althoughmanyof th etestsof thedifferences intheloadingswerestatistically significant,mostdifferences weresmallin magnitude (lessthan.05).Themostnotabledifferencesoccurredfo ratestofaviationknowledge.Itha dalowerloadingfo r men thanfo rwomenon bothhierarchicalgan don thelower-orderTechnicalKnowledgefactor.Italsohad alowerloadingfo rWhitesthanforBlacksan dAsian-Americans on thelower-orderTechnicalKnowledgefactor.Correlationsbetweenfactorloadingsfor thesexgroupsandfo rallpairsofethnicgroupswereveryhigh,approachingr=1.0.Regressionsbetweenpairs ofgroupsindicatedthattherewas no meandifferenceinloadingsbetweenmalesandfemalesandamongth eethnicgroups.Thesefindings,alongwithpreviousresearch,presentaconsistentpictureofnearidentityofstructureof intellectfo rsexandethnicgroups.

INTRODUCTION Therehavebeennumerousstudiesofgroupdifferencesinmeanaptitudescores.Willerman (1979)reviewedseveralstudiescomparingaveragedifferencesinfactorsfo rAmericanan dforeignracialandethnicgroups.Similarly,Loehlin,Lindzey,an dSpuehler(1975)providedasummaryofmeandifferencestudiesamonggroups.Whileinformative,thesestudiesdidno taddressdifferencesinfactorstructureamonggroups.Thecurrenteffortremediesthatdeficitby

providinganinvestigation of thestructureofcognitiveabilitiesforsexan dracial/ethnicgroups.StudiesofMeanDif ferences

Therehavebeenseveralstudiesofaptitudedifferencesbetweensexan dethnicgroups.For example,Hyde(1981)inameta-analysisofmale-femaledifferencesincognitiveabilities,reported mediandifferencesof.45sforverbal,.43sforquantitative,an d.45sfo rvisual-spatialability.Lesser,Fifer,andClark(1965)investigateddifferencesamongethnicgroupson abilities.TheycomparedmeanprofilesofscoresforAmericansidentifiedasBlack,Chinese,Jewish,an d


7/18

PuertoRican.Theyfoundnot onlymeandifferences,bu talsodifferencesinth epatternofmeanprofiles amongth egroups.For example,theBlackan dJewishgroupshadtheirhighestaverage scoreson th everbalfactor,whileth eChinesean dPuertoRicangroupshad theirlowestaveragescoreson thisfactor.Spaceperception,in contrast,was relatively lowerfortheBlackan dJewishgroupsan dhighestfo rth eChinesean dPuertoRicangroups.Morerecently,LynnandOwen(1994)studiedthreeethnicgroupsinSouthAfrica(Whites,Indians,and Blacks).ThemeanstandardizedWhite-Blackdifferencewasabout2.1c ran dth emeanstandardizedWhite-Indiandifferencewas1.0c.neachcaseth eWhitemeanwashigher.Itshouldbe notedthatlackofEnglishas afirstlanguagefortheBlackgroupmayhavecontributedto th eobservedWhite-Blackdifferenceon the tests.In Americansamples,similarpatternsofintercorrelationsamongaptitudemeasureshave

beenfoundforBlacksan dWhitesinstudiesby Nichols(1970)an dScarr-Salapatek (1971).Bothstudiesconcludedthatitwas unlikely thatunderlyingdimensionsofabilityvaried.hesedifferencesindimensionalitywouldgiverisetodifferentfactorstructuresthatcouldbe statisticallytested.StudiesofFactorialSimilarity

Severalstudiesoffactorialsimilarity havebeenconducted.Michael(1949)comparedthefactorstructureofWorldW ar IIU.S.ArmyAirForcepilotcandidateselectiontestsforBlacksandWhitesan dfoundvirtuallynodifferences.Humphreysan dTaber(1973)comparedfactorstructuresforhighan dlowsocio-economicstatusboysfromProjectTalentan dfoundno discernibledifferences.Loehlin,Lindzey,andSpuehler(1975)notedthatalthough th eethnicityof theparticipantsinProjectTalentwas no tspecifically identified,itcouldbe expectedthattheethniccompositionofthetw ogroupswoulddiffersignificantly.

DeFriesetal.(1974)comparedthestructureofabilityfortw oethnicgroupsasmeasuredby 15cognitivetests.TheirparticipantswereHawaiiansofeitherJapaneseor Europeanancestry.Theyconcludedthatthesamefourfactorswerepresentinbothgroupsan dthatthefactorloadingswerenearlyidenticalfo r thetw ogroups.

Thesepreviousstudiesallinvestigatedtherelationshipsamonggroupsforcommonfactors.Eachofthesecommonfactorswas mostlygeneralcognitiveability(Jensen,1980;Ree & Earles,1991).MoreinformativeanalyseswereconductedbyRee an dCarretta(1995)whoinvestigatedthecomparativestructureofanothermultipleaptitudebattery,theASVAB,acrossse xan dethnicgroupsusingahierarchicalmodelofability(seeVernon,1969)an dconducted statisticaltestsofthesimilarityofthefactorloadings.Theyfoundonlysmalldifferenceson th eVerbal/Mathan dSpeedfactors.However,therewerenotabledifferenceson theTechnicalKnowledgefactor.Mostof thedifferencesinthisfactorweredu etoasinglespecificknowledgetestofelectronicsinformation.


8/18

TheASVAB isrelatively factorpoor.The AirForceOfficerQualifyingTest(AFOQT)ha sagreaternumberof testsand agreaterdiversityoffactors.Skinneran dRee (1987)conductedexploratoryfactoranalysesoftheAFOQT usingasample of3,000USAirForceofficerapplicants.Thecorrelationsamongthe16testsrangedfrom.1 7to.77,withan average correlationof.43.asedon aprincipalfactorsanalysiswithcommunalities intheprincipaldiagonalan dKaiser-HarrisTypeIIobliquerotation,theydescribeda5-factorsolution:Verbal,Math,Spatial,TechnicalKnowledge(Aircrew Interest/Aptitude), and PerceptualSpeed.Thesefactorsshowedan averagecorrelationof .36witha rangeof.22to.50.Althoughth esample containedboth menand womenan dseveralethnicgroups,no group-specificanalyseswereconducted.

Earlesan dRee(1991)ina reanalysisof theSkinneran dRee (1987)data,demonstratedthehierarchicalstructure of th eAFOQT.hey foundan hierarchicalestimateofgeneralcognitiveability,g, basedon correlatedlower-orderfactors.swithSkinneran dRee(1987),theydidnot testthefitof themodelor performgroupspecificanalyses.

Morerecently,Carrettaan dRee (1996)usedconfirmatoryfactormethods(Bentler,1989)totestthestructure of th eAFOQTin theSkinner an dRee (1987)sample.Theyexaminedsevenstructuralmodelsincluding:gonly,tw omodelsbasedon th eoperationalAFOQTcomposites (notincludingtheAcademicAptitudecompositewhichwouldhavebeenredundantwiththeVerbalan dQuantitativecompositesandwouldhavecausedalineardependencyin th edata),tw omodelsbasedon Skinneran dRee (1987),andtw omodelsthatsimplifiedth eSkinneran dRee (1987)solutionbycombiningsomeofthelower-orderfactors.

Thetw omodelsbasedon operationalcompositeswerea4-factorsolution(Verbal,Quantitative,Pilot,an dNavigator-Technical) an dthesamewitha hierarchicalgadded.hetw oSkinneran dRee (1987)modelswere5-factors(Verbal,Math,Spatial,TechnicalKnowledge,andPerceptualSpeed)an d5-factorswithahierarchicalgadded.Model6wasasimplificationofSkinneran dRee (1987)withhierarchicalgthatcombinedtheSpatialan dPerceptualSpeedfactors,bu tkeptalltheothers.Model7combinedtheVerbalan dMathfactorsto produceaVernon-likemodel(Vernon,1969).Thisresultedinamodelwithhierarchicalgan dresidualizedlower-orderVerbal/Math(combinedfromSkinner& Ree,1987),Spatial/PerceptualSpeed(combinedfromSkinner& Ree,1987),andTechnicalKnowledgefactors.TheSkinneran dRee (1987)modelwithhierarchicalgandfive5lower-orderfactorswasselectedas itshowedbothgoodfitan dinterpretability.TheCarrettaan dRee(1996)analysesdidnotaddresspossiblegroup differencesassexan dethnicgroupcomparisonswereno tmade.

Thecurrentstudyinvestigated whetherthefactorstructureofabilitieswasth esamefo rmenandwomenan dforfiveethnicgroups.Ifthefactorsorloadingsoftestson thefactorsdifferamongor betweengroups,scoresbasedon thefactorscannothavethesamemeaninginresearchorinpracticalapplication.Differencesinfactorstructureorloadingscouldoccurasaresultofdifferencesinpersonalchoice(e.g.,courseenrollment,leisureinterests)an dopportunity.


9/18

Theconsequencesofgroupdifferencesin aptitudemeasurementon operationalscoresar eafunctionof th emagnitude of thedifferencesin factorloadingsan dth enumberoffactorsthatcontributeto operationalscores.Theseconsequences extendto usingth etestforresearchpurposes.Finally,differencesin factorloadingscannotnecessarilybe interpretedtoindicatetestbias.

TheCarrettaandRee(1996)factorstructurewasusedasth ebasisforth ecurrentstudyasithadbeenstatisticallyconfirmedandwasmeaningfully interpretable.Thesequalitiesmadeitsuperior toth eothermodels.Thegoalof thisstudywas todetermine if thefactorstructurewasthesamefo rm enan dwomenan dfor fiveethnicgroups.METHOD

ParticipantsTheparticipantswere269,968applicantsfo rUSAirForceofficercommissionstested

between1981an d1993.Theywere81.4% malean d78.7% White,12.2% Black,4.7%Hispanic,3.5% Asian-American,an d0.9% Native-American.Theiragesattime of testingrangedbetween18and 27 years.Theage,sex,an dethnicityof theparticipantswerecollectedby self-report. Measures

TheAFOQT isamultiple-aptitudetestbatteryusedfor th ecommissioningofU SAirForceofficersthroughOfficerTrainingSchool(OTS)an dtheReserveOfficerTrainingCorps(ROTC).AFOQT scoresareusedalongwithothermeasuresofaptitudean deducationalachievement (e.g.,collegegradepointaverage,typeofdegree,previousflyingexperience)toqualifyapplicantswho passmedicalan dphysicalrequirements forpilotan dnavigatortraining.TheAFOQTissimilartomanyothermultipleaptitude testssuchastheDifferentialAptitudeTestsGeneralAptitudeTestBattery,an dtheArmedServicesVocationalAptitudeBatterybecauseit'containsmeasuresofverbal,mathematical,andspatialfactors.Itdiffersfromth eothersinitsmeasurementofspecializedaviationknowledge.Allofthesetestsstemfrom th etheoreticalmodelofferedbyKelley(1928)andcontainseveralofthefactorshe positedsuchasspatial,quantitative,verbal,andspeed.

TheAFOQT consistsof16teststhatincludeVerbalAnalogies(VA),ArithmeticReasoning(AR),ReadingComprehension(RC),DataInterpretation (DI),WordKnowledge(WK),Math Knowledge(MK),MechanicalComprehension(MC),ElectricalMaze(EM),ScaleReading(SR),InstrumentComprehension(IC),BlockCounting(BC),TableReading(TR),Aviation Information(AI),RotatedBlocks(RB),GeneralScience(GS),an dHiddenFigures(HF).Amoredetaileddescriptionofthetestswasprovidedelsewhere(Carretta& Ree,1995;Skinner& Ree 1987).Itsreliabilityhasbeenexamined(Earles& Ree,1991;Skinner& Ree,1987)an dithas'beenvalidatedforpilotandnavigatortraining(Arth,Steuck,Sorrentino,& Burke,1990;


10/18

Carretta,1992;Carretta& Ree,1994,1995,Koonce,1982;Olea& Ree,1994).All16AFOQT testswereusedinthisstudy.Procedures

Thefactorstructuretestedacrossgroupsinthisstudywas confirmedby Carrettaan dRee (1996)an drepresenteda6-factorsolution (SeeFigure1).Thehierarchicalfactoris g(all16tests)an dth efivelower-orderfactorsareVerbal(VA,RC,W K,GS),Math(AR,DI,M K,SR),Spatial(MC,EM ,BC,RB,HF),TechnicalKnowledge(MC,IC ,AI,GS),an dPerceptualSpeed(DI,SR,BC,TR).Themodelisspecifiedin residualizedform(Schmid& Leiman,1957)withth eeffectsof th ehierarchicalfactorremovedfrom thefivelower-orderfactors.Maximum likelihoodestimation procedureswereused.

VA RC WK GS \R D l V IK SR MC EM BC RB HF IC AI TR Figure1.StructuralModeloftheAirForceOfficerQualifyingTest

Achi-squaretestwasconductedtodetermineif theloadingsforavariable on afactorfo rtw ogroupswerethesame(Bentler,1989).Todetermineifthefactorloadingsfo ratestwerethesamefortwogroups(e.g.,WhitesandBlacks),itisnecessarytoestimatethestructuralmodelparameterswiththeloadingsforbothgroupsconstrained tothesamevaluean dthentoestimatethemodelparameterswiththeconstraintremoved.Thisprocedureletseachgroup-specific(e.g.,


11/18

Whitesor Blacks)factorloadingestimatehave adifferentvalue.If th echi-squaretestissignificant,separateparameterestimatesforeachgroupprovideabetterfitand ar estatistically justifiable.In thisstudy,th eloadingsofmalesversusfemalesforeachteston eachof th esix'factorsweretested.Themostimportantethniccomparisonsarethosebetweenthemajorityan deachminoritygroup.Thetestsofequalityof th eloadingsweredoneby separatelycomparingthemajoritygroup(Whites)witheachof th efourminoritygroups:WhitesversusBlacks,WhitesversusHispanics,WhitesversusAsian-Americans,an dWhitesversusNative-Americans.TypeIerrorrateofp


12/18

Table1 .PercentofTotalVarianceAccountedforbv FactorsWithin theGroups

Percent ofTotalVariance

Group gerbalathpatial Technical PerceptualKnowledge SpeedMales 38.15 6.81 3.59 2.58 4.30 2.69Females 39.97 5.16 3.62 2.75 5.57 2.58 Whites 33.05 7.81 3.78 2.90 5.11 2.88 Blacks 35.42 7.94 3.92 3.20 6.41 2.79Hispanics 33.52 7.94 3.88 2.92 5.42 2.78 Asian-Americans 32.93 7.79 3.73 2.87 5.90 2.75Native-Americans 33.00 7.88 " 3.73 2.87 5.31 2.77

The loadingsof thetestson thefactorsfo rmalesan dfemalesareshownin Tables2(g )an d3(fivelower-order factors).Althoughmany of theunivariatetestsfo rdifferencesinfactorloadingsbetweenmalesandfemaleswerestatisticallysignificant(30of37),onlyfourpairsofloadingsdifferedbymorethan.05.Theseweretheloadingsfo rEM,IC ,an dAIon gan dforAIon thelower-orderTechnicalKnowledgefactor.AIshowedthelargestmale/femaledifference,.684versus.800on thislower-orderfactor.

Tables2(g )and3(fivelower-orderfactors)alsoshow theloadingsforth efiveethnic groups. lthoughmanyoftheunivariatetestsfordifferencesin loadingsbetweenWhitesan dtheotherethnicgroupswerestatisticallysignificant,few werelarger than.05inmagnitude.TheseincludedEM andICon gforWhitesversusBlacks,EMon thelower-orderSpatialfactorforWhitesversusBlacks,andAIon thelower-orderTechnicalKnowledgefactorforWhitesversusBlacksandWhitesversusAsian-Americans.Thelargestdifferences inloadingsoccurredfortheAIteston thelower-orderTechnicalKnowledgefactor(Whites,.695;Blacks,.795;Hispanics,.717;Asian-Americans,.772;Native-Americans,.709).Pearsoncorrelationsbetweenfactorloadingsforthesexgroupsan damongtheethnicgroupswereveryhigh,approachingr=1.0.hecorrelationoftheloadingsformalesan dfemalesforthehierarchicalgfactorwas.97andrangedfrom.97to.99for pairsof thefiveethnicgroups.Resultsfortheresidualizedlower-orderfactorsshowedthesamerange.


13/18

Table2.""""""s

Sex Ethnicity

Test Male Female White Black Hispanic Asian-Am. Native-Am.

VA .715 .712 .635 .618 .623 .619 .627AR - .753 .759 .730 .738 .738 .725 .724RC .615 .619 .521 .520 .513 .521 .516DI .710 .705 .667 .692 .682 .667 .673W K .518 .512 .432 .442 .446 .435 .441M K .723 .724 .707 .730 .710 .705 .705M C .606 .637 .556 .584 .568 .564 .564EM .457 .539 .448 .524 .477 .457 .456SR .737 .744 .706 .730 .713 .693 .705IC .604 .661 .572 .646 .589 .585 .585BC .660 .652 .615 .633 .607 .609 .610TR .534 .512 .463 .432 .446 .446 .446AI .375 .439 .326 .372 .336 .362 .332RB .611 .634 .588 .601 .574 .577 .578 GS .596 .628 .570 .612 .588 .583 .580HF .527 .531 .494 .485 .488 .484 .484

Regressionswerecomputedbetweenth esexgroupsan dbetweenallpairsof theethnicgroups.Alltheregressionsyieldedinterceptsofapproximately zero,indicatingthattherewas no meandifference inloadings.DISCUSSION

Fitstatisticsindicatedthatthedatagaveagoodfittothemodelfo rbothsexesandforth efiveethnicgroups.TheAFOQT issomewhatlessg-saturatedthanthemilitaryenlistmenttestbattery,theArmedServicesVocationalAptitudeBattery(Carretta& Ree,1996).Hierarchicalgaccountedforth egreatestamountofvarianceforbothsexesan dallethnicgroupsand differed littlebysexor ethnicity.Further,gaccountedfo rmoreof thetotalvariance thandidthesu mofR e e rcSn^SS") 8 " C O n S i S t e m W l t h r e S U l t S f r t h e r 3 P t i t U d e t C S t S ( J C n S e n ' ' 9 8 0 ;


14/18

Table3.FactorLoadingsbySexandEthnicity for theResidualizedLower-OrderFactors

Sex Ethnicity

Factor/Test Male Female White Black Hispanic Asian-Am Native-Am

VerbalVA- - .406 .404 .464 .451 .455 .452 .458 RC .573 .577 .630 .628 .621 .629 .624W K .736 .728 .764 .781 .788 .768 .778GS .234 .247 .232 .249 .239 .237 .236 MathAR .646 .651 .667 .675 .675 .662 .661DI .298 .296 .310 .321 .317 .310 .312M K .214 .214 .206 .213 .207 .206 .205SR .148 .149 .146 .151 .147 .143 .145SpatialMC .197 .207 .241 .253 .246 .244 .244EM .265 .313 .292 .342 .312 .299 .298BC .278 .274 .281 .289 .277 .278 .279RB .391 .405 .422 .432 .413 .415 .415HF .270 .272 .253 .248 .250 .247 .248TechnicalKnowledgeMC .301 .317 .371 .390 .379 .376 .376IC .284 .311 .337 .381 .347 .345 .345AI .684 .800 .695 .795 .717 .772 .709GS .222 .234 .288 .310 .298 .295 .294PerceptualSpeedDI .094 .093 .106 .110 .109 .106 .107SR .323 .326 .322 .333 .326 .317 .322BC .328 .324 .338 .348 .334 .335 .335TR .458 .440 .481 .449 .463 .464 .464


15/18

Amongtheresidualizedlower-orderfactors,thefollowingstandout.Thetw olargestsourcesof testvarianceforal lgroupsareVerbalan dTechnicalKnowledge.TheVerbalfactoraccountedformorevariancethandidth eTechnicalKnowledgefactorfo r al lgroupsexceptfemales,whereth ereversewas true.However,themostnotablefinding was thegreatsimilarity in percent-of-total-variance-accounted-forby eachof th efactorsfo ral lgroups.

Thesimilarityin th eproportionsof totalan dcommonvarianceforgforth esexan dethnicgroupssuggested thatthecombinationofunreliability an duniquevariancewassimilaracrossgroups.Also,th eproportionsofvarianceaccountedfo rby th elower-orderfactorsweresimilaracrossgroups.Thelargestsexand ethnicgroupdifferencesoccurredfortheTechnicalKnowledgefactorwhichaccountedfo rlesstotalan dcommon-varianceformales(4.3% an d7.4%>thanforfemales(5.6% and 9.3%)an dlessfo rWhites(5.1% and9.2%)thanforBlacks(6.4% an d10.7%),Hispanics(5.4% an d9.6%),Asian-Americans(5.9% and10.5%),and Native-Americans(5.3% an d9.6%).Itislikelythatthesedifferencescanbe attributedtoopportunityorindividualdecisionstolearnaviationprinciplesand flying.

Themostnotabledifferencesin factorloadingsoccurredforth eElectricalMaze(EM),InstrumentComprehension(IC),an dAviation Information(AI)tests.Allthreeof thesetestshad lowerloadingsfo rmalesthanfo rfemaleson gan dAIhadalowerloadingformalesonth elower-orderTechnicalKnowledgefactor.Thesedifferencesinloadingsaremostlikelyth econsequencesofpersonalchoicean dopportunity.Forexample,Reean dCarretta,1994)founddifferencesinloadingsfo rmalesandfemaleson technicalknowledgetestsonasimilarbatteryan dattributedthem todifferencesinhighschoolcourseenrollmentpatterns.

For theethnicgroupcomparisons,EM an dIC had lowerloadingson gfo rWhitesthanfo rBlacks.EM alsohad alowerloadingon thelower-orderSpatialfactorforWhitesthanforBlacks.Onthelower-orderTechnicalKnowledgefactor,AI had alowerloadingforWhitesthanforBlacksandAsian-Americans.Thereasonsforthesedifferencesar enotapparent.Thecorrelationalandregressionanalyseswereconsistentwiththefindingsof th ecomparisonsof thestructuralmodel.

Theresultsofthisstudyad dtothefindings(DeFriesetal.,1974;Michael,1949)thatthestructureofcognitiveabilityisnearlyidenticalacrosssexan dethnicgroups.Theseresults furtherexpandour understandingbyincludingadditionalethnicgroupsan dbyusingahierarchicalmodelofhumanabilities.

10


16/18

REFERENCESArth,T.0.,Steuck,K.W .,Sorrentino,C.T.,& Burke,E.F.(1990).AirForceOfficerQualifying Test(AFOQT):Predictorsofundergraduate pilottrainingandundergraduatenavigatortrainingsuccess(AFHRL-TP-89-52).BrooksAFB,TX:Manpoweran dPersonnelDivision,Air ForceHumanResourcesLaboratory.Bentler,P.M.(1989).EQSstructural equations programmanual .Los Angeles,CA,BMDPStatisticalSoftware.Bentler,P.M .(1990).Comparativefitindexesinstructuralmodels.PsychologicalBulletin

1-07 ,238-246.Browne,M.W .,& Cudeck,R.(1993).Alternativewaysofassessingmodelfit.In K.A.Bollen

an dJ.S.Lang(Eds.),Testingstructuralequationmodels .NewburyPark,CA:Sage.Carretta,T.R.(1992).Understanding th erelationsbetweenselectionfactorsan dpilottraining

performance:Doesth ecriterionmakeadifference?TheInternationaljournalofAviation Psychology,2,95-105.Carretta,T.R.,& Ree,M .J.(1994).Pilot-candidate selectionmethod:Sourcesofvalidity.Th e

InternationaljournalofAviation Psychology,4,103-117.Carretta,T.R.,& Ree,M.J.(1995).AirForceOfficerQualifying Testvalidityfo rpredictingpilottrainingperformance.JournalofBusinessandPsychology,9,379-388.Carretta,T.R.,& Ree,M.J.(1996).FactorstructureoftheAir ForceOfficerQualifyingTest:Analysisan dcomparison.MilitaryPsychology,8 ,29-42.DeFries,J.C,Vandenberg,S.G.,McClearn,G.E.,Kuse,A.R.,Wilson,J.R.,Ashton,G.C,&

Johnson,R.C.(1974).Nearidentityofcognitivestructureintw oethnicgroups.Science 183 ,338-339.Earles,J.A.,& Ree,M.J.(1991).AirForceOfficerQualifyingTest(AFOQT):Estimating th e

generalability component(AL-TP-1991-0039).BrooksAFB,TX:ManpowerandPersonnelResearchDivision,HumanResourcesDirectorate,ArmstrongLaboratory.

Humphreys,L.G.,& Taber,T.(1973).Abilityfactorsasafunctionofadvantagedan ddisadvantagedgroups.JournalofEducationalMeasurement, 0 ,107-115.Hyde,J.S.(1981).How largearecognitivegenderdifferences? Ameta-analysisusinga2an dd.American Psychologist,36,892-901.

1 1


17/18

Jensen,A.R.(1980).Biasin mentaltesting.NewYork:TheFreePress.Kelley,T.L.(1928).Crossroadsin th emindofman:Astudyofdifferentiatementalabilities.PaloAlto,CA:StanfordUniversityPress.Koonce,J.M ..(1982).Validation ofaproposedpilot-traineeselectionsystem.Aviation,Space,andEnvironmentalMedicine,53 ,1166-1169.Lesser,G.S.,Fifer,G.,& Clark,D.H.(1965).Mentalabilitiesofchildrenfromdifferentsocial-

classan dculturalgroups.Monographsofth e SocietyforResearch in ChildDevelopment ,

Loehlin,J.C,Lindzey,G.,& Spuehler,J.N.(1975).Racedifferencesin intelligence.San Francisco:Freeman.Lynn,R.,& Owen,K.(1994).Spearman'shypothesis an dtestscoredifferencesbetweenWhitesIndians,and BlacksinSouth Africa.Th e Journal ofGeneral Psychology,21 ,27-36.Marsh,H.W .,Balla,J.R.,& McDonald,R.P.(1988).Goodness-of-fitindexesin confirmatoryfactoranalysis:Theeffectofsamplesize.Psychological Bulletin, 0 3 ,391-410.Michael,W .B.(1949).Factoranalysesof testsan dcriteria:Acomparativestudyof tw oAAF

pilotpopulations.Psychological Monographs , 3 ,55-84.Nichols,P.L.(1970).Theeffectsofheredityan denvironmenton intelligencetestperformance

in4an d7yearwhitean dNegrosiblingpairs.Unpublisheddoctoraldissertation.UniversityofMinnesota,AnnArbor,M I:UniversityMicrofilms71-18,874.Olea,M .,& Ree,M.J.(1994).Predicting pilotan dnavigatorcriteria:Notmuchmorethan g

JournalofAppliedPsychology,79,845-851.Ree,M .J.,& Carretta,T.R.(1994).FactoranalysisoftheASVAB:Confirming aVernon-

hkestructure.Educational andPsychological Measurement ,54 ,459-463.Ree,M & Carretta,T.R.(1995).Groupdifferencesinaptitude factorstructure.Educational

andPsychologicalMeasurement ,55 ,268-277.Ree,M .J & Earles,J.A.(1991).Predictingtrainingsuccess:N ot muchmorethan?.PersonnelPsychology,44,321-332.Scarr-Salapatek,S.(1971).Race,socialclass,andIQ .Science, 7 4 ,1285-1295.

12


18/18

Schmid,J.,& Leiman,J.M .(1957).Thedevelopmentofhierarchicalfactorsolutions.Psychometrika,22,53-61.Skinner,J.,& Ree,M .J.(1987). AirForceOfficerQualifyingTest(AFOQT):Itemandfactoranalysis ofform O(AFHRL-TR-86-68).BrooksAFB,TX:Manpowerand PersonnelDivision,AirForceHumanResourcesLaboratory.Tucker,L.R.,& Lewis,C.(1973).A reliabilitycoefficientfo rmaximumlikelihoodfactoranalysis.Psychometrika,38 ,1-10.Vemon,P.E.(1969).Intelligenceandculturalenvironment.London:Methuen.Willerman,L.(1979).he psychologyofindividual andgroupdifferences.an FranciscoCA: W.H.Freeman

1 3

Carretta & Ree 1995 Near Identity of Cognitive Structures in Sex and Ethnic Groups (PUBLIC VERSION)

Documents