Extendingtheexaminationoftheholvle-Shortform Toa Furtherdataset

ExtendingtheExaminationoftheHOlVlE-ShortForm toa FurtherDataset:

FactorAnalysesandExaminationof-InternalConsistency intheJOBSDescriptiveStudy

CarrieL.MarinerandMarthaJ.Zaslow

ChildTrends,Inc. MethodsWorkingPaper#98.3

ThereisaclearneedtoextendpsychometricanalysesoftheHOME-ShortFormto samplesbeyondtheNationalLongitudinalSurveyofYouth-ChildSupplementinordertoassure thatfindingsregardingthismeasurearenotsamplespecific.·Inthepresentworkingpaper,we extendtheexaminationofthefactorstructureandinternalconsistencyoftheearlychildhood versionoftheHOME-ShortFormtoa furtherdataset:theJOBSDescriptiveStudy. Thisisa sampleofmotherswhohadappliedfororwerereceivingwelfare,allwithachildofbetween about3and5yearsofage. FamiliesinthesamplewerefromtheFultonCounty,Georgiaarea, andnearlyallofthefamiliesinthesamplewereAfricanAmerican.

AnalysesrevealtheessentialsimilarityoffactorsunderlyingtheHOME-ShortFormin theDescriptiveStudysampleandthesubgroupofAfrican-Americanfamilieswithpreschoolers inthe1988NationalLongitudinalSurveyofYouth-ChildSupplement. However,internal consistencyofthetwooriginalHOME-ShortFormsubscales,EmotionalSupportandCognitive Stimuilation,islowerintheDescriptiveStudysample.

Extending theExamination oftheHOME-ShortForm toaFurtherDataset:

FactorAnalysesandExamination ofInternal Consistency

intheJOBSDescriptiveStudy

(:'arrieL.MarinerandMarthaJ.Zaslow1

MethodsWorkingPaper #98.3

Introduction

OfthemanyresearcherswhohaveusedtheHOMEShortForminanalyzingdatafromthe NLSY-CS,severalhavedemonstrateddissatisfactionwiththetwosubscalesprovidedbythe survey(Baker,Keck,Mott,Quinlan,1993)bycreatingtheirownconceptually-basedsubscales (Barratt.1991)orbyfactoranalyzingtheHOME-SFandusingfactor-basedsubscales

(MenaghanParcel,1991;ParcelMenaghan,1989;Quint,Polit,Bos,Cave,1994;Sugland etal.,1995). Thisismostltkelyduetothelackofconceptualspecificityorclarityofthetwo subscales,CognitiveSimulationandEmotionalSupport. Thesubscalescreatedbythese researchersthroughfactoranalyses,however,haveneverbeenvalidatedusingotherdatasets.2

TheJOBSDescriptiveStudy(seeMooreetal.,1995fordetailsregardingthepurposes

andproceduresofthestudy)usestheearlychildhoodformoftheHOME-SFinanalmost

1ThispaperwascompleledaspartoftheworkonNICHDGrantNo.ROlHD31056.ThedatasetispartoftheChild OutcomeStudyoftheNationalEvaluationofWelfare-to-Work Strategies.TheauthorsareextremelygratefultotheFoundation forChildDevelopment,theWilliamT.GrantFoundation,andtheGeorgeGundFoundationforthefundingfortheJOBS ObservationalStudy,andtotheU.S.DepartmentofHealthandHumanServicesandtheU.S.DepartmentofEducationfor fundingoftheNationalEvaluationofWelfare-to-WorkStrategies,includingtheChildOutcomescomponentofthatstudy, withinwhichtheJOBSObservationalStudyisembedded.TheauthorsarealsogratefultoKathrynToutforhelpfulfeedback andtoJulieA.Floryanforhelpinfinalizingthemanuscript.

2TheNLSY-CSandtheNewChanceDemonstration (Quintetal.,1994)areamongthefewpublisheddatasetswhich includetheShortFormoftheHOME,and,todate(tothebestofourknowledge)noonehaspublishedacomparisonoffactor analysesofHOME-SFdatafrom morethanonedataset.

identicalformattotheNLSY,andcanbeusedtoextendthepsychometricexaminationofthe

HOME-SF.

Method

SampleforthePresentAnalyses

Thesampleforthepresentanalysesconsistedofthe790familieswhoparticipatedinthe Descriptive Study. TheDescriptiveStudywascarriedoutasaspecialsubstudywithintheChild OutcomesStudyoftheNationalEvaluationofWelfare-to-WorkStrategies(NEWWS,see Hamiltoll;Brock,Farrell,Friedlander,andHarknett,1997). TheChildOutcomesStudyofthe NationalEvaluationofWelfare-to-workStrategiesaskswhetherthereareimpactsonpreschool­ agechildren'sdevelopmentaloutcomestwoandfiveyearsafterrandomassignmentoftheir motherswithintheevaluation. Allthemothersintheevaluationhadappliedfororwere receivingAidtoFamiliesWithDependentChildren(AFDC)atthestartoftheevaluation. The evaluationinvolvesacontrastofacontrolgroupandtwoexperimentalgroups. Thetwo experimental groupstakecontrastingprogrammaticapproachestoencouragingatransitionfrom welfaretowork:alaborforceattaclunentapproach,inwhichmothersareencouragedtomakea rapidtransitionintothelaborforce;andahumancapitaldevelopmentapproach,inwhich

mothersareencouragedfirsttopursuebasiceducationandjobskillstraining,inordertoenhance theirlongtermemploymentprospects.

TheDescriptiveStudy,as itsnameimplies,wascarriedoutsoonafterrandomassignment withtheaimofdescribingthewell-beingofthefamiliesandchildren,nearthestartofthe

evaluation. WhiletheChildOutcomesStudyoftheNEWWS wascarried outinthreestudysites

(Atlanta, Georgia; GrandRapids,Michigan; andRiverside, California),theDescriptiveStudy wascompletedonlyintheAtlantasite.

EachofthefamiliesintheDescriptiveStudysamplehadayoungestchildofbetween about3and5yearsatthetimeofenrollmentinthestudy. Thischildserved asthe"focal"child fortheDescriptiveStudy,i.e.,thechildfocuseduponintheinterviewmeasureswiththemother andgivenchildassessments.Wheretherewasmorethanonechildofbetween 3and5yearsin thefamily atthetimeofenrollment,onewasrandomly selected tobethefocalchild.

Nearly allofthemothersintheDescriptiveStudysample (96percent)wereAfrican­ American.TheDescriptiveStudysampledidnotincludemothers whowereteenagerswhen

theyenrolledinthestudy. Indeed, mostofthemothersinthesamplewerebetween 25and34at baseline.However 40percentofthemothersinthesamplehadbeen19oryounger atthebirth of theoldest childlivinginthehousehold. Manyofthemothersinthesample (61percent) had

completedhighschool,andafurther5percenthadobtained aGED.Twenty-sevenpercent ofthe mothershadonlyonebirthchildlivinginthehousehold,while38percent hadtwo. Only35 percent ofthemothers hadthreeormorebirthchildren livinginthehousehold. Approximately equalproportionsofthefocalchildreninthesamplearemales(48percent) andfemales (52 percent). Thirty-fourpercentofthechildreninthesamplewere3-year-oldsatbaseline, 42 percentwere4-year-olds,and23percentwere5-yearolds,whileatotalofninechildren were already 6atbaseline.

Wewillcontrasttheresultsofthepresentanalyseswithparallelanalysescompletedwith asubsampleofAfricanAmericanfamiliesinthe1988waveoftheNationalLongitudinalSurvey ofYouth-ChildSupplementwhohadapreschool-agechild. Theseresultsarereportedinsome detailbySuglandandcolleagues(1995),andbrieflysummarizedinMethodsWorkingPaper

#98.2. Table1notesselectedcharacteristicsofthetwosamples. Ascanbeseen,since approximately 96%ofthe JOBSDescriptiveStudysampleisAfrican-American,thissampleis similartotheAfrican-AmericansubsampleoftheNLSY-CSinrace/ethnicity. YettheJOBS

DescriptiveStudysampleisrelativelylesseducatedandlesslikelytobeemployedthanthe

African-American subsampleofthe 1988NLSY-CS.

Procedure

ThefamiliesintheDescriptiveStudywerevisitedintheirhomestobeinterviewed approximately3monthsafterenrollingintheevaluation. Theinterviewcoveredarangeof topics,includingparentingandthehomeenvironment(seealso,MethodsWorkingPaper#98.6), mothers'psychologicalwell-being,contactwithextendedfamilyandthechild's father,

emotionalsupport,educationalattainmentandfamilyeconomicself-sufficiency. Direct assessmentsofthechildren'sverbalability(thePPVT-R)andschoolreadiness(theCaldwell PreschoolInventory)werecarriedout,andmothersreportedonthechild's healthandsocial behavior.

Aspartoftheinterviewandhomevisit,motherswereaskedthematernalreportitemsof theearlychildhoodversionoftheHOME-ShortForm. Inaddition,basedontheirobservations ofthehomeenvironmentandmother-childinteractionduringthehomevisit(whichlastedabout

1Y2hours),interviewerscompletedtheobservation-basedratingsitemsoftheHOME-Short

Form.

Strategy ofAnalysis

Thispaperwillcomparethefactorstructureof theHOME-SFintheJOBSDescriptive StudySampleandtheAfrican-Americansubsampleofthe1988waveoftheNLSY-CS,toask whetherthefactorstructureoftheHO:I.\1E-SFintheNLSYdataisreplicatedintheJOBS DescriptiveStudysample. Specifically,wewillcanyoutPrincipalComponentsAnalysis,with varimaxrotation,usingunweighteddata. Varimaxrotationwaschosentosimplifytheseparation ofitemsontofactors,sincethistechniqueseldomgivesmorethanonehighloadingperitem. In addition,wewillexaminetheinternalconsistency(ascalculatedbyCronbach'salpha)ofthetwo originalHOME-SFsubscales,EmotionalSupportandCognitiveStimulation,withintheJOBS DescriptiveStudydataset,againaskingiffindingsfromtheNLSY-CSarereplicated.

Results

FactorAnalyses

TheresultsoffactoranalysisoftheAfrican-AmericansubsampleofNLSY-CSchildren aged3to6bySuglandandcolleagues(1995)aredisplayedinTable2, andtheresultsoffactor analysisoftheJOBSDescriptiveStudysamplearesummarizedinTable3. Whiletheorderof thefactorsissomewhatdifferent,itcanbeseenbylookingatthesetwotablesthatthefive factorsfromtheNLSY-CSdataarealsofoundintheJOBSDescriptiveStudydata.TheJOBS

DescriptiveStudydatashowtwoadditionalsubscales,fathercontactorinvolvement,andanother subscalewhichisdifficulttointerpretandhasalowalpha(.29). Theadditionofafathercontact

factorin theJOBSDescriptiveStudydatamaybearesultofthespecificsampleused. Mothers intheJOBSDescriptiveStudyweremostlylivingassinglemothersandwerereceivingorhad appliedforAFDC. Dailycontactwiththefatherislikelytohaveadifferentmeaningforthese childrenthanforthechildrenintheAfrican-AmericansubsampleoftheNLSY-CS,37%of whomhadmarriedmothers(seeTable1).Also,thequestionsintheHOME-SFencompassboth

fathersandfather-figures,sothemothersintheJOBSDescriptiveStudymayhavebeenreporting morefrequentlyonfather-figuresratherthanfathers,andthesefather-figure/child relationships couldbedifferentfromthosewithfathers.

Acomparisonofthealphasofthefactor-basedsubscalesalsodemonstratesthe similaritiesbetweenthefactorsinthesetwodatasets. AsseeninTables2and3, thealphasof thefivesimilarfactorsarequitesimilarinthetwodatasets,withtheexceptionofthePositive PaternalInvolvementsubscale. In theNLSY-CS,thissubscalehasanalphaof.76basedon3 items,whilethesubscalehasanalphaofonly.42basedon4itemsintheJOBSDescriptive Studydata.

InternalConsistency:EmotionalSupportandCognitiveStimulation

Tables4and5showalphasforthetwooriginalsubscalesoftheHOME,Emotional SupportandCognitiveStimulation. Ascanbeseeninthesetables,thealphasforthetwo originalHOME-SFsubscalesshowadifferenceinthesamedirectioninthetwodatasets,with EmotionalSupportapproximately.20lowerthanCognitiveStimulationinboth,andtheJOBS DescriptiveStudyalphasapproximately.20belowtlieNLSYalphasonbothsubscales. Itis

noteworthythatintheJOBSDescriptiveStudydataset,alphasforthetwooriginalsubscalesare

inthelowtomoderaterange(.55for CognitiveStimulationandonly.32forEmotionalSupport).

Discussion

TheessentialsimilarityofthefactorsunderlyingtheHOME-SFinthesetwodatasets confirmsthefactorstructurefoundusingtheNLSY-CSdata,andsupportsthecomparabilityof themeasureinthesetwostudies. LowalphasforthetwooriginalHOME-SFsubscalesandfor theMaternalInvolvementfactor-basedsubscaleintheJOBSDescriptiveStudyshouldbekeptin mindwhenutilizingdatafromeitherdataset,sinceitis impossibletobecertainofthe importanceofthelowalphasintheJOBSDescriptiveStudydatawithoutinformationfrommore

thantwodatasets. Theselowalphascouldmeanthatthereisaproblemwiththetwooriginal subscalesandtheMaternalInvolvementsubscaleof theHOME-SFthatisspecificto administrationintheJOBSDescriptiveStudysample,or thattheHOME-SFactuallyhaslower alphasingeneralthanareevidentintheNLSY-CS. Onlyby collectingHOME-SFdatafrom anothersamplecanthisbeclarified.

Theimplicationsofthiscomparativefactoranalysisarethreefold. First,theconfirmation ofthebasicfactorstructurefoundintheNLSY-CSlendssupporttothereliabilityofthemeasure. Second,thefactthateachfactoranalysisfoundatleastfivefactorstobemeaningfulsupportsthe judgementofthoseresearcherswhoshoweddissatisfactionwiththeoriginaltwosubscales providedbytheNLSY-CS. Third,in futuremeasuresdevelopment,weshouldbecertaintobase oursubscalesonclearlydefinedconstructstoavoidthenecessityofcontinuallyre-factoringand

re-scalingmeasures.

TableI: CharacteristicsofAfrican-American MothersinNLSY-CS1988, OneSiblingStudySample1 ComparedtoJOBSDescriptiveStudySample

CharacteristicPercentofPercentofNLSYJOBS Sample Sample

CurrentlySingle

Youngerthan25years

2:3BiologicalChildren2

Education:

12years

12years

12years

CurrentlyEmployed

African-American

63%81%

21%17%

36%35%

19%40%

51%53%

30%8%

49%17%

100%96%

1. InformationtakenfromTable1bofSugland,B.W.,Zaslow,M.,Smith,J.R.,Brooks-Gunn,J., Coates,D.,Blumenthal,C.,Moore, K.A.,Griffm,T.,Bradley,R.H.(1995),inJournalof FamilyIssues,16(5),632-663.

2.IntheNLSY-CS,thisnumberreferstothepercentageofhouseholdsinwhichthreeormore childrenhaveeverbeenborntothemother,whileintheJOBSDescriptiveStudysample,the childrenmustbelivinginthehousehold. Therefore,wewouldexpecttheactualpercentof householdswiththreeormorechildreneverbornintheJOBSDescriptiveStudysampletobe slightlyhigherthanisreportedhere.

Table2:FactorLoadingsandInternal ConsistencyofFactor-BasedSubscales: DatafromChildrenAges3through5years11months

BorntoAfrican-AmericanMothers intheNLSY-CS19881

FactorSubscale / Alpha / Nwnber ofltems / ItemsinSubscale / Factor
Loadings
Fl:School
Preparation / .82 / 3 / Numbers Alphabet Colors / .88
.75
.74
F2:Positive Maternal Involvement / .76 / 3 / MotherTalkstoChild
MotherAnswered ChildVerbally
PositiveToneofVoicetoChild / .75
.69
.69
F3:Stimulation / .62 / 6 / Mother ReadstoChild
TripstoMuseum
ChildTakenonOuting
ChildHas5+Tapes/Records
ChildOwns10+Books
Family gets1+MagazinesRegularly / .53
.49
.42
.41
.39
.35
F4:Physical
Environment / .68 / 4 / Clean
NotCluttered
Safe
NotDarkorMonotonous / .75
.71
.68
.45
FS:NoObserved
Punitiveness / .59 / 3 / MotherPhysically RestrictedChild
MotherSlappedChild
Spanked 0or1TimesInLastWeek / .58
.50
.49

1. InformationtakenfromTable3bandTable4bofSugland, B.W.,Zaslow, M.,Smith, J.R, Brooks-Gunn,J.,Coates,D.,Blwnenthal,C.,Moore,K.A.,Griffin,T.,Bradley, R.H.(1995), inJournal ofFamilyIssues.16(5),632663.

Table3: FactorLoadingsandInternal Consistency ofFactor-BasedSubscales: DatafromJOBSDescriptiveStudySample,N=7901

Factor Subscale / Alphaor Pearson Correlation / Number ofltems / ItemsinSubscale / Factor
Loadings
Fl:School
Preparation / .80 / 4 / Numbers Alphabet Colors Shapes / .89
.88
.85
.69
F2:Physical
Envirorunent / .62 / 4 / Clean
NotCluttered
Safe
NotDarkorMonotonous / ..82
.78
.60
.48
F3:Stimulation / .50 / 5 / ChildHas5+Tapes/Records
ChildOwns10+Books
Familygets1+MagazinesRegularly
TripstoMuseum
MotherReadstoChild / .63
.59
.54
.47
.45
F4:Positive Maternal Involvement / .42 / 4 / MotherShowsAffection
MotherTalkstoChild
PositiveToneofVoicetoChild
Introduces ChildtoInterviewer / .67
.65
.58
.38
F5:Father
Contact / .56 / 2 / ChildSeesFatherDaily
ChildEatswithBothParentsDaily / .81
.81
F6:NoObserved
Punitiveness / .53 / 2 / MotherPhysically Restricted Child
MotherSlapped Child / .78
.75
F7:Hitback; Spanking,TV / .29 / 3 / Spanked0or1TimesInLastWeek MotherWouldNotHitChildBack Nwnber ofHoursTVonPerWeek / .66
.55
.54

1. Pearsoncorrelationsareprovidedwhenonly2itemsarepresentinafactor-basedsubscale.

Note:Twoitems,ChoiceofFoodandChildGoesonOutings,didnotloadonariyfactor higher than.35,andweretherefore notincluded inthetable.

Table4: FactorAnalysisandInternal Consistency ofHOME-SFSubscales: DatafromJOBSDescriptive StudySample,N=790

Subscale / Alpha / NumberofItems
Cognitive
Stimulation / .55 / 14
Emotional
Support / .32 / 12

Table5: FactorAnalysisand InternalConsistencyofHOME-SFSubscales: DatafromChildrenAges3through5years11months

BorntoAfrican-AmericanMothersintheNLSY-CS19881

Subscale / Alpha / NumberofItems
Cognitive
Stimulation / .72 / 14
Emotional
Support / .52 / 12

1. InformationtakenfromTable4bofSugland,B.W.,Zaslow,M.,Smith,J.R.,Brooks-Gunn, J., Coates,D.,Blumenthal,C.,Moore,K.A., Griffin,T.,Bradley,R.H.(1995);inJournalof FamilyIssues.16(5),632-663.

References

Baker,P.C.,Keck,C.K.,Mott,F.L.Quinlan,S.V.(1993). TheNLSYchildhandbook. revisededition:Aguidetothe1986-1990NationalLongitudinalSurveyofYouthchildda Columbus,OH:CenterforHumanResourceResearch.

Barratt,M.S. (1991). School-ageoffspringofadolescentmothers:Environmentsand outcomes. Family Relations.40,442-447.

Hamilton,G.,Brock,T.,Farrell,M.,Friedlander,D.,Harknett,K.(1997).National EvaluationofWelfare-to-WorkStrategies. NewYork:ManpowerDemonstrationResearch Corporation.

Menaghan,E.G.,Parcel,T.L. (1991). Determiningchildren'shomeenvironments: Theimpactofmaternalcharacteristicsandcurrentoccupationalandfamilyconditions. Journal ofMarriageandtheFamily,53,417-431.

Moore,K.A.,Zaslow,M.J.,Coiro,M.J.,Miller,S.M.,Magenheim,E.B.(1995). How

wellaretheyfaring? AFDCfamilieswithpreschool-agedchildreninAtlantaattheoutsetoftheJOBSevaluation.Washington,DC:U.S.DepartmentofHealthandHumanServices,U.S. DepartmentofEducation.

Parcel,T.L.,Menaghan,E.G. (1989). ChildhomeenvironmentasamediatingconstructbetweenSESandchildoutcomes. Columbus,Ohio:TheOhioStateUniversity, DepartmentofSociology. _

Quint,J.C.,Polit,D.F.,Bos,H.Cave,G.(1994). NewChance:Interimfindingsona comprehensiveJ2rogramfordisadvantaged·youngmothersandtheirchildren. NewYork,NY:

ManpowerDemonstrationResearchCorporation.

Sugland,B.W.,Zaslow,M.,Smith,J.R.,Brooks-Gunn,J., Coates,D.,Blumenthal,C., Moore,K.A.,Griffin,T.,Bradley,R.H. (1995). TheearlychildhoodHOMEinventoryand HOME-shortformindifferingracial/ethnicgroups. Aretheredifferencesinunderlying

structure,internalconsistencyofsubscales,andpatternsofprediction? JownalofFamilyIssues,

632-663.