[PDF] duolingo demographics
[PDF] durée de vol paris los angeles
[PDF] durée du vol paris new york
[PDF] duree du voyage paris ile maurice
[PDF] durée trajet dijon paris tgv
[PDF] durée trajet orleans paris tgv
[PDF] duree trajet paris ile de re
[PDF] durée trajet paris nantes tgv
[PDF] duree trajet paris orleans tgv
[PDF] durée trajet paris rennes tgv
[PDF] durée vol paris ile maurice
[PDF] durée vol paris kuala lumpur
[PDF] durée vol paris los angeles
[PDF] durga puja permission form 2019
[PDF] dynapos am/at r
Academicachievementof studentsindual languageimmersion
JohannaWatzinger-Tharp
a ,KristinSwenson b andZacharyMayne c a DepartmentofLinguistics, Universityof Utah,SaltLake City,UT,USA; b
UtahDepartmentof HumanServices,
Universityof Utah,SaltLake City,UT,USA;
c UtahEducationPolicy Center,University ofUtah,Salt LakeCity,UT, USA
ABSTRACT
Thisarticlereports onastudy thatinvestigatedachievement inmathof thirdandfourth gradeduallanguage immersion(DLI)students, building onresearchthat hasdemonstrated theacademicachievement of studentswhoreceive contentinstructionpredominantly inthetarget language.Ourstudy expandsthescope andmethodologyof prior researchbyincluding one-wayprograms inthreelanguages (Chinese, FrenchandSpanish) andtwo-waySpanish-English programs;andby relyingonpropensity matchingto mitigatepossibleeffects ofschool andstudentdifferences. Inourthird gradestudy,we compared students'mathscoresin relationto theirEnglishLanguage Arts(ELA) achievementtocontrol forpre-existingdifferences betweenDLIand non-DLIstudents.DLI studentswho attainedthesame levelsinELA, and whoreceivedmath instructionina targetlanguage, performedatthe samelevelas theirnon-DLIpeers inthirdgrade mathtestsgiven in English.Forthe fourthgradestudy, wecomparedDLI studentstoa propensity-matchednon-DLI group.DLIstudents grewmorein math thantheircounterparts notinDLI. Theresults fromthisnatural experimentindicatethat studentsina DLIprogramthat hasbeen implementedstate-wide wereableto succeedacademicallyin math.
ARTICLEHISTORY
Received26August 2015
Accepted13July 2016
KEYWORDS
Duallanguageimmersion;
academicachievement; contentandlanguage integratedlearning;bilingual education;content-based instruction;immersion education
Introduction
Languageeducationscholars generallyconsider thecreationof Frenchimmersionprograms inthe
1960sin Canadaas thebeginningofimmersion educationinNorth America.
1
St.LambertElementary,
oneofthe first,andperhaps thebest-known Frenchimmersionsch ool,was openedinQuebec in
1965inrespon setodema ndsbyparentsofEngli sh-speakingchildren toprovidetheirstudents
theopportunity tobecomebilingualinFrenchand English.At St.Lambert,in asuburb ofMontreal, English-speakingchildrenlearnedsubject mattercontentalmostexclusivelyin Frenchin theearly grades,and thenshiftedtoward equallydistributed instructionin Frenchan dEnglish startingin secondor thirdgrade;thus, theprogramprovided whatbecame tobeknown as'total'and 'partial'immersion.Aroundthesame time,and inresponseto asimilargrassroots effortby parentsofEngli sh-speakingchildren,Cor alWayElementaryinMiami-DadeCounty establishedits English-Spanishbilingualprogram forbothnativeEnglish-speakingand nativeSpanish-spea kingstu- dents(Fortuneand Tedick2008). Thesetwoprograms inCanada andtheUS eachrepresent aparticularimmersion education programtyp e:One-wayprogramsthatserve studentswhospeak themajoritylanguage(e.g. Englishnative speakersmovingtowardproficiencyinChi neseorGerm an),an dtwo-way programs foraling uisticallyheterogeneous group,withstudents'movingin twodistinctdirect ionstoward thenativelangua geof theirlinguisticallydifferentpeers '(Fortuneand Tedick2008,6).To achieve ©2016Informa UKLimited,trading asTaylor& FrancisGroup
CONTACTJohannaWatzinger-Tharp j.tharp@utah.edu
INTERNATIONALJOURNAL OFBILINGUAL EDUCATIONANDBILINGUALISM, 2016 amutuallybenefic iallearningenvi ronment,two-wayprogramsrequirea balanced distributionofstu- dents'languagebackgrounds, usuallyoperationalized asroughlyaone-thirdto two-thirdsminimum ratio. Educationalpolicyin theUSdidnotpromotethe benefitsofbilin gualeducation formonoling ual Englishspeakersto thesameexten tasCanada (CohenandSwain 1976).Whileimmers ionprograms flourishedinCanadain theearly1970s, onlyafew USprograms followedtheCanadi anmodel. Cohen andSwai n(1976)point toaK-4Spanishprogramin Californiaand anelementary Frenchimmersion programinMaryl andas theonlyexamplesofbilingual educationtha ttargetedma joritygroup English-speakingstudents(49). Moreover,forminoritystuden ts,language-focus ededucationwas oftenlimited topull-outremedialEnglishas aSecondLa nguage(ESL) lessons,whichstigmatizestu- dentsnotonly aslinguistically, butalsoas academically deficient.Incontrast,two-w ayimmersion programs,whichstart edtogrow inthe1990s,explicitlyaim atthe integrationoflanguage minority studentswithmajo rityEnglishspeakers ,whilealsoaligningwith thegoals ofhighlevelsofprofi- ciencyandachievem entinacademi careas(Christian1996). Sincetheseearlybeginnings, asignificant numberof programsthatfallunder thebroadrubric of immersioneducation 2 havebeen establishedin theUS.Startingwithjust afewin the1970sand
1980s,immersion programsmorethandoubledin the1990sfrom119in 1991to278 in1999.By
2011,448immers ionprograms wererecordedbytheCenter forAppliedLinguistics(CAL) 2011,
butitshoul dbeemphasiz edthatthesenumbersarelikely tobemuch higher,sinceCALrelies exclu- sivelyonself-re porting.Moreove r,immersioneducationhasrecentlyexperienced significantgrowth acrosstheUS. Forexample, NorthCarolina'sdual language/immersion programshavegrownfrom ninetoover 90since2005. InUtah,DLI programsmore thandoubledfrom 58to 138between
2011and 2015.Indeed,a'majorityofstatesinthe UnitedStatesreporte dthat,during the2012-
2013school year,districts intheir statewereimplementingatleas tonedual languageprogram'
(Boyleetal. 2016,x).
Background
Researchrelated tobilingualism,bilingualandimmers ioneducationadd ressesa widerangeofinter- connectedissues,but ofparticularinterest forthepresent studyaretheoreti caland experimental studiesonthe cognitive benefitsofbilingual ism,andstudiesthatexa minetheeffects ofdual languageimmersioneduc ationon academicachievement,inparticular thoseconductedon a largescale. Bilingualandimmersioneducation aregrounded inasignificantbodyof researchthathas demon- stratedcognitiveadvantages ofbilinguals. Afteranumbe rof studieshadissuedwarning sabout potentiallydetrimental effectsofbilingualismon students'intellectualfunctioning,Pealand Lam- bert's(1962)semina lstudyshowedthatbal ancedbilinguals withequalproficiencyintwolanguages wereable tooutperformmonolingualpee rson variousverbalandnonve rbaltestsofintelligence. Sincethattime, manystudies havedemonstrated cognitiveadvantages ofbilinguals, which,in turn,mayenhance academicperformance, inparticular inmathandreading (Bialystok1991;Bialys- tokan dMajumder1998;Bialystok etal.2009;Bialystokand Craik2010;Esposito andBaker-Ward
2013;Foyand Mann2014).Fromtheir reviewofliteratu reonthe relationshipbetwee nbilingualism
andcognitiveproce sses,Bialystok etal.(2009)conclude thatbilingualspeakers, bynecessity,develop anenhanced abilitytoinhibitorsuppr essmaterialthatisnotreleva ntto ataskat hand,andto select informationthatis.Togeth erwithother functionssuchas monitoring,theprocessesof inhibitionand selectioncontri butetobilinguals'advantageinexecutivecontrol. Tobeable toestimate theeffects ofcognitivecorrelates associatedwithbilingualism,Adesope etal.( 2010)conducteda meta-analysisof 63studies thathadreportedonmeasur esof cognitive benefitssuch asattentional control,workingmemo ry,metalinguisticawareness,an dproblem solving(212). Thougheffectsizesvaried significantly,thissystematicreviewof awide rangeof
2J.WA TZINGER-THARPETAL.
studiesconfirmedthe positiveassociationof bilingualismwith cognitivebenefi ts,inparticular meta- linguisticandmetacognitiv eawareness, attentionalcontrol,andproblemsolving. Thepotential ofbilingualismtoimpactstudents'academicperfo rmancepositivelyismanifestedin thefocusof observationalimmers ioneducationresearch onacademicachievement(Swainand Lapkin1982;Thomas andCollier1997,2002;Collier andThomas2004).Forminori tylanguage stu- dents,programs withformalschoolingintheir nativelanguagehave beenfoundtobe mosteffect ive forensuring long-termacademicsuccess (LindholmandAclan1991;Thomasand Collier1997,2002; Alanis2000;Lindholm-Leary andHoward2008;Lindholm-Leary andBlock2010). IntheUS, researchonthe effectsand effectivenessof duallanguage immersionedu cation,as measuredbyacademic outcomes,is oftensituated inthecontextofthe persistentEnglish Learner (EL)academicach ievementgapin readingandmath.Someofthe mostsign ificantfindingson aca- demicachievement ofELscomefroma seriesofcomprehe nsivecross-site studieswhichexamineda varietyofschoolin gtypes,inc ludingbilingualandimmersion.Building onresearch thathadstartedin
1985,Thomas andCollier(2002)collected dataonlanguage minoritystuden tachievement fromfive
schooldistricts representingnorther nandsouthernregions,andurban andruralareasin theUS between1996 and2001.Lang uagesincludedEnglish, French,and Spanish.Theirstudyincludeda totalofeight majorprogram types:Four developmentalbilin gualorduallanguageimm ersionpro- grams(90/10and 50/50,andone-w ayand two-way 3 );twotransit ionalbilingual(90/10 and50/50); ESLthroughacademic content;and Englishmainstream. Two-waybilingualimmersion(90/10 and
50/50)and90/ 10one-way developmentalbilingualeducationwere foundtobe effectivelong-
term,asmeasured bylanguage minority studentsre achingandmaintainingthe50thperce ntilein bothlanguagesin allsubjects (7).In contrast,transitional, ESLandEnglishmainstreamprogram s wereshown nottobeeffectivein closingthe achievementgap forELs.Based ontheirconsistent results,Collierand Thomas (2004)conclude thatduallanguageschooling, bothone-and two-way, canclose70 -100%ofthe achievementgap forELsby grade5(15). Anum berofstudiesinCanadianand UScontextshave demonstratedthat immersionstudent s performatsimilar levels,or outperformtheirnon-imme rsionpeers intestedacademi ccontent areas,mostcommonly Englishliteracy andma th(Turnbull, Lapkin,andHart2001;Lazaruk 2007;Lind- holm-LearyandHerna ndez2011;Marian, Shook,andSchroeder2013;Padillaet al.2013;Steele etal.
2015).Inthei rprovince-wide studyinOntario,Turnbullandhis colleagues(2001)found thatthird
gradeFrenchimmersion students, enrolledin arangeofprogramtypes, performedaswell on reading,writin gandmathtestsasEnglish students.Immersion studentswho hadnotreceived anyEnglish instructionbythetimethey tookthetests slightlylaggedbehi ndinliteracy, butforall others,evenif theyhadonly startedEnglish instructionin thirdgrade, therewas noevidencethat thelevelof Englishinstruction hadaffected theirtestperformance. Inmathematics,studentsdid aswell astheirnon-immersion peers,eve nwhentheyhadreceived noinstructioninEnglish. Further- more,itdid notmatterwhethe rstudents tookthema thtests inEnglish,theirnativelanguage,but not thelanguage ofinstruction,orwhether theytookthe testsinFrench, theirsecond languageandthe languageofinstructi on. Bournot-TritesandReeder(2001)comp aredacohortofCanadianFrench immersionstudentswho received20%of theirmathem aticsinstruction inEnglish and80%inFrenchwith anothercohortthat wasinstructed equallyin EnglishandFrench(50%each) .Thenat iveEnglish-speakingstudents who received80%of theirmathematic sinstruction inFrench performedsignificantlybetterthanthe 50/
50cohort.
Theexisting bodyofresearchsuggeststhatstudents induallang uageimmersion programs are abletoachieve academically,and perhapseven outperformtheirpeersin monolingualprogram s. Lindholm-LearyandGenesee( 2014)concludefrom theirmeta-analysis ofinternational research onstudent outcomestha tduallanguageimmersioneducation ,inone-way,two-wayand indigenous programs,clearly confersacademi candlinguisticbenefits.The benefitshavebeenfoundfor majority andminor itylanguagespeakers,forstudent sfromarangeof ethnicbackgrounds andfor students whohavespecial educationalneeds (Lindholm-Learyand Genesee2014,175). INTERNATIONALJOURNALOFBILINGUALEDUCATI ONANDBILINGUALISM3 However,asSteeleetal. (2015)point out,themethodologyofstudies ontheeffect sofdual languageimmersion educationisoftennotsufficientl yrobusttocontrolfor pre-existingconditions orselection bias,whichlimi tstheabilit ytoiden tifycausalprogrameffects ortogeneralize beyonda study'sparticularcontext (9).In theirlarge-scale longitudinalstudy ofstudentsin gradeseight throughkindergarten inPortlandPublicSchools,Steeleand hercolleagues capitalized onlottery systemdata tocontrolfor unobserveddifferen cesbetween studentsrandomly assignedtodual languageimmersion inkindergartenandstudents notselectedfor theprogram(4).Thedual languageimmersion cohortinthisstudyoutperformed thenon-immersion cohortin reading,or EnglishLangua geArts(ELA),withtheir advantageequivalent toaboutsevenmonthsof instruction ingradefive, androughly oneacademicyear ingradeeight (21).Thepositive effectsonreading were observedforboth native speakersofEnglish andstudentsclassifiedasELs, andacrossthe different programsandlang uages(predom inantlyMandarinandJapaneseone- way,andSpanishtwo-way). Positiveimmersion effectswerenotfoundonmath orscience. However, itiswo rthnoting that targetlanguage instructioninthesecontentareas, fullyorpartially until gradefive,also didnotdis- advantagestudents (Steeleetal.2015,25). ThePortlandstudy, asoneof fewconductedwith alargerandomi zedsample andmultiple languages,provide sparticularlystrongevidence ofthepositiveeffectsof duallanguage immersion education.Italsounders corestheneed foradditional systematicresearchthattightlycontr olsfor selectionbias orpre-existing between-groupdifferenc es.Beforeweturntoour study,wenext provideanoverview oftheUtah duallang uageimmersioncontext.
Utah'sdual languageimmersioneducation model
Amajor developmentinimmersioneducation duringthepastfivetoseven yearsis theemerg enceof state-fundedinitiatives tohelpestablishimmersion programsin publicschool s.Until2009,Utah, muchlikeother states,hadcreated bilingualeducationprogr amswithoutsufficient fundingor infra- structurefor some30years (LeiteandCook 2015).Withthe passageofSenat eBill41 in2008,the Utah legislaturefunded theestablishment ofpublicelemen taryschoolDLIprograms inChinese, French, andSpanish, startinginfirstgra deor,forsome,inkindergar ten.Other states(e.g.Delawa reand
Georgia)havesince followedUtah'sexample.
4 Asof 2015-2016,andfollowi ngthe additionofPortuguesein2012 andGerman in2014,Utah had
111elementary, grades1-6,and27 secondary schools,grades7 -9,in5 differentlanguages; Arabic
andRussianprogram sare expectedtostartin 2017.Elementary DLIprogramsarelocatedacross the statein22 urbanandrural schooldistricts,out ofatotal of36dis trictswith elementarysch ools.Utah schoolsestablish aDLIprogrambysubmittingan applicationto theUtah StateBoardof Education requestingoneof twoprogramtypes, one-wayor two-way,andone ofthe fiveimmersionlanguages. Schoolsand districtsarefr eetouse theirow nprocedurestoenrollstudentsonce theirapplicationhas beenapproved. Whendemandexceedsava ilableslots, afewdistrictsusealottery system,butmost establishwaiting lists. Inadditionto one-wayprogramsin fivelanguages,Utah schoolshousetwo-w ayelementary Spanishprograms(30 outof57 in2015-2016),which alsoemploythe 50/50two-teacher model. Two-wayprograms requirethatatleasta thirdofthe enrolledstudent smustbe nativespeakers ofthetarget language(Spanish orEnglish), althougha1:1 ratioisdesired.One-wa yandtwo-way pro- gramssharethe maingoalsof duallanguage immersioneducation: Forall studentsto succeedaca- demically;toatta inhighlevels ofproficiencyintwolanguages;and, ultimately, toemerge fromtheir immersioneducation bilingualandbiliterate,andequipped withcrosscultura lunders tanding (Howard,Olague,and Rogers2003). Tomeet itssignificant demandfor qualifiedteachers,Utahrelieson bothdomesticandinter- nationalguestteach ers.AllDLI teachersmustbelicensed toteachin publicschools, anddomestic teachersmustalso holdaworld languageand aduallanguage immersionendorseme ntwhen hired,orobtai nthes ecredentialswithina specifiedtimeframe.UtahDLI teachersparticipatein
4J.WA TZINGER-THARPETAL.
mandatorypre-andin-service trainingand professionaldevelopme nttogetherwiththeir English classroomcounterparts. Inaddition,regularmeetings ofthestate-wi deadv isorycouncilserveto reinforcethe DLImodelwith school-anddistrict-l evelprincipals andadministrators. AllDLIprogr amsinUtah thatreceivelegislativefundingarerequ iredtoalign withthe 50/50two- teachermodel andtodemonstratefidelity to9progr amassurances forgrades1through6. 5
Inthis
model,students spendhalfoftheirschool daywithexclusiveinstruction inthe targetlanguage by oneteacherand theotherhalf inEnglishwith anotherteacher. Utahhas implementeduniform cur- riculathatare designedto promoteliteracy inbothlanguagesand tomeetestab lishedproficiency targetsforall fourskillsand foreachgra delevel.Students 'developmentofproficiency inthe targetlanguage ismeasuredwiththeACTF LAssessment ofPerformance ofProficiency inLanguages (AAPPL)test ingrades3-9,with thegoalof 80%ofDLI studentsmeeting theproficiency benchmarks foreachgrade level.Academiccontent curriculaandmater ialsforma th,social studiesan dscience havebeen translatedintoeachimmersionlangua ge,andare supportedby lessonplans. Theoriginal andthetranslate dversionsalign withthecommon(orstate) core,and DLIandnon-DL Istudentsalike havetomeet contentstandar ds.Math andallothercontentareas, includingsocialstudies and science,aretaught nearlyexclusivelyin thetargetlanguage infirstgrade throughthird grade.Stu - dentsreceiveabout 70minutesof dailymathinstruction, whichaccou ntsfor roughly20% ofthe schoolday,as shownin Figure1 .TheEngli shclassroom focusesonbuildingEnglishliteracy, with ELAaccounting forabout35%oftheschool day,whichtransl atesinto some140 minuteseach day.Reinforcement inEnglishofallconten tareasthat aretaughtinthetarget language(math, scienceandsocia lstudies)accounts forabout15%oftheday. Infourth andfifthgra de(seeFigure2),mathinstruction changesto60 minutesdailyin theEnglish classroom,and 30minutesin thetarget languagefocused onactivitiesand practicalapp licationto Figure1.Duallanguage immersioninstructional time:grades1 -3. INTERNATIONALJOURNALOFBILINGUALE DUCATIONANDBILINGUALISM 5 promoterichlanguage use.Targ etlanguage literacyinstructionincreasesfrom 15%ingradesone throughthreeto 25%(about90 minutesofdaily instruction)in gradesfour andfive. Asawhole, Utah'sDLI curriculumpromotesliteracyinstruction inEnglishand inthetarge t language,andits integrationwithcontent instruction.Italigns withthepremise that languagedevel- opmentisat thecore ofstudents 'academiclearningor,put differently,that 'everycontent lesson mustbea languagelessonas well'(Met2008,56).
Thestudy
Ourstu dysoughttodetermine theeffectof duallanguageimmersionona cademicachievement inmatha crossthr eedifferentduallanguageimmersionlanguages(Ch inese,French,and Spanish)and twoprogram types(one-wayandt wo-way).Itbuilds onpreviousr esearch, whichhas shownthatstudent sin duallanguageimme rsionsucceedacademicallywhen contentistaug htfullyorpre dominantlyinthetargetlan guage(e.g.Bou rnot-Tritesa nd
2013;Steeleetal.2015).I talsoe xpandsonthes copeofpriorresearchi ntwoway s:Thestud y
examinedtheachievement ofstude ntsenrolledinaDLIprogramthat hasbeenimplem ented state-wide,across threelanguagesandwithtwoprogram types(one-way andtwo-way);and itemployedpro pensitymatch ingofDLIandnon-DLI schools,a ndDLI andnon -DLIstudents tomi tigatepotentialeffectsofpr e-existingdifferencesat theschoolands tudentlevels.This researchisalsomotiv atedby theneedtode monstratetostakeholdersthatst udents whopar- ticipatein duallanguagei mmersion,andin arel ativelyrec entlyestablishedprogram,are per- formingatt hesame levelastheirno n-immersionp eersincontent areas,asm easuredby state-wideaccountabilityt esting . Figure2.Duallanguageimmersion instructionaltime: grades4-5.
6J.WA TZINGER-THARPETAL.
Tomeasureacad emicperformance ofstudentsinone-wayand two-wayUtahDLIprograms, we formulatedthefollowing researchquestions: (1)How dothirdgrade studentswhoare andarenot inUtahDLI programsperform inmath, relative totheirperformance inELA? (2)Howmuch growthin mathisobserved inthefourth gradeyearfor studentsin UtahDLIprogr ams comparedtosimi larstudents whoarenotinDLI? (3)Whateffect doesDLItarget language orprogramtyp e(one-way ortwoway) haveonthirdand fourthgrade students'academicperformancein math?
Methods
Weuse drigorousstatisticalmethods toensureanequitablecomparison ofstudentsparticipating in DLIwithstudents notinDLI. Inthethird-gra destudy,student sprovidedwithin- subjectcontrol groupsfor themselvesand wewereableto considera student'smathperformance relativeto his orherperformance inELA. Inthe fourth-gradestudy, weusedpropensitymatching.This method identifiedstudent pairsthatincludedtwostuden ts,oneDLI studentand onenon-DLIstudent, whoweresimilar academically anddemographic allyinthethirdgrade.Theanalysis thenfocused ondifferences ingrowthinmathfrom thethirdto thefourthgradeyear. Thepropensity matching approacheliminatedpre-existing differences indemographic characteristics,andtestscores inELA andmath, betweenDLIandnon-DLI studentspriortofourthgrade.
Samples
Twosampleswere usedin thisresearch: athirdgrade sample,whichconsisted ofthirdgrade stu- dentswhoatten dedDLIschool sinthe2011-2012schoolyear, anda fourthgrade sample,whichcon- sistedoffourth gradestudentswho werein DLIinthe 2012-2013school year,and theirpropensity- matchedpeers. Inbothsamp les,students fromDLIprograms wereidentifiedfromwithin26 schools, whichinc luded6Chinese,5French, 6one-waySpanish and9two-way Spanishprogr ams.
Thirdgradesample
Thethirdgrade samplewascompr isedofall studentsenroll edin thirdgradeinthe2011-2012school yearwho metthreeinclusion criteria:(1) studentswere enrolledina Utahpublicschoolthatoffered UtahmodelDLI programstothird gradestuden ts;(2)student sdidnot changeschools betweenfirst andthirdgrade s;and(3) studentsreceivedthirdgrade scoresfromUtah 'sstanda rdizedELAtestand
Utah'sstanda rdizedmathtests.
Atotalof 2524students metthe inclusioncriteria.A majorityofthesestudentsidentified aseither non-HispanicWhite(73%) orHispanicethnicity(20%),with 37%ofthe studentsqualifyi ngforfree or reducedpriceslunch (FRL)and7% qualifyingforEL servicesinthe thirdgrade. Thedemographic characteristicsofthisstudentsample arerepresentative ofstudentdemog raphicswithinthe state. Ofthesamp le,47.3% ofthestudents(N=1195)were enrolled inDLIprograms and52.7%ofthestu- dents(N=1329)were not.
Fourthgradesample
Thefourthgrade samplecomprisedstudents whowere enrolledina DLIprogram asfourth graders duringthe2012 -2013school year,andwho hadthirdgrade andfourthgrade ELAandmath scores onUtah'sstandardized achievementtests.Thefourthgrade studentswere pairedwith propensity- matchedstudentswho hadsimilardem ographicsandsimilar testscores. INTERNATIONALJOURNALOFBILINGUALEDUCATI ONANDBILINGUALISM7 Overall,2287studentswere inthefourth gradesample, including1148DLIstuden tsand1139 non-DLIpropensity-matched students.Consistentwithdemog raphicsinthestate,a majorityof thestudent sinthefourthgradesampl eidentified aseithernon-Hispa nicWhite(73%) orHispanic ethnicity(21%), with35%qualifyingforfreeor reducedpriced lunchand7% qualifyingfor ELservices inthefourth grade.
Academicoutcomes
Students'ELAandmath scoreson Utah'sstanda rdizedCriterionReferenceTests(CRTs) servedasthe academicoutcomes inthisstudy.Untilrecently, CRTswere administeredeach spring;theCRTs have beenreplacedby anew setofstand ardizedtestsaligne dtoa newsetof corestandards implemented inUtah.Scaled scoresonthe CRTsrangedfrom 130to190 points,witha meanof 160anda standard deviationof10. CRTscoreswe reusedto analyzemath scoresof thirdgradestudentsandcre ate StudentGrowthPercentage s(SGPs) toanalyzemathlearning duringthefourthgrade.SGPs were createdbybinning studentsinto 100quantiles basedontheirthirdgradetest results,and calculating thepercentilefor eachstudentwithin eachbinusing fourthgradetest results.For example,ifa studentwasin the45thpercentile inmath inthethird grade,tha tstudent 'sfourthgrade math scoreswerecompared toallother studentsin thestatewho werealso inthe45th percentilein thethirdgrade. Eachstudent'sfourthgra deSGPcoul drangefromthe1st tothe100th percentile.
Propensitymatching
Tofinda sampleof non-DLIstudents forcomparisonwithDLIstuden tsin thefourth gradestudy,we useda two-stepprocessfor propensityscore matching.In thefirststep, weidentifiedasimilarnon - DLIschoolfor eachDLIschool bymatchingon schoolsize,perce ntof studentsfrom traditionallymar- ginalizedracesor ethnicities,mobility-rate, percentofstudents whoqualifiedforfreeor reduced pricelunch, andperformance ofthirdand fourthgradestudentsprior tothestudy. Aftermatching eachDLI schoolwith anon-DLIschool,wema tchedDLIstudents fromwithin eachDLIschoolwith peersfromwithin thematched non-DLIschool .Student matchingwasconductedusing gender,eth- nicity,qualification forfreeandreducedlunch,qualification forspecial educationand ELservices,and thirdgrade CRTscores.Asthefocusof theanalysis wasmath learninginthe fourthgrade, anexact matchwasrequired, bydecile,for mathscores. Inotherwords, astudentwho scoredinthe 80th-89th decilein mathin thethirdgradewasma tchedtoanothe rstudentwho scoredin the80th-89thdecile onthemath testin thethirdgrade .Weuse dcalipermatching andrequired adifferen ceofnomore than.1on thecombinationof allpropensity matchingvariablein orderto pairstudents. Usingthis method,weobtained suitablematches for1139of the1148DLIstudents. Propensitymatchingwasassessedby comparingthe absolutestand ardizedbias (ASB)estimates ofDLI studentsandnon-DLIstuden tspriorto matchingandafter matching.Standardizedbiasis a measurethatis notinfluencedby samplesize and,thus, canbe usedtocompare matchedand unmatchedsamples(Stuart2010).Theuse ofASB removedthesign fordirection fromthestandar - dizedbiasresu ltsallowing foradirectcomparisonof themagnitudeofbia sand,importantl y,allowed formeaningful averagingofmagnitudes acrosscovariates.
Dataanalysis
Toanswerthe researchquestions relatedtothird grademath scores,datawereanalyzedwith multi- levelregression.To answertheresearc hquestions relatedtofourth gradetest scores,multiple regressionwas usedtocomp areDLIstuden tswith theirpropensity-matchedpeers,andmulti-level multipleregression wasusedtodetectpossible effectsoftarget language (i.e.Chinese,French, or Spanish)andprogr amtype (i.e.one-wayortwo-wayimmersion) onmathlearning.Mult i-level regressionsforboth thirdandfourth gradeanalyses includedstudent-l evelandschool-l evel
8J.WA TZINGER-THARPETAL.
models,whichwe reestimatedwith HLMsoftware(Raudenbush, Bryk,and Congdon2004).First,null models(i.e.model swith nopredictors)werefitto gainabaselin eestimate ofvarianceatthe student- levelandto determinewhat proportionof thestudent-levelvariancecoul dbeaccou ntedforatthe schoollevel. Next,means-as-outcomesmodelswithschool -levelpredictors (i.e.targetlanguageand programtype) wererun toestimatetheproportion ofvariance inschool-level averagesthatcouldbe accountedforby targetlanguage andprogramtyp e.Finally,full modelswererun toestimatethe effectofDLI participationonstuden tmathscores .quotesdbs_dbs17.pdfusesText_23
[PDF] Effects of Dual-Language Immersion Programs on Student - ERIC