Citation
Reading Performance and High-Stakes Statewide Assessment in a Juvenile Corrections Facility

Material Information

Title:
Reading Performance and High-Stakes Statewide Assessment in a Juvenile Corrections Facility
Creator:
Gaddis, Justin G
Place of Publication:
[Gainesville, Fla.]
Florida
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (15 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
School Psychology
Special Education, School Psychology and Early Childhood Studies
Committee Chair:
KRANZLER,JOHN H
Committee Co-Chair:
GAGNON,JOSEPH
Committee Members:
BEAULIEU,DIANA JOYCE
MILLER,DAVID
Graduation Date:
8/9/2014

Subjects

Subjects / Keywords:
Academic achievement ( jstor )
Criminal punishment ( jstor )
Disabilities ( jstor )
Grade levels ( jstor )
Intelligence quotient ( jstor )
Modeling ( jstor )
Open reading frames ( jstor )
Reading comprehension ( jstor )
Special education ( jstor )
Special needs students ( jstor )
Special Education, School Psychology and Early Childhood Studies -- Dissertations, Academic -- UF
assessment -- delinquent -- ed -- education -- fcat -- juvenile -- reading -- sld -- special -- statewide
City of Tallahassee ( local )
Genre:
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
born-digital ( sobekcm )
Electronic Thesis or Dissertation
School Psychology thesis, Ph.D.

Notes

Abstract:
Recent legislation has ushered in an era of unprecedented accountability that has produced a paradigm shift within the field of education. Highlighting the push for accountability within schools is the reliance on student statewide assessment performance as a means to inform high-stakes decision making. Despite increased efforts by education researchers to establish links between reading performance and statewide assessment performance, extant literature within juvenile corrections in regard to statewide assessment as well as the overall reading abilities possessed by these youth remains in its infancy. The educational deficits of delinquent youth, their proclivity to recidivate, and the overrepresentation of marginalized minority and disability groups within juvenile correctional facilities, all provide sufficient rationale for further inquiry into reading skills of these youth. Further, studying the reading performance of incarcerated youth is of particular importance as the juvenile correctional system is often the last chance that these students have to be successful in school. Using a subset of participant data from Project LIBERATE, a randomized controlled trial federally funded by the U.S. Department of Education, Institute of Education Sciences (IES) (R324A080006), the present study sought to determine which specific academic and reading skills are most predictive of success on the reading section of the FCAT 2.0 for individuals in a juvenile corrections facility. Linear regression modeling, stepwise multiple regression model-fitting procedures utilizing backward elimination (based on bias-corrected Akaike Information Criterion [AICc]), and predictive discriminant analysis (PDA) were employed to address study questions. Incarcerated students within the study presented with significant intellectual and academic deficits. Deficits were prevalent across all academic and cognitive measures, as well as the FCAT 2.0 Reading. Further, incarcerated special education students performed significantly worse across all measures compared to their non-disabled incarcerated peers. IQ, Sight Word Efficiency (SWE), Reading Comprehension (RC), and Receptive Vocabulary (RV) were all statistically significant predictor variables. Overall, IQ proved to be the single most important predictor of FCAT 2.0 Reading success. Last, results indicated that the Florida Assessments for Instruction in Reading (FAIR) served as an adequate predictor of FCAT 2.0 Reading proficiency in regard to hit rate, sensitivity, and specificity. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (Ph.D.)--University of Florida, 2014.
Local:
Adviser: KRANZLER,JOHN H.
Local:
Co-adviser: GAGNON,JOSEPH.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2015-08-31
Statement of Responsibility:
by Justin G Gaddis.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Embargo Date:
8/31/2015
Resource Identifier:
969976926 ( OCLC )
Classification:
LD1780 2014 ( lcc )

Downloads

This item has the following downloads:


Full Text

PAGE 1

Journal of Special Education Leadership The Journal of the Council of Administrators of Special Education A Division of the Council for Exceptional Children IDEA-RelatedProfessionalDevelopmentin JuvenileCorrectionsSchools JosephCalvinGagnon,Ph.D., MaryAnneSteinberg,Ph.D.,and JeanCrockett,Ph.D. SchoolofSpecialEducation,SchoolPsychology, andEarlyChildhoodStudies,UniversityofFlorida KristinM.Murphy,M.S.,Ed.M. DepartmentofCurriculumandInstruction, UniversityofMassachusetts,Boston JustinGaddis,M.Ed. SarahA.ReedChildrenÂ’sCenter,Erie, Pennsylvania

PAGE 3

IDEA-RelatedProfessionalDevelopmentin JuvenileCorrectionsSchools N JosephCalvinGagnon,Ph.D., MaryAnneSteinberg,Ph.D.,and JeanCrockett,Ph.D.SchoolofSpecialEducation,SchoolPsychology, andEarlyChildhoodStudies,UniversityofFloridaKristinM.Murphy,M.S.,Ed.M.DepartmentofCurriculumandInstruction, UniversityofMassachusetts,BostonJustinGaddis,M.Ed.SarahA.ReedChildren’sCenter,Erie, Pennsylvania N Juvenilecorrections(JC)schoolshavearecordofnoncompliancewithrequirementssetforthbythe IndividualswithDisabilitiesEducationAct(IDEA)regulations(IDEA,2006),resultinginlengthyandcostly litigation. N Atleast56lawsuitsagainstJCschoolshavefocusedonvariousaspectsofthesixprinciplesofIDEA:(a)zero reject/childfind;(b)nondiscriminatorytesting;(c)individualeducationprogram(IEP);(d)leastrestrictive environment;(e)proceduraldueprocess;and(f)parentparticipation. N PossessionofIDEA-relatedknowledgebyJCschoolleadershipisessentialforimplementationofIDEA policyattheschoollevel.However,schoolleaderstypicallylackacomprehensiveunderstandingofIDEA requirements. N ItisessentialthatJCschoolandfacilityleadershipteamsgainknowledgeofIDEAviacomprehensive professionaldevelopment(PD).ThePDshouldalsoprovidetheguidancenecessaryforJCleadershipteams todevelopandimplementclearactionplansfocusingonimplementingIDEArequirements.OnlinePDisa promisingdeliverymethodworthfurtherinvestigationforthissetting.N EducationforStudentsWith DisabilitiesinJuvenileCorrections Schools:PastandPresentI ncarceratedyouthareamongtheleastacademically andbehaviorallycompetentstudentsintheUnited States.Inspiteofjuvenilejusticereformefforts, includingstateandfederalguaranteesofappropriate education,educationalservicesinjuvenile corrections(JC)schools,especiallyforyouthwith disabilities,arelacking(Houchins,Jolivette,Shippen, &Lambert,2010;Houchins,Shippen,&Jolivette, 2006).Thefailureofstatesandlocaljurisdictions toimplementtheIndividualswithDisabilities EducationAct(IDEA)2004statuteand2006 regulationsinJCisasignificantnationalpolicy concern(Gagnon,2010).Furthermore,JCschools haveadisturbingrecordofnoncompliancewith federalspecialeducationrequirementsdue,inpart, toinadequateoversightandenforcementatthelocal educationagency(LEA)andstatelevels(Gagnon& Barber,2010). Throughoutthepast40years,legislationand litigationhaveshapedtheAmericaneducation system,particularlyforstudentswithdisabilities (West&Schaefer-Whitby,2008).Forexample,the passageoftheCivilRightsofInstitutionalized PersonsActbyCongressin1980underscoredthe federalcommitmenttoprotectingbasicrights, includingtheeducationofstudentswithdisabilities inyouthprisonsandotherinstitutions.However,in JC,litigationrelatedtocomplianceratherthanthe actualfederalregulationsservesastheprimaryforce forshapingeducationservicesforyouthwithand withoutdisabilities(Gagnon,2010;Mathur& Schoenfeld,2010).NearlyallsignificantreformofJC educationservicesforyouthwithdisabilitieshave occurredfollowinglengthyandcostlylitigation; oftenthesereformsfollowedtheinvestigationof complaintsbytheU.S.DepartmentofJusticethrough N JournalofSpecialEducationLeadership26(2) N September201393 N

PAGE 4

theCivilRightsofInstitutionalizedPersonsAct (Gagnon,Barber,VanLoan,&Leone,2009; Houchins,Puckett-Patterson,Crosby,Shippen,& Jolivette,2009). Inadequatestaffdevelopmentisonemajor contributortothepoorstateofeducationservicesfor youthwithdisabilitiesinJC(Gagnon,Houchins,& Murphy,2012).TheisolationofJCprofessionalsisa long-standingconcernthatremainsproblematic.In manyjurisdictionsJCeducatorsareisolatedfrom LEAs,regionaleducationagencies,andstate departmentsofeducationthattypicallyprovide staffdevelopmentopportunitiesforteachersand administrators(Leone&Cutting,2004).Furthermore, theuniqueattributesofJCschoolsoftenpresent administrativeandinstructionalchallengesnotfound incomprehensivemiddleandsecondaryschools (Gagnonetal.,2012;Mathur,GrillerClark,& Schoenfeld,2009;Mathur&Schoenfeld,2010). Nationalorganizationshavecontributed somewhattoJCcompliancewithIDEA.Specifically, in2004theCorrectionalEducationAssociation reviseditsstandardsforeducationprogramsinboth juvenileandadultcorrectionalfacilities.Although severalofthesestandardsaddressedspecial educationandareperiodicallyincludedin discussionsatconferencesandseminars,the standardsarebroadanddonotprovidesufficient detailandrelevantpracticalinformationconcerning implementationofIDEAinJC.TheJohnD.and CatherineT.MacArthurFoundation’s(2004) Juvenile CourtTrainingCurriculum alsoprovidesacurriculum forcourtjudges,defenseattorneys,prosecutors,and probationstaffrepresentingyouthwithdisabilitiesin delinquencyproceedings.However,theSpecial EducationandDisabilityRightsmoduleprovides onlygeneralinformation,ratherthancomprehensive coverageofIDEA(Leone,Zablocki,Wilson,Mulcahy, &Krezmien,2009).Asof2004(Leone&Cutting, 2004),therewerenocomprehensiveandsystematic staffdevelopmentopportunitiesavailablethat focusedonIDEAforschooladministratorsandkey educationalandfacilitypersonnelinJC.Sincethat time,fewpreserviceandnoknowncomprehensive IDEA-focusedstaffdevelopmentprogramshavebeen developedspecifictotheJCschoolsetting(Mathur etal.,2009). SeriousconcernsexistthatJCadministrators andkeyschool/facilitypersonnelareuninformed, arecommonlynoncompliantregardingIDEA regulations,andmayfeelisolatedandill-equipped toimplementschool-levelchangesinpolicyand practicesrelatedtoIDEAcompliance(Gagnonetal., 2009;Gagnon,Haydon,&Maccini,2010).Failure toadheretoIDEAregulationsareparticularly problematic,giventhereareaboutfourtimesas manyyouthwithdisabilitiesinJCthaninregular publicschools(Gagnonetal.,2009).Inappropriate educationandspecialeducationservices undoubtedlycontributetothefactthatfew incarceratedyouthreturn,stayinschool,andearn adiplomauponexit(GrillerClark,Rutherford,& Quinn,2004).Failureofyouthtoearnahighschool diplomanotonlyaffectsrecidivismrates,italso increasesthelikelihoodthatyouthwillnotholdfulltimeemploymentandsubsequentlyliveinpoverty (Harlow,2003)........................................... Inappropriateeducationandspecialeducation servicesundoubtedlycontributetothefactthat fewincarceratedyouthreturn,stayinschool,and earnadiplomauponexit(GrillerClark,Rutherford, &Quinn,2004).Assuch,thegoalofthisarticleistoidentify criticalareasofneedandtosuggestaplanfor improvingknowledgeofIDEA,aswellastoprovide informationandrecommendationsthatcould facilitateaninstitutionalreviewandrevisionofJC school-levelpolicyandpractice.Ourpurpose, therefore,istoidentifyanddiscuss(a)concerns relatedtoIDEAregulations;(b)issuesrelatedto transferofadministratorandkeypersonnel knowledgeofIDEAtoschool-levelpolicyand practice;(c)keycomponentsofeffectiveprofessional development(PD)inJC;and(d)thepotentialfor onlinePD.ConcernsRelatedto IDEARegulationsOurfirstgoalistoidentifyspecificIDEAprovisions thatarecommonlyviolatedinJC.Althoughwewill notaddresseveryprovisionwithinIDEA,ourgoalis tofocusonkeyareasthatareofparticularconcern withinthissetting.WebaseourdiscussiononBoyle andWeishaar’s(2001)sixprinciplesthatserveasthe N IDEA-RelatedProfessionalDevelopmentinJuvenileCorrectionsSchoolsN N 94 JournalofSpecialEducationLeadership26(2) N September2013N

PAGE 5

Table1: KeyIDEAprinciples,JCissues,andIDEAregulations IDEAPrinciple JCIssue IDEARegulation Zeroreject Additionalproceduresfor identifyingchildrenwith specificlearningdisabilities (LD) JCschoolleadersneedtounderstandadditionalproceduralrequirementsforidentifying youthwithLDinIDEA(34C.F.R.71 1 300.307). Nondiscriminatory testing Evaluationsand reevaluations AStateeducationalagency,otherStateagency,orlocaleducationalagencyshallconduct afullandindividualinitialevaluationbeforetheinitialprovisionofspecialeducationand relatedservicestoachildwithadisability(34C.F.R.71 1 614(a)(1)(A)(B)) Alocaleducationalagencyshallensurethatareevaluationofeachchildwithadisabilityis conductedifthelocaleducationalagencydeterminesthattheeducationalorrelatedservices needs,includingimprovedacademicachievementandfunctionalperformance,ofthechild warrantareevaluation;orifthechildsparentsorteacherrequestsareevaluation.Areevaluation shallnotoccurmorefrequentlythanonceayear,unlesstheparentandthelocaleducational agencyagreeotherwise;andatleastonceevery3years,unlesstheparentandthelocal educationalagencyagreethatareevaluationisunnecessary(34C.F.R.71 1 614(2)(A)(B)) Individual education program(IEP) Definitionand componentsofIEP TheindividualizededucationprogramorIEPisawrittenstatementforeachchildwitha disabilitythatisdeveloped,reviewed,andrevisedinameetingthatmustinclude: N Astatementofthechildspresentlevelsofacademicachievementandfunctional performance N Astatementofmeasurableannualgoals,includingacademicandfunctionalgoals designedtomeetthechildsneedsthatresultfromthechildsdisabilitytoenablethe childtobeinvolvedinandmakeprogressinthegeneraleducationcurriculum;andmeet eachofthechildsothereducationalneedsthatresultfromthechildsdisability.For childrenwithdisabilitieswhotakealternateassessmentsalignedtoalternate achievementstandards,adescriptionofbenchmarksorshort-termobjectives; N Adescriptionofhowthechildsprogresstowardmeetingtheannualgoalswillbe measured;andwhenperiodicreportsontheprogressthechildismakingtoward meetingtheannualgoals(suchasthroughtheuseofquarterlyorotherperiodicreports, concurrentwiththeissuanceofreportcards)willbeprovided. N Astatementofthespecialeducationandrelatedservicesandsupplementaryaidsand services,basedonpeer-reviewedresearchtotheextentpracticable,tobeprovidedtothe child,oronbehalfofthechild N Astatementofanyindividualappropriateaccommodationsthatarenecessaryto measuretheacademicachievementandfunctionalperformanceofthechildonStateand districtwideassessmentsconsistentwithsection612(a)(16)oftheAct;andiftheIEPTeam determinesthatthechildmusttakeanalternateassessmentinsteadofaparticular regularStateordistrictwideassessmentofstudentachievement,astatementofwhythe childcannotparticipateintheregularassessmentandwhytheparticularalternate assessmentselectedisappropriateforthechild(34C.F.R. 1 300.320(a)) Development,review, andrenewalofIEPs WhendevelopingtheIEP,theIEPTeammustconsider (i)Thestrengthsofthechild; (ii)The concernsoftheparentsforenhancingtheeducationoftheirchild; (iii)Theresultsofthe initialormostrecentevaluationofthechild;and (iv)Theacademic,developmental,and functionalneedsofthechild.(34C.F.R.71 1 300.324) TheIEPTeammust (i)reviewthechildsIEPperiodically,butnotlessthanannually,to determinewhethertheannualgoalsforthechildarebeingachieved;and (ii)revisetheIEP, asappropriate,toaddress (A)anylackofexpectedprogresstowardtheannualgoals,and inthegeneraleducationcurriculum,ifappropriate (B)theresultsofanyreevaluation conducted; (C)informationaboutthechildprovidedto,orby,theparents; (D)thechilds anticipatedneeds;or (E)othermatters.(34C.F.R.71 1 300.324) IEPteam TheIEPTeamforeachchildwithadisabilityincludes:(1)Theparentsofthechild;(2)Not lessthanoneregulareducationteacherofthechild(ifthechildis,ormaybe,participating intheregulareducationenvironment);(3)Notlessthanonespecialeducationteacherof thechild,orwhereappropriate,notlessthanonespecialeducationproviderofthechild; (4)Arepresentativeofthepublicagency;(5)Anindividualwhocaninterpretthe instructionalimplicationsofevaluationresults(6)Atthediscretionoftheparentorthe agency,otherindividualswhohaveknowledgeorspecialexpertiseregardingthechild, includingrelatedservicespersonnelasappropriate;and(7)Wheneverappropriate,the childwithadisability.(34C.F.R.71 1 300.321) N IDEA-RelatedProfessionalDevelopmentinJuvenileCorrectionsSchoolsN N JournalofSpecialEducationLeadership26(2) N September201395 N

PAGE 6

IDEAPrinciple JCIssue IDEARegulation ParentalparticipationEachpublicagencymusttakestepstoensurethatoneorbothoftheparentsofachild withadisabilityarepresentateachIEPTeammeetingorareaffordedtheopportunityto participate.(34C.F.R.71 1 300.322) Eachpublicagencymustensurethattherightsofachildareprotectedwhen (1)No parent(asdefinedin 1 300.30)canbeidentified; (2)Thepublicagency,afterreasonable efforts,cannotlocateaparent; (3)ThechildisawardoftheStateunderthelawsofthat State;or (4)Thechildisanunaccompaniedhomelessyouth.(34C.F.R.71 1 300.519) WhenIEPsshouldbein effect IEPsshouldbeineffectatthebeginningofeachschoolyear.Eachpublicagencymusthavein effect,foreachchildwithadisabilitywithinitsjurisdiction,anIEP.(34C.F.R.71 1 300.323) ForinitialIEPsandprovisionofservices,eachpublicagencymustensurethat: (1)AmeetingtodevelopanIEPforachildisconductedwithin30daysofadetermination thatthechildneedsspecialeducationandrelatedservices;and (2)AssoonaspossiblefollowingdevelopmentoftheIEP,specialeducationandrelated servicesaremadeavailabletothechildinaccordancewiththechildsIEP. (d)AccessibilityofchildsIEPtoteachersandothers.Eachpublicagencymustensurethat (1)ThechildsIEPisaccessibletoeachregulareducationteacher,specialeducation teacher,relatedservicesprovider,andanyotherserviceproviderwhoisresponsibleforits implementation;and (2)Eachteacherandproviderisinformedof (i)HisorherspecificresponsibilitiesrelatedtoimplementingthechildsIEP;and (ii)Thespecificaccommodations,modifications,andsupportsthatmustbeprovidedfor thechildinaccordancewiththeIEP. WhenIEPsshouldbein effectforin-statetransfers (e)IEPsforchildrenwhotransferpublicagenciesinthesameState.Ifachildwitha disability(whohadanIEPthatwasineffectinapreviouspublicagencyinthesameState) transferstoanewpublicagencyinthesameState,andenrollsinanewschoolwithinthe sameschoolyear,thenewpublicagency(inconsultationwiththeparents)mustprovide FAPEtothechild(includingservicescomparabletothosedescribedinthechildsIEPfrom thepreviouspublicagency),untilthenewpublicagencyeither (1)AdoptsthechildsIEPfromthepreviouspublicagency;or (2)Develops,adopts,andimplementsanewIEPthatmeetstheapplicablerequirements WhenIEPsshouldbein effectforout-of-state transfer (f)IEPsforchildrenwhotransferfromanotherState.Ifachildwithadisability(whohadan IEPthatwasineffectinapreviouspublicagencyinanotherState)transferstoapublic agencyinanewState,andenrollsinanewschoolwithinthesameschoolyear,thenew publicagency(inconsultationwiththeparents)mustprovidethechildwithFAPE (includingservicescomparabletothosedescribedinthechildsIEPfromtheprevious publicagency),untilthenewpublicagency (1)Conductsanevaluationpursuantto 11 300.304through300.306(ifdeterminedtobe necessarybythenewpublicagency);and (2)Develops,adopts,andimplementsanewIEP,ifappropriate. (g)Transmittalofrecords.Tofacilitatethetransitionforachilddescribedinparagraphs(e) and(f)ofthissection: (1)Thenewpublicagencyinwhichthechildenrollsmusttakereasonablestepsto promptlyobtainthechildsrecords,includingtheIEPandsupportingdocumentsandany otherrecordsrelatingtotheprovisionofspecialeducationorrelatedservicestothechild, fromthepreviouspublicagencyinwhichthechildwasenrolled;and (2)Thepreviouspublicagencyinwhichthechildwasenrolledmusttakereasonablesteps topromptlyrespondtotherequestfromthenewpublicagency.(34C.F.R.71 1 300.323) Transitionservices Transitionservicesmustincludeanindividualizedplan,beinitiatedatintake,andarea coordinatedsetofactivitiesandsupports(34C.F.R.71 1 300.43) Leastrestrictive environment(LRE) Leastrestrictive environment YouthinJCareguaranteedaccesstothegeneraleducationcurriculum(34C.F.R.71 1 300.39(b)(3)(iii);C.F.R.71 1 300.320(a)(2)(i)(A)) Accesstoeducationand specialeducationservices Thereshouldexistacontinuumofspecialeducationservicesandtothemaximumextent possible,youthareeducatedwithnondisabledpeers(34C.F.R.71 1 300.114). Table1.—Continued. N IDEA-RelatedProfessionalDevelopmentinJuvenileCorrectionsSchoolsN N 96 JournalofSpecialEducationLeadership26(2) N September2013N

PAGE 7

foundationofIDEA:‘‘(a)zeroreject/childfind;(b) nondiscriminatorytesting;(c)IndividualEducation Program(IEP);(d)leastrestrictiveenvironment;(e) proceduraldueprocess;and(f)parentparticipation’’ (p.3).Inthissection,wediscusstheseprinciplesand theIDEA-relatedregulationsandissuesofconcern thatexistinJCschools.Theseprinciplesandtheir relatedIDEAstatutesarepresentedin Table1 .It shouldbenotedthatoverlapexistswithspecific IDEAprovisionsandtheseprinciples.Forexample, youthinJCareguaranteedaccesstothegeneral educationcurriculum(34C.F.R.71 1 300.39(b)(3)(i–ii); 34C.F.R.71 1 300.320(a)(2)(i)(A)).Thismaybe inhibitedbytheinadequateadherencetoastudent’s IEPandmayalsobeimpactedifayouthisnot providedinstructionintheleastrestrictive environment.Aswediscussthekeyprinciples,we willtypicallylimitdiscussionofspecificIDEA provisionstoasingleprinciple.Foreachprinciple, weprovideabriefoverviewandexplanationofthe provision.Next,wediscussidentifiedandpotential causesforalackofadherence.Then,weidentify implicationsoftheproblems,intermsofthepossible effectsonstudentsandsociety.ZeroReject/ChildFindThefirstprincipleofIDEAiszeroreject/childfind. Withregardtothisprinciple,IDEA-relatedconcerns withinJCschoolsincludetheprovisionsoffreeand appropriateeducation(FAPE)andtheidentification ofstudentswithdisabilitiesthroughchildfind. FAPEisanongoingissuewithinJCschoolsandhas beenthesubjectofatleast56lawsuits(Gagnon, Leone,&Jossi,2013).FAPEspansseveralofthe principles,andwediscussspecificissuesand recommendations(e.g.,providingconformitywith IEPrequirements)inthesectionthatfollows. Broadlyspeaking,Krezmien(2008)notedinhis evaluationofTexasfacilitiesthattheorganizationof courseenrollmentbasedonsecurityandnotstudent academicneedshinderedprovisionofFAPE.More specifically,henotedthatinstructionalpractices wererarelybasedonempiricalresearchandalso concernswithchildfind.Wediscussthesetwo FAPE-relatedissuesinthissection........................................... Freeandappropriateeducation(FAPE)isan ongoingissuewithinjuvenilecorrectionsschools andhasbeenthesubjectofatleast56lawsuits (Gagnon,Leone,&Jossi,2013).Appropriateinstructionaladaptationsforyouth withdisabilitiesareguaranteedunderIDEA(34 C.F.R.71 1 300.39(b)(3)(i–ii)).Asnoted,thereisan overrepresentationofyouthwithdisabilitiesinJC andresearchershaveidentifiedpervasiveacademic deficitsofincarceratedyouth(Krezmien,Mulcahy, &Leone,2008).However,effectiveinstructional practiceswithinJCarerareandmanyteachersdonot havetheskillsnecessarytoprovideresearch-based instruction(Houchinsetal.,2009).Thesupportthat teachersreceivefromadministratorsmayalsobe lacking,giventhatmanyJCprincipalsdonothave adequateknowledgeoftheIDEAandmaynot understandthatresearch-basedinstructional adaptationsareactuallyrequired,asappropriate,for youthwithdisabilities(Gagnonetal.,2009; Gagnonetal.,2010).Withoutprovisionof appropriateinstructionaladaptations,itishighly Table1.—Continued. IDEAPrinciple JCIssue IDEARegulation Proceduraldue process DisciplineproceduresIDEAemphasizesproactivebehaviorimprovementplans(34C.F.R.71 1 300.324(a)(2)(i)) Individualizedbehavioralinterventionsandmanifestationdeterminations(34C.F.R.71 1 300.530(e)(1)(iii)) ProceduralsafeguardsProceduralsafeguardsanddueprocessforparentsandchildrenwithinIDEA(34C.F.R.71 1 300.500.529). ParentparticipationConfidentialityof information ConfidentialityofinformationisrequiredwithinIDEA(34C.F.R.71 1 300.610.627). Parentalandsurrogate parentparticipation SchoolsmustattempttoinvolveParent(andsurrogateparents)intheeducationalprocess (34C.F.R.71 1 300.322; 1 300.519). Note. IDEA 5 IndividualswithDisabilitiesEducationAct;JC 5 juvenilecorrections;FAPE 5 freeandappropriateeducation. N IDEA-RelatedProfessionalDevelopmentinJuvenileCorrectionsSchoolsN N JournalofSpecialEducationLeadership26(2) N September201397 N

PAGE 8

unlikelythatyouthwithdisabilitieswillaccess andbesuccessfulinthegeneraleducation curriculum. .......................................... Broadlyspeaking,Krezmien(2008)notedinhis evaluationofTexasfacilitiesthattheorganization ofcourseenrollmentbasedonsecurityandnot studentacademicneedshinderedprovisionof FAPE. Childfindistheothermajorconcerninthis principle(notedundertheIEPsectionof Table1 ). Appropriateandsystematicchildfindprocesses mustbeinplacetoidentifystudentsalready classifiedasneedingspecialeducationservicesandto recordtransfersbetweenJCandpublicschools (Storandt,2007).However,alackofchildfind policiesandpracticesisaseriousissuewithinJC schools(Krezmienetal.,2008).TheFamily EducationalRightsandPrivacyActof1974often interruptstheseprocesses,whichisfrequently interpretedinappropriatelytomeanthatpublicand JCschoolscannottransferschoolrecords(Leone& Weisberg,2010).Furthercomplicatingtheissueis thatincarceratedyouthmaybewardsofthestate. However,‘‘Congressmadeitclearthatchildrenwith disabilitieswhoarewardsofthestateandinneedof specialeducationmustbeidentified,located,and evaluated’’(Leone&Weisberg,2010,p.24).WhenJC schoolsdonotreceivestudentrecordsinatimely manner,itcangreatlyinhibittheidentificationof youthwhoareclassifiedwithadisabilityand subsequentlyresultininadequateacademic andbehavioralsupports(Krezmienetal.,2008; Leone&Weisberg,2010).Assuch,anincreased understandingofchildfindpoliciesandprocedures arenecessarycomponentsofacomprehensive professionaldevelopmentprograminJC.NondiscriminatoryTestingThesecondprincipleofIDEAisnondiscriminatory testing.Specificallywithinthisprinciple,IDEA-related concernsincludealackofadherencetothe requirementsofstudentevaluationandreevaluation. Todate,15lawsuitsagainstJCfacilitieshavecited inappropriatepoliciesandpracticeswithregardto evaluationandreevaluation(Gagnonetal.,2013).IDEA mandatesthatnondiscriminatoryevaluationbeutilized indeterminingthepresenceandextentofadisability andwhetherspecialeducationand/orrelatedservices arenecessary.IDEAalsorequiresthatevaluationsbe conductedinamannerthatensuresresultsarenot biasedduetocultural,language,orsocioeconomic factors(Turnbull,Wilcox,&Stowe,2002). Adherencetothisprincipleisparticularlycritical inJCschools,giventhedisproportionatenumberof minoritystudentsfromhigh-povertybackgrounds (Quinn,Rutherford,Leone,Osher,&Poirer,2005). Moreover,IDEAregulationsincludetherequirement thatassessmentsare‘‘intheformmostlikelyto yieldaccurateinformationonwhatthechildknows andcandoacademically,developmentally,and functionally’’(34C.F.R.71 1 300.304(1)(ii)).Providing appropriateassessmentsandaccommodationsare criticaltoobtainingaccurateinformation,particularly forstudentswithdisabilities,anotherpopulationthat isoverrepresentedinJC.However,guidanceforJC administratorsandteachersisnecessary,because theirabilitytochooseassessmentsanddevelop accommodationswithoutguidancehasbeencalled intoquestion(Gagnon,Maccini,&Haydon,2011). Krezmien(2008)alsonotedissueswith appropriateassessmentinhisanalysisofTexasJC schools.HereportedthattheTestofAdultBasic Educationwasthesoleassessmenttoolandthat unqualifiedpersonneladministeredtheassessment. Moreover,inclearviolationoftheIDEAprovision that‘‘Thechildisassessedinallareasrelatedtothe suspecteddisability,including,ifappropriate,health, vision,hearing,socialandemotionalstatus,general intelligence,academicperformance,communicative status,andmotorabilities’’(34C.F.R.71 1 300.304(5)), facilitiesreliedontheTestofAdultBasicEducation asthesoleassessmentforidentifyingyouthwitha disability.Inappropriateassessmentsoralackof comprehensiveassessmentcouldpotentiallyaffect whetherayouthisidentifiedasneedingspecial educationservices.Further,theseissuesmayleadto inappropriatelevelsofserviceand/oraffect conclusionsconcerningyouthprogressonIEPgoals andobjectives.IndividualEducationProgramThereareseveralaspectsofIEPsidentifiedwithin IDEAregulations.Infact,concernswithdeveloping andimplementingappropriateIEPshavebeen citedasacomplaintin13lawsuitsagainstJC N IDEA-RelatedProfessionalDevelopmentinJuvenileCorrectionsSchoolsN N 98 JournalofSpecialEducationLeadership26(2) N September2013N

PAGE 9

schools(Gagnonetal.,2013).IEPviolationscitedin DepartmentofJusticefindingsrelatedtoadherence ofbasicIEPprovisions,including(a)lackof individualization;(b)decisionsaboutspecial educationservicedeliverybasedontheavailable resources,ratherthanstudentneeds;(c) inconsistenciesbetweencurrentandpastIEPs(e.g., numberofminutesofdirectservicesprovided); (d)lackofIEPimplementationandrelateddata collectionandanalysis;and(e)absenceof participationbyrequiredteammembersinIEP meetings(e.g.,seeU.S.DepartmentofJustice, 2011).WithoutJCfacilityadherencetothesemost basicIEP-relatedprovisions,itisunlikelythat youthwithdisabilitieswillexperienceacademic success. Inadditiontomorebasiccomplianceissues,there aretwoIEP-relatedissuesthatareparticularly concerningwithinJC:(a)accesstothegeneral educationcurriculum;and(b)transitionplanning. First,itisguaranteedwithinIDEAthatyouthwith disabilitieswillbeprovidedaccesstothegeneral educationcurriculum(34C.F.R.71 1 300.320).Access toacurriculumthatisalignedwithstatestandards andassessmentsisguaranteedandnecessaryfor youthtoearnahighschooldiploma.However,only aboutthreefourthsofJCprincipalsreportedthatthey useddistrictorstatecurriculumandthattheirschool wasalignedwithstateassessments toagreatextent (Gagnonetal.,2009).Granted,JCprincipalsand teachersfacesetting-specificcomplicationswith curriculum.Forexample,incarceratedyouth typicallyhavefewacademiccredits(Leone& Cutting,2004).Whereasitwouldbenefitmany incarceratedyouthtoenrollinprevocationaland vocationaltraining,paidworkexperience,and GeneralEducationalDevelopment(GED)test preparation,maintainingcompliancewithIDEA requiresthatthesecoursessupplementandnot supplantaccesstothegeneraleducationcurriculum (Gagnonetal.,2009).IEP-relatedprofessional developmentwouldneedtodelineatethisimportant distinctionandensureadherencetoIDEA requirements. Thetransitionplanisanessentialcomponentof theIEPforsecondary-school-agedstudents.This planshouldbeindividualizedandcontaina coordinatedsetofactivitiesandsupports(34C.F.R. 71 1 300.43).Unfortunately,transitionplansand servicesinJCarecommonlygeneric,lackspecificity, andarenotavailabletoallyouthwhoneedthem (Krezmien,2008).Infact,JCadministratorsand teachersidentifiedasignificantneedforprofessional developmentthatfocusedspecificallyontransition services(Mathuretal.,2009).Transitionisoften complicatedforincarceratedyouthwithdisabilities. Forexample,theseyouthrequireatransitionplan thatincludesasetofactivitiesandsupportsthat outlinewhatwilloccurfromschooltoschoolorfrom schooltopostschool;includingpostsecondary education,vocationaltraining,oremployment (Leone&Cutting,2004).Administratorsand teachersshouldbeawareofotheruniquechallenges thattheseyouthfaceincluding‘‘(a)howtorespond toemployersaboutpreviousinvolvementwith juvenilejustice;(b)howtogetjuvenilerecords sealedandexpunged;and(c)howtogetsuchitems asasocialsecuritycard,financialassistance(e.g., healthcare,housingassistance,foodassistance)’’ (Gagnon&Richards,2008,p.16).Addressingboth theIDEArequirementsanduniquetransitionneeds forincarceratedyouthwillpromotelong-term reintegrationintoschool,theworkforce,andsociety asawhole.LeastRestrictiveEnvironmentThefourthprincipleofIDEAistheprovisionofthe leastrestrictiveenvironment.Thereshouldexista continuumofspecialeducationservicesandyouth mustbeeducatedwithnondisabledpeerstothe maximumextentpossible(34C.F.R.71 1 300.114). Theconcernisthat,withina‘‘closed’’facilitywith limitedspaceandstaff,studentsmaybecompelled tolearninasetting(e.g.,self-contained,total inclusion)thatisinappropriatefortheirindividual needs(Krezmien,2008;Leone&Cutting,2004). Althoughthisisadifficultlogisticalsituation,itis essentialthatJCadministratorsunderstandthat resolutionisnecessarytomaintaincompliancewith IDEArequirementsandtobestserveincarcerated youthwithdisabilities.ProceduralDueProcessUnderproceduraldueprocesstherearetwo requirementsthatthatareparticularlyproblematic withinJC:proactivebehaviorimprovementplans(34 C.F.R.71 1 300.324(a)(2)(i)),aswellasindividualized behavioralinterventionsandmanifestation determinations(34C.F.R.71 1 300.530(e)(1)(i–ii)). N IDEA-RelatedProfessionalDevelopmentinJuvenileCorrectionsSchoolsN N JournalofSpecialEducationLeadership26(2) N September201399 N

PAGE 10

Giventhattherateofyouthwithemotional disturbance(ED)inJCisaboutsixtimesthatof regularschools,thereareconcernsthattheseyouth aredisproportionallyaffectedbyinadequate behavioralpoliciesandpractices(Gagnon&Barber, 2010).Unfortunately,JCadministratorsandteachers rarelyhavethetrainingneededtodevelopand implementacomprehensiveapproachtopromoting positiveyouthbehavior(Oliver&Reschly,2010). Developmentandimplementationofbehavioral interventionsinJCisalsocomplicatedbytheintegral roleofsecuritypersonnelintheJCsetting. Competingphilosophiesofbehavior(i.e.,proactive andpositive,reactiveandpunitive)cancause conflictsbetweeneducationalandcorrectionalstaff. Furthermore,thereareoftensignificantdifferencesin educatorversuscorrectionalofficerbackground knowledgeofyouthwithdisabilitiesandbehavioral interventions.Infact,excludingeducators,lessthan twothirdsofJCpersonnelhavetrainingfocused onyouthwithdisabilities(Kvarfordt,Purcell,& Shannon,2005).Additionally,Stinchcomb(2002) identifiedthatinalmost90%offacilities,correctional officersobtainonlyahighschooldiplomafor employment.Assuch,forappropriatebehavioral policiesandpracticestobedevelopedand implementedwithfidelity,thereisneedinJCfor cross-disciplineprofessionaldevelopmentamong administrators,educators,andsecuritypersonnel (Jurich,Casper,&Hull,2001).ParentParticipationThesixthandfinalprincipleofIDEAisparticipation ofparents,includingguardians/surrogateparents. Schoolsmustattempttoinvolveparents(and guardians/surrogateparents)intheeducational process(34C.F.R.71 1 300.322;34C.F.R.71 1 300.519). However,parentalinvolvementineducational programmingandspecialeducationservicesisoften lackingwithinJC(Houchinsetal.,2009).In particular,participationiscomplicatedwhenyouth areconsideredwardsofthestateand/orwhenyouth areincarceratedinafacilitythatisfarfromtheir home.JCeducatorsandadministratorsmayalsonot havetheknowledgeorskillsnecessarytonavigate setting-specificlogisticaldifficultiesrelatedto includingparents(PublicEducationNetwork,2006). Inadditiontoparent/guardianparticipationas perrequirementswithinIDEA,familialinvolvement forconfinedyouthisevenmorecritical.Forexample, parent/guardiansupportisnecessaryforeffective family-basedcognitive-behaviortreatmentofyouth mentaldisordersanddrugproblems(Hoagwood, Burns,Kiser,Ringeisen,&Schoenwald,2001). Moreover,parentalinvolvementduringyouth incarcerationhasbeenindicatedasafactorfor reducingrecidivismrates(NationalCenterforMental HealthandJuvenileJustice,2003). Clearly,lackofadministrator,educator,and correctionalstaffunderstandingofthecurrentfederal regulationsnegativelyaffectIDEAimplementationinJC. Moreover,thereareseveraluniqueattributesofthe settingandstudentsthatcomplicateprovisionofIDEA.It isbeyondthescopeofourcurrentdiscussiontoaddress eachattributeandthewaysinwhichsafetyand appropriateeducationandspecialeducationcanbe ensured.Rather,inthesectionsthatfollow,weprovidea planforhowtoprovideprofessionaldevelopment concerningIDEAinamannerthatisaccessibletoJC personnel,whoareoftenisolatedfromtypical professionaldevelopment.Furthermore,weaddress methodsofpromotingfacilitydiscussion,aswellas policyandpracticechangebyincludingkeystakeholders withintheJCsettingdecision-makingprocess.TransferofAdministratorandKey PersonnelKnowledgetoSchoollevelPolicyandPracticeAneffectiveschoolleadershipteamisessentialto modifyexistingIDEA-relatedorganizational practicesandincreasecapacityofpersonnel.School leadershipteamsinfluenceclassroompractices throughthecreationofworkingconditionsthat motivateprofessionalsandfacilitateeffective practices(Boydetal.,2011;Guskey,2000).Broadly, professionallearningteamsareacrucialcontextual facilitatorthatmayimproveeducator,administrator, andcorrectionalstafflearning,beliefs,andpractices, andsubsequently,students’outcomes(Hochberg& Desimone,2010).Specifically,theseteamshavethe potentialtoinfluencestudentoutcomesindirectly throughproximalimpactinseveralareasthatare relatedtoIDEAimplementation,including(a) teachermotivationandpractice;(b)curriculum;(c) assessment;and(d)discipline. InordertoimproveJCschoolconditionsand compliancewithIDEApolicy,JCschoolswouldbe wellservedtoformcollectiveleadershipteams N IDEA-RelatedProfessionalDevelopmentinJuvenileCorrectionsSchoolsN N 100 JournalofSpecialEducationLeadership26(2) N September2013N

PAGE 11

(CLT).ACLTisaprofessionalcollaborative communitycharacterizedbyprincipalssharing ownershipwithotherkeyschooladministrators, teachers,andserviceproviders(Louis,Leithwood, Wahlstrom,&Anderson,2010).Resultsfromthe largeststudyinvestigatinglinksbetweenschool leadershipandstudentlearningindicatethataCLT modelmaybeanadvantageouswayofaddressing someofthechallengesinJCschools(Louisetal., 2010).ACLTmodelcanreducethecommonisolation experiencedbysomemembersoftheJCstaff(e.g., correctionalstaff)andreducetheconflictinggoals thatmayexistacrossJCserviceproviders (DelliCarpini,2008;Leone&Weinberg,2010). Security,administrative,andeducationpersonnel servingvariousroleshavedifferentbackground trainingandphilosophiestowardtreatmentand educationofstudents.Whentheseassorted stakeholdershavetheopportunitytoworkcollectively insteadofseparatelyfromoneanother,theycanshare theirdifferentphilosophiesandgain abetterunderstandingofoneanother’sduties. Inaddition,thiscanhelptofacilitatebetter communicationandpotentiallybreakdownpunitive philosophiespresentinsomestaff(Leone&Weinberg, 2010).WithintheJCschoolcontext,CLTsshould includerepresentationofkeyfacilitystakeholders, includingtheschoolandfacilityadministration,general andspecialeducationteachers,counselorsand caseworkers,andsecuritypersonnel(Gagnonetal., 2012).GagnonandRichards(2008)emphasizedthat withinJC,‘‘collaborativeeffortsshouldinclude discussionofpolicyandpractice,methodsfor implementation,andaccountabilityforprogram effectiveness’’(p.40).Yet,onequestionremains:How doesaJCschoolfacilitateandsustainaCLT?Unique difficultiesdoexistwithregardtobuildingCLTsacross keyJCpersonnelinaschool(Leone&Cutting,2004). PDrepresentsanopportunitytobuildsuchateam.ProfessionalDevelopmentfor SchoolLeadersinJuvenile CorrectionsSchoolsFor30years,researchhasindicatedthatschoolleaders needongoingandeffectivePDtobuildandsustaina positiveschoolenvironment(Crockett,Billingsley,& Boscardin,2012;Guskey,2003).ThegenerallyagreeduponcharacteristicsofeffectivePDareextended duration;activelearning;collectiveparticipation; coherencewithlocal,state,andfederalpolicy;and specificcontentfocus(Desimone,2009;Gagnonetal., 2012;Mathuretal.,2009).PDforJCeducatorsalso needstoaddresssalientissuesrelevanttotheirspecific worksettingandpopulation(Fichman-Dana, Tricarico,&Quinn,2009).Theseresearchersalso supportPDthatprovidesspecificlinksbetween theoryandpracticeandincludesstrategiesthat enhancecriticalthinkingandreflection.Well-designed PDfacilitatesopportunitiestoreflect,discuss,and problemsolvewithmembersoftheleadershipand otherconcernedadults(Fichman-Danaetal.,2009). IDEA-relatedPDforJCschoolleadershipteams shouldbedesignedtoincreaseknowledgeand promotechangeviainformationandactivities designedforCLTs.Topromotebothunderstanding ofIDEAandpositivestepsforschool/facilitychange inpolicyandpractice,thePDprogramshouldconsist ofthreecriticalsegments:(a)introducingthePD, IDEA,andseriousnessofviolationsinJC;(b) promotingFAPE;and(c)planning,implementing, andassessingimprovementsinschoolpoliciesand practicesattheschoollevel. DespiteourunderstandingoftheurgencyforPD relatedtoIDEAimplementationinJC,thelandscape ofPDofferingsforeducatorsinJCschoolsisseverely limited.EducatorsinJCtypicallyreceive‘‘one-shot’’ workshopsthattheymayfindinterestingbutnot relevanttotheirwork.Additionally,intheeventthat theylearnpotentiallyusefulstrategies,thereisoften nofollow-uptoensureproperimplementationand sustainabilityofthenewinnovation(Mathuretal., 2009;Nelson,Jolivette,Leone,&Mathur,2010). Relatedtocontent,PDforschoolleadersisoften centeredonacademicsandlackssufficient experiencesandcourseworktoleadeffortstosupport studentswithdisabilities(Villani,2006).Assuch, thereisanurgentneedforPDinJCthatfocuses onstudentswithdisabilities,relatedpolicy,and subsequentlyprovidesopportunitiestolearn strategiesthatsupportyouthacademiclearningand socialbehavior(Grissom&Harrington,2010).OnlineProfessionalDevelopmentfor JCSchoolCollectiveLeadershipTeamsAsnoted,schoolleadersrequirecomprehensive traininginordertoleadeffortstodevelopandenact schoolactionplansthatalignpoliciesandpractices N IDEA-RelatedProfessionalDevelopmentinJuvenileCorrectionsSchoolsN N JournalofSpecialEducationLeadership26(2) N September2013101 N

PAGE 12

withIDEA.Theuseofonlinelearningisonepromising deliverymodelthatensurestheprogramiseasily accessibleandpromotesthenecessarycollaborationof keyplayerswithinandacrossJCfacilities(Nelsonet al.,2010).Althoughnoavailabledataexistregarding onlinePDforprincipalsofJCschools,thereisa growingbodyofresearchsupportingtheeffectiveness ofonlinelearningacrossothersettings. SomeofthebenefitsofonlinePDincludegreater flexibility,providing‘‘just-in-time’’support,reducing costs,allowingforreflectiontime,transcendingrural andisolatedcommunities,andprovidingadegreeof individualization(Dede,Ketelhut,Whitehouse,Breit, &McCloskey,2009).OnlinePDalsoallowsaccessto expertsandarchivalresourcesthatfiscalandlogistical constraintsmayotherwiselimit(Dedeetal.,2009). Whenconsideringtheeffectivenessandbenefitsof onlinePDobservedinothersettings,thereis considerablepromiseforitsapplicationwithschool leadersacrossJCschoolstoaddressgapsintraining. Despiteoverwhelmingempiricalsupportforthe useofonlineeducation,determiningtherequisite componentsforahigh-qualityonlinePDprogramis challenging.Perhapsthegreatestconsiderationwhen creatingasuccessfulprogramislinkingonlinecourse designtoapplicationandevaluationoflearning outcomes(Swan,Matthews,Bogle,Boles,&Day, 2012).Thefirststeptoensuringappropriatelearning outcomesisprovidingasoundpedagogical foundationspecifictothecontext.Thisfoundation canbefosteredbypromotingtwo-wayinteraction betweenfacilitatorandJCleadershipteams,thus providing‘‘amorehumanfeel’’totheonline material,promotingindividualizedlearning,and incorporatingrelevantmultimediacontent(Boling, Hough,Krinsky,Saleem,&Stevens,2012). Sustainedandcontinuedparticipationinonline coursesisvitaltoachievingexpectededucational outcomes(Reeves&Pedulla,2011).Learningneeds tobeconnectedtoparticipants’livesandprofessional needs(Kenner&Weinerman,2011).Creatorsof onlinecontentmustworktoincreaseparticipant satisfactionandcommitmenttopromotecontinued participation(Rovai,2002).Toensureasuccessful onlineexperience,PDfacilitatorsshouldalsofoster asenseofbelongingandconnectionamong participantsofthecourseandtheironlineclassmates (Baker,2010).Theutilizationofsocialmediainonline communitiesisonemethodtoachievethesefeelings ofbelongingandconnection.Forexample,social mediacanbeusedtopromotepeer-to-peer mentoringandinstructor-to-participantmentoring (Hoffmanetal.,2008).Othersocialmediadriven opportunitiesincludesharedreflectionsand enhancedcriticalthinkingactivitiesamongpeers. Thesecomponentscanincreasetheparticipationand benefitsofanonlinePD(Holley&Dobson,2008). AnonlinePDprogramforJCschoolleadersalso representsatrainingeffortthatisuserfriendlyfor schoolleadershipaudiencesandcanbefeasibly implementedinJC.Forexample,thisformatof trainingwouldbeadvantageousacrosstimeand location.Theonlineformathasthepotentialto facilitateparticipationfromconvenientlocations withouttheneedfortravelandtheopportunityto interactwithleadershipteamsinJCfacilitiesacross thecountry.Leadershipteamswillhaveautonomyto integratethePDintotheirschedulewhilepromoting gainsinIDEApolicyknowledge,aswellasactual changesinJCschoolpolicyandpractice.Additionally, becausethePDwouldprovideongoingsupportand feedback,onebenefitmaybeareductionofstaff burnoutandanincreaseinjobsatisfaction(Mathur& Schoenfeld,2010).WhenonlinePDisdelivered effectivelytoJCschoolleadershipteamsinorderto providekeyIDEAinformation,aswellastopromote implementationandevaluationofpositivechangesin schoolpolicyandpractice,weanticipatethatitwill leadtoimprovedimplementationofIDEAand ultimately,studentoutcomes.FinalThoughtsStudentsinJCaresomeofthemostvulnerableand academicallydisadvantagedindividualsinthe country.Althoughtheyareaffordedthesamerights asallstudentsinpublicschoolsunderIDEA(2006) andNoChildLeftBehind(2002),JCschoolshavea longhistoryofnoncompliancewithpolicy.Often, theyonlycomeintocomplianceinlightofreform drivenbyfederallitigation(Gagnon,2010;Houchins etal.,2009). IDEA-relatedPDinJCschoolsisavalidsolution toaddressgapsinknowledge,practice,and compliance.Specifically,ongoingandcomprehensive trainingisrecommendedthataddressesIDEA regulationsandapproachesincluding(a)zeroreject/ childfind;(b)nondiscriminatorytesting;(c)IEP;(d) leastrestrictiveenvironment;(e)proceduraldue process;and(f)parentparticipation. N IDEA-RelatedProfessionalDevelopmentinJuvenileCorrectionsSchoolsN N 102 JournalofSpecialEducationLeadership26(2) N September2013N

PAGE 13

CLTshaveshownpromiseforstudent achievementinrecentlarge-scaleresearchingeneral publicschoolsettings(Louisetal.,2010).This collaborativeapproachtoleadershipresponsibilities hasthepotentialtobehighlybeneficialintheJC schoolsetting,particularlyiftheCLTincludesthe schoolandfacilityadministration,generaland specialeducationteachers,counselorsand caseworkers,andsecuritypersonnel(Gagnonetal., 2012).However,theformationandsustainabilityofa CLTinaJCschoolisacomplexendeavor.These stakeholdersoftenworkinseparationfromeach otherwithdifferentandsometimesconflictinggoals andpurposes(Leone&Weinberg,2010).PDcan serveasafacilitatorforbuildingaCLTinJCschools. Nevertheless,PDthatprovidesJCschoolswith astructuredopportunitytoformaCLT,learn IDEA-contentknowledge,receiveguidancein implementationofIDEA-relatedmeasures,and providesfrequentopportunitiesforevaluationof progressrepresentsafeasiblewaytohelpJCschools moveintocompliancewithIDEA. OnlinePDasdeliverymethodisanimportant venueworthyoffurtherexploration.Asopposedto traditionalface-to-facePD,onlinePDcantakeplace anytimeandanywhereandcanprovideaforumfor collaborativerelationshipsacrossdistance.By combiningonlinePDwithCLTs,multiple stakeholdersinaJCsettingcanactivelyparticipatein trainingactivitiesandimplementingnewpractices; ensuringsustainabilityoflearnedinnovations. Currently,littleresearchexiststhatexamines PDpracticesinJCschools(Gagnonetal.,2012). However,wehaveidentifiedimportantIDEA-related concernsandsetforthastructureforPDthatutilizes CLTsandonlinePDtoincreasethelikelihoodof increasedknowledge,collaborativedecisionmaking, andimplementationofIDEA,aswellassustaining appropriatepoliciesandpractices.Throughthese actions,wewillhavethegreatestpotentialto positivelyaffectincarceratedyouthacademicand behavioralprogress.References Baker,C.(2010).Theimpactofinstructorimmediacyand presenceforonlinestudentaffectivelearning, cognition,andmotivation. JournalofEducatorsOnline , 7 (1),1. Boling,E.C.,Hough,M.,Krinsky,H.,Saleem,H.,& Stevens,M.(2012).Cuttingthedistanceindistance education:Perspectivesonwhatpromotespositive, onlinelearningexperiences. TheInternetandHigher Education , 15 ,118. Boyd,D.,Grossman,P.,Ing,M.,Lankford,H.,Loeb,S.,& Wyckoff,J.(2011).Theinfluenceofschool administratorsonteacherretentiondecisions. AmericanEducationalResearchJournal , 48 , 303. Boyle,J.R.,&Weishaar,M.K.(2001). Specialeducationlaw withcases .Boston,MA:Pearson. CorrectionalEducationAssociationStandards Commission.(2004). Performancestandardsfor correctionaleducationprograms:Backgroundinformation andexamplestandards .Elkridge,MD:Author. Crockett,J.B.,Billingsley,B.,&Boscardin,M.L.(2012). Handbookofleadershipandadministrationforspecial education .NewYork,NY:Routledge. Dede,C.,Ketelhut,D.J.,Whitehouse,P.,Breit,L.,& McCloskey,E.M.(2009).Aresearchagendaforonline teacherprofessionaldevelopment. JournalofTeacher Education , 60 ,8. DelliCarpini,M.(2008).Creatingcommunitiesof professionalpracticeinthecorrectionaleducation classroom. JournalofCorrectionalEducation , 59 ,219. Desimone,L.(2009).Improvingimpactstudiesofteachers’ professionaldevelopment:Towardbetter conceptualizationsandmeasures. Educational Researcher , 38 ,181. FamilyEducationalRightsandPrivacyAct,20U.S.C. 1 1232g,34CFR,pt.99.(1974). Fichman-Dana,N.,Tricarico,K.,&Quinn,D.M.(2009). Theadministratorasactionresearcher:Acasestudyof fiveprincipalsandtheirengagementinsystematic, intentionalstudyoftheirownpractice. JournalofSchool Leadership , 19 ,232. Gagnon,J.C.(2010).State-levelcurricular,assessment,and accountabilitypolicies,practices,andphilosophiesfor exclusionaryschoolsettings. JournalofSpecial Education , 43 ,206. Gagnon,J.C.,&Barber,B.R.(2010).Characteristicsofand servicesprovidedtoyouthinsecurecarefacilities. BehavioralDisorders , 36 ,7. Gagnon,J.C.,Barber,B.R.,VanLoan,C.L.,&Leone,P.E. (2009).Juvenilecorrectionalschools:Characteristics andapproachestocurriculum. EducationandTreatment ofChildren , 32 ,673. Gagnon,J.C.,Haydon,T.,&Maccini,P.(2010).Juvenile correctionalschools:Assessmentandaccountability policiesandpractices. JournalofCorrectionalEducation , 61 ,23. Gagnon,J.C.,Houchins,D.E.,&Murphy,K.M.(2012). Currentjuvenilecorrectionsprofessionaldevelopment practicesandfuturedirections. TeacherEducationand SpecialEducation , 35 ,3334. N IDEA-RelatedProfessionalDevelopmentinJuvenileCorrectionsSchoolsN N JournalofSpecialEducationLeadership26(2) N September2013103 N

PAGE 14

Gagnon,J.C.,Leone,P.E.,&Jossi,M.(2013). Ananalysisof disabilityrelatedlitigationconcerningjuvenilecorrections Unpublishedmanuscript. Gagnon,J.C.,Maccini,P.,&Haydon,T.(2011).Assessment andaccountabilityinpublicandprivatesecondary daytreatmentandresidentialschoolsforstudents withemotionalandbehavioraldisorders. Journalof SpecialEducationLeadership , 24 ,79. Gagnon,J.C.,&Richards,C.(2008). Makingtheright turn:Aguideaboutyouthinvolvedinthejuvenile correctionssystem pp.1.Washington,DC:National CollaborativeonWorkforceandDisabilityforYouth, InstituteforEducationalLeadership. GrillerClark,H.,Rutherford,R.B.,&Quinn,M.M.(2004). Practicesintransitionforyouthinthejuvenilejustice system(pp.247).InD.Cheney(Ed.). Transitionof studentswithemotionalorbehavioraldisabilitiesfrom schooltocommunity:Currentapproachestopositive outcomes .Arlington,VA:DivisionofCareer DevelopmentandTransition,CouncilforChildren withBehavioralDisorders. Grissom,J.A.,&Harrington,J.R.(2010).Investingin administratorefficacy:Anexaminationofprofessional developmentasatoolforenhancingprincipal effectiveness. AmericanJournalofEducation , 116 , 583. Guskey,T.R.(2000). Evaluatingprofessionaldevelopment . ThousandOaks,CA:CorwinPress. Guskey,T.R.(2003).Analyzinglistsofthecharacteristicsof effectiveprofessionaldevelopmenttopromote visionaryleadership. NASSPBulletin , 87 (637),4. Harlow,C.W.(2003,January). Educationandcorrectional populations. BureauofJusticeStatisticsSpecialReport. Washington,DC:U.S.DepartmentofJustice. Hoagwood,K.J.,Burns,B.J.,Kiser,L.,Ringeisen,H.,& Schoenwald,S.K.(2001).Evidence-basedpracticein childandadolescentmentalhealthservices. Psychiatric Services , 52 ,1179. Hochberg,E.D.,&Desimone,L.M.(2010).Professional developmentintheaccountabilitycontext:Building capacitytoachievestandards. EducationalPsychologist , 45 ,89. Hoffman,E.S.,Menchaca,M.P.,Eichelberger,A.,Cordiro, E.,Note-Gressard,S.L.,&Yong,L.(2008).Pickingtools fordistancelearning:Aviewfromthetrenches. ProceedingsoftheTechnology,Colleges&Community WorldwideOnlineConference ,175–183. Holley,D.,&Dobson,C.(2008).Encouragingstudent engagementinablendedlearningenvironment:The useofcontemporarylearningspaces. Learning,Media, &Technology , 33 ,139. Houchins,D.,Puckett-Patterson,D.,Crosby,S.,Shippen, M.,&Jolivette,K.(2009).Barriersandfacilitatorsto providingincarceratedyouthwithaqualityeducation. PreventingSchoolFailure , 53 ,159. Houchins,D.E.,Jolivette,K.,Shippen,M.E.,&Lambert,R. (2010).Theadvancementofhighqualityliteracy researchinjuvenilejustice:Methodologicaland practicalconsiderations. BehaviorDisorders , 36 ,61. Houchins,D.E.,Shippen,M.,&Jolivette,K.(2006).System reformandjobsatisfactionofjuvenilejusticeteachers. TeacherEducationandSpecialEducation , 29 ,127. IndividualsWithDisabilitiesEducationActof2004,Pub. L.No.108-446,118Stat.2658(2004). IndividualsWithDisabilitiesEducationAct,34C.F.R.pts. 300and301(2006). John,D.,&Catherine,T.MacArthurFoundation.(2004). Juvenilecourttrainingcurriculum (2nded.).Chicago,IL: Author. Jurich,S.,Casper,M.,&Hull,K.A.(2001).Training correctionaleducators:Aneedsassessmentstudy. JournalofCorrectionalEducation , 52 ,23. Kenner,C.,&Weinerman,J.(2011).Adultlearningtheory: Applicationstonon-traditionalcollegestudents. JournalofCollegeReadingandLearning , 41 ,87. Krezmien,M.P.(2008).Areviewofeducationprograms forstudentsintheTexasYouthCommissionstate schools:Aspecialreportoftheindependent ombudsman.Austin,TX:TexasYouthCommission. Krezmien,M.P.,Mulcahy,C.A.,&Leone,P.E.(2008). Detainedandcommittedyouth:Examiningdifferences inachievement,mentalhealthneeds,andspecial educationstatus. EducationandTreatmentofChildren , 31 ,445. Kvarfordt,C.L.,Purcell,P.,&Shannon,P.(2005).Youth withlearningdisabilitiesinthejuvenilejusticesystem: Atrainingneedsassessmentofdetentionandcourt servicespersonnel. Child&YouthCareForum , 34 , 272. Leone,P.E.,&Cutting,C.A.(2004).Appropriateeducation, juvenilecorrections,andNoChildLeftBehind. BehavioralDisorders , 29 ,260. Leone,P.E.,&Weinberg,L.(2010). Addressingtheunmet needsofchildrenandyouthinthejuvenilejusticeandchild welfaresystems .Washington,DC:CenterforJuvenile JusticeReform. Leone,P.E.,Zablocki,M.,Wilson,M.,Mulcahy,C.,& Krezmien,M.(2009).Specialeducationanddisability rights:Modulethree.In Towarddevelopmentally appropriatepractice:Ajuvenilecourttrainingcurriculum (pp.90).Washington,DC:NationalJuvenile DefenderCenterandtheJuvenileLawCenter. Louis,K.S.,Leithwood,K.,Wahlstrom,K.L.,&Anderson, S.E.(2010,July). Learningfromleadership:Investigating thelinkstoimprovedstudentlearning .RetrievedJune28, 2013fromhttp://www.wallacefoundation.org/ knowledge-center/school-leadership/key-research/ Documents/Investigating-the-Links-to-ImprovedStudent-Learning.pdf. N IDEA-RelatedProfessionalDevelopmentinJuvenileCorrectionsSchoolsN N 104 JournalofSpecialEducationLeadership26(2) N September2013N

PAGE 15

Mathur,S.R.,GrillerClark,H.G.,&Schoenfeld,N.A. (2009).Professionaldevelopment:Acapacity-building modelforjuvenilecorrectionaleducationsystems. JournalofCorrectionalEducation , 60 ,164. Mathur,S.R.,&Schoenfeld,N.(2010).Effective instructionalpracticesinjuvenilejusticefacilities. BehavioralDisorders , 36 ,20. NationalCenterforMentalHealthandJuvenileJustice. (2003). Nationalpolicyforumonmentalhealthandjuvenile justice:Movingtowardanintegratedpolicyforyouth . Delmar,NY:PolicyResearchAssociates. Nelson,C.M.,Jolivette,K.,Leone,P.E.,&Mathur,S.R. (2010).Meetingtheneedsofat-riskandadjudicated youthwithbehavioralchallenges:Thepromiseof juvenilejustice. BehavioralDisorders , 36 ,70. NoChildLeftBehindActof2001,Pub.L.No.107 (2002). Oliver,R.M.,&Reschly,D.J.(2010).Specialeducation teacherpreparationinclassroommanagement: Implicationsforstudentswithemotionaland behavioraldisorders. BehavioralDisorders , 35 ,188. PublicEducationNetwork.(2006). TitleII:Teacherand principalpreparation .WashingtonDC:Author. Quinn,M.,Rutherford,R.,Leone,P.E.,Osher,D.,&Poirer, J.(2005).Youthwithdisabilitiesinjuvenile correctional:Anationalsurvey. ExceptionalChildren , 71 ,339. Reeves,T.D.,&Pedulla,J.J.(2011).Predictorsofteacher satisfactionwithonlineprofessionaldevelopment: EvidencefromtheUSA’se-LearningforEducators Initiative. ProfessionalDevelopmentinEducation , 37 , 591. Rovai,A.P.(2002).Buildingsenseofcommunityata distance. InternationalReviewofResearchinOpenand DistanceLearning , 3 (1),1. Stinchcomb,J.B.(2002).Fromrehabilitationtoretribution: Examiningpublicpolicyparadigmsandpersonnel educationpatternsincorrections. AmericanJournalof CriminalJustice , 27 ,1. Storandt,J.(2007). Toolsforpromotingeducationalsuccessand reducingdelinquency .Alexandria,VA:National AssociationofStateDirectorsofSpecialEducation. Swan,K.,Matthews,D.,Bogle,L.,Boles,E.,&Day,S.(2012). Linkingonlinecoursedesignandimplementationto learningoutcomes:Adesignexperiment. TheInternet andHigherEducation , 15 ,81–88. Turnbull,H.R.,Wilcox,B.L.,&Stowe,M.J.(2002).Abrief overviewofspecialeducationlawwithfocusonautism. JournalofAutismandDevelopmentalDisorders , 32 , 479. U.S.DepartmentofJustice,CivilRightsDivision.(2012, August22).Findingsletter:Investigationofthe PendletonJuvenileCorrectionalFacility,Indianapolis, Indiana.RetrievedApril4,2013fromhttp://www. justice.gov/crt/about/spl/documents/pendleton_ findings_8-22-12.pdf. Villani,S.(2006). Mentoringandinductionprogramsthat supportnewprincipals .ThousandOaks,CA:Sage. West,J.E.,&Schaefer-Whitby,P.J.(2008).Federalpolicy andtheeducationofstudentswithdisabilities: Progressandthepathforward. FocusonExceptional Children , 41 (3),1.AbouttheAuthorsJosephCalvinGagnon,Ph.D.,isAssociateProfessor attheUniversityofFlorida,SchoolofSpecial Education,SchoolPsychology,andEarlyChildhood Studies,1403NormanHall,P.O.Box117050, Gainesville,FL32611-7050.E-mail: jgagnon@coe.ufl.edu. KristinM.Murphy,M.S.,Ed.M.isaLecturerin SpecialEducation,WheatleyHall,2ndFloor,Room 143,DepartmentofCurriculum&Instruction, CollegeofEducationandHumanDevelopment, UniversityofMassachusettsBoston,100Morrissey Blvd.,Boston,MA02125-3393.E-mail: kristin.murphy@umb.edu. MaryAnneSteinberg,Ph.D.isattheUniversityof Florida,SchoolofSpecialEducation,School Psychology,andEarlyChildhoodStudies,1403 NormanHall,P.O.Box117050,Gainesville,FL326117050.E-mail:MASteinberg@coe.ufl.edu. JustinGaddis,M.Ed.isattheSarahA.Reed Children’sCenter,2445West34thSt.,Erie,PA16506. E-mail:jggadd2@gmail.com. JeanCrockett,Ph.D.isProfessorandDirectorofthe SchoolofSpecialEducation,SchoolPsychology,and EarlyChildhoodStudiesattheUniversityofFlorida, 1403NormanHall,P.O.Box117050,Gainesville,FL 32611-7050.E-mail:jcrockett@coe.ufl.edu. N IDEA-RelatedProfessionalDevelopmentinJuvenileCorrectionsSchoolsN N JournalofSpecialEducationLeadership26(2) N September2013105 N



PAGE 1

READING PERFORMANCE AND HIGH STAKES STATEWIDE ASSESSMENT IN A JUVENILE CORRECTIONS FACILITY By JUSTIN GRANT GADDIS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2014

PAGE 2

© 2014 Justin Grant Gaddis

PAGE 3

3 ACKNOWLEDGMENTS I want to extend my sincere gratitude to the members of my committee for their continued guidance and unwavering support. Each member has not only been essential to the development and completion of my dissertation, but also to my growth as a professional throughout my time at the University of Florida. Foremost, I thank my co chair, Dr. Joseph Gagnon. His generosity is seemingly limitless, and for that I am beyond appreciative. I also thank Dr. Joyce for opening countless doors for me and for serving as a continued pillar of support. I also would like to thank my advisor and co chair, Dr. John Kranzler, for helping me mature as a scholar and as a person. I also thank Dr. David Miller for providing much needed assistance with my research design and statisti cal analysis. Lastly, I thank Dr. Holly Lane, Dr. David Houchins, Dr. Yaacov Petscher, and Dr. Joseph Torgesen for their contributions. I thank my family, who are still not entirely sure what it is that I was doing in Florida from the years of 2008 to 201 3. I cherish my mother, Julia Dohn, whose love persevered even through my tumultuous childhood. Words cannot describe how thankful I am. I also thank my father and stepmother, Martin Gaddis and Sara Gaddis. They taught me many lessons about life regardless of whether I thought I needed to hear them or not. I am also blessed with two grandmothers: Nora Gaddis and Barbara Cox. I thank them for being tremendously loving and supportive, always. I thank my grandfather, James Cox, for teaching me all sorts of man ly things, and for being the same man on Sunday mornings that he was on Saturday nights. I thank my Jeffrey Heston, who never expected me to turn the other cheek and for giving so much more than he signed up for. He is truly a man for all seasons.

PAGE 4

4 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 3 LIST OF TABLES ................................ ................................ ................................ ........................... 6 LIST OF FIGURES ................................ ................................ ................................ ......................... 8 ABSTRACT ................................ ................................ ................................ ................................ ..... 9 CHAPTER 1 REVIEW OF THE LITERATURE ................................ ................................ ........................ 11 High Stakes Statewide Assessments ................................ ................................ ...................... 11 Reading Achievement and Statewide Assessments ................................ ................................ 15 High Stakes Testing and Youth in Juvenile Corrections ................................ ........................ 20 Characteristic of Youth in Juvenile Corrections ................................ ............................. 21 Incarcerated Youth and Special Education Disability ................................ ..................... 23 Incarcerated Youth and Reading Ability ................................ ................................ ......... 25 The Florida Comprehensive Assessment Test (FCAT) ................................ .......................... 28 Academic Performance Variables and the FCAT SSS Reading ................................ ..... 31 Ethnicity and the FCAT DSS Reading ................................ ................................ ............ 39 Exceptional Student Status (ESE) and the FCAT DSS Reading ................................ .... 40 FAIR Progress Monitoring and the FCAT Reading ................................ ........................ 40 Limitations of Prior Research ................................ ................................ ................................ . 43 Purpose of the Current Study ................................ ................................ ................................ .. 45 Research Study Questions and Rationale ................................ ................................ ............... 47 Research Question One ................................ ................................ ................................ ... 47 Research Question Two ................................ ................................ ................................ ... 47 Resear ch Question Three ................................ ................................ ................................ . 47 2 RESEARCH METHODOLOGY ................................ ................................ ........................... 49 Participants ................................ ................................ ................................ ............................. 49 Sampling and Setting ................................ ................................ ................................ .............. 49 Research Question One ................................ ................................ ................................ ... 50 Research Question Two ................................ ................................ ................................ ... 51 Research Question Three ................................ ................................ ................................ . 52 Instrumentation ................................ ................................ ................................ ....................... 53 Intelligence Quotient (IQ) ................................ ................................ ............................... 54 Re ceptive Vocabulary (RV) ................................ ................................ ............................ 55 Sight Word Efficiency (SWE) and Phonemic Decoding Efficiency (PDE) .................... 55 Letter Word Identification (LWI), Reading Fluency (RF), Passage Comprehension (PC), & Word Attack (WA) ................................ ................................ ......................... 56 Oral Reading Fluency (ORF) & Reading Comprehension (RC) ................................ ..... 58

PAGE 5

5 Scholastic Reading Inventory (SRI) ................................ ................................ ................ 59 Florida Assessments for Instruction in Reading (FAIR) ................................ ................. 60 Florida Comprehensive Assessment Test (FCAT) ................................ .......................... 61 Research Design ................................ ................................ ................................ ..................... 62 Data Collection ................................ ................................ ................................ ....................... 63 Data Analysis ................................ ................................ ................................ .......................... 64 Preliminary An alyses ................................ ................................ ................................ ....... 64 Research Question One ................................ ................................ ................................ ... 66 Research Question Two ................................ ................................ ................................ ... 67 Research Question Three ................................ ................................ ................................ . 70 3 RESULTS ................................ ................................ ................................ ............................... 77 Preliminary Analys es ................................ ................................ ................................ .............. 77 Data Analysis ................................ ................................ ................................ .......................... 78 Research Question One ................................ ................................ ................................ ... 78 Research Question Two ................................ ................................ ................................ ... 80 Research Question Three ................................ ................................ ................................ . 89 4 DI SCUSSION ................................ ................................ ................................ ....................... 125 General Strengths of this Study ................................ ................................ ............................ 125 Research Question One ................................ ................................ ................................ ......... 126 Research Question Two ................................ ................................ ................................ ........ 134 Resear ch Question Three ................................ ................................ ................................ ...... 138 General Limitations ................................ ................................ ................................ .............. 141 Summary and Implications for Future Research ................................ ................................ .. 143 LIST OF REFERENCES ................................ ................................ ................................ ............. 148 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ....... 167

PAGE 6

6 LIST OF TABLES Table page 2 1 Overall sample demographics: Frequencies and percentages ................................ ............ 72 2 2 Subsample demographics for question one descriptive analyses: Frequencies and percentages ................................ ................................ ................................ ......................... 73 2 3 Sample demographics for question one ANOVA groupings: Frequencies and percentages ................................ ................................ ................................ ......................... 74 2 4 Sample demographics for question two multiple regression models: Frequencies and percentages ................................ ................................ ................................ ......................... 75 2 5 Sample demographics for question three simple linear regression models: Frequencies and percentages ................................ ................................ .............................. 76 3 1 ICC estimates (one way random effects model) for inter rater reliability on academic assessments ................................ ................................ ................................ ........................ 92 3 2 Absolute agreement estimates for inter rater rel iability on academic assessments ........... 93 3 3 Mean score differences between overall sample and imputed subsample ......................... 94 3 4 Descriptive statistics for imputed subsample by grade ................................ ...................... 95 3 5 Descriptive statistics for imputed subsample by ESE classification ................................ . 96 3 6 Descriptive statistics for imputed subsample by race/ethnicity ................................ ......... 97 3 7 Descriptive statistics for ANOVA of FCAT DSS Reading ................................ ............... 98 3 8 One way ANOVA procedure for FCAT DSS Reading ................................ ..................... 99 3 9 Shaffer Holm procedure of pairwise comparisons for FCAT DSS Reading .................. 100 3 10 Pearson product moment correlation between variables ................................ ................. 101 3 11 Simultaneous multiple regression analysis for predicting FCAT DSS Reading for imputed subsample ................................ ................................ ................................ ........... 1 02 3 12 Summary of backward elimination stepwise multiple regression models for predicting FCAT DSS Reading for imputed subsample ................................ ................. 103 3 13 AIC c model of best fit for predicting FCAT DSS Reading for imputed subsample ....... 104 3 14 Simultaneous multiple regression analysis for predicting FCAT DSS Reading for Grade 8 subsample ................................ ................................ ................................ ........... 105

PAGE 7

7 3 15 Summary of backward elimination stepwise multiple regression models for predicting FCAT DSS Reading for Grade 8 subsample ................................ .................. 106 3 16 AIC c model of best fit for predicting FCAT DSS Reading for Grade 8 subsample ........ 107 3 17 Simultaneous multiple regression analysis for predicting FCAT DSS Reading for Grade 9 subsample ................................ ................................ ................................ ........... 108 3 18 Summary of backward elimination stepwise multiple regression models for predicting FCAT DSS Reading for Grade 9 subsample ................................ .................. 109 3 19 AIC c model of best fit for predicting FCAT DSS Reading for Grade 9 subsample ........ 110 3 20 Simultaneous multiple regression analysis for predicting FCAT DSS Reading for Grade 10 subsample ................................ ................................ ................................ ......... 111 3 21 Summary of backward elimination stepwise multiple regression models for predicting FCAT DSS Reading for Grade 10 subsample ................................ ................ 112 3 22 AIC c model of best fit for predicting FCAT DSS Reading for Grade 10 subsample ...... 113 3 23 Descriptive statistics for SRI and FSP subsamples ................................ .......................... 114 3 24 Linear regression analysis for SRI predicting FCAT DSS Reading ............................... 115 3 25 Linear regression analysi s for FSP predicting FCAT DSS Reading ............................... 116 3 26 PDA of the SRI and FSP in relation to predicting FCAT DSS Reading Level ( > 3) ...... 117

PAGE 8

8 LIST OF FIGURES Figure page 3 1 Linear plot of conditional FCAT DSS Reading means for imputed subsample ............. 118 3 2 Residual plot for imputed subsample. ................................ ................................ .............. 119 3 3 Residual plot for Grade 8 subsample. ................................ ................................ .............. 120 3 4 Residu al plot for Grade 9 subsample. ................................ ................................ .............. 121 3 5 Residual plot for Grade 10 subsample. ................................ ................................ ............ 122 3 6 Residual plot for SRI subsample. ................................ ................................ ..................... 123 3 7 Residual plot for FSP subsample. ................................ ................................ .................... 124

PAGE 9

9 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy READING PERFORMANCE AND HIGH STAKES STATEWIDE ASSESSMENT IN A JUVENILE CORRECTIONS FACILITY By Justin Grant Gaddis August 2014 Chair: John H. Kranzler Cochair: Joseph C. Gagnon Major: School Psychology Recent legislation has ushered in an era of unprecedented accountability that has produced a paradigm shift within the field of education. Highlighting the push for accountability within schools is the reliance on stude nt statewide assessment performance as a means to inform high stakes decision making. Despite increased efforts by education researchers to establish links between reading performance and st atewide assessment performance , extant literature within juvenile corrections in regard to statewide assessment as well as the overall reading abilities possessed by these youth remains in its infancy. The educational deficits of delinquent youth, their proclivity to recidivate, and the overrepresentation of marginalized minority and disability groups within juvenile correctional facilities, all provide sufficient rationale for further inquiry into reading skills of these youth. F urther, studying the reading performance of incarcerated youth is of particular importance as the juvenile correctional system is often the last chance that these students have to be successful in school. Using a subset of participant data from Project LIBERATE, a randomized controlled trial federally funded by the U.S. Department of Education, Institute of Education Sciences (IES) (R324A080006), the present study sought to determine which specific academic and reading

PAGE 10

10 skills are most predictive of success on the reading section of the FCAT 2.0 for individuals in a juvenile corrections facility. L inear regression modeling , stepwise multiple regression model fitting procedures utilizing backward elimination (based on bias corrected Akaike Information Criterion [ AIC c ]), and predictive discriminant analysis (PDA) were employed to address study questi ons . Incarcerated s tudents within the study presented with significant intellectual and academic deficits. Deficits were prevalent across all academic and cognitive measures, as well as the FCAT 2.0 Reading . Further, incarcerated special education students performed significantly worse across all measures compared to their non disabled incarcerated peers. IQ, Sight W ord E fficiency (SWE), Reading C omprehension (RC), and Receptive V ocabulary (RV) were all statistically significant predictor variables. Overall , IQ proved to be the single most important predictor of FCAT 2.0 Reading success . Last, results indicated that the Florida Assessments for Instruction in Reading (FAIR) served as an adequate predictor of FCAT 2.0 Reading proficiency in regard to hit rate, sensitivity, and specificity.

PAGE 11

11 CHAPTER 1 REVIEW OF THE LITERATURE High Stakes Statewide Assessments Recent legislation (No Child Left Behind Act [NCLB], 2002; Individuals with Disabilities Education Improvement Act [IDEIA], 2006) has compelled stakeholders across practice, research, and policy spheres of public education to reconceptualize what constitut es high quality educational practices for all students. A paradigm shift within education, underscored by these legislative measures, has ushered in an era of unprecedented accountability, highlighted by stringent requirements for and increased reliance on high stakes statewide assessments (Gagnon et al., 2002; Guilfoyle, 2006; Liston, Whitcomb, & Borko, 2007; Sunderman, Kim, & Orfield, 2005). These legislative regulations have unequivocally altered the agnon, Mason Williams, 2012), and have created intense pressure for school personnel to improve academic outcomes, of all students, including those with disabilities (Jackson & Neel, 2006). Foundationally, NCLB is a continuation of a historic promise of access to a high quality However, ensuring access to high quality education is only the first step. NCLB also requires the creation of, and adherence to, standards and accountability systems designed to ensure greater educational outcomes for all students. Thus, NCLB at its core is designed to raise achievement levels across schools; with a specific focus on ensuring that all students reach grade level proficiency. This includes closing the achievement gap that parallels racial and socioeconomic distinctions (Darling Hammond, 2004; NCLB, 2002). Although NCLB includes a large number of regulations, the proceeding regulations discussed c oncern high stakes assessments. NCLB mandates (i.e., unless the state has received a flexibility waiver) that schools

PAGE 12

12 must achieve adequate yearly progress (AYP). The driving provision of establishing AYP is the requirement that all public schools must adm inister statewide standardized achievement tests annually . In order to attain AYP, schools must demonstrate that (a) at least 95% of students targets for a give n year, and (c) all subgroups (i.e., American Indian, Asian, Hispanic, African American, White, Limited English Proficient, Special Education, Migrant Status, and Free and Reduced Priced Lunch) achieve targets for graduation and/or attendance (Yell, Katsiy annas, & Shiner, 2006). Last, NCLB mandates that statewide assessments must be given to all students in Grades 3 through 8, and at least once after Grade 9 (U.S. Department of Education, 2007). st be included in statewide accountability systems, IDEIA contains a number of regulations designed to promote greater alignment with NCLB (Gagnon, Maccini, & Mulcahy, 2014). Specifically, IDEIA aligns with the assessment and accountability provisions of N CLB for students who have disabilities (Gagnon, 2008). For example, states must establish performance goals for students with dropout rates for students with d isabilities, and to the extent appropriate, maintain all general goals and standards established for all students, regardless of disability classification (34 C.F.R. § 300.157[a][1 4]). IDEIA also mandates that students with disabilities are included in al l state and district wide assessment programs, including those aligned with NCLB (i.e., with any necessary accommodations outlined in individualized education programs [IEP] as necessary) (34 C.F.R. § 300.160[a]) . Additionally, IDEIA provides that states m ust produce comprehensive public reports annually that detail the performance of students who have disabilities on statewide assessment, the number of students with disabilities who participated in assessments, including

PAGE 13

13 the number participating with accom modations, and the number who participated in alternative assessments (34 C.F.R. § 300.602[b][1]) . This accountability provision was enacted to ensure that states promote learning outcomes for all students, which includes reporting on the progress of stude nts who have disabilities towards pre established indicators of performance (Cortiella, 2006) . Overall, the provisions included in IDEIA are designed to increase achievement, and to promote greater accessibility to general education curriculum for students with disabilities. The provisions outlined within IDEIA and NCLB pertaining to high stakes statewide assessments not only afford protections to students who have disabilities (Gagnon, Maccini, & Haydon, 2011), but they also include students who are incarc erated (Leone & Cutting, 2004). Although NCLB and IDEIA afford the same protections to incarcerated students such as access to high quality educational services, juvenile correctional facilities have a historic record of noncompliance with education legisl ation, as evidenced by at least 56 lawsuits against these institutions (Gagnon, Leone, & Jossi, 2014, as cited by Gagnon, Murphy, Steinberg, Gaddis, & Crockett, 2013). One area of noncompliance concern is the lack of participation by delinquent students in statewide assessments (Maccini et al., 2012). Many juvenile correctional facilities also employ inconsistent, or simply do not participate in, statewide assessment reporting practices (Gagnon, 2010; Gagnon, Haydon, Maccini, 2010). The lack of consistent p ublic reporting practices is troublesome because these students overwhelmingly perform well below their non incarcerated counterparts on statewide assessments (Florida Office of Program Policy Analysis & Government Accountability, 2010; Forsyth, Asmus, Sto kes, & Forsyth , 2010). Further, the unavailability of yearly statewide assessment data not only prevents assessment of performance, but it promotes an educational climate where there is little impetus for change (Gagnon et al. 2010; Leone & Cutting, 2004).

PAGE 14

14 Despite legislation that requires juvenile correctional facilities to meet the same standards required of public schools, there are a number of issues that contribute to the underreporting of statewide assessment results (Blomberg, Blomberg, Waldo, Pesta , & Bellows, 2006). For instance, AYP and test scores may not be reported because students are often held within the facility for less than one year. Further, disability subgroups of students are often unreported due to their disaggregated numbers being be low mandated reportable thresholds (Leone & Cutting, 2004). Compounding the issue of underreporting, statewide assessments are based on grade level curricula that may not be addressed in multigrade classrooms typically utilized in juvenile correctional fac ilities (Leone & Cutting, 2004). Despite noted issues within juvenile corrections, there is no sufficient legal justification for denying these students equal access to high quality educational services, such as participation in statewide academic assessme nts (Gagnon et al., 2010). The lack of participation in and equal access to high quality educational services may not only inhibit successful community reintegration (Blomberg et al., 2006), but may also interfere with the attainment of a standard diploma. Despite survey results obtained by researchers Gagnon, Barber, Van Loan, and Leone (2009) indicating that principals within correctional settings felt that their primary responsibility was to help students earn a high school diploma, a vast majority of th ese students never receive a standard diploma (U.S. Department of Education, 2006). Documenting the performance of these youth on statewide assessments, particularly in the area of reading, is an important step in ensuring that educators have the informati on necessary to develop instructional practices to help prepare these youth for life beyond the confines of correctional facilitates. Across educational settings, the nationwide implementation of state testing and accountability practices has ensured that assessments are routinely utilized as a basis for making

PAGE 15

15 policy decisions, program and school funding, educational placements, disability classifications, and even as a prerequisite for graduation (Figlio & Getzler, 2002; Jacob, 2005; Shapiro et al., 2008). Notwithstanding a number of social, political, and philosophical issues (e.g., see Darling Hammond, 2004; Fritz berg, 2004; Gay, 2007), federal legislation has uneq uivocally furthered the discourse and quality of reading research and accountability practices. However, despite the wealth of research generated by NCLB that focuses on improving the reading and literacy skills of students (Foorman & Nixon, 2006), extant literature in regard to the reading ability of incarcerated youth has been largely underwhelming (Krezmien, Mulcahy, Travers, Wilson, & Wells, 2013). Understanding which specific academic and reading skills promote educational achievement, in respect to hi gh stakes testing, is critical as juvenile justice populations are overwhelming at risk for academic failure and disproportionally comprised of individuals who have disabilities. Although a plethora of literature exists in regard to high stakes assessment, research concerning the academic and reading skills related to high stakes test performance for students within the juvenile justice system is especially lacking. Reading Achievement and Statewide Assessments Reading and literacy skills are predictive of a number of factors including academic achievement, success on statewide assessments, and general life outcomes (Feifer, 2010; Fuchs & Fuchs, 2006; Lyon, 1998; Riddle Buly & Valencia, 2002; Stanovich, 1986). The ability to read, and to read well, is inextr icably related to school achievement, so much so that Snow, 18). Although the neurological processes underlying the ability to read are complex (Feifer, 2010; Goswa mi, 2007; Shaywitz & Shaywitz, 2005; Sousa, 2011), the ability to read written text can be conceptually divided into five core areas outlined by the National Reading Panel (2000).

PAGE 16

16 These areas include: phonological awareness, phonics (i.e., alphabetic princ iple), fluency, vocabulary, and reading comprehension (National Reading Panel, 2000). Phonological awareness and phonics are foundational components of fluent reading and the acquisition of vocabulary, whereas fluency and vocabulary directly contribute to understand and comprehend written text (LPA, 2004). These core areas and the overall ability to read are inextricably linked (Eldredge, 2005). Thus, mastery of these five areas is critical for demonstrating proficiency on reading port ions of statewide assessments. Children begin to develop proficiency in phonics and phonemic awareness at an early age (Carroll, Bowyer Crane, Duff, Hulme, & Snowling, 2011; Yopp, 1992). Due to the qualitative shift in reading development and academic stan dards which occurs around the third grade (i.e., instruction moves towards more cognitively complex tasks such as reading comprehension; Shapiro et al., 2008) , measures of oral reading fluency (ORF) and reading comprehension are often used as predictor var iables when studying performance on statewide assessments because they are theoretically aligned with reading speed and comprehension, which are requisite skills to demonstrate proficiency on statewide assessments. Several studies have examined the relati onships between ORF, reading comprehension, or both, and high stakes statewide assessments. However, there is still much to be learned. An overwhelming majority of research studies focus solely on populations in primary grades. Restricting samples to stude nts in primary grades, and primarily to measures of ORF, is problematic because the predictive validity of ORF on statewide assessments diminishes as students matriculate through school (Shapiro et al., 2008; Schatschneider et al., 2004; Tighe & Schatschne ider, 2013; Wood, 2006). Silberglitt, Burns, Madyun, and Lail (2006) noted that the relationship between ORF and the statewide Minnesota Comprehensive Assessments Reading

PAGE 17

17 (MCA R) decreased from .71 in third grade to .51 in eighth grade. While there is a mo derate as great a value for the purpose of predicting statewide achievement test scores in the later grades easures are less predictive of statewide achievement test performance as children age because even struggling readers typically demonstrate a reasonable mastery of phonemic awareness, phonics, and more automaticity in regard to fluency as they reach adoles cence (Roberts, Torgesen, Boardman, & Scammacca, 2008) . A narrow focus on ORF and comprehension may also be limiting as a number of reading skills including decoding, word recognition, vocabulary, fluency, word expression, comprehension, and overall readin g competence all contribute to success on statewide assessments across grade levels (Hosp & Fuchs, 2005; Riddle Buly & Valencia, 2002). Additionally, the relationships between both of these skills and statewide assessment performance changes as a function of the increased cognitive complexity required by statewide assessments as students age ( Schatscneider et al., 2004; Silberglitt et al., 2006; Tighe and Schatschneider, 2013). The concurrent and predictive validity of using ORF assessments as predictors of statewide high stakes assessments, especially those conducted throughout the primary grades, is documented extensively (Buck & Torgesen, 2003; Crawford, Tindal, & Stieber, 2001; Goffreda, 2009; Good et al., 2001; Hunley, Davies, & Miller, 2013; McGlinc hey & Hixson, 2004; Merino & Beckman, 2010; Roehrig, Petscher, Nettles, Hudson, & Torgesen, 2008; Schilling, Carlisle, Scott, & Zeng, 2007; Shapiro, Keller, Lutz, Edwards, & Hintze, 2006; Shapiro et al., 2008; Stage & Jacobsen, 2001; Wood, 2006). In respec t to ORF and statewide assessments across states, most studies report overall correlations of .60 to .75 (Shapiro et al., 2006). While ORF is strong

PAGE 18

18 predictor of statewide assessment performance as it is comprised of a blend of necessary word reading skill s, including phonics, word recognition, word meaning, and contextual clues, other academic variables also possess predictive utility in regard to statewide assessment performance (McGlinchey & Hixson, 2004). For instance, researchers have demonstrated that measures of reading comprehension are also reliable predictors of performance on statewide assessments, both in isolation as well as in conjunction with ORF measures (Allison & Johnson, 2011; Ardoin et al., 2004; Deno et al., 2009; Denton et al., 2011 ; Ma rcotte & Hintze, 2009; Merino & Ohmstede Beckman., 2010; Shapiro et al., 2008; Siberglitt, Burns, Madyun, & Lail 2006; Wiley & Deno, 2005). Interestingly, the predictive utility of using reading comprehension measures to predict high stakes assessment scor es increases steadily as students matriculate through the primary grades, due to the more demanding nature and cognitive complexity of statewide assessments as students age. Further, only a small number of research studies have focused on the relationshi ps among other important educational performance factors and statewide achievement tests. For example, chometric g ) become increasingly important predictors of success as students are administered the Florida Comprehensive Assessment Test (FCAT) from third grade to tenth grade. Interestingly, despite the relationship between intelligence and school achievem ent (Gottfredson, 2002; Jensen, 1969; Jensen, 2002; Rohde & Thompson, 2007; Watkins, Lei, & Canivez, 2007), the complex interactions between age, disability, intelligence, and the predictive utility of reading skills on statewide assessment performance has not been sufficiently established.

PAGE 19

19 Last, there is little research regarding the predictive utility of specific reading skills on statewide assessments for students who have disabilities. This is troublesome considering that historically students with dis abilities demonstrate poor performance across statewide high stakes assessments (Altman, Vang, & Thurlow, 2012; Katsiyannis, Zhang, Ryan, & Jones, 2007). When students who have disabilities are included in research samples, the data are seldom disaggregate d by special education classification. The aggregation of all disability classifications erroneously categorizes individuals belonging to one or more of 13 categories outlined by IDEIA (i.e., Specific Learning Disability [SLD], Intellectual Disability [In D], Emotional Disturbance [ED], Deaf or Hard of Hearing [DHH], Visually Impaired [VI], Dual Sensory Impaired [DSI], Orthopedic Impairment [OI], Other Health Impairment [OHI], Traumatic Brain Injury [TBI], Speech Impairment [SI], Language Impairment [LI], A utism Spectrum Disorder [ASD], Developmentally Delayed [DD]) as a single homogenous group. Aggregating achievement data for students who have disabilities into a singular category ignores the inherent differences in intellectual, academic, and behavioral functioning found across disability categories (Caffrey & Fuchs, 2007; Saborine, Cullinan, Osborne, & Brock, 2005). In a meta analysis of 27 studies which examined the predictive utility of curriculum based measures (CBM) on statewide assessments from 14 s tates, Yeo (2009) noted that the characteristics of students who have disabilities were found to act as a moderator between CBM and performance on statewide achievement tests. Yeo (2009) reported that studies containing large proportions of students with d isabilities produced weaker correlation coefficients than did studies with lower proportions of these students. Further, Yeo (2009) found a correlation of .31 between sample size and the overall proportions of students who have disabilities within the met a analysis, indicating that the number of studies possessing sample sizes with high

PAGE 20

20 proportions of students with disabilities was small. These results prompted Yeo (2009) to call for further research with larger sample sizes comprised of higher proportions of special education need to include data for youth with special needs wit hin studies of reading and to disaggregate the data categorically. Understanding the unique relationships between literacy skills and disability categories are particularly important considering that less than 20% of high school students who have disabilit ies receive satisfactory scores on high stakes statewide assessments (Thurlow & Wiley, 2006). Given that students who have disabilities are exceptionally vulnerable to the possible consequences of failing high stakes assessments, it is important that attem pts are made to address their respective academic deficiencies in an effort to increase proficiency levels (Katsiyannis et al., 2007). More information regarding the specific skills of these students could lead to the development and implementation of mean ingful academic interventions aimed at preventing failure on statewide assessments. Although a plethora of literature exists in respect to high stakes assessment, research concerning how specific reading skills relate to high stakes test performance for st udents within the juvenile justice system is especially lacking. The following section describes the academic, social, and psychiatric characteristics of typical populations confined within juvenile correctional facilities. A section providing a comprehens ive review of salient studies regarding the FCAT, an NCLB accountability assessment, follows. High Stakes Testing and Youth in Juvenile Corrections Amidst deep cuts in federal funding to the Department for Juvenile Justice, more than 144,000 adolescents ar e committed to juvenile correctional facilities annually (Snyder & Sickmund, 2006). This number indicates an increase of 44% over the last two decades (Gagnon

PAGE 21

21 & Richards, 2008). The rapid increase in youth housed in correctional facilities presents numerou s societal and financial costs. Although accurately approximating social costs is difficult, the American Correctional Association estimated that in 2008 the per diem financial cost of detaining a juvenile in a residential setting was $241.29, and that one year costs were approximately $88,000 and rising (Petteruti, Walsh, & Velazquez, 2009). Estimates of the per diem expenses for incarcerated juveniles within the United States are upwards of $6 billion each year ( Sickmund, Sladky, & Kang, 2008 ). To put the costs of incarceration into perspective, in 2010 the state of Florida spent an average of $8,741 per student attending public school (Dixon, 2012). Further, it is estimated that the typical career criminal costs taxpayers between $2.6 and $5.3 million (Co hen & Piquero, 2009). Considering that a majority of juvenile offenders are expected to reenter the juvenile system or enter the adult correctional system, current recidivism rates ensure a persistent culture of crime and failure for these youth (Bullis, Y ovanoff, & Havel, 2004; Myner, Santman, Cappelletty, & Perlmutter, 1998). Characteristic of Youth in Juvenile Corrections Adolescent males present the highest risk for both committing crimes and becoming victims of crime themselves (Archwamety & Katsiyann is, 2000). In a study conducted by Zhang, Hsu, Katsiyannis, Barrett, and Ju (2011), the researchers noted an average age of approximately 14 years for first time offenders referred to the juvenile justice system. Incarcerated adolescents can be detained fo r a variety of offenses, including violent felonies, property crimes, crimes against people, misdemeanor offenses, and status related offenses (Shelley & Langhinrichsen Rohling, 2007). Research indicates that youth confined within juveni le correctional facilities present disturbingly analogous demographic characteristics (Wilson, 2013). These youth are disproportionately male, minority, from low socioeconomic backgrounds, and demonstrate significant learning and behavioral problems that m ake them disproportionately

PAGE 22

22 eligible for special education services (Leone, Meisel, & Drakeford, 2002). Concerning minority groups, arrest data indicate that youth of color, especially those identifying as African American, are overrepresented within juven ile justice populations (Jenson, Potter, & Howard, 2001; OJJDP, 2012; Snyder & Sickmund, 2006). In regard to academic ability, incarcerated youth are among the least academically competent students in the United States (Gagnon et al., 2013). The educationa l deficits exhibited by incarcerated youth are in part explained by various levels of educational disenfranchisement and in part by adverse demographic characteristics (Leone, Krezmien, Mason, & Meisel, 2005; Myner et al., 1998); both are significant barri ers to successful reintegration of juvenile offenders into society. These youth often exhibit a complex array of educational, social, behavioral, and psychiatric issues (Gagnon & Barber, 2010). Approximately 70% of youth in the juvenile justice system me et diagnostic criteria for at least one mental health disorder (Shufelt & Cocozza, 2006). Specifically, delinquent adolescents exhibit high rates of attention deficit/hyperactivity disorder (ADHD) and conduct disorder (CD); they are also at higher risk for antisocial behavior and criminality in adulthood, and are especially at risk for the development of personality disorders (e.g., borderline and antisocial personality disorders) (Bussing, Mason, Bell, Porter, and Garvan, 2010; Elmund, Melin, von Knorring, Proos, & Tuvenmo, 2004; Vermeiren, Jespers, & Moffitt, 2006) . Further, youth in secure care typically have especially high rates of comorbid psychiatric disorders (Abram, Teplin, McClelland, & Duncan, 2003; Kroll et al., 2002; Shufelt & Cocozza, 2006). De linquent youth also have disproportionate experience with abuse, neglect, and violence (Gagnon & Barber, 2010). Youth in secure care often exhibit neuropsychological profiles that are consistent with low levels of achievement and rife with deficiencies in cognitive functioning. An inverse

PAGE 23

23 relationship between intelligence and delinquency is documented extensively (Elmund et al., 2004; Foley, 2001; Lynam, Moffitt, & Stouthamer Loeber, 1993; Moffit & Silva, 1998 ). Lynam et al. (1993) postulated that the dire ction of effect of the well established link between IQ and delinquency runs from low IQ to delinquency, and found no evidence that the link was spurious in nature or that factors related to delinquency lead to lower IQ scores. Studies of delinquent youth indicate that these individuals score within the below average to average levels on standardized tests of intelligence, and often achieve scores of as much as one standard deviation below the mean when compared to non incarcerated peers (Foley, 2001; Elmun d et al., 2004; Rincker, 1990). Additionally, these youth exhibit a significantly higher prevalence of intellectual disabilities than their non incarcerated peers (Kroll et al., 2002). Researchers have also established a relationship between delinquency an d both cognitive and behavioral impulsivity (White et al., 1994). Delinquent youth are also prone to hyperactivity and attention problems (Lynam, 1996), more likely to exhibit impairments in memory and spatial ability (Raine et al., 2005), and deficits in effortful control and prosocial skills (e.g., empathy) (Jiron, 2010). Further, the psychosocial and psychiatric issues presented by incarcerated youth are often compounded by the overrepresentation of youth who have one or multiple educational disabilities . Incarcerated Youth and Special Education Disability Educationally, incarcerated youth are qualitatively different from their non incarcerated peers. The prevalence of students who have disabilities in juvenile correctional facilities greatly exceeds t he national average. Specifically, prevalence rates of disability within juvenile correctional facilities across states are estimated to range from about 9% to 77%, with a median of approximately 33% (Quinn, Rutherford, Leone, Osher, & Poirier, 2005), whil e the national average of youth who have educational disabilities is approximately 13% (Snyder & Dillow, 2013). Of the 13 categories of disability defined by IDEIA (2006), delinquent youth most often

PAGE 24

24 are classified as having an ED, SLD, or both (Quinn, Rut herford, & Leone, 2001; U.S. Department of Education, 2011). Quinn et al., (2005) reported that youth who have ED or SLD comprise approximately 47.7% and 38.6% of youth with disabilities in the juvenile correctional populations, respectively. Comorbidity e stimates reported indicate that approximately 24% to 54% of adolescents with SLD meet classification criteria for ED, while 38% to 75% of adolescents identified with ED meet criteria for SLD (Rock, Fessler, & Church, 1997). The large numbers of ED youth in secure care is especially troubling for educators. Hayling, Cook, Gresham, State, and Kern (2008) reported that students with ED educated in residential settings performed significantly lower across academic areas than students with ED in other settings. In a survey conducted by Wilkerson, Gagnon, Mason Williams, and Lane (2012), juvenile correctional teachers reported that at minimum 50% of their students with high incidence disabilities could not read well enough to glean rudimentary information from tex t. Harris, Baltodano, Bal, Jolivette, and Malchy (2009) studied the performance of 398 juveniles, ages 14 to 17 years, on three Woodcock Johnson III Tests of Achievement (WJ III ACH) subtests who were incarcerated in three different states. According to H arris et al. (2009), individuals within the sample scored approximately one standard deviation below the mean across subtests. Harris et al. (2009) also found significant performance differences between students with educational disabilities and nondisable d students within the sample (i.e., between one and two standard deviations). On the Letter Word Identification subtest general education students attained an average standard score of 90.1 ( SD = 15.9), while students with special education exceptionalities achieved an average standard score of 77.1 ( SD = 20.4). On the Word Attack subtest general education students achieved an average standard score of 90.1 ( SD = 15.7), compared to an average of 77.6 ( SD = 19.3) for students receiving special education

PAGE 25

25 services. Similarly, incarcerated general education students ( M = 87.1; SD = 16.0) significantly outperformed their peers ( M = 77.2; SD = 19.2) with special education exceptionalities on the Pa ssage Comprehension subtest. Harris et al. (2009) noted that on Letter Word Identification and Word Attack, 9% of the variance in scores was accounted for by student special education status alone, whereas 7% was attributed to special education status on t he Passage Comprehension subtest. In respect to specific special education exceptionality subtypes, Harris et al. (2009) reported that students with a primary disability label of ED outperformed those with a primary label of SLD, who subsequently outperfor med students with a primary label of InD across all subtests. Likewise, on the same three subtests of the WJ III ACH, Krezmien et al. (2013) found that special education students scored significant lower than their incarcerated general education peers . The deficient academic performance of youth in secure care settings with special education disabilities often manifest outside of correctional facility walls. For example, Bullis and Yovanoff (2006) found that youth with educational disabilities were 2.5 time s less likely to be employed and received a lower hourly wage than delinquent youth without educational disabilities after release . Incarcerated Youth and Reading Ability The long term social and economic trajectory of poor readers is concerning. For example, 43% of adults with the lowest literacy skills are impoverished; likewise, 70% of these individuals do not hold a full or part time job (Humboldt Literacy Project, 200 7). A large proportion of incarcerated juveniles are often only marginally literate, have often experienced school failure, and many have been subject to grade retention (Center on Crime, Communities, and Culture, 1997). The average reading ability of thes e youth ranges from fourth to ninth grade (Baltodano, Harris, & Rutherford, 2005; Brunner, 1993; Foley 2001; Hodges, Giuliotti, & Porpotage, 1994). However, regardless of the mean reading level demonstrated by respective populations, these

PAGE 26

26 youth are often several years below grade level expectations in regard to reading ability (Foley, 2001). O n standardized achievement tests, juvenile justice populations often perform one standard deviation or more below the overall population mean (Harris et al., 2009; Kr ezmien, Mulcahy, & Leone, 2008; Snowling, Adams, Bowyer Crane, & Tobin, 2000). For example, one study indicated that only 66% of students achieved scores above the first grade reading level when the r eading comprehension of 71 juvenile justice students, 13 to 18 years of age, across four states was assessed (Drakeford & Krezmien, 2004, as cited by Houchins, Jolivette, Krezmien, & Baltodano, 2008). In regard to specific reading skills, Krezmien et al. (2013) found that the mean performance of 533 males withi n a juvenile correctional facility on selected subtests of the WJ III ACH was uniformly more than one standard deviation below the expected mean of the general population. Krezmien et al. (2013) reported the overall standard scores of subtests performance for participants as follows: Letter Word Identification 80.9 ( SD = 18.0), Word Attack 84.0 ( SD = 18.2), and Passage Comprehension 81.1 ( SD = 17.7). Krezmien et al. (2013) also noted a disparity between the expected and actual performance of African America n students in both general and special education. More than 33% of African American general education students achieved a standard score below 85 across subtests, more than double what would be expected given the normative distribution of the assessment me asure. Likewise, 50% of African American students in special education attained a standard score below 85, tripling what would be expected. Conversely, researchers noted that 15% of White general education students achieved below 85, matching the expected distribution. While only 5.6% to 8.7% of White students in special education achieved a standard score below 85. Krezmien et al. (2013) reported that regression models containing age, race, and special education significantly

PAGE 27

27 predicted performance on the s ubtest as follows: Letter Word Identification ( R 2 = .16), Word Attack ( R 2 = .15), and Passage Comprehension ( R 2 = .13), indicating that comparatively older students performed more poorly than their younger counterparts across all subtests. Finally, Krezmie n et al. (2013) reported a negative predictive relationship between age and reading performance, as well as between special education and reading performance, indicating that comparatively older students and students within special education performed more poorly than their counterparts across all subtests On statewide high stakes assessments in Louisiana, Forsyth et al. (2010) found that juvenile delinquents performed significantly lower than non incarcerated students across Grades 3 to 11. Specifically, r esearchers noted that only 30.2% of juvenile delinquents demonstrated at bility to read well is demonstrative of the poor educational attainment and lack of grade promotion characteristically found among youth in correctional settings. Likewise, concerning educational attainment, a recent study conducted by Cavendish (2013) in the state of Florida ( N = 4,066) yielded an average GPA range of 1.76 to 1.91 among delinquent youth across subject areas, There is an inextricable link between poor reading achievement and deleterious outcomes f or delinquent youth, especially in regard to recidivism rates (Brunner, 1993; Katsiyannis, Ryan, Zhang, & Spann, 2008; Unruh, Gau, Waintrup, 2009; Vacca, 2008). As such, the current j uvenile hs who have reading (Christle & Yell, 2008, p. 148). Despite the knowledge that incarcerated youth are

PAGE 28

28 youth is largely nonexistent (Krezmien et al., 2 013). Developing appropriate reading remediation strategies for delinquent youth is contingent on first determining what skills these individuals possess and which skills they lack. Additionally, it is important to establish which of these skills are most predictive of demonstrating proficiency on statewide high stakes assessments, as these tests are indicative of overall reading ability and are often required for grade promotion and to receive a standard diploma upon high school graduation. The Florida Co mprehensive Assessment Test (FCAT) One such statewide assessment is the Florida Comprehensive Assessment Test (FCAT). ncarcerated juveniles as they are held to the same general stand ards as adolescents who are educated in traditional settings. Expectedly, students within the juvenile justice system score significantly lower on this assessment across grades compared to non incarcerated students. Data from the 2007 2008 school year prov ided by the Florida Office of Program Policy Analysis & Government Accountability (2010) indicate that only 14% of incarcerated students passed the FCAT on average across grades, compared to 64% of their non incarcerated peers. Understanding the specific r eading abilities possessed by students within the juvenile justice system, and which abilities are necessary to demonstrate sufficient knowledge on the FCAT may help guide instruction and intervention across grades for these students. Beginning in 1995, t he Florida Educational Reform and Accountability Commission recommended the development of a statewide assessment system designed to increase educational achievement for students across the state. These recommendations, known as the Florida Comprehensive A ssessment Design, subsequently produced the Sunshine State Standards (SSS) and led to the development of the Florida Comprehensive Assessment Test® (FCAT) (State of Florida Department of State, 2007) . The FCAT is a high stakes annual assessment

PAGE 29

29 designed to increase student achievement by implementing higher educational standards across the state. The FCAT was first administered statewide in 1998, and has continued every year since its inception. The FCAT aligns with NCLB as a form of statewide accountabili ty for student achievement. The FCAT is a challenging test, and is generally accepted to be among the more comprehensive high stakes statewide assessments (Figlio & Getzle, 2002). The test is administered every spring, typically over a two week period. The FCAT is comprised of reading, math, writing, and science content. Passing the reading section of the FCAT is required for promotion in third grade and graduation in the tenth grade. Students who fail to pass either the reading or mathematics section of th e FCAT may retake either section up to five times prior to graduation. If the student fails to complete the FCAT during the maximum number of allotted attempts they are issued a certificate of completion in lieu of a standard diploma. The FCAT is criterio n based assessment, with benchmarks set by experts through an extensive item construction process. The reading portion of the FCAT is group administered and contains 6 to 8 informational and literary reading passages (FL DOE, 2005). For example, the readin g section of the FCAT 2.0 is comprised of the following content: vocabulary (15 25%), reading application (approximately 20 35%), fiction and nonfiction literary analysis (approximately 20 35%), and informational text and research process approximately (15 35%) across grade levels (FLDOE, 2011). The FCAT Reading section is comprised of multiple choice items and short and extended response items; these sections contain a gradient of questions which vary in complexity (FLDOE, 2005). Of note, beginning in 201 1, the Florida Department of Education released a more rigorous version of the FCAT, FCAT 2.0. The increased rigor must not be understated, when cut points were revised in 2012 the percentage of Grade 3 students meeting grade level standards on the FCAT 2. 0 fell 19% from the previous year (Foorman,

PAGE 30

30 Kershaw, & Petscher, 2013. Due to structural changes, the FCAT 2.0 no longer utilizes the Sunshine State Standard Score (FCAT SSS) from the FCAT 1.0 as the primary outcome score for students. The FL DOE now repor ts the Developmental Scale Score (FCAT DSS) from FCAT 2.0 as the primary outcome score. For the remainder of the study, when distinctions are necessary (e.g., when discussing prior research), FCAT SSS Reading is used when referring to the FCAT 1.0, and FCA T DSS Reading will is used when referring to the FCAT 2.0. Student scores on the FCAT are interpreted through the use of scaled scores which range from 100 to 500 at each grade level, and provide a subsequent designation as to l evel of content mastery (e. g., L evel 1 5) (FLDOE, 2005). FCAT 2.0 achievement level designations are illustrated by the following (FLDOE, 2014): Level 1 (indicates an inadequate level of success with the challenging content of the Next Generation Sunshine State Standards [NGSSS]), Level 2 (indicates a below satisfactory level of success with the challenging content of the NGSSS), Level 3 (indicates a satisfactory level of success with the challenging content of the NGSSS), Level 4 (indicates an above satisfactory level of success wi th the challenging content of the NGSSS), Level 5 (indicates mastery of the most challenging content of the NGSSS) (p. 1 2). Until the end of the 2009 2010 school year, the last year of FCAT 1.0 administration, the following score ranges represented achiev ement of a Level 3/satisfactory score on the FCAT (SSS): Grade 6 (296 338), Grade 7 (344 388), Grade 8 (350 393), Grade 9 (354 381), Grade 10 (355 371). Although the FCAT 2.0 measures student achievement based on academic standards created by the state ( i.e., NGSSS), as previously mentioned, the FCAT 2.0 scoring metric assigned to one of five achievement level categories. The FCAT 2.0 (DSS) Developmental Scale S cores (DSS) Level 3/satisfactory score ranges are as follows: Grade 6 (222 236), Grade 7 (228 242), Grade 8 (235 248), Grade 9 (240 252), and Grade 10 (245 255) (FLDOE, 2012a).

PAGE 31

3 1 Academic Performance Variables and the FCAT SSS Reading Buck and Torgesen (20 03) used one minute measures of ORF to determine whether ORF measures are valid and reliable predictors of FCAT SSS Reading scores. They administered ORF prompts to third grade students from 13 schools in the state of Florida ( N = 1102) and used ORF benchm arks (i.e., high risk [< 80 correct words per minute], some risk [80 109 cwpm], and to determine sensitivity and specificity in regard to FCAT SSS Reading performance by level. The researchers reported a correlation of .70 between ORF scores and FCAT SSS Reading performance. Additionally, the authors found that the ORF benchmarks reliably predicted 91% of students who attained a Level 3 or higher, and 81% of students wh o achieved at levels 1 or 2. This study provides important information regarding the utility of ORF prompts as predictors of FCAT SSS Reading performance. Although 19% of the students in the research sample were classified as having an educational disabili ty under IDEIA, data for all participants were collapsed into a singular group. Another limitation is that this study only provides information in relation to the performance of third grade students. Exploring how these relationships differ amongst grade l evels and disability categories could provide valuable information to educators regarding group performance. Schatscneider et al. (2004) examined multiple academic performance variables within the domains of reading, language, and cognitive ability to determine the predictive utility of these variables in respect to FCAT SSS Reading success. These variables inclu ded: verbal knowledge and reasoning (i.e., Wechsler Abbreviated Scale of Intelligence [WASI] Vocabulary and Similarities subtests; listening comprehension with FCAT passages), text reading fluency (i.e., ORF FCAT passages, grade level texts, Gray Oral Read ing Test), phonemic decoding efficiency (i.e., Test of Word Reading Efficiency [TOWRE] Phonemic Decoding Efficiency subtest),

PAGE 32

32 nonverbal reasoning (i.e., WASI Matrix Reasoning and Block Design subtests), and working memory (i.e., an adapted version of the C ompeting Language Processing Task including Reading Span and Listening Span measures). Utilizing dominance analysis (DA), the researchers reported that during third grade, ORF was the dominant factor in explaining performance differences on the FCAT SSS Re ading. Specifically, Schatscneider et al. (2004) reported that when measured individually, ORF accounted for 56% of the reported variance, verbal knowledge and reasoning accounted for 44%, while 25% and 14% were accounted for by nonverbal reasoning and wor king memory, respectively. At the seventh grade, the researchers found that fluency and verbal knowledge were shown to be codominant. Lastly, in tenth grade Schatschneider et al. (2004) reported that verbal knowledge and reasoning was the dominant factor a ccounting or 52% of the variance, while fluency accounted for 32% of the total variance. Additionally, the researchers reported that FCAT SSS Reading comprehension levels were comparable to reading comprehension scores from the Stanford Achievement Test Ni nth Edition (SAT 9) test, which is nationally standardized. The correlations between the FCAT SSS Reading and SAT 9 scores were .86, .78, and .74 at Grades 3, 7, and 10, respectively; indicating that the FCAT SSS Reading is a valid measure of reading abili ty and thus comparable to other statewide assessments. Expanding upon Schatschneider et al. (2004) , Tighe and Schatschneider (2013) utilized the same dataset to provide a more detailed analysis in regard to the requisite skills for FCAT SSS and SAT 9 succ ess across grades. Specifically, they employed exploratory factor analysis (EFA), utilizing principal axis factoring (PAF) and dominance analysis (DA), to identify significant predictors of reading comprehension (i.e., performance on the FCAT SSS and SAT 9 ). They included 12 variables (i.e., TOWRE [PDE and SWE], three reading fluency composites

PAGE 33

33 [AIMSweb, textbook passages from the state adoption list, FCAT SSS practice items], WASI [Similarities, Vocabulary, Matrix Reasoning, Block Design], and the R eading Span and Listening Span measures tenth grade models. A four factor solution accounting for 82% of the variance (i.e., of the original measures) was selected for third grade. The factors, l isted in order of highest eigenvalues, were as follows: fluency (AIMSweb, textbook passages, FCATS SSS practice items, and TOWRE), verbal reasoning (listening comprehension passages from the FCAT, Similarities, Vocabulary), nonverbal reasoning (Matrix Reas oning and Block Design), and working memory (Reading and Listening Span). The authors noted that all factors were moderately to highly correlated with both FCAT SSS and SAT 9 performance (.57 to .78). The four factor third grade model accounted for 71% of the total variance in regard to FCAT SSS and SAT 9 performance. Specifically, fluency uniquely accounted for 61% of the variance, verbal reasoning uniquely accounted for 9% beyond fluency, nonverbal reasoning accounted for 1% when added to the model, and w orking memory did not uniquely account for any variance. Fluency exhibited complete dominance over nonverbal reasoning and working memory across all subset models generated through the DA procedure; however, complete dominance could not be established betw een fluency and verbal reasoning. For seventh grade, a three factor model accounting for 74% of the variance was selected. The selected factors for seventh grade were as follows: reasoning (WASI), fluency (AIMSweb, textbook passages, FCATS SSS practice ite ms, and TOWRE), and working memory (Reading and Listening span). All three factors were moderately to highly correlated with FCAT SSS and SAT 9 performance (.31 to .74). The three factor seventh grade model accounted for 63% of the total variance in regard to FCAT SSS and SAT 9 performance. Both reasoning and fluency exhibited complete dominance over working

PAGE 34

34 memory; however, reasoning did not completely dominate fluency. Reasoning was found to be the strongest predictor of FCAT SSS and SAT 9 performance as it explained more unique variance (17%) when controlling for all other predictors than did fluency (11%). Likewise, a three factor model accounting for 71% of the variance was selected for tenth grade. The selected factors for tenth grade were as follows: fluency (AIMSweb, textbook passages, FCATS SSS practice items, and TOWRE), reasoning (WASI and listening comprehension passages from the FCAT), and working memory (Reading and Listening Span). All three factors were moderately to highly correlated with FCA T SSS and SAT 9 performance (.48 to .80). The three factor tenth grade model accounted for 68% of the total variance in regard to FCAT SSS and SAT 9 performance. Reasoning was completely dominant of both fluency and working memory; while fluency exhibited complete dominance over working memory. Reasoning was found to be the strongest predictor of FCAT SSS and SAT 9 performance as it explained more unique variance (30%) when controlling for all other predictors than did fluency (7%). Schatschneider et al. (2 004) and Tighe and Schatschneider (2013) demonstrated that general cognitive ability (i.e., psychometric g ) are increasingly dominant predictors of success as students take more complex versions of the FCAT. While these studies provide a solid foundation f or studying FCAT performance differences across grade levels, there are a number of limitations. Specifically, it is unclear if students with disabilities were included in the sample. Moreover, if these students were included, the data were not disaggregat ed in a way that would allow for the examination of achievement abilities across groups. This study also omitted Grades 6, 8, and 9. Understanding the skills required at these grades may prove beneficial to researchers, as well as educators. Although this study was to be replicated in 2004, to examine the specific performance deficits of students at levels 1 and 2 on the FCAT, it was not.

PAGE 35

35 Similar to Buck and Torgesen (2003), Schatschneider et al. (2004), and Tighe and Schatschneider (2013), Torgesen, Nettl es, Howard, and Winterbottom (2003) found a predictive relationship between ORF prompts and FCAT SSS Reading scores. The correlations ranged between .55 and .62 across Grades 4 ( n = 88), 6 ( n = 252), 8 ( n = 161), and 10 ( n = 98). They hypothesized that a s mall sample size likely contributed to the lower relationship found between ORF and FCAT SSS Reading scores compared to the relationships reported in similar studies reported by Buck and Torgesen (2003), Schatschneider et al. (2004), and Tighe and Schatsch neider (2013) . They also found a relationship between maze passages constructed from FCAT passages and FCAT SSS Reading scores. Specifically, the authors noted the following correlations between maze prompts and FCAT SSS Reading: fourth grade (.54), sixth grade (.67), eighth grade (.63), and tenth grade (.32). This study provides important information regarding the relationships between ORF and maze prompts as predictors of performance for the FCAT SSS Reading across grades. However, this study either did n ot include students with disabilities or if these students were included in the sample, the data were not disaggregated. This study also only provided simple correlations, the use of more advanced statistical modeling techniques could have provided more in formation in regard to prediction of FCAT SSS Reading success. Roehrig et al. (2008) examined the validity of the Dynamic Indicators of Basic Early Literacy Skills (DIBELS) ORF for predictive bias in relation to FCAT SSS Reading and Stanford Achievement T est (SAT 10) outcomes. The sample included over 17,000 third grade students. The researchers reported correlational coefficients between ORF and both the FCAT SSS Reading and SAT 10 ranging between .66 and .71. These correlations indicate that the ORF is a good measure of predictive performance on a reading comprehension measure used across

PAGE 36

36 states, as well as the FCAT. Using logistic regression models the researchers reported no evidence of bias in relation to a number of demographic groups (i.e., races/eth nicities, socioeconomic status, and language status). Although special education students were included in the sample, the data were not disaggregated by special education classification. Examining special education student data would have provided useful data to examine the relationship between special education status, ORF, and FCAT SSS Reading performance. Similar to other studies, only third grade students were included in this sample. Tannenbaum, Torgesen, and Wagner (2006) explored the relationship between multiple measures of reading comprehension, fluency, and word knowledge. These included measures of word knowledge (i.e., PPVT III, WISC III Vocabulary subtest, and the Multiple Mean ings and Attributes subtests of the LPT R, WUF subtest of DIBELS, and Semantic Category Fluency [specially designed for the study]) and reading comprehension measures (i.e., FCAT SSS Reading, and SAT 9 Reading Comprehension Test) were explored. The tests w ere administered to 203 third grade students. Although the study focused on the relationship between measures of word knowledge and reading comprehension, information relevant to the current study was also provided. Specifically the authors provided the fo llowing correlations between the given measures and the FCAT SSS Reading: PPVT III (.54), WISC III Vocabulary (.61), and SAT 9 Reading (.78). However, the correlations were only provided for third grade students. Although these correlations were moderate i n nature, the relationship between these variables likely changes as the FCAT SSS Reading increases in complexity in higher grades. Similarly to other studies, Tannenbaum et al. (2006) did not disaggregate data by disability categories. Researchers Stan ley and Stanley (2011) used the Reading Level Indicator (RLI) to predict FCAT SSS Reading performance. The RLI is a 10 15 minute paper and pencil test

PAGE 37

37 administered in the classroom. The RLI is purported to measure reading comprehension and vocabulary simil ar to the FCAT SSS Reading. The RLI consists of 20 sentence comprehension (i.e., similar to a maze format) items and 20 vocabulary items (i.e., multiple choice). The authors assessed RLI performance in relation to FCAT SSS Reading performance on fourth gra de students ( N = 38). The researchers found a correlation of .79 between the RLI and the FCAT SSS Reading. Using regression analysis, Stanley et al. (2011) reported a moderately strong linear relationship of R 2 = .63. This indicates that approximately 63% of the variance in FCAT SSS Reading scores may be attributed to the constructs measured by the RLI. The researchers also developed RLI cut point categories to predict students who may be at risk for failing the FCAT SSS Reading. Specifically, Stanley and S tanley (2011) noted the following RLI scoring categories for fourth grade students: High Risk (0 17), Average Risk (18 25), and Low Risk (26 40). Although Stanley and Stanley (2011) noted the utility of using the RLI for prediction of FCAT SSS Reading perf ormance, the sample was restricted to students within the fourth grade and did not include students with special education exceptionalities. Thus, results from this study in regard to the validity of using RLI scores to predict FCAT SSS Reading scores are only representative of regular education students who are in the fourth grade. Additional research is needed to understand the differences in FCAT SSS Reading performance across grade and disability category. Algozzine, Wang, and Boukhtiarov (2011) assesse d the predictive validity of both the STAR Reading 2.0 (STAR Reading) and the Scholastic Reading Inventory Interactive (SRI I) in respect to performance on the FCAT SSS Reading. The STAR Reading assessment is a norm and criterion referenced adaptive compu terized test designed to quickly determine appropriate student instructional level. It is available for students across all grade levels. The authors noted

PAGE 38

38 that the technical adequacy of the test is well documented and that STAR Reading scores closely corr elate to other widely used reading measures. The SRI I is also computerized, adaptive, and measures reading comprehension level. Similarly to the STAR Reading, the SRI strong internal and external validity, and operates on the Lexile framewor k. The authors noted the following correlations between the STAR Reading and the FCAT SSS Reading: Grade 6 (.75), Grade 7 (.61), Grade 8 (.71). Likewise, the authors found the following correlations between the SRI I and the FCAT SSS Reading: Grade 6 (.76) , Grade 7 (.58), Grade 8 (.68). Algozzine et al. (2011) used multiple linear regression to further explore the relationship between predictor variables and the FCAT SSS Reading. The authors included gender, ethnicity, free/reduced lunch status, and the res pective assessment instruments as independent variables within stepwise multiple regression models. For male participants, the authors reported the following predictive models in regard to FCAT SSS Reading performance: R 2 = .50 for Model 1 (STAR predicting FCAT SSS Reading), R 2 = .57 for Model 2 (STAR and SRI I predicting FCAT SSS Reading), and R 2 = .58 for Model 3 (STAR, SRI I, and race predicting FCAT SSS Reading). The authors reported an adjusted R 2 R 2 = .50 for Mod R 2 R 2 R 2 across models was statistically significant. The researchers also employed predictive discriminant analysis (PDA) to assess how well predetermined cut points on the STAR Reading and SRI I predict student performance on the FCAT SSS Reading based on FCAT Achievement Levels. The authors reported that both the SRI I and STAR Reading were accurate in predicting performance demonstrated by an average hit rate of 76% across grades. Algozzine et al. (2011) also reported sensitivity rates ranging from 62% to 85% and specificity rates ranging from 75% to 99% for the or SRI I. Overall, the authors concluded that both assessments were strong

PAGE 39

39 predictors of the FCAT SSS Reading sco res across grades and adequately predicted FCAT SSS Reading Level proficiency across grades. Although this study provides evidence regarding the predictive utility of both progress monitoring tools beyond the primary grades, a number of limitations are pr esent. For instance it is unclear if students with exceptionalities were present, and if they were present the data were not disaggregated to represent their inclusion. Additionally, this study did not explore the specific skills needed to demonstrate prof iciency on the reading section of the FCAT SSS. Considering that the state of Florida currently uses the Florida Assessments for Instruction in Reading (FAIR) to predict FCAT success, more research is needed regarding the instruments predictive utility. Comparing the FAIR to an established instrument, such as the SRI I, may E thnicity and the FCAT DSS Reading According to data obtained from the Florida Department of Education (2012), males between Grades 6 to 10 achieved a Level 3 or higher (i.e., Level > 3 indicates a passing score) on the 2012 administration of the FCAT DSS R eading at a rate of 52.6% with a mean DSS of 234.2. Disaggregated ethnicity data for males achieving a Level 3 or higher on the FCAT DSS Reading are expressed by rate as follows: Asian (70.8%), White (64.4%), Multiracial (58.8%), American Indian or Alaska Native (52.8%), Native Hawaiian or Other Pacific Islander (51.2%), Hispanic/Latino (48%), Unknown Race/Ethnicity (36.4%), and African American (31.2%) (FLDOE, 2012b). Likewise, the disaggregated mean DSS on the FCAT DSS Reading for males between Grades 6 t o 10 are as follows: Asian (244.6), White (240.4), Multiracial (237.6), Native Hawaiian or Other Pacific Islander (234.2), American Indian or Alaska Native (233.8), Hispanic/Latino (231.2), Unknown Race/Ethnicity (224.4), and African American (224.4) (FLDO E, 2012b).

PAGE 40

40 Exceptional Student Status (ESE) and the FCAT DSS Reading In the state of Florida, students who meet eligibility criteria for one or more of the disability categories outlined by IDEIA are given an Exceptional Student Status (ESE) designation. According to data obtained from the Florida Department of Education (2012), male students without disabilities between Grades 6 to 10, including students with a gifted classification, achieved a Level 3 or higher on the 2012 administration of the FCAT DSS Reading at a rate of 58.6% with a mean DSS of 237.8. Contrastingly, students with disabilities achieved a Level 3 or higher at a rate of 22% with a mean DSS of 217.8 (FLDOE, 2012b). Disaggregated data by ESE disability classifications for individuals achi eving a Level 3 or higher on the FCAT are expressed by rate as follows: SI (55%), VI (49.6%), ASD (43%), OI (36%), OHI (24.8%), DHH (23.6%), ED (22.6%), SLD (19.8%), TBI (19.2%), LI (12%), and InD (1%) (FLDOE, 2012b). Disaggregated mean DSS on the FCAT DSS Reading for individuals between Grades 6 to 10 are as follows: SI (236.0), ASD (227.2), VI (231.2), OI (224.2), OHI (219.8), SLD (216.6), ED (216.4), DHH (216.2), TBI (215.6), LI (213.6), and InD (154.4) (FLDOE, 2012b). Estimates were not available for Du al Sensory Impaired (DSI), Developmentally Delayed (DD), or students in Grade 10 with an InD designation. FAIR Progress Monitoring and the FCAT Reading The Florida Assessments for Instruction in Reading (FAIR) is a statewide progress monitoring tool desig ned to help guide reading instruction in Florida. The FAIR is administered of year benchmarks, identify learning needs, and monitor instructional progress. The Bro ad Screen/Progress Monitoring Tool portion of the FAIR measures the complex reading comprehension that is assessed by the FCAT Reading. The Broad Screen/Progress Monitoring tool is computer adaptive and is comprised of reading passages that vary in categor y, length, and

PAGE 41

41 difficulty across grade levels. The Broad Screen/Progress Monitoring tool yields the FCAT Success Probability (FSP) score. The FSP score is an estimation of the probability that a student will be successful in passing the reading portion of of year recent FCAT performance (i.e., FCAT Reading) (Florida Department of Education, 2009). For example, an FSP score of .30 indicat es that a student has a 30% chance of achieving a DSS at the respective grade level; scores of less than 85% are considered to be at risk of not meeting grade level standards on the FCAT DSS (Foorman et al., 2013). Due to the recent development and adoption of the FAIR (i.e., 2009), extant research regarding the utility of the FAIR is minimal. Researchers Petscher and Foorman (2011) studied the utility of the FAIR for predicting FCAT SSS Reading proficiency. The authors reported that across all grades, more than 90% of students who attained an FSP of at least 0.85 passed the cut point of .70, indicating that at this cut point a st rong prediction for FCAT SSS Reading still exists (i.e., > 80%). Foorman and Petscher (2010) employed multiple regression analysis to SSS Reading performance. According to the aut hors, prior FCAT SSS Reading across Grades 4 10 accounts for a majority of the variance in predicting current/future performance, although the FAIR Broad Screen Reading adds 3.8% unique variance at Grade 4, 3.9% at Grade 5, 7.3% at Grade 6, 3.0% at Grade 7 , 3.7% at Grade 8, 1.7% at Grade 9, and 2.3% at Grade 10. Further, Foorman and Petscher (2010) reported the following prediction estimates (based on negative predictive power) SSS Reading score and the FAIR Broad Screen Read ing to

PAGE 42

42 predict current FCAT SSS Reading Level performance: 98% at Grade 4, 98% at Grade 5, 93% at Grade 6, 85% at Grade 7, 92% at Grade 8, 90% at Grade 9, and 54% at Grade 10. The decrease in predictive power beyond the primary grades underscores the need for more research regarding the academic variables and specific reading skills that contribute to FCAT success as students matriculate through school, as educators can use this information to screen for students who are at risk for school failure. Foorman et al. (2013) examined the predictive utility of the FAIR in approximating FCAT DSS (i.e., FCAT 2.0) performance scores. The researchers reported a strong correlation between FSP scores and 2012 FCAT DSS performance across grades and multiple assessment p eriods (i.e., ranging from .67 to .79). Additionally, Foorman et al. (2013) provided predictive estimates for using FSP cut points of .85 and .70 in relation to 2012 FCAT DSS Reading Level. Estimates for the .85 cut point, across Grades 4 10, were as follo ws: sensitivity (93 99%), specificity (39 71%), positive predictive power (60 68%), and negative predictive power (94 98%). Estimates for the .70 cut point, across Grades 4 10, were as follows: sensitivity (84 97%), specificity (52 84%), positive predictiv e power (70 77%), and negative predictive power (89 95%). The researchers endorsed the use of a FAIR FSP cut point of .70, in lieu of the .85 suggested by the Florida Department of Education (2009), although the .70 cut point slightly lowered the accuracy for sensitivity and negative predictive power, it markedly increased the accuracy for positive predictive power and specificity, which resulted in greater predictive accuracy. Lastly, the authors noted that using both the FAIR Broad Screen and the 2011 FCA T DSS lowered the underidentification rate of at risk students by 12 20% (i.e., on the 2012 FCAT) in contrast to using 2011 FCAT DSS performance scores solely as the means of prediction.

PAGE 43

43 These results indicate the valuable utility of the FAIR for predictin g future FCAT DSS Reading performance. Joyce, Tropf, and Gaddis (2013) reported the test retest reliability (.81 to .90), convergent validity (i.e., each result significant at the p = 0.01 level), and predictive utility (i.e., each result significant at the p = 0.001 level) of the FAIR when they compared performance on the measure to scores on the Wide Range Achievement Test 4th Edition (WRAT 4), and scores on the FCAT SSS Reading on a population of individuals who were either at risk for, or had a curr ent classification of ED. Although current research points to the efficacious nature when using the FAIR to predict FCAT SSS Reading performance, more research is needed regarding the utility of the FAIR. Specifically, more research needs to be conducted b y independent researchers to replicate the findings provided by the developers of the FAIR. In addition, more research is needed to determine the predictive utility of the FAIR when it is used with specific categories of ESE students, including those in se cure care settings. Limitations of Prior Research There are several limitations in the research literature surrounding statewide assessments. First, there is limited extant research on high stakes assessments across states, including achievement test FCAT. Second, although concurrent and predictive validity estimates of ORF prompts (Buck & Torgesen., 2003; Crawford et al., 2001; Good et al., 2001; Hunley et al., 2013; McGlinchey et al., 2004; Merino et al., 2010; Roehrig et al., 2008 ; Schilling et al., 2007; Shapiro et al., 2006; Shapiro et al., 2008; Stage & Jacobsen, 2001; Wood, 2006) and reading comprehension measures (Allison & Johnson., 2011; Ardoin et al., 2004; Denton et al., 2011 ; Marcotte & Hintze, 2009; Merino et al., 2010; Shapiro et al., 2008; Siberglitt, Burns, Madyun, & Lail 2006; Wiley & Deno, 2005) are both documented extensively, an overwhelming majority of research studies focus solely on the primary grades. Narrowly

PAGE 44

44 focusing on primary grades is problematic because t he predictive validity of ORF and comprehension measures weaken as the cognitive complexity, or cognitive load, increases on statewide assessments as students age (Shapiro et al., 2008; Schatschneider et al., 2004; Tighe and Schatschneider, 2013; Wood, 200 6). Thus, more research is needed concerning the predictive relationships of ORF, reading comprehension, and other reading skills as indicators of performance on statewide beyond the primary grades. Third, researchers have not sufficiently examined the relationship between a number of educational performance factors and statewide achievement tests. For example, despite the relationship between intelligence and school achievement (Gottfredson, 2002; Jensen, 1969; Jensen, 2002; Rohde & Thompson., 2007; Wat kins, et al. 2007), the complex interactions between age, disability, and the predictive utility of reading skills on statewide assessment performance in relation to intelligence has not been sufficiently established. Fourth, regarding the predictive util ity of academic and reading skills on statewide assessments, students who have disabilities are rarely included in datasets. Yeo (2009) illustrated this point in his meta analysis, noting that reading studies including students who have disabilities are of ten comprised of small sample sizes. If these students are included, the data are rarely disaggregated by respective disability categories. This incorrectly implies that individuals who fall into one or more of the 13 disability categories outlined by IDEI A represent a homogenous population. Not only is more research needed regarding the predictive utility of specific reading skills on statewide assessments, but this research should extend beyond the primary grades, and likewise for students who have disabi lities. This will allow researchers and educators to examine the complex interplay between disability categories, reading ability, and scores on statewide assessments across grade levels. It is crucial that we understand the academic

PAGE 45

45 needs of these student s as there are deleterious outcomes associated with failure on high stakes exams (Katsiyannis et al., 2007). A final and important limitation of the literature concerning academic performance variables and statewide assessments is the lack of extant litera ture in regard to incarcerated youth. Although Harris et al. (2009) and Krezmien et al. (2013) studied the relationship between a small array of specific reading abilities and special education status within individual juvenile correctional facilities, nei ther study included critical reading skills such as phonemic awareness, vocabulary, or fluency. Additionally, neither study compared the reading abilities of delinquent youth to an outcome variable such as a high stakes statewide assessment. Further, Krezm ien et al. (2013) did not provide disaggregated data in regard to specific special education classification(s). Thus, more research is needed concerning the specific reading skills possessed by incarcerated youth, and how these skills are related to specif ic disability categories and to attaining proficiency on statewide assessments across grades. Considering the aforementioned lack of information related to statewide assessments and disability categories, and considering the high concordance between disabi lity and delinquency, this population is ideal for exploring these relationships. Moreover, a better understanding of reading in this population could further research and intervention initiatives, possibly leading to positive life outcomes for these youth . Purpose of the Current Study programs for delinquent youth, it is clear that high quality research and evidenced based reading instruction within juvenile corrections both remain in infancy. Althoug h highly structured evidenced based reading instruction can effectively increase reading performance of incarcerated students (Allen DeBoer, Malmgren, & Glass, 2006; Coulter, 2004; Houchins et al., 2008), the specific skills required to demonstrate profici ency on statewide assessments for this population

PAGE 46

46 have not been established. Understanding the literacy performance of these youth is essential for improving educational outcomes as it provides important information that can be used to make decisions based on empirical data that ultimately can enhance reading instruction (Coyne & Harn, 2006) . Establishing the requisite skills for statewide high stakes assessment success is particularly important as the juvenile correctional system is often the last chance t hat juveniles have to be successful in school; notably, Cavendish (2013) found that only 44% of school age youth in secure care facilities return to school after release in a large study conducted in Florida. The educational deficits of delinquent youth, t heir proclivity to recidivate, and the overrepresentation of marginalized minority and disability groups both of which are falling tremendously behind academically provide an important backdrop as a rationale for further inquiry. The present study see ks to provide a more comprehensive depiction of the academic and reading abilities of incarcerated youth than is available in the extant literature. Further, the present study explores the relationships between academic and reading skills and their utility in predicting performance on statewide assessment in Florida across both grade level and educational disability category in a population of juvenile delinquents, as well as the predictive validity of progress monitoring instruments in regard to FCAT perfo rmance. The relationships explored in this study may produce information applicable to populations beyond incarcerated juveniles, and to states beyond Florida. The guiding research question of this study was as skills are most predictive of success on the reading section of the FCAT

PAGE 47

47 Research Study Questions and Rationale Research Question One What are the relationships between student demographic variable s (i.e., age, race/ethnicity, grade, exceptional student education classification(s) [ESE]), academic performance variables, and FCAT DSS Reading scores for students? Further, are there significant mean score differences on the FCAT DSS Reading between ind ividuals within different special education disability classifications? Data gathered by this question will be used to construct a comprehensive profile of the reading abilities of incarcerated youth. These questions also explore the complex relationships between the independent and dependent variables, and how these relationships are related to disability classification. These analyses will also examine for collinearity between academic performance variables, as well as how these relationships change as a function of age, disability, and grade level. Further, this question serves as the foundation for subsequent regression modeling. Research Question Two Which student academic performance variables (i.e., Intelligence Quotient [IQ], Receptive Vocabulary [R V], Sight Word Efficiency [SWE], Phonemic Decoding Efficiency [PDE], Letter Word Identification [LWI], Reading Fluency [RF], Passage Comprehension [PC], Word Attack [WA], Oral Reading Fluency [ORF], and Reading Comprehension [RC]) are most predictive of FC AT DSS Reading scores? This question explores which statistical models, involving the independent variables, account for the most variance within grade levels in regard to FCAT DSS Reading performance. Research Question Three Of the selected progress moni toring measures (Scholastic Reading Inventory [SRI], Florida Assessments for Instruction in Reading [FAIR]), which measure is more predictive of

PAGE 48

48 FCAT DSS Reading scores? Further, which of these measures is more accurate in predicting adequate performance o n the FCAT DSS Reading (i.e., Level > 3)? Predictive discriminant analysis (PDA) determines the number of true positive (TP), true negative (TN), false positive (FP), and false negative (FN) scores on the FCAT Reading across grades. PDA will be utilized to help determine which measure is a better predictor of FCAT DSS Reading Level.

PAGE 49

49 CHAPTER 2 RESEARCH METHODOLOGY Participants A sample consisting of archival data for 283 male participants was gathered from a moderate risk juvenile correctional facility, demographic information is represented in Table 2 1. The mean age for participants was 16.4 ( SD = 1.3) years. With regard to race/ethnicity, participants were African American (50%), followed by White/Non Hispanic (40%), Hispanic (9%), Mixed (1%) , and Asian (.4%). The grade levels of participants were distributed across seven grade levels: sixth (3%), seventh (10%), eighth (23%), ninth (35%), tenth (19%), eleventh (8%), and twelfth (1%). The majority of students in the sample were classified as ha ving at least one educational disability (56%). A disaggregated breakdown of ESE classification was as follows: ED (32%), SLD (17%), OHI (3%), InD (3%), LI (1%), and ASD (1%). The rate of comorbid educational disability classifications was 10%. Due to miss ingness of data, the number of participants varied throughout analyses depending on research question, estimates for respective subsamples are available below. Sampling and Setting Only data for individuals for whom the superintendent consented and for whom the individuals themselves assented were included in the sample for the study. Additionally, data were only included from students who completed the initial academic assessments u pon facility intake, which was required for participation in the original research grant. Of note, all students elected to participate in the initial assessments. Intake assessments were administered by trained graduate students within two weeks of the stu and assent were both obtained. Only students with dependent variable scores (i.e., FCAT DSS Reading score) were included in the data sample. Further, students who had passed the tenth

PAGE 50

50 grade FCAT, or ha d previously received a passing score on the General Education Development (GED) Language Arts Reading Test, were excluded from the research grant and subsequently the current study. The juvenile correctional facility is a secure 165 bed facility housin g male youth ages 14 18 with a grade range of 6 to 12. Students attend school year round across three equal semesters. On November 4, 2009 a total of 145 youth were housed at the facility. Of the 145 youth housed, approximately 38% were classified with a d isability. Of those with disabilities, primary ESE classifications included ED (58%), SLD (30%), OHI (6%), LI (4%), and InD (2%). In the subsequently decreased the number of youths housed to approximately 90 individuals. The participants in this research study were incarcerated at the correctional facility for varied sentencing lengths typically ranging between six and nine months. Research Question One Data for 263 partic ipants were included for the descriptive analyses conducted in question one, demographic information is represented in Table 2 2. The mean age for participants was 16.3 ( SD = 1.2) years. With regard to race/ethnicity, participants were African American (49 %), followed by White/Non Hispanic (39%), Hispanic (9%), Mixed (1%), and Asian (.4%). Participants were distributed across seven grade levels: sixth (3%), seventh (11%), eighth (24%), ninth (35%), tenth (17%), eleventh (8%), and twelfth (1%). The majority of students in the subsample were classified as having at least one educational disability (57%). A disaggregated breakdown of ESE classification of the subsample was as follows: ED (33%), SLD (17%), OHI (3%), InD (2%), LI (1%), and ASD (1%). Following des criptive analysis, an analysis of variance (ANOVA) for three different groups (i.e., No ESE, ED, and SLD) was

PAGE 51

51 conducted; demographic information for each grouping is represented in Table 2 3 and summarized below. Archival data for 112 participants was inc luded for the Non ESE ANOVA group. The mean age for participants was 16.3 ( SD = 1.2) years. With regard to race/ethnicity, participants were African American (54%), followed by White/Non Hispanic (31%), Hispanic (12%), and Mixed (3%). The grade levels of p articipants were distributed across seven grade levels: sixth (5%), seventh (8%), eighth (21%), ninth (38%), tenth (17%), eleventh (10%), and twelfth (1%). For the ED ANOVA group, which included data for 88 participants, the mean age was 16.3 ( SD = 1.2) y ears. With regard to race/ethnicity, participants were African American (48%), followed by White/Non Hispanic (44%), and Hispanic (8%). The grade levels of participants were distributed across seven grade levels: sixth (2%), seventh (13%), eighth (22%), ni nth (35%), tenth (18%), eleventh (8%), and twelfth (2%). For the SLD ANOVA group, which included data for 45 participants, the mean age was 16.3 ( SD = 1.3) years. With regard to race/ethnicity, participants were African American (47%), followed by White/N on Hispanic (51%), and Hispanic (2%). The grade levels of participants were distributed across five grade levels: seventh (13%), eighth (33%), ninth (29%), tenth (20%), and eleventh (4%). Research Question Two The analyses conducted for this question consi sted of four multiple regression models. The mean age, standard deviation of age, frequencies, and percentages for the overall model and the subsample, which contained 263 participants, are included above. Estimates for the three remaining models are repor ted in Table 2 4 and summarized below. For the eighth grade multiple regression model, which included data for 63 participants, the mean age was 15.9 ( SD = 1.2) years. With regard to race/ethnicity, participants were African

PAGE 52

52 American (44%), followed by Wh ite/Non Hispanic (48%), Hispanic (6%), and Mixed (2%). The majority of students in the subsample were classified as having at least one educational disability (63%). A disaggregated breakdown of ESE classification of the subsample was as follows: ED (30%), SLD (24%), OHI (6%), and InD (3%). For the ninth grade multiple regression model, which included data for 93 participants, the mean age was 16.4 ( SD = 1.1) years. With regard to race/ethnicity, participants were African American (54%), followed by White/N on Hispanic (33%), Hispanic (11%), and Mixed (2%). The majority of students in the subsample were classified as having at least one educational disability (53%). A disaggregated breakdown of ESE classification of the subsample was as follows: ED (33%), SLD (14%), OHI (2%), InD (2%), LI (1%), and ASD (1%). For the tenth grade multiple regression model, which inc luded data for 45 participants, the mean age was 16.6 ( SD = 1.2) years. With regard to race/ethnicity, participants were African American (51%), foll owed by White/Non Hispanic (38%), Hispanic (9%), and Mixed (2%). The majority of students in the subsample were classified as having at least one educational disability (58%). A disaggregated breakdown of ESE classification of the subsample was as follows: ED (36%), SLD (20%), and InD (2%). Research Question Three The analyses conducted for this question consisted of two simple linear regression models and predictive discriminant analysis of FCAT DSS Reading proficiency for two respective instruments. Compl ete demographic information for both models is presented in Table 2 5. For the SRI regression model, which included data for 282 participants, the mean age was 16.3 ( SD = 1.3) years. With regard to race/ethnicity, participants were African American (49%), followed by White/Non Hispanic (40%), Hispanic (9%), Mixed (1%), and Asian (1%). The

PAGE 53

53 grade levels of participants were distributed across seven grade levels: sixth (3%), seventh (10%), eighth (23%), ninth (35%), tenth (19%), eleventh (8%), and twelfth (1% ). The majority of students in the subsample were classified as having at least one educational disability (56%). A disaggregated breakdown of ESE classification of the subsample was as follows: ED (32%), SLD (17%), OHI (3%), InD (2%), LI (1%), and ASD (1% ). For the FSP regression model, which included data for 174 participants, the mean age was 16.3 ( SD = 1.2) years. With regard to race/ethnicity, participants were African American (54%), followed by White/Non Hispanic (37%), Hispanic (7%), and Mixed (2%). The grade levels of participants were distributed across seven grade levels: sixth (2%), seventh (13%), eighth (21%), ninth (40%), tenth (17%), eleventh (6%), and twelfth (1%). The majority of students in the subsample were classified as having at least o ne educational disability (62%). A disaggregated breakdown of ESE classification of the subsample was as follows: ED (36%), SLD (19%), OHI (3%), InD (3%), and LI (1%). Instrumentation To better understand the variables affecting overall reading achievemen t, data from a set of individually administered academic performance assessments was chosen for this research study. As specified below, in some instances only subtests from larger assessment batteries were used for analyses. The use of individual subtests is aimed at identifying the specific reading skills that are essential to achieving success on the reading portion of the FCAT DSS. All tests and subtests were chosen based on theoretical considerations including alignment with the five core areas outline d by the National Reading Panel (2000), prior research, as well as the internal structure of each measure. Additionally, with the exception of AIMSweb and SRI scores, all scores on academic performance assessments were calculated and subsequently reported utilizing

PAGE 54

54 age based norms. Tests are listed below categorically under headings that specify each respective skill domain of interest relevant to the study. Intelligence Quotient (IQ) Wechsler Abbreviated Scale of Intelligence (WASI; Weschler, 1999). The W echsler Abbreviated Scale of Intelligence (WASI) is an individually administered, nationally standardized measure that generates three traditional Verbal, Performance, and Full Scale IQ scores. The WASI consists of four subtests including: Vocabulary, Bloc k Design, Similarities, and Matrix Reasoning. For the purposes of this study, only scores from the Vocabulary and Matrix Reasoning subtests were used. These two subtests comprise the two subtest form, and provide an estimate of general cognitive ability (i .e., psychometric g ) expressed by the Full Scale IQ (FSIQ 2) composite. The Vocabulary subtest is a 42 item task in which the examinee is visually and orally presented words that the examinee defines. The Vocabulary subtest is a measure of verbal comprehen sion and knowledge, as well as psychometric g . In regard to psychometric g , the Vocabulary subtest is highly g loaded (.87) (Canivez, Konold, Collins, & Wilson, 2009). The Matrix Reasoning subtest is a series of 35 incomplete gridded patterns that the exam inee completes by selecting from five total possible choices. The Matrix Reasoning subtest measures nonverbal fluid reasoning, as well as general intellectual ability with a high g loading of .74 (Canivez et al., 2009). The FSIQ 2 , demonstrates excellent internal consistency .96, test retest reliability of .88, and a reported inter rater reliability of .98 for Vocabulary (Weschler, 1999). Additionally, Weschler (1999) reported the following correlations between the FSIQ 2 and three related measures: Weschl er Intelligence Scale for Children (WISC III; r = .81), Weschler Adult Intelligence Scale (WAIS III; r = .87), and Weschler Individual Achievement Test (WIAT; r = .47 to .72). These correlations suggest the FSIQ 2 possesses adequate convergent validity wit h

PAGE 55

55 similar cognitive measures (i.e., WISC III and WAIS III) and adequate discriminant validity with a moderately related measure (i.e., WIAT). FSIQ 2 scores were reported in the form of standard scores ( M = 100; SD = 15) for this study. Receptive Vocabulary (RV) Peabody Picture Vocabulary Test (PPVT 4; Dunn & Dunn, 2007). The Peabody Picture Vocabulary Test (PPVT 4) is an individually administered, norm referenced instrument designed to measure receptive vocabulary. The PPVT 4 contains 228 test items, which are divided into 19 item sets. During administration, the examiner says a word, and the examinee responds by selecting one of four full Dunn & Dunn (2007) reported that the PPVT 4 pos sesses an internal consistency .94 to .95, a test retest reliability of .93, and alternate form reliability of .89. Additionally, the strong correlations reported by Dunn and Dunn (2007) between the PPVT 4 and the Expressive Vocabulary Test (EVT 2; r = .82 ), the Comprehensive Assessment of Spoken Language (CASL; r = .41 to .79), the Clinical Evaluation of Language Functions (CELF 4; r = .67 to .75), the Group Reading Assessment and Diagnostic Evaluation (GRADE; r = .40 to .79), and the Peabody Picture Vocab ulary Test (PPVT III; r = .84) indicate that the PPVT 4 is a valid measure 4 scores were reported in the form of standard scores ( M = 100; SD = 15) for this study. Sight Word Efficiency (SWE) and Phon emic Decoding Efficiency (PDE) Test of Word Reading Efficiency (TOWRE; Torgesen, Wagner, & Rashotte, 1999). The Test of Word Reading Efficiency is an individually administered, norm referenced assessment designed for evaluating fluency and decoding accurac y. The TOWRE contains two subtests, Sight Word Efficiency (SWE) and Phonemic Decoding Efficiency (PDE). The Sight Word Efficiency subtest measures both accuracy and fluency by requiring individuals to read a series

PAGE 56

56 of printed words as quickly as possible w ithin a 45 second time frame. The Phonemic Decoding Efficiency subtest measures word decoding by requiring individuals to read a series of printed nonsense words as quickly as possible within a 45 second time frame. All reliability coefficients for the mea sure exceed .90 (Torgesen et al., 1999). In regard to the concurrent validity of the TOWRE, the correlation between the PDE subtest and the Woodcock Johnson Reading Mastery (WRMT R) Word Attack subtest was .85 as reported by Torgesen et al. (1999). Additio nally, Torgesen et al. (1999) reported a correlation of .89 between the SWE subtest and the WRMT R Word Identification subtest. SWE and PDE scores from the TOWRE were reported individually in the form of standard scores ( M = 100; SD = 15) for this study. Letter Word Identification (LWI), Reading Fluency (RF), Passage Comprehension (PC), & Word Attack (WA) Woodcock Johnson III Tests of Achievement (WJ III ACH; Woodcock, McGrew, & McGrew, 2007). The Woodcock Johnson Tests of Achievement (WJ III ACH) is an in dividually administered, norm referenced test designed to measure academic achievement (Mather & Woodcock, 2001). The WJ III ACH includes 22 tests measuring five curricular domains reading, mathematics, written language, oral language, and academic knowl edge (Mather & Woodcock, 2001). For the purposes of this study, only the following subtests were used: Letter Word Identification, Reading Fluency, Passage Comprehension, and Word Attack. The first three of these subtests make up the WJ ding cluster. The Broad Reading cluster is a broad measure of reading ability which assesses reading decoding, reading rate, and comprehension abilities (Mather & Woodcock, 2001). The Broad Reading cluster has a reported median reliability of .93, and subt est test retest reliability of .97 (McGrew, Schrank, & Woodcock, 2007). The last subtest, Word Attack, is administered from the extended battery portion of the WJ III ACH.

PAGE 57

57 Letter Word Identification assesses reading decoding through the use of word identi fication tasks (McGrew et al., 2007). The initial items require the subject to identify individual letters, while the remaining items require the person to pronounce whole words correctly (Mather & Woodcock, 2001). Test retest reliability for the subtest i s reported to be .95 while the median reliability is .94 by McGrew and colleagues (2007). The Reading Fluency correct, and then provide a response by circli 2001). Additionally, the subject must attempt to complete as many items as possible with a 3 minute time limit. McGrew et al. (2007) reported a mean reliability of .95 for Reading Fluency, and a test retest relia bility of .88. Passage Comprehension differs in composition and complexity with actual pictures of the objects, as the test progresses individuals must point to pictures which represent presented phrases, the remaining items require the subject to read short passages and identify missing words that which are contextually aligned with the passage (Mather & Woodcock, 2001). McGrew et al. (2007) reported that the median reliability for Passage Comprehension was .88, and that test retest reliability was .92. The Word Attack subtest words (Mather & Woodcock, 2001). The ini tial prompts call for the individual to produce the sounds for individual letters, while the remaining items require the person to read combinations of letters that are phonetically consistent but are actually non or low frequency words (Mather & Woodcock , 2001). Word Attack has a median reliability of .87 and test retest reliability of .83 (McGrew et al., 2007). LWI, RF, PC, and WA scores were reported individually in the form of standard scores ( M = 100; SD = 15) for this study.

PAGE 58

58 Oral Reading Fluency (ORF ) & Reading Comprehension (RC) AIMSweb (AIMSweb; Pearson Inc., 2012). The AIMSweb is an online progress monitoring system designed to facilitate tiered interventions that are consistent with models of Response to Intervention (RtI). The system is comprised of a variety of curriculum based measurements (CBM), and can be utilized to monitor student progress in both reading and mathematics. Two different assessments from the AIMSweb system are included in this study, R CBM and Maze CBM. R CBM is an individuall y administered standardized measure of oral reading fluency (Howe & Shinn, 2002). Students read appropriate R CBM grade level passages aloud for 1 minute while the examiner scores their reading efficiency. Misread, skipped, and omitted words are considered incorrect, while self corrections are considered correct (Howe & Shinn, 2002). Reported SEM values were between 6 and 13 words read correctly per minute across first through eighth grade, and average reliability values for alternate forms are range from . 80 to .90 across grades (Howe & Shinn, 2002). R CBM data for this study were derived from three consecutive prompt administrations, and expressed as a single median raw score, for each participant. The median of three R CBM prompts provides an appropriate approximation of within the literature (Petscher & Kim, 2011). AIMSweb Maze CBM is a standardized, multiple choice cloze task designed to measure reading comprehe nsion skills (Shinn & Shinn, 2002). In Maze CBM passages, every seventh word is replaced by a multiple choice prompt containing three options enclosed by parentheses. Students have three minutes to complete each prompt by choosing only one word which best completes each sentence. Average reliability values of pairwise comparisons for Maze CBM tasks range from .70 to .91 (Tolar et al., 2011). Maze CBM data for this study included raw scores from single prompt administrations.

PAGE 59

59 In regard to methodological con siderations, sixth grade level prompts for ORF (i.e., R CBM) and RC (i.e., Maze CBM) were chosen due to the high variability in reading ability found within juvenile justice populations. Research indicates average reading ability found within ranges from fourth to ninth grade (Brunner, 1993; Hodges, Giuliotti et al., 1994; Foley 2001; Baltodano et al., 2005). ORF and RC scores were reported individually in the form of raw scores for this study. ORF and RC data was not standardized as the majori ty of the participants were beyond the sixth grade in terms of placement and age, thus the use of traditional standardization estimates would have been inappropriate. Scholastic Reading Inventory (SRI) Scholastic Reading Inventory (SRI; Scholastic Inc., 20 01, 2006). The SRI is a computer adaptive reading assessment program for students that uses the Lexile framework to measure reading comprehension (Scholastic, Inc., 2001b, 2006). The Lexile scale is the most widely adopted reading measure currently in use (Scholastic Inc., 2008). Reading assessment instruments such as TerraNova, Iowa Tests, Stanford Achievement Test, and the Metropolitan Achievement Tests, among others, report reading scores on the Lexile Scale (Scholastic Inc., 2008). The SRI is a criterio n referenced measure typically requiring 20 30 minutes for completion and consisting of 20 determined by passages at which they can read and understand with moderate success (i.e., 75% comprehens ion) (Scholastic Inc., 2008). Lexile proficiency standards are computed approximately as follows: Grade 1 100 to 400; Grade 2 300 to 600; Grade 3 500 to 800; Grade 4 600 to 900; Grade 5 700 to 1,000; Grade 6 800 to 1,050; Grade 7 850 to 1,100; Grade 8 900 to 1,150; Grade 9 1,000 to 1,200; Grade 10 1,025 to 1,250; Grade 11 10 50 to 1300; and Grade 12 1100 to 1349 (Scholastic Inc., 2006). The reported SRI test retest reliability is .89 (Renaissance

PAGE 60

60 Learning, Inc., 2000). Additionally, the SRI possesses a repo rted criterion validity of .76 with the FCAT SSS Reading (Scholastic Inc., 2001a). SRI scores were reported in standard score form ( M = 100; SD = 15) for this study. To report SRI (i.e., Lexile scores) scores as standard scores, Lexile scores were transfo rmed to corresponding normal curve equivalent (NCEs) provided by the publisher (Scholastic Inc., 2008). The NCEs were then transformed linearly into standard scores to facilitate interpretation. The use of NCEs and standard scores allowed the author to col lapse SRI scores across grade levels into a single group. SRI categories (e.g., Beginning Reader, Below Basic, Basic, Proficient, and Advanced) were also used to satisfy portions of research question three. Florida Assessments for Instruction in Reading (F AIR) in number based on different studies, measurements, and subgroups. The sample included a statewide representative sample of students from diverse ethnic backgroun ds and included students with ESE classifications. Reliability estimates for the grade FAIR were based on an Item Response Theory (IRT), which calculates precision estimates. Estimates of general reliability were approximately .88 for each grade level for the broad screening tool. Reliability for the subtests over grade levels ranges from .77 to .95. Correlations between the Broad Screen and the FCAT SSS Reading varies between grades from .75 to .83. The cut points selected for the FAIR were based on a nega tive predictive power of 0.85, indicating that 85% or more of 3 on the reading section of the FCAT SSS by the end of the year. With the exception of Grade 10, the negative predictive power met or exceeded 0.85 criteria across all grades (Florida comparing the FCAT, differential analyses on subgroups, and examining converge nt validity

PAGE 61

61 with other reading measures (Florida Department of Education, 2009). Carlson et al. (2010) conducted a comprehensive review of the FAIR for the Buros Testing Center. Overall, the authors noted that psychometric data provided by the test develop ers appeared sound and complete. In this study FAIR scores were reported by corresponding FSP (i.e., the FSP is designed to predict future FCAT DSS Reading performance) for each participant, in the form of a percentile. Of note, FSP estimates were unavaila ble for the 2011 school year due to the restructuring of FCAT 1.0 to the FCAT 2.0, therefore FSP are only reported for the 2010 2011 and 2012 2013. Additionally, FSP Zones (i.e., Red [ < 15%], Yellow [16 84%], and Green [85%+]) were also used to satisfy portions of research question three. Florida Comprehensive Assessment Test (FCAT) The FCAT was originally created by the Florida Department of Education through a contract with CTB/McGraw Hill Company. Experts, including committees of Florida educators and citizens, refined the test by conducting both qualitative and quantitative review processes (FLDOE, 2005). Although panels of experts are involved in the yearly construction of the FCAT, there has been some criticism regarding the small amount of outside technical advisors involved this process (Buros, 2007). There are eight stages in the development of each FCAT item including: item development, initial testing, expert review, testing, s tatistical analysis, test construction, operational testing, and item release or reuse (FLDOE, 2005). The validity of the FCAT is supported by content connection with state standards through review committees, statistical review IRT, differential item func tioning, and item total correlations. Correlations between both the criterion referenced component and the norm referenced tests (SAT 9) of the FCAT for all students in Grades 4, 5, 8, and 10 are between .70 and .81 (FLDOE, 2001). Internal consistency esti mates calculated for the FCAT SSS Reading ranged from .86 to .91 (FLDOE,

PAGE 62

62 2007). Likewise, IRT marginal reliability of the FCAT SSS Reading ranged from .87 to .92. Further content, criterion, and construct validity of the FCAT have been documented extensive ly (FLDOE, 2007). Due to the incomparable nature of the FCAT across both year and grade levels, traditional FCAT DSS Reading scaled scores are reported for descriptive purposes only. Outside of initial descriptive comparisons, the FCAT DSS score variable in this study is reported in a standardized form. To approximate statistical comparisons between different grade levels and administration years, a standardized rendition of the FCAT DSS Reading score was created to serve as the primary outcome variable th roughout the study. To derive standard scores from FCAT DSS Reading scores, mean scores and standard deviations (i.e., respective to year and grade level) were used to transform FCAT DSS scores linearly (i.e., M = 100; SD = 15). The respective means and st andard deviations were requested by the author and subsequently provided by the Florida Department of Education. Research Design The study was non experimental, utilized correlational, multivariate, regression, and predictive discriminant analyses with arc hival data collected over a three year period (2010 2011, 2011 2012, 2012 2013) from a juvenile correctional facility in the state of Florida. The relationships between academic performance variables and demographic variables in regard to FCAT DSS Reading performance were explored. The primary dependent variable for this study was the FCAT DSS Reading score from the reading portion of the FCAT 2.0. The independent variables were age at intake, months between intake and FCAT DSS administration, grade, except ional student education classification (ESE), IQ, Receptive Vocabulary [RV], Sight Word Efficiency [SWE], Phonemic Decoding Efficiency [PDE], Letter Word Identification [LWI], Reading Fluency [RF], Passage Comprehension [PC], Word Attack [WA], Oral Reading Fluency

PAGE 63

63 [ORF], and Reading Comprehension [RC]. Additionally, the predictive utility of multiple progress monitoring instruments (e.g., Scholastic Reading Inventory [SRI], and Florida Assessments for Instruction in Reading [FAIR]) was investigated in relat ion to their viability for predicting FCAT DSS Reading scores, as well as their ability to predict FCAT DSS Reading proficiency. Data Collection All data reported in this study were gathered from Project LIBERATE, a large U.S. Department of Education, Ins titute of Education Sciences (IES) federally funded research grant (R324A080006) which utilized a randomized controlled trial. Subsequent analyses were secondary in nature to the original research analyses planned by the principal investigators. The author of this study was an employee of the research grant from August 2011 until August 2013. As a graduate assistant to the grant the author participated in assessment training, administration of instruments, scoring of instruments, database creation, data ent ry, and subsequent statistical analyses. Data in regard to academic performance variables, demographic variables, progress monitoring data, and FCAT DSS Reading scores for the study were gathered from information originally collected for grant purposes bet ween 2010 and 2013. The assessment instruments that were employed to measure academic performance variables included: Wechsler Abbreviated Scale of Intelligence (WASI), Woodcock Johnson III Tests of Achievement (WJ III ACH), Peabody Picture Vocabulary Test (PPVT 4), Test of Word Reading Efficiency (TOWRE), AIMSweb Curriculum Based Measurements (AIMSweb), and the Scholastic Reading Inventory (SRI). All assessment data was originally compiled by members of the grant team. In addition to these measures, the mo st recent FCAT DSS Reading data and FAIR data was collected from databases housed at the University of Florida. This data

PAGE 64

64 was originally requested by the grant team and the request was fulfilled by the registrar of the facility. The following demographic variables were required for use in the study: age at intake, race/ethnicity, grade, and ESE classification. All demographic information for the study was collected from multiple databases currently housed at the University of Florida. Relevant data were in itially requested as part of the research grant and were provided by the registrar at the juvenile corrections facility. The registrar gathered this information from internal documents as well as from district and statewide educational databases. Prior to analyses, all required data were compiled by the author at the University of Florida. All participant data were housed securely at the University of Florida unless otherwise identified. Data Analysis Preliminary Analyses To ensure the integrity and re producibility of the data, intraclass coefficients (ICC) were calculated using SPSS version 21.0 (IBM, 2012) to provide inter rater reliability estimates for 20 percent of all academic assessments administered. ICCs are commonly used in behavioral measurem ent and psychometrics to provide inferences about homogeneity among variables of a common class (Mcgraw & Wong, 1996). Additionally, absolute agreement estimates between database entry personnel were calculated and reported for 20% of the data using the su bsequent formula: number of agreements/(number of agreements + number of disagreements) x 100 (House, House, & Campbell, 1981). Error rates were also calculated using the following formula: number of errors/number of agreements + number of disagreements) x 100. Information contained within academic databases included raw data input and subsequent standardized score

PAGE 65

65 transformation(s) for all academic assessments (i.e., WASI, PPVT 4, TOWRE, WJ III ACH, etc.). All observed discrepancies between personnel were corrected post hoc. All data points were excluded from the analysis for those individuals who did not complete the outcome variable (i.e., FCAT DSS Reading score), and for those individuals who did not complete a single independent variable. In dealing wit h missing data, researchers often employ statistical techniques such as multiple imputation (MI), which serves as a principled method to assist in the analysis of complex incomplete data problems (Rubin, 1996; Schafer & Olsen, 1998; van Buuren & Groothuis Oudshoorn, 2011). Imputation provides researchers will constructing plausible models based on the variable means of complete cases within datasets (Zhang, Liao, & Zhu, 200 8). Of note, including imputed dependent variables in analyses can Therefore, imputed dependent variables were not included within the analysis. Consistent with suggestio ns provided by researchers Schafer and Graham (2002), the imputation model utilized within this study included input from all available variables that were included within the analysis model. Imputation modeling was completed using R 2.15.1 (R Core Team, 2 012) via single imputation by chained equations approach available within the mice 2.17 package (van Buuren & Groothuis Oudshoorn, 2012). Although multiple imputation by chained equations (MICE) method is superior in nature to a single imputation by chaine d equations approach, both techniques can perform equally well with datasets containing proportionally small numbers of missing values (van der Heijden, Donders, Stijen, & Moons, 2006).

PAGE 66

66 Before subsequent statistical analyses, all models were checked for v iolations of the four principal assumptions (i.e., linearity, homoscedasticity, independence, and normality) underlying regression models outlined by Fox (2008, pp. 100 102). Research Question One Descriptive statistics for all demographic, predictor, and outcome variables of interest were collected and reported through respective means and standard deviations. Additionally, a one way between subjects analysis of variance (ANOVA), followed by pairwise comparisons, was conducted to examine the extent to whic h reading performance was statistically similar across student disability classification(s). The ANOVA model included the following groups: Non ESE, ED, and SLD. Post hoc comparisons of each ANOVA group were conducted using the Shaffer Holm (Holm, 1979a; 1 979b; Shaffer, 1980; 1986) procedure to control for the familywise error rate. SPSS 21 (IBM, 2012) and the PROC GLM function in SAS 9.3 (SAS Institute Inc., 2012) were used to conduct these analyses. Last, utilizing the eta 2 ) statistic, effect s ize was calculated to determine the unique contribution of group membership in regard to predicting FCAT DSS Reading score. Eta squared in ANOVA, or the correlation ratio, is calculated by dividing the sum of squares (SS) for an effect by the SS total (Nix & Barnette, 1998). Measures of effect size provide estimates of practical significance, that are independent of sample size, as they estimate the magnitude to which a phenomenon exists in a study (Nix & Barnette, 1998). In contrast, tests of statistical si gnificance (e.g., p values) are bound to sample size, and only meaningfully approximate how improbable it is to obtain observed sample data if the null hypothesis is true (Fan, 2001). The utility and necessity of calculating effect size, in lieu of or in c onjunction with p values, has been extensively documented (e.g., Cohen, 1992; Fan,

PAGE 67

67 2 A correlational matrix was then constructed to examine the relationship between independent predictor variables and multicollinearity, and to provide r ationale for subsequent regression modeling. To examine for multicollinearity, the Variance Inflation Factor (VIF) was calculated for each salient independent variable (i.e., each academic assessment) in relation to all other independent variables of inter est. Although there remains little consensus in the professional literature regarding suitable VIF values and subsequent tolerance levels, a VIF of 10 multicollinea collinearity and may prompt consideration for variable exclusion or other procedural safeguards (Hair, Anderson, Tatham, & Black, 1995; Kleinbaum, Kupper, Nizam, & Muller, 2007 ). Research Question Two Simultaneous multiple regression analysis was conducted on the overall imputed subsample to identify which predictor variables within the sample explained the most variance in terms of FCAT DSS Reading score. A model fitting proced ure utilizing backward elimination stepwise multiple regression was then conducted in R 2.15.1 (R Core Team, 2012) to reduce the number of predictor variables included, thus producing a more parsimonious model. Stepwise multiple regression, by way of backw ard elimination subset selection, systematically deletes variables from the original model based on estimates of regressor importance (Burnham & Anderson, 2002; Fox, 2008). The criterion of model quality was based on the Akaike Information Criterion (AIC) , one of the most commonly used model selection criteria (Fox, 2008, p. 611). AIC, introduced by Akaike (1973), seeks to derive the most parsimonious model by providing an estimation of fit

PAGE 68

68 based on an equation that combines a negative log likelihood and a penalty term (i.e., the number of free parameters [ k ] within a given model) (Bozdogan, 1987; 2000). The formula for calculating AIC using least squares (LS) estimation can be expressed as follows: n log(RSS/ n ) + 2 k (Burnham & Anderson, 2002, p. 63). If ba sed on a strong theoretical foundation and well justified criterion, AIC can provide accurate approximations of model selection in regard to predictive phenomena within a given sample (Burnham & Anderson, 2004; Yamashita, Yamashita, & Kamimura, 2007). Mode ls with lower AIC values indicate a better goodness of fit compared to models with higher values, as AIC estimates the Kullback Leibler divergence between the sample and the model that approximates it (Akaike, 1973; 1978; Bozdogan, 1987; Buckland, Burnham, & Augustin, 1997). In contrast to R 2 , which necessarily increases with variable inclusion(s), AIC penalizes overparameterization as it selects the simplest model of fit (Bozdogan, 1987; Fox, 2008, p. 611 614). However, the use of AIC criteria for model se lection with relatively small sample sizes (i.e., n / k = < 40) is problematic, as it may produce biased estimates which subsequently lead to model overfitting (Hurvich & Tsai, 1989; Burnham & Anderson, 2002). Thus, Hurvich and Tsai (1989; 1995) proposed a b ias corrected formula for AIC (i.e., AIC c ). The following example of the bias correction formula for least squares regression with finite sample sizes was provided by researchers Burnham, Anderson, and Huyvaert (2011): AIC c = n log(RSS/ n ) + As Hurvich and Tsai (1989) note, AIC c

PAGE 69

69 Thus, for this study, inferences about individual models were based on bias corrected AIC c and i AIC c (i.e., i = [AIC ci min AIC c ]) ; a measure of the relative plausibility between each model (e.g., see Burnham & Anderson, 2002, p. 71; Wagenmakers & Farrell, 2004). To interpret i AIC c , empiric al support between models were used, the plausibility levels are: i (0 2) = substantial; i (4 7) = considerably less; i ( > 10) = essentially none. Further, relative likelihood ratios (estimates that are invariant to other models), which demonstrate the probability that the model minimizes the relative estimation of information lost, for each model (i.e., exp[ i ]) were ca lculated and then normalized to derive Akaike weights ( w i AIC c ) for each model (e.g., see Akaike, 1983; Burnham & Anderson, 2002; Turkheimer, Hinz, & Cunningham, 2003). Providing weights are treated as probabilities; although invariant to other models, the w i AIC c depends on the full model set as they collectively sum to 1 (Burnham & Anderson, 2004). Researchers Burnham and Anderson (2002; 2004) provided the following formula for d eriving w i AIC c : w i AIC c = After are Akaike weights were computed, evidence ratios (ER) were provided for each AIC c model based on respective w i AIC c , the following formula was used (Burnham & Anderson, 200 2; Wagenmakers & Farrell, 2004): ER = or more simply ER =

PAGE 70

70 i of 2 between w i (i.e., model of best fit) and w j (i.e., all other models) indicates that w i is 2.7 times more likely to be model of best fit when compared to w j , while a i of 10 indicates that w i is approximately 148.4 times more likely to be the model of best fit (Burnham & Anderson, 2004, p. 78; Burnham et al., 2011). Methods for the model fitting procedure conducted in this study were outlined by Fox and Weisberg (2011, pp. 208 213). Three addi tional multiple regression models, employing the same backward elimination stepwise multiple regression procedure, were also constructed based on grade level (i.e., Grade 8, Grade 9, and Grade 10) to determine if predictor variables remained stable across grade levels in relation to FCAT DSS Reading scores. Global effect sizes, utilizing 2 regression model selected by the AIC c procedure in order to determine the utility in predicting FCAT criteria, in respect to the 2 four ANOVAs were constructed to calculate eta 2 ) values in order to quantify the magnitude of effect sizes for individual regressors within models. Specifically, eta squared values were use d to determine the unique contribution, or effect size, of each explanatory variable within each selected final simultaneous multiple regression model (i.e., overall imputed 2 statistic, was used to interpret effect sizes for individual regressors. Research Question Three Due to missingness of data and the use of two different subsamples for this analysis, salient descriptive statistics are provided as necessary. Two separate simple linear regression

PAGE 71

71 models were constructed in R 2.15.1 (R Core Team, 2012) to assess the predictive validity of each progress monitoring measure (i.e., SRI, FAIR) in regard to FCAT DSS Reading score. 2 st atistic, were then calculated in order to interpret the practical utility of each measure (i.e., SRI, FAIR) in predicting FCAT DSS Reading score. Additionally, predictive discriminant analysis (PDA), a technique outlined by researchers Hosp and Fuchs (200 5), was used to assess the prediction accuracy or specificity of each progress monitoring measure in regard to prediction of adequate performance on the FCAT DSS Reading (i.e., Level > 3). Two cut point of .85, were tested as thresholds for FSP predictive power in regard to FCAT point of confidence by the Florida Department of Education. However, a reduced cut point of .70 still possesses strong predictive ability in regard to FCAT DSS designation or higher was used as the cut point on the SR I as the predictive threshold for FCAT DSS Reading success. Frequencies based on the number of true positives (TP), true negatives (TN), false positives (FP), and false negatives (FN) in relation to FCAT DSS Reading performance were reported. Hit rate, sen sitivity, and specificity estimates of each measure were calculated and reported. Hosp and Fuchs (2005) provide the following formulas to calculate PDA: Hit Rate = (TP) + (TN) / N; Sensitivity = TP / TP + (FN); and Specificity = TN / TN + (FP)

PAGE 72

72 Table 2 1 . Overall sample d emographics : Frequencies and p ercentages Demographic Variables Frequency ( N = 283) Percentages Race/ethnicity African American 141 49.8 White a 112 39.6 Hispanic 25 8.8 Mixed 4 1.4 Asian 1 0.4 Grade Level Sixth 8 2.8 Seventh 29 10.2 Eighth 65 23.0 Ninth 100 35.3 Tenth 54 19.1 Eleventh 23 8.1 Twelfth 4 1.4 ESE Classification Non ESE 124 43.8 Special Ed. 159 56.2 ED 92 32.5 SLD 47 16.6 OHI 10 3.5 InD 7 2.5 LI 2 0.7 ASD 1 0.4 Note. White a = Non Hispanic. ESE = Exceptional Student Education; ED = Emotional Disturbance; SLD = Specific Learning Disability; OHI = Other Health Impairment; InD = Intellectual Disability; LI = Language Impairment; ASD = Autism Spectrum Disorder.

PAGE 73

73 Table 2 2. Subsample demographics for question one descriptive analyses: Frequencies and p ercentages Demographic Variables Frequency ( n = 263) Percentages Race/ethnicity African American 130 49.4 White a 103 39.2 Hispanic 25 9.5 Mixed 4 1.5 Asian 1 0.4 Grade Level Sixth 8 3.0 Seventh 29 11.0 Eighth 63 24.0 Ninth 93 35.4 Tenth 45 17.1 Eleventh 21 8.0 Twelfth 4 1.5 ESE Classification Non ESE 112 42.6 Special Ed. 151 57.4 ED 88 33.5 SLD 45 17.1 OHI 9 3.4 InD 6 2.3 LI 2 0.8 ASD 1 0.4 Note. White a = Non Hispanic. ESE = Exceptional Student Education; ED = Emotional Disturbance; SLD = Specific Learning Disability; OHI = Other Health Impairment; InD = Intellectual Disability; LI = Language Impairment; ASD = Autism Spectrum Disorder.

PAGE 74

74 Tabl e 2 3. Sample demographics for question one ANOVA g roupings: Frequencies a nd p ercentages Non ESE ( n = 112) ED ( n = 88) SLD ( n = 45) Demographic Variables Frequency Percentages Frequency Percentages Frequency Percentages Race/ethnicity African American 60 53.6 42 47.7 21 46.7 White a 35 31.3 39 44.3 23 51.1 Hispanic 14 12.5 7 8.0 1 2.2 Mixed 3 2.7 Asian Grade Level Sixth 6 5.4 2 2.3 Seventh 9 8.0 11 12.5 6 13.3 Eighth 23 20.5 19 21.6 15 33.3 Ninth 43 38.4 31 35.2 13 28.9 Tenth 19 17.0 16 18.2 9 20.0 Eleventh 11 9.8 7 8.0 2 4.4 Twelfth 1 0.9 2 2.3 Note. White a = Non Hispanic

PAGE 75

75 Table 2 4. Sample demographics for question two multiple regression models: Frequencies and p ercentages Overall Model ( n = 263) Grade 8 ( n = 63) Grade 9 ( n = 93) Grade 10 ( n = 45) Demographic Variables Frequency Percentages Frequency Percentages Frequency Percentages Frequency Percentages Race/ethnicity African American 130 49.4 28 44.4 50 53.8 23 51.1 White a 103 39.2 30 47.6 31 33.3 17 37.8 Hispanic 25 9.5 4 6.3 10 10.8 4 8.9 Mixed 4 1.5 1 1.6 2 2.2 1 2.2 Asian 1 0.4 Grade Level Sixth 8 3.0 Seventh 29 11.0 Eighth 63 24.0 63 100 Ninth 93 35.4 93 100 Tenth 45 17.1 45 100 Eleventh 21 8.0 Twelfth 4 1.5 ESE Classification Non ESE 112 42.6 23 36.5 43 46.2 19 42.2 Special Ed. 151 57.4 40 63.5 50 53.8 26 57.8 ED 88 33.5 19 30.2 31 33.3 16 35.6 SLD 45 17.1 15 23.8 13 14.0 9 20.0 OHI 9 3.4 4 6.3 2 2.2 InD 6 2.3 2 3.2 2 2.2 1 2.2 LI 2 0.8 1 1.1 ASD 1 0.4 1 1.1 Note. White a = Non Hispanic. ESE = Exceptional Student Education; ED = Emotional Disturbance; SLD = Specific Learning Disability; OHI = Other He alth Impairment; InD = Intellectual Disability; LI = Language Impairment; ASD = Autism Spectrum Disorder.

PAGE 76

76 Table 2 5. Sample demographics for question three s imple linear r egression models: Frequencies and percentages SRI Model ( n = 282) FSP Model ( n = 174) Demographic Variables Frequency Percentages Frequency Percentages Race/ethnicity African American 140 49.6 94 54.0 White a 112 39.7 64 36.8 Hispanic 25 8.9 13 7.5 Mixed 4 1.4 3 1.7 Asian 1 0.4 Grade Level Sixth 8 2.8 4 2.3 Seventh 29 10.3 23 13.2 Eighth 65 23.0 36 20.7 Ninth 99 35.1 70 40.2 Tenth 54 19.1 29 16.7 Eleventh 23 8.2 11 6.3 Twelfth 4 1.4 1 0.6 ESE Classification Non ESE 124 44.0 66 37.9 Special Ed. 158 56.0 108 62.1 ED 91 32.3 63 36.2 SLD 47 16.7 34 19.5 OHI 10 3.5 5 2.9 InD 7 2.5 5 2.9 LI 2 0.7 1 0.6 ASD 1 0.4 Note. White a = Non Hispanic. ESE = Exceptional Student Education; ED = Emotional Disturbance; SLD = Specific Learning Disability; OHI = Other Health Impairment; InD = Intellectual Disability; LI = Language Impairment; ASD = Autism Spectrum Disorder.

PAGE 77

77 CHAPTER 3 RESULTS Preliminary Analyses Table 3 1 presents the ICC estimates for inter rater reliability between academic assessment administrators, utilizing one way random effects models and the average measures estimate, ranged from to .98 to 1.00 ( p = < .001) (T able 3 1). ICC results indicate high levels of homogeneity and reproducibility, these estimates are in concordance with published estimates. The mean absolute agreement between database entry personnel across databases was approximately 99%, with an approx imate mean error rate of 0.5 per 100 instances (Table 3 2). These results purport high levels of absolute agreement, and support the overall integrity of the data. Preliminary analysis of the dataset indicated that 20 individuals were missing data across all independent variables related to academic assessment; these cases were not included in the subsample used in research questions one and two. To prevent further deletion of cases containing only partial academic assessment data, single imputation by cha ined equations approach was conducted to impute missing values; imputation was utilized to salvage 1.3% of the data. Table 3 3 presents the comparison of the mean score differences between the overall sample ( N = 283) and the imputed subsample ( n = 263). Regression models were within normal limits in respect to violations of assumptions. To check the assumption that the conditional means of response should be a linear function of FCAT DSS Reading, conditional means were plotted linearly; results are presen ted in Figure 3 1. Residuals were plotted against fitted values to assess for homoscedasticity within each 2 through 3 7. Skewness and kurtosis values were calculat ed to assess normality, values for each

PAGE 78

78 regression model follow: Imputed subsample (skewness = 0.01; kurtosis = 0.63), Grade 8 (skewness = 0.05; kurtosis = 0.81), Grade 9 (skewness = 0.07; kurtosis = 0.86), Grade 10 (skewness = 0.61; kurtosis = 0.15), SRI (skewness = 0.04; kurtosis = 0.72), and FSP (skewness = 0.02; kurtosis = 0.77). Data Analysis Research Question One Overall descriptives for the outcome and predictor variables of interest for the imputed subsample are reported through respective means and standard deviations below. Further, disaggregated breakdowns by grade and disability category are presented in Table 3 4 and Table 3 5, respectively. Additionally, descriptive statistics disaggregated by race/ethnicity are presented in Table 3 6. The overall descriptive statistics for the imputed subsample, comprised of 263 participants, are provided within each table to facilitate relative comparisons concerning the sample. For the overall imputed subsample, the mean age for participants was 16.3 ( SD = 1.2). The mean grade level for participants comprising the imputed subsample was 8.8 ( SD = 1.2). On average 2.8 months ( SD = 5.2) elapsed between the intake date and the administration of the FCAT DSS Reading. When all participants from the impu ted subsample were included, the mean FCAT DSS Reading score was 214.5 ( SD = 21.9); expressed as a standard score, the mean FCAT DSS Reading performance was 83.2 ( SD = 14.8). The means and standard deviations for standardized independent academic variables for the subsample were as follows: IQ ( M = 87.9; SD = 12.2), RV ( M = 86.2; SD = 13.0), SWE ( M = 85.2; SD = 11.7), PDE ( M = 85.2; SD = 14.9), LWI ( M = 85.5; SD = 16.7), RF ( M = 84.7; SD = 12.9), PC ( M = 84.1; SD = 15.0), WA ( M = 92.3; SD = 14.5). Overall, participants in the subsample achieved an average mean score of approximately 86.4 across standardized reading measures, or about one standard deviation below

PAGE 79

79 the mean. The mean raw score for participants in regard to ORF was 157.3 ( SD = 49.9), while the m ean raw RC score was 24.1 ( SD = 10.4). National norms provided by AIMSweb in 2012 indicate that sixth grade students achieved an ORF mean of 154.0 ( SD = 40) and an RC mean of 20 ( SD = 10.0). On average, when compared to national mean scores, individuals in this sample scored slightly above sixth graders in regard to ORF and below sixth graders in RC performance. ANOVA: Non ESE vs. ED, Non ESE vs. SLD, and ED vs. SLD. A one way ANOVA was conducted to compare mean differences on FCAT DSS Reading scores acros s three different groups. Table 3 7 presents descriptive statistics for those without an ESE designation ( M = 88.1; SD = 14.4), those with a primary classification of ED ( M = 79.7; SD = 14.5), and those with a primary classification of SLD ( M = 80.6; SD = 14.3). The results of the ANOVA are presented in Table 3 8. A statistically significant effect was found for at least one group in regard to FCAT DSS Reading scores ( F [2, 242] = 9.53, p = < .001, 2 = .07). Further the 2 value of .07 is characterized as a small effect size, indicating that approximately 7% of the variance in FCAT DSS Reading score can be attributed to group membership. Due to a significant F value, pairwise post hoc compari sons utilizing the Shaffer Holm procedure were conducted controll fw / C = .05 for all three family pairs (Table 3 9). Comparisons indicated that the mean score difference of 8.4 between Non ESE and ED was statistically significant at p = < .001 ( t [242] = 4.07). Likewise, the mean score diffe rence of 7.5 between Non ESE and SLD was significant at p = < .004 ( t [242] = 2.74). Lastly, the mean score difference ( 0.9) between ED and SLD was not statistically significant ( t [242] = 0.35, p = < .120), indicating that the difference between ED and SL D was likely due to chance.

PAGE 80

80 The results from the correlational analysis are presented in Table 3 10. As expected, many of the academic variables were moderately to highly correlated. To examine for the presence of multicollinearity, VIF was calculated for each academic variable of interest by means of multiple regression analysis. None of the independent variables approached the critical VIF value set at 10. Therefore, multicollinearity was deemed to be within acceptable limits for the dataset. However, LW I approached moderate VIF levels of 5.0 to 5.2 with respect to the following variables: IQ, SWE, PDE, RF, RC, ORF, and RV. Considering that LWI measures decoding, a foundational component to reading, moderate levels of collinearity are to be expected. Of n ote, although LWI was included within all initial models, the variable was excluded by stepwise multiple regression procedures in all final models except for Grade 10. Research Question Two Four multiple regression analyses (i.e., Overall imputed subsample , Grade 8 subsample, Grade 9 subsample, and Grade 10 subsample), utilizing backward elimination stepwise multiple regression based on bias corrected Akaike information criterion, were constructed to determine if predictor variables remained stable across g rade levels in relation to FCAT DSS Reading scores. All predictor variables were initially included in modeling procedures for each group. The predictor variables were age at intake, months between intake and FCAT DSS Reading administration, IQ, RV, SWE, P DE, LWI, RF, PC,WA, ORF, and RC. The sequence of presentation for each group of models is as follows: (1) multiple simultaneous regression analysis of initial model, (2) comparison of competing models selected via stepwise multiple regression based on AIC c , and (3) multiple simultaneous regression analysis of model of best fit. When all regressors were included in the initial model which included all grades and all participants within the subsample (Table 3 11), the model accounted for approximately 38% of the variance within FCAT DSS Reading scores (multiple R 2 = .38, adjusted R 2 = .35). The overall

PAGE 81

81 model was highly predictive and significant at p = < .001 ( F [12, 250] = 12.8). Three of the independent variables exhibited significant main effects in regard t o prediction of the outcome variable (i.e., IQ, RC, and months between intake and FCAT DSS Reading administration). IQ was a significant predictor of FCAT DSS Reading performance ( t [250] = 3.81, p = < .001), while both RC ( t [250] = 2.60) and months between intake and FCAT DSS Reading administration ( t [250] = 2.55) were significant at p = .010 and p = .011, respectively. Although results indicate that scoring highly on the aforementioned variables is related to achieving higher FCAT DSS Reading scores, the r elative weight, or predictor importance, within the model cannot be determined due to the use of unstandardized regression coefficients. A total of seven subsequent models were generated via stepwise multiple regression based on a criterion of relative AI C c . The aforementioned model, which included all predictor variables, produced a relative AIC c c of 1319.25 represents a i of 9.36 when compared to the (i.e., AIC c = 1309.89). A i of 9.36 is substantial, such that one can assume that the initial model is 107.89 (ER) times less probable to represent the most parsimonious model when compared to the model of best fit selected in Step 7. Likewise, the initial model accounted for an ove rall w i AIC c of .003 , indicating that there is only a .003% chance that this model represents the most parsimonious model. Table 3 12 presents the results of the stepwise procedures for the seven subsequent models generated via stepwise multiple regression based on a criterion of relative AIC c . A summary of the subsequent models produced through the stepwise procedure follows: Step 1: excluded age ( AIC c = 1317.10 ; i = 7.22); Step 2: excluded age and PC ( AIC c = 1315.16 ; i = 5.28); Step 3: excluded age, PC, and WA ( AIC c = 1313.41 ; i = 3.53); Step 4: excluded age, PC, WA, and LWI ( AIC c = 1312.17 ; i = 2.28); Step 5: excluded age, PC, WA, LWI, and RF ( AIC c =

PAGE 82

82 1311.11 ; i = 1.22); Step 6: excluded age, PC, WA, LWI, RF, and ORF ( AIC c = 1310.40 ; i = .51); Step 7 : excluded age, PC, WA, LWI, RF, ORF, and PDE ( AIC c = 1309.89 ; w i AIC c = .343 ). The final model (see Step 7) included IQ, SWE, RC, months between intake and FCAT DSS Reading administration, and RV, arranged serially by relative importance. Although this model was selected as the most parsimonious, providing meaningful inferences is problematic. For instance, there is strong evidence that the failure to include variables contained within Step 4 ( i = 2.28), Step 5 ( i = 1.22), and/or Step 6 ( i = .51) sign ificantly increases the likelihood that the final model is errant in its approximation. However, if a relative heuristic of .90 (acc w i ) is employed, a reasonable assertion can be made that the variables contained within Steps 4 7 provide a 90.4% chance of containing the best model of fit. Further, using this heuristic we can assert with confidence that the initial model and the models contained within Steps 1 3 can all be rejected as possible models of best fit. Table 3 13 presents the simultaneous multi ple regression analyses of the final model of best fit generated through the use of the stepwise model fitting procedure. The final model included IQ, SWE, RC, months between intake and FCAT DSS Reading administration, and RV as predictor variables. The re duced model accounted for approximately 37% of the variance within FCAT DSS Reading scores (multiple R 2 = .37, adjusted R 2 = .36). The overall model was highly predictive and significant at p = < .001 ( F [5, 257] = 28.89, 2 = .59). Examination of the 2 va lue indicated that the effect size for the overall model was large in terms of prediction. Four of the independent variables exhibited significant main effects in regard to prediction of the outcome variable. These included IQ ( t [257] = 4.01, p = < .001, 2 = .23), SWE ( t [257] = 3.58, p = < .001, 2 = .08), RC ( t [257] = 3.40, p = < .001, 2 = .03), and months between intake and FCAT DSS Reading administration ( t [257] = 2.54, p = .012, 2 = .01). Examination of the

PAGE 83

83 predictor variables indicated the effect size for IQ was medium to large, explaining 23% of the variance, while SWE (8%) and RC (3%) both had small effect sizes . Although the months between intake and FCAT DSS Reading administration significantly predicted the outcome variable, it did not produce a substantive ly significant effect size. The reduced model, when F = 16.09 , R 2 = R 2 = .01. Thus, a small change in adjusted R 2 (.01) indicates that not onl y is the selected model more parsimonious, but it is also approximately equivalent to the original model in regard to prediction of the outcome variable. Table 3 14 presents the simultaneous multiple regression analyses for the initial eighth grade model, which included all twelve regressors. The model accounted for approximately 47% of the variance within FCAT DSS Reading scores (multiple R 2 = .47, adjusted R 2 = .34). The overall model was significant at p = < .001 ( F [12, 50] = 12.8). However, only one of the independent variables exhibited a significant main effect in relation to predicting FCAT DSS Reading scores: IQ ( t [50] = 3.81, p = < .008). Results indicate that not only is scoring highly on IQ related to achieving higher FCAT DSS Reading scores, but that it is the most important predictor within the model. The initial model for Grade 8, comprised of all predictor variables, produced a relative AIC c c i of 18.36 compared to the final model in Step 8 (i.e., AIC c = 311.19). A i of 18.36 is substantial, indicating that the final model is 9,702.83 (ER) times more likely to represent the most parsimonious model compared to the model containing all twelve predictor variables. Likewise, the initial model ac counted for a meager overall w i AIC c of less than .0001 , indicating that there is less than a .0001% chance that this model represents the most parsimonious model.

PAGE 84

84 Table 3 15 presents the results of the stepwise procedures for the eight successive models g enerated via stepwise multiple regression based on a criterion of relative AIC c . A summary of the subsequent models produced through the stepwise procedure follows: Step 1: excluded age ( AIC c = 326.62 ; i = 15.43); Step 2: excluded age and PDE ( AIC c = 323. 78 ; i = 12.59); Step 3: excluded age, PDE, and RF ( AIC c = 321.16 ; i = 9.97); Step 4: excluded age, PDE, RF, and RV ( AIC c = 318.70 ; i = 7.51); Step 5: excluded age, PDE, RF, RV, and SWE ( AIC c = 316.57 ; i = 5.38); Step 6: excluded age, PDE, RF, RV, SWE, and ORF ( AIC c = 314.45 ; i = 3.26); Step 7: excluded age, PDE, RF, RV, SWE, ORF, and PC ( AIC c = 313.94 ; i = 2.75); Step 8: excluded age, PDE, RF, RV, SWE, ORF, PC, and LWI ( AIC c = 311.19 ; w i AIC c = .645 ). The model of best fit (see Step 8) included IQ, RC, months between intake and FCAT DSS Reading administration, and WA (arranged serially by relative importance). The model selected in Step 8, accounted for a w i AIC c of .645 , or 64.5% chance being the tru e model of best fit. If the relative heuristic of .90 (acc w i ) is employed, a reasonable assertion can be made that the variables contained within Steps 6 8 provide a 93.5% chance of containing the best model of fit. Further, using this heuristic we can co nfidently affirm that the initial model and the models produced in Steps 1 5 can all be rejected as possible models of best fit. The final model of best fit for Grade 8 retained months between intake and FCAT DSS Reading administration, IQ, WA, and RC as predictor variables (Table 3 16). The reduced model accounted for approximately 44% of the variance within FCAT DSS Reading scores (multiple R 2 = .44, adjusted R 2 = .41). The final model was also highly predictive, had a large effect size, and was signific ant at p = < .001 ( F [4, 58] = 11.57, 2 = .79). In the final model, IQ ( t [58] = 3.51, p = < .001, 2 = .23) remained a significant predictor and had medium to large effect size, explaining 23% of the variance. RC ( t [58] = 2.39, p = < .001, 2 = .05) also was a statistically significant

PAGE 85

85 predictor of FCAT DSS Reading scores, with a small effect size of 5%. Although neither variable was a statistically significant predictor of the outcome variable, both months between intake and FCAT DSS Reading administration and WA had small effect sizes of .0 8 and .07, respectively . This model, when compared to the Grade 8 model containing all twelve predictor F = R 2 = R 2 = .06. Despite removing eight predictor varia bles, the model remained approximately equivalent to the original model, while gaining a small amount of predictive power. Table 3 17 displays the simultaneous multiple regression analyses for the initial model for Grade 9, which included all regressors. T he model accounted for approximately 48% of the variance within FCAT DSS Reading scores (multiple R 2 = .48, adjusted R 2 = .40). The model was highly predictive and significant at p = < .001 ( F [12, 80] = 6.20). Two independent variables (i.e., RC and RV) ex hibited significant main effects in regard to prediction of the outcome variable: RC was significant ( t [80] = 2.93) at p = .004, while RV ( t [80] = 2.01) was significant at p = .042. A total of nine subsequent models were generated utilizing a stepwise mod el fitting procedure. The initial Grade 9 model, which included all predictor variables, carried a relative AIC c c i of 18.25 compared to final model (Step 9). A i of 18.25 indicates that the initial model is 9,161.15 (ER) times less likely to represent the most parsimonious model when compared to the model of best fit selected in Step 9. Table 3 18 presents the results of the stepwise procedures for the nine successive models generated via stepwise multiple r egression based on a criterion of relative AIC c . A summary of the subsequent models produced through the stepwise procedure follows: Step 1: excluded PC ( AIC c = 458.36 ; i = 15.64); Step 2: excluded PC and RF ( AIC c = 455.87 ; i = 13.16); Step 3:

PAGE 86

86 excluded PC, RF, and age ( AIC c = 453.61 ; i = 10.89); Step 4: excluded PC, RF, age, and PDE ( AIC c = 451.45 ; i = 8.74); Step 5: excluded PC, RF, age, PDE, and WA( AIC c = 449.38 ; i = 6.66); Step 6: excluded PC, RF, age, PDE, WA, and LWI ( AIC c = 447.17 ; i = 4.45); S tep 7: excluded PC, RF, age, PDE, WA, LWI, and months between intake and FCAT DSS Reading administration ( AIC c = 446.23 ; i = 3.52); Step 8: excluded PC, RF, age, PDE, WA, LWI, months between intake and FCAT DSS Reading administration, and IQ ( AIC c = 443.7 4 ; i = 1.03); Step 9: excluded PC, RF, age, PDE, WA, LWI, months between intake and FCAT DSS Reading administration, IQ, and ORF ( AIC c = 442.71 ; w i AIC c = .517 ). The final model in Step 9 included RC, RV, and SWE (presented serially by relative importance within the model). However, failure to retain ORF as a predictor, which was removed following Step 8 ( i = 1.03), significantly increases the likelihood that the final model is missing a variable that could add important predictive value. If the relative heuristic of .90 (acc w i ) is again used as a cut point, the candidate models contained within Steps 7 9 pr ovide a 91.6% chance of containing the best model of fit. It is important to note, however, that Step 7 accounts for less than 10% of the overall weight ( w i AIC c = 8.27 ), indicating that its inclusion may not be necessary if parsimony is desired. The final model for Grade 9 retained RC, RV, and SWE as predictor variables (Table 3 19). These predictors accounted for approximately 46% of the variance within FCAT DSS Reading scores (multiple R 2 = .46, adjusted R 2 = .45). The model was highly predictive and sig nificant at p = < .001 ( F [3, 89] = 25.66, 2 = .85). Examination of the 2 value indicated that the effect size for the overall model was large. All three of the retained variables exhibited significant main effects: RC ( t [89] = 3.67, p = < .001, 2 = .08) , RV ( t [89] = 2.87, p = .005, 2 = .16), and SWE ( t [89] = 2.49, p = .015, 2 = .22). Examination of the predictor variables indicated

PAGE 87

87 the effect sizes for SWE and RV were both medium, explaining 22% and 16% of the variance , respectively . While the effect size for RC was small, explaining 8% of the variance. Of note, interpret each variables relative importance to the model. For instance, if both RV and SWE ar e observed FCAT DSS Reading score by .38 standard deviations. Likewise, if all other variables 3) by one standard deviation would change the FCAT DSS Reading score by .26 or .23 standard deviations , respectively. Lastly, when compared to the initial model containing all twelve predictor variables, the reduced F R 2 = .02, a R 2 = .05. A small change in adjusted R 2 (.05) indicates that not only is the selected model more parsimonious, but that it also gained predictive power. When all regressors were included in the initial model for Grade 10 (Table 3 20), the model a ccounted for approximately 54% of the variance within FCAT DSS Reading scores (multiple R 2 = .54, adjusted R 2 = .36). The overall model was predictive of the outcome variable and significant at p = .005. ( F [12, 32] = 3.10). Two of the independent variables exhibited significant main effects in regard to prediction. Of these, IQ was the more significant predictor ( t [32] = 3.19, p = .003) compared to LWI ( t [32] = 2.6, p = .037). Originally eight models were chosen using model fitting procedure based on selec tion criterion of AIC. However, post hoc analysis indicated that two of these models increased the AIC c and therefore needed to be excluded due to selection bias. This is common when AIC is used as sole criteria on models with small sample sizes. Thus, onl y six models were included in the analysis, results are summarized in Table 3 21. The initial model produced a relative AIC c of

PAGE 88

88 c i of 15.86 when compared to the model of best fit selected in Step 6 (i.e., AIC c = 235.35). A summary of the models produced through the stepwise procedure follows: Step 1: excluded age ( AIC c = 247.48 ; i = 12.13); Step 2: excluded age and PDE ( AIC c = 244.04 ; i = 8.69); Step 3: excluded age, PDE, and SWE ( AIC c = 240.79 ; i = 5.44); Step 4: excluded age, PDE, SWE, and months between intake and FCAT DSS Reading administration ( AIC c = 238.34 ; i = 2.99); Step 5: excluded age, PDE, SWE, months between intake and FCAT DSS Reading administration, and RF ( AIC c = 237.21 ; i = 1.86 ); Step 6: excluded age, PDE, SWE, months between intake and FCAT DSS Reading administration, RF, and RC ( AIC c = 235.35 ; w i AIC c = .588 (Step 6) included IQ, LWI, WA, ORF, PC and RV within the respective model (arranged se rially by relative importance). Table 3 22 presents the simultaneous multiple regression analyses for the final reduced model. The model accounted for approximately 50% of the variance within FCAT DSS Reading scores (multiple R 2 = .50, adjusted R 2 = .42). The model was highly predictive, had a large effect size, and significant at p = < .001 ( F [6, 38] = 6.29, 2 = 1.0). Four of the independent variables exhibited significant main effects in regard to prediction of the outcome variable. These include IQ ( t [38] = 3.30, p = < .002, 2 = .31), LWI ( t [38] = 2.49, p = < .017, 2 = .00), WA ( t [257] = 2.17, p = < .036, 2 = .06), and ORF ( t [38] = 2.20, p = .034, 2 = .09). Examination of the predictor variables indicated the effect size for IQ was large, explaini ng 31% of the variance, while WA and ORF had small effects, explaining 6% and 9% of the variance , respectively . The remaining variab les did not produce substantively significant effect sizes. For this model, etric to compare the relative importance for each significant variable. The relative importance of each variable in regard to FCAT DSS Reading

PAGE 89

89 reduced model, when compared to the initial model containing all twelve predictor variables, F R 2 = R 2 = .12. In regard to prediction, based on the change in adjusted R 2 (.12), this model possesses more predictive utility than original model which included all twelve variables. Although this model was selected as the most parsimonious for Grade 10, due to the small sample size, providing meaningful inferences is troublesome. For instance, there is evidence that the failure to include variabl es contained within Step 4 ( i = 2.99) and Step 5 ( i = 1.86) significantly increases the likelihood that the final model is errant in its approximation. Further, if a relative heuristic of .90 (acc w i ) is used (i.e., Steps 4 6 provide a 95.2%), we are only able to confidently exclude four regressors from the initial model with any reasonable certainty. Research Question Three Two separate simple linear regression models were constructed to identify the predictive validity of both the SRI and the FAIR in regard to FCAT DSS Reading score. Table 3 23 contains relevant descriptive statistics, describing both regression model subsamples. Simple linear regression models were for both the SRI and the FSP. PDA w as also conducted on both instruments to quantify the respective hit rate, sensitivity, and specificity of each instrument in relation to FCAT DSS Reading performance prediction. Table 3 24 presents the results of the simple linear regression model for SR I as the predictor variable and the FCAT DSS as the dependent variable. SRI scores significantly predicted FCAT DSS Reading scores ( B = 0.39, t [280] = 8.80, p = < .001, 2 = .28). Additionally, the squared multiple correlation was R 2 = .22 (adjusted R 2 = 0 .21). Results indicate that SRI scores are significant predictors of FCAT DSS Reading scores, accounting for

PAGE 90

90 approximately 22% of the variance within the model. Further, examination of the 2 value indicated that the SRI had a medium effect size on FCAT DS S Reading scores. Table 3 25 presents the results of the simple linear regression model for FSP as the predictor variable and the FCAT DSS as the dependent variable. The regression coefficient for FSP was 0.31, and was highly predictive of FCAT DSS Reading scores ( t [172] = 8.33, p = < .001, 2 = .41). FSP scores accounted for approximately 29% of the variance within FCAT DSS Reading scores (multiple R 2 = .29, adjusted R 2 = .28). Results indicate that achieving higher FSP scores are significantly rela ted to achieving higher scores on the FCAT DSS Reading. Further, examination of the 2 value indicated that the FSP had a large effect size on FCAT DSS Reading scores. Table 3 26 presents the relevant statistics for the PDA conducted for both the SRI and t he FAIR as predictors of FCAT DSS Reading Level. Descriptive analyses indicate that individuals within the SRI subsample passed the FCAT DSS Reading with a Level 3 or higher at a rate of 14.9%, with a mean FCAT DSS Reading Level of 1.6 ( SD = .9). Likewise , individuals belonging to the FSP subsample achieved an average overall FCAT DSS Reading Level of 1.5 ( SD = .8), and passed the FCAT DSS Reading at a rate of 12.1%. determina nt cut point effectively predicted FCAT DSS Reading achievement with 73.8% accuracy (i.e., hit rate). The sensitivity index, or true positive rate, for the SRI indicated that the measure was 59.5% accurate in identifying students who had mastered the skill s necessary to pass the FCAT DSS Reading (i.e., Level > 3). In regard to specificity, or the true negative rate, score on the FCAT DSS Reading.

PAGE 91

91 Using the .85 thre shold as a cut point for the FSP yielded a 91.9% hit rate in terms of achievement prediction on the FCAT DSS Reading. The high number of false negatives (i.e., students who did not meet the .85 threshold but achieved a Level > 3) provided for a low sensiti vity index (41.2%). The .85 cut point demonstrated excellent specificity as 100% of students who achieved an FSP of .85 or above achieved a passing score on the FCAT DSS Reading. The .70 threshold proved equally accurate at predicting FCAT DSS performance . The evidenced hit rate indicates that an FSP cut point of .70 was 93.1% accurate in predicting student outcomes on the FCAT DSS Reading. Similar to the .85 FSP threshold, the high number of false negatives (i.e., proportional to the true positives) led t o a low sensitivity index (52.4%). In regard to specificity, only 1.3% of students who achieved an FSP of .70 or higher did not achieve at least a Level 3 on the FCAT DSS Reading. Despite the greater chance for false positives due to a lower threshold, the .70 cut point outperformed the .85 cut point in terms of both hit rate and sensitivity. Overall, results from the sample indicate that an FSP cut point of .70 had more overall predictive utility in terms of FCAT DSS Reading achievement than using an FSP t hreshold of .85 for this sample.

PAGE 92

92 Table 3 1. ICC estimates (one way random effects m od el) for inter rater reliability on academic a ssessments Academic Assessment ICC Single Measures ICC Average Measures ICC Significance WASI .97 .98 < .001*** PPVT 4 .99 .99 < .001*** TOWRE .99 .99 < .001*** WJ III ACH 1.00 1.00 < .001*** AIMSweb ORF 1.00 1.00 < .001*** AIMSweb RC .99 .99 < .001*** Note. WASI = Wechsler Abbreviated Scale of Intelligence; PPVT 4 = Peabody Picture Vocabulary Test Fourth Edition; TOWRE = Test of Word Reading Efficiency; Woodcock Johnson Tests of Achievement Third Edition; ORF = Oral Reading Fluency; RC =Reading Comprehen sion. * p < .05., ** p < .01., *** p < .001.

PAGE 93

93 Table 3 2. Absolute agreement estimates for inter rater reliability on academic a ssessments Academic Databases Percent Agreement Error Rate Assessment Raw Data WASI Score Translation PPVT 4 Score Translation TOWRE Score Translation WJ III ACH Score Translation 99.9 .05 99.1 .9 100 .0 99.3 .7 99.3 .7 Note. WASI = Wechsler Abbreviated Scale of Intelligence; PPVT 4 = Peabody Picture Vocabulary Test Fourth Edition; TOWRE = Test of Word Reading Efficiency; Woodcock Johnson Tests of Achievement Third Edition.

PAGE 94

94 Table 3 3. Mean score differences between overall sample and imputed s ubsample Overall Sample [ N = 283] Imputed Subsample [ n = 263] Overall vs. Imputed Subsample Variables Mean ( SD ) Mean ( SD ) Mean Difference Age at intake 16.4 (1.2) 16.3 (1.2) .1 Months a 2.7 (5.0) 2.8 (5.2) +.1 FCAT DSS std. 83.5(15.0) 83.2 (14.8) .3 FCAT DSS Level 1.6 (.84) 1.5 (.84) .1 IQ 88.2 (12.0) 87.9 (12.2) .3 RV 86.3 (13.1) 86.2 (13.0) .1 SWE 85.2 (11.5) 85.2 (11.7) .0 PDE 85.3 (14.7) 85.2 (14.9) .1 LWI 85.5 (16.7) 85.5 (16.7) .0 RF 84.5 (12.9) 84.7 (12.9) +.2 PC 84.1 (15.1) 84.1 (15.0) .0 WA 92.2 (14.5) 92.3 (14.5) +.1 ORF 157.4 (50.1) 157.3 (49.9) .1 RC 24.1 (10.4) 24.1 (10.4) .0 Note. Months a = Number of months between intake and FCAT DSS Reading. FCAT DSS std. = Florida Comprehensive Test 2.0 Reading Developmental Scaled Score transformed standardized score; IQ = WASI FSIQ 2; RV = PPVT 4 Receptive Vocabulary; SWE = TOWRE Sight Word Efficien cy; PDE = TOWRE Phonemic Decoding Efficiency; LWI = WJ III ACH Letter Word Identification; RF = WJ III ACH Reading Fluency; PC = WJ III ACH Passage Comprehension; WA = WJ III ACH Word Attack; ORF = AIMSweb Oral Reading Fluency (R CBM); RC = AIMSweb Reading Comprehension (Maze CBM).

PAGE 95

95 Table 3 4. Descriptive statistics for imputed subsample by g rade Subsample [ N = 263] Grade 6 [ n = 8] Grade 7 [ n = 29] Grade 8 [ n = 63] Grade 9 [ n = 93] Grade 10 [ n = 45] Grade 11 [ n = 21] Grade 12 [ n = 4] Variables Mean ( SD ) Mean ( SD ) Mean ( SD ) Mean ( SD ) Mean ( SD ) Mean ( SD ) Mean ( SD ) Mean ( SD ) Age at intake 16.3 (1.2) 15.7 (1.4) 15.8 (1.4) 15.9 (1.2) 16.4 (1.1) 16.8 (1.2) 17.1 (.8) 17.7 (.5) Months a 2.8 (5.2) 2.4 (3.3) 1.8 (6.3) 2.9 (5.0) 3.2 (5.2) 3.2 (4.6) 1.4 (5.9) 3.3 (4.1) FCAT DSS b 214.5 (21.9) 199.1 (17.2) 208.4 (21.9) 214.5 (22.0) 213.1 (21.1) 215.9 (22.5) 229.9 (19.3) 227.9 (18.6) FCAT DSS std. 83.2 (14.8) 81.8 (12.1) 84.1 (15.5) 85.2 (14.7) 82.0 (14.2) 79.3 (16.3) 89.2 (14.0) 88.0 (13.0) IQ 87.9 (12.2) 78.6 (13.4) 85.3 (13.6) 87.9 (11.9) 86.7 (11.1) 90.7 (12.2) 94.8 (12.5) 88.0 (9.6) RV 86.2 (13.0) 85.0 (11.6) 84.0 (12.7) 88.1 (12.8) 83.9 (13.7) 87.8 (11.4) 91.7 (14.2) 82.5 (10.1) SWE 85.2 (11.7) 78.4 (12.6) 84.0 (11.6) 86.2 (11.9) 84.1 (11.7) 86.7 (10.6) 88.8 (12.9) 80.5 (5.3) PDE 85.2 (14.9) 76.3 (12.5) 83.4 (16.3) 85.2 (14.8) 84.3 (14.6) 86.9 (14.6) 92.3 (15.9) 80.0 (9.1) LWI 85.5 (16.7) 80.3 (17.3) 79.2 (19.6) 86.9 (16.3) 84.0 (17.7) 89.2 (13.3) 90.8 (14.1) 82.3 (12.2) RF 84.7 (12.9) 75.6 (7.4) 81.3 (13.5) 86.1 (12.2) 82.5 (13.2) 88.6 (11.9) 89.6 (13.7) 83.5 (14.6) PC 84.1 (15.0) 75.0 (16.1) 78.9 (14.1) 85.6 (15.3) 82.2 (15.9) 87.4 (10.7) 91.7 (15.5) 85.8 (14.1) WA 92.3 (14.5) 86.0 (15.1) 87.8 (15.4) 93.3 (13.7) 91.5 (14.1) 95.2 (13.5) 96.8 (18.4) 86.0 (9.4) ORF 157.3 (49.9) 119.5 (58.4) 136.3 (46.1) 150.5 (41.7) 154.2 (50.1) 181.2 (49.4) 181.7 (53.4) 168.0 (21.3) RC 24.1 (10.4) 15.1 (8.7) 19.9 (9.2) 22.9 (9.9) 23.5 (10.2) 29.3 (10.1) 27.1 (10.6) 27.3 (5.6) Note. Months a = Number of months between intake and FCAT DSS Reading. FCAT DSS b = Scaled score provided by FLDOE. FCAT DSS std. = Florida Comprehensive Test 2.0 Reading Developmental Scaled Score transformed standardized score; IQ = WASI FSIQ 2; RV = PPVT 4 Receptive Vocabulary; SWE = TOWRE Sight Word Efficiency; PDE = TOWRE Phonemic Decoding Efficiency; LWI = WJ III ACH Letter Word Identification; RF = WJ III ACH Reading Fluency; PC = WJ III ACH Passage Comprehension; WA = WJ III ACH Word Attack; ORF = AIMSweb Oral R eading Fluency (R CBM) ; RC = AIMSweb Reading Comprehension (Maze CBM).

PAGE 96

96 Table 3 5. Descriptive statistics for imputed subsample by ESE c lassification Subsample [ N = 263] Non ESE [ n = 112] All ESE [ n = 151] ED [ n = 88] SLD [ n = 45] OHI [ n = 9] InD [ n = 6] Variables Mean ( SD ) Mean ( SD ) Mean ( SD ) Mean ( SD ) Mean ( SD ) Mean ( SD ) Mean ( SD ) Age at intake 16.3 (1.2) 16.3 (1.2) 16.3 (1.2) 16.3 (1.2) 16.3 (1.3) 16.4 (1.6) 16.7 (1.4) Months a 2.8 (5.2) 3.0 (5.3) 2.6 (5.1) 2.8 (5.0) 2.9 (5.3) 1.0 (5.2) 1.0 (4.8) FCAT DSS std. 83.2 (14.8) 88.1 (14.4) 79.6 (14.1) 79.7 (14.5) 80.6 (14.3) 78.0 (13.6) 70.8 (12.5) IQ 87.9 (12.2) 91.8 (11.1) 85.1 (12.2) 85.8 (12.9) 84.9 (10.7) 87.0 (12.8) 76.8 (12.5) RV 86.2 (13.0) 88.3 (12.3) 84.7 (13.4) 85.1 (13.6) 83.9 (12.9) 88.1 (10.6) 75.3 (17.3) SWE 85.2 (11.7) 88.3 (10.6) 82.9 (11.9) 83.6 (13.1) 82.4 (10.4) 84.6 (9.1) 74.2 (9.3) PDE 85.2 (14.9) 90.1 (12.6) 81.6 (15.6) 83.9 (16.6) 78.1 (13.2) 82.4 (11.3) 69.8 (9.9) LWI 85.5 (16.7) 90.9 (11.3) 81.5 (18.8) 82.3 (21.2) 80.5 (14.8) 83.7 (12.9) 70.2 (20.4) RF 84.7 (12.9) 89.0 (10.5) 81.4 (13.7) 81.9 (15.4) 81.9 (11.5) 80.6 (8.2) 74.2 (9.7) PC 84.1 (15.0) 87.8 (10.7) 81.4 (17.1) 81.4 (17.8) 83.1 (13.5) 86.6 (10.6) 61.8 (29.4) WA 92.3 (14.5) 97.8 (12.4) 88.2 (14.6) 88.9 (16.1) 87.4 (12.0) 90.2 (8.5) 79.5 (17.5) ORF 157.3 (49.9) 171.8 (39.6) 146.5 (54.1) 147.9 (59.5) 146.4 (47.4) 145.4 (31.2) 115.5 (47.9) RC 24.1 (10.4) 27.0 (9.8) 21.9 (10.4) 22.4 (11.2) 21.7 (9.6) 20.9 (7.8) 15.8 (7.8) Note. Months a = Number of months between intake and FCAT DSS Reading. FCAT DSS std. = Florida Comprehensive Test 2.0 Reading Developmental Scaled Score transformed standardized score; IQ = WASI FSIQ 2; RV = PPVT 4 Receptive Vocabulary; SWE = TOWRE Sight Word Efficien cy; PDE = TOWRE Phonemic Decoding Efficiency; LWI = WJ III ACH Letter Word Identification; RF = WJ III ACH Reading Fluency; PC = WJ III ACH Passage Comprehension; WA = WJ III ACH Word Attack; ORF = AIMSweb Oral Reading Fluency (R CBM); RC = AIMSweb Reading Comprehension (Maze CBM).

PAGE 97

97 Table 3 6. Descriptive statistics for imputed subsample by r ace/ethnicity Imputed Subsample [ N = 263] African American [ n = 130] White a [ n = 103] Hispanic [ n =25] Variables Mean ( SD ) Mean ( SD ) Mean ( SD ) Mean ( SD ) Age at intake 16.3 (1.2) 16.3 (1.2) 16.2 (1.3) 16.9 (1.4) Months b 2.8 (5.2) 3.7 (5.3) 2.6 (4.9) 1.4 (5.6) FCAT DSS std. 83.2 (14.8) 82.2 (13.8) 84.1 (15.6) 85.4 (17.9) IQ 87.9 (12.2) 86.1 (11.8) 90.0 (12.3) 89.4 (13.3) RV 86.2 (13.0) 81.4 (12.4) 93.2 (10.7) 81.8 (12.7) SWE 85.2 (11.7) 85.3 (11.7) 85.0 (11.8) 84.6 (12.1) PDE 85.2 (14.9) 83.4 (14.9) 86.6 (15.1) 87.3 (14.1) LWI 85.5 (16.7) 84.0 (16.7) 86.5 (17.2) 87.4 (16.2) RF 84.7 (12.9) 83.4 (11.4) 86.3 (14.5) 83.6 (13.2) PC 84.1 (15.0) 82.3 (15.5) 87.6 (14.1) 80.3 (14.8) WA 92.3 (14.5) 89.4 (17.8) 93.7 (13.6) 93.2 (16.6) ORF 157.3 (49.9) 152.1 (46.6) 162.5 (53.6) 156.1 (52.5) RC 24.1 (10.4) 22.1 (9.3) 26.4 (11.0) 23.8 (11.5) Note. White a = Non Hispanic. Months b = Number of months between intake and FCAT DSS Reading. FCAT DSS std. = Florida Comprehensive Test 2.0 Reading Developmental Scaled Score transformed standardized score; IQ = WASI FSIQ 2; RV = PPVT 4 Receptive Vocabulary; SWE = TOWRE Sight Word Efficien cy; PDE = TOWRE Phonemic Decoding Efficiency; LWI = WJ III ACH Letter Word Identification; RF = WJ III ACH Reading Fluency; PC = WJ III ACH Passage Comprehension; WA = WJ III ACH Word Attack; ORF = AIMSweb Oral Reading Fluency (R CBM) ; RC = AIMSweb Reading Comprehension (R Maze).

PAGE 98

98 Table 3 7. Descriptive s tatistics for ANOVA of FCAT DSS Reading Group N Mean SD Non ESE 112 88.1 14.4 ED 88 79.7 14.5 SLD 45 80.6 14.3 Note. n = 245. ESE = Exceptional Student Education; ED = Emotional Disturbance; SLD = Specific Learning Disability.

PAGE 99

99 Table 3 8. One way ANOVA p rocedure for FCAT DSS Reading df Sum of Squares Mean Square F Pr> F Between 2 3953.85 1976.85 9.53 < .001*** Within 242 50226.57 207.55 Corrected Total 244 54180.42 Note. n = 245. FCAT DSS = Florida Comprehensive Test 2.0 Reading Developmental Scaled Score.

PAGE 100

100 Table 3 9. Shaffer Holm procedure of pairwise c omparisons for FCAT DSS Reading Rank Family Comparison Mean Difference F T p C fw / C 1 Non ESE vs. ED 8.4 16.55 4.07 < .001*** 1 .05 2 Non ESE vs. SLD 7.5 8.54 2.74 .004** 1 .05 3 ED vs. SLD 0.9 0.12 0.35 .120 1 .05 Note. n = 245. FCAT DSS = Florida Comprehensive Test 2.0 Reading Developmental Scaled Score. * p < .05., ** p < .01., *** p < .001.

PAGE 101

101 Table 3 10. Pearson product moment correlation between v ariables Age Months IQ SWE PDE LWI RF PC WA RC ORF RV FCAT Age a 1.00 Months b .08 1.00 IQ .13 .08 1.00 SWE .29 .03 .34 1.00 PDE .19 .01 .37 .80 1.00 LWI .18 .02 .49 .71 .73 1.00 RF .17 .05 .46 .70 .62 .67 1.00 PC .10 .02 .53 .55 .53 .74 .64 1.00 WA .12 .00 .44 .62 .74 .81 .58 .60 1.00 RC .02 .02 .46 .52 .54 .56 .70 .55 .54 1.00 ORF .07 .04 .45 .79 .71 .74 .78 .64 .64 .71 1.00 RV .16 .09 .57 .30 .38 .49 .41 .49 .37 .41 .38 1.00 FCAT .14 .10 .47 .44 .45 .42 .47 .42 .42 .48 .44 .39 1.00 Note. n = 263. Age a = Chronological age at time of intake. Months b = Number of months between intake and FCAT DSS Reading. FCAT = Florida Comprehensive Test 2.0 Reading Developmental Scaled Score transformed standardized score; IQ = WASI FSIQ 2; RV = PPVT 4 Recept ive Vocabulary; SWE = TOWRE Sight Word Efficiency; PDE = TOWRE Phonemic Decoding Efficiency; LWI = WJ III ACH Letter Word Identification; RF = WJ III ACH Reading Fluency; PC = WJ III ACH Passage Comprehension; WA = WJ III ACH Word Attack; ORF = AIMSweb Ora l Reading Fluency (R CBM) ; RC = AIMSweb Reading Comprehension (Maze CBM).

PAGE 102

102 Table 3 11. Simultaneous multiple regression analysis for predicting FCAT DSS Reading for imputed s ubsample Predictor Variables B SE t value Pr(>| t |) Intercept 8.85 17.19 ----Age a 0.14 0.67 0.22 .829 Months b 0.37 0.15 2.55 .011* IQ 0.31 0.08 3.81 .001*** SWE 0.23 0.14 1.68 .093 PDE 0.12 0.10 1.21 .226 LWI 0.12 0.10 1.23 .220 RF 0.10 0.10 0.99 .322 PC 0.04 0.08 0.50 .618 WA 0.06 0.10 0.64 .521 RV 0.11 0.08 1.46 .144 RC 0.30 0.11 2.60 .010* ORF 0.03 0.03 1.02 .309 Multiple R 2 0.38 Adjusted R 2 0.35 F 12.8*** Note. n = 263. Age a = Chronological age at time of intake. Months a = Number of months between intake and FCAT DSS Reading. FCAT DSS = Florida Comprehensive Test 2.0 Reading Developmental Scaled Score transformed standardized score; IQ = WASI FSIQ 2; RV = PPVT 4 Receptive Vocabulary; SWE = TOWRE Sight Word Efficiency; PDE = TOWRE Phonemic Decoding Efficiency; LWI = WJ III ACH Letter Word Identification; RF = WJ III ACH Reading Fluency; PC = WJ III ACH Passage Comprehension; WA = WJ III ACH Word Attack; ORF = AIMSweb Oral Reading Fluency (R CBM) ; R C = AIMSweb Reading Comprehension (Maze CBM). * p < .05., ** p < .01., *** p < .001.

PAGE 103

103 Table 3 12. Summary of backward elimination stepwise multiple regression models for predicting FCAT DSS Reading for imputed s ubsample Stepwise Models k AIC AIC c i AIC c Relative likelihood w i AIC c acc w i ER Imputed a 12 1318.00 1319.25 9.36 0.009 0.003 1.000 107.89 Step 1 b 11 1316.05 1317.10 7.22 0.027 0.009 0.997 36.89 Step 2 c 10 1314.29 1315.16 5.28 0.071 0.025 0.988 13.99 Step 3 d 9 1312.70 1313.41 3.53 0.172 0.059 0.963 5.83 Step 4 e 8 1311.60 1312.17 2.28 0.320 0.110 0.904 3.13 Step 5 f 7 1310.67 1311.11 1.22 0.542 0.186 0.795 1.84 Step 6 g 6 1310.07 1310.40 0.51 0.774 0.265 0.608 1.29 Step 7 h 5 1309.38 1309.89 0.00 1.000 0.343 0.343 Note. n = 263. a Predictors excluded: none; b Predictors excluded: age; c Predictors excluded: age, PC; d Predictors excluded: age, PC, WA; e Predictors excluded: age, PC, WA, LWI; f Predictors excluded: age, PC, WA, LWI, RF; g Predictors excluded: age, PC, WA, LWI, RF, ORF; h Predictors excluded: age, PC, WA, LWI, RF, ORF, PDE.

PAGE 104

104 Table 3 13. AIC c model of best fit for predicting FCAT DSS Reading for imputed s ubsample Predictor Variables B SE t value Pr(>| t |) Intercept 14.7 7.48 ----Months a 0.36 0.14 2.54 .01* IQ 0.31 0.08 4.01 < .001*** SWE 0.27 0.07 3.58 < .001*** RV 0.11 0.07 1.63 .105 RC 0.30 0.09 3.40 < .001*** Multiple R 2 0.37 Adjusted R 2 0.36 F 28.89*** Note. n = 263. Months a = Number of months between intake and FCAT DSS Reading. FCAT DSS = Florida Comprehensive Test 2.0 Reading Developmental Scaled Score transformed standardized score; IQ = WASI FSIQ 2; SWE = TOWRE Sight Word Efficiency; RV = PPVT 4 Receptive Vocabulary; RC = AIMSweb Reading Comprehension (Maze CBM). * p < .05., ** p < .01., *** p < .001.

PAGE 105

105 Table 3 14. Simultaneous multiple regression analysis for p redictin g FCAT DSS Reading for Grade 8 s ubsample Predictor Variable B SE t value Pr(>| t |) Intercept 9.27 42.60 ----Age a 0.51 1.62 0.32 .754 Months b 0.69 0.35 1.96 .056 IQ 0.42 0.15 2.74 .008** SWE 0.20 0.32 0.62 .537 PDE 0.08 0.24 0.32 .753 LWI 0.20 0.23 0.85 .397 RF 0.08 0.24 0.35 .727 PC 0.09 0.14 0.63 .531 WA 0.33 0.20 1.66 .102 RV 0.10 0.19 0.55 .589 RC 0.39 0.24 1.63 .110 ORF 0.06 0.07 0.88 .384 Multiple R 2 0.47 Adjusted R 2 0.34 F 3.71*** Note. n = 63. Age a = Chronological age at time of intake. Months a = Number of months between intake and FCAT DSS Reading. FCAT DSS = Florida Comprehensive Test 2.0 Reading Developmental Scaled Score transformed standardized score; IQ = WASI FSIQ 2; RV = PPVT 4 Receptive Vocabulary; SWE = TOWRE Sight Word Efficiency; P DE = TOWRE Phonemic Decoding Efficiency; LWI = WJ III ACH Letter Word Identification; RF = WJ III ACH Reading Fluency; PC = WJ III ACH Passage Comprehension; WA = WJ III ACH Word Attack; ORF = AIMSweb Oral Reading Fluency (R CBM) ; RC = AIMSweb Reading Comp rehension (Maze CBM). * p < .05., ** p < .01., *** p < .001.

PAGE 106

106 Table 3 15. Summary of backward elimination stepwise multiple regression models for p redictin g FCAT DSS Reading for Grade 8 s ubsample Stepwise Models k AIC AIC c i AIC c Relative likelihood w i AIC c acc w i ER Grade 8 a 12 323.31 329.55 18.36 0.000 0.000 1.000 9702.83 Step 1 b 11 321.44 326.62 15.43 0.000 0.000 1.000 2238.16 Step 2 c 10 319.55 323.78 12.59 0.002 0.001 1.000 542.16 Step 3 d 9 317.76 321.16 9.97 0.007 0.004 0.998 145.95 Step 4 e 8 316.03 318.70 7.51 0.023 0.015 0.994 42.67 Step 5 f 7 314.53 316.57 5.38 0.068 0.044 0.979 14.71 Step 6 g 6 312.95 314.45 3.26 0.196 0.126 0.935 5.10 Step 7 h 5 311.66 313.94 2.75 0.253 0.163 0.809 3.96 Step 8 i 4 310.50 311.19 0.00 1.000 0.645 0.645 Note. n = 63. a Predictors excluded: none; b Predictors excluded: age; c Predictors excluded: age, PDE; d Predictors excluded: age, PDE, RF; e Predictors excluded: age, PDE, RF, RV; f Predictors excluded: age, PDE, RF, RV, SWE; g Predictors excluded: age, PDE, RF, RV, SWE, ORF; h Predictors excluded: age, PDE, RF, RV, SWE, ORF, PC; i Predictors excluded: age, PDE, RF, RV, SWE, ORF, PC, and LWI.

PAGE 107

107 Table 3 16. AIC c model of best fit for p redicting FCAT DSS Reading for Gra de 8 s ubsample Predictor Variables B SE t value Pr(>| t |) Intercept 16.61 13.17 ----Months a 0.55 0.30 1.87 .066 IQ 0.45 0.13 3.51 < .001*** WA 0.20 0.12 1.68 .099 RC 0.40 0.17 2.39 .020* Multiple R 2 0.44 Adjusted R 2 0.41 F 11.57*** Note. n = 63. Months a = Number of months between intake and FCAT DSS Reading. FCAT DSS = Florida Comprehensive Test 2.0 Reading Developmental Scaled Score transformed standardized score; IQ = WASI FSIQ 2; WA = WJ III ACH Word Attack; RC = AIMSweb Reading Comprehension (R Maze) . * p < .05., ** p < .01., *** p < .001.

PAGE 108

108 Table 3 17. Simultaneous multiple regression analysis for p redictin g FCAT DSS Reading for Grade 9 s ubsample Predictor Variable B SE t value Pr(>| t |) Intercept 25.69 34.50 ----Age a 0.50 1.33 0.38 .708 Months b 0.15 0.23 0.69 .493 IQ 0.08 0.14 0.58 .561 SWE 0.29 0.24 1.21 .230 PDE 0.08 0.17 0.47 .637 LWI 0.07 0.17 0.42 .678 RF 0.05 0.18 0.25 .800 PC 0.03 0.14 0.19 .849 WA 0.11 0.17 0.64 .523 RV 0.23 0.11 2.01 .042* RC 0.54 0.18 2.93 .004** ORF 0.05 0.06 0.89 .374 Multiple R 2 0.48 Adjusted R 2 0.40 F 6.20*** Note. n = 93. Age a = Chronological age at time of intake. Months a = Number of months between intake and FCAT DSS Reading. FCAT DSS = Florida Comprehensive Test 2.0 Reading Developmental Scaled Score transformed standardized score; IQ = WASI FSIQ 2; RV = PPVT 4 Receptive Vocabulary; SWE = TOWRE Sight Word Efficiency; P DE = TOWRE Phonemic Decoding Efficiency; LWI = WJ III ACH Letter Word Identification; RF = WJ III ACH Reading Fluency; PC = WJ III ACH Passage Comprehension; WA = WJ III ACH Word Attack; ORF = AIMSweb Oral Reading Fluency (R CBM) ; RC = AIMSweb Reading Comp rehension (Maze CBM). * p < .05., ** p < .01., *** p < .001.

PAGE 109

109 Table 3 18. Summary of backward elimination stepwise multiple regression models for p redictin g FCAT DSS Reading for Grade 9 s ubsample Stepwise Models k AIC AIC c i AIC c Relative likelihood w i AIC c acc w i ER Grade 9 a 12 457.06 460.96 18.25 0.000 0.000 1.000 9161.15 Step 1 b 11 455.10 458.36 15.64 0.000 0.000 1.000 2495.78 Step 2 c 10 453.19 455.87 13.16 0.001 0.001 1.000 719.96 Step 3 d 9 451.44 453.61 10.89 0.004 0.002 0.999 232.08 Step 4 e 8 449.74 451.45 8.74 0.013 0.007 0.997 79.03 Step 5 f 7 448.06 449.38 6.66 0.036 0.018 0.990 27.98 Step 6 g 6 446.19 447.17 4.45 0.108 0.056 0.972 9.26 Step 7 h 5 444.74 446.23 3.52 0.172 0.089 0.916 5.81 Step 8 i 4 443.47 443.74 1.03 0.599 0.310 0.827 1.67 Step 9 j 3 442.26 442.71 0.00 1.000 0.517 0.517 Note. n = 93. a Predictors excluded: none; b Predictors excluded: PC; c Predictors excluded: PC, RF; d Predictors excluded: PC, RF, age; e Predictors excluded: PC, RF, age, ORF; f Predictors excluded: PC, RF, age, ORF, WA; g Predictors excluded: PC, RF, age, ORF, WA, LWI; h Predictors excluded: PC, RF, age, ORF, WA, LWI, months; i Predictors excluded: PC, RF, age, ORF, WA, LWI, months; IQ; j Predi ctors excluded: PC, RF, age, ORF, WA, LWI, months, IQ, ORF.

PAGE 110

110 Table 3 19. AIC c model of best fit for p redictin g FCAT DSS Reading for Grade 9 s ubsample Predictor Variables B SE t value Pr(>| t |) Intercept 24.33 10.79 ------SWE 0.27 0.11 0.23 2.49 .015* RV 0.27 0.09 0.26 2.87 .005** RC 0.52 0.14 0.38 3.67 < .001*** Multiple R 2 0.46 Adjusted R 2 0.45 F 25.66*** Note. n = 93. FCAT DSS = Florida Comprehensive Test 2.0 Reading Developmental Scaled Score transformed standardized score SWE = TOWRE Sight Word Efficiency; RV = PPVT 4 Receptive Vocabulary; RC = AIMSweb Reading Comprehension (R Maze). * = p < .05., ** = p < .01 ., *** = p < .001.

PAGE 111

111 Table 3 20. Simultaneous multiple regression analysis for p redicting FCAT DSS Reading for Grade 10 s ubsample Predictor Variable B SE t value Pr(>| t |) Intercept 24.96 52.50 ----Age a 0.19 1.8 0.10 .919 Months b 0.28 0.46 0.60 .555 IQ 0.73 0.23 3.19 .003** SWE 0.15 0.48 0.32 .751 PDE 0.07 0.24 0.27 .789 LWI 0.69 0.32 2.18 .037* RF 0.34 0.29 1.17 .250 PC 0.29 0.25 1.19 .242 WA 0.41 0.28 1.45 .156 RV 0.28 0.25 1.08 .287 RC 0.43 0.35 1.24 .224 ORF 0.12 0.10 1.20 .240 Multiple R 2 0.54 Adjusted R 2 0.36 F 3.10** Note. n = 45. Age a = Chronological age at time of intake. Months a = Number of months between intake and FCAT DSS Reading. FCAT DSS std. = Florida Comprehensive Test 2.0 Reading Developmental Scaled Score transformed standardized score; IQ = WASI FSIQ 2; RV = PPVT 4 Receptive Vocabulary; SWE = TOWRE Sight Word Efficien cy; PDE = TOWRE Phonemic Decoding Efficiency; LWI = WJ III ACH Letter Word Identification; RF = WJ III ACH Reading Fluency; PC = WJ III ACH Passage Comprehension; WA = WJ III ACH Word Attack; ORF = AIMSweb Oral Reading Comprehension (R CBM) ; RC = AIMSweb R eading Comprehension (Maze CBM). * p < .05., ** p < .01., *** p < .001.

PAGE 112

112 Table 3 21. Summa ry of backward elimination stepwise multiple regression models for p redicting FCAT DSS Reading for Grade 10 s ubsample Stepwise Models k AIC AIC c i AIC c Relative likelihood w i AIC c acc w i ER Grade 10 a 12 241.46 251.21 15.86 0.000 0.000 1.000 2778.70 Step 1 b 11 239.48 247.48 12.13 0.002 0.001 1.000 430.41 Step 2 c 10 237.57 244.04 8.69 0.013 0.008 0.998 77.09 Step 3 d 9 235.65 240.79 5.44 0.066 0.039 0.991 15.20 Step 4 e 8 234.34 238.34 2.99 0.224 0.132 0.952 4.46 Step 5 f 7 234.18 237.21 1.86 0.395 0.232 0.820 2.53 Step 6 g 6 233.14 235.35 0.00 1.000 0.588 0.588 Note. n = 45. a Predictors excluded: none; b Predictors excluded: age; c Predictors excluded: age, PDE; d Predictors excluded: age, PDE, SWE; e Predictors excluded: age, PDE, SWE, months; f Predictors excluded: age, PDE, SWE, months, RF; g Predictors excluded: age, PDE, SWE, months, RF, RC.

PAGE 113

113 Table 3 22. AIC c model of best fit for p redicting FCAT DSS Reading for Grade 10 s ubsample Predictor Variables B SE t value Pr(>| t |) Intercept 2.74 20.40 ------IQ 0.70 0.21 0.52 3.30 .002** LWI 0.74 0.30 0.60 2.49 .017* WA 0.50 0.30 0.41 2.17 .036* ORF 0.13 0.06 0.39 2.20 .034* PC 0.31 0.23 0.21 1.40 .17 RV 0.22 .21 0.15 1.06 .30 Multiple R 2 0.50 Adjusted R 2 0.42 F 6.39*** Note. n = 45. FCAT DSS = Florida Comprehensive Test 2.0 Reading Developmental Scaled Score transformed standardized score; IQ = WASI FSIQ 2; LWI = WJ III ACH Letter Word Identification; WA = WJ III ACH Word Attack; ORF = AIMSweb Oral Reading Fluency (R CBM) ; PC = WJ III ACH Passage Comprehension; RV = PPVT 4 Receptive Vocabulary. * p < .05., ** p < .01., *** p < .001.

PAGE 114

114 Table 3 23. Descriptive statistics for SRI and FSP s ubsamples SRI ( n = 282) FSP % ( n = 174) Demographic Variables Mean ( SD ) [ n ] Mean ( SD ) [ n ] Race/ethnicity African American 87.4 (16.1) [140] 19.5 (25.7) [188] White a 95.6 (18.3) [112] 27.6 (24.6) [128] Hispanic 87.3 (16.1) [25] 28.5 (32.3) [26] Grade Level Sixth 84.3 (19.7) [8] 40.3 (39.9) [4] Seventh 85.8 (18.9) [29] 42.3 (28.2) [23] Eighth 90.5 (173) [65] 33.9 (29.7) [36] Ninth 91.1 (17.7) [100] 13.7 (18.1) [70] Tenth 92.4 (17.9) [54] 16.4 (21.9) [29] Eleventh 94.5 (18.0) [23] 14.8 (21.6) [11] Twelfth 91.3 (11.2) [4] 28.0 [1] ESE Classification No Disability 95.1 (17.1) [124] 27.1 (29.1) [66] With Disability 87.3 (17.5) [158] 20.3 (23.5) [108] ED 88.6 (18.2) [91] 20.3 (24.0) [63] SLD 85.3 (16.8) [47] 21.6 (20.9) [34] OHI 90.6 (17.0) [10] 27.2 (41.5) [5] InD 76.0 (8.8) [7] 8.6 (10.5) [5] Note. White a = Non Hispanic. ESE = Exceptional Student Education; ED = Emotional Disturbance; SLD = Specific Learning Disability; OHI = Other Health Impaired; InD = Intellectual Disability.

PAGE 115

115 Table 3 24. Linear regression analysis for SRI p redicting FCAT DSS Reading Predictor Variable B SE t value Pr(>| t |) Intercept 47.98 4.13 ----SRI 0.39 0.04 8.80 < .001*** Multiple R 2 0.22 Adjusted R 2 0.21 F 77.20*** Note. n = 282. FCAT DSS = Florida Comprehensive Test 2.0 Reading Developmental Scaled Score transformed standardized score; SRI = Scholastic Reading Inventory. * p < .05., ** p < .01., *** p < .001.

PAGE 116

116 Table 3 25. Linear regression analysis for FSP p redicting FCAT DSS Reading Predictor Variable B SE t value Pr(>| t |) Intercept 74.14 1.30 ----FSP 0.31 0.04 8.33 < .001*** Multiple R 2 0.29 Adjusted R 2 0.28 F 69.37*** Note. n = 174. FCAT DSS = Florida Comprehensive Test 2.0 Reading Developmental Scaled Score transformed standardized score; FSP = FCAT Success Probability score. * p < .05., ** p < .01., *** p < .001.

PAGE 117

117 Table 3 26. PDA of the SRI and FSP in relation to p redicting FCAT DSS Reading Level ( > 3) FCAT DSS Reading Pass Hit Rate Sensitivity Specificity Academic Assessments % TP FN TN FP (%) (%) (%) 14.9 25 17 158 82 73.8 59.5 64.9 FSP .70 12.1 11 10 151 2 93.1 52.4 98.7 FSP .85 12.1 7 14 153 0 91.9 41.2 100 Note. SRI ( n = 282). FSP ( n = 174). SRI = Scholastic Reading Inventory; FSP FCAT Success Probability score. TP = true positives; FN = False negatives; TN = true negative; FP = false positives.

PAGE 118

118 Figure 3 1. Linear plot of c onditional FCAT DSS Reading means for imputed s ubsample .

PAGE 119

119 Figure 3 2. Residual plot for imputed s ubsample.

PAGE 120

120 Figure 3 3. Residual plot for Grade 8 s ubsample.

PAGE 121

121 Figure 3 4. Residual plot for Grade 9 s ubsample.

PAGE 122

122 Figure 3 5. Residual plot for Grade 10 s ubsample.

PAGE 123

123 Figure 3 6. Residual plot for SRI s ubsample.

PAGE 124

124 Figure 3 7. Residual plot for FSP s ubsample.

PAGE 125

125 CHAPTER 4 DISCUSSION The present study sought to provide a more comprehensive depiction of the academic and reading skills of incarcerated youth than is available in the extant literature by exploring, and potentially establishing, the specific reading skills required to demonstrate proficiency on a statewide reading assessment within the state of Florida. Specifically, the present study explored t he relationships between academic and reading skills and FCAT performance across grades and special education disability categories in a population of incarcerated juveniles , as well as the predictive validity of multiple progress monitoring instruments in regard to FCAT performance. This study attempted to examine these relationships by addressing the following questions: Which specific academic and reading skills are most predictive of success on the reading section of th e FCAT DSS for individuals in a juvenile corrections facility? What are the relationships between student demographic variables, academic performance variables, and FCAT DSS Reading scores for these students? Are there significant mean differences on the F CAT DSS Reading between individuals within different special education disability categories? Which student academic performance variables are most predictive of FCAT DSS Reading scores? Of the selected progress monitoring measures (SRI and FAIR), which me asure is more predictive in regard to FCAT DSS Reading scores? Which progress monitoring measure is more accurate in predicting adequate performance on the FCAT DSS Reading? General Strengths of this Study The present study is perhaps one of the most com prehensive examinations of the specific reading skills exhibited by students within a juvenile correctional facility to date. Further, this study is one of the few studies that has examined the specific reading abilities of delinquent youth disaggregated b y grade level and special education disability subtype. Additionally, this

PAGE 126

126 study is the first to examine the multifarious relationships between academic and reading skills and FCAT performance across both grade levels and special education disability categ ories in a population of juvenile delinquents. This study is also the first study to examine the predictive validity of multiple progress monitoring instruments in regard to FCAT performance within this population. The present study obtained a comprehensiv e profile of the reading skills possessed by incarcerated youth, and explored how these skills predict performance on a high stakes statewide test. Research Question One Similar to the findings of previous researchers who studied the academic abilities of incarcerated juveniles (e.g., Baltodano et al., 2005; Brunner, 1993; Drakeford & Krezmien, 2004; Foley 2001; Harris et al., 2009; Hodges et al., 1994; Krezmien et al. , 2013), the observed abilities of delinquent youth within this study were significantly lower across nearly all cognitive and academic domains when compared to normative population means. Specifically, participants achieved on average approximately one standard deviation below the normative mean on each respective measure. These scores indica te that juveniles within this study possess substantial deficits across multiple domains related to reading proficiency. In regard to intelligence, individuals within the study scored within the low average range, indicating that the IQ of participants is considerably below the normative mean. The estimate of the cognitive capacity of juvenile delinquents provided by the current study is consistent with numerous previous studies which also estimated intelligence of incarcerated youth (Foley, 2001). As expec ted, the academic performance variables included in the study all demonstrated moderate to strong inter variable correlations. All academic performance variables were also positively correlated with FCAT DSS Reading performance (i.e., moderate correlations ).

PAGE 127

127 In respect to spe cific reading skills, the mean Receptive V ocabulary (RV) scores of participants, measured by the PPVT 4, was in the low average range. Indicating that delinquent youth within the study possess a below average vocabulary which likely in hibits their ability to read and comprehend grade level texts ( National Reading Panel , 2000; Fowlert & Swainson, 2004). Participants within the study also achieved mean scores within the low average range on measures of both Sight Word E fficiency (SWE) and Phonemic Decoding E fficiency (PDE), as measured by the TOWRE. Deficiencies in sight word and phonemic decoding efficiency indicate that participants struggle to read individual words accurately and fluently when timed; this is particularly troublesome as inefficient phonemic processing contributes to poor reading fluency (Ashby, Dix, Bontrager, Dey, & Archer, 2013). A recent study conducted by Wilkerson et al. (2012) underscores the need for systematic phonics instruction within juvenile correctional setti ngs as teachers reported that it was one of the least utilized instructional strategies, with 22.4% of teachers reporting that they never explicitly teach these skills. The ability of incarcerated youth within the study to correctly decode individual lette rs and words, measured by Letter Word Identification (LWI) subtest on the WJ III ACH, was in the low average range as well. Deficiencies on decoding at the word level impede reading fluency and can render the nearly impossible (Hudson, Lane, & Pullen, 2005). Participants attained average mean scores within the below average range on the Reading Fluency (RF) subtest on the WJ III ACH , which required them to quickly read simple sentences, decide if the answer wa s correct, and provide a response . The importance of reading fluency is well established, including its foundational nature in regard to the acquisition of reading Kim, Wagner, & Lopez, 2012 ). The ability of participants to read and comprehend text, as measured by the

PAGE 128

128 Passage Comprehension (PC) subtest on the WJ III ACH, was in the below average range. 3), is often participants to decode unfamiliar words was within the average range on the Word Attack (WA) subtest of the WJ III ACH. The performances by participant s on the Word Attack are not consistent with the results reported by researchers Harris et al. (2009) or Krezmien et al. (2013), as individuals within the study scored significantly higher on this measure. Although a high degree of concordance exists betwe en the present research and the results of the aforementioned researchers, it is unclear why significant performance differences were observed on the Word Attack subtest. The mean scores observed on the AIMSweb subtests measuring Oral Reading F luency (ORF ; R CBM) and Reading C omprehension (RC; Maze CBM) indicated that participants fell far below grade level expectations. Due to the discord between grade level placement and actual age (i.e., participants were on average two grades lower than what would be e xpected by average age), and that only sixth grade prompts were administered, typical standardization procedures were not appropriate in regard to data interpretation. Thus raw data comparisons were utilized. Using national norms provided by AIMSweb (2012) , individuals in this sample scored slightly above sixth graders in regard to ORF and below sixth graders in RC . This is especially troubling considering that the actual average grade level placement of participants was between the eighth and ninth grades. In addition, the average age of individuals within the study is consistent with individuals who are typically tenth or possibly even eleventh graders. These results are consistent with published estimates that place the overall reading ability of incarcer ated youth between the

PAGE 129

129 fourth and ninth grade (e.g., Baltodano et al., 2005; Brunner, 1993; Foley 2001; Hodges et al., 1994). On the dependent variable, the FCAT DSS Reading, study participants scored approximately one standard deviation below the normativ e average when compared to all test takers who took the FCAT in the state of Florida. Researchers such as Forsyth et al. (2010) have detailed the poor performance of incarcerated youth on high stakes statewide assessments; however, the participants within the present study attained at a least basic level of proficiency at only half the rate as the incarcerated youth in Louisiana studied by Forsyth et al. (2010). The percentage of incarcerated youth within this study who demonstrated at least a basic level o f proficiency on the FCAT DSS Reading was nearly identical to the overall passage rate of incarcerated youth across the state reported by the Florida Office of Program Policy Analysis and Government Accountability in 2010, indicating that the sample was re presentative of incarcerated youth across the state in terms of FCAT performance. These results indicate that not only did incarcerated youth within this study score significantly lower than their non incarcerated peers, but that they were also four times less likely to pass the FCAT DSS Reading than peers who were not incarcerated. The unbalanced numbers of participants within the study across grade levels, and the wide range of ages present within each grade, make inferences in regard to grade level comparisons difficult . H owever, chronological age can be used as a rough metric to determine if a linear relationship between academic performance and age, and subsequently grade exists. Although Harris et al. (2009) did not observe significant differences in reading performance across age when studying incarcera ted youth, Krezmien et al. (2013 ) found age to be a significant predictor of reading performance. The results of the current study align with those

PAGE 130

130 pr esented by Krezmien et al. (2013 ), as age proved to have a negative predictive relationship in regard to all academic performance variables, as well as performance on the FCAT DSS Reading. These results indicate that the abilities of older incarcerated youth in this study appear to remain stagnant and m ay be prone to regression when compared to younger incarcerated peers within the study . Although Krezmien et al. (2013 ) surmised that the negative relationship found in this setting are not learning the necessary skills to improve reading performance (p. 79); the present author exercises a more cautious approach. For instance, data pr esented by Krezmien et al. (2013 ) and the current study possessed insufficient information in regard to potential qua litative similarities and differences between older and younger students within the respective samples, such as: age at first incarceration, recidivism rates, length of stay, comorbid psychiatric diagnoses, attitudes towards reading and school, etc. Althou gh older youth within the sample demonstrated poorer performances across all independent and dependent academic variables when compared to their younger counterparts, a longitudinal approach and data on additional variables would be necessary to determine causality. In regard to race/ethnicity, African American students were over represented within the sample. H owever, overrepresentation of African American and minority students within juvenile justice populations is common (Jenson et al., 2001; Leone et a l., 2002; OJJDP, 2012; Snyder & Sickmund, 2006). Due to the auxiliary nature of race/ethnicity to the main purposes of this study, data were not disaggregated between different racial/ethnic groups in respect to special education group membership. However, mean scores on all academic performance variables as well as the FCAT DSS Reading were disaggregated by race/ethnicity. Of note, the means for African American students were lower across all academic performance variables compared to their

PAGE 131

131 White/Non Hispa nic peers. Likewise, White/Non Hispanic participants scored higher on average than their African American counterparts on the FCAT DSS Reading. The majority of these differences were slight and nonsignificant in nature. However, the differences between the observed mean scores between African American and White/Non Hispanic participants on the PPVT 4, Passage Comprehension (PC) subtest of the WJ III ACH, and the AIMSweb (ORF; R CBM) were more substantial. White/Non Hispanic participants also achieved higher mean scores across all academic performance variables compared to their Hispanic peers, except for the Phonemic Decoding Efficiency (PDE) subtest of the TOWRE and the Letter Word Identification (LWI) subtest of the WJ III ACH. Hispanic students achieved s lightly higher mean scores on the FCAT DSS Reading than White/Non Hispanic peers. Hispanic students scored higher across all academic performance variables compared to their African American peers, except for Sight Word Efficiency (SWE) subtest of the TOWR E and the Passage Comprehension (PC) subtest of the WJ III ACH. Hispanic students also attained, on average, higher scores on the FCAT DSS Reading compared to their African American peers. While racial differences in performance were observed, these differ ences were smaller than the disparities documented by other researchers studying youth in juvenile correctional facilities (e.g., Harris et al., 2009; Krezmien et al., 2013). Possible explanations of racial disparities evidenced on measures of academic per formance (e.g., cultural factors, environmental factors, disparate educational opportunities, and differences in intelligence) are commonplace throughout the literature and will not be reiterated here as they are of little consequence to the current study. The prevalence rate of individuals possessing special education exceptionalities was approximately 58%. The proportion of individuals within special education was significantly higher than the prevalence rates observed in contemporary studies such as Ha rris et al. (2009)

PAGE 132

132 and Krezmien et al. (2013), who reported prevalence rates of 33% and 42%, respectively. Likewise, the prevalence rate of the sample was higher than the median national average of 33% for incarcerated youth reported by Quinn et al. (2005) ; however, the aforementioned researchers documented prevalence rates as high as 77%. The observed prevalence rate of the current sample dwarfs the national average of youth who have educational disabilities within public education, which is estimat ed at 1 3 % ( Snyder & Dillow , 2013 ). Similar to others studies (e.g., Quinn et al. 2005), ED and SLD were the most common special education exceptionalities observed within the sample. The prevalence rates of ED and SLD within special education were approximately 58% and 30% , respectively. Youth in the sample with special education exceptionalities scored lower across all academic performance variables in comparison to their non disabled peers, and in many cases the se differences were significant. T hese results are consistent with those of Harris et al. (2009) and Krezmien et al. (2013). Overall, special education students score nearly one and a half standard deviations below the normative mean on each respective measure. Variations in observed mean scores across ac ademic performance variables were inconstant relative to specific special education exceptionalities and were generally statistically small. However, as expected, individuals with an InD classification scored significantly lower than other students within special education across all academic performance variables and on the FCAT DSS Reading. Interestingly, the reported mean IQ of individuals with an InD exceptionality was smaller than two standard deviations below the normative population mean, indicating that these individuals may possess greater cognitive abilities than expected . H owever, due to the small sample size of this subgroup, the results are likely attributable to regression to the mean. Students possessing special education exceptionalities also scored lower than their non disabled peers on the FCAT DSS Reading. Due to small participant numbers across specific special education categories, it

PAGE 133

133 was only possible to conduct a one way ANOVA which compared the mean differences on FCAT DSS Reading scor es between the following groups: Non ESE vs. ED, Non ESE vs. SLD, and ED vs. SLD. Results indicated that group membership had a statistically significant effect, with a small effect size on FCAT DSS Reading performance. Further, Comparison testing also ind icated that the mean score differences of Non ESE vs. ED, and Non ESE vs. SLD were both statistically significant. It is hypothesized that mean performance differences were due to one, or multiple specific educational deficits that are cumulatively reducti ve in nature and further manifest into statistically poorer performance on omnibus measures of reading and language ability, such as the FCAT DSS Reading. Of note, the mean score differences observed between ED and SLD were not statistically significant an d likely due to chance. The lack of statistical difference on the FCAT DSS Reading scores between the ED and SLD group is of interest, as differences have been documented in studies of non incarcerated individuals between these groups on measures of IQ (e. g., McHale, Obrzut, & Sabers, 2003). Performance differences have also been noted between these special education groups on measures of academic achievement on studies including incarcerated youth (e.g., Harris et al., 2009). Further, 10% of the individual s within this study have both primary and secondary educational disability classifications, most commonly ED and SLD in varying order. The comorbidity rate between ED and SLD was 12% within the sample, thus the possibility of a complex interplay between ED and SLD may confound results. I t is difficult to determine which disability (i.e., if multiple disabilities are Of note , IDEIA ( 2006 ) allows for the use of three theoretically di fferent methods to determine eligibility for SLD (i.e., traditional ability achievement discrepancy approach, Response to Intervention [RtI], and for alternative research . F . R . § 300. 309 ).

PAGE 134

134 Discord exists t hroughout professional practice and research regarding not only the definitional ambiguity of SLD criteria, but also as to which method provides the most accurate assessment of individuals suspected of meeting SLD criteria (Hale, Kaufman, Naglieri, & Kaval e, 2006). Definitional ambiguity, as well as the discord between methods used to determine eligibility, often leads to SLD samples that are heterogeneous (Flanagan, Ortiz, Alfonso, & Dynda, 2006). Due to the archival nature of the current study and that da ta came from multiple school districts throughout the state of Florida, d etermining which method was used for SLD eligibility determination was not possible. Research Question Two Understanding the specific academic characteristics of incarcerated youth i s critical to not only predict future academic performance, but also if researchers are to design evidenced based curricula and instructional interventions (Gagnon & Barber, 2010). Similar to researchers Harris et al. (2009) and Krezmien et al. (2013), thi s study attempted to provide a contemporary profile of the reading skills of delinquent youth; however, the present study expanded on the research of the aforementioned authors by including more predictor variables (i.e., age at intake, months between inta ke and FCAT DSS Reading administration, IQ, RV, SWE, PDE, LWI, RF, PC,WA, ORF, and RC) and by constructing a series of stepwise multiple regression models that utilized a statewide omnibus reading measure as the outcome variable. Regression models, based o n AIC c , were conducted on the overall imputed subsample and additionally on three grade levels (i.e., Grades 8, 9, 10) in an effort to identify which predictor variables explained the most variance in terms of FCAT DSS Reading performance. The overall reg ression model, which included all regressors and all participants within the subsample, was highly predictive of FCAT DSS Reading performance. Three of the independent variables (i.e., IQ, RC, and months between intake and FCAT DSS Reading

PAGE 135

135 administration) exhibited significant main effects in regard to FCAT DSS Reading performance. These results indicate that scoring highly on the aforementioned variables is related to achieving higher FCAT DSS Rea ding scores. H owever, predictor importance within the model could not be determined due to the use of unstandardized regression coefficients. When stepwise multiple regression based on bias corrected Akaike information criterion was utilized to determine the most parsimonious model, the following variables (present ed serially in terms of significance) were retained: IQ, SWE, RC, months between intake and FCAT DSS Reading administration, and RV were retained. The effect size for IQ was medium to large, while the effect size for SWE was small. The two remaining predic tors (i.e., RC and months between intake and FCAT DSS Reading administration) did not prod uce substantively significant effect sizes. The final model, selected based on AIC c , had a statistically significant effect on FCAT DSS Reading scores, with a large e ffect size. A large effect size indicates that this model has practical utility in predicting FCAT DSS Scores. Generally speaking, these results are consistent with the results reported by researchers Tighe and Schatschneider (2013), as the researchers doc umented the increased role of intelligence in predicting FCAT Reading outcomes as the test increases in complexity. This model is only illustrative in nature, as interpretations of data resulting from collapsing grade levels into a singular analysis can be fallacious as different FCAT tests are not equivalent. Thus, the only meaningful inference in regard to this model is that SWE, RC, RV, and in particular IQ represent important constructs that are predictive of FCAT DSS Reading success across grades . Addi tionally, interpreting the importance of the months between intake and FCAT DSS Reading administration variable is difficult . It is hypothesized that the significance of the variable is related to a possible stabilization effect on student performance, as participants received daily academic instruction and possibly medication while they were at the secure care facility .

PAGE 136

136 H owever, it is possible that the variable is a statistical anomaly or that another explanation exists . Grade level models were constructed to provide a more detailed analysis of the specific reading skills necessary to demonstrate proficiency at each grade level on the FCAT DSS Reading. In regard to performance on the eighth grade FCAT DSS Reading, the most parsimonious model included IQ, R C, months between intake and FCAT DSS Reading administration, and WA (arranged serially by relative predictive importance). The final model had a substantial and statistically significant effect on FCAT DSS Reading scores, with a large effect size. In the final model, IQ was a significant predictor and had medium to large effect size, while RC, months between intake and FCAT DSS Reading administration, and WA had small effect sizes . Despite removing eight predictor variables, the model remained approximatel y equivalent to the original model, while gaining a small amount of predictive power. The final model for the ninth grade FCAT DSS Reading retained the following as predictor variables: RC, RV, and SWE . All three of the retained variables exhibited signif icant main effects. Examination of the predictor variables indicated that the effect sizes for SWE and RV were both medium, while the effect size for RC was small. Further, the final model had a substantial and statistically significant effect on FCAT DSS Reading scores, and also had a large effect size. DSS Reading included IQ, LWI, WA, ORF, PC and RV within the respective model. The model was highly predictive and had a large effect size. Four of the independent variables exhibited significant main effects in regard to prediction of the outcome variable (i.e., IQ, LWI, WA, and ORF). Examination of the predictor variables indicated the effect size for IQ was large, while WA and ORF had small effects on the outcome variable. None of remaining variables produced substantive ly significant

PAGE 137

137 effect sizes. Although this model was selected as the most parsimonious for Grade 10, due to the small sample size, results should be interpreted cautiously. The results of grade level analyses must be interpreted with caution due to a larger than expected degree of discordance between respective grade level models. Specifically, no single variable was retained as a predictor across all four models after stepwi se procedures selected for parsimony, although correlations between measures across grade levels remained relatively stable. Further , because the FCAT grows increasingly complex as students matriculate through school, there should be a gradual shift away f rom basic skills and towards skills that are more cognitively complex, or g loaded, in regard to predictive importance when using modeling procedures (Schatscneider et al., 2004; Tighe & Schatschneider, 2013) . However, in the current study at Grade 9, SWE was retained over other variables which have higher g loadings, such as IQ. Likewise, at Grade 10, LWI and WA were retained over other variables that are more indicative of higher order thinking skills such as RC. Thus, the grade level models constructed w ithin the study may not provide wholly accurate estimates in regard to the requisite skills related to FCAT DSS Reading performance at respective grade levels outside of the sample. The general discordance between expectations and results in respect to gra de level model selection may have resulted from small sample sizes, sampling error, or model overfitting. However, many of the variables that emerged as significant predictors of FCAT performance within grades were similar to those outlined by previous res earchers (e.g., Schatschneider et al., 2004 ; Tighe & Schatschneider 2013). For the overall model which included all students, and within all but one grade level model (i.e., Grade 9), psychometric g measured by the FSIQ on the WASI proved to be the single most important predictor of FCAT DSS Reading scores. Further, higher order skills related to psychometric g (e.g., reading comprehension) also proved to be significant predictors

PAGE 138

138 in relation to FCAT DSS Reading performance. These skills become increasingly important predictors of FCAT performance beyond the primary grades, while ORF ability is typically the most predictive reading skill related to FCAT performance for students who have yet to reach middle school (Schatschneider et al., 2004 ; Tighe & Schatsc hneider 2013). For example, SWE, specified by Tighe and Schatschednieder (2013) as a measure of reading fluency, was a significant predictor of FCAT DSS Reading performance; however, its predictive importance was significantly less in the present study com pared to other studies which utilized measures of reading fluency as FCAT Reading predictors within the primary grades (Buck & Torgeson, 2003; Schatschneider et al., 2004 ; Tighe & Schatschneider 2013; Torgeson et al. 2003 ). Of note, due to range restrictio ns on some achievement measures, the observed magnitude of relationships may be limited. Research Question Three Although researchers have begun to establish the test retest reliability, convergent validity, and predictive utility of the FAIR (e.g., Carl son, et al. 2010; Foorman et al. 2013; Foorman & Petscher, 2011 ; Joyce et al., 2013 ; Petscher & Foorman, 2011), there is little information in the extant literature regarding the predictive utility of the FAIR for exceptional populations. Further, there ar e currently no published studies which examine the predictive utility of the FAIR that include incarcerated students within study samples. This research question serves as an evaluation of the FAIR FSP as a means to predict future FCAT DSS Reading performa nce for delinquent youth. The rationale for inclusion of the SRI as a comparison measure was based on the work of Algozzine et al. (2011), which indicated that the SRI I was a significant predictor of FCAT SSS Reading performance in Grades 6, 7, and 8. Thu s, the SRI served as a litmus test for comparison, as the SRI is a nationally standardized assessment tool for measuring reading proficiency.

PAGE 139

13 9 When utilizing linear regression models, both measures were statistically significant predictors of FCAT DSS perf ormance in isolation. Not surprisingly, the FSP accounted for more variance in respect to overall FCAT DSS Reading performance compared to that explained by the SRI. The FSP also exhibited a large effect size on FCAT DSS Reading, while the SRI had a medium effect size. These results indicate that students who achieve higher SRI and/or FSP scores are likely to achieve higher scores on the FCAT DSS Reading. The predictive properties of the SRI found within this study were similar to those reported by Algozzin e et al. (2011). Likewise, the predictive properties of the FAIR FSP found within this study were similar to those reported by previous researchers (i.e., Foorman et al. 2013; Foorman & Petscher, 2011 ; Joyce et al. 2013 ). These results indicate that both m easures adequately predict FCAT DSS Reading performance, although the FAIR FSP would be indicated for use between the two. Predictive discriminant analysis (PDA) was employed to determine the hit rate, sensitivity, and specificity of each instrument for p redicting FCAT DSS Reading proficiency. In regard to PDA, hit rate is indicative of how well each respective measure predicted overall FCAT DSS Reading performance. The sensitivity index, or the true positive rate, is a measure of how well each measure acc urately predicted the percentage of students who mastered the skills measured by the FCAT DSS Reading. Specificity, or the true negative rate, reflects how well each measure accurately identified students who did not master the skills measured by the FCAT DSS Reading. Of note, proficiency on the outcome measure was defined by a designation of Level 3 or above on the FCAT DSS Reading. Proficiency criteria on the independent points of .70 and .85 on the FAIR FSP.

PAGE 140

140 Results of the present study indicate that both measures are highly discriminant in terms of FCAT DSS Reading proficiency prediction. Both measures are more accurate in predicting specificity (i.e., true negatives) than predicting sensitivit y (i.e., true positives) on the FCAT DSS Reading. The SRI demonstrated a higher rate of sensitivity compared to the sensitivity rates e xpressed by both FSP cut points . Indicating that the SRI is slightly more effective at identifying students who have mast ered the skills necessary to achieve a roficient designation on the FCAT DSS Reading. However, the SRI produced a significantly lower hit rate and a significantly lower specificity rate compared to the rates expressed by both the .70 and .85 cut point d esignations of the FSP. This indicates that both FSP cut points, overall, serve as better predictive measures in terms of overall FCAT DSS Reading proficiency. The FAIR FSP cut point of .70 outperformed the FSP cut point of .85 in respect to hit rate and s ensitivity, while producing an equivalent specificity rate. These results indicate that an FSP cut point of .70 was the most efficacious of the three independent variables in terms of FCAT DSS Reading proficiency prediction. Using a cut point of .70 is con tradictory to suggestions made by the Florida Department of Edcuation (2009) which endorses using an FSP cut point of .85; however, it is congruent with Foorman et al. (2013), who found that the use of the .70 cut point produced better predictive balance. The sensitivity and specificity estimates produced by the SRI were significantly lower than those reported by Algozzine et al. (2011). This could be due in part to population differences, as well as potential differences in grade sampling, and/or by the in creased rigor of the FCAT 2.0 as the aforementioned researchers used the FCAT 1.0 as the primary outcome variable. The results produced by research question three indicate that both measures reliably predicted future performance on the FCAT DSS Reading wit hin the sample; however, a FAIR

PAGE 141

141 FSP cut point of .70 was the most efficacious in terms of overall predictive power. Although a smaller amount of variance was accounted for by the FAIR FSP compared to some of the more complex regression models outlined in r esearch question two, which typically contained multiple assessment tools, the FAIR FSP likely has more practical utility in terms of FCAT DSS Reading prediction. This is due to a number of factors, including: the fact that the FAIR is already being admini stered within the state of Florida, it is supported by an online data collection system, it is more sensitive to slight changes in performances, it is less cost prohibitive than using a menagerie of measures, and it requires less training. However, despite its utility, the FAIR FSP does not describe the specific reading skills necessary to demonstrate proficiency on the FCAT DSS Reading that are accounted for by the more complex regression models. General Limitations The current study is subject to severa l limitations. The data presented are only representative of incarcerated male students within a single facility in one state. Thus, the sample may not be representative of the demographic characteristics or reading ability of incarcerated youth outside of the facility or state where data was collected. Likewise, due to relatively small sample sizes across some special education disability subtypes (e.g., InD) and grades levels, interpretations of academic and reading abilities and their respective function s in relation to FCAT performance disaggregated by grade level and disability subtype must be made with caution as data may not provide an accurate representation beyond the sample. Replication studies are needed to validate the findings of the present stu dy. Missing data compounded by the necessary use of imputation also encourage caution in interpreting the results provided by the present study. Specifically, preliminary analysis of the dataset concluded that 20 individuals, comprising 7.1% of the sample, were missing data across all independent variables related to academic assessment and were therefore excluded from

PAGE 142

142 research questions one and two. To prevent further deletion of cases containing only partial academic assessment data, single imputation by chained equations approach was conducted to impute additional missing values (i.e., 1.3%). For research question three, only one individual, or 0.4% of the sample, was excluded from the SRI model. However, 109 individuals (i.e., 38.5% of the sample) were n ot included in the FSP model due to missing data. Although the mean score differences between the original dataset and the imputed dataset used for research questions one and two were nonsignificant in nature, it is undeterminable if the FSP regression mod el results reported for research question three are representative of the overall sample due to the large amount of missing data. Although appropriate steps and analyses were conducted to ensure the fidelity, reliability, and reproducibility of the data, it is potentially problematic that a large number of test administrators collected participant data over the course of three years. To reduce possible sources of bias, the assessments would have preferably been completed utilizing a more limited number of raters over the course of a single assessment period. Additionally, the use of a comprehensive battery such as the Woodcock Johnson III Tests of Cognitive Abilities Normative Update (WJ III COG; Woodcock, McGrew, & Mather, 2007) in lieu of the abbreviated WASI, may have provided a more representative picture of the cognitive abilities of youth within the sample. Although the WASI provides an excellent approximation of psychometric g (Canivez et al., 2009), the FSIQ 2 only measures two of the broad Cattell Horn Carroll (CHC) abilities (i.e., comprehension knowledge [ Gc ], and fluid reasoning [ Gf ]). As outlined by Newton and McGrew (2010, p. 622 ), the contemporary CHC taxonomy of cognitive Horn Gf Gc (Horn, 1989) and the Carroll (1993) three stratum models of human cognitive abilities

PAGE 143

143 assessment of bot h broad and narrow CHC abilities may have provided information about the specific cognitive abilities of this population beyond psychometric g in regard to school achievement (e.g., see Evans, Floyd, McGrew, & Leforgee, 2002; Flanagan et al., 2006; Floyd, Keith, Taub, & McGrew, 2007; McGrew and Wendling 2010; Newton & McGrew, 2010). The inclusion of data concerning possible comorbid psychiatric diagnoses and secondary educational disabilities may have been useful in providing a more comprehensive profile o f individuals within the sample. Further, the possibility that different classification criteria were used to determine educational disabilities (e.g., SLD) may inhibit generalizations. Data with respect to SES, as well as student perceptions of school and reading may have also been valuable. Additionally, the study only included males which prevented the capability to explore possible gender effects. Finally, the inclusion of the aforementioned variables as well as the use of more complex multivariate anal yses may have yielded a more precise and informative representation of the intricate relationships between the predictor and dependent variables. Summary and Implications for Future Research Consistent with other published studies detailing the characteri stics of delinquent youth, the participants comprising the study sample were disproportionally minority, disproportionally received special education services, and presented with significant intellectual and academic deficits compared to non incarcerated p eers. These deficits were prevalent across all measures and manifested into a collective lack of roficient designations on the Florida statewide omnibus measure of reading performance. Further, incarcerated students within special education fared signif icantly worse across all measures compared to their incarcerated counterparts. Although these results echo those of previous researchers (e.g., Baltodano et al., 2005; Brunner, 1993; Drakeford & Krezmien, 2004; Foley 2001; Harris et al., 2009; Hodges et al ., 1994; Krezmien, et al. 2013), they are no less harrowing as the marriage between poor academic

PAGE 144

144 performance and criminality is rarely fortuitous. The results are also particularly troubling as the academic deficiencies demonstrated by delinquent youth of ten serve as harbingers for future deleterious life outcomes (Gagnon & Barber, 2010). In an effort to quantify the importance of specific reading skills in relation to FCAT performance, a host of academic performance variables where used as predictor varia bles to predict FCAT DSS Reading scores. When all FCAT data across grades was combined, the present study found that IQ, SWE, RC, and RV were the most important academic predictor variables in regard to FCAT DSS reading performance . Over a ll, IQ or psychome tric g , measured by the WASI, proved to be the single most important predictor of FCAT DSS Reading scores . Lastly, a subsequent predictive discriminant analysis indicated that the FAIR FSP (i.e., at levels .70 and .85) served as an adequate predictor of FC AT DSS Reading proficiency in respect to hit rate, sensitivity, and specificity . Although both cut points were adequate in prediction of FCAT scores, a cut point of .70 is recommended when working within juvenile corrections as it possessed the greatest pr edictive utility for this population . In addition to conducting studies that address the limitations of the present study, future researchers should use the results of this study to facilitate the development, implementation, and assessment of evidence b ased academic curriculums and interventions in an effort to improve the overall academic and reading abilities of incarcerated youth. Specifically, these interventions should possess practical utility for educational practitioners and focus primarily on im proving the knowledge and skills necessary to increase the reading comprehension and higher order thinking skills of youth within juvenile correctional facilities across grade levels. However, because the ability to read fluently and comprehend text hinges on previous acquisition of basic foundational reading skills, which these youth are severely deficient, interventions must first place emphasis

PAGE 145

145 on improving basic reading skills in an effort to bolster the ability of these youth to problem solve and compr ehend text. In addition to performing below grade level on measures of reading comprehension, results of this study indicate that incarcerated youth exhibit deficiencies in phonemic processing, word decoding, reading fluency, and possess immature vocabular ies. To help youth within juvenile corrections gain the knowledge and basic reading skills necessary to read fluently and comprehend text proficiency, the following areas of instructional strategies outlined by the National Reading Panel (2000) should be u tilized in (i.e., in addition to reading comprehension instruction): systematic phonics instruction, fluency instruction, and vocabulary instruction. Although systematic reading interventions within juvenile corrections are not commonplace (e.g., see Gagno n & Barber, in press), educational practitioners within these settings should borrow and utilize established evidence based reading instruction strategies from domains outside of juvenile corrections to teach incarcerated youth basic reading skills, as wel l as reading comprehension strategies. For example, the What Works Clearinghouse (U. S. Department of Education, 2014) provides information on numerous evidence based reading strategies and programs for adolescents, some of these include: Repeated reading, Student team reading and writing, Peer Assisted Learning Strategies (PALS), SuccessMaker®, Fast ForWord®, Read 180®, Lindamood Phoneme Sequencing® (LiPS), and SpellRead TM . Though the literature outside of juvenile corrections is rich with instructional st rategies, many teachers within juvenile corrections never explicitly teach the basic reading skills that are critical to reading and comprehending text ( Wilkerson et al., 2012) . Further, in an effort to reduce recidivism rates and promote positive life out comes, special consideration should be given to employing systematic reading interventions with first time offenders and youth in the early

PAGE 146

146 stages of adolescence within juvenile correctional facilities, as a negative predictive relationship exists between age and reading achievement. Subsequently, the overall effectiveness of these programs should be weighed against student performance on omnibus measures of academic achievement such as statewide assessments, and any additional appropriate long te rm outcome data available (e.g., recidivism rates). An increased emphasis on conducting empirically based studies of reading interventions within juvenile correctional facilities is crucial as the current literature base is severely deficient. The few empi rically based reading intervention studies within the extant literature are rife with methodological concerns and thus lack generalizability. Although previous researchers have conducted studies of reading interventions within juvenile corrections populati ons, these studies rarely utilize randomized designs, lack statistical rigor, and are composed of especially small sample sizes (e.g., Allen DeBoer et al., 2006; Drakeford, 2002; Houchins et al., 2008; Malmgren & Leone, 2000; Shippen, Morton, Flynt, Houchi ns, & Smitherman, 2012). Last , future efforts to improve student achievement should be tied to professional development within juvenile correctional facilities, as suggested by Gagnon et al. (2012). Increased attention should also be given to designing studies that focus on improving the academic outcomes of delinquent students who possess special education exceptionalities, as these individuals have demonstrated great need in regard to high quality academic instruction . Encouragingly, it is possible tha t many of students across special education categories will benefit from similar intervention strategies due to a high degree of concordance in regard to specific reading deficits , overall reading ability, and intelligence. The present study, along with ot hers, has established the demographic profiles and adverse academic realities of delinquent

PAGE 147

147 youth. It is now time for researchers to uphold their social contract to society and begin to change these realities.

PAGE 148

148 LIST OF REFERENCES Abram, K. M., Teplin, L. A., McClelland, G. M., & Duncan, M. K. (2003). Comorbid psychiatric disorders in youth in juvenile detention. Archives of General Psychiatry , 60 , 1097 1108. AIMSweb. (2012). AIMSweb national norms . Pearson Education, Inc. Retrieved from http://aimsweb.com/wp content/uploads/AIMSweb National Norms Technical Documentation.pdf Akaike, H. (1973). Information theory and an extension of the maximum likeli hood principle. In B. N. Petriv & F. Csaki (Eds.), Second international symposium on information theory , Academiai Kiado, Budapest, (pp.267 281). Akaike, H. (1978). A Bayesian analysis of the minimum AIC procedure. Ann, Inst. Statist. Math, A, 9 14. Al gozzine, B., Wang, C., & Boukhtiarov, A. (2011). A comparison of progress monitoring scores and end of grade achievement. New Waves Educational Research & Development , 14 (1), 3 21. Allen DeBoer, R. A., Malmgren, K. W., & Glass, M. (2006). Reading instruc tion for youth with emotional and behavioral disorders in a juvenile correctional facility. Behavioral Disorders , 32 (1), 18 28. Allison, J. R., & Johnson, E. S. (2011). Identifying struggling readers in middle school with ORF, Maze and prior year assessment data. Journal of Educational and Developmental Psychology , 1 (1), 35 44. Altman, J., Vang, M., & Thurlow, M. (2012). 2008 2009 APR snapshot #1: State assessment participation and performance of special education students . Minneapolis, MN: University of Minnesota, National Center on Educational Outcomes. Retrieved from http://cehd.umn.edu/NCEO/APRsnapshot/partperf/default.html Anthony, E. K ., Samples, M. D., de Kervor, D. N., Ituarte, S., Lee, C., & Austin, M. J. (2010). Coming back home: The reintegration of formerly incarcerated youth with service implications. Children and Youth Services Review, 32 , 1271 1277. Archwamety, T., & Katsiyan nis. (2007). Academic remediation, parole violations, and recidivism rates among delinquent youths. Remedial and Special Education , 21 (3), 161 170. Ardoin, S. P., Witt, J. C., Suldo, S. M., Connell, J. E., Koenig, J. L., Resetar, J. L., et al. (2004). E xamining the incremental benefits of administering a maze and three versus one curriculum based measurement reading probes when conducting universal screening. School Psychology Review , 33 , 218 233.

PAGE 149

149 Ashby, J., Dix, H., Bontrager, M., Dey, R., & Archer, A. (2013). Phonemic awareness contributes to text reading fluency: Evidence from eye movements. School Psychology Review, 42 (2), 157 170. Baltodano, H. M., Harris, P. J., & Rutherford, R. B. (2005). Academic Achievement in Juvenile Corrections: Examinin g the impact of age, ethnicity, and disability. Education and Treatment of Children , 28 (4), 361 379. Blomberg, T. G., Blomberg, J., Waldo, G. P., Pesta, G., & Bellows, J. (2006). Juvenile justice education, no child left behind, and the national collabo ration project. Corrections Today , 63 , 143 146. theory and its analytical extensions. Psychometrika, 52 (3), 345 370. complexity. Journal of Mathematical Psychology, 44, 62 91. Brunner, M. S. (1993). Reading recidivism and increased employment opportunity through research based reading instruction . NCJ Publication No. 141324. Washington, DC: U.S. Department of Justice, Office of Juvenile Justice and Delinquency Prevention. Buck, J., & Torgesen, J. (2003). The relationship between performance on a measure of oral reading fluenc y and performance on the Florida Comprehensive Assessment Test (Technical Report 1). Tallahassee, FL: Florida Center for Reading Research. Buckland, S. T., Burnham, K. P., & Augustin, N. H. (1997). Model selection: An integral part of inference. Biometri cs, 54 (2), 603 618. Bullis, M., & Yovanoff, P. (2006). Idle hands: Community employment experiences of formerly incarcerated youth. Journal of Emotional and Behavioral Disorders, 14 (2), 71 85. Bullis, M., Yovanoff, P., & Havel, E. (2004). The important of getting started right: Further examination of the facility to community transition of formerly incarcerated youth. Journal of Special Education , 38 (2), 80 94. Burnham, K. P., & Anderson, D. R. (2002). Model selection and inference: A practical information theoretic approach (2nd ed.). New York: Springer. Burnham, K. P., & Anderson, D. R. (2004). Multimodel inference: Understanding AIC and BIC in model selection. Sociological Methods & Research, 33 , 261 304. Buros Center for Testing. (2007). I nitial r eport to the F lorida Department of Education: R ecommendations on FCAT for 2007 2008 .

PAGE 150

150 Bussing, R., Mason, D. M., Bell, L., Porter, P. & Garvan, C. (2010). Adolescent outcomes of childhood attention deficit/hyperactivity disorder in a diverse community sample. Journal of the American Academy of Child & Adolescent Psychiatry , 49 (6), 595 605. Caffrey, E., & Fuchs, D. (2007). Differences in performance between students with learning disabilities and mild mental retardation: Implications for categorical instruction. Learning Disabilities Research & Practice , 22 (2), 119 128. Canivez, G. L., Konold, T. R., Collins, J. M., & Wilson, G. (2009). Construct validity of the Wechsler Abbrev iated Scale of Intelligence and Wide Range Intelligence Test: Convergent and structural validity. School Psychology Quarterly, 24 (4), 252 265. Carlson, J. F., Romhild, A., McCormick, C., Chin, K, & Geisinger, K. F. et al., (2010). Evaluating reading test s for the state of Florida: Dynamic Indicators of Basic Early Literacy Skills, 6th Edition (DIBELS) and Florida Assessments for Instruction in Reading (FAIR) . Lincoln, NE: Buros Center for Testing. Carroll, J. B. (1993). Human cognitive abilities: A surv ey of factor analytic studies . New York: Cambridge University Press. Carroll, J. M., Bowyer Crane, Duff, F. J., Hulme, C., & Snowling, M. J. (2011). Developing language and literacy: Effective intervention in the early years . Hoboken, NJ: Wiley. Cavendish, W. (2013). Academic attainment during commitment and postrelease education related outcomes of juvenile justice youth with and without disabilities. Journal of Emotional and Behavioral Disorders, 22 (1), 41 52. Christle, C. A., & Yell, M. L. (2 008). Preventing youth incarceration through reading remediation: Issues and solutions. Reading & Writing Quarterly: Overcoming learning difficulties , 24 (2), 148 176. Coffey, O. D., & Gemignani, M. G. (1994). Effective practices in juvenile correctional education: A study of the literature and research, 1980 1992 . Washington, DC: U.S. Department of Justice. The National Office for Social Responsibility. Cohen, J. (1992). A Power Primer. Psychological Bulletin , 112 (1), 155 159. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. Cohen, M. A., & Piquero, A. R. (2009). New evidence on the monetary value of saving a high risk youth. Journal of Quantitat ive Criminology , 25 , 25 49. Cortiella, C. (2006). NCLB and IDEA: What parents of students with disabilities need to know and do. Minneapolis: University of Minnesota, National Center on Educational Outcomes.

PAGE 151

151 Coulter, G. (2004). Using one to one tutoring and proven reading strategies to improve reading performance with adjudicated youth. Journal of Correctional Education, 55 (4), 321 333. Coyne, M. D., & Harn, B. A. (2006). Promoting beginning reading success through meaningful assessment of early literacy skills. Psychology in the Schools , 43 , 33 43. Crawford, L., Tindal, G., & Stieber, S. (2001). Using oral reading rate to predict student performance on statewide achievement tests. Educational Assessmen t , 7 Darling collision of new standards and old inequalities. In D. Meier & G. Wood (Eds.), Many children left behind: How the No Child Left Behind Act is damaging our children and our schools (pp. 3 32). Boston, MA: Beacon Press. Deno, S. L., Reschley, A. L., Lembke, E. S., Magnusson, D., Callender, S. A., Windram, H., & Stachel, N. (2009). Developing a school wide progress monitoring system. Psychology in the Schools, 46 (1), 44 55. Denton, C. A., Barth, A. E., Fletcher, J. M., Wexler, J., Vaughn, S., & Cirino, P. T. et al. (2011). The relations among oral reading fluency and silent reading fluency and comprehension in middle school: Implications for identific ation and instruction of students with reading difficulties. Scientific Studies of Reading , 15 (2), 109 135. Dixon, M. (2010). Public Education Finances: 2010 . Washington, DC: U.S. Department of Commerce, U.S. Census Bureau. Drakeford, W. (2002). The im pact of an intensive program to increase the literacy skills of youth confined to juvenile corrections. Journal of Correctional Education , 53 (4), 139 144. Dunn, L. M., & Dunn, D. M. (2007). Peabody Picture Vocabulary Test, Fourth Edition . Minneapolis, MN: Pearson Assessments. Durkin, D. (2003). Teaching them to read (6th ed.). Boston, MA: Allyn & Bacon. Eldredge, J. L. (2005). Foundations of fluency: An exploration. Reading Psychology , 26 , 161 181. Elmund, A., Melin, L., von Knorring , A. L., Proos, L., Tuvemo, T. (2004). Cognitive and neuropsychological functioning in transnationally adopted juvenile delinquents. Acta Paediatrica, 93, 1507 1513. Evans, J. J., Floyd, R. G., McGrew, K. S., & Leforgee, M. H. (2001). The relations betw een measures of Cattell Horn Carroll (CHC) cognitive abilities and reading achievement during childhood and adolescence. School Psychology Review, 31 (2), 246 262.

PAGE 152

152 Fan, Xitao. (2001). Statistical significance and effect size in education research: Two si des of a coin. Journal of Educational Reseatch, 94 (5), 275 282. Feifer, S. G. (2010). Assessing and intervening with children who have reading disorders. In D. C. Miller (Ed.), Best practices in school neuropsychology: Guidelines for effective practice, assessment, and evidence based interventions (pp. 359 386). Hoboken, NJ: Wiley. Figlio, D., & Getzle, L. (2002). Accountability, ability and disability: Gaming the system? National Bureau of Economic Research working paper 9307. Flanagan, D. P., Ortiz, S. O., Alfonso, V. C., Dynda, A. A. (2006). Integration of response to intervention and norm referenced tests in learning disability identification: Learning from the Tower of Babel. Psychology in the Schools, 43 (7), 807 825. Florida Department of Education (2001). FCAT briefing book . Retrieved from http://fcat.fldoe.or g/pdf/fcataabb.pdf Florida Department of Education (2005). FCAT Handbook A resource for educators . Retrieved from http://fcat.fldoe.org/handbk/complete.pdf Florida Department of Education (2007). Assessment and Accountability Briefing Book: FCAT . Retrieved from http://fcat.fldoe.org/pdf/BriefingBook07web.pdf Florida Department of Education (2009). Florida Assessments for Instruction in Reading: Technical Manual, 2009 2010 Edition (Grade 3 1: Bureau of Exceptional Education and Student Services. Retrieved from http://fcrr.org/FAIR/%20312_Technical_Manual_ FINAL.pdf Florida Department of Education (2014). FCAT 2.0 and Florida EOC Assessments: Achievement L evels . Retrieved from http://fcat.fldoe.org/fcat2/pdf/achlevel.pdf Florida Department of Education (2011). Test design summary: 2011 operational assessments FCAT/FCAT 2.0/end of course . Retrieved from http://fcat.fldoe.org/pdf/1011designsu mmary.pdf Florida Department of Education (2012a). FCAT 2.0 Parent Information Sheet Developmental Scale Scores . Retrieved from http://fcat.fldoe.org/fcat2/pdf/pidss final.pdf Florida Department of Education (2012b). FCAT 2.0 Student Performance Results: Demographic Report . Retrieved from http://app1.fldoe.org/FCATDemographics/ Florida Office of Program Policy Analysis & Government Accountability . (2010). Youth Available data is insufficient to assess learning gains ( OPPAGA Report No. 10 07) . Retrieved from http://oppaga.state.fl.us/MonitorDocs/Reports/pdf/1007rpt.pdf

PAGE 153

153 Floyd, R. G., Keith, T. Z., Taub, G. E., & McGrew, K. S., (2007). Cattell Horn Carroll cognitive abilities and their effects on reading decoding skills: g has indirect effects, more specif ic abilities have direct effects. School Psychology Quarterly, 22 (2), 200 233. Foley, R. M. (2001). Academic characteristics of incarcerated youth and correctional education programs: A literature review. Journal of Emotional & behavioral Disorders , 9 , 2 48 260. Foorman, B. R., Kershaw, S., & Petscher, Y. (2013). Evaluating the screening accuracy of the Florida Assessments for Instruction in Reading (FAIR). (REL 2013 008). Washington, DC: U.S. Department of Education, Institute of Education Sciences, Nat ional Center for Education Evaluation and Regional Assistance, Regional Educational Laboratory Southeast. Retrieved from http://ies.ed.gov/ncee/edlabs/ Foorman, B. R., & Nixon, S. M. (2006). The influence of public policy on reading research and practice. Top Language Disorders , 26 (2), 157 171. Foorman, B. R., & Petscher, Y. (2010). The unique role of the broad screen in predicting FCAT reading comprehension. Tallahassee, FL: Florida Center for Reading Research. Retrieved from http://fcrr.org/fair/index.shtm Forsyth, C. J., Asmus, G., Stokes, B. R., & Forsyth, Y. A. (2010). Comparing performance test scores of juvenile delinquents with the general population of students. Deviant Behavior , 31 , 303 313. Fowlert, A. E., & Swainson, B. (2004). Relationships of naming skills to reading, memory, and receptive vocabulary: Evidence for imprecise phonological representations of words by poor readers. Annals of Dyslexia, 54 (2), 247 280. Fox, J. (2008). Applied regression models and generalized linear model (2nd ed.). Thousand Oaks, CA: SAGE Publications. Fox, J. & Wesberg, S. (2011). An R companion to appli ed regression (2nd ed.). Thousand Oaks, CA: SAGE Publications. Fritz Educational Foundations , 18 (3), 7 24. Fuchs, D., & Fuchs, L. S. (2006). Introduction to Response to Intervention: What why, and how valid is it? Reading Research Quarterly, 41 , 93 99. Gagnon, J. C. (2008). High stakes testing and students with disabilities: Challenges and Promises. Journal of Disability Policy Studies , 18 (3), 160 167. Gagnon, J. C. (2010). State level curricular, assessment, and accountability policies, practices, and philosophies for exclusionary school settings. Journal of Special Education, 43 (4), 206 219.

PAGE 154

154 Gagnon, J. C., & Barber, B. R. (2010). Characteristics of and services provided to youth in secure care facilities. Behavioral Disorders , 36 (1), 7 19). Gagnon, J. C., & Barber, B. R. (in press). Instructional practice guide for teaching reading and mathematics in juvenile correctional schools. Journal of Correctional Educat ion. Gagnon, J. C., Barber, B. R., & Van Loan, C., Leone, P. E. (2009). Juvenile correction schools: Characteristics and approaches to curriculum. Education and Treatment of Children, 32 (4), 673 696. Gagnon, J. C., Haydon, T., & Maccini, P. (2010 ). Juvenile correctional schools: Assessment and accountability policies and practices. Journal of Correctional Education, 61 (1), 23 45. Gagnon, J. C., Maccini, P., & Haydon, T. (2011). Assessment and accountability in public and private secondary day t reatment and residential schools for students with emotional and behavioral disorders. Journal of Special Education Leadership , 24 (2), 79 91. Gagnon, J. C., Maccini, P., & Mulcahy, C. (2014). Mathematics in secondary psychiatric schools: Curricular and assessment policies and practices. Exceptionality, 22 (1), 51 67. Gagnon, J. C., McLaughlin, M. J., Rhim, L. M., & Davis, G. A. (2002). Standards driven reform policies at the local level: Report on a survey of local special educat ion directors in large districts. Journal of Special Education Leadership, 15 (1), 3 9. Gagnon, J. C., Murphy, K. M., Steinberg, M. A., Gaddis, J., & Crockett, J. (2013). IDEA related professional development in juvenile corrections schools. Journal of S pecial Education Leadership , 26 (2), 93 105. Gagnon, J. C., & Richards, C. (2008). Making the right turn: A guide about improving transition outcomes of youth involved in the juvenile corrections system. Washington, DC: National Collaborative on Workforce and Disability for Youth, Institute for Educational Leadership. Gay, G. (2007). The rhetoric and reality of NCLB. Race Ethnicity and Education , 10 (3), 279 293. Goffreda, C. T. (2009). Preventive screening for early readers: Predictive validity of the Dynamic Indicators of Basic Early Literacy Skills (DIBELS). Psychology in the Schools, 46 (6), 539 332. Good, R. H., Simmons, D., & Kame'enui, E. (2001). The importance of decision making utility of a continuum of fluency based indicators of foundation al reading skills for third grade high stakes outcomes. Scientific Studies of Reading , 5

PAGE 155

155 Summary of decision rules for intensive, strategic, and benchmark instruct ional recommendations in kindergarten through third grade (Technical Report No. 11). Eugene, OR: University of Oregon. Goswami, U. (2007). Typical reading development and developmental dyslexia across languages. In D. Coch, G. Dawson, & K. W. Fischer (Ed s.), Human behavior, learning, and the developing brain (pp. 117 144). New York: Guilford. Gottfredson, L. S. (2002). Where and why g matters: Not a mystery. Human Performance , 15 (1/2), 25 46. Grigorenko, E. L. (2006). Learning disabilities in juvenil e offenders. Child and Adolescent Psychiatric Clinics of North America , 15 , 353 371. Guilfoyle, C. (2006) NCLB: Is there life beyond testing? Educational Leadership , 64 (3), 8 13. Guthrie, J. T. (2002). Preparing students for high stakes test taking in reading. In A. E. Farstrup & S. J. Samuels (Eds.), What research has to say about reading instruction (3rd ed.) (pp. 370 391). Newark, DE: International Reading Association. Hair. J. F. Jr., Anderson, R. E., Tatham, R. L., & Black, W. C. (1995). Multivariate data analysis (3rd ed.). New York: Macmillan. Hale, J. B., Kaufman, A., Naglieri, J. A., Kavale, K. A. (2006). Implementation of IDEA: Integrating response to intervention and cognitive assessment methods. Psychology in the Schools, 43 (7), 753 770. Harris, P. J., Baltodano, H. M., Bal, A., Jolivette, K., & Malcahy, C. (2009). Reading achievement of incarcerated youth in three regions. The Journal of Correctional Education , 60 (2), 120 145. Hayling, C. C., Cook, C., Gresham, F. M., State, T., & Kern, L. (2008). An analysis of the status and stability of the behaviors of students with emotional and behavioral difficulties. Journal of Behavioral Education , 17 , 24 42. Hodges, J., Giuliotti, N., & Porpotage II, F. M. (1994). Improving liter acy skills of juvenile detainees . NCJ Publication No. 150707. Washington, DC: U.S. Department of Justice, Office of Juvenile Justice and Delinquency Prevention. Holm, S. (1979a). A simple sequentially rejective multiple test procedure. Scandinavian Journa l of Statistics, 6, 65 70. Holm, S. (1979b). A stagewise directional test based on t statistics (Statistical Research Report No. 1979 3). Gothenburg, Sweden: Chalmers University of Technology, Institute of Mathematics.

PAGE 156

156 Horn, J. L. (1989). Measurement of intellectual capabilities: A review of theory. In K. S. McGrew, J. K. Werder, & R. W. Woodcock (Eds.), WJ R technical manual (pp. 197 245). Chicago, IL: Riverside. Hosp, M. K., & Fuchs, L. S. (2005). Using CBM as an indicator of decoding, word reading, and comprehension: Do the relations change with grade? School Psychology Review, 34 (1), 9 26. Houchins, D., Jolivette, K., Krezmien, M. & Baltodano, H. (2008). A multi state study examining reading instruction with incarcerated students. The Journal of Correctional Education, 59 (1), 65 85. House, A. E., House, B. J., & Campbell, M. B. (1981). Measures of interobserver agreement: Calculation formulas and distribution effects. Jour nal of Behavioral Assessment, 3, 37 57. Howe, K. B., & Shinn, M. M. (2002). Standard reading assessment passages (RAPs) for use in general outcome measurement: A manual describing development and technical features . Retrieved from http://teacher.scholastic.com/products/fluencyformulapdfs/OFATechMa nual.pdf Huck, S. W. (2012). Reading Statistics and research (6th ed.). Boston, MA: Allyn & Bacon. Hudson, R. F., Lane, H. B., & Pullen, P. C. (2005). Reading fluency assessment and instruction: What, why, and how? Reading Teacher, 58 (8), 702 714. (2007). Retrieved from http://eurekawebs.com/humlit/fast_facts.htm Hunley, S. A., Davies, S. C., & Miller, C. R. (2013) The relationship between curriculum based measures in oral reading fluency and high stakes tests f or seventh grade students. RMLE Online , 36 (5), 1 8. Hurvich, C. M., & Tsai, C. L. (1989). Model selection for extended quasi likelihood models in small samples. Biometrics , 51 (2), 1077 1084. Hurvich, C. M., & Tsai, C. L. (1995). Regression and time seri es model selection in small samples. Biometrika , 76 (2), 297 307. IBM Corp. Released 2012. IBM SPSS Statistics for Windows, Version 21.0. Armonk, NY: IBM Corp. Individuals with Disabilities Improvement Education Act, 34 C.F.R. pts. 300 and 301 (2006).

PAGE 157

157 Jackson, H. G., & Neel, R. S. (2006). Observing mathematics: Do students with EBD have access to standards based mathematics instruction. Education and Treatment of Children, 29 (4), 593 614. Jacob, B. (2005). Accountability, incentives and behavior: The impact of high stakes testing in the Chicago public schools. Journal of Public Economics , 89 , 761 796. Jensen, A. R. (1969). How much can we boost IQ and scholastic achievement? Harvard Educ ational Review, 39 (1), 1 123. Jensen, A. (2002). Psychometric g: Definition and substantiation. In R. J. Sternberg & E. L. Grigorenko (Eds.), The general factor of intelligence: How general is it? (pp. 39 55). Mahwah, NJ: Erlbaum. Jenson, J. M., Potter, C. C., & Howard, M. O. (2001). American juvenile justice: Recent trends and issues in youth offending. Social Policy & Administration, 35 (1), 48 68. Jiron, C. (2010). Assessing and intervening with children with externalizing disorders. In D. C. Miller (Ed.), Best practices in school neuropsychology: Guidelines for effective practice, assessment, and evidence based interventions (pp. 359 386). Hoboken, NJ: Wiley. Joyce, D., Tropf, L., & Gaddis, J. (2013). Comparison of the WRAT 4 and FAIR Reading Sc reening with ED students at risk for high stakes test failure. Manuscript submitted for publication. Katsiyannis, A. Ryan, J. B., Zhang, D., Spann, A. (2008). Juvenile delinquency and recidivism: The impact of academic achievement. Reading & Writing Quar terly, 24 (2), 177 196. Katsiyannis, A., Zhang, D., Ryan, J. B., & Jones, J. (2007). High stakes testing and students with disabilities: Challenges and promises. Journal of Disability Policy Studies , 18 (3), 160 167. Kim, Y., Wagner, R. K., & Lopez, D. (2012). Developmental relations between reading fluency and reading comprehension: A longitudinal study from Grade 1 to Grade 2. Journal of Experimental Child Psychology, 113 (1), 93 111. Kleinbaum, D. G., Kupper, L. L., Nizam, A., & Muller, K. E. (2007). Applied regression analysis and other multivariate methods (4th ed.). Belmont, CA: Thomson. Krezmien, M. P., Mulcahy, C. A., & Leone, P. E. (2008). Detained and committed youth: Examining differences in achievement, mental health needs, and sp ecial education status. Education and Treatment of Children , 31 (4), 445 464. Krezmien, M. P., Mulcahy, C. A., Travers, J., Wilson, M., & Wells, C. (2013). Reading performance of incarcerated youth: Understanding and responding to a unique population of re aders. Journal of Special Education Leadership , 26 (2), 71 81.

PAGE 158

158 Kroll, L., Rothwell, J., Bradley, D., Shah, P, et al. (2002). Mental health needs of boys in secure care for serious or persistent offending: A prospective, longitudinal study. The Lancet , 359 , 1975 1979. Learning Point Associates. (2004). A closer look at the five essential components of effective reading instruction: A review of scientifically based reading research for teachers . Naperville, IL: Author. Retrieved from http://Learningpt.org/pdfs/literacy/components .pdf Leo ne, P. E., & Cutting, C. A. (2004). Appropriate Education, juvenile corrections, and No Child Left Behind. Behavioral Disorders, 29 (3), 260 265. Leone, P. E., Krezmien, M., Mason, L., Meisel, S. M. (2005). Organizing and delivering empirically based lit eracy instruction to incarcerated youth. Exceptionality , 13 (2), 89 102. Leone, P. E., Meisel, S. M., Drakeford, W. (2002). Special education programs for youth with disabilities in juvenile corrections. Journal of Correctional Education, 53 (2), 46 50. Liston, D., Whitcomb, J., & Borko, H. (2007). NCLB and scientifically based research: Opportunities lost and found. Journal of Teacher Education , 58 (2), 99 107. Lynam, D. R. (1996). Early identification of chronic offenders: Who is the fledgling psychopath? Psychological Bulletin , 120 (2), 209 234. Lynam, D. R., Moffitt, T., & Stouthamer Loeber. (1993). Explaining the relationship between IQ and delinquency: Cla ss, race, test motivation, school failure, or self control? Journal of Abnormal Psychology , 102 (2), 187 196. Lyon. G. R. (1998). Overview of reading and literacy initiatives. Report presented before the committee on Labor and Human Resources, U.S. Senate , Washington, DC. Retrieved from http://cdl.org/resource library/pdf/lyon_testimonies.pdf Maccini, P., Gagnon, J. C., & Mason Williams, L. (2012). Mathematics and juvenile correction s: Curricular and assessment policies and practices. Learning Disabilities, 18 (1), 11 21. Malmgren, K. W., & Leone, P. E. (2000). Effects of a shirt term auxiliary reading program on the reading skills of incarcerated youth. Education & Treatment of Chi ldren, 23 (3), 239 247. Marcotte, A. M., & Hintze, J. M. (2009). Incremental and predictive validity of formative assessment methods of reading comprehension. Journal of School Psychology , 47 , 315 335.

PAGE 159

159 Mather, N., & Woodcock, R. W. (2001). anual . Woodcock Johnson III . Itasca, IL: Riverside Publishing. McGlinchey, M. T., & Hixson, M. D. (2004). Using curriculum based measurement to predict performance on state assessments in reading. School Psychology Review , 33 McGraw, K. O., & Wong, S. P. (1996). Forming inferences about some intraclass coefficients. Psychological Methods, 1 (1), 30 46. McGrew, K. S., Schrank, F. A., & Woodcock, R. W. (2007). Technical manual: Normative update . Woodcock Johnson III . Itasca, IL: Riverside Publ ishing. McGrew, K. S., & Wendling, B. J. (2010). Cattell Horn Carroll cognitive achievement relations: What we have learned from the past 20 years of research. Psychology in the Schools, 47 (7), 651 675. McHale, B. G., Obrzut, J. E., & Sabers, D. L. (20 03). Relationship of cognitive functioning and aggressive behavior with emotionally disabled and specific learning disabled students. Journal of Developmental and Physical Disabilities , 15 (2), 123 140. Merino, K., & Ohmstede Beckman, T. (2010). Using re ading curriculum based measurements as predictors for the Measure Academic Progress (MAP) standardized test in Nebraska. International Journal of Psychology , 6 , 85 98. Moffitt, T. E., & Silva, P. A. (1988). IQ and delinquency: A test of the differential detection hypothesis. Journal of Abnormal Psychology, 97 (3), 330 333. Myner, J., Santman, J., Cappelletty, G., & Perlmutter, B. (1998). Variables associated with recidivism among juvenile offenders. International Journal of Offender Therapy and Comparative Criminology, 42, 65 80. National Computer Systems Pearson, Inc. (2012). AIMSweb progress monitoring and improvement system . Retrieved from http://www.aimsweb.com National Reading Panel. (2000). Report of the National Reading Panel. Teaching children to read: An evidence based assessment of the scientific research literature on reading and its implications for reading instruction (NIH Publication No. 00 4769). Wash ington, DC: National Institute of Child Health and Human Development. Newton, J. H., & McGrew, K. S. (2010). Introduction to the special issue: Current research in Cattell Horn Carroll based assessment. Psychology in the Schools, 47 (7), 621 634.

PAGE 160

160 Nix, T. W., & Barnette, J. J. (1998). The data analysis dilemma: Ban or abandon. A review of null hypothesis significance testing. Research in the Schools, 5 (2), 3 14. No Child Left Behind Act of 2001, 20 U.S.C. § 6319 (2008). ., Haussmann, A., & Kloos, H. (2013). Using complexity metrics to assess silent reading fluency: A cross sectional study comparing oral and silent reading. Scientific Studies of Reading, (ahead of print), 1 20. rules of thumb for variance inflation factors. Quality & Quantity, 41, 673 690 Office of Juvenile Justice and Delinquent Prevention. (2006). Statistical briefing book . Retrieved from http://ojjdp.gov/ojstatbb/crime/JAR_Display.asp?ID=qa05260 Petscher, Y., & Foorman, B. R. (2011). Summary of the predictive relationship between the FAIR and the FCAT in Grades 3 10: 2010 2011. Tallahassee, FL: Florida Center for Reading Research. Retrieved from http://fcrr.org/fair/index.shtm Petscher, Y., & Kim, Y. (2011). The utility and accuracy of oral reading fluency score ty pes in predicting reading comprehension. Journal of School Psychology , 49, 107 129. Petteruti, A., Walsh, N., & Velazquez, T. (2009). The costs of confinement: Why good juvenile justice policies make good fiscal sense . Justice Policy Institute. Retrieved from http://justi cepolicy.org/images/upload/09_05_REP_CostofCofinement _JJ_PS.pdf Quinn, M. M., Rutherford, R. B., & Leone, P. E. (2001). Students with disabilities in correctional facilities. Arlingto n, Va.: Council for Exceptional Children, The ERIC Clearinghouse on Disabilities and Gifted Education. Quinn, M. M., Rutherford, R. B., Leone, P. E., Osher, D. M., & Poirier, J. M. (2005). Youth with disabilities in juvenile corrections: A national survey. Exceptional Children , 71 (3), 339 345. Raine, A., Moffitt, T. E., Caspi, A., Loeber, R., Stouthamer Loeber, et al. (2005). Neurocognitive impairments in boys on the life course persistent antis ocial path. Journal of Abnormal Psychology , 114 (1), 38 49. Renaissance Learning, Inc. (2000). Comparison of the STAR Reading Computer Adaptive Reading Test and the Scholastic Reading Inventory Test . Wisconsin Rapids, WI: Author. R Core Team. (2012). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from http://www.r project.org/

PAGE 161

161 Riddle Buly, M., & Valencia, S. W. (2002). Below the bar: P rofiles of students who fail state reading assessments. Educational Evaluation and Policy Analysis , 24 (3), 219 239. Rincker, J. L. (1990). Academic and intellectual characteristics of adolescent juvenile offenders. Journal of Correctional Education, 41 (3), 124 131. Roberts, G., Torgesen, J. K., Boardman, A., & Scammacca, N. (2008). Evidence based strategies for reading instruction of older students with learning disabilities. Learning Disabilities Research & Practice , 23 (2), 63 69. Rock, E. E., Fessler, M. A., & Church, R. P. (1997). The concomitance of learning disabilities and emotional/behavioral disorders: A conceptual model. Journal of Learning Disabilities , 30 (3), 245 263. Roehrig, A., Petscher, Y., Nettles, S., Hudson, R., & Torgesen, J. (2008). Accuracy of the DIBELS oral reading fluency Measure for predicting third grade reading comprehension outcomes. Journal of School Psychology , 46 , 343 366. Rohde, T. E., & Thompson, L. (2007). Predicting academic achievement with cogn itive ability. Intelligence , 35 (1), 83 92. Rubin, D. B. (1996). Multiple imputation after 18+ years. American Statistical Association , 91 (434), 473 489. Saborine, E. J., Cullinan, D., Osborne, S. S., & Brock, L. B. (2005). Intellectual, academic, and behavioral functioning of students with high incidence disabilities: A cross cultural meta analysis. Exceptional Children , 72 (1), 47 63. SAS Institute Inc. (2012). SAS© 9.3. Cary, NC: SAS Institute Inc. Schafer, J. L., & Graham, J. W. (2002). Missing dat a: Our view of the state of the art. Psychological Methods, 7 (2), 147 177) Schafer, J. L., & Olsen, M. K. (1998). Multiple imputation for multivariate missing data Multivariate Behavioral Research , 33 (4), 545 571. Schatschneider, C., Buck, J., Torgesen, J. K.,Wagner, R. K., Hassler, L., Hecht, S., et al. (2004). A multivariate study of individual differences in performance on the reading portion of the Florida Comprehensive Assessment Test: A brief report (FCRR R eport No. 5). Tallahassee, FL: Florida Center for Reading Research Retrieved from http://fcrr.org/tech nicalreports/multi_variate_study_december2004.pdf Schilling, S. G., Carlisle, J. F., Scott, S., & Zeng, J. (2007). Are fluency measures accurate predictors of reading achievement? Elementary School Journal , 107

PAGE 162

162 Scholastic, Inc. (2001a). Scholastic Reading Inventory (SRI) . New York, NY: Author. Scholastic, Inc. (2001b). Scholastic Reading Inventory: Technical guide . New York, NY: Author. Scholastic, Inc. (2006). software and teacher support . New York, NY: Author. Scholastic, Inc. (2008). The Lexile framework for reading: A system for measuring reader ability and text complexity . New York, NY: Author. Shufelt, J. L., & Cocozza, J. J. (2006). Youth with mental health disorders in the juvenile justice system: Resu lts from a multi state prevalence study . Delmar, NY: National Center for Mental Health and Juvenile Justice. Shaffer, J. P. (1980). Control directional errors with stagewise multiple test procedures. Annals of Statistics, 8, 1342 1347. Shaffer, J. P. (1986). Modified sequentially rejective multiple test procedures. Journal of the American Statical Association, 81, 826 831. Shapiro, E. S., Keller, M. A., Edwards, L., Lutz, G., & Hintze, J. M. (2006). General outcome measures and perfor mance on state assessment and standardized tests: Reading and math performance in Pennsylvania. Journal of Psychoeducational Assessment , 42 Shapiro, E. S., Solari, E., & Petscher, Y. (2008). Use of a measure of reading comprehension to enhance prediction on the state high stakes assessment. Learning and Individual Differences , 18 , 316 328. Shaywitz, S. E. & Shaywitz, B. A. (2005). Dyslexia: Specific reading disability. Biological Psychiatry , 57 , 1301 1309. Shelley Rohling, J. (2007). Reading disability in adjudicated youth: Prevalence rates, current models, traditional and innovative treatments. Aggression and Violent Behavior , 12 , 376 392. Shinn, M. M., & Shinn, M. R. (2002). AIMSweb Training Workbook: Administration and scoring of Reading Curriculum Based Measurement (R CBM) for use in general outcome measurement . Eden Prairie, MN: Edformation, Inc. Shippen, M. E., Morton, R. C., Flynt, S. W., Houchins, D. E., & Smitherman, T. (2012). Efficacy of a computer based program on acquisition of reading skills of incarcerated youth. Remedial and Special Education, 33 (1), 14 22.

PAGE 163

163 Shufelt, J. L., & Cocozza, J. J. (2006). Youth with mental health disorders in the juv enile justice system: Results from a multi state prevalence study . Delmar, NY: National Center for Mental Health and Juvenile Justice. Sickmund, M., Sladky, T. J., & Kang, W. (2008). Census of juveniles in residential placement databook . Washington, DC: U.S. Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention. Silberglitt, B., Burns, M. K., Madyun, N. H., & Lail, K. E. (2006). Relationship of reading fluency assessment data with state accountabili ty test scores: A longitudinal comparison of grade levels. Psychology in the Schools , 43 , 527 535. Snow, C.E., Porche, M.V., Tabors, P.O., & Harris, S.R. (2007). Is literacy enough? Baltimore, MD: Brookes Publishing Co. Snowling, M. J., Adams, J. W., Bow yer Crane, C., & Tobin, V. (2000). Levels of literacy among juvenile offender: The incidence of specific reading difficulties. Criminal Behavior and Mental Health, 10 (4), 229 241. Snyder, H. N., & Sickmund, M. (2006). Juvenile offenders and victims: 200 6 national report . Washington, DC: U.S. Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention. Snyder, T.D., & Dillow, S.A. (2013). Digest of education s tatistics 2012 (NCES 2014 015). National Center for Education Statistics, Institute of Education Sciences, U.S. Department of Education. Washington, DC. Sousa, D. A. (2011). How the brain learns (4th ed.). Thousand Oaks, CA: Corwin Press Stage, S., & Jacobsen, M. (2001). Predicting student success on a state mandated performance based assessments using oral reading fluency. School Psychology Review , 30 Stanovich, K. E. (1986). Matthew effects in reading: Some consequences of individual differences in the acquisition of literacy. Reading Research Quarterly , 21 , 360 496. State of Florida Department of State. (2007). FCAT reading lessons learned: 2001 2005 data analyses and instructional implications . Retrieved from http://education.ucf.edu/mirc /Research/FCAT%20Reading%20Lessons%20Learned%20 %202001 2005%20Data%2 0Analyses%20and%20Instructional%20Implications.pdf Sunde rman, G. L., Kim, J. S., & Orfield, G. (2005). NCLB meets school realities: Lessons from the field. Thousand Oaks, CA: Corwin Press. Tannenbaum, K. R., Torgesen, J. K., & Wagner, R. K. (2006). Relationships between word knowledge and reading comprehensi on in third grade children. Scientific Studies of Reading , 10 (4), 381 398.

PAGE 164

164 Thompson, B. (1999). If statistical significance tests are broken/misused, what practices should replace them? Theory & Psychology, 9 (2), 165 181. Thompson, B. (1998). Statistical significance and effect size reporting: Portrait of a possible future. Research in the Schools, 5 (2), 33 38. Thurlow, M. L., & Wiley, H. I. (2006). A baseline perspective on disability subgroup reporting. The Journal of Special Education , 39 (4), 246 254. Tighe, E. L., & Schatschneider. (2013). A dominance analysis approach to determining predictor importance in third, seventh, and tenth grade reading comprehension skills. Reading and Writing , 27 (1), 101 127. Tolar, T. D., Barth, A. E., Francis, D. J., Fletcher, J. M., Stuebing, K. K., & Vaughn, S. (2011). Psychometric properties of maze tasks in middle school students. Assessment for Effective Intervention , 37 (3), 131 146. Torgesen, J. K. , Nettles, S., Howar d, P., & Winterbottom, R. (2003) . Brief report of a study to investigate the relationship between several brief measures of reading fluency and performance on the Florida Comprehensive Assessment Test Reading in 4th, 6th, 8th, and 10th grades ( FCRR Report No. 6) . Tallahassee, FL: Florida Center for Reading Research. Retrieved from http://fcrr.org/TechnicalReports/Progress_monitoring_eport. pdf Torgesen, J. K., Wagner, R. K., & Rashotte, C. A. (1999). Test of Word Reading Efficiency (TOWRE), Austin, TX: PRO ED. Turkheimer, F. E., Hinz, R., & Cunningham, V. J. (2003). On the undecidability among kinetic models: From model selection to model averaging. Journal of Cerebral Blo od Flow & Metabolism, 23, 490 498. Unruh, D. K., Gau, J. M., & Waintrup, M. G. (2009). An exploration of factors reducing recidivisim rates of formerly incarcerated youth in disabilities participating in a re entry intervention. Journal of Child & Family Studies, 18 (3), 284 293. U.S. Department of Education. (2006). Twenty seventh annual report to Congress on the implementation of the Individuals with Disabilities Education Act . Washington, DC: U.S. Government Printing Office. U.S. Department of Educat ion. (2007). Standards and assessments Peer review guidance: Information and examples for meeting requirements of the No Child Left Behind Act of 2001 . Washington, DC: U.S. Government Printing Office.

PAGE 165

165 U.S. Department of Education. (2011). Thirtieth annual report to Congress on the implementation of the Indviduals with Disabilities Education Act . Washington, DC: U.S. Government Printing Office. U.S. Department of Education , Institute of Education Sciences, National Center for Education, Evaluation and Regional Assistance, What Works Clearing House . ( 2014). Retrieved from http://ies.ed.gov/ncee/wwc Vacca, J. S. (2008). Crime can be prevented if schools teach juvenile offenders to read. Childr en and Youth Services Review , 30 , 1055 1062. van Buuren, S., & Groothuis Oudshoorn, K. (2011). Mice: Multivariate imputation by chained equations in R. Journal of Statistical Software, 45 (5), 1 67. van Buuren, S., & Groothuis Oudshoorn, K. (2012). M ic e 2.17. Retrieved from http://cran.r project.org/web/packages/mice/index.html van der Heijden, G. J., Donders, A. R., Stijnen, T. & Moons, K. G. (2006). Imputing missing values is su perior to complete case analysis and the missing indicator method in multivariable diagnostic research: A clinical example. Journal of Clinical Epidemiology, 59 , 1102 1109. Vermeiren, R., Jespers, I, & Moffitt, T. (2006). Mental health problems in juvenil e justice populations. Child and Adolescent Psychiatry Clinics of North America , 15 , 333 351. multiply imputed data. Sociological Methodology, 37 , 83 117. Wagenma kers, E. J., & Farrell, S. (2004). AIC model selection using Akaike weights. Psychonomic B ulletin & Review, 11, 192 196. Watkins, M. W., Lei, P., & Canivez, G. L. (2007). Psychometric intelligence and achievement: A cross lagged panel analysis. Intelligence , 35 (1), 59 68. Weschler, D. (1999). Weschler Abbreviated Scale of Intelligence . New York, NY: The Psychological Corporation. White, J. L., Moffitt, T. E., Caspi, A., Bartusch, D. J., & Needles, D. J., et al. (1994). Measuring impulsivity a nd examining its relationship to delinquency. Journal of Abnormal Psychology , 2 , 192 205. Wiley, H. I., & Deno, S. (2005). Oral reading and maze measures as predictors of success for English learners on a state standards assessment. Remedial and Special Education , 26 (4), 207 214.

PAGE 166

166 Wilkerson, K. L., Gagnon, J. C., Mason Williams, L. & Lane, H. B. (2012). Reading instruction for students with high incidence disabilities in juvenile corrections. Preventing School Failure , 56 (4), 219 231. Wood, D. E. (2006). Modeling the relationship between oral reading fluency and performance on a statewide reading test. Educational Assessment , 11 Wood, G. (2004). Introduction. In D. Meier & G. Wood (Eds.), Many children left behind: How the No Child Left Behind Act is damaging our children and our schools (pp. vii). Boston, MA: Beacon Press. Woodcock, R. W., McGrew, K. S., & Mather, N. (2007). Woodcock Johnson III Tests of Achievement Normative Update . Itasca, IL: Riverside Publishing. Wo odcock, R. W., McGrew, K. S., & Mather, N. (2007). Woodcock Johnson III Tests of Cognitive Abilities Normative Update. Itasca, IL: Riverside. Yamashita, T., Yamashita, K., Kamimura, R. (2007). A stepwise AIC method for variable selection in linear regr ession. Communication in Statistics Theory and Methods, 36 (13), 2395 2403. Yell, M. L., Katsiyannas, A., & Shiner, J. G. (2006). The No Child Left Behind Act, adequate yearly progress, and students with disabilities. Teaching Exceptional Children , 38 ( 4), 32 39. Yeo, S. (2009). Predicting performance on state achievement tests using curriculum based measurement in reading: A multilevel meta analysis. Remedial and Special Education , 31 (6), 412 422. Zhang, D., Hsu, H., Katsiyannis, A., Barrett, D. E., & Ju, S. (2011). Adolescents with disabilities in the juvenile justice system: Patterns for recidivism. Exceptional Children , 77 (3), 283 298. Zhang, S., Liao, J., & Zhu, X. (2008). A SAS macro for single imputation. Pap. SP10. Merck & Co., Upper Gwynedd, PA. Retrieved from http://lexjansen.com/pharmasug/2008/sp/sp10 .pdf

PAGE 167

167 BIOGRAPHICAL SKETCH Justin Grant Gaddis was born in Louisville, Kentucky. He received his Bachelor of Arts degree i n education from the University of Kentucky in 2008, his Master of Education degree in school psychology from the University of Florida in 2011 , and his Doctor of Philosophy degree in school psychology from the University of Florid a in 2014 . During his time at the University of Florida he was awarded the Norman F. Nelson Fellowship (2013), the Fien Dissertation Award (2012), the College of Education Alumni Endowment Scholarship (2012), the Office of Educational Research Travel Award (2011), the SESPECS Research Travel Award (2011), and the Phi Delta Theta Graduate Academic Scholarship (2008 2011). In addition to serving as both a Research and Graduate Assistant on research projects while at the University of Florida, Justin also held positions within the departments of Pediatric Endocrinology, Psychiatry, Psychology, and Human Growth and Organizational Studies. Justin completed an APA/APPIC accredited r in Erie, PA in 2014 . His research interests include classification practices and assessment instruments, emotional/behavioral disorders, and pediatric neuro psychology.