FaithSchoolsand Pupils’ Progress through Primary Education
By
TilayeYeshanew
IanSchagen
NationalFoundationforEducationalResearch
TheMere,UptonPark,Slough,SL12DQ.
SuzanneEvans
SchoolofEconomics,MathematicsandStatistics, Birkbeck,
UniversityofLondon,MaletStreet,LondonWC1E7HX
Paper presented at the British Educational Research Association Annual Conference, Institute of Education, University of London, 5-8 September 2007
Abstract
The impact offaithschools on the performanceandprogress of their pupils has been studied, using data from the National Pupil Database (NPD). The value-added analysiswascarriedoutusing multilevel modelling, controlling for prior attainment as well as a range of backgroundvariablesincluding ethnicity,sex, eligibility for free school meals (FSM), alternativemeasuresofdeprivation based on census information, special educational needs (SEN)and English as an additional language (EAL).
Theanalysisconfirmedthatallfaithschools, in particular RomanCatholicandChurchofEngland schools, madeslightly more progress with their pupils thannon–faithschools. It alsoshowedthatpupilswith SENattendingfaithschoolsperformedbetterinkeystage2thanpupilswith SENinnon-faithschools.
Keywords:faithschool,MultilevelModelling,Fixedeffect,Randomeffect
Introduction
Faithschoolstendtobepopularwithparents,andobtaingoodresultsinnationalleaguetables.Questionsaresometimesaskedastowhetherthisisduetotheirstatusasfaithschools,ortootherfactors.Inparticular,ithasbeensuggestedthattheirsuccessisduetothefactthattheyhavea'better'intakethanotherschools.Inordertoexplorethisissue,itwasnecessarytoundertakeavalue-addedanalysiswhichtookpriorattainmentandotherkeyfactorsintoaccount,suchasEnglishfluency,ethnicityandSEN.InEnglandthereisnotmuchresearchaboutfaithschools:Schagenetal.(2002)showthatpupilsinfaithsecondaryschoolsprogressslightlyfasterinEnglish(butnotinmathsandscience),andalsoseemtopassmoresubjectsoverallintheirage-16exams.Bentonetal.(2003)reportthatfaithsecondaryschoolsareassociatedwithfastergradeprogressionbetweenage11and13,andage13and16;yetthisisconfinedtoschoolsaffiliatedwithJewishandOtherChristiandenominations(i.e.noncatholicandchurchofEngland).Thisresearchcomprisedonlyprimaryanalysisbecausethereislittlepublishedevidenceoftheirspecificimpact.
Thedatausedinthisstudywascollectedandcollatedduring2000to2004fromage7wheretheywereYear2toage11wheretheywereYear6.Thedatasetconsistsofavarietyofbackgroundvariablesaswellasoutcomevariables,includingkeystage2averagescore,keystage2English,keystage2mathematics,andkeystage2science.Thestudyaimedtoidentifywhichfactors,includingfaithschoolandbackgroundfactors,affectperformance.
Themultilevelmodelwasfittedatthreelevels:LEA,schoolandpupil.Overandabovethe‘fixed’partofthemodel,whichconsistsoftheregressionagainstthebackgroundfactors,therewasarandompartofthemodelwhichrelatestovariationsinoutcomewhicharespecifictoparticularpupils,schoolsorLEAs. Therewasarandomvarianceateachlevel:i.e.weassumedthatpupilsdifferintheirKS2outcomes,overandabovewhatmightbepredicted;schoolsdifferedbetweenthemselvesinaverageoutcomes,asdidLEAs. Inaddition,therelationshipbetweenkeystage1priorattainmentandkeystage2outcomeswasassumedtovaryfromschooltoschool.Themodelwasfittedtodatafrom548743pupilsin15285schoolsin148LEAs.
Methodology
ThemainthrustofthisstudywasintheformofstatisticalanalysisofthemostcurrentNationalPupilDataset(2004). Thisincludedpupil-levelinformationregardingprogress(KeyStage1(2000)toKeyStage2(2004))aswellasimportantbackgrounddataonaspectssuchasfaithschoolasawhole,individualreligiousgroup,andothercontextualinformation.SchoolidentifiersallowedustolinkthisinformationtotheDepartmentforEducationandSkills(DfES)schooldatabasetoprovideschoolcontextualdata. TheanalysiswascarriedoutusingAnalysisofvarianceandmultilevelmodelling(Goldstein,1995;Hecketal,2000),whichiswidelyrecognisedasoneofthemostappropriatemethodsforlongitudinaldataofthistype,withresultsforpupilsnestedwithinschoolsandLocalEducationAuthorities(LEAs).
Preliminarydataexplorationusedanalysisofvarianceisanappropriatestatisticaltechniqueforlookingattherelationshipbetweenanumericalvariableandacategoricalvariable.InthiscaseweareinterestedinthemeanvalueofKeyStage2(KS2)attainmentbetweendifferentgroupsofpupils,definedbyacategoricalvariable(pupilattendingfaithschoolandpupilattendingnon-faithschool).TheANOVAstatisticsinvestigatedamongdifferentgroupofschools;1)iftherewasanydifferenceinoutcomesbetweenthoseattendingfaithschoolsandthoseattendingnon-faithschools,and2)Iftherewasanydifferencebetweentypesoffaithschool(RomanCatholic,ChurchofEngland,Jewishschool,combinedfaith,otherandnon-faithschools).
Multilevelmodellingisadevelopmentof‘regressionanalysis’,andisanimportanttechniqueineducationalresearch,aswellasothersocialsciences. Multilevelmodellingallowsustotakeaccountofthecomplexandhierarchicalstructureofdata. ThistechniquetakesintoconsiderationthevariationsbetweenLEAs,schoolsandpupils,andproducespredictionsandestimatesthesizeofthesevariationsateachlevelintermsof‘randomvariances’(thefluctuationorvaryingofmeasuresduetochance).
Withinthecomplexstructureofthedatathereisasinglemeasureofinterest;inthiscaseitwasoneofKeyStage2average,KeyStage2EnglishScoreorKeyStage2MathsScore.ThesescoresarebasedonthestandardconversionofthenationalcurriculumlevelsawardedattheendofKeyStage2toequivalent‘pointsscores’,bymultiplyingby6andadding3(
Appendix1containsfulldetailsofallthevariablesusedinsettingupthemultilevelmodels.Thesevariablescanbedividedintofourmaingroups:
- Outcomemeasuresattheendofthekeystage;
- Priorattainmentmeasuresatthestartofthekeystage;
- Pupilbackgroundinformation;
- Schoolbackgroundandcontextualinformation.
PriorattainmentwasmeasuredintermsofattainmentattheendofKeyStage1(KS1).Previousresearch(Spielhoferetal,2002)hasshownthepossibilityofnon-linearrelationshipsbetweentheoutcomevariablesandtwobackgroundvariables(priorattainmentandschool-levelpercentageeligibleforfreeschoolmeals(FSM)).Forthisreasonquadratictermsforthesefactorswereincludedinthemodels.Onefinaloptionof“Interactionterms”wasalsoincludedinthemodels.Thesetermsallowedforthepossibilitythatcertainfactorsmighthaveanimpactwhichdependedonwhetherthepupilattendingfaithschoolornot.ForexamplethestrengthoftherelationshipbetweenSENandtheKS2outcomevariablesmightbedifferentforpupilwithSENattendingfaithschoolandpupilwithSENattendingnonfaithsschools.
Theanalysiswasthereforeruninfivedistinctstages,asfollows:
- Withnobackgroundvariables(‘baseline’);
- Includingpriorattainmentatthestartofthekeystage;
- Includingotherpupil-levelbackgroundvariables;
- Includingallschool-andpupil-levelvariables.
- Includingallinteractionvariables.
The models were also run assuming that the relationship between prior attainment and outcome was random at the school and LEA levels – in other words the strength of this relationship varied from school to school within each LEA as well as from LEA to LEA.
BackgroundData
Table1showstheoveralldifferencesbetweenthosepupilswhoareinfaithschoolandthosewhoarenotinfaithschoolsforkeystage12000tokeystage22004.Therewere5604faithschoolswithinthemaintainedprimarysector;theyaccountedformorethanathirdofprimaryschools.Table1breaksdownfaithschoolsintovariouscategories.Aswemightexpect,thetwolargestcategoriesareRomanCatholic(1624schools)andChurchofEngland(3855Schools).Twenty-threeJewishschoolsareidentifiedandtreatedhereasaseparategroup.The‘CombinedChristian’categoryincludesjointRomanCatholicandChurchofEnglandschoolsaswellasschoolsbelongingtoothercombinedChristiantraditions.OtherfaithincludesSikh,Muslim,SeventhdayAdventistchurchandothers.Weacknowledgethatthisisnotanentirelysatisfactorygroupingbutitisnotfeasibletofurther subdivide a categorywhichisalreadyverysmall.
Table 1Pupilsandschoolswithdifferentreligiousaffiliations
ReligiousAffiliation / Number
ofschools / Percentage
ofschools / Number
Ofpupils / Percentage
ofpupils
Non-religious / 9681 / 63.3 / 389365 / 70.9
RomanCatholic / 1624 / 10.6 / 52474 / 9.6
ChurchofEngland / 3855 / 25.2 / 103757 / 18.9
Combinefaith / 40 / 0.3 / 1371 / 0.2
Otherfaith / 62 / 0.4 / 898 / 0.2
Jewish / 23 / 0.2 / 941 / 0.2
Total / 15285 / 100 / 548743 / 100
Table 2belowshows thepercentageofSEN, EAL and FSM pupils indifferent school types.The highest proportionofSENpupil(schoolactionplus)attending afaithschooltype is 18%, compared with 19% for non-faith schools . The highest percentageofpupils eligibleforfreeschoolsmeals in faith schools is forRomanCatholic and other schools(15%), while thelowest percentage is forJewish schools (4%); for non-faith schools it is 19%. ThenumberofEALpupilattending faithschools ranges from 13% (other) down to 1% (combined Christian) – for non-faith schools it is 10%.
Table 2: PercentageofPupilswithSEN,EALandFSMbydifferentreligiousaffiliations
Category / Church of England / Roman Catholic / Combined Christian / Other / Jewish / Non-faithEligible for FSM / 11% / 15% / 11% / 15% / 4% / 19%
SEN (School action plus) / 18% / 17% / 14% / 17% / 18% / 19%
SEN (statement) / 2% / 2% / 3% / 3% / 2% / 4%
EAL / 5% / 6% / 1% / 13% / 7% / 10%
In order to compare the social backgrounds of different school types more precisely, it is important to use measures of deprivation other than eligibility for free school meals. The latter is an indicator of high deprivation, but does not give any gradation across the social context scale. Alternative measures, based on linking pupil postcodes to census data, have the potential to give a more finely-graded indication of social background. 10 census variables (listed in Table 3 below) were linked to pupil postcodes for the cohort being studied, and principal factor analysis with Varimax rotation was used to define a suitable subset of factors from this data.
Table 3shows the outcomes of the principal factor analysis, leading to the derivation of two main factors. Loadings greater than 0.3 in absolute value are shown and these seem to indicate how these factors could be interpreted:
- Overall measure ofdeprivation ;
- High occupancy area with high non-white ethnicity.
Factor scores for each pupil based on this decomposition of census variables were included in the modelling.
Table 3:Principal Factor Loadings for Census Variables
Census Variables / Factor1 / 2
%ofpeopleaged16-74withnoqualificationsinOA / .907
%ofhouseholdsinOAthatarerentedfromLA/HAorregisteredsociallandlord / .751
%ofpeopleaged16-74inroutineoccupationsinOA / .747
%ofpeopleinOAwithnotgoodhealth / .693
%ofpeopleaged16-74whoareunemployedinOA / .599 / .419
%ofhouseholdsthathaveloneparentwithdependentchildreninOA / .578
%ofhouseholdswithoccupancyrating-2orless / .812
%whiteinOA / -.709
numberofpersonsperhectareinoutputarea / .450
%ofhouseholdsclassifiedaswhollymoving
RESULTS
UsingAnalysisofvariancethereisevidencetosuggestthereisanapparentdifferencebetweenfaithschoolandnon-faithschools.Faithschoolsappeartohaveahigherperformancemeanscorethannonfaithschools.Table4showsthemeanscoreandstandarddeviationforfaithschoolsandnonfaithschools.Itshouldbeborneinmindthatthedescriptivestatisticsresultsarelimitedfortworeasons:
- Nopriorattainmentorotherbackgroundfactorsaretakenintoaccount
- Descriptivestatisticsdonottakeintoconsiderationclusteringpupilsintoschools.
Table4KS2ResultsforFaithnon-faithschools
KS2Averagepointscore / KS2Englishpointscore / KS2Maths pointscoreFaithSchools / Mean
Std.Deviation / 28.14
4.36 / 27.56
5.20 / 26.36
5.99
Non-faithSchools / Mean
Std.Deviation / 27.07
5.11 / 26.32
5.99 / 26.53
5.82
Overall / Mean
Std.Deviation / 28.14
4.41 / 27.41
5.30 / 27.44
5.42
ThethreeattainmentoutcomevariableswhichwereusedinthemultilevelmodellinganalysestomeasuretheimpactoffaithschoolwereKS2Average,KS2EnglishandKS2mathematics.Thefaithschoolsshowedasignificantdifference(p0.05)inattainmentcomparedtonon-faithschools.
Table5presentsthemeanscore,standarddeviationandconfidenceintervalforsubdivisionsfaithschools.TheDuncantest(Duncan,D.B.(1955).MultipleRangeandMultipleF-tests.Biometrics,11,1-42)wasusedtocomparetheMeankeystage2scores.UsingtheDuncantestthereisevidencetosuggestthatthereisanapparentdifferencebetweenJewishschoolsandotherfaithschools;RomanCatholicschoolsaresignificantlydifferencefromChurchofEnglandschoolsforalloutcomevariablesexceptKeyStage2Science.
Table5DifferentFaithschoolperformancebyattainment
N / Mean / Std.Deviation / Std.Error / 95%ConfidenceIntervalforMeanLowerBound / UpperBound
KeyStage2Average / Non-religious / 383218 / 27.08 / 5.11 / 0.01 / 27.06 / 27.09
ChurchofEngland / 102303 / 28.01 / 4.44 / 0.01 / 27.98 / 28.04
RomanCatholic / 51942 / 28.36 / 4.21 / 0.02 / 28.32 / 28.39
Jewish / 937 / 29.56 / 3.45 / 0.11 / 29.34 / 29.78
Combinefaith / 1346 / 28.25 / 4.36 / 0.12 / 28.02 / 28.48
Otherfaith / 831 / 28.28 / 4.00 / 0.14 / 28.01 / 28.55
Total / 540577 / 27.38 / 4.93 / 0.01 / 27.37 / 27.40
KeyStage2English / Non-religious / 386736 / 26.32 / 5.99 / 0.01 / 26.30 / 26.34
Churchofengland / 103155 / 27.39 / 5.32 / 0.02 / 27.36 / 27.43
RomanCatholic / 52187 / 27.85 / 4.97 / 0.02 / 27.80 / 27.89
Jewish / 940 / 29.11 / 4.00 / 0.13 / 28.85 / 29.36
Combinefaith / 1363 / 27.68 / 5.42 / 0.15 / 27.39 / 27.97
Otherfaith / 838 / 27.62 / 4.77 / 0.16 / 27.30 / 27.95
Total / 545219 / 26.68 / 5.80 / 0.01 / 26.67 / 26.70
KeyStage2Maths / Non-religious / 386507 / 26.46 / 6.07 / 0.01 / 26.45 / 26.48
Churchofengland / 103206 / 27.42 / 5.43 / 0.02 / 27.39 / 27.46
RomanCatholic / 52277 / 27.82 / 5.20 / 0.02 / 27.77 / 27.86
Jewish / 937 / 29.36 / 4.34 / 0.14 / 29.08 / 29.64
Combinefaith / 1362 / 27.54 / 5.38 / 0.15 / 27.26 / 27.83
Otherfaith / 833 / 27.75 / 4.98 / 0.17 / 27.41 / 28.09
Total / 545122 / 26.79 / 5.89 / 0.01 / 26.77 / 26.80
KeyStage2Science / Non-religious / 387470 / 28.31 / 5.07 / 0.01 / 28.29 / 28.33
Churchofengland / 103248 / 29.12 / 4.41 / 0.01 / 29.09 / 29.15
RomanCatholic / 52307 / 29.33 / 4.27 / 0.02 / 29.30 / 29.37
Jewish / 940 / 30.19 / 3.71 / 0.12 / 29.95 / 30.42
Combinefaith / 1356 / 29.39 / 4.25 / 0.12 / 29.17 / 29.62
Otherfaith / 837 / 29.40 / 4.22 / 0.15 / 29.12 / 29.69
Total / 546158 / 28.57 / 4.89 / 0.01 / 28.56 / 28.58
Thethreeattainmentoutcomevariableswhichwereusedinthemultilevelmodellinganalysistomeasuretheimpactoffaithschoolwere:KS2average,KS2EnglishandKS2mathematics.
Appendix2presentsthedetailedresultsofthemultilevelmodelfittedtotheKS2averagescore.DetailedresultofthemultilevelmodellingforKS2EnglishandKS2mathematicsarealsoprovidedinappendix2.Intechnicallanguage,thistableshowstherandomvariancesateachlevelateachstageofthemodelfitting,plusthecoefficientsofthebackgroundvariablesinthe‘fullmodel’.Itisalsoshowswhetherornotthevariancesorcoefficientsarestatisticallysignificantatthe5%level,aswellas95%confidenceintervalsforeachparameter.
Thetable, while showing full results of all the modelling carriedoutatthisstage,maynotbeeasytointerpretforallreaders.Tohelpwiththis,therefore,thecoefficientswhichexpresstheestimatedrelationshipsbetweenKS2averagescoreandeachofthebackgroundvariableshavebeenconvertedinto‘normalisedcoefficients’whichrepresentthe‘strength’ofeachrelationshipasapercentage,andwhichallowthedifferentvariablestobecomparedintermsoftheirapparentinfluenceonthetestoutcome,whenallothervariablesaresimultaneouslytakenintoaccount(SchagenElliot,2004).
Normalisedcoefficientsareplottedinfigure1,forthemodeldescribedinappendix;thenormalisedcoefficientforKS2Englishandmathematicsarelocatedinappendix1.Foreachvariable,theestimatednormalizedcoefficientsareplottedasadiamond,withaverticallineindicatingthe95%confidenceintervalfortheestimate.Anyvariablewhoselineintersectsthehorizonzeroaxiscanberegardedasnotstatisticallysignificant(atthe5%level).PositivevaluesimplyapositiverelationshipwiththeKS2averagescore;negativevaluesimplythatKS2averagescoretendstodecreasewithhighervaluesofthegivenbackgroundvariable.
AlthoughwehaveavarietyofbackgroundvariableswhichshowaneffectonKS2performance,thevariablesthatareparticularinterestarefaithschool,andtheinteractionbetweenfaithschoolandspecialeducationalneeds.Infigure1itcanbeseenthatfaithschoolsjustasignificantpositiveimpactonKS2performance.Fromtable1itisclearthatSENpupilattendinginfaithschoolperformedbetterinKS2thanPupilswithSENinnon-faithschools,whereasSENpupilhadasignificantnegativeimpactonKS2performance.AsimilarpatternofresultappearedinthemultilevelmodellingresultsfortheoutcomevariableofKS2English,MathematicsandScience(appendix2).
FIG1AverageKeystage2pointscore
Conclusion
Theresultofthemultilevelmodellinganalysisthatcontrolforallthecontextualbackgroundfactorsatpupil,schoolandpostcodelabelshowthatfaithschoolsarepositivelyrelatedtokeystage2 performance.
Theanalysisconfirmedthatallfaithschools,inparticularRomanCatholicandChurchofEnglandschools,madeslightlymoreprogresswiththeirpupilsthannon–faithschools.ItalsoshowedthatpupilswithSENattendingfaithschoolsperformedbetterinkeystage2thanpupilswithSENinnon-faithschools.
Acknowledgements
IwouldliketoacknowledgemycolleaguesinstatisticalandanalysisgroupatNFERwhoadvisedandsupportedmethroughoutthis.I alsowishtothank Joanna Luke and her team at the DCSFforgrantingaccesstotheanonymisednationaldatasetforthisresearch.
References
CHILD,D.(December1969)TheEssentialsofFactorAnalysis
DUNCAN,D.B.(1955).MultipleRangeandMultipleF-tests.Biometrics,11,1–42
GOLDSTEIN,H.(1995)MultilevelStatisticalModels2ndEdition. London,EdwardArnold.
GOLDSTEIN,H.SPIEGELHALTER,D.J.(1996)Leaguetablesandtheirlimitations:statisticalissuesincomparisonofinstitutionsperformance,JournaloftheRoyalStatisticalSociety,SeriesA(StatisticsinSociety).Pp.385-443
HAMMOND,P,Yeshanew,T(2007)Theimpactoffeedbackonschoolperformance
HECK,R.H.andTHOMAS,S.L.(2000)AnIntroductiontoMultilevelModellingTechniques. NewJersey,LawrenceErlbaumAssociates.
MORRIS,M.,RUTT,S.andYeshanew,T.Aimhigher:ExcellenceChallenge:pupiloutcomesoneyearon.
SCHAGEN,I.,andELLIOT,K.,(2004)(Eds)Butwhatdoesitmean?Theuseofeffectsizesineducationalresearch”.Slough,NFER.
SCHAGEN,I.(1996)(QUASE:QuantitativeAnalysisforSelf–Evaluation.
SPIELHOFER,T.,O’DONNEL,L.,BENTOM,T.,SCHAGEN,S.,SCHAGEN,I.,(2002)TheImpactofSchoolSizeandSingle-SexEducationonPerformance.Slough,NFER.
TOM,A.,SNIJDERSandROELJ.BOSKER.(1999)MultilevelAnalysis
Appendix1. Listofvariablesusedinmultilevelmodelling
Variables defined for multilevel modellingRange
No / Name / Description / Minimum / Maximum
176 / PUPILID / Pupil identifier / 1 / 548806
1 / LEA / Local educational Authority Identifier / 202 / 938
2 / SCHOOL / School identifier / 1100 / 7510
3 / K2ENG / Key Stage 2 English / 9 / 33
4 / K2MATH / Key Stage 2 Maths / 9 / 33
5 / K2SCI / Key Stage 2 Science / 9 / 33
19 / SEX / Sex of student / 1 / 2
20 / SEN / Special Educational Needs / 0 / 2
21 / FSM / Eligible for free school meals / 0 / 1
23 / EAL / English as an additional language / 0 / 1
28 / FIRSTMID / First & Middle School / 0 / 1
29 / JUNIOR / Junior school / 0 / 1
30 / MIDDLE / Middle school / 0 / 1
31 / SPECSCH / Special school / 0 / 1
32 / OTHSCH / Other non Infant & Junior primary school / 0 / 1
33 / FAITH / Faith school / 0 / 1
34 / CENGLAND / Church of England / 0 / 1
35 / CATHOLIC / Catholic / 0 / 1
36 / JEWISH / Jewish / 0 / 1
39 / BOYSCH / Boys' school / 0 / 1
40 / GIRLSCH / Girls' school / 0 / 1
41 / MIXSCH / mix school / 0 / 1
42 / PCFSM / % entitled to FSM / 0 / 100
43 / PCSEN / % of pupils with special needs with statements / 0 / 100
44 / PCEAL / % of EAL pupils / 0 / 100
49 / K2AV / Key Stage 2 Average / 9 / 33
53 / K1AV / Key Stage 1 Average / 3 / 27
56 / FEMALE / Female pupil / 0 / 1
57 / SENSA / SEN - School Action/Plus / 0 / 1
58 / SENSTAT / SEN - Statement / 0 / 1
59 / WHITUK / Ethnicity -White UK / 0 / 1
60 / WHITOTH / Ethnicity - White Non-UK / 0 / 1
61 / GYPSY / Ethnicity - Gypsy/Roma / 0 / 1
62 / ETHMIX / Ethnicity - Mixed / 0 / 1
63 / ASIANI / Ethnicity - Asian Indian / 0 / 1
64 / ASIANP / Ethnicity - Asian Pakistani / 0 / 1
65 / ASIANB / Ethnicity - Asian Bangladeshi / 0 / 1
66 / ASIANO / Ethnicity - Asian Other / 0 / 1
67 / BLACKC / Ethnicity - Black Caribbean / 0 / 1
68 / BLACKA / Ethnicity - Black African / 0 / 1
69 / BLACKO / Ethnicity - Black Other / 0 / 1
70 / CHINESE / Ethnicity - Chinese / 0 / 1
71 / ETHOTH / Ethnicity - Other / 0 / 1
72 / ETHREF / Ethnicity - Refused / 0 / 1
73 / ETHMISS / Ethnicity - Unknown / 0 / 1
74 / PUPMOB / Pupil moved schools between KS1 and KS2 / 0 / 1
129 / DEP1 / REGR factor score 1 for analysis / -2.06156 / 3.272273
130 / DEP2 / REGR factor score 2 for analysis / -1.59323 / 9.000828
135 / CONS / Constant term / 1 / 1
Interaction Terms
Variables defined for multilevel modelling
Range
No / Name / Description / Minimum / Maximum
136 / FAIFSM / FAITH * FSM / 0 / 1
137 / FAIEAL / FAITH * EAL / 0 / 1
138 / FAISEN / FAITH * SENA / 0 / 1
139 / FAISENTA / FAITH * SENSTAT / 0 / 1
140 / FAIDEP1 / FAITH * DEP1 / -2.17032 / 3.070899
141 / FAIDEP2 / FAITH * DEP2 / -1.60714 / 8.725002
142 / FAIBLACA / FAITH * BLACKA / 0 / 1
143 / FAIBLACO / FAITH * BLACKO / 0 / 1
144 / FAIBLACC / FAITH * BLACKC / 0 / 1
145 / FAICHIN / FAITH * CHINA / 0 / 1
146 / FAIWHOTH / FAITH * WHOTH / 0 / 1
147 / K1AVSQ / Key stage one average square / 9 / 729
148 / CHENGFSM / CENGLAND * FSM / 0 / 1
149 / CHENGEAL / CENGLAND * EAL / 0 / 1
150 / CHENGSENA / CENGLAND * SENA / 0 / 1
151 / CHESENSTA / CENGLAND * SENSTA / 0 / 1
152 / CHEDEP1 / CENGLAND * DEP1 / -2.17032 / 3.067291
153 / CHEDEP2 / CENGLAND * DEP2 / -1.60714 / 8.725002
154 / CHEBLACA / CENGLAND * BLACKA / 0 / 1
155 / CHEBLACO / CENGLAND * BLACKO / 0 / 1
156 / CHEBLACC / CENGLAND * BLACKC / 0 / 1
157 / CATFSM / CATHOLIC *FSM / 0 / 1
158 / CATEAL / CATHOLIC *EAL / 0 / 1
159 / CATSEN / CATHOLIC *SENA / 0 / 1
160 / CATSENTA / CATHOLIC *SENSTA / 0 / 1
161 / CATDEP1 / CATHOLIC *DEP1 / -2.02672 / 3.070899
162 / CATDEP2 / CATHOLIC *DEP2 / -1.49128 / 8.692374
163 / CATBLACA / CATHOLIC *BLACKA / 0 / 1
164 / CATBLACO / CATHOLIC *BLACKO / 0 / 1
165 / CATBLACC / CATHOLIC *BLACKC / 0 / 1
166 / JEWFSM / JEWISH * FSM / 0 / 1
167 / JEWEAL / JEWISH * EAL / 0 / 1
168 / JEWSEN / JEWISH * SENA / 0 / 1
169 / JEWSENTA / JEWISH * SENSTA / 0 / 1
170 / JEWDEP1 / JEWISH * DEP1 / -1.8572 / 1.682314
171 / JEWDEP2 / JEWISH * DEP2 / -0.78969 / 4.301277
172 / JEWBLACA / JEWISH * BLACKA / 0 / 1
173 / JEWBLACO / JEWISH * BLACKO / 0 / 1
174 / JEWBLACC / JEWISH * BLACKC / 0 / 0
Appendix2:Detailedresultsfrommultilevelmodelling
KS2Averagescoredetailedresults
AveargeKS2 / Multilevelresults95%Confidenceinterval
Parameter / Estimate / Standarderror / Sig. / Min. / Max.
Basecase
Leavariance / 3.260 / 0.054 / * / 3.154 / 3.366
Schoolvariance / 9.475 / 0.119 / * / 9.242 / 9.708
Pupilvariance / 19.372 / 0.038 / * / 19.298 / 19.446
Finalmodel
LEAvariance / 0.754 / 0.115 / *
LEAks2cov / -0.037 / 0.006 / *
LEAks2variance / 0.002 / 0.000 / *
schoolvariance / 12.550 / 0.201 / * / 12.156 / 12.944
schoolks2cov / -0.618 / 0.011 / * / -0.640 / -0.596
Schoolks2variance / 0.032 / 0.001 / * / 0.030 / 0.034
Pupilvariance / 6.450 / 0.013 / * / 6.425 / 6.475
Fixedcoefficients
CONS / 14.823 / 0.095 / * / 14.637 / 15.009
AverageKS1score / 0.970 / 0.007 / * / 0.956 / 0.984
Sex / -0.318 / 0.009 / * / -0.336 / -0.300
k1avsq / -0.004 / 0.000 / * / -0.004 / -0.004
FreeSchoolMeals / -0.379 / 0.011 / * / -0.401 / -0.357
EnglishasanAdditionalLanguage / 0.265 / 0.026 / * / 0.214 / 0.316
senscoolaction/plus / -1.840 / 0.013 / * / -1.865 / -1.815
sen–statement / -3.781 / 0.031 / * / -3.842 / -3.720
Female / -0.455 / 0.007 / * / -0.469 / -0.441
Middle / -0.294 / 0.055 / * / -0.402 / -0.186
faithschool / 0.064 / 0.018 / * / 0.029 / 0.099
Boysch / 1.616 / 0.266 / * / 1.095 / 2.137
percentageoffsm / -0.010 / 0.001 / * / -0.012 / -0.008
percentageofeal / -0.042 / 0.001 / * / -0.044 / -0.040
WhiteOther / 0.448 / 0.031 / * / 0.387 / 0.509
Gypsy / -0.894 / 0.130 / * / -1.149 / -0.639
Mixethnicity / 0.123 / 0.022 / * / 0.080 / 0.166
Indian / 0.104 / 0.033 / * / 0.039 / 0.169
Pakistani / -0.267 / 0.035 / * / -0.336 / -0.198
Bangladish / 0.228 / 0.048 / * / 0.134 / 0.322
asianother / 0.501 / 0.056 / * / 0.391 / 0.611
BlackCarribbean / -0.424 / 0.038 / * / -0.498 / -0.350
Blackother / -0.250 / 0.062 / * / -0.372 / -0.128
Chinese / 0.971 / 0.069 / * / 0.836 / 1.106
otherethnicity / 0.505 / 0.050 / * / 0.407 / 0.603
deprivation1 / -0.201 / 0.005 / * / -0.211 / -0.191
deprivation2 / 0.072 / 0.006 / * / 0.060 / 0.084
Interaction:faithv.EAL / -0.120 / 0.040 / * / -0.198 / -0.042
Interaction:faithv.SENACTION / 0.263 / 0.023 / * / 0.218 / 0.308
Interaction:faithv.SENSTATEMENT / 0.247 / 0.058 / * / 0.133 / 0.361
Interaction:faithv.Blackcarribean / -0.158 / 0.073 / * / -0.301 / -0.015
pupilmobility / -0.249 / 0.011 / * / -0.271 / -0.227
KS2Englishscoredetailedresults
KS2english / Multilevel results95% Confidence interval
Parameter / Estimate / Standard error / Sig. / Min. / Max.
Base case
Lea variance / 3.680 / 0.059 / * / 3.564 / 3.796
School variance / 9.370 / 0.122 / * / 9.131 / 9.609
Pupil variance / 28.083 / 0.055 / * / 27.975 / 28.191
Final model
LEA variance / 1.320 / 0.192 / *
LEA ks2cov / -0.066 / 0.010 / *
LEA ks2 variance / 0.003 / 0.001 / *
school variance / 16.027 / 0.277 / * / 15.484 / 16.570
school ks2cov / -0.806 / 0.015 / * / -0.835 / -0.777
School ks2 variance / 0.043 / 0.001 / * / 0.041 / 0.045
Pupil variance / 11.266 / 0.022 / * / 11.223 / 11.309
Fixed coefficients
CONS / 11.987 / 0.123 / * / 11.746 / 12.228
Average KS1 score / 1.189 / 0.010 / * / 1.169 / 1.209
k1avsq / -0.011 / 0.000 / * / -0.011 / -0.011
Free School Meals / -0.422 / 0.014 / * / -0.449 / -0.395
English as an Additional Language / 0.276 / 0.037 / * / 0.203 / 0.349
sen scool action/plus / -2.600 / 0.017 / * / -2.633 / -2.567
sen -statement / -4.885 / 0.040 / * / -4.963 / -4.807
female / 0.716 / 0.009 / * / 0.698 / 0.734
percentage of fsm / -0.010 / 0.001 / * / -0.012 / -0.008
junior / -0.094 / 0.034 / * / -0.161 / -0.027
middle / -0.329 / 0.067 / * / -0.460 / -0.198
faith school / 0.135 / 0.023 / * / 0.090 / 0.180
boysch / 1.205 / 0.321 / * / 0.576 / 1.834
percentage of sen / -0.003 / 0.001 / * / -0.005 / -0.001
percentage of eal / -0.028 / 0.001 / * / -0.030 / -0.026
White Other / 0.466 / 0.041 / * / 0.386 / 0.546
gypsy / -0.877 / 0.166 / * / -1.202 / -0.552
Mix ethnicity / 0.327 / 0.029 / * / 0.270 / 0.384
Indian / 0.249 / 0.046 / * / 0.159 / 0.339
Pakistani / 0.151 / 0.048 / * / 0.057 / 0.245
Bangladish / 0.567 / 0.065 / * / 0.440 / 0.694
asian other / 0.482 / 0.075 / * / 0.335 / 0.629
Black Africa / 0.546 / 0.049 / * / 0.450 / 0.642
Chinese / 0.451 / 0.092 / * / 0.271 / 0.631
pupil mobility / -0.243 / 0.015 / * / -0.272 / -0.214
deprivation 1 / -0.206 / 0.007 / * / -0.220 / -0.192
deprivation 2 / 0.078 / 0.008 / * / 0.062 / 0.094
Interaction: faith v. EAL / -0.189 / 0.052 / * / -0.291 / -0.087
Interaction: faith v. SEN ACTION / 0.493 / 0.030 / * / 0.434 / 0.552
Interaction: faith v. SEN STATEMENT / 0.357 / 0.075 / * / 0.210 / 0.504
Interaction: faith v. Black carribean / -0.257 / 0.082 / * / -0.418 / -0.096
KS2 Mathematics score detailed results
KS2 Maths / Multilevel results95% Confidence interval
Parameter / Estimate / Standard error / Sig. / Min. / Max.
Base case
Lea variance / 0.361 / 0.058 / * / 0.247 / 0.475
School variance / 8.702 / 0.115 / * / 8.477 / 8.927
Pupil variance / 29.400 / 0.057 / * / 29.288 / 29.512
Final model
LEA variance / 0.908 / 0.143 / *
LEA ks2cov / -0.044 / 0.007 / *
LEA ks2 variance / 0.002 / 0.000 / *
school variance / 15.267 / 0.279 / * / 14.720 / 15.814
school ks2cov / -0.746 / 0.015 / * / -0.775 / -0.717
School ks2 variance / 0.038 / 0.001 / * / 0.036 / 0.040
Pupil variance / 13.388 / 0.026 / * / 13.337 / 13.439
Fixed coefficients
CONS / 13.800 / 0.112 / * / 13.580 / 14.020
Average KS1 score / 0.977 / 0.010 / * / 0.957 / 0.997
k1avsq / -0.001 / 0.000 / * / -0.001 / -0.001
Free School Meals / -0.311 / 0.015 / * / -0.340 / -0.282
English as an Additional Language / 0.454 / 0.038 / * / 0.380 / 0.528
sen scool action/plus / -1.958 / 0.019 / * / -1.995 / -1.921
sen -statement / -3.676 / 0.038 / * / -3.750 / -3.602
female / -1.324 / 0.010 / * / -1.344 / -1.304
percentage of fsm / -0.009 / 0.001 / * / -0.011 / -0.007
middle / -0.447 / 0.064 / * / -0.572 / -0.322
boysch / 1.231 / 0.328 / * / 0.588 / 1.874
percentage of sen / -0.003 / 0.001 / * / -0.005 / -0.001
percentage of eal / -0.035 / 0.001 / * / -0.037 / -0.033
White Other / 0.485 / 0.045 / * / 0.397 / 0.573
gypsy / -0.930 / 0.182 / * / -1.287 / -0.573
Indian / 0.360 / 0.050 / * / 0.262 / 0.458
Pakistani / -0.229 / 0.052 / * / -0.331 / -0.127
Bangladish / 0.353 / 0.070 / * / 0.216 / 0.490
asian other / 0.930 / 0.081 / * / 0.771 / 1.089
Black Carribean / -0.776 / 0.047 / * / -0.868 / -0.684
Black Africa / -0.190 / 0.053 / * / -0.294 / -0.086
Black other / -0.479 / 0.088 / * / -0.651 / -0.307
Chinese / 1.875 / 0.099 / * / 1.681 / 2.069
pupil mobility / -0.353 / 0.015 / * / -0.382 / -0.324
KS2 Science score detailed results
ks2sci / Multilevel results95% Confidence interval
Parameter / Estimate / Standard error / Sig. / Min. / Max.
Base case
Lea variance / 0.283 / 0.049 / * / 0.187 / 0.379
School variance / 9.817 / 0.123 / * / 9.576 / 10.058
Pupil variance / 19.008 / 0.037 / * / 18.935 / 19.081
Final model
LEA variance / 0.497 / 0.095 / *
LEA ks2cov / -0.025 / 0.005 / *
LEA ks2 variance / 0.001 / 0.000 / *
school variance / 18.274 / 0.298 / * / 17.690 / 18.858
school ks2cov / -0.911 / 0.016 / * / -0.942 / -0.880
School ks2 variance / 0.047 / 0.001 / * / 0.045 / 0.049
Pupil variance / 10.404 / 0.020 / * / 10.365 / 10.443
Fixed coefficients
CONS / 18.530 / 0.095 / * / 18.344 / 18.716
Average KS1 score / 0.770 / 0.009 / * / 0.752 / 0.788
k1avsq / -0.002 / 0.000 / * / -0.002 / -0.002
Free School Meals / -0.428 / 0.013 / * / -0.453 / -0.403
sen scool action/plus / -1.013 / 0.017 / * / -1.046 / -0.980
sen -statement / -2.808 / 0.038 / * / -2.882 / -2.734
female / -0.760 / 0.009 / * / -0.778 / -0.742
percentage of fsm / -0.012 / 0.001 / * / -0.014 / -0.010
Junior / 0.064 / 0.031 / * / 0.003 / 0.125
middle / -0.151 / 0.063 / * / -0.274 / -0.028
boysch / 1.827 / 0.313 / * / 1.214 / 2.440
percentage of sen / 0.004 / 0.001 / * / 0.002 / 0.006
percentage of eal / -0.054 / 0.001 / * / -0.056 / -0.052
White Other / 0.427 / 0.037 / * / 0.354 / 0.500
gypsy / -0.941 / 0.160 / * / -1.255 / -0.627
mix ethnicity / 0.084 / 0.028 / * / 0.029 / 0.139
Indian / -0.241 / 0.036 / * / -0.312 / -0.170
Pakistani / -0.676 / 0.037 / * / -0.749 / -0.603
Bangladish / -0.154 / 0.055 / * / -0.262 / -0.046
asian other / 0.142 / 0.068 / * / 0.009 / 0.275
Black Carribean / -0.568 / 0.042 / * / -0.650 / -0.486
Black Africa / -0.239 / 0.044 / * / -0.325 / -0.153
Black other / -0.341 / 0.078 / * / -0.494 / -0.188
Chinese / 0.644 / 0.084 / * / 0.479 / 0.809
pupil mobility / -0.182 / 0.014 / * / -0.209 / -0.155
deprivation 1 / -0.209 / 0.006 / * / -0.221 / -0.197
deprivation 2 / 0.063 / 0.008 / * / 0.047 / 0.079
Interaction: faith v. SEN ACTION / 0.159 / 0.029 / * / 0.102 / 0.216
Interaction: faith v. SEN STATEMENT / 0.351 / 0.072 / * / 0.210 / 0.492
KS2 English Score normalised coefficients
KS2 Science Score normalised coefficients
1