Widening Participation in Higher Education: Analysis using Linked Admin Data Institute of Education Institute for Fiscal Studies Centre for Economic Performance

Research Team Haroon Chowdry Claire Crawford Lorraine Dearden Alissa Goodman Anna Vignoles

Background and Motivation Rapid expansion of HE in the UK –43% of year olds participated in But widening participation still cause for concern –Socio-economic gap in HE participation appeared to widen in mid to late 1990s Blanden & Machin (2004); Galindo-Rueda et al (2004); Glennerster (2001); Machin & Vignoles (2004) –Although may have narrowed somewhat since then Raffe et al. (2006)

Background and Motivation Concerns increased following introduction of tuition fees in 1998 –But did not deter low income students (who were protected by increased loan availability) (Dearden, Fitzsimons & Wyness, 2008) Recent policy developments may affect future participation –e.g reforms (top-up fees) –Will soon be evaluated using this data

Research Questions How does the likelihood of HE participation vary by socio-economic background? How much of this gap can be explained by prior achievement? How does the type of HE participation vary across socio-economic groups?

Methodology Linear probability regression model –Easier to include school fixed effects Two models: –HE participation (at age 19/20) –HE participation in a high status institution Dependent variables are binary –1 if participates, 0 otherwise

New longitudinal admin data Linked individual-level administrative data –School, FE and HE records Data on participants AND non-participants Consider two cohorts: –In Year 11 in or –Potential age 18/19 HE entry in or (age 19/20 entry / ) State and private school students

Data Socio-economic background –State school analysis: Free school meals status from PLASC IMD quintiles based on home postcode (age 16) –State and private school analysis: Assume FSM = 0 for all private school kids IMD quintiles based on school postcode (age 16) –47% of state school kids are in same quintile using home or school postcode 81% are in same or adjacent quintile

Data Gender, MOB and school ID available for all –School fixed effects for state school analysis –School type dummies when include private school kids Ethnicity, EAL, SEN from PLASC –Missing for private school kids Neighbourhood measure of parental education based on 2001 Census –Based on home postcode for state school analysis –Based on school postcode when include private school kids

Data Prior attainment –State school analysis: Quintiles (based on APS) at Key Stage 2, 3, 4 and 5 (plus indicators of reaching expected level at Key Stage 4 and 5) –Private school analysis: Key Stage 4 and 5 results only Exclusion of Key Stage 2 and 3 results makes negligible difference Use of school rather than home postcode reduces raw differences (but end result similar) –Essentially eliminating within-school differences

Male HE participation, by deprivation quintile

HE participation (state school males) No controls Individual and school controls Plus Key Stage 2 results Plus Key Stage 3 results Plus Key Stage 4 results Plus Key Stage 5 results 4 th deprivation quintile0.065** [0.003] 0.048** [0.002] 0.029** [0.002] 0.017** [0.001] 0.003* [0.001] [0.001] 3 rd deprivation quintile0.134** [0.003] 0.085** [0.002] 0.055** [0.002] 0.035** [0.002] 0.010** [0.002] [0.001] 2 nd deprivation quintile0.201** [0.004] 0.118** [0.002] 0.079** [0.002] 0.052** [0.002] 0.017** [0.002] [0.002] Least deprived quintile0.288** [0.006] 0.160** [0.003] 0.110** [0.003] 0.076** [0.002] 0.031** [0.002] 0.007** [0.002] Observations550,972 R-squared F-test of extra controls0.000

HE participation (state and private school males) No controlsIndividual and school controls Plus Key Stage 4 results Plus Key Stage 5 results 4 th deprivation quintile0.109** [0.008] 0.098** [0.006] 0.016** [0.003] 0.004* [0.002] 3 rd deprivation quintile0.130** [0.008] 0.108** [0.006] 0.015** [0.003] [0.002] 2 nd deprivation quintile0.173** [0.008] 0.143** [0.007] 0.019** [0.003] [0.002] Least deprived quintile0.223** [0.008] 0.171** [0.007] 0.026** [0.004] [0.002] Observations584,259 R-squared F-test of extra controls0.000

Type of Participation Also consider type of HE participation, because: –Students at less prestigious institutions more likely to drop out and/or achieve lower degree classification –Graduates from more prestigious institutions earn higher returns in the labour market Define high status university as: –Russell Group university (20 in total) –Any UK university with an average 2001 RAE score greater than lowest found amongst Russell Group Adds Bath, Durham, Lancaster, York, etc (21 in total)

Female high status participation, by deprivation quintile

High status HE participation (state school females) No controls Individual and school controls Plus Key Stage 2 results Plus Key Stage 3 results Plus Key Stage 4 results Plus Key Stage 5 results 4 th deprivation quintile0.049** [0.005] 0.031** [0.004] 0.019** [0.004] 0.012** [0.004] [0.004] [0.004] 3 rd deprivation quintile0.101** [0.005] 0.048** [0.004] 0.032** [0.004] 0.021** [0.004] 0.013** [0.004] 0.009* [0.004] 2 nd deprivation quintile0.148** [0.005] 0.063** [0.005] 0.043** [0.005] 0.029** [0.004] 0.018** [0.004] 0.012** [0.004] Least deprived quintile0.200** [0.007] 0.076** [0.005] 0.054** [0.005] 0.038** [0.005] 0.026** [0.005] 0.017** [0.004] Observations181,391 R-squared F-test of extra controls0.000

High status HE participation (state and private school females) No controlsIndividual and school controls Plus Key Stage 4 results Plus Key Stage 5 results 4 th deprivation quintile0.087** [0.012] 0.062** [0.010] 0.018** [0.006] [0.005] 3 rd deprivation quintile0.120** [0.012] 0.073** [0.010] 0.022** [0.006] 0.012* [0.006] 2 nd deprivation quintile0.144** [0.012] 0.073** [0.009] 0.020** [0.006] [0.006] Least deprived quintile0.177** [0.011] 0.097** [0.010] 0.031** [0.007] 0.019** [0.006] Observations205,523 R-squared F-test of extra controls0.000

Conclusions Widening participation in HE to students from deprived backgrounds is largely about tackling low prior achievement Focusing policy interventions post compulsory schooling unlikely to eliminate raw socio-economic gap in HE participation –But does not absolve universities

Limitations Young participants only –But other work looks at mature students Limited information on private school students

HE participation (state school males without Key Stage 2 & Key Stage 3 results) No controlsIndividual and school controls Plus Key Stage 4 results Plus Key Stage 5 results 4 th deprivation quintile0.065** [0.003] 0.048** [0.002] 0.003** [0.001] [0.001] 3 rd deprivation quintile0.134** [0.003] 0.085** [0.002] 0.010** [0.002] [0.001] 2 nd deprivation quintile0.201** [0.004] 0.118** [0.002] 0.018** [0.002] [0.002] Least deprived quintile0.288** [0.006] 0.160** [0.003] 0.032** [0.002] 0.007** [0.002] Observations550,972 R-squared F-test of extra controls0.000

HE participation (state school males without KS2 & KS3 and using school postcode) No controlsIndividual and school controls Plus Key Stage 4 results Plus Key Stage 5 results 4 th deprivation quintile0.098** [0.007] 0.100** [0.006] 0.014** [0.002] 0.004* [0.002] 3 rd deprivation quintile0.122** [0.007] 0.115** [0.006] 0.015** [0.002] [0.002] 2 nd deprivation quintile0.165** [0.008] 0.155** [0.006] 0.018** [0.003] [0.002] Least deprived quintile0.213** [0.008] 0.187** [0.007] 0.027** [0.003] [0.002] Observations539,298 R-squared F-test of extra controls0.000