Socio-economic differences in higher education ...€¦ · Chalk + Talk: The University Challenge...

Chalk + Talk: The University Challenge Socio-economic differences in higher education participation and outcomes Speaker: Dr Claire Crawford, University of Warwick and Institute of Fiscal Studies Chair: Nida Broughton, Chief Economist, SMF

@SMFthinktank | smf.co.uk #SMFchalktalk

Wi-Fi Network: SMF Password: SMF2014Market!

Socio-economic differences in higher education participation and outcomes

Claire Crawford

University of Warwick and Institute for Fiscal Studies

Motivation: rising socio-economic inequalities in HE participation and degree acquisition over time

18.9ppts 22.6ppts

0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

40.0

45.0

NCDS BCS BHPS Change: NCDS to BHPS

Perc

enta

ge p

oint

dif

fere

nce

Difference in HE participation/degree acquisition rates between those in the top and bottom income quintile groups

HE participation at age 19 Degree acquisition by age 23

Source: Blanden & Machin (2004), Educational inequality and the expansion of UK higher education, Scottish Journal of Political Economy, Vol. 51, pp. 230-249.

Motivation: what has happened since then?

• Participation in higher education has continued to rise

• Fees and student support arrangements have changed significantly

• SES differences in some measures of attainment have been falling

SES gap in terms of % getting 5 A*-C grades in GCSEs and equivalents has fallen substantially

0%

5%

10%

15%

20%

25%

30%

35%

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

2004 2005 2006 2007 2008 2009 2010 2011 2012

% pupils getting 5 A*-C grades in GCSEs and equivalents

FSM Non-FSM Difference (RH axis)

2010-2012 figures based on SFR 04/2013: GCSE and Equivalent Attainment by Pupil Characteristics in England. 2006-2009 figures based on SFR 37/2010: GCSE and Equivalent Attainment by Pupil Characteristics in England.

2004-2005 figures based on authors’ calculations using Key Stage 4 and PLASC data.

Motivation: what does this mean for SES gaps in HE participation and outcomes?

• Prior attainment:

– Given key role in driving HE participation and outcomes, poorer students “catching up” with their better off peers may decrease SES gaps

• Changes to student finance:

– Concerns that prospect of high fees/debt levels would create a barrier to participation/retention for poorer students and hence increase SES gaps

• Mass HE participation:

– Potential “selection effects”: lower ability students may be less likely to complete their degree and less likely to graduate with a 1st or a 2:1

• Effect on SES gaps ambiguous (depends where new participants are drawn from)

• Empirical question . . .

Plan for today

• Document socio-economic gaps in HE participation, drop-out, degree completion and degree class

• Explore the extent to which these gaps can be explained by differences in other characteristics, especially prior attainment

• Compare these results to differences by school characteristics (including differences in labour market outcomes)

• What policy implications can be drawn from our results?

Data

• Analysis of HE participation, drop-out, degree completion and degree class uses linked NPD-HESA data

– NPD: population of pupils attending schools in England

– HESA: population of students attending universities in the UK

→ Allows us to follow the population of pupils attending schools in England from age 11 through to potential degree completion

• Analysis of labour market outcomes uses DLHE data

– Intended to be a census of those graduating from university (2007 here)

– Information on activity status collected after 6 months and 3.5 years, and matched to HESA data, so have some background characteristics

Key covariates of interest

• Socio-economic status

– Combine FSM eligibility at age 16 with measures of local area deprivation based on pupils’ home postcode at age 16

– Split state school population into quintile groups based on this index

– Add private school students to top quintile group

• Secondary school performance:

– % of pupils in school achieving at least 5 A*-C grades at GCSE

– Pupils split into quintile groups based on this measure

• Type of school in which sat A-levels (for labour market analysis):

– State vs. private school

HE participation

Outcomes: HE participation

• Participation at any UK university for the first time at age 18 or 19, for participants first eligible to go to university 2004-05 to 2010-11

• Participation at a “high status” institution, where high status is:

– Russell Group institutions (20 in total pre-2012)

– Plus any UK university with a 2001 average RAE score higher than the lowest amongst the Russell Group (an extra 21 institutions)

HE participation: how has it changed amongst state school students over time?

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

2005 2006 2007 2008 2009 2010 2011

Participation at age 18 or 19 High status participation at age 18 or 19

Source: authors’ calculations based on linked schools and universities administrative data for the cohorts first eligible to start university between 2004-05 and 2010-11 (who sat their GCSEs between 2002 and 2008)

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

2005 2006 2007 2008 2009 2010 2011


Participation: most deprived High status participation: most deprived

Participation: least deprived High status participation: least deprived

Source: authors’ calculations based on linked schools and universities administrative data for the cohorts first eligible to start university between 2004-05 and 2010-11 (who

sat their GCSEs between 2002 and 2008)


40.0ppts 36.6ppts

19.5ppts 18.4ppts

0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

40.0

45.0

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

2005 2006 2007 2008 2009 2010 2011

Difference in participation (least-most) Difference in high status participation (least-most)


Participation: most deprived High status participation: most deprived

Participation: least deprived High status participation: least deprived

Source: authors’ calculations based on linked schools and universities administrative data for the cohorts first eligible to start university between 2004-05 and 2010-11 (who sat their

GCSEs between 2002 and 2008)


HE participation overall and at high status institutions for state school pupils first eligible to go in 2010-11, by SES

19.1%

25.6%

34.1%

42.3%

55.7%

36.6ppts

2.9% 4.7% 7.7% 12.1%

21.3% 18.4ppts

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

70.0%

Lowest SES quintile group

2nd 3rd 4th Highest SES quintile group

Difference (highest - lowest)

% pupils going to university at age 18/19: highest SES quintile group including state school pupils only

HE participation overall HE participation at a high status institution

Source: authors’ calculations based on linked schools and universities administrative data for the cohort first eligible to start university in 2010-11 (who sat their GCSEs in 2007-08)

HE participation overall and at high status institutions for all pupils first eligible to go in 2010-11, by SES

19.1%

25.6%

34.1%

42.3%

60.9%

41.8ppts

2.9% 4.7% 7.7% 12.1%

28.1% 25.2ppts

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

70.0%

Lowest SES quintile group

2nd 3rd 4th Highest SES quintile group

Difference (highest - lowest)

% pupils going to university at age 18/19: highest SES quintile group including state and private school pupils

HE participation overall HE participation at a high status institution


What explains differences in HE participation between state school pupils from most and least deprived backgrounds?

36.6ppts

18.2ppts

3.8ppts 3.0ppts

18.4ppts

7.0ppts

1.6ppts 1.3ppts 0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

40.0

Raw Plus individual and school characteristics

and KS2 results

Plus Key Stage 4 and equivalent results

Plus Key Stage 5 and equivalent results

Perc

enta

ge p

oint

dif

fere

nce

HE participation overall Participation at high status institutions


Summary

• Large differences in HE participation overall and at high status institutions on the basis of socio-economic status

• But these gaps can largely be explained by differences in prior attainment between pupils from different backgrounds

– Particularly emphasise the substantial explanatory power of KS4

• Suggests that before the end of secondary school is a potentially vital period for interventions to “widen” participation in HE

Drop-out, degree completion and degree class

Outcomes: drop-out

• Drop-out in first or second year:

– Defined only for those who went to university at age 18 or 19

– Focus on those who leave the sector completely; anyone who transfers to another university is included in the zeroes

• Need to be able to observe three years of data to define measure

– Means focus on those first eligible to go 2004-05 to 2008-09

• 11.5% drop-out on our measure

• Slightly lower (9.7%) if we focus on full-time first degree entrants

Outcomes: degree completion and degree class

• For both outcomes, focus on those completing within 5 years

– Means need to be able to observe 5 years of data to define measure

– Hence focus on those first eligible to go 2004-05 to 2006-07

• Degree completion:

– Defined for those who went to university at age 18 or 19 to study full-time for a first degree in a non-medical subject

– 78.2% complete their degree within 5 years on our definition

• Graduate with a 1st or a 2:1:

– Sample as above but additionally restricted to those who complete their degree within 5 years

– 64.6% of degree completers graduate with a 1st or a 2:1 on our definition

% of HE participants from state schools who drop-out, complete their degree and graduate with a first or 2:1, by percentile of socio-economic background

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

1

4

7

10

13

16

19

22

25

28

31

34

37

40

43

46

49

52

55

58

61

64

67

70

73

76

79

82

85

88

91

94

97

10

0

Percentile of SES distribution (1=least deprived; 100=most deprived)

Drop-out within 2 years Complete degree within 5 years Graduate with a first or 2:1

Source: authors’ calculations based on linked NPD-HESA data for the cohorts first eligible to start university between 2004-05 and 2008-09 for drop-out, and between 2004-05 and

2006-07 for degree completion and degree class

What explains differences in university outcomes between pupils from high and low SES backgrounds?

-8.4ppts -7.2ppts

-3.9ppts -3.7ppts

-3.4ppts

13.3ppts

9.3ppts

6.1ppts 5.5ppts 5.3ppts

22.9ppts

10.1ppts

5.2ppts 4.3ppts 3.7ppts

-10.0

-5.0

0.0

5.0

10.0

15.0

20.0

25.0

Raw (accounting for cohort)

Plus individual and school

characteristics and KS2 results

Plus KS4 and equivalent results


Plus university attended and

subject studied

Perc

enta

ge p

oint

dif

fere

nce

Dropout within 2 years Complete degree within 5 years Graduate with a first or 2:1



How does this compare to the differences between pupils from the highest and lowest performing schools?

-6.3ppts

-2.3ppts

2ppts 1.8ppts 1.8ppts

9.8ppts

2.7ppts

-2.7ppts -2.6ppts

-2.9ppts

17.5ppts

3ppts

-7.3ppts -7.4ppts -7.8ppts -10.0

-5.0

0.0

5.0

10.0

15.0

20.0

Raw (accounting for cohort)

Plus individual characteristics and

KS2 results



Plus university attended and

subject studied

Perc

enta

ge p

oint

dif

fere

nce

Dropout within 2 years Complete degree within 5 years Graduate with a first or 2:1



Summary

• Differences in HE outcomes smaller, on average, than participation, and in expected direction (but amongst selected sample)

• Controlling for attainment on entry to university substantially reduces SES differences; comparing students on the same courses makes little difference over and above accounting for attainment

– Students from high SES backgrounds still, on average, less likely to drop out, more likely to complete degree and more likely to get first or 2:1 than students from low SES backgrounds

• Different picture when comparing outcomes by school performance

– Students from high-performing schools are, on average, more likely to drop out, less likely to complete degree and less likely to get first or 2:1 once we account for differences in attainment prior to university entry

Policy implications?

• Attainment during secondary school still a key driver of progression and performance at university, so SES gaps in these outcomes may fall if attainment rises earlier in the school system

• Differences by school characteristics suggest that pupils from low performing schools with the same attainment as those from high performing schools have, on average, higher “potential”

– Universities may wish to account for this in making entry offers

– If they do, they are likely to get it right on average

• Same is not true for individual/neighbourhood measures of SES

– Does not mean that no students from lower SES backgrounds will go on to outperform students from higher SES backgrounds at university

– But it is not true on average: makes it more challenging for universities to identify low SES students with high potential to do well

Early labour market outcomes

Is HE a route to social mobility?

• Returns to education in the UK largely accrue to qualifications

– Also vary by institution, subject and degree class

• But those from lower SES backgrounds and lower performing schools are less likely to attend a high status institution, less likely to complete their degree and less likely to get a first or 2:1

• Do such differences persist into the labour market too?

• Look at earnings differences 6 months and 3.5 years after graduation for those who attended private vs. state schools

Differences in earnings between graduates who previously attended private vs. state schools

12.8% 11.7%

6.7% 5.8%

17.2%

12.7%

7.1% 6.0%

0%

2%

4%

6%

8%

10%

12%

14%

16%

18%

20%

Raw Plus background controls

Plus prior attainment Plus occupation

6 months 3.5 years

Source: authors’ calculations based on Destination of Leavers from Higher Education data for UK-domiciled students who studied full-time for a first degree and graduated in 2006-07.

Summary

• Private school students earn more than state school students, even when we compare those who went to the same universities, studied the same subjects and went into the same occupations

• Why?

– Better social networks?

– Better non-cognitive skills?

– Or are we still not measuring ability well enough? • Can check this now that NPD-HESA and DLHE have been linked

• But as things stand, those from lower socio-economic backgrounds and state schools do not appear to benefit to the same extent from higher education: challenge for social mobility?

Relevant published work

• Chowdry, H., C. Crawford, L. Dearden, A. Goodman and A. Vignoles (2013), Widening participation in higher education: analysis using linked administrative data, Journal of the Royal Statistical Society: Series A, Vol. 176, pp. 431-457.

• Crawford, C. (2012), Socio-economic gaps in HE participation: how have they changed over time?, IFS Briefing Note BN133.

• Crawford, C. (2014a), The link between secondary school characteristics and HE participation and outcomes, CAYT Research Report (https://www.gov.uk/government/publications/secondary-school-characteristics-and-university-participation).

• Crawford, C. (2014b), Socio-economic differences in university outcomes in the UK: drop-out, degree completion and degree class, IFS Working Paper No. 14/31.

• Crawford, C. and A. Vignoles (2014), Heterogeneity in graduate earnings by socio-economic background, IFS Working Paper No. 14/30.

• Crawford, C., L. Macmillan and A. Vignoles (2014), Progress made by high-attaining children from disadvantaged backgrounds, CAYT Research Report (https://www.gov.uk/government/publications/high-attaining-children-from-disadvantaged-backgrounds).

Socio-economic differences in higher education ...€¦ · Chalk + Talk: The University Challenge...

Documents

Transcript of Socio-economic differences in higher education ...€¦ · Chalk + Talk: The University Challenge...