Neighbours, Peers and Educational Achievement2015/01/22  · Effect of neighbour changes on own ks3...

Suggested hashtag for Twitter users: #LSEworks

Neighbours, Peers and Educational AchievementLSE Works: Spatial Economics Research Centre public lecture

Professor Stephen GibbonsProfessor of Economic Geography Director, Spatial Economics Research Centre, LSE

Dr Tim LeunigChief Analyst and Chief Scientific Adviser Department for Education

Professor Ruth LuptonProfessor of Education University of Manchester

Professor Julia BlackChair, LSE

Neighbours, peers and children’s education:Quantitative evidence from the Spatial Economics Research Centre

Steve Gibbons

January 2015

Outline of this talk

Is there a ‘transformative’ role for neighbours and peers – do they really matter for children’s education outcomes?

Overview of recent quantitative empirical work on England from SERC researchers

1. ‘Neighbourhood effects’: Who your neighbours are 2. ‘Peer effects’: The ability of your child’s school-mates 3. ‘Mobility’: How fast neighbours and peers come and go 4. Comparisons with other interventions 5. Conclusions and policy implications

1

‘Neighbourhood effects’: Who your neighbours are

?

?

Does it matter which neighbourhood?

2

0 .1 5

0 .2

0 .2 5

0 .3

0 .3 5

0 .4

0 .4 5

0 .5

0 .5 5

0 0 .0 5 0 .1 0 .1 5 0 .2 0 .2 5

A g e 1 3 W a rd , p ro p o rtio n w ith A L e ve ls & h ig h e r

A L

evel

s &

hig

her a

t age

33

Older evidence

Association between neighbourhood quals. (1970s) and child’s adult attainment (1990s), Gibbons (2002)

1971, age -13 ward, proportion of adults with A Levels +

Pro

babi

lity

of A

Lev

els

+ at

age

33

3

Adjust for family background

Smaller effects: ‘Sorting’ matters - higher educated families choose higher educated neighbourhoods

0 .1 5

0 .2

0 .2 5

0 .3

0 .3 5

0 .4

0 .4 5

0 .5

0 .5 5

0 0 .0 5 0 .1 0 .1 5 0 .2 0 .2 5

A g e 1 3 W a rd , p ro p o rtio n w ith A L e ve ls & h ig h e r

A L

evel

s an

d hi

gher

at a

ge 3

3

1971, age -13 ward, proportion of adults with A Levels +

Pro

babi

lity

of A

Lev

els

+ at

age

33

4

0

0 .0 5

0 .1

0 .1 5

0 .2

0 .2 5

0 .3

0 .3 5

0 .4

0 0 .0 5 0 .1 0 .1 5 0 .2

A g e 1 3 W a rd , p ro p o r tio n w ith A L e v e ls & h ig h e r

A L

evel

s an

d hi

gher

at a

ge 3

3

Social tenants in the 1970s

Arguably, social tenants had less choice over housing location: less ‘sorting’

1971, age -13 ward, proportion of adults with A Levels +

Pro

babi

lity

of A

Lev

els

+ at

age

33

5

Parent’s education versus neighbours

Parent’s education matters way more than neighbours’ education in 1970s

6

More recent evidence on neighbours

Ideally we need an experiment e.g. Moving to Opportunity Programme in US, which finds no

systematic impacts on substantive outcomes

No explicit experiments in Britain Alternative research designs using ‘random’ sources of variation

in neighbourhood quality Big administrative data on England– the National Pupil

Database

7

KS1, age 6‐7 KS2, age 8‐11 KS3, age 12‐14 KS4, 15‐16

Primary school Secondary school

KS1 test

KS2 test

KS3 test

GCSEs

Timing of National Curriculum tests

Neighbours change between ks2 and ks3

8

Neighbourhood at ks3Neighbourhood at ks2

?

Effects of changing neighbours on stayers

9

Neighbourhood at ks2Neighbourhood at ks3

?

Effects of changing neighbours on stayers

10

Effect of neighbour changes on own ks3

8.79

0.05 0.00 0.02

-2

0

2

4

6

8

10

ks3,unadjusted

ks3-ks2 value-added

Same school Same schooland neighb

ks1FSMSENMale

Neighbours change from average to top 10% ks1

Neighbours change from 25% to zero FSM

Neighbours change from 20% to zero SEN

Neighbours change from 50% to 70% boys

11

Does moving into social housing matter?

Compare KS3 for movers into social housing, with movers into non-social housing, before and after Key Stage 3 tests

Pupils move to social/ non-social housing between ks2 and ks3

Pupils move to social/non-social housing after ks3

KS1, age 6‐7 KS2, age 8‐11 KS3, age 12‐14 KS4, 15‐16

Primary school Secondary school

KS1 test

KS2 test

KS3 test

GCSEs

12

Move after ks3 33.07 45.42 -12.35

Social n’hood Non-social n’hood Difference

Move before ks3 34.20 45.71 -12.51

Evidence from movers into social housing

Effects of moving to social housing neighbourhoods and non-social housing neighbourhoods on KS3 scores (in percentiles)

‘Dif-in-dif ’

-0.16

13

‘Peer effects’: The ability of your child’s school mates

Pupils move from primary to secondary school – change of peer group

KS1, age 6‐7 KS2, age 8‐11 KS3, age 12‐14 KS4, 15‐16

Primary school Secondary school

KS1 test

KS2 test

KS3 test

GCSEs

New peer groups at secondary school

14

11.63

3.94

0.85 0.910.38

0.0

2.0

4.0

6.0

8.0

10.0

12.0

ks3,unadjusted

ks3-ks2value added

same schoolpair

same schoolpair +

controls

sameindividual,differentsubjects

Effect of peer group ks2 on own ks3

Secondary school peers: average compared to top 10%

Secondary school peer group: fewer poor performing peers

15

‘Mobility’: How fast neighbours and peers come and go

People move, neighbourhood stays the same

Neighbourhood at ks3

?

Neighbourhood at ks2

16

Effect of neighbour mobility, ks2 to ks3

2.23

0.63

0.13 0.060.0

0.5

1.0

1.5

2.0

2.5

ks3,unadjusted

ks3-ks2 value-added

same school same school +neighb

Neighbourhood change from 16% turnover per year to zero

17

Comparisons with other inputs

Comparison with other inputs

Effects of various education related inputs (0-100 scale)

0.02 0.060.91

0.07 0.36

5.63

7.03

0123456789

18

Conclusion and policy implications

Neighbourhood composition irrelevant for a child’s education Turnover seems to matter more, in neighbourhoods and

schools Disruption in teaching and breaking of social ties potentially important

– theory rather than evidence

School peer groups moderately important, though account for only 0.04% of the variance in pupil achievement

No role for engineering neighbourhood mixing to address education performance or inequality E.g. mixed communities planning

We should be more much concerned about school resources and teacher quality

19

Works by SERC researchers referred to Gibbons, S. (2002) Neighbourhood Effects on Educational Achievement: Evidence from the

Census and National Child Development Study, Centre for Economics of Education DP18 Cheshire, P., S. Gibbons and I. Gordon (2008) Policies for 'mixed communities': A critical

evaluation, Spatial Economics Research Centre, Policy Paper 2 Gibbons, S. and S. Telhaj, (2011), Pupil mobility and school disruption, Journal of Public

Economics Gibbons, S., S. McNally and M. Viarengo (2011), Does Additional Spending Help Urban

Schools? An Evaluation Using Boundary Discontinuities, SERC DP0090 Lavy, V., O.Silva, F. Weinhardt (2012), The Good the Bad and the Average: Evidence on

Ability Peer Effects in Schools, Journal of Labor Economics Gibbons, S., O. Silva and F. Weinhardt (2013), Everybody Needs Good Neighbours?,

Economic Journal (revised version of SERC DP0063 Weinhardt, F. (2014) Social Housing, Neighborhood Quality and Student Performance, Journal

of Urban Economics (revised version of SERC DP 0044) Gibbons, S. and S. Telhaj, (2014) Peer Effects: Evidence from Secondary School Transition in

England, forthcoming Oxford Bulletin of Economics and Statistics (revised version of SERC DP0001

Gibbons, S., O.Silva and F. Weinhardt (2014) Neighbourhood Turnover and Teenage Attainment, SERCDP0163 20

Suggested hashtag for Twitter users: #LSEworks

Neighbours, Peers and Educational AchievementLSE Works: Spatial Economics Research Centre public lecture

Professor Stephen GibbonsProfessor of Economic Geography Director, Spatial Economics Research Centre, LSE

Dr Tim LeunigChief Analyst and Chief Scientific Adviser Department for Education

Professor Ruth LuptonProfessor of Education University of Manchester

Professor Julia BlackChair, LSE