university access for socio-economically disadvantaged children: a  comparison across anglophone...

Post on 25-Feb-2016

41 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

DESCRIPTION

University access for socio-economically disadvantaged children: A  comparison across Anglophone countries. John Jerrim Anna Vignoles Ross Finnie. Background. Social mobility has emerged as one of the key academic and political topics over the past decade - PowerPoint PPT Presentation

TRANSCRIPT

University access for socio-economically disadvantaged children: A comparison across Anglophone countries

John JerrimAnna Vignoles

Ross Finnie

1

Background• Social mobility has emerged as one of the key academic and political

topics over the past decade

• Large economic literature on the link between fathers’ and sons’ incomes.

• There has been a suggestion by economists that this association is stronger in the US/UK than other countries

• ……..though some sociologists would disagree (Erickson and Goldthorpe 1992, Breen 2004)

• Recent evidence suggests that, despite similar levels of income inequality, US and UK less socially mobile than Canada and Australia

2

DKNOFI

SE

JP

DE

CA

AUNZ

ESFR

PKCH

UKUS

SG

CN

AR

PE

CL

BR

0.2

0.4

0.6

0.8

Inco

me

elas

ticity

20.0 30.0 40.0 50.0 60.0Inequality (Gini)

Inequality versus intergenerational income mobility

Anglophone countries (broadly) similar in terms of income inequality…..

….. but intergenerational income elasticity bigger in UK/ US than Australia or Canada

3

Background

• Small, but growing, literature attempting to explain why US/UK less socially mobile then Canada / Australia

• Bradbury et al (2012), Blanden et al (2012), Haveman et al (2012) some recent examples. Focus on the early years.

• We attempt to add to this debate by investigating how the link between family background and access to university varies across these four countries.

4

Why is university access important for social mobility?

• University is one of the main mechanisms by which skills developed in school converted into valued labour market qualifications

• Large economic returns to university (particularly UK/US)

• Large socio-economic differences in access

• Hence university considered to be a key driver of intergenerational persistence

5

Why are there SES differences in access to university?

6

Two broad schools of thought.....

1. Childhood inputs• In-uteri experiences• Parent-child interactions• Child care• Heredity• Pre-school• Schools

Influence university entry via impact upon school

achievement7

2. Towards point of decision• Credit constraints• Relative risk aversion• Lack of information• Peer influences • Educational aspirations

Influence university entry

over and above school achievement

A model of intergenerational persistence(adapted from Haveman and Wolfe 1995)

Family backgroundTime inputsGoods inputs

Heredity

university entry

Child’s teenage skills

STAGE 1 (Early investments)

university graduation

Labour market outcomes

STAGE 3 (Labour market entry)STAGE 2 (Point of decision)

NOTE: Key role of prior achievement in determining university entry (and thus social mobility)

8

Why might SES gaps in university access differ across countries?

9

1. Differences in childhood inputs• Schooling systems• Child care support • Teenage pregnancy / age of parenthood• Pre-school support / child care• Healthcare • Maternity leave• School segregation

Leads to cross-national variation in school achievement towards end of secondary school (+ size of SES gaps)

10

Socio-economic gaps in prior achievement (PISA test scores)

Luxemburg

Portugal

Iceland

Scotland

Austria

Netherlands

Spain

Canada

Korea

Sweden

Finland

Italy

Greece

Australia

Belgium

New Zealand

Switzerland

Turkey

Ireland

France

Norway

Germany

Northern Ire-land

Denmark

Slovakia

Poland

England

USA

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Standard deviations change in children's maths score with a 4 year increase in highest parental educa-tion

11

2. Differences in structure of higher education systems• Tuition fee levels• Private sector provision• Financial support (e.g. Loans and scholarships)• Provision of information about university• Distance to education institutions• School – to – university pathways (e.g. decision points)• Entry criteria (e.g. school grades, SAT tests)• Length of degree courses

Lead to SES differences in university access over and above role of secondary school achievement

12

United Kingdom (before September 2012)Fees: Maximum ≈ $4,800 per year (no variation by institution).

Typical tuition fee cost of 3 year degree ≈ $14,000.

Finance: Income contingent loans + grants for poor students.

Pathways: TWO school leaving decisions (age 16 AND age 18). Study 3/4 subjects of choice between 16 and 18.

Decisions: Both subject and institution at age 18.

Entry criteria: Based upon predicted school grades at age 18.13

HE systems – key features

United StatesFees: Large variation by institution (Issue “net” vs “sticker” price).

Typical tuition fee cost of 4-year degree ≈ $47,000.

Finance: Mortgage-style loans (no income contingency). Also grants / scholarship programmes.

Pathways: Single school leaving decision (graduate high school).

Decisions: Institution only at age 18. Two-tier HE system (Community vs 4-year college).

Entry criteria: GPA , SAT / ACT scores, Carnegie units 14

HE systems – key features

AustraliaFees: Large variation by subject (little by institution). E.g. $4,000 for sciences. $9,000 for Law. Typical cost 4- year degree ≈ $18,000

Finance: Income contingent loans + grants for poor students

Pathways: Single school leaving decision (graduate high school).

Entry criteria: Tertiary entry rank (a scaling of school grades)

15

HE systems – key features

Implications:

1. There is clear SES inequality in academic achievement, and the extent of this inequality varies across countries.

2. Higher education systems (e.g. Fees, finance, entry criteria etc) also differ dramatically across countries.

Hence university access for socio-economically disadvantaged children could vary across countries because

of both of the above.

16

Research questions

17

Research questions1. Is the SES gap in university participation greater in England and the

United States than in Canada and Australia?

2. Do disadvantaged children in England/US have particularly low chances of entering a “selective” university?

3. To what extent can each of the above be explained by differences in children’s academic achievement at age 15?

4. Do schools play a role in explaining SES inequality in university access (beyond their role in developing young people’s academic ability)?

5. Does any SES difference in university access remain once school grades (at age 18) have been controlled?

18

Model

19

Model

20

Where:Π (E) = Probability of the child going to university

F = Socio – economic status (measured by parental education level)

A = Children’s academic achievement as teenagers

C = A vector of basic control variables (dummy variables for gender and language)

G = School grades at age 18

µ = School level fixed effect

K = Country K

Estimated via linear probability model (response = 1 if entered university , 0 otherwise)

Model Specification

21

Four specificationsSpecification 1 = No control for prior achievement (γ, δ, µ constrained to equal 0). Raw socio-economic gap in university access.

Specification 2 = Prior achievement controlled. Socio-economic gap in university access, conditional upon teenage skills.

Specification 3 = Also include a school fixed effect.

Specification 4 = Also include school grades at age 18

Each specification estimated for each of the four countries

• How do the SES parameters (β’s) change once controls included?

Data

22

Datasets

• UK = LSYPE 2004 (n ≈ 8,000)• US = ELS 2002 (n ≈ 13,000)• Australia = LSAY 2003 (n ≈ 6,500)• Canada = YITS 2000 (n ≈ 11,000)

Sample and design• Nationally representative• Similar sample designs and response rates• Longitudinal follow-up of 15/16 year old children into

university (age 20)• Sample sizes several thousand in each study

23

Measures – Family background

• Family background = Highest parental education level.

24

  Australia England United States Canada

Low 13 12 7 9

Medium 47 68 56 63

High 40 20 38 27

Low = Below ISCED 3 = Below high school

Medium = ISCED 3 – 5B = High school to associates degree

High = ISCED 5A / 6 = Bachelor ‘s degree or higher

Measures• Academic achievement = PISA math and reading test

scores at age 15 (NOTE: proxies for England)……

• university access = Enrolled on a bachelor's degree course by age 20 (refers to a 4 year university degree in the US).

25

  Australia England United States Canada

Average PISA math 523 494 491 533

Average PISA reading 524 493 525 534

% Attending university 39   37  45  43

“Selective” universities

• All university degrees not necessarily of equal value.

• Some evidence that returns vary by institutional quality. Maybe higher returns at “elite” universities (e.g. Oxford, Yale, Melbourne).

• Probability of graduation also tends to be higher.

• Socio-economically disadvantaged children may have particular difficulties accessing these institutions (e.g. higher costs, less geographically accessible etc).

• Hence want to consider “elite” university access in paper26

Definition of “Elite”

• UK = “Russell Group” universities. - Self-selected alliance of the top 20 research universities- 10% of the population attend• Australia = “Group of 8” universities- Self-selected alliance of the top 8 research universities- 12% of the population attend• Canada = “U15” universities- Self-selected alliance of the top 15 research universities• US = “Highly selective” (Carnegie classification)- Based upon entry test scores- 13% of the population attend

27

Results

Access to a bachelor’s (“four year”) degree

28

Raw SES gap (specification 1)

29

SES gap in university access notably smaller in Australia than other countries

Top – Middle gap greater than the Bottom – Middle gapin all countries (though particularly Australia)

Canada England United States

Australia

-1.5

-1

-0.5

0

0.5

1

1.5

Diff

eren

ce (l

og -

odds

)

PISA controls included (specification 2)

30

Bottom – Middle gap now small in each of the countries.England / US now very similar to Australia.

Middle –Top gap has been reduced, but still quite large.

Australia still significantly different to England / US

Canada England United States

Australia

-1.5

-1

-0.5

0

0.5

1

1.5

Diff

eren

ce (l

og -

odds

)

School fixed effect (specification 3)

31

Little change in results once school FE is included.

School-level factors play a relatively modest role in explaining SES inequality in university access (over and above influence on cognitive skills)

Canada England United States

Australia

-1.5

-1

-0.5

0

0.5

1

1.5

Diff

eren

ce (l

og -

odds

)

School grades at age 18 (specification 4)

32

Difference between low and middle SES now statistically insignificant

Remains a sizeable and statistically significant difference between most advantaged and other groups

Canada England United States

Australia

-1.5

-1

-0.5

0

0.5

1

1.5

Diff

eren

ce (l

og -

odds

)

Access to selective HE institutionsConditional upon university attendance

Raw SES gap (Unconditional on uni entry)

34

Canada England United States

Australia

-2

-1

0

1

2

Diff

eren

ce (l

og -

odds

)

Raw SES gap (Conditional on uni entry)

35

Little difference between low and middle SES groups in access to elite institutions

Canada England United States Australia

-2

-1

0

1

2

Diff

eren

ce (l

og -

odds

)

Conditional (PISA test scores controlled)

36

Canada England United States

Australia

-2

-1

0

1

2D

iffer

ence

(log

- od

ds)

Conditional (School grades controlled)

37

Canada England United States Australia

-2

-1

0

1

2

Diff

eren

ce (l

og -

odds

)

Conclusions• Large SES gaps in university access in each country, though

particularly in England and US

• Roughly half the SES gap can be explained by academic skill at age 15 (as measured by PISA test scores)

• School level factors explain a rather modest amount of the SES gap in each of the three countries

• Almost no difference in university access between low and middle SES groups once academic achievement has been controlled (based upon the definitions used)

• High SES children still much more likely to enter university (including selective institutions) even once academic achievement taken into account 38

Policy recommendations

• Reducing SES gap in school achievement important for reducing SES gap in university access.

• Particularly important to raise low SES children’s skills

to close gap with middle class peers

• Interventions at the point of entry should not focus solely upon the most disadvantaged children – there is a large SES gap (conditional upon achievement) between high SES children and everyone else

39

Questions??

40

Appendix – proxy PISA scores for England

41

My predictions vs actual PISA 2006

42

0.0

01.0

02.0

03.0

04D

ensi

ty

0 200 400 600 800 1000PISA points

Actual PISA 2006 Predicted score 1Predicted score 2

Males (Predictions vs actual 2006)

43

0.0

01.0

02.0

03.0

04D

ensi

ty

0 200 400 600 800 1000PISA points

Actual PISA 2006 Predicted score 1Predicted score 2

44

Females (Predictions vs actual 2006)0

.001

.002

.003

.004

Den

sity

0 200 400 600 800 1000PISA points

Actual PISA 2006 Predicted score 1Predicted score 2

top related