Widening Participation in Higher Education: A Quantitative

Report
Widening Participation in
Higher Education: A
Quantitative Analysis
Institute of Education
Institute for Fiscal Studies
Centre for Economic Performance
Aims
•
To provide a theoretically based analysis
of HE participation for different types of
student
– the factors influencing the likelihood of
entering higher education
– the quality and nature of the higher
education experienced by different types of
student
– the determinants of and barriers to
progression in HE.
Key Objective
• To identify at what stage in the lifecourse,
and for which groups of students,
interventions to widen participation in
higher education might be best focussed.
Participation in HE at age 18 by Alevel score and parents’ SEG
A-level point score
25+
13 to 24
1 to 12
0
10
20
30
40
50
Per cent
low er
higher
60
70
80
Methods – Integrated data set
• School
• Pupil Level Annual School Census: ALL PUPILS in State
schools in England in Year 11 in 2001/02
• School Attainment records: Key Stages 2, 3, 4 and 5
• Further Education
• Individual Learner Records
• National Information System for Vocational Qualifications
• Higher Education
• UCAS records: all applicants to HE
• HESA records: all attendants in HE
• [Student Loan Book records]
Summary of progress
•
•
•
•
Major task is to collect and merge the
necessary data
We are well advanced in this and are
working closely with the DfES and
HEFCE
Analysis of the main data set is
underway
Some results for the project are available
– Work on teacher expectations
Methods – integrated dataset
• Work ongoing on school and FE data
– Data work
• Construction of post-16 participation variables
• Attainment post-16
• Background characteristics
– Modelling:
• Decision to remain in post-compulsory education
• Attainment post-16
Methods – integrated dataset
• Construction of post-16 participation variables
– School: attendance in Y12 or Y13 PLASC
– FE: presence of ILR record
• Gives crude measure of participation BUT:
– ILR and PLASC have very different structures so
measure is not consistent across different forms of
post-16 participation
– Transition from state to private => counted as nonparticipation!
– Cannot YET distinguish full-time vs. part-time FE
Methods – integrated dataset
• Attainment post-16
– Work in progress…
– Need consistent measure across vocational
and non-vocational qualifications
– DfES constructed data contains derived
measures of government targets (e.g. L2
attainment by 19) but not very detailed scores
Methods – integrated dataset
• Individual background characteristics
– Gender, ethnicity, FSM and EAL status
• Previous attainment
– Results at Key Stages 2 and 3
• School characteristics
– Ethnic, FSM and EAL composition; and
ranking at GCSE level
• EMA availability
Preliminary findings – integrated
dataset
At 16
At 17
% participating
81.2
72.2
- School
27.5
21.1
- FE
38.3
34.9
- WBL
7.1
9.2
- Some combination thereof
8.3
7.0
Sample size
548,749 548,749
These match official DfES post-16 statistics quite closely
Post-16 participation (1)
Participation in post-compulsory education: at 16
Quintile of KS2 distribution
5
4
FSM pupils
3
Non-FSM pupils
2
1
0
0.2
0.4
0.6
Proportion participating
0.8
1
Post-16 participation (2)
Participation in post-compulsory education: at 17
Quintile of KS2 distribution
5
4
FSM pupils
3
Non-FSM pupils
2
1
0
0.2
0.4
0.6
Proportion participating
0.8
1
Ongoing work
• HE records to be added
• Model sequence of decisions
• Sub-group analysis
The role of teacher and pupil
expectations in students' HE
decisions
– Can teachers' perceptions about the student's
ability explain inequalities in HE participation?
– Can students’ perceptions about their own
abilities
explain
inequalities
in
HE
participation?
Steve Gibbons and Arnaud Chevalier
Teacher expectations and pupil
attainment
Motivation and questions
• How important are teacher expectations in influencing
pupils’ education decisions and outcomes?
• How well do teacher expectations reflect actual student
achievement?
• Do expectations differ across demographic groups?
Methods and data: teacher expectations
• Quantitative (regression) approach based on differences
between Teacher Assessment of pupil attainment level at
Ks3, and actual attainment
• Investigates Teacher Assessment at age-14 relative to
what we would expect of pupils given past, current and
future attainment
• PLASC/NPD data: around 1.1.million pupils without special
needs in non-special schools. Cohorts in year 11 in
2001/2, 2002/3 and 2003/4
• Staying-on based on attendance in year 12. No other postcompulsory schooling data yet
Teacher Expectations
• Teachers tend to under estimate the
educational potential of certain groups of
students, across a range of subjects
Teacher Assessment of pupils at age 14 (English). Demographic
groups relative to white British girls, not on free meals.
1
0.8
Teacher assessed Ks3 points
0.6
0.4
0.2
0
Male
-0.2
Free
meals
Other
white
Asian
Black
Chinese
Mixed
Other
EAL
-0.4
-0.6
-0.8
-1
Estimates control for Ks3 level actually
attained; outline-only non-significant
Teacher Assessment of pupils at age 14 (Science). Demographic
groups relative to white British girls, not on free meals.
1
0.8
Teacher assessed Ks3 points
0.6
0.4
0.2
0
Male
-0.2
Free
meals
Other
white
Asian
Black
Chinese
Mixed
Other
EAL
-0.4
-0.6
-0.8
-1
Estimates control for Ks3 level actually
attained; outline-only non-significant
Teacher Assessment of pupils at age 14 (Maths). Demographic
groups relative to white British girls, not on free meals.
1
0.8
Teacher assessed Ks3 points
0.6
0.4
0.2
0
Male
-0.2
Free
meals
Other
white
Asian
Black
Chinese
Mixed
Other
EAL
-0.4
-0.6
-0.8
-1
Estimates control for Ks3 level actually
attained; outline-only non-significant
Teacher Assessment of pupils at age-14 (English). Estimates
conditional on all past and current attainment
0.8
Teacher assessed Ks3 points
.
0.6
0.4
0.2
0
Male
-0.2
-0.4
-0.6
-0.8
Free
meals
Other
white
Asian
Black
Chinese
Mixed
Other
Controls are Ks2 & Ks3 levels and test scores, Teacher
Assessment at Key Stage 2; outline-only non-significant
EAL
Teacher Assessment of pupils at age-14 (Science). Estimates
conditional on all past, current attainment
0.8
Teacher assessed Ks3 points
0.6
0.4
0.2
0
Male
-0.2
-0.4
-0.6
-0.8
Free
meals
Other
white
Asian
Black
Chinese
Mixed
Other
EAL
Controls are Ks2 & Ks3 levels and test scores, Teacher
Assessment at Key Stage 2, GCSE points (all subjects);
outline-only non-significant
Teacher Assessment of pupils at age-14 (Maths). Estimates
conditional on past and current attainment
0.8
Teacher assessed Ks3 points
0.6
Controls are Ks2 & Ks3 levels and test
scores, Teacher Assessment at Key
Stage 2; outline-only non-significant
0.4
0.2
0
Male
-0.2
-0.4
-0.6
-0.8
Free
meals
Other
white
Asian
Black
Chinese
Mixed
Other
EAL
Summary
• For some demographic groups, teachers’ assessment of
age-14 attainment differs systematically from what we
would expect based on based on past assessment, past
test scores and current test scores
• Teacher assessments are systematically below test scorebased assessments:
– For those on free meals
– For boys, in English and Maths
Summary
• Picture more mixed for ethnic groups
– Teacher assessments are systematically above test scorebased assessments for Chinese pupils
• Discrepancies are fairly small:
– e.g. average under-assessment in English for FSM-boys is
only about 0.9 KS3 points (< 1 term’s progress)
Do teacher expectations matter?
• Do teacher expectations influence pupil
decisions and outcomes?
Association between Teacher Assessment at age-14 and age-16
outcomes, conditional on test scores and prior assessment
GCSE
points
percentile
Probability
of staying
on to year
12
Stay on to
year 12,
conditional
on GCSE
points
Teacher Assessment
English (points)
+0.47
+0.37%
+0.10%
Teacher Assessment
Science (points)
+0.32
+0.28%
+0.09%
Teacher Assessment
Maths (points)
+0.38
+0.32%
+0.11%
Estimates control for pupil characteristics, Ks2 and Ks3 levels and test scores, teacher
assessment at Key Stage 2. Last column controls for GCSE points. All significant at
<0.1%
Summary
• Low teacher expectations translate into lower pupil GCSE
point scores and lower staying on rates (at school)
– A pupil assessed 1 level down (6 points) in English Maths
and Science can expect to be 7 percentiles down in GCSE
points
– And around 1.8 percentage points less likely to stay on at
school (on a base of 20% for boys on FSM)
Summary
• Two competing hypotheses:
• 1. Teacher under-assessment at age-14 is rational and
based on information not available to us
– e,g. past teacher assessment, and actual past and current
attainment are worse predictors of the ability of boys on FSM
than of girls not on FSM
– difficult to see why different demographic groups should differ
in this way
• 2. Expectations of attainment of these groups are lower
than they should be, and become self-fulfilling

similar documents