You are in :

July 2003/32
Issues paper
This report is for information


Schooling effects on higher education achievement

This report describes research to determine whether the characteristics of an applicant's school or college can be used in an assessment of his or her potential in higher education.

Executive summary (read on-line)


[ MS Word 228K | Zipped Word 91K | Adobe PDF 144K | Zipped PDF 129K ]

Description of annexes (read on-line)



1.    This report describes research to determine whether the characteristics of an applicant's school or college can be used in an assessment of his or her potential in higher education (HE).

Key points

2.    At the 'Fair Enough' conference on admissions to higher education sponsored by Universities UK in January 2003, it was recommended that school performance data be used to identify applicants with relative educational disadvantage, and, in conjunction with other information, to decide whether some applicants should be made lower offers. The justification for making such differential offers is based on the belief that students from poorly performing schools will achieve better results within higher education than students from better performing schools, all other things being equal.

3.    This document reports the findings from our attempts to assess such 'school performance' effects along with the effects of 'school type', such as whether the school is independent or the responsibility of the Local Education Authority (LEA). To reduce the complexity of the analysis we have restricted it to 18 year-old entrants with A-level qualifications to degree courses in 1997-98. These entrants were followed through to 2001-02 and their HE achievement was assessed by whether they had discontinued their studies, whether they had gained a qualification and, if they graduated, the class of degree they obtained. We found that:

  1. A-level grades, as summarised by A-level points, were the single most important factor in determining the expected HE achievement. However, other factors - like the gender of the student, characteristics of the school and the university, subject studied - were also associated with HE achievement, in a rather complex way.
  2. The effect of school performance was inconsistent. That is, under certain conditions, students from poorly performing schools are likely to do less well in HE than similar students from better performing schools. Whether a student from the poorly performing school does better or worse can depend on A-level points, the gender of the student, the subject of study in HE, and the measure of HE achievement used.
  3. Students from independent schools appear to consistently do less well than students from other schools and colleges, when compared on a like-for-like basis. For all but those students attending the most highly selective institutions, we would expect a student from an independent school to achieve as well as a student from an LEA school who has between one and four fewer A-level points.
  4. For the most highly selective higher education institutions the effect of having been to a further education college or grant maintained school is unclear, though we still find that students from LEA schools do consistently better than similar students from independent schools. However, at these most selective institutions, it is not possible to measure this difference in HE achievement in terms of equivalent A-level points, because the majority of the students have 30 A-level points, the maximum recorded value.

4.    We have not attempted to establish the cause of the lower expected HE achievement for students from independent schools, and have simply discussed the explanations that others have offered. We think it unlikely that there will ever be clear evidence to answer such questions, and we leave it to the judgements of those deciding whether and how to use the evidence we have presented.

5.    Aspects of our analysis could usefully be taken further including: extending the analysis to other cohorts, looking at other attributes of schools and exploring whether schooling effects related to a student's school at age 16 are important.

6.    More generally, our experience in assessing schooling effects suggests that assessments of new ways of selecting applicants to HE are unlikely to be conclusive if based on small-scale studies. The models presented here provide a possible framework for systematic studies to find out what works and what does not in identifying potential.

Action required

7.    No action is required in response to this document.


8.    Comments or questions about this study should be sent to: Mark Gittoes, tel 0117 931 7052, e-mail

Description of annexes

Annex A - Definitions and data sources
Explains what cohort of students is being studied and the data definitions for that cohort.

Annex B - Tabulations of HE achievement and main report tables/figures
Contains tabulations of the five measures of student achievement split in a number of ways. It also contains the figures (and related data), and tables from the main report.

Annex C - Modelling of HE achievement
Describes the statistical models used in the main report and the methods used to examine the results of these models. The main spreadsheet in Annex C contains data on the statistical models and the results derived from them. It contains two types of worksheet:

  1. The worksheets named using the notation C(x) are either figures or tables that appear in the Word document part of Annex C.
  2. Those named using the notation R(x) are referenced in the Word document but appear only in the spreadsheet.

The additional spreadsheet shows a contour plot of the estimated higher education outcomes for different types of students (see paragraphs 28-29 of Annex C).