Cambridge University Letter on AS Levels

Embed Size (px)

Citation preview

  • 7/28/2019 Cambridge University Letter on AS Levels

    1/3

    Dr Mike Sewell

    Director of Admissions for the Cambridge Colleges

    31 May 2013

    Kevin Brennan MP

    House of Commons

    London

    SW1A 0AA

    Dear Mr Brennan,

    As promised, we should like to offer some further reflection on the analysis put forward to you byDavid Laws MP in his letter of 14 May.

    An initial analysis of the DfE study suggests that this could be a case of policy-based evidence.No attempt appears to have been made to recognise the effect of the variables at any level ofdetail (as far as can be determined given the lack of information about the base data employed),and that in turn suggests that the original proposition being explored was either poorly consideredor loaded towards a particular result or, more charitably, that the raw data the researchers hadto work with was simply not up to the job.

    In our judgment the study has three fundamental design flaws:-

    - The authors choose two oversimplified measures of GCSE and AS attainment, andcompare them to a crude measure of degree success. Selective HEIs do not admit on a chanceof getting a II.1. They seek to admit the students who are likeliest to perform really well whichmeans a good II.1 (65%+) or First.

    - Extrapolating from logistic regression in the manner in which this study does is highlyproblematic, since the binary outcome treats the lowest II.1 as equivalent to the highest First, andsimply asks whether AS and KS4 performance can correctly predict which side of the II.2/II.1divide a student will end up on. A linear regression model incorporating the full range of degreeresults (including failures) allows for finer distinctions between student performances, and sowould have been much more appropriate to the question of whether AS Levels provide universityadmissions officers with useful additional information about student potential, over and aboveGCSE.

    - The model fails to adjust for the fact that, at any given institution, students will tend to have

    roughly similar levels of prior attainment. This is because students in large part select HEIs basedon their own level of academic attainment. Since, as the study admits, a II.1 is a relative measureof performance within an institutions cohort rather than a nationally calibrated standard, lumpingall universities together, as this analysis does, is destined significantly to weaken correlations. Thisis because, while students obtaining II.1s at a highly competitive-for-entry university might averagegrades AAAA at AS Level, students obtaining II.1s at a less competitive-for-entry university mightonly average grades BBBB. A serious and rigorous analysis would have run multiple separateregression models to analyse the relationships within each institution.

    Cambridge Admissions Office

    Fitzwilliam House

    32 Trumpington Street

    Cambridge CB2 1QY

    Tel: +44 (0) 1223 333308

    Fax: +44 (0) 1223 746868

    Email: [email protected]

    Cambridge Admissions Office

  • 7/28/2019 Cambridge University Letter on AS Levels

    2/3

    Page 2 of 3

    The reporting of the studys findings is incomplete in a number of respects. In particular the reportprovides very little information about the factors considered when securing base data andconducting the analysis.

    A complete report should, for example, provide information on how the analysts counted GCSEresults and AS results: which grades did they include, A*-C, or A*-F? Did they exclude subjectslike General Studies and Critical Thinking which are not accepted by most selective universities?Did they include short courses, and did they cap the total number they counted? If they used apoints system for each grade, what points did they allocate to each grade?

    Similarly for AS level, the key information should include whether capping was applied and thepoints allocated per grade. Finally, does this analysis include degree failure or drop-out?

    A good sense of what the data being analysed actually is and how it is being used is essential to asound statistical report. We therefore requested the missing information, and the data-set, fromthe DfE in order to carry out our own modelling.

    The DfEs reply confirmed that the analysis is based on uncapped average scores in both cases,that KS4 qualifications are being used rather than GCSE in isolation, and that no subjects havebeen excluded.

    The University of Cambridge uses an average score calculated across the best three, or threemost relevant, AS UMS scores (depending on whether the application is for arts or sciences), andexcludes General Studies and Critical Thinking.

    While we await a decision on our request for the full data-set, we have run a similar logisticregression model to that to which Mr Laws refers using our own data, and using the yardstick of agood II.1/First, which around 50% of our students achieve.

    In this case we see a clear effect when adding UMS to GCSE: the predictive validity of the modelis improved and the GCSE coefficient is rendered insignificant. That is, if we have UMS, GCSEdoes not add value, but if we already have GCSE, UMS adds value over GCSE while alsoeradicating any value GCSE had.

    So far as Mr Laws argument that GCSE and AS tell us the same thing goes, we can show thatUMS tells us everything that GCSE does, and much more besides.

    Our position remains that AS UMS is key to accurate and fair university admissions. If otheruniversities also adopted UMS as a key selection-tool, we believe that they too would find that itwas much more valuable than GCSE.

    We hope that this is helpful.

    As ever, if you would like to discuss the issues in more detail, please dont hesitate to get in touch.

    Yours sincerely,

    Jon Beard Dr Mike Sewell

    Director of Undergraduate Director of AdmissionsRecruitment for the Cambridge Colleges

    cc:

    David Willetts MP, Minister of State for Universities and Science, Department for Business,

  • 7/28/2019 Cambridge University Letter on AS Levels

    3/3

    Page 3 of 3

    Innovation & Skills, 1 Victoria Street, London SW1H 0ET

    Michael Gove MP, Secretary of State for Education, Department for Education, Sanctuary

    Buildings, Great Smith Street, London SW1P 3BT

    David Laws MP, Minister of State for Schools, Department for Education, Sanctuary Buildings,Great Smith Street, London SW1P 3BT

    Dr Julian Huppert MP, 16 Signet Court, Swann Road, Cambridge CB5 8LA

    Dr Wendy Piatt, Director General, The Russell Group, 1 Northumberland Avenue, TrafalgarSquare, London WC2N 5BW

    Nicola Dandridge, Chief Executive, Universities UK, Woburn House, 20 Tavistock Square, LondonWC1H 9HQ

    Mary Curnock Cook, Chief Executive, UCAS, Rosehill, New Barn Lane, Cheltenham GL52 3LZ

    Janet Graham, Director, SPA, Rosehill, New Barn Lane, Cheltenham GL52 3LZ