Statistician Peter Tymms’s attempt to build a case against the Office for Standards in Education’s primary teacher training inspection methodology, (“Statistics suggest odds are stacked”, TES, December 12) is founded on manifestly unsustainable premises.
His first assertion that if an institution “fails” in any one of the 14 “cells” inspected, the institution “fails overall” is simply not true. Dr Tymms’s own institution, the University of Durham, for example, received a grade profile comprising nine good grades, four adequates and one poor quality (a grade 4 in OFSTED parlance). Nowhere in the report does it say that Durham University “failed” its inspection. Indeed, while OFSTED could hardly ignore the fact that in one aspect the university was failing to comply with the Secretary of State’s requirements, the report highlights a number of strengths and presents a balanced picture of the university’s provision.
Dr Tymms then repeats the canard that providers are “failed” on the strength of one borderline student underperforming. Professors Peter Mortimore and Ted Wragg have both energetically peddled this myth recently - but that does not make it true. OFSTED has not awarded a grade 4 in any cell on the strength of a single unsatisfactory student. Where we have given a grade 4 for poor quality assessment of students, this has been based on checking out all of the provider’s assessments against a sample of 16 students (selected in consultation with the provider) and finding significant discrepancies - always affecting at least two of the students and sometimes more than half of them.
Furthermore, Dr Tymms makes much of the alleged request by OFSTED to see borderline students. We do not. We ask to see students whom the providers judge to be fully deserving of Qualified Teacher Status. True, we ask for a sample representing the range of competence , but we specifically ask institutions not to include any students about whom there were any doubts. All of OFSTED’s grade 4 judgments, therefore, have related to students graded by providers as being at least adequate.
Dr Tymms has thus built his statistical edifice on very shaky foundations. It is hardly surprising that the inferences drawn by him are not only inherently implausible but, wildly inaccurate. The Romans had a phrase for it “res ipsa loquitur” - the facts speak for themselves. If Dr Tymms’s methodology had even a fraction of the robustness of that of OFSTED, one would expect at least some match between his predictions and the actual outcomes of inspections. There is none.
DAVID TAYLOR
Head of teacher education team
Office for Standards in Education
London WC2
