HEPI’s last foray into the debate on academic selection suggested grammar schools are successful in helping their poorer pupils reach highly-selective universities.
In this response, a diverse set of voices use the latest evidence to challenge the idea that grammar school systems serve pupils better than comprehensive schools.
This collection of essays also asks what lessons comprehensive schools might offer for the UK’s highly selective university system.
These papers point out a number of weaknesses in Mansfield’s analysis. For example, that the comparisons take ‘no account of the differences between selective and non-selective areas’ (page 19). However, even if we did make the ‘heroic assumption’ that areas choosing to keep grammar schools do not differ in other respects, we would still have a difficulty. Areas do not form neat, self-contained selective systems. “Of the 150 Local Authorities in England only 36 actually have grammar schools of their own, but 80 Local Authorities have more than 1% of the pupils who live in their area attending grammar schools” (Sutton Trust, 2008, ‘Social selectivity of state schools and the impact of grammars’.) And even for the Local Authorities which have grammar schools, not all have secondary moderns and a full selective system. This means that the comparison table (page 19 of current paper) does NOT show figures for non-selection and selection. The figures are almost certainly biased against ‘comprehensives’ and bias for ‘grammar schools and secondary moderns taken together’. All pupils at comprehensives are counted as non-selective, even if the comprehensives lose potential pupils to grammar schools in their or other Local Authorities. This weakness in Mansfield’s analysis is not highlighted.
I have raised this issue with Mansfield and he accepts that the comparison he used would favour ‘grammar schools and secondary moderns taken together’. He believes solving the ‘leakage issue’ would reduce the size of the apparent selection advantage but still leave it ‘both large and significant’. I’m not so sure. Access to data and more sophisticated analysis could provide the answer.