This article has two chief purposes. It presents a substantive reappraisal of a decade of school choice research in the UK. This reappraisal is used as a case study illustrating the elasticity of the notion of social science "evidence", when wielded by academics in an area where strong ideological preconceptions struggle with the lack of a sound quantitative tradition of research. The focus here is on the changing socio-economic compositions of schools in an era of choice. A prediction from theory and from small-scale studies had been that schools in England and Wales would become more segregated in terms of indicators of socio-economic disadvantage after the Education Reform Act 1988. The first large-scale study of the actual compositions of schools suggested that this did not happen. This study was then subjected by a majority of UK academics in the field to a level of criticism that was not applied by them to subsequent, but seemingly inferior, studies that reached an opposite conclusion. The criticism involved widespread misquotation and misunderstanding that was not picked up by "peer" review. What, therefore, counts as evidence in the school choice debate? (Contains 4 notes.)