Abstract
This paper uses the notion of ‘pathological’ social science, wherein large claims to knowledge are made on the basis of very small differences in the data, to consider the notion of the school mix effect. It describes a variety of plausible alternative explanations for the same sets of findings, including the school mix effect, but also errors in measurement and in the specification of the model. The conclusion is that belief in a school mix effect is difficult to sustain, and that the other explanations are just as probable. In these circumstances, it could be ‘pathological’ to expect much practical progress from work based on the validity of this effect, and the situation is compared to the kinds of vanishing breakthroughs occurring in other fields.
Notes
1. For example, Harker (Citation2004), whose definition of the school mix effect appears earlier, was working with one dataset in which 24% of the schools who had been approached refused to take part.