That said, Feinstein and Symons (1999) Attainment in secondary schools Oxford Economic Papers 51:300-321 did find better English test scores for schools that stream.

And I understand (but haven’t checked this) positive effects for setting were found by:

Kulik, James A.; Chen-Lin C. Kulik (1992). “Meta-analytic findings on grouping programs”. Gifted Children Quarterly 36 (2): 73-77

Argys, L. M.; Rees, D. I., & Brewer, D. J. (1996). “Detracking America’s Schools: Equity at Zero Cost?”. Journal of Policy Analysis and Management 15 (4): 623-645.

However, what has discredited mixed ability teaching is actually at a far more fundamental level. Hattie (2003) found that the things that made the most difference to learning that a teacher controlled were:

Providing feedback.
Quality of instruction.
Direct instruction.

Both experience of teaching, and a quick look at the most commonly promoted methods of mixed ability teaching suggests that these are fair better provided in to a class which has a limited range of abilities.

I’d be keen to know what research proved mixed ability was best. There has been what seems to be a lot of publications done by a researcher called (I believe) Jo Boaler based on comparing two awful schools and discovering that an awful school with mixed ability teaching in maths is only slightly worse than an awful school with setting, and that this can be reversed if you ignore exam results and make up your own tests. Despite the obvious inadequacies of research done in this way trainee teachers are still given this stuff as if it wasn’t embarrassingly bad nonsense.