I do think that a rigorous impact evaluation of a major program such as Reading First is capable of producing valid evidence ... there have been a number of other such big program evaluations at ED and elsewhere Almost all have produced disappointing findings about program effectiveness.  The problem with such studies is that these programs are each funding streams of many different interventions, so evaluations of them average the effect of the few effective interventions with the many ineffective ones.That’s why, (I) generally discourage such whole program evaluations and instead suggest focusing rigorous evaluations on promising interventions (e.g., Reading Recovery), so as to grow the body of research-proven interventions which, if then disseminated, can then improve the overall effectiveness of a big program like Reading First.