ABSTRACT
Results on the relationship between intervention and outcome in this sample were disquieting to speech and language therapists: the children who had the most intervention had the worst outcomes. If, as appears to have been the case in the Bishop and Edmundson study in the 1980s, therapists work most with the more serious cases, then there is a confound between severity of language problems and receipt of therapy, and this can make intervention appear ineffective. This gives a very clear illustration of the perils of observational studies. The basic problem is the same: treated and untreated groups are not comparable, and so comparing them will give misleading results. There are just too many potential confounds that could cause bias. Here again, though, the comparison has the potential to mislead. But there would still be concerns about possible differences between the therapists and their practices that might be influencing results.
