Cooper's analysis is incorrect

Abstract

In his papers, Richard Cooper finds positive associations between health care quality and both specialist and generalist physicians, but he misinterprets his results. Instead of undermining the findings of our study, which found higher quality in areas with more generalists relative to specialists, his results bolster ours: they suggest that the effect of generalists on quality is ten times larger than that of specialists. Furthermore, his rejection of multiple regression in favor of exclusive reliance on isolated correlations precludes him from gauging the relative contributions of specialists, generalists, and other factors. Unfortunately, these deficiencies mean that we can learn little from Cooper's analyses.