In drawing entirely different conclusions from the same bank of data, Conroy saw consistent evidence favoring the alternative CLAs over the small ICFs in Pennsylvania while Crinella et al. found a large regression artifact produced by matching the ICF and CLA residents on adaptive behavior scores. Ten areas of flawed logic, analysis, and interpretation have undermined both sides of the argument. Conroy's matching—which might not have been necessary in the first place-does not pose a serious threat of regression artifacts. Because of two subtle but fatal errors, the Monte Carlo experiment of Crinella et al. provides no important evidence against Conroy's findings. Other problems, however, require a small reduction in his pro-CLA results, along with a substantial limit on their generalizability.

This content is only available as a PDF.