Sex as a possible source of group inequivalence in Lovaas (1987)

The possibility of differential sex proportions as a confounding factor in the Lovaas (1987) study is raised in this paper. It is argued that the chi-square analysis reported in the original study was inadequate and that the appropriate comparison should be made not between the experimental group and primary control group (Control Group 1) utilizing expected cell frequencies estimated from sample data, but between these two groups using population data to estimate expected cell frequencies. Implications of this interpretation are discussed.