For in vivo data in the main study, although all measurements were recorded from individual animals (dams or pups), individual pup measurements were used to calculate litter means, and litter means were the unit of analysis. Thus the numbers per group are the number of litters (as opposed to number of pups). Significance for treatment effects on organ weights and anogenital distance data were determined using an analysis of covariance with body weight as the covariate.
For testicular histopathology analysis, vacuolization or atrophy of the seminiferous tubule epithelium was scored by Experimental Pathology Laboratories from 1 to 5, where a score of 1 was mildest and a score of 5 was most severe. We tested for treatment-related differences in percentages of affected animals using a chi-square statistic. Group-specific percentages of affected animals were compared using Fisher exact tests. In one analysis, animals receiving scores of 1 or higher were considered affected. In a second analysis, animals receiving scores of 2 or higher were considered affected.