The effect of treatment across all cows was assessed by fitting treatment in univariable models. Hypothesized interactions between treatment and potential interacting factors were examined using separate models for each risk factor. Continuous variables were assessed for linear relationships with the logit of conceived to AI by categorizing into equally spaced groups and assessing plots of logits versus median value for the risk factor for each group and by assessing plots of predicted logits after fitting fractional polynomials. Variables that did not display linearity in the logit were converted into categorical variables. For each potential risk factor, the main effects of both treatment and the factor were fitted along with the interaction term(s). Likelihood ratio test P-values for the interaction term (or for risk factors with more than 2 categories, the joint likelihood ratio test P-value for all interaction terms) were used. The 95% confidence intervals and Wald P-values were reported for the estimated effects of treatment.