Summary of binomial GLM with logit link function.

The response variable is estimated number of males killed in all spots (successes) relative to the number released but not killed (failures), and predictors were application density (230, 110, or 50 spots per km2) and state (California or Florida). Estimates and significance values for each treatment and for state are relative to 110 spots per km2 and California. Null deviance was 18281 on 11 df and residual deviance was 14254 on 8 df. A post hoc test using least-squares means showed that all pairwise comparisons of treatment and state combinations were statistically significantly different from each other except for within treatment comparisons between states (e.g., California 50 was not different from Florida 50; the same was true for the other two treatment levels).