To manage the risk of an experiment-wise type 1 error, we used a modified Bonferroni approach (Judd&McClelland, 1989), inwhich the strongest results of the total number of regression analyses were compared to the selected alpha level of .05, repeating the process for each test.