Hi, I have a set of hypotheses about main effects and a set of hypotheses about moderating effects. One of the main effects is insignificant (H rejected). Namely, Capability does not predict Fraud (in my data). However, the moderating hypothesis "Machiavellian personality moderates effect of Capability on Fraud" is significant. According to my calculation, the statistical power in my regression models should be good, therefore my findings reliable. Still, I am not sure how to interpret this situation. So MACH changes (negatively) the slope of CAP=>FRAUD. But there is no reliable slope! Operationalization of CAP and MACH is consistent with the literature; FRAUD is measured through a lab experiment. Any ideas on how to interpret this? Also, I had to run regressions with moderators without the main effects due to huge multicollinearity. So I have two sets of models, models with main effects (just IVs / IVs + controls) and models with moderators (just MODs / MODs + controls).