Surprise: Why is my factor not significant? I am analysing the performance of a large population in exams using a binary (pass / fail) outcome. In the majority of exams, women out-perform men. However, when gender is included with other factors in multiple logistic regression, it is not a significant predictor of performance.
I have some conceptual understanding of why this is but want to know the best way of explaining this to seniors. It goes against general understanding that gender is not an influential factor. Happy to expand following questions.
 A: There is something else about women that means that they outperform men, rather than just being a woman. 
Imagine running a logit model with just gender as an explanatory variable. The results suggest being female is a strong predictor of exam success. However, there must be lots of things that are correlated with being a woman and also predict exam success. For example, time spent studying; in a world where men are too busy playing video games. 
When running a multivariate logit that includes these omitted variables, it becomes clear that it is the other stuff about being a woman that explains exam success, rather than being a woman per se. 
A: Remember the interpretation of a coefficient in a multivariate model. It is the effect of a 1-unit increase in it's corresponding independent variable, $X_i$, holding all other variables constant. If a certain coefficient is significant in a univariate regression setting, but not in a multivariate, that means that the data suggests that some of the other variables are what's really driving whether or not you pass an exam and gender is just a consequence of some other variable that's the real driver.
Another explanation is that gender is related to "exam performance" (as you say) but not whether or not you pass. Imagine that half the woman in your sample fail with a high D and the other half score a high A. Half the men fail with a low D and the other half score a low A - then gender is related to exam performance but not whether or not you pass.
