Assume a logisitc regression model includes a continuous variable like age and a categorical variable like gender and an interaction...

Assume a logisitc regression model includes a continuous variable like age and a categorical variable like gender and an interaction term for these. If the P value for any of the main effects helpful in determining whether (a) the interaction is significant, (b) whether the main effect should be included in the model? (c) Would your answers to (a) or (b) change if the categorical variable were RACE with three categories?