What's the right stats test? Could anyone tell me what would be a suitable stats test to look at relationships between 4 continuous IVs and 1 continuous DV between 2 groups each of which has 2 levels i.e. male - female and internet users - non internet users?
 A: There are several approaches, depending on what you hope to understand and what you expect out of your continuous IVs. One approach would be a split-plot or mixed ANOVA where each case yields a single point (mean, for instance) for the DV, each IV, and a categorical label for male/female and user/not user. You’ll then get main effects and interactions for each combination of 6 IVs (4 continuous, 2 nominal) on the DV. 
A: Among the other possibilities that @Heitz alludes to are:
a) "Regular" regression with dummy variables and interactions between the dummies and the continuous variables. (I believe this is mathematically equivalent to the ANOVA he recommends, but the output will look different). This produces one regression equation, but interpretation is a little tricky if you are not use to looking at interactions.
b) Stratifying on the categorical variables. This would produce four different regressions, which is simpler to look at that than 1) but it won't produce statistics for the interactions so you won't know things like p values and effect sizes and standard errors for those (or for the main effects of the categorical variables themselves).
c) Regression trees and their offshoots (forests etc.) 
