0
$\begingroup$

Possible Duplicate:
How can adding a 2nd IV make the 1st IV significant?

I am looking at the prediction of relapse by gender. In a hierarchical log regression, with gender entered in first block, it is significant. When anxiety is entered in second block, both anxiety and gender are insignificant. Interaction terms are also insignificant. What does this mean? It is clearly not a moderator or mediator effect because anxiety does not predict relapse (even when entered before gender). Is this simply an effect of multicollinearity between gender and anxiety? Because gender is more strongly related to anxiety than to relapse?

$\endgroup$
0
0
$\begingroup$

Yes I think it must be because gender and anxiety are correlated. Probably both have a small but statistically significant effect when entered separately but when together in the model the individual effect of either one is not significant. the tests don't measure their combined effect.

$\endgroup$
1
  • $\begingroup$ Are gender and anxiety are correlated? What happens if you do the analysis separately by gender? What size N do you have? Are plots of the data helpful? $\endgroup$ – Joel W. May 21 '12 at 14:40

Not the answer you're looking for? Browse other questions tagged or ask your own question.