Regression Interpretation Intercept I am running a regression on the impact of Gender on unemployment,  however when I control for Test score the intercept increases.  When I interpret it, it means that controlling for test score increases unemployment - which doesn't seem to make sense.  Therefore, I am wondering as to why it would be the case that when you control for a variable the intercept increases?
 A: The intercept increases because you are factoring in the intercept in your model. You interpret effects in the context of the intercept plus your coefficients. 
For example, your gender coefficient is 10, this means that the mean difference between the two is 10. With a model of outcome = gender, and an intercept of 5. That means that the mean for whatever you coded as 1 is 15 and the mean for what you coded as 0 is 5. Now, I add in test score to this. In a perfectly independent world, my coefficient for gender wont change, but my testing coefficient is 1. Now I have scaled test from 0 to 100. The average on the test is 70. So now my intercept might be -50. The intercept went down because you are now putting a predictor on a 0 to 100 scale and the intercept is calculated at 0 for predictors (in this case, even though no one on your test got a 0, your model is still running as though this was the case). 
What you might want to do is center your test to have a bit more reasonable coefficient. 
