I'm looking to conduct regression analysis using three dummy variables for independent variables: (1) ethnicity, (2) GLOBE category (i.e. country of origin), and (3) international/domestic student status.

I realize that when using dummy variables, one control variable is excluded. However, can someone explain how you know whether the excluded variable is significant? For example, if the 'white' ethnicity category is excluded, how do you know whether or not being a white participant predicts behavior?

Secondly, as many of my variables risk auto-correlation due to thematic similarities, should I consider stepwise regression?