How Do You Know You Don Have A Multicollinearity Problem In A Multiple Regression Model?
Multicollinearity can be a problem for many empirical researchers, and this problem will influence how you are able to interpret the coefficients in your model. I will show you how to check for this problem using Stata! There are similar procedures to check for this problem in programs like SPSS or SAS. Run your regression in Stata. I regressed age on income and marital status in the 1940 census and got what is in the picture. The command was reg age marst incwage Directly after that command type VIF. This will show the Variance Inflation Factors for yor variables. 1/VIF is called the tolerance. As long as your average VIF is under 5 or your tolerance is above .1, you should be fine. In this example, the VIF values are quite low, so there is probably not a multicollinearity problem here. When you are running regressions make sure that you *always* check for multicollinearity! Making a simple check for this issue can save you a lot of trouble in the analysis.
