**Multi-Collinearity**

When two independent variables are correlated with each other, it is difficult to estimate their individual effects, controlling for each other. Usually you just have to accept this (e.g. it does not matter too much if the variables are in the equation as possible cofounders). If it is crucial:

Use tabulation | |

Split one variable into two groups (e.g. good/ bad) and run the regression for the other variable within these new categories |

Make an index, e.g. add together similar variables using factor analysis, if this is familiar (not recommended practice) |