Multicollinearity occurs when two or more predictor variables in a regression model are redundant. It is a real problem, and it can do terrible things to your results. But it is uncommon, and is often misdiagnosed.