The general linear model made many more assumptions other than the normality of the error term. One of the other assumptions is that there is no multicollinearity among the regressors in the model. Multicollinearity can be detected by several methods.