As lp251 says, the determinant method is not a good idea. Checking the condition number of the covariance matrix of the LS estimates (X'X)^{-1} is the way R do

As lp251 says, the determinant method is not a good idea. Checking the condition number of the covariance matrix of the LS estimates (X'X)^{-1} is the way R does it.