R Programming By Example

Checking no collinearity with correlations

To check no collinearity, we could use a number of different techniques. For example, for those familiar with linear algebra, the condition number is a measure of how singular a matrix is, where singularity would imply perfect collinearity among the covariates. This number could provide a measure of this collinearity. Another technique is to use the Variance Inflation Factor, which is a more formal technique that provides a measure of how much a regression's variance is increased because of collinearity. Another, and a more common, way of checking this is with simple correlations. Are any variables strongly correlated among themselves in the sense that there could be a direct relation among them? ...