Factor-based comparison of several populations using the COMPAR software (Q1965950)

The author proposes an interactive software, COMPAR, written for PCs with the GAUSS programming language, in order to gather recent inferential tools for comparing several populations using Principal Component Analysis results. The methods used for this comparison are based on the comparison of covariance matrices under normality assumptions. They include the test of equality, the test of proportionality and the test of comparison of sets of eigenvectors and eigenspaces. The design of the software and examples, which illustrate the aim of this software, are presented.