Supplement 1. MATLAB simulation.

File List Evans_Collinearity_Supplement.m read me.txt Description The MATLAB script Evans_Collinearity_Supplement.m generates two random normal variables <i>x</i> and <i>y</i> with the correlation between them specified by the user. It then creates a third ‘response’ variable <i>z</i> that is a function of <i>x</i> and <i>y</i>. The exact function used to create <i>z</i> can be chosen by the user. Finally, it runs a set of linear regressions of <i>z</i> vs. various combinations of <i>x</i>, <i>y</i>, and their interactions. The regressions are used to demonstrate how using <i>P</i> values to select variables in a regression model can lead to retaining falsely significant parameters, resulting in an incorrect interpretation of the data. Model selection using Akaike’s ‘An Information Criterion’ (AIC) is illustrated as an alternative method. More complete details are given in Appendix B and in the comments of the file itself.