What To Do If Your Results Are Statistically (In)Significant

You're a proactive employer, and are in the process of examining your compensation structure for internal pay equity. You've assembled and cleaned your data, the statistical models have been carefully constructed and the regressions have been performed. You're now presented with a summary of the regression results. Some of those results indicate statistically significant differences and some don't.

What do you do now?

This is THE question - not just for compensation regression, but for any statistical analysis of any kind of employment decision.

If you know what to do next, you'll be able to evaluate potential problem areas, take appropriate action, and do what's in the best interest of your organization and your employees. 

If you don't know what to do next, you may overreact and make wide-sweeping changes that can make things worse. You may choose to do nothing, lulled into thinking there are no issues because you've overestimated what statistics can tell you.

In this week's installment of The Proactive Employer Podcast, we'll be talking about what to do when your results are statistically (in)significant. We'll discuss the important role that practical significance plays in interpreting the results, the limitations of statistics and touch on how the number of events you're studying can influence the results. We'll conclude with a discussion of five simple questions you can ask to help you decide what to do next.