Regression to the mean effect: When to be concerned and how to correct for it

Abstract

The paper deals with the problem of regression to the mean when testing hypotheses of change dependent on the level of the measure of interest at the time of the first measurement in longitudinal studies. The paper includes: A discussion on the nature of regression to the mean and when to be aware of it in relation to different fields of psychological research. An example of the severe implications regression to the mean can have for interpretation of results, based on Nielsen’s research on changes in the learning styles of Danish university students. Finally, a relatively easy way to correct for regression to the mean in connection with paired sample t-tests of no change for persons with either low or high scores at the first measurement is demonstrated.