Aguinis, H., & Pierce, C. A.
Applied Psychological Measurement Vol. 30, Issue 3, p. 440-442
[available at http://carbon.ucdenver.edu/~haguinis/pubs.html ]

We provide a user-friendly description of a modified f 2 effect-size estimate for categorical moderator variables in multiple regression. We recommend that this modified f 2 statistic be used when the homogeneity of error variance assumption is violated. Many researchers in the social sciences may not report effect-size estimates because the tools needed to compute these estimates are not readily available. To increase researchers’ awareness of f 2 in general and to increase the accessibility of computing modified f 2 values, we also describe an on-line program that performs all computations needed and offer this program free-of-charge at the following URL: http://carbon.ucdenver.edu/~haguinis/mmr/fsquared. A benefit of using this program is that, once computed, researchers are able to compare the magnitude of their modified f 2 values to values commonly observed in the social sciences as determined by Aguinis, Beaty, Boik, and Pierce’s (2005) 30-year review of the size of moderating effects of categorical variables.