Abstract
We examined the measurement invariance of organizational citizenship behavior (OCB) across employee gender using six different datasets. Based on self-reported OCB data, we found that two datasets that used the Lee and Allen (2002) scale with collectivistic and feminine country samples showed scalar invariance across employee gender. However, the other four datasets showed poor model fit in configural invariance. Next, based on supervisor-reported OCB data, we found poor model fit in configural invariance across employee gender, and poor model fit in configural invariance between gender-matched dyads and gender-dissimilar dyads. To examine the source of the poor model fit in configural invariance testing, we conducted permutation tests. With the exception of three cases, permutation results suggested that the poor model fit in configural invariance testing originated from poor overall model fit rather than from actual group discrepancy. As supplemental analyses, we revised OCB scales and recommended shortened OCB scales that are likely to meet scalar invariance. The current study uncovers possible psychometric issues related to popular OCB measures, proposes alternative measures, and provides suggestions for OCB-gender research and practice.