Other Adjustments to the Meta-Analytic Correlation Matrix
One limitation we discovered in previous metanalyses involving job attitudes and behavioral criteria was failure to appropriately specify the level of analysis of primary studies involved in the calculation of the final metaanalytic estimate (Ostroff & arrison, 1999). For example, one of the prior metaanalytic estimates in the literature was based on a total N of approximately 3,000, but 1,244 of the data points were from Angle and Perry’s (1981) unit-level rather than individual-level analysis. In several cases, inclusion of this correlation created significant bias in the published estimate. Thus, we removed Angle and Perry’s (1981) result from the meta-analytic estimates that included it, permitting our final model to reflect individual-level relationships