Mend It or End It
指出组织科学中调节变量研究的实证记录令人沮丧，因统计功效不足且效应量极小，建议仅在测量、设计及成本效益评估更优时才进行此类研究。
Moderator variables are widely hypothesized and studied in the organizational sciences, but the empirical track record of moderator variable studies is very discouraging. These studies often lack sufficient statistical power and the type of designs and measures common in organizational research virtually guarantee that the moderator effects that are found are usually extremely small. We recommend that future attempts to identify and estimate moderator effects should be limited to situations where better measures, stronger research designs and a realistic cost-benefit assessment are available. Researchers should avoid moderator hypotheses in contexts where the measures and research designs employed do not allow them to be tested in a meaningful way, and should be cautious about interpreting the very small effects they are likely to find.