Wei Xuhua, Allen Natalie J, Liu Yongmei
Business School, Central South University, Changsha, Hunan, 410083, People's Republic of China.
Department of Psychology, The University of Western Ontario, London, Ontario, Canada.
Behav Res Methods. 2016 Mar;48(1):72-90. doi: 10.3758/s13428-015-0585-0.
Researchers can use the coefficient of variation (CV), Gini coefficient, standard deviation (SD), Theil index, or relative mean deviation (RMD) to measure organizational disparity. Because these five measures have different properties, however, using them interchangeably may lead to inconsistent findings. Using simulated team pay data, we conducted two simulation studies to examine the similarities and potential differences among these measures. The results showed that CV, Gini, Theil, and RMD were strongly related in most circumstances and that interchanging them had little impact on their relations with outcome variables. Differences were observed, however, when interchanging any of these four measures (CV/Gini/Theil/RMD) with SD, especially when samples were characterized by a seriously skewed distribution, a wide pay gap, and a high sample disparity. Given that SD does not meet some of the properties of disparity, and that it may underestimate correlations between disparity and outcome variables, we suggest that researchers use CV, Gini, Theil, or RMD, rather than SD, to assess organizational disparity.
研究人员可以使用变异系数（CV）、基尼系数、标准差（SD）、泰尔指数或相对平均偏差（RMD）来衡量组织差异。然而，由于这五种度量具有不同的属性，互换使用它们可能会导致结果不一致。我们使用模拟的团队薪酬数据进行了两项模拟研究，以检验这些度量之间的异同。结果表明，在大多数情况下，CV、基尼系数、泰尔指数和RMD密切相关，互换使用它们对其与结果变量的关系影响不大。然而，当将这四个度量（CV/基尼系数/泰尔指数/RMD）中的任何一个与SD互换时，会观察到差异，尤其是当样本具有严重偏态分布、较大薪酬差距和较高样本差异的特征时。鉴于SD不满足差异的某些属性，并且它可能会低估差异与结果变量之间的相关性，我们建议研究人员使用CV、基尼系数、泰尔指数或RMD，而不是SD，来评估组织差异。
