An Alternative Approach to Method Effects by Using Latent-Variable Models: Applications in Organizational Behavior Research
An Alternative Approach to Method Effects by Using Latent-Variable Models: Applications in Organizational Behavior Research
复制标题
DOI：
10.1037/0021-9010.79.3.323
复制
发表时间：
1994-06
影响因子：
9.9
通讯作者：
L. J. Williams;S. Anderson
中科院分区：
文献类型：
--
作者：
L. J. Williams;S. Anderson
回顾了近年来组织行为学领域关于社会期望和消极情感作为人为协方差潜在来源的研究。接下来，潜变量结构方程方法的一个例子，作为一个更合适的策略，分析方法的影响，规避与以前的统计技术相关的问题。本例说明了具有和不具有方法效应的结构方程模型的规格，并描述了模型比较的顺序，该顺序提供了方法效应的存在以及这些效应对实质性关系估计值的影响的直接检验。最后，这种潜变量的方法进行了讨论，涉及方法的影响方面的其他潜在的应用。在组织行为学研究中，方法差异的一个重要研究方向是试图直接测量与方法的共同效应相关的一些变量。
Recent research in the area of organizational behavior on social desirability and negative affectivity as potential sources of artifactual covariance is reviewed. Next, an example of a latent-variable structural equation approach is presented as a more appropriate strategy for analyzing method effects that circumvents problems associated with prior statistical techniques. This example illustrates the specification of a structural equation model with and without method effects and describes the sequence of model comparisons that provides direct tests for the presence of method effects and for the impact of these effects on estimates of substantive relationships. Finally, this latent-variable approach is discussed with regard to other potential applications involving method effects. An important stream of research on method variance in organizational behavior studies has attempted to directly measure some of the variables associated with common effects of method