Improving the performance of the unmeasured latent method construct technique in common method variance detection and correction
提出通过添加识别指标并利用Rindskopf重新参数化来改进未测量潜在方法构念技术，模拟表明该方法在共同方法变异检测和校正中表现良好，优于CFA标记技术。
Summary In behavioral research, common method variance (CMV) is likely to occur when data are obtained from the same sources. Ignoring CMV can lead to biases in parameter estimates. Although the unmeasured latent method construct (ULMC) technique, based on confirmatory factor analysis (CFA), is frequently used for CMV detection and correction, previous research has indicated that its performance is rather poor. In this study, we propose an approach to improve the performance of ULMC by adding identifying indicators together with the use of Rindskopf's reparameterization of the CFA model for error variances and implicit constraints for correlations. The chi‐square test for model fit and the chi‐square difference test are used to test for the existence of CMV and to determine the number of method constructs. The simulation results indicated that, given adequate sample size, the ULMC technique with the proposed approach performs well in CMV detection and correction. Moreover, its performance in CMV correction is superior to the performance of the CFA marker technique. The approach, illustrated with two datasets, is strongly recommended for empirical studies that suffer CMV.