Partial Least Squares Approaches for Detecting and Controlling Common Method Variance: A Monte Carlo Assessment of All Previous Approaches Plus a New CLC2 Approach

Common method variance (CMV) can bias the estimates of structural paths and loadings, and therefore can yield misleading conclusions and lead to potential threats to the validity of research findings (Bagozzi et al. 1991, p.421). Among the statistical techniques used to mitigate the effects of CMV, there are only two approaches to handle CMV using partial least squares (PLS), namely, the PLS instantiation of the unmeasured latent variable (UMLV) technique and the measured latent marker variable (MLMV) technique. In this paper, we assess the utility of the two MLMV-based techniques introduced by Chin et al. (2013), namely construct-level correction (CLC) and item-level correction (ILC), using Monte Carlo simulations. Moreover, we introduce a new variation of the MLMV called CLC2 and showcase its superior performance in detecting, estimating, and controlling for the effects of CMV including statistical power which heretofore has never been incorporated in past studies. Our research contributes to the Information Systems and Research Method literatures by offering a comprehensive assessment of current PLS-based techniques to handle CMV, which leads to more robust and accurate research outcomes.