Simplifying Common Method Variance Mitigation: The Role of Additional Variables

Abstract This study examines how adding independent variables can mitigate the influence of common method variance (CMV) on observed relationships. We investigate the effects of adding additional predictors in both structural equation modeling and regression under conditions of imperfect measurement, sampling error, and correlated predictors. Using an extensive simulation, the study indicates that the use of additional independent variables in structural equation models can notably reduce both the error and bias caused by CMV when four or more additional variables are present. The presence of CMV produces errors in terms of accuracy (that is, if the confidence interval of the parameter estimate contains the true parameter estimate) roughly to the same magnitude expected from Type I and Type II errors. We argue and demonstrate that results from typical models tested in the organizational sciences are unlikely to be biased by CMV because the designs of these models can mitigate CMV, even in the absence of other statistical and procedural approaches. Furthermore, if one were only concerned with appropriate conclusions of statistical significance, tests are correct consistently over 80% of the time. In contrast, CMV may be a greater concern in meta-analyses which rely on bivariate effect sizes (i.e., typically correlations) corrected for unreliability. The study has implications for researchers where CMV may be a risk and allows scholars to gauge the potential threat of CMV in prior published research that did not take steps to test for or address potential CMV concerns.