Moreover, we checked for the existence of common method bias in the data. Harman’s one-factor test was employed, in which no single, general factor was extracted (Podsakoff and Organ 1986).In addition, we re-estimated our structural equation model with all the indicator variables loading on an unmeasured latent method factor (MacKenzie et al. 1993). For identification purposes, it was necessary to constrain factor loadings within constructs to be equal when estimating this model. The results showed that none of the individual path coefficients corresponding to relationships between the indicators and the method factor was significant.Moreover, the overall pattern of significant relationships was not affected by common method variance (i.e., all of the paths that were significant when common method variance was not controlled remained significant even when common method variance was controlled). Overall, we conclude that common method bias does not seem to be a serious concern for this study.