“…Although our proposed model operated at the individual level, our data had a nested structure (i.e., an average of 2.76 subordinates were nested within the same supervisor), which created the potential for non-independence in the data, downwardly biasing the standard errors (Bliese, 2000). Following Muthén and Muthén's (2015) recommendation, we first clustered employees by supervisors and then adopted maximum likelihood parameter estimates with standard errors (MLR) using a sandwich estimator to account for the non-independence of observations due to cluster sampling, which can provide a more conservative and accurate test (Bliese, 2000;Huber, 1967;Lin et al, 2020;Muthén & Muthén, 2015;Welsh et al, 2020). We conducted all the analyses using a latent variable structural equation model in Mplus 8 (Muthén & Muthén, 2015), which can provide a more robust approach than path analysis by mitigating potential measurement bias (Cole & Preacher, 2014;Cortina et al, 2021;Preacher et al, 2010).…”