“…Second, measurement invariance across type of facility, professional category and sex was performed following the common sequence explained previously (e.g., Vandenberg & Lance, ), by comparing progressively more constrained nested models across groups. For model identification, we used the factor‐variance strategy rather than the marker‐variable strategy (see, e.g., Byrne, ; Kim & Yoon, ); for the first step (configural model), this strategy consists of freely estimating all factor loadings and intercepts, whereas factor variances and latent means are fixed, respectively, at 1 and 0 in both groups (for more detailed model identification, see, for example, Ezpeleta & Penelo, ), preventing using as marker item that has not been proven to be noninvariant. For comparison between nested models, a decrease in CFI or TLI > .010 and an increase in RMSEA > .015 would indicate a meaningful decrement in fit and, therefore, noninvariance (Cheung & Rensvold, ).…”