Skip to main content
. 2020 Jun 23;17(12):4507. doi: 10.3390/ijerph17124507

Table 1.

Comparison of measurement models.

Model Description χ2 df Δχ2 RMSEA SRMR CFI TLI
The baseline five-factor model ST, SS, AN, DE, SH. 1443.052 395 0.059 0.070 0.911 0.902
The four-factor model a ST and SH were combined into one factor. 2932.681 400 1489.629 ** 0.090 0.329 0.785 0.766
The four-factor model ST and SS were combined into one factor. 5469.909 400 4026.857 ** 0.128 0.133 0.569 0.532
The four-factor model AN and DE were combined into one factor. 2599.756 400 1156.704 ** 0.084 0.127 0.813 0.797
The four-factor model AN and SH were combined into one factor. 2327.843 400 884.791 ** 0.079 0.129 0.836 0.822
The four-factor model DE and SH were combined into one factor. 1886.997 400 443.945 ** 0.069 0.086 0.874 0.863
The three-factor model AN, DE, and SH were combined into one factor. 3163.953 404 1720.901 ** 0.094 0.125 0.766 0.748

Note: ST = internalized stigmatization; SS = social support; AN = anxiety; DE = depression; SH = shame. RMSEA = root mean square error of approximation; SRMR = standardized residual mean root; CFI = comparative fit index; TLI = non-normed fit index (NNFI). ** p < 0.01. a Consistent with the practice used in previous studies, the comparison of measurement models is conducted based on the fact that the time-lag design ensured the acceptable discriminate validity of variables measured at different time points. Thus, the comparison of measurement models focuses on the test of discriminate validity of variables measured at the same time point (at Time 1 or Time 2 for this study). Even so, the test of discriminate validity of internalized stigmatization and shame, suggested by the reviewer, had to be conducted. Thus, an alternative four-factor model that combined internalized stigmatization and shame into one factor was used to compare with the baseline five-factor model.