Table 3. Confirmatory factor analysis model fit parameters for 1-, 2-, 3-, and 4-factor models.
| Model | Group of patients | ||
|---|---|---|---|
| Inpatients | Outpatients | Medical students | |
| 1-factor model | |||
| χ2 (df) | 384.7 (135) | 332.2 (135) | 398.2 (135) |
| P-value | <0.0001 | <0.0001 | <0.0001 |
| χ2/df | 2.85 | 2.46 | 2.95 |
| RMSEA (90% CI) | 0.164 (0.149–0.180) | 0.142 (0.126–0.158) | 0.163 (0.148–0.179) |
| TLI | 0.427 | 0.464 | 0.319 |
| CFI/PCFI | 0.495 / 0.358 | 0.529 / 0.369 | 0.400 / 0.288 |
| AIC / BIC | 4.52 / 5.46 | 4.26 / 5.23 | 4.52 / 5.44 |
| SRMR | 0.135 | 0.140 | 0.157 |
| 2-factor model | |||
| χ2 (df) | 286.2 (134) | 271.4 (134) | 297.5 (134) |
| P-value | <0.0001 | <0.0001 | 2.22 |
| χ2/df | 2.14 | 2.03 | <0.0001 |
| RMSEA (90% CI) | 0.114 (0.098–0.131) | 0.114 (0.096–0.131) | 0.122 (0.106–0.138) |
| TLI | 0.647 | 0.623 | 0.572 |
| CFI/PCFI | 0.692 / 0.489 | 0.672 / 0.460 | 0.627 / 0.435 |
| AIC / BIC | 3.57 / 4.53 | 3.64 / 4.64 | 3.57 / 4.52 |
| SRMR | 0.143 | 0.147 | 0.129 |
| 3-factor model | |||
| χ2 (df) | 186.4 (132) | 197.8 (132) | 232.0 (132) |
| P-value | 0.0013 | 0.0002 | <0.0001 |
| χ2/df | 1.41 | 1.50 | 1.76 |
| RMSEA (90% CI) | 0.054 (0.027–0.076) | 0.070 (0.047–0.090) | 0.091 (0.073–0.108) |
| TLI | 0.870 | 0.815 | 0.733 |
| CFI/PCFI | 0.890 / 0.615 | 0.843 / 0.564 | 0.772 / 0.524 |
| AIC / BIC | 2.62 / 3.63 | 2.90 / 3.96 | 2.98 / 3.98 |
| SRMR | 0.079 | 0.091 | 0.096 |
| 4-factor model | |||
| χ2 (df) | 181.7 (129) | 191.9 (129) | 217.3 (129) |
| P-value | 0.0016 | 0.0003 | <0.0001 |
| χ2/df | 1.41 | 1.49 | 1.68 |
| RMSEA (90% CI) | 0.055 (0.027–0.077) | 0.067 (0.044–0.088) | 0.082 (0.063–0.100) |
| TLI | 0.871 | 0.819 | 0.759 |
| CFI/PCFI | 0.894 / 0.607 | 0.850 / 0.560 | 0.799 / 0.533 |
| AIC / BIC | 2.63 / 3.72 | 2.90 / 4.04 | 2.90 / 3.97 |
| SRMR | 0.079 | 0.090 | 0.093 |
χ2 –chi-square statistics.
df–degrees of freedom.
CI–confidence intervals.
RMSEA–Root Mean Square Error of Approximation. The strict cut-off point estimate close to 0.06 with a lower 90% CI limit close to 0 and the upper limit less than 0.08 is currently considered “a good fit”. In the past, however, the recommendations were less strict: a point estimate below 0.08 was considered “a good fit”, whereas between 0.08 to 0.10 “a mediocre fit”.
TLI–Tucker-Lewis index or Non-Normed Fit Index. Values at least 0.95 are preferred, but values as low as 0.80 were also acceptable.
CFI–Comparative Fit Index. Values at least 0.95 are presently recognized as indicative of “a good fit”, but the limit of 0.90 was proposed in the past.
PCFI–Parsimony-corrected Comparative Fit Index. While no threshold levels have been recommended for parsimony-corrected indices, it is suggested that the values of at least 0.50 or, even better, 0.60 should be obtained.
AIC–Akaike’s Information Criterion.
BIC–Bayesian Information Criterion. Both AIC and BIC are used for assessment relative to other models. The smaller the values, the better and more parsimonious the model fit.
SRMR–standardized root mean square residual. Values less than 0.05 represent “a good fit”, however, values as high as 0.08 are deemed “acceptable”.
The above recommendations for the model fit indices were reported in Hooper et al. [33].