Table 2.
Assessment of unidimensionality for all subscales in the 32-item HIV Stigma Scale
| Subscale | Fit measures from CFAa | |||
|---|---|---|---|---|
| All data combined | Indian cohort | Swedish cohort | US cohort | |
| Personalized stigma | ||||
| n | 960 | 244 | 157 | 559 |
| Chi-square(df = 44) | 148.37* | 75.0544* | 67.12* | 150.40* |
| CFI | 0.998 | 0.995 | 0.999 | 0.998 |
| TLI | 0.998 | 0.993 | 0.999 | 0.997 |
| RMSEA | 0.050 | 0.054 | 0.058 | 0.066 |
| Disclosure concerns | ||||
| n | 992 | 250 | 168 | 574 |
| Chi-square(df = 20) | 134.66* | 24.74, ns | 51.87* | 90.50* |
| CFI | 0.991 | 0.996 | 0.992 | 0.992 |
| TLI | 0.997 | 0.994 | 0.989 | 0.989 |
| RMSEA | 0.076 | 0.031 | 0.098 | 0.078 |
| Concerns about public attitudes | ||||
| n | 965 | 246 | 159 | 560 |
| Chi-square(df = 9) | 39.48* | 4.38, ns | 38.38* | 34.63* |
| CFI | 0.997 | 1.000 | 0.987 | 0.997 |
| TLI | 0.995 | 1.008 | 0.979 | 0.995 |
| RMSEA | 0.059 | < 0.001 | 0.114 | 0.071 |
| Negative self-image | ||||
| n | 1010 | 248 | 180 | 582 |
| Chi-square(df = 14) | 45.65* | 35.79* | 9.72, ns | 38.28* |
| CFI | 0.997 | 0.981 | 1.000 | 0.996 |
| TLI | 0.995 | 0.972 | 1.003 | 0.994 |
| RMSEA | 0.047 | 0.079 | < 0.001 | 0.055 |
| Number of factors proposed to retain by EKCb | ||||
|---|---|---|---|---|
| Personalized stigma | 1 | 1 | 1 | 1 |
| Disclosure concerns | 1 | 2 | 2 | 1 |
| Concerns about public attitudes | 1 | 1 | 1 | 1 |
| Negative self-image | 1 | 2 | 1 | 1 |
Figures in bold indicate that the subscale was not unidimensional according to EKC
*p < 0.001
aConfirmatory factor analysis of unidimensional models, fitted using the weighted least squares mean and variance (WLSMV)
bThe Empirical Kaiser Criterion