Table 6.
Comparison of nested MIMIC models to assess differential item functioning of core stigma items
| Pathway from country to item | Likelihood ratio test of constrained and full model* | ||||
|---|---|---|---|---|---|
| “Feeling detached” | “Blaming yourself” | “Feeling rejected” | “Feeling stigma” | ||
| Full model | − 0.482 (0.135), 0.000 | − 0.488 (0.140), 0.000 | 0.254 (0.121), 0.035 | 0.370 (0.115), 0.001 | – |
| Constrained model 1 | INVARIANTa | − 0.398 (0.137), 0.004 | 0.363 (0.119), 0.002 | 0.480 (0.115), 0.000 | 12.672 (df 1), p = 0.002 |
| Constrained model 2 | − 0.395 (0.132), 0.003 | INVARIANT | 0.363 (0.122), 0.003 | 0.480 (0.114), 0.000 | 12.122 (df 1), p = 0.002 |
| Constrained model 3 | − 0.526 (0.132), 0.000 | − 0.532 (0.140), 0.000 | INVARIANT | 0.316 (0.113), 0.005 | 4.431 (df 1), p = 0.1412 |
| Constrained model 4 | − 0.553 (0.133), 0.000 | − 0.559 (0.137), 0.000 | 0.167 (0.119), 0.161 | INVARIANT | 10.453 (df 1), p = 0.005 |
aThe pathway from country to this item was not estimated, as it was treated as invariant in this model
*p values have been multiplied by the number of tests (n = 4) as a Bonferroni correction