Table 3.
Fit Parameters of the DRKA Scale Compared to the Rasch Model Before and After Modification (n = 200)
| Parameters | Rasch Model | Knowledge (n = 28 Items) | Knowledge Revised (n = 22 Items)a | Attitudes (n = 9 Items) | Attitudes Revised (n = 9 Items)a |
|---|---|---|---|---|---|
| Item no. | 1–28 | 1–14, 16, 18–22, 27–28 | 1–9 | 1–9 | |
| Disordered thresholds | No | No | No | Yes | No |
| Person separation index | >1.5 | 1.06 | 1.09 | 1.18 | 1.24 |
| Person reliability | >0.7 | 0.53 | 0.54 | 0.58 | 0.60 |
| Item fit (infit MnSq) | >0.7 and <1.3 | None | None | None | None |
| Principal components analysis | |||||
| Variance by the first factor | >50% | 37.8% | 38.1% | 48.2% | 46.2% |
| Eigenvalue for first contrast | <2.0 | 2.63 | 2.19 | 1.71 | 1.62 |
| Targeting, difference between person and item mean | <1.0 | 0.82 | 0.92 | 0.93 | 1.32 |
| Differential Item Functioningb | <1.0, P > 0.05 | Age (item 10) Gender (item 20) Education (item 11) | None | None | None |
Bolded values indicate misfit to the Rasch model.
a
Columns for which the scales achieved the best fit to the Rasch model.
b
DIF was tested for age group (≤50 years, >50 years), gender, and education level (none or primary, secondary, or higher).