Table 3 Model performance based on the estimation sample (n = 1001)
From: Estimating EQ-5D utilities based on the Short-Form Long Term Conditions Questionnaire (LTCQ-8)
EQ-5D-5L utility | Model rank | EQ-5D-3L utility | Model rank | |||
|---|---|---|---|---|---|---|
Model | MSE (MAE) | MSE | MAE | MSE (MAE) | MSE | MAE |
1. OLS | 0.0390 (0.1486) | 2 | 2 | 0.0525 (0.1732) | 2a | 3 |
2. Tobit | 0.0397 (0.1490) | 4 | 3 | 0.0536 (0.1744) | 5 | 5 |
3. CLAD | 0.0436 (0.1566) | 7 | 9 | 0.0574 (0.1808) | 7 | 7 |
4. GLM (Gamma; Log link) | 0.0414 (0.1515) | 6 | 5 | 0.0540 (0.1742) | 6 | 4 |
5. GLM (Gaussian; Log link) | 0.0399 (0.1527) | 5 | 6 | 0.0533 (0.1747) | 4 | 6 |
6. Fractional logit | 0.0392 (0.1492) | 3 | 4 | 0.0525 (0.1724) | 2a | 2 |
7. Two-part model with OLS | 0.0384 (0.1474) | 1 | 1 | 0.0516 (0.1716) | 1 | 1 |
8. Two-part model with beta regression | 0.0473 (0.1575) | 10 | 10 | 0.0903 (0.2157) | 10 | 10 |
9. Multinomial logit | 0.0451 (0.1550) | 8 | 7 | 0.0675 (0.1972) | 8 | 8 |
10. Ordered logit | 0.0458 (0.1563) | 9 | 8 | 0.0705 (0.1999) | 9 | 9 |