Table III.
Results from simulation study with approximately 10 studies
| µ | τ2 | Naïve random-effects | Fixed-effect | Random-effects | BT | Approx. semi-Bayes | Semi-Bayes | |
|---|---|---|---|---|---|---|---|---|
| 0 | 0 | Studies at stopping | 10.9 | 10.5 | 11.9 | 11.9 | 12.5 | 13.9 |
| P(benefit|µ) | 0.075 | 0.032 | 0.028 | 0.027 | 0.023 | 0.015 | ||
| Coverage, all CIs | 84.7 | 93.2 | 93.9 | 93.9 | 95.1 | 96.9 | ||
| Coverage, last CI | 84.7 | 93.2 | 93.9 | 93.9 | 95.1 | 96.9 | ||
| τ2 median | 0.022 | 0 | 0.025 | 0.022 | 0.033 | 0.067 | ||
| τ2 mean | 0 | 0 | 0 | 0 | 0.015 | 0.065 | ||
| V at stopping | 46.3 | 46.4 | 46.6 | 46.6 | 46.7 | 46.1 | ||
| 0.0625 | Studies at stopping | 11.6 | 10.3 | 13.4 | 13.4 | 14.0 | 14.5 | |
| P(benefit|µ) | 0.108 | 0.057 | 0.046 | 0.046 | 0.037 | 0.027 | ||
| Coverage, all CIs | 77.4 | 88.5 | 90.9 | 90.9 | 92.5 | 94.5 | ||
| Coverage, last CI | 77.4 | 88.5 | 90.9 | 90.9 | 92.5 | 94.5 | ||
| τ2 median | 0.049 | 0 | 0.057 | 0.057 | 0.067 | 0.079 | ||
| τ2 mean | 0.0066 | 0 | 0.024 | 0.024 | 0.038 | 0.071 | ||
| V at stopping | 46.0 | 46.4 | 46.5 | 46.5 | 46.5 | 46.0 | ||
| 0.25 | Studies at stopping | 15.1 | 9.7 | 19.5 | 19.5 | 19.9 | 17.5 | |
| P(benefit|µ) | 0.177 | 0.136 | 0.069 | 0.070 | 0.060 | 0.063 | ||
| Coverage, all CIs | 64.9 | 71.8 | 85.9 | 85.8 | 87.8 | 87.7 | ||
| Coverage, last CI | 64.9 | 71.8 | 85.9 | 85.8 | 87.8 | 87.7 | ||
| τ2 median | 0.16 | 0 | 0.19 | 0.19 | 0.20 | 0.15 | ||
| τ2 mean | 0.13 | 0 | 0.18 | 0.18 | 0.19 | 0.12 | ||
| V at stopping | 45.1 | 45.9 | 45.7 | 45.8 | 45.8 | 45.7 | ||
| 0.5 | 0 | Studies at stopping | 5.1 | 7.0 | 7.9 | 7.9 | 8.5 | 9.4 |
| P(benefit|µ) | 0.958 | 0.933 | 0.943 | 0.944 | 0.942 | 0.947 | ||
| Coverage, all CIs | 85.9 | 93.6 | 94.6 | 94.6 | 95.5 | 97.4 | ||
| Coverage, last CI | 96.3 | 97.7 | 98.1 | 98.1 | 98.2 | 99.0 | ||
| τ2 median | 0.043 | 0 | 0.031 | 0.031 | 0.048 | 0.073 | ||
| τ2 mean | 0 | 0 | 0 | 0 | 0.023 | 0.069 | ||
| V at stopping | 16.1 | 30.0 | 29.7 | 29.7 | 29.6 | 29.5 | ||
| 0.0625 | Studies at stopping | 5.6 | 7.0 | 9.0 | 9.0 | 9.5 | 9.7 | |
| P(benefit|µ) | 0.945 | 0.903 | 0.922 | 0.922 | 0.926 | 0.932 | ||
| Coverage, all CIs | 78.2 | 87.9 | 90.8 | 90.9 | 92.7 | 94.6 | ||
| Coverage, last CI | 94.3 | 95.7 | 96.8 | 96.8 | 97.4 | 98.1 | ||
| τ2 median | 0.075 | 0 | 0.061 | 0.061 | 0.077 | 0.085 | ||
| τ2 mean | 0 | 0 | 0.013 | 0.014 | 0.037 | 0.076 | ||
| V at stopping | 15.4 | 29.9 | 29.7 | 29.7 | 29.7 | 29.4 | ||
| 0.25 | Studies at stopping | 7.1 | 6.8 | 12.3 | 12.4 | 12.7 | 11.3 | |
| P(benefit|µ) | 0.938 | 0.861 | 0.909 | 0.909 | 0.909 | 0.901 | ||
| Coverage, all CIs | 65.0 | 72.1 | 85.4 | 85.2 | 87.0 | 86.5 | ||
| Coverage, last CI | 89.9 | 90.2 | 95.5 | 95.5 | 95.9 | 95.5 | ||
| τ2 median | 0.18 | 0 | 0.18 | 0.18 | 0.19 | 0.14 | ||
| τ2 mean | 0.11 | 0 | 0.15 | 0.15 | 0.18 | 0.11 | ||
| V at stopping | 13.5 | 29.1 | 28.1 | 28.1 | 28.3 | 28.4 |
BT, Biggerstaff and Tweedie.