Table 3 Percentage of valid simulation replicates across nsim=1000 replicates. Bayesian replicates were excluded if they yielded any divergent transitions, effective sample sizes were too low (below 400), or split-\(\hat {r}\) values were too large (above 1.01). REML replicates were excluded from calculations for \(\hat {r}\) if both cluster and cluster-period variances were estimated as 0, yielding an invalid estimate of r
r | |||||||
|---|---|---|---|---|---|---|---|
0.8 | 1 | ||||||
ρ1 | T | m | S | Bayesian | REML | Bayesian | REML |
0.05 | 5 | 10 | 1 | 99.5 | 88.5 | 94.6 | 100.0 |
2 | 100.0 | 97.4 | 99.8 | 100.0 | |||
5 | 100.0 | 100.0 | 100.0 | 100.0 | |||
100 | 1 | 90.1 | 99.8 | 44.2 | 100.0 | ||
2 | 100.0 | 100.0 | 94.6 | 100.0 | |||
5 | 100.0 | 100.0 | 100.0 | 100.0 | |||
9 | 10 | 1 | 100.0 | 99.6 | 99.1 | 100.0 | |
2 | 100.0 | 99.9 | 100.0 | 100.0 | |||
5 | 100.0 | 100.0 | 100.0 | 100.0 | |||
100 | 1 | 99.8 | 100.0 | 86.9 | 100.0 | ||
2 | 100.0 | 100.0 | 100.0 | 100.0 | |||
5 | 100.0 | 100.0 | 100.0 | 100.0 | |||
0.1 | 5 | 10 | 1 | 99.8 | 95.2 | 97.0 | 100.0 |
2 | 100.0 | 99.9 | 100.0 | 100.0 | |||
5 | 100.0 | 100.0 | 100.0 | 100.0 | |||
100 | 1 | 98.0 | 100.0 | 67.8 | 100.0 | ||
2 | 100.0 | 100.0 | 99.0 | 100.0 | |||
5 | 100.0 | 100.0 | 100.0 | 100.0 | |||
9 | 10 | 1 | 100.0 | 100.0 | 99.6 | 100.0 | |
2 | 100.0 | 100.0 | 100.0 | 100.0 | |||
5 | 100.0 | 100.0 | 100.0 | 100.0 | |||
100 | 1 | 100.0 | 100.0 | 96.8 | 100.0 | ||
2 | 100.0 | 100.0 | 100.0 | 100.0 | |||
5 | 100.0 | 100.0 | 100.0 | 100.0 | |||