Table 2.
Method robustness measured in stability of inference by model and prior setting
| Setting (abbreviation) | Stability |
|---|---|
| Phenotype selection | 1.00 |
| Add. tree expert (A-tree*) | 1.00 |
| Add. tree default (A-tree) | 0.99 |
| Add. maximum likelihood (A-ML) | 0.99 |
| Non-add. tree expert (ADX-tree*) | 0.98 |
| Add. + dom. tree expert (AD-tree*) | 0.97 |
| Add. comp. expert (A-comp*) | 0.94 |
| Add. + dom. maximum likelihood (AD-ML) | 0.88 |
| Add. comp. default (A-comp) | 0.86 |
| Non-add. tree expert opposite (ADX-tree-opp*) | 0.86 |
| Non-add. comp. expert (ADX-comp*) | 0.80 |
| Non-add. maximum likelihood (ADX-ML) | 0.79 |
| Add. + dom. comp. expert (AD-comp*) | 0.69 |
| Add. + dom. tree default (AD-tree) | 0.51 |
| Non-add. tree default (ADX-tree) | 0.23 |
| Add. + dom. comp. default (AD-comp) | 0.13 |
| Non-add. comp. default (ADX-comp) | 0.04 |
a
As a proportion of analyses with less than 1% divergences for the Bayesian approach and as a proportion of analyses with convergence for the ML approach.