Table 4. Bayesian analysis model comparison.
| Model (M) | P(M) | P(M|data) | BFM | BF01 | Error % |
|---|---|---|---|---|---|
| C | 0.20 | 0.733 | 10.990 | 1.000 | |
| C + E | 0.20 | 0.181 | 0.883 | 4.05 | 0.990 |
| C + E + C X E | 0.20 | 0.086 | 0.376 | 8.53 | 1.390 |
| Null model | 0.20 | 4.109×10–82 | 1.643×10–81 | 1.784×10+81 | 1.994×10–85 |
| E | 0.20 | 1.071×10–83 | 4.285×10–83 | 6.844×10+82 | 0.012 |
C = consultation style. E = participant ethnicity. C X E = the Interaction between C and E. Model (M) the variables included in each model. P(M) The prior model probability. P(M|data) Posterior model probability. BFM the posterior model odds. BF01 represents the Bayes factors, which indicates how much more likely the best model is (the model on the top row) compared with the model on the current row; for instance, the data is 8.53 times more likely under the model with just consultation style, compared with the model with both main effects and an interaction (third row from the top). Error % error in computation of the Bayes factor.