Table 1. Bayesian ANOVA Analysis of Effects.
| Effects | Calibration (QSR) | Metacognitive bias (confidence level) | Metacognitive efficiency log (meta-d′/d′) | |||
|---|---|---|---|---|---|---|
| BFinclusion | Evidence | BFinclusion | Evidence | BFinclusion | Evidence | |
| Note. Evidence in support of including different explanatory variables in models of metacognitive calibration in the experimental group. We obtained positive evidence against inclusion of a Training × Stimulus interaction term for all measures, indicating the best-fitting model is one in which the training effect is similar for both stimulus types. There was positive evidence against inclusion of a Training × Domain interaction term (indicating transfer across domains) in models of calibration (quadratic scoring rule [QSR] score) and equivocal evidence for or against this term in models of both metacognitive bias and metacognitive efficiency. Strength of evidence is evaluated using Kass and Raftery’s (1995) interpretation of the Bayes factor. ANOVA = analysis of variance; BF = Bayes factor. | ||||||
| Training | 1.09e+10 | Very strong for | ∞ | Very strong for | 5.55 | Positive for |
| Domain | .08 | Positive against | .46 | Insubstantial | 2348.77 | Very strong for |
| Stimulus | .09 | Positive against | .08 | Positive against | .20 | Positive against |
| Training × Domain | .10 | Positive against | .59 | Insubstantial | 1.18 | Insubstantial |
| Training × Stimulus | .13 | Positive against | .09 | Positive against | .13 | Positive against |
| Domain × Stimulus | .01 | Strong against | .04 | Strong against | .46 | Insubstantial |
| Training × Domain × Stimulus | 3.66e-4 | Very strong against | 4.55e-4 | Very strong against | .07 | Positive against |