Table 2.
AIC and BIC weights for Models V1 and V2 (“drift” models) and B1 and B2 (“threshold” models) for each of the four data sets.
| Data sets |
Σ | |||||
|---|---|---|---|---|---|---|
| I | II | III | IV | |||
| AIC | Model V1 | 0.059 (0) | 0.050 (0) | 0.185 (4) | 0.152 (2) | 6.9% |
| Model V2 | 0.106 (4) | 0.094 (2) | 0.119 (3) | 0.116 (2) | 12.6% | |
| Model B1 | 0.412 (8) | 0.498 (9) | 0.468 (12) | 0.419 (8) | 42.5% | |
| Model B2 | 0.422 (13) | 0.358 (8) | 0.228 (6) | 0.313 (6) | 38.0% | |
| B vs. V | 5.06 | 5.94 | 2.29 | 2.73 | ||
| BIC | Model V1 | 0.027 (0) | 0.015 (0) | 0.129 (3) | 0.085 (1) | 4.6% |
| Model V2 | 0.313 (8) | 0.221 (5) | 0.389 (10) | 0.337 (7) | 34.5% | |
| Model B1 | 0.123 (3) | 0.255 (4) | 0.216 (4) | 0.153 (1) | 13.8% | |
| Model B2 | 0.537 (14) | 0.509 (10) | 0.267 (8) | 0.425 (9) | 47.1% | |
| B vs. V | 1.94 | 3.24 | 0.934 | 1.37 | ||
The row labeled B vs. V shows how much more likely that either of Models B1 or B2 is the true model compared to Model V1 or V2 according to AIC and BIC. The values in parentheses represent the number of participants best fit by each model in each data set. The sum column (Σ) shows the percentage of participants for which each model provided best fit.