Table 2.
Goodness-of-fit values for the models
| Data fit | Group | SS |
MS1 |
MS2 |
|||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| p | ln(L) | AIC* | BIC† | p | ln(L) | AIC* | BIC† | p | ln(L) | AIC* | BIC† | ||
| Aggregated | KOR (z = 1) | 5 | −5172.7 | 10,355.4 | 10,383.2 | 5 | −5196.7 | 10,403.4 | 10,431.3 | 6 | −5171.5 | 10,355.1 | 10,388.5 |
| CON (z = 1) | 5 | −5035.2 | 10,080.4 | 10,108.2 | 5 | −5042.7 | 10,095.4 | 10,123.2 | 6 | −5034.8 | 10,081.6 | 10,115.0 | |
| Individual | KOR (z = 15) | 5 | −2925.5 | 6001.1 | 6382.8 | 5 | −2943.3 | 6036.7 | 6418.4 | 6 | −2922.1 | 6024.2 | 6482.3 |
| CON (z = 19) | 5 | −3444.8 | 7079.6 | 7585.6 | 5 | −3446.2 | 7082.4 | 7588.4 | 6 | −3443.2 | 7114.5 | 7721.7 | |
For the aggregate fits, data from all 24 participants are modeled as if from one participant (hence z = 1, where z is the (effective) number of participants modeled in each experiment). For the individual fits, it was not possible to model participants who had zero hit, miss, false alarm, or correct rejection responses (hence z values <24). A smaller AIC or BIC value indicates greater support for a model. BOLD indicates that the model fit the data best according to the AIC measure.
*The AIC is calculated as follows: AIC = −2ln(L) + 2P, where P = p × z is the total number of free parameters for each fit, p is the number of free parameters for each model.
†The BIC is calculated as follows: BIC = −2ln(L) + Pln(q), where q is the number of observations [q(Aggregated, KOR group) = 1920, q(Aggregated, CON group) = 1920, q(Individual, KOR group) = 1200, q(Individual, CON group) = 1520].