Table 2. The difference in deviance information criterion (ΔDIC) between the full model (i.e., the model that includes both me and z) and either reduced model (me-only or z-only), for experimental data and data simulated using each reduced model.
Negative/positive values favor the full/reduced model. Note that the ΔDIC values for the experimental data were all strongly negative, favoring the full model. In contrast, the ΔDIC values for the simulated data were all positive, implying that this procedure did not simply prefer the more complex model.
Table 2—source data 1. DIC for model fitting to the monkeys’ data and to the simulated data.
elife-36018-table2-data1.zip (901B, zip)
DOI: 10.7554/eLife.36018.015
| Experimental data | Simu: me model | Simu: z model | ||||||
|---|---|---|---|---|---|---|---|---|
| ∆DIC: full - me | ∆DIC: full - z | ∆DIC: full - me | ∆DIC: full - z | |||||
| Mean | Std | Mean | Std | Mean | Std | Mean | Std | |
| Monkey F | −124.6 | 2.3 | −2560.4 | 5.2 | 3.1 | 9.8 | 0.2 | 11.8 |
| Monkey C | −1700.4 | 2.1 | −6937.9 | 1.3 | 17.5 | 11.3 | 1.8 | 1.3 |
| Monkey A | −793.6 | 3.4 | −2225.7 | 4.0 | 25.4 | 9.0 | 1.2 | 3.4 |