# BIC loss
```r
> d1[, .(loss = mean(loss, na.rm = T)), keyby = .(model)]
model loss
1: drift_perc_acc 110.2713
2: drift_perc_accDR 109.8721
3: drift_perc_accDR_ndt0 443.6314
```
# Compare BICs
```r
> m1 <- lmer(loss ~ model + (1| study/id), data = d1)
> summaryh(m1)
term results
1: (Intercept) b = 100.69, SE = 9.36, t(10) = 10.76, p < .001, r = 0.96 # perc_acc
2: modeldrift_perc_accDR b = −0.40, SE = 5.29, t(2737) = −0.08, p = .940, r = 0.001
3: modeldrift_perc_accDR_ndt0 b = 336.92, SE = 5.65, t(2745) = 59.65, p < .001, r = 0.75
> m2 <- lmer(loss ~ model + platformZ + (1| study/id), data = d1)
> summaryh(m2)
term results
1: (Intercept) b = 102.43, SE = 8.33, t(8) = 12.30, p < .001, r = 0.98
2: modeldrift_perc_accDR b = −0.40, SE = 5.29, t(2737) = −0.08, p = .940, r = 0.001
3: modeldrift_perc_accDR_ndt0 b = 336.99, SE = 5.65, t(2739) = 59.68, p < .001, r = 0.75
4: platformZ b = 12.55, SE = 7.60, t(6) = 1.65, p = .151, r = 0.56
> m3 <- lmer(loss ~ model + (1| platform/study/id), data = d1)
> summaryh(m3)
term results
1: (Intercept) b = 100.16, SE = 13.07, t(1) = 7.66, p = .071, r = 0.99
2: modeldrift_perc_accDR b = −0.40, SE = 5.29, t(2737) = −0.08, p = .940, r = 0.001
3: modeldrift_perc_accDR_ndt0 b = 336.94, SE = 5.65, t(2740) = 59.67, p < .001, r = 0.75
```