[[220502_192121 parameter correlations and crt-parameter model#parameter bivariate correlations|Boundary and non-decision time correlate at 0.6]], we refit models without non-decision time as predictor below. Results are very similar to the [[220502_192121 parameter correlations and crt-parameter model#crt parameters|model with non-decision time included]]. Boundary effect is slightly smaller in this model (b = 0.11 vs 0.12) but error bars are smaller as well.
```r
m_crt2 <- brm(crt ~ boundz + biasz + driftz + bound_countryz + bias_countryz + drift_countryz +
(1 + boundz + biasz + driftz | Country), d1)
Population-Level Effects:
Estimate Est.Error l-95% CI u-95% CI Rhat Bulk_ESS Tail_ESS
Intercept 1.50 0.06 1.39 1.62 1.00 6638 3053
boundz 0.11 0.07 -0.02 0.25 1.00 4161 2870 # smaller error (vs 0.08)
biasz -0.02 0.07 -0.16 0.12 1.00 4818 2935
driftz 0.31 0.07 0.16 0.45 1.00 4274 3114
bound_countryz 0.01 0.06 -0.11 0.12 1.00 5733 2953
bias_countryz -0.01 0.06 -0.12 0.11 1.00 4928 2686
drift_countryz -0.01 0.06 -0.12 0.10 1.00 5312 2841
```
![[fold0_bayesian_intervals_crt-params_no-tau.png]]
# should we exclude non-decision time (ndt)?
ndt (`ndtz`) correlates with age (e.g., [this paper](https://mdpi-res.com/d_attachment/jintelligence/jintelligence-08-00033/article_deploy/jintelligence-08-00033.pdf?version=1598953696)) and we see it in our data as well
- also correlates with [[220504_103514 predict other stuff with parameters#need for cognition nfc|need for cognition]] and possibly [[220504_103514 predict other stuff with parameters#accimp|accimp]]
model: age ~ parameters
![[fold0_bayesian_intervals_age-params 1.png]]
and max age in our dataset is 86?!
![[s20220504_150554.png]]