- N in dataset: 3247
- excluded 73 with incomplete data (left 3174)
- excluded 428 with < 15 trials after RT exclusion (left 2746)
# Results
Bayesian mixed-effects models: No "significant" effects of condition (maybe slightly smaller mean drift rate?).
- `parameter ~ condition + (1 + condition | study)`
- condition coded $[-0.5, 0.5]$
```r
# condition effects
boundary: b = 0.09, [−0.30, 0.44], BF = 0.32
weight on accuracy: b = 0.07 [−0.19, 0.32], BF = 0.36
mean drift rate: b = −0.11 [−0.34, 0.13], BF = 0.81
# frequentist mixed model: b = −0.12, SE = 0.04, t(28) = −2.75, p = .010
non-decision time: b = 0.07 [−0.33, 0.49], BF = 0.24
starting-point: b = −0.004 [−0.11, 0.11], BF = 0.2
```
![[param__posteriors 2.png|1400]]
![[param__posteriors_study 1.png|1400]]
![[parameters_all.png|800]]
# Correlations between parameters and behavior
`supplement__pause.Rmd`
People with larger weight on accuracy are more discerning news sharers.
![[Pasted image 20220222001326.png]]
People with larger boundaries are **less** discerning?! Probably because of outliers. See figure below
![[Pasted image 20220222001422.png]]
After removing outliers (boundary > 5), small positive effect of boundary on discernment.
![[Pasted image 20220222002144.png]]
Exclude even more outliers (boundary > 3).
![[Pasted image 20220222002339.png]]