- [[20210802_133212 exclusion for prereg and manuscript - methods]]
- [public osf prereg link](https://osf.io/6vmfa)
# Model fit
![[rating2_pred 9.png|800]]
![[rt_pred 9.png|800]]
![[rt0_pred 8.png|800]]
![[rt1_pred 9.png|800]]
```r
# model fit (BIC)
> d[condition %in% c(0, 1), summaryh(lmer(loss ~ conditionEC * platformZ + (1 + conditionEC | study)))]
boundary (singular) fit: see ?isSingular
term results
1: (Intercept) b = 113.96, SE = 4.96, t(4) = 22.97, p < .001, r = 1.00
2: conditionEC b = 2.07, SE = 1.02, t(12) = 2.03, p = .066, r = 0.50
3: platformZ b = 8.03, SE = 4.31, t(4) = 1.86, p = .136, r = 0.68
4: conditionEC:platformZ b = −0.38, SE = 0.97, t(17) = −0.39, p = .702, r = 0.09
```
# DDM Results
```r
> d[, lapply(.SD, mean, na.rm = T), keyby = .(Platform, condition), .SDcols = params]
Platform condition driftb0 driftb1 B nondectime x0
1: Lucid 0 -0.1639949 0.1618512 3.079581 2.928859 -0.03622875
2: Lucid 1 -0.1843951 0.4143044 2.995991 2.861511 -0.05673250
3: MTurk 0 -0.3880884 0.1101174 3.216906 3.236276 0.01852434
4: MTurk 1 -0.2774960 0.4165748 2.867304 3.339186 0.04178184
```
![[param-driftb1 1.png|800]]
![[param-driftb0 1.png|800]]
![[param-b 1.png|800]]
![[param-startingpoint 1.png|800]]
![[param-ndt 1.png|800]]
## Frequentist mixed models
Predictor coding
```r
> d[, .(Platform, Condition, platformZ, conditionEC)] %>% distinct() %>% arrange(conditionEC)
Platform Condition platformZ conditionEC
1: Lucid Control -1.467939 -0.5
2: MTurk Control 0.681227 -0.5
3: Lucid Treatment -1.467939 0.5
4: MTurk Treatment 0.681227 0.5
```
Model: `parameter ~ conditionEC * platformZ + (1 + conditionEC | study)`
```r
# driftb1
> d[condition %in% c(0, 1), summaryh(lmer(driftb1 ~ conditionEC * platformZ + (1 + conditionEC | study)))]
term results
1: (Intercept) b = 0.27, SE = 0.09, t(4) = 3.07, p = .038, r = 0.84
2: conditionEC b = 0.20, SE = 0.05, t(4) = 3.73, p = .019, r = 0.88 # barely significant!
3: platformZ b = −0.03, SE = 0.08, t(4) = −0.34, p = .753, r = 0.17
4: conditionEC:platformZ b = 0.009, SE = 0.05, t(5) = 0.19, p = .854, r = 0.09
# bayesian results
effect result
1: conditionEC b = 0.19, 95% HPD [0.007, 0.35], BF = 3.97, d = 0.19, 95% HPD [0.009, 0.34]
1: platformZ b = −0.02, 95% HPD [−0.25, 0.22], BF = 0.32, d = −0.02, 95% HPD [−0.24, 0.22]
1: conditionEC:platformZ b = 0.006, 95% HPD [−0.14, 0.15], BF = 0.2, d = 0.007, 95% HPD [−0.14, 0.15]
```
![[Pasted image 20220118153451.png]]
```r
# b
> d[condition %in% c(0, 1), summaryh(lmer(B ~ conditionEC * platformZ + (1 + conditionEC | study)))]
boundary (singular) fit: see ?isSingular
term results
1: (Intercept) b = 3.07, SE = 0.18, t(4) = 17.24, p < .001, r = 0.99
2: conditionEC b = −0.06, SE = 0.07, t(22) = −0.85, p = .405, r = 0.18
3: platformZ b = 0.06, SE = 0.16, t(4) = 0.39, p = .715, r = 0.19
4: conditionEC:platformZ b = −0.05, SE = 0.07, t(32) = −0.73, p = .472, r = 0.13
# bayesian results
effect result
1: conditionEC b = −0.06, 95% HPD [−0.24, 0.14], BF = 0.13, d = −0.02, 95% HPD [−0.11, 0.06]
1: platformZ b = 0.05, 95% HPD [−0.43, 0.56], BF = 0.24, d = 0.03, 95% HPD [−0.20, 0.24]
1: conditionEC:platformZ b = −0.05, 95% HPD [−0.22, 0.12], BF = 0.11, d = −0.02, 95% HPD [−0.10, 0.06]
```
![[Pasted image 20220118193722.png]]
```r
# driftb0
> d[condition %in% c(0, 1), summaryh(lmer(driftb0 ~ conditionEC * platformZ + (1 + conditionEC | study)))]
boundary (singular) fit: see ?isSingular
term results
1: (Intercept) b = −0.29, SE = 0.06, t(4) = −4.64, p = .010, r = 0.92
2: conditionEC b = −0.01, SE = 0.03, t(12) = −0.49, p = .635, r = 0.14
3: platformZ b = −0.09, SE = 0.06, t(4) = −1.61, p = .182, r = 0.62
4: conditionEC:platformZ b = 0.02, SE = 0.03, t(17) = 0.91, p = .374, r = 0.22
# starting point
> d[condition %in% c(0, 1), summaryh(lmer(x0 ~ conditionEC * platformZ + (1 + conditionEC | study)))]
term results
1: (Intercept) b = −0.002, SE = 0.02, t(4) = −0.12, p = .912, r = 0.06
2: conditionEC b = −0.01, SE = 0.02, t(3) = −0.86, p = .446, r = 0.43
3: platformZ b = 0.03, SE = 0.02, t(4) = 1.85, p = .137, r = 0.67
4: conditionEC:platformZ b = 0.006, SE = 0.01, t(4) = 0.40, p = .708, r = 0.19
# non-decision time
> d[condition %in% c(0, 1), summaryh(lmer(nondectime ~ conditionEC * platformZ + (1 + conditionEC | study)))]
boundary (singular) fit: see ?isSingular
term results
1: (Intercept) b = 3.13, SE = 0.24, t(4) = 13.13, p < .001, r = 0.99
2: conditionEC b = 0.13, SE = 0.07, t(6) = 1.86, p = .112, r = 0.60
3: platformZ b = 0.25, SE = 0.21, t(4) = 1.20, p = .299, r = 0.52
4: conditionEC:platformZ b = 0.04, SE = 0.07, t(9) = 0.57, p = .585, r = 0.19
```
# Parameter ~ covariate
Model: `lmer(parameter ~ covariate * platform + (1 + covariate | study))`
Only 2018 participants (control condition).
## CRT
- `driftb1 ~ crt_accZ`: p = .065
![[param_crt_accZ 1.png]]
## Conservative
![[param_conservativeZ.png]]
## Age
![[param_ageZ.png]]
## Gender
![[param_genderZ.png]]
## Race (white/non-white)
![[param_whiteZ.png]]
## Education (college or not)
![[param_collegeZ.png]]
## Value placed on accuracy
![[param_accimpZ.png]]