stanmodel.R

f.10 <- log.auc.positive ~ log(CueOrder)*Congruent + (1|Participant) + (0+log(CueOrder)*Congruent|Participant)  


fit.10.stan <- stan_lmer(f.10, data=dftouse, iter=5000, chains=3)
summary(fit.10.stan)

fit.10.stan.preds <- posterior_linpred(fit.10.stan, 
                                     transform=F, 
                                     re.form=NA) %>% 
  t %>% 
  as.data.frame() %>% 
  bind_cols(dftouse, .) %>% 
  reshape2::melt(id.vars=c("Participant", 
                           "BlockNo", "Trial", "Congruent"))  %>% 
  rename(fit=value, sim=variable)


fit.10.stan.preds.wide <-  fit.10.stan.preds %>% 
  reshape2::dcast(BlockNo+Trial+sim~Congruent, 
                  fun.aggregate = mean, 
                  value.var='fit') %>% 
  ungroup() %>% 
  mutate(Effect = (`TRUE`-`FALSE`), D = Effect/sd(`FALSE`))

fit.10.stan.preds.wide %>%
  group_by(BlockNo, Trial) %>%
  tidybayes::median_qi(D) %>% 
  ggplot(aes(Trial, D, ymin=conf.low, ymax=conf.high)) + 
  geom_hline(yintercept = 0) +
  geom_ribbon(alpha=.2) + geom_line() + facet_wrap(~BlockNo, ncol=4) +
  scale_y_continuous(breaks=seq(-3, 5, by=.5)) +
  ylab("Congruency Effect")