added pilot data & exploration scrip

analyses/01-data-exploration-pilot-01.r

@@ -0,0 +1,130 @@
+# load packages
+library(tidyverse)
+library(ggbeeswarm)
+library(brms)
+library(tidyboot)
+library(tidyjson)
+library(tidybayes)
+library(patchwork)
+library(GGally)
+library(cowplot)
+library(BayesFactor)
+library(aida)   # custom helpers: https://github.com/michael-franke/aida-package
+library(faintr) # custom helpers: https://michael-franke.github.io/faintr/index.html
+library(cspplot)
+
+##################################################
+
+# these options help Stan run faster
+options(mc.cores = parallel::detectCores(),
+        brms.backend = "cmdstanr")
+
+# use the CSP-theme for plotting
+theme_set(theme_csp())
+
+# global color scheme from CSP
+project_colors = cspplot::list_colors() |> pull(hex)
+
+# setting theme colors globally
+scale_colour_discrete <- function(...) {
+  scale_colour_manual(..., values = project_colors)
+}
+scale_fill_discrete <- function(...) {
+  scale_fill_manual(..., values = project_colors)
+}
+
+##################################################
+
+
+
+# import data
+d <- read.csv('../data/01-pilot/01-data-raw-pilot-01.csv')
+
+# identify if response is correct (for comprehension questions)
+d <- d %>% mutate(correct = 
+                  ifelse(is.na(correctResponse), 1,
+                         correctResponse==response))
+
+# identify participants failing comprehension question(s)
+getCorrectness <- function(id){
+  g <- d %>% filter(prolific_pid==id)
+  if(sum(g$correct)==nrow(g)){return(1)}
+  else{return(0)}
+
+}
+
+d <- d %>% mutate(participantCorrect = unlist(
+  pmap(
+    list(prolific_pid),
+    getCorrectness
+  )
+))
+
+# remove bad participants
+d <- d %>% filter(participantCorrect==1)
+# n for final sample
+length(unique(d$prolific_pid))
+
+# focus on critical trials, recode response variable
+d2 <- d %>%  filter() %>%
+  filter(trialType=='critical') %>% 
+  mutate(response=ifelse(actual_cause=='red', 8-as.numeric(response),
+                                  as.numeric(response)))
+# figure
+d2 %>% 
+  ggplot(aes(x=mechanism, y=response, color=effect_valence))+
+  stat_summary(fun='mean')+
+  stat_summary(fun='mean', geom='line', aes(group=effect_valence))+
+  stat_summary(fun.data='mean_se')+
+  scale_color_manual(values=c('blue', 'darkgreen', 'red'))+
+  scale_y_continuous(breaks=1:7)+
+  coord_cartesian(ylim=c(1,7))+
+  geom_hline(aes(yintercept=4), linetype='dashed')+
+  theme_classic()
+
+# t-test in the neutral condition
+dneutral <- d2 %>% filter(effect_valence=='neutral')
+t.test(response~mechanism, data=dneutral)
+
+# "omnibus" regression model:
+fit <- brms::brm(response ~ mechanism * effect_valence,
+                 data=d2 |> mutate(response = factor(response, ordered=T)),
+                 family=brms::cumulative("logit"),
+                 iter=4000, warmup=2000)
+
+# some tests
+
+## unpleasant > neutral? (yes)
+faintr::compare_groups(
+  fit, 
+  higher = effect_valence == 'unpleasant', 
+  lower = effect_valence == 'neutral')
+
+## neutral > pleasant? (yes)
+faintr::compare_groups(
+  fit, 
+  higher = effect_valence == 'neutral', 
+  lower = effect_valence == 'pleasant')
+
+## disjunctive vs conjunctive for each level of effect_valence
+
+### pleasant (no effect)
+faintr::compare_groups(
+  fit, 
+  higher = effect_valence == 'pleasant' & mechanism == 'conjunctive',
+  lower  = effect_valence == 'pleasant' & mechanism == 'disjunctive' 
+)
+
+### neutral (no effect)
+faintr::compare_groups(
+  fit, 
+  higher = effect_valence == 'neutral' & mechanism == 'conjunctive',
+  lower = effect_valence == 'neutral' & mechanism == 'disjunctive' 
+  )                       
+
+### unpleasant (no effect)
+faintr::compare_groups(
+  fit, 
+  higher = effect_valence == 'unpleasant' & mechanism == 'disjunctive',
+  lower  = effect_valence == 'unpleasant' & mechanism == 'conjunctive'
+)