title: “SI_graphs” output: html_document — Horowitz, Mekler, Schneider, Frank, etc. Analyses for Scalar Implicature, Give-Quantifier, and DCCS

Set up Rmd parameters

library(knitr)

## Warning: package 'knitr' was built under R version 3.1.3

opts_chunk$set(cache=TRUE, message=FALSE, warning=FALSE)

Preliminaries

rm(list=ls())
library(ggplot2)
library(reshape)
library(plyr)
library(dplyr)
library(stringr)
library(tidyr)
library(markdown)
library(directlabels)
library(magrittr)
library(bootstrap)
library(RCurl)
library(langcog)
library(RColorBrewer)


theme_set(theme_bw())

Read in individual dataframes from all studies

Experiment 1: Adhoc and scalar items

d <- read.csv("../data/implicatures_adhocScalar_data.csv")

#fill in the selection types for easier analyses
d_exp1=melt.data.frame(d,c("Subj_ID", "age","agegroup", "list"),c("carrots", "hats", "cookies", "trains", "cats", "purses",  "keys", "shirts", "breads", "horses", "bears", "frogs", "plates", "books", "elephants", "lamps", "bananas", "butterflies", 
"carrots_condition", "hats_condition", "cookies_condition", "trains_condition", "cats_condition", "purses_condition",  "keys_condition", "shirts_condition", "breads_condition", "horses_condition", "bears_condition", "frogs_condition", "plates_condition", "books_condition", "elephants_condition", "lamps_condition", "bananas_condition", "butterflies_condition", 
"carrots_type", "hats_type", "cookies_type", "trains_type", "cats_type", "purses_type",  "keys_type", "shirts_type", "breads_type", "horses_type", "bears_type", "frogs_type", "plates_type", "books_type", "elephants_type", "lamps_type", "bananas_type", "butterflies_type"))

data_exp1 <- d_exp1[1:864,] 
data_exp1$condition <- d_exp1$value[865:1728]
data_exp1$type <- d_exp1$value[1729:2592]
names(data_exp1)[5] <- "item"
names(data_exp1)[6] <- "correct"

attach(data_exp1)
data_exp1$trial_type[type == "implicature"] <- "implicature"
data_exp1$trial_type[type != "implicature"] <- "control"
detach(data_exp1)

#rename agegroups for prettier graphs
data_exp1$agegroup %<>%
  str_replace("4.0--4.5", "4-4.5 years")%>%
  str_replace("4.5--5.0", "4.5-5 years")

data_exp1 %<>%
  rename(Age = agegroup)

#age groups
# data_exp1 %<>%
#   filter(Subj_ID != "", correct != "NA") %>%
#   mutate(agesplit = cut(age, breaks=c(4, 4.5, 5)),
#          agesplit = factor(agesplit,
#                            labels=c("4-4.5 years", "4.5-5 years"))) %>%
#   filter(agesplit != "NA")

#renaming for graphs
data_exp1$type <- factor(data_exp1$type, 
                    levels = c("implicature","control_comparison", 
                    "control_distractor",
                    "control_none","control_unambiguous", "control_all"), 
                    labels = c("Implicature","Comparison", 
                    "Distractor",
                    "None","Unambig. Some", "All"))

data_exp1$condition <- factor(data_exp1$condition, 
                    levels = c("adhoc","scalar"), 
                    labels = c("Ad-hoc","Scalar"))

# data_exp1 %<>%
#   dplyr::rename(Age = agesplit)

Experiment 2: Scalar items only

d2 <- read.csv("../data/implicatures_scalarOnly_data.csv")

data_exp2=melt.data.frame(d2,c("Subj_ID","age","agegroup","list"),c("carrots", "hats", "cookies", "trains", "cats", "purses",  "keys", "shirts", "breads", "horses", "bears", "frogs", "plates", "books", "elephants", "lamps", "bananas", "butterflies", 
"carrots_selectionType", "hats_selectionType", "cookies_selectionType", "trains_selectionType", "cats_selectionType", "purses_selectionType",  "keys_selectionType", "shirts_selectionType", "breads_selectionType", "horses_selectionType", "bears_selectionType", "frogs_selectionType", "plates_selectionType", "books_selectionType", "elephants_selectionType", "lamps_selectionType", "bananas_selectionType", "butterflies_selectionType", 
"carrots_type", "hats_type", "cookies_type", "trains_type", "cats_type", "purses_type",  "keys_type", "shirts_type", "breads_type", "horses_type", "bears_type", "frogs_type", "plates_type", "books_type", "elephants_type", "lamps_type", "bananas_type", "butterflies_type"))

data <- data_exp2[1:918,] 
data$selectionType <- data_exp2$value[919:1836]
data$type <- data_exp2$value[1837:2754]
names(data)[5] <- "item"
names(data)[6] <- "correct"

data_exp2 <- data

attach(data_exp2)
data_exp2$trial_type[type == "some"] <- "implicature"
data_exp2$trial_type[type != "some"] <- "control"
detach(data_exp2)

#rename agegroups
data_exp2$agegroup %<>%
  str_replace("3.0--3.5", "3-3.5 years")%>%
  str_replace("3.5--4.0", "3.5-4 years")%>%
  str_replace("4.0--4.5", "4-4.5 years")%>%
  str_replace("4.5--5.0", "4.5-5 years")

data_exp2 %<>%
  rename(Age = agegroup)

# data_exp2 %<>%
#   filter(Subj_ID != "", correct != "NA") %>%
#   mutate(agesplit = cut(age, breaks=c(3, 3.5, 4, 4.5, 5)),
#          agesplit = factor(agesplit,
#                            labels=c("3-3.5 years", "3.5-4 years", "4-4.5 years", "4.5-5 years"))) %>%
#   filter(agesplit != "NA")

#renaming for graphs
data_exp2$type <- factor(data_exp2$type, 
                    levels = c("all","none", 
                    "some"), 
                    labels = c("All","None", 
                    "Some"))
data_exp2$selectionType <- factor(data_exp2$selectionType, 
                    levels = c("all","none", 
                    "some"), 
                    labels = c("All","None", 
                    "Some"))
# data_exp2$trial_type <- factor(data_exp2$trial_tyçpe, 
#                                levels = c ("control", "implicature"), 
#                                labels = c("Control", "Implicature")) 

# data_exp2 %<>%
#   rename(Age = agesplit)

Exp 2 age details

age2 <- data_exp2 %>%
  group_by(Age)%>%
  summarise(mean = mean(age), sd = sd(age), median = median(age))

Experiment 3: SI, GQ, DCCS Scalar Implicature first

#reading in the original wide-form dataframe
d1 <- read.csv("../data/SI_coding.csv")

#fill in the selection types for easier analyses
d1 %<>%
  mutate(carrot_selection = factor(carrot_selection, levels=c(1, 2, 3), labels = c("none", "some", "all")), 
         hat_selection = factor(hat_selection, levels=c(1, 2, 3), labels = c("all", "some", "none")), 
         cookies_selection = factor(cookies_selection, levels=c(1, 2, 3), labels = c("none", "some", "all")),
         trains_selection = factor(trains_selection, levels=c(1, 2, 3), labels = c("none", "all", "some")), 
         cats_selection = factor(cats_selection, levels=c(1, 2, 3), labels = c("all", "some", "none")),
         purses_selection = factor(purses_selection, levels=c(1, 2, 3), labels = c("none", "some", "all")),
         keys_selection = factor(keys_selection, levels=c(1, 2, 3), labels = c("none", "some", "all")),
         shirts_selection = factor(shirts_selection, levels=c(1, 2, 3), labels = c("some", "all", "none")),
         breads_selection = factor(breads_selection, levels=c(1, 2, 3), labels = c("all", "some", "none")), 
         horses_selection = factor(horses_selection, levels=c(1, 2, 3), labels = c("all", "some", "none")), 
         bears_selection= factor(bears_selection, levels=c(1, 2, 3), labels = c("none", "some", "all")),
         frogs_selection = factor(frogs_selection, levels=c(1, 2, 3), labels = c("all", "some", "none")), 
         plates_selection = factor(plates_selection, levels=c(1, 2, 3), labels = c("all", "some", "none")), 
         books_selection = factor(books_selection, levels=c(1, 2, 3), labels = c("none", "all", "some")), 
         elephants_selection = factor(elephants_selection, levels=c(1, 2, 3), labels = c("some", "all", "none")),
         lamps_selection = factor(lamps_selection, levels=c(1, 2, 3), labels = c("none", "all", "some")), 
         bananas_selection = factor(bananas_selection, levels=c(1, 2, 3), labels = c("some", "all", "none")), 
         butterflies_selection = factor(butterflies_selection, levels=c(1, 2, 3), labels = c("none", "some", "all")))

d2=melt.data.frame(d1,c("sub_ID","test_age", "agegroup","training"),c("carrot", "hat", "cookies", "trains", "cats", "purses",  "keys", "shirts", "breads", "horses", "bears", "frogs", "plates", "books", "elephants", "lamps", "bananas", "butterflies", "carrot_selection", "hat_selection", "cookies_selection", "trains_selection", "cats_selection", "purses_selection",  "keys_selection", "shirts_selection", "breads_selection", "horses_selection", "bears_selection", "frogs_selection", "plates_selection", "books_selection", "elephants_selection", "lamps_selection", "bananas_selection", "butterflies_selection", "carrot_type", "hat_type", "cookies_type", "trains_type", "cats_type", "purses_type",  "keys_type", "shirts_type", "breads_type", "horses_type", "bears_type", "frogs_type", "plates_type", "books_type", "elephants_type", "lamps_type", "bananas_type", "butterflies_type"))


#making a respectable df  
onethird <- (1/3)*nrow(d2)
data <- d2[1:onethird,]
data$selection <- d2$value[(onethird+1):(onethird*2)]
data$type <- d2$value[(2*onethird+1):nrow(d2)]
names(data)[5] <- "item"
names(data)[6] <- "SI_correct"
data$correct <- data$SI_correct==1

#filtering out NAs, adding agesplit
data %<>%
  filter(sub_ID != "", correct != "NA") 

data %<>%
  rename(Age = agegroup)

#renaming for better-looking graphs
data$type %<>%
  str_replace("control_all", "All")%>%
  str_replace("control_none", "None")%>%
  str_replace("implicature", "Implicature")

data$selection %<>%
  str_replace("all", "All")%>%
  str_replace("none", "None")%>%
  str_replace("some", "Some")

#Find the Mean and median ages
age <- data %>%
  dplyr::select(sub_ID, test_age, Age) %>%
  dplyr::distinct() %>%
  group_by(Age)%>%
  summarise(n=n(), avg = mean(test_age))

#use this dataframe for stats
data_SIexp3 <- data %>%
  mutate(test = "Scalar Implicature")

#age stats
age <- data %>%
  group_by(Age)%>%
  summarise(mean = mean(test_age), median = median(test_age), sd = sd(test_age))

GIVE-QUANTIFIER

#read in and establish dataframe
d1 <- read.csv("../data/GQ_coding.csv")
d1[,c(8,10,12,14,16,18,20,22, 31, 32, 33, 34)] <- sapply(d1[,c(8,10,12,14,16,18,20,22, 31, 32, 33, 34)],as.character) #for unknown reasons this is necessary for melt.data.frame now

#change to long form
d2=melt.data.frame(d1,c("sub_ID", "test_age","agegroup", "condition", "training"),c("X1_condition", "X2_condition", "X3_condition", "X4_condition", "X5_condition", "X6_condition",  "X7_condition", "X8_condition", "X1_given", "X2_given", "X3_given", "X4_given", "X5_given", "X6_given",  "X7_given", "X8_given", "X1_correct", "X2_correct", "X3_correct", "X4_correct", "X5_correct", "X6_correct",  "X7_correct", "X8_correct"))

#making a respectable df  
onethird <- (1/3)*nrow(d2)
data <- d2[1:onethird,]
data$selection <- d2$value[(onethird+1):(onethird*2)]
data$type <- d2$value[(2*onethird+1):nrow(d2)]
names(data)[6] <- "trial"
names(data)[7] <- "prompt"
names(data)[8] <- "num_given"
names(data)[9] <- "GQ_correct"

#filtering data, creating age breaks
data %<>%
  filter(sub_ID != "", num_given != "N/A")%>%
  rename(Age = agegroup)

#this is because num_given is a factor - need to change to a number - need to do this for standard error
data$num_given <- as.character(data$num_given)
data$num_given <- as.numeric(data$num_given)
data$GQ_correct <-as.character(data$GQ_correct)
data$GQ_correct <- as.numeric(data$GQ_correct)

#renaming things for prettier graphs
data$prompt %<>%
  str_replace("all", "All")%>%
  str_replace("most", "Most")%>%
  str_replace("some", "Some")%>%
  str_replace("none", "None")

#creating dataframe for analyses
data_gq <- data %>%
  mutate(test = "Give-Quantifier")

DCCS

#wide-form
d1 <- read.csv("../data/DCCS_coding.csv")
#long form
d2=melt.data.frame(d1,c("sub_ID","test_age", "agegroup", "condition"),c("pre_1", "pre_2", "pre_3", "pre_4", "pre_5", "pre_6",  "post_1", "post_2", "post_3", "post_4", "post_5", "post_6"))

#filtering data, creating age breaks
data <- d2 %>%
  filter(sub_ID != "") 

data %<>%
  rename(Age = agegroup)

#renaming things
names(data)[5] <- "trial_type"
names(data)[6] <- "DCCS_correct"

#df for further analyses
data_dccs <- data
data_dccs$test <- "DCCS"
  
#rename trials
tmp1 <- unlist(data$trial_type)%>%
  str_replace("pre_1", "pre")%>%
  str_replace("pre_2", "pre")%>%
  str_replace("pre_3", "pre")%>%
  str_replace("pre_4", "pre")%>%
  str_replace("pre_5", "pre")%>%
  str_replace("pre_6", "pre")%>%
  str_replace("post_1", "post")%>%
  str_replace("post_2", "post")%>%
  str_replace("post_3", "post")%>%
  str_replace("post_4", "post")%>%
  str_replace("post_5", "post")%>%
  str_replace("post_6", "post")

tmp <- as.data.frame(tmp1) %>%
  dplyr::rename(switch_type = tmp1)
tmp2 <- bind_cols(data, tmp)
data <- tmp2

#renaming for prettier graphs
data$switch_type %<>%
  str_replace("post", "Post-switch")%>%
  str_replace("pre", "Pre-switch")

data$switch_type <- as.factor(data$switch_type)
dccs <- data

GRAPHS

Performance on experiment 1

agg.1<- data_exp1 %>%
  mutate(correct = as.integer(correct)) %>%
  group_by(condition, type, trial_type, Age)%>%
  multi_boot_standard("correct", na.rm=T)

# agg.data <- aggregate(data_exp1$correct, list(data_exp1$condition, data_exp1$type, data_exp1$trial_type, data_exp1$Age), FUN=sum)
# agg.data.len <- aggregate(data_exp1$correct, list(data_exp1$condition, data_exp1$type, data_exp1$trial_type, data_exp1$Age), FUN=length)
# agg.data$x <- agg.data$x 
# names(agg.data) <- c("implicature_type", "type", "trial_type", "Age", "correct")
# agg.data$total <- agg.data.len$x
# agg.data$prop.corr <- agg.data$correct / agg.data$total

# 
# #find the total number of kids in each age group and n total
# num_kids <- d_exp1 %>%
#   dplyr::select(Subj_ID, agegroup) %>%
#   dplyr::distinct() %>%
#   dplyr::group_by(agegroup)%>%
#   dplyr::summarize(n=n())%>%
#   dplyr::mutate(total.n = sum(n))
# 
# agg.data$sd <- sd(agg.data$prop.corr, na.rm = FALSE)
# 
# #prop correct and standard error
# agg.data$q <- 1 - agg.data$prop.corr
# agg.data$err <- sqrt((agg.data$prop.corr * agg.data$q) / agg.data$total)
# agg.data$CIerr <- qnorm(0.975)*agg.data$error/sqrt(n)


# #limits for graphing
# dodge <- position_dodge(width=0.9) 
# limits <- aes(ymax = ci_upper, ymin=ci_lower) 

#save df for later
# agg.exp1 <- agg.data

#graph
quartz()
ggplot(data = agg.1, 
       aes(x=type, y=mean, fill=Age)) +
  geom_bar(stat="identity", position = position_dodge(1)) +
  geom_linerange(aes(ymin = ci_lower,
                      ymax = ci_upper),
                  size = .8,
                  show_guide = FALSE,
                 position=position_dodge(1)) +
  ylab("Proportion correct") + 
  xlab("Trial Type") +
  theme_bw(base_size = 17) + theme(axis.title.x = element_text(size=18),
           axis.title.y  = element_text(size=18)) +
  scale_fill_manual(values = c("#4EB3D3", "#0868AC")) + facet_grid(~condition, scale="free_x") + geom_hline(yintercept = .5, linetype = "dashed")

Histogram of bimodal responses

imp_hist <- data_exp1 %>%
  filter(condition == "Scalar" & type == "Implicature")%>%
  group_by(Subj_ID)%>%
  mutate(correct=as.numeric(correct))%>%
  summarise(mean = mean(correct))

quartz()
hist(imp_hist$mean, main = paste('"Some" trials'), xlab = 'Performance on "some" trials', cex.lab=1.5, cex.axis=1.5, cex.main=1.5, cex.sub=1.5)

#ggplot
# quartz()
# ggplot(d, aes(x = correct), width = 50, fill = type) + geom_histogram(stat = "bin", position = "dodge") + ylab("Frequency") + xlab('Proportion of trials correct') + theme_bw(base_size=18) + facet_grid(~type)

none_hist <- data_exp1 %>%
  filter(condition == "Scalar" & type == "None") %>%
  group_by(Subj_ID)%>%
  mutate(correct = as.numeric(correct))%>%
  summarise(mean = mean(correct))

quartz()
hist(none_hist$mean, main = paste('"None" trials'), xlab = 'Performance on "none" trials', cex.lab=1.5, cex.axis=1.5, cex.main=1.5, cex.sub=1.5)

Individual differences

data_exp1 %<>%
  mutate(correct = as.numeric(correct))

ms <- ddply(data_exp1, .(type, condition, Subj_ID), summarise, 
    mean = mean(correct),
    Age = Age[1])

names(ms) <- c("type", "condition", "Subj_ID", "mean", "Age")
cs <- cast(ms, Subj_ID + Age ~ type + condition, value="mean")

#individual differences plot
quartz()
ggplot(cs, aes(x=None_Scalar, y=Implicature_Scalar, colour=Age)) + 
  geom_point(stat = "identity", position = position_jitter(.02), shape=16) +
  geom_smooth(method="lm", col="black", lty=1) + 
  geom_smooth(aes(col=Age, group=Age), 
                se=FALSE, method="lm",lty=3) + scale_colour_manual(values = c("#4EB3D3", "#0868AC")) + theme_bw(base_size=20) + theme(axis.title.y = element_text(size = 18), 
                                      axis.title.x = element_text(size = 18)) +
  labs(x = 'Proportion of "none" trials correct', y = 'Proportion of "some" trials correct')

Experiment 2

agg.2<- data_exp2 %>%
  mutate(correct = as.integer(correct)) %>%
  group_by(type, trial_type, Age)%>%
  multi_boot_standard("correct", na.rm=T)



# agg.data <- aggregate(data_exp2$correct, list(data_exp2$type, data_exp2$trial_type, data_exp2$Age), FUN=sum)
# agg.data.len <- aggregate(data_exp2$correct, list(data_exp2$type, data_exp2$trial_type, data_exp2$Age), FUN=length)
# agg.data$x <- agg.data$x 
# agg.data.len$x <- agg.data.len$x 
# 
# names(agg.data) <- c( "trial_type", "implicature", "Age", "correct")
# agg.data$total <- agg.data.len$x
# agg.data$prop.corr <- agg.data$correct / agg.data$total
# 
# agg.data$q <- 1 - agg.data$prop.corr
# agg.data$err <- sqrt((agg.data$prop.corr * agg.data$q) / agg.data$total)
# 
# #save df for later
# agg.exp2 <- agg.data

#graph
quartz()
ggplot(data = agg.2, 
       aes(x=type, y=mean, fill=Age)) +
  geom_bar(stat="identity", position = position_dodge(1)) +
  geom_linerange(aes(ymin = ci_lower,
                      ymax = ci_upper),
                  size = .8,
                  show_guide = FALSE,
                 position=position_dodge(1)) +
  ylab("Proportion correct") + 
  xlab("Trial Type") +
  theme_bw(base_size=17) + theme(axis.title.x = element_text(size=18), 
                                 axis.title.y = element_text(size=18)) +
  scale_fill_manual(values = c("#4EB3D3", "#2B8CBE", "#0868AC", "#084081")) + geom_hline(yintercept = .5, linetype = "dashed")

# mss.acc <- data_exp2 %>%
#   group_by(Subj_ID, type)%>%
#   dplyr::summarise(correct=mean(correct), 
#                    Age = Age[1])%>%
#   spread(type, correct)

# #individual differences plot
# ggplot(mss.acc, aes(x=None, y=Some, colour=Age)) + 
#   geom_point(stat = "identity", position = position_jitter(.05), shape=16) + 
#   geom_smooth(method="lm", col="black", lty=1) + 
#   geom_smooth(aes(col=Age, group=Age), 
#               se=FALSE, method="lm",lty=3) + 
#   labs(x = "Proportion of SI 'none' trials correct", y = "Proportion of SI 'implicature' trials correct") + scale_fill_manual(values = c("#4EB3D3", "#2B8CBE", "#0868AC", "#084081"))



##analysis of what kinds of items kids chose when they picked the wrong thing
exp2_wrong <- data_exp2 %>%
  filter(correct != 1) %>%
  group_by(Age, type, selectionType)%>%
  dplyr::summarise(n=n())%>%
  mutate(n.total = sum(n), prop = n/n.total)

wrong <- exp2_wrong[c(1:12, 14:20),] 
    
#now graph
quartz()
ggplot(data = wrong, 
       aes(x=selectionType, y=n, fill=Age)) +
  geom_bar(stat="identity", position = position_dodge())  +
  ylab("Count of children choosing") + 
  xlab("Selection Type") + facet_wrap(~type) + theme_bw(base_size=17) + 
  theme(axis.title.x = element_text(size=18),
        axis.title.y = element_text(size=18))+
  scale_fill_manual(values = c("#4EB3D3", "#2B8CBE", "#0868AC", "#084081"))

Experiment 3 Scalar Implicatures

# #accuracy by age group in different trials 
# agg.data <- aggregate(data_SIexp3$correct, list(data_SIexp3$type, data_SIexp3$Age), FUN=sum)
# agg.data.len <- aggregate(data_SIexp3$correct, list(data_SIexp3$type, data_SIexp3$Age), FUN=length)
# agg.data$x <- agg.data$x 
# agg.data.len$x <- agg.data.len$x 
# 
# names(agg.data) <- c("type", "Age", "correct")
# agg.data$total <- agg.data.len$x
# agg.data$prop.corr <- agg.data$correct / agg.data$total
# 
# #prop correct and standard error
# agg.data$q <- 1 - agg.data$prop.corr
# agg.data$err <- sqrt((agg.data$prop.corr * agg.data$q) / agg.data$total)
# 
# #save for later
# agg.exp3.si <- agg.data
agg.3 <- data_SIexp3 %>%
  mutate(correct = as.integer(correct)) %>%
  group_by(type, Age)%>%
  multi_boot_standard("correct", na.rm=T)

#plot
quartz()
ggplot(data = agg.3, 
       aes(x=type, y=mean, fill=Age)) +
  geom_bar(stat="identity", position = position_dodge(1)) +
  geom_linerange(aes(ymin = ci_lower,
                      ymax = ci_upper),
                  size = .8,
                  show_guide = FALSE,
                 position=position_dodge(1)) +
  ylab("Proportion correct") + 
  xlab("Trial Type") +
  theme_bw(base_size=17) + theme(axis.title.x = element_text(size = 18),
                                 axis.title.y = element_text(size = 18)) +
  scale_fill_manual(values = c("#4EB3D3", "#2B8CBE", "#0868AC", "#084081")) + geom_hline(yintercept = .5, linetype = "dashed")

#What kinds of items are children giving in incorrect trials? 
si_wrong <- data_SIexp3 %>%
  filter(correct != "TRUE") %>%
  group_by(Age, type, selection)%>%
  dplyr::summarise(n=n())%>%
  mutate(n.total = sum(n), prop = n/n.total)

si_wrong$selection %<>%
  str_replace("all", "All")%>%
  str_replace("none", "None")%>%
  str_replace("some", "Some")

si_wrong$type %<>%
  str_replace("Implicature", "Some")



quartz()
ggplot(data = si_wrong, 
       aes(x=selection, y=n, fill=Age)) +
  geom_bar(stat="identity", position = position_dodge())  +
  ylab("Count of children choosing") + 
  xlab("Selection Type") + facet_wrap(~type) + theme_bw(base_size=17) + theme(axis.title.x = element_text(size = 18), 
                                                                              axis.title.y = element_text(size = 18))+
  scale_fill_manual(values = c("#4EB3D3", "#2B8CBE", "#0868AC", "#084081"))

DCCS

agg.DCCS <- dccs %>%
  mutate(DCCS_correct = as.integer(DCCS_correct)) %>%
  group_by(switch_type, Age)%>%
  multi_boot_standard("DCCS_correct", na.rm=T)

agg.DCCS$switch_type <- factor(agg.DCCS$switch_type, levels=c("Pre-switch","Post-switch"))
agg.DCCS %<>%
  filter(switch_type=="Post-switch")


quartz()
ggplot(data = agg.DCCS, 
       aes(x=switch_type, y=mean, fill=Age)) +
  geom_bar(stat="identity", position = position_dodge(1)) +
  geom_linerange(aes(ymin = ci_lower,
                      ymax = ci_upper),
                  size = .8,
                  show_guide = FALSE,
                 position=position_dodge(1)) +
  ylab("Proportion correct") + 
  xlab("Trial Type") +
  theme_bw(base_size=17) + theme(axis.title.x = element_text(size = 18),
                                 axis.title.y = element_text(size = 18)) +
  scale_fill_manual(values = c("#4EB3D3", "#2B8CBE", "#0868AC", "#084081")) + geom_hline(yintercept = .5, linetype = "dashed")

GQ

agg.gq <- data_gq %>%
  mutate(GQ_correct = as.integer(GQ_correct)) %>%
  group_by(prompt, Age)%>%
  multi_boot_standard("GQ_correct", na.rm=T)

quartz()
ggplot(data = agg.gq, 
       aes(x=prompt, y=mean, fill=Age)) +
  geom_bar(stat="identity", position = position_dodge(1)) +
  geom_linerange(aes(ymin = ci_lower,
                      ymax = ci_upper),
                  size = .8,
                  show_guide = FALSE,
                 position=position_dodge(1)) +
  ylab("Proportion correct") + 
  xlab("Quantifier Prompt") +
  theme_bw(base_size=17) + theme(axis.title.x = element_text(size=18),
                                 axis.title.y = element_text(size=18)) +
  scale_fill_manual(values = c("#4EB3D3", "#2B8CBE", "#0868AC", "#084081")) + geom_hline(yintercept = .5, linetype = "dashed")

hist_data <- data_gq %>%
  filter(prompt != "Most")%>%
  group_by(Age, prompt, num_given)%>%
  dplyr::summarise(n = n()) %>%
  group_by(Age, prompt) %>%
  mutate(n.total = sum(n), prop = n/n.total)


quartz()
ggplot(data = hist_data, 
       aes(x=num_given, y=prop, fill=Age)) +
  geom_bar(stat="identity", position = position_dodge(1)) +
  ylab("Proportion of children giving number") + 
  xlab("Number of items") +
  theme_bw(base_size=17) + theme(axis.title.x = element_text(size = 18),
                                 axis.title.y = element_text(size=18)) +
  scale_fill_manual(values = c("#4EB3D3", "#2B8CBE", "#0868AC", "#084081")) + facet_grid(prompt~Age) + theme(legend.position="none")