# Analysis for Stothart, Simons, Boot, and Kramer (2014)

# Author: Cary Stothart
# Date: 12/05/2013

library(ggplot2)
library(reshape)
library(gmodels)
library(scales)
library(extrafont)

my.desc <- function(x) {
  return(c(Mean=mean(x), SD=sd(x), Median=median(x), IQR=IQR(x)))
}

# Open and create datasets.
init.data <- read.delim("Data_S1_Dataset.txt", header=TRUE, sep="\t")
init.data$count <- 1

##### FILTERS ##### 

# Only good participants without knowledge about exercise literature.
w.data <- init.data[init.data$exclude==0, ]

# Only good sedentary participants.
#w.data <- init.data[init.data$exclude==0 & init.data$does_part_exer=="No", ]

###################

w.data$part_num <- factor(w.data$part_num)
w.data$exer_group_plot <- NA
w.data$exer_group_plot[w.data$exer_group=="Walking"] <- "Aerobic"
w.data$exer_group_plot[w.data$exer_group=="Toning"] <- "Nonaerobic"
w.data$exer_group_plot <- factor(w.data$exer_group_plot)
w.data$exer_group_plot <- relevel(w.data$exer_group_plot, ref="Nonaerobic")
improve.l.data <- melt(w.data, variable_name="task", 
                       measure.vars=c("exer_improve_rt", "exer_improve_switch",
                                      "exer_improve_memory"))
improve.l.data$part_num <- factor(improve.l.data$part_num)

amount.l.data <- melt(w.data, variable_name="task", 
                       measure.vars=c("exer_improve_rt_amount", 
                                      "exer_improve_switch_amount",
                                      "exer_improve_memory_amount"))
amount.l.data$part_num <- factor(amount.l.data$part_num)
l.data <- data.frame(improve.l.data$part_num, improve.l.data$exer_group, 
                     improve.l.data$age, improve.l.data$task, improve.l.data$value, 
                     amount.l.data$value, improve.l.data$does_part_exer,
                     improve.l.data$exclude_knowledge)
colnames(l.data) <- c("part_num", "exer_group", "age", "task", "improve", 
                      "improve_amount", "does_part_exer", "exclude_knowledge")
l.data$count <- 1
l.data$task <- revalue(l.data$task, c("exer_improve_rt"="Reaction Time",
                                      "exer_improve_switch"="Task Switching",
                                      "exer_improve_memory"="Relational Memory"))
l.data$cells <- paste(l.data$exer_group, l.data$task, l.data$improve)
l.data$exclude_knowledge <- factor(l.data$exclude_knowledge)

# Descriptives
aggregate(count~exer_group, FUN=sum, data=w.data)
aggregate(count~improve:task:exer_group, FUN=sum, data=l.data)
aggregate(improve_amount~exer_group:task, FUN=my.desc, 
          data=l.data[l.data$improve=="Yes", ])
aggregate(count~exer_group:gender, FUN=sum, data=w.data)
aggregate(age~exer_group, FUN=function(x) c(mean=mean(x), sd=sd(x)), 
          data=w.data)

# Theme for plots
font_import(pattern="[A/a]rial")
y
loadfonts(device="win")
font <- "Arial"
plot.theme <- theme(panel.background=element_rect(fill="#ffffff", color="#ffffff"),
                    axis.title.x=element_text(face="plain", size=14, color="#000000",
                                              vjust=0, family=font),                                              
                    axis.title.y=element_text(face="plain", size=14, color="#000000",
                                              vjust=.25, family=font),
                    axis.text.x=element_text(face="plain", size=14, color="#000000",
                                             family=font),
                    axis.text.y=element_text(face="plain", size=14, color="#000000",
                                             family=font),
                    legend.title=element_text(face="plain", size=14, color="#000000",
                                              family=font),                     
                    legend.text=element_text(face="plain", size=14, color="#000000",
                                             family=font), 
                    axis.ticks=element_blank(),
                    panel.grid.minor=element_blank(),
                    panel.grid.major=element_blank())

# Improvement rate plot
l.data$exer_group_plot <- NA
l.data$exer_group_plot[l.data$exer_group=="Walking"] <- "Aerobic"
l.data$exer_group_plot[l.data$exer_group=="Toning"] <- "Nonaerobic"
l.data$exer_group_plot <- factor(l.data$exer_group_plot)
l.data$exer_group_plot <- relevel(l.data$exer_group_plot, ref="Nonaerobic")
bar.data <- aggregate(count~task:exer_group_plot, FUN=sum,
                      data=l.data[l.data$improve=="Yes", ])
bar.data$total <- nrow(w.data[w.data$exer_group_plot=="Nonaerobic", ])
bar.data$total[bar.data$exer_group_plot=="Aerobic"] <- 
                      nrow(w.data[w.data$exer_group_plot=="Aerobic", ])
bar.data$percentage <- bar.data$count / bar.data$total
bar.plot1 <- ggplot(data=bar.data, aes(x=task, y=percentage, ymax=1)) +
             geom_bar(aes(fill=exer_group_plot), position=position_dodge(0.9)) +
             #geom_text(aes(y=percentage+.025, x=task, group=exer_group_plot,
             #               label=paste(sprintf("%.2f", percentage*100), "%", sep="")),
             #           position=position_dodge(0.9), size=4.94) +
             xlab("") +
             ylab("Percentage of Participants who Expected Improvement") +
             scale_y_continuous(labels=percent, expand=c(0, 0), breaks=seq(from=0, to=1, by=.2)) +
             scale_x_discrete(expand=c(0, 0)) +
             scale_fill_manual("Intervention\nPresented",
                               values=c("#12006f", "#8f9a9d")) +
             plot.theme

bar.plot1
ggsave(filename="FigureX_ImproveRate.tiff", dpi=300)

# Improvement amount plot
bar.data2 <- aggregate(improve_amount~task:exer_group_plot, 
                       data=l.data[l.data$improve=="Yes", ],
                       FUN=function(x) c(mean=mean(x), 
                       ci.error=qt(.975, df=length(x)-1)*sd(x)/sqrt(length(x))))
bar.plot2 <- ggplot(data=bar.data2, aes(x=task, y=bar.data2$improve_amount[, 1])) +
  geom_bar(aes(fill=exer_group_plot), position=position_dodge(0.9)) +
  geom_errorbar(aes(x=task, group=exer_group_plot, ymax=bar.data2$improve_amount[, 1]+
                    bar.data2$improve_amount[, 2], ymin=bar.data2$improve_amount[, 1]-
                    bar.data2$improve_amount[, 2], width=.5), size=.5, position=position_dodge(0.9)) +
  xlab("") +
  ylab("Amount of Expected Improvement") +
  scale_y_continuous(expand=c(0, 0)) +
  scale_x_discrete(expand=c(0, 0)) +
  scale_fill_manual("Intervention\nPresented",
                    values=c("#12006f", "#8f9a9d")) +
  plot.theme
bar.plot2
ggsave(filename="FigureX_ImproveAmount.tiff", dpi=300)

# Chi-squares for improvement data.
rt <- l.data[l.data$task=="Reaction Time", ] # Complex RT
CrossTable(rt$exer_group, rt$improve, fisher=TRUE, chisq=TRUE, expected=TRUE)
switching <- l.data[l.data$task=="Task Switching", ] # Task Switching
CrossTable(switching$exer_group, switching$improve, fisher=TRUE, chisq=TRUE, 
           expected=TRUE)
memory <- l.data[l.data$task=="Relational Memory", ] # Relational Memory
CrossTable(memory$exer_group, memory$improve, fisher=TRUE, chisq=TRUE, 
           expected=TRUE)

# Logistic regressions for improvement data.
rt_fit <- glm(improve~exer_group, data=l.data, family=binomial, 
            subset=task=="Reaction Time")
summary(rt_fit)
ts_fit <- glm(improve~exer_group, data=l.data, family=binomial, 
            subset=task=="Task Switching")
summary(ts_fit)
rm_fit <- glm(improve~exer_group, data=l.data, family=binomial, 
            subset=task=="Relational Memory")
summary(rm_fit)

# T-tests for improvement amount data.
t.test(improve_amount~exer_group, data=rt, 
       subset=improve=="Yes", var.equal=TRUE) # Complex RT
t.test(improve_amount~exer_group, data=switching, 
       subset=improve=="Yes", var.equal=TRUE) # Task Switching
t.test(improve_amount~exer_group, data=memory, 
       subset=improve=="Yes", var.equal=TRUE) # Relational Memory