makeover: convert from two groups of side-by-side vertical bar charts to a more readable dot plot

median_hourly_earnings.R

# from Conrad Hacket
# Median hourly earnings
# \url{https://twitter.com/conradhackett/status/748884076493475840}
# makeover: convert from two groups of side-by-side vertical bar charts to a more readable dot plot
# Demonstrates:
#   Use of in ggplot2
#   Creating dot plots 
#   Combining color and shape in a single legend
#   Sorting a dataframe so that categorical data in one column is ordered by a second numerical column
# Note: resulting graph displays best at about 450 pixels x 150 pixels

library(tidyr)
library(dplyr)
library(ggplot2)

shape_func <- function(x) {
  y <- rep(15, times = length(x))
  y[x == "Male"] <- 1
  y[x == "Female"] <- 4
  return(y)
}

earnings_df <- data_frame(gender = factor(rep(c("Male","Female"), each = 4)),
                          race = factor(rep(c("White","Black","Hispanic","Asian"), times = 2)),
                          earnings = c(21,15,14,24,
                                       17,13,12,18)) %>% 
  mutate(race = reorder(race, earnings))

ggplot(earnings_df) + 
  aes(y = earnings, x = race) + 
  #geom_hline(yintercept = earnings_df$earnings[which(earnings_df$gender == "Male" & earnings_df$race == "White")],
  #           colour = "grey80", linetype = "dashed", alpha = 0.75) +
  geom_point(aes(colour = gender, shape = gender), size = 2.5) +
  scale_y_continuous(minor_breaks = seq(12, 24, by = 1)) +
  scale_color_discrete(name = "Gender") +
  scale_shape_manual(name = "Gender", values = c(4, 1)) +
  coord_flip() +
  ylab("Median Earnings per Hour ($/hr)") +
  theme_bw() +
  theme(axis.title.y = element_blank())