Visualizing Proportions using Stacked bars

There are many ways to present proportions.

1. Pie charts:

- work well when the goal is to emphasize simple fractions, such as one-half, one-third, or one-quarter

- work well when we have very small datasets - are not effective for comparing proportions

2. Stacked bars : work for side-by-side comparisons of multiple conditions or in a time series

3. Side-by-side bar charts: to directly compare the individual fractions to each other.

In this post, stacked bars are used when we have only two bars in each stack.

Example: to examine the difference in income between male and female (data source: BRFSS2018)

Here is how the aggregated data look like

df_GenderIncome <- BRFSS %>% 
  group_by(GENDER, INCOME) %>% 
  dplyr::summarise(Count =  n()) %>% 
  ddply(.(INCOME),transform, Percentage = Count*100/sum(Count)) %>% 
  mutate(pos = cumsum(Percentage) - (0.5 * Percentage))
                      
df_GenderIncome

##    GENDER     INCOME Count Percentage       pos
## 1  Female $20K-<$35K 29608   57.43996  28.71998
## 2    Male $20K-<$35K 21938   42.56004  78.71998
## 3  Female $35K-<$50K 20421   52.79882 126.39941
## 4    Male $35K-<$50K 18256   47.20118 176.39941
## 5  Female $50K-<$75K 23800   50.50184 225.25092
## 6    Male $50K-<$75K 23327   49.49816 275.25092
## 7  Female      $75K+ 47630   46.60561 323.30280
## 8    Male      $75K+ 54568   53.39439 373.30280
## 9  Female      <$20K 24445   61.07585 430.53793
## 10   Male      <$20K 15579   38.92415 480.53793

To create stacked bar chart

ggplot(data = df_GenderIncome)+
  geom_bar(aes(x = INCOME, 
               y = Percentage,
               fill = GENDER),
               stat = "identity") +
   labs(y = "Relative proportion (%)", 
        x = "Salary Bracket") +
  scale_fill_manual(values = c("#A9A9A9", "#FF8C00")) +
  scale_y_continuous(breaks = seq(0, 100, 25)) +
  geom_hline(yintercept = 50,  linetype = "dashed", color = "blue")+
  theme(axis.text = element_text(size = 12, color = "black"),
        axis.title = element_text(size = 12, face = "bold"),
        legend.text = element_text(colour="black", size = 12),
        legend.position = "right")