Centering labels at the bars using ggplot2 and moving labels to the top of the errorbar in R

With the code below,

library(ggplot2)
load(url("http://murraylax.org/datasets/cps2016.RData"))

ggplot(df, aes(industry, usualhrs, fill=as.factor(sex))) +
  stat_summary(geom = "bar", fun = mean, position = "dodge", width=0.7) +
  stat_summary(geom = "errorbar", fun.data = mean_se, position = "dodge", width=0.7) + 
  stat_summary(aes(label = round(..y..,0)), fun = mean, geom = "text", size = 3, vjust = -1) +  
  xlab("Industry") + ylab("Usual Hourly Earnings") +  
  scale_x_discrete(labels = function(x) str_wrap(x, width = 12)) +
  theme(legend.position = "bottom") + 
  labs(fill = "Gender")  +
  theme_bw() 

I am producing this barplot (with error bars):

Output

The labels are centered according to the x-axis, but I would like to have the labels centered in each bar. In the first two bars, for example, I would like to have 27 at the center of the "Female" bar and 46 at the center of "Male" bar. I would also like to move the labels to the top of the error bars.

1 answer

  • answered 2020-05-22 12:47 Greg

    Add position = position_dodge(width = 1)) to your stat_summary(aes(label...)) call, outside of aes to move the labels above their respective bars.

    To move the labels above the error bars I used geom_text with a y position slightly above the error bars, which required calculating the error bar position ahead of time using dplyr::summarize

    library(dplyr)
    df %>% 
      group_by(industry, sex) %>% 
      summarise(usualhrs_mean = mean(usualhrs, na.rm = TRUE),
                count = n(),
                usualhrs_se = sd(usualhrs, na.rm = TRUE)/sqrt(count)) %>% 
      ggplot(aes(x = industry, y = usualhrs_mean, fill = as.factor(sex))) +
      geom_bar(stat = "identity", position = position_dodge(width = 1)) +
      geom_errorbar(aes(ymin = usualhrs_mean - usualhrs_se,
                        ymax = usualhrs_mean + usualhrs_se), 
                    position = position_dodge(width = 1)) +
      geom_text(aes(label=round(..y.., 0), y = (usualhrs_mean + usualhrs_se + 0.1)), vjust = -1.5, position = position_dodge(width = 1)) +
      scale_x_discrete(
        labels = function(x)
          str_wrap(x, width = 12)
      ) +
      coord_cartesian(ylim = c(0, 55)) +
      theme(legend.position = "bottom") +
      labs(fill = "Gender",
           y = "Usual Hourly Earnings")  +
      theme_bw() 
    

    enter image description here