GGPLOT box plot subdivided by color with means in middle of boxplot

Question

I've got data with two categorical variables. I can boxplot these but I can't get the means to display in the correct position. I've created the effect in the iris dataset (the red rectangles are added by hand, not in ggplot).

Iris <- iris %>%
        mutate(SepalLengthType = ifelse(Sepal.Length > 5.8, "high", "low"))

means <- Iris %>% 
        group_by(Species, SepalLengthType) %>% 
        summarise(Sepal.Width = mean(Sepal.Width), .groups = "keep")
plot <- ggplot(data = Iris, aes(y=Sepal.Width, x = SepalLengthType, colour = Species))+
        geom_boxplot()

Now I want to add the means to each box plot These lines below all work, but the mean is not centred on the box plot but on the SepelLengthType category.

plot + stat_summary(fun = "mean" , aes(color = Species), shape = 15)
plot + stat_summary(fun = "mean" , aes(group = Species), shape = 15)
plot + stat_summary(fun.y = "mean", shape = 15) # this works, but is deprecated
plot + geom_point(data = means, aes(color = Species), shape = 15)

How can the means be displayed in the middle of each box plot? I appreciate I could re-arrange the data so each set of data points is in it's own column, but as they are not all the same length, this needs it's own work-arounds.

When I use fun = "mean" I get a warning message "Removed 5 rows containing missing values (geom_segment)." Why is that? The 'means' line does not have this problem but I'd rather not have to calculate the means myself.

Bappa Das Bappa Das · Accepted Answer · 2021-04-05T11:08:56

You can use position=position_dodge(0.9) like the following code

library(tidyverse)

Iris <- iris %>%
  mutate(SepalLengthType = ifelse(Sepal.Length > 5.8, "high", "low"))

means <- Iris %>% 
  group_by(Species, SepalLengthType) %>% 
  summarise(Sepal.Width = mean(Sepal.Width), .groups = "keep")

plot <- ggplot(data = Iris, aes(y=Sepal.Width, x = SepalLengthType, colour = Species))+
  geom_boxplot(position=position_dodge(0.9))

plot + geom_point(data = means, aes(color = Species), shape = 15, 
                  position = position_dodge2(width = 0.9))

or using stat_summary as

plot + stat_summary(fun = "mean", aes(group = Species), shape = 15, 
                  position = position_dodge2(width = 0.9))

GGPLOT box plot subdivided by color with means in middle of boxplot

1 Answers