How do I plot the mean instead of the median with geom_boxplot?

Question

for some inane reason, I need to create a boxplot where the middle line is the mean instead of the median. I checked stackoverflow already and found examples of adding a mean line, but not exactly what I need. I tried using stat_smooth but no dice. Any ideas?

code below:

dust <- c(4.5, 3.7, 5, 5.2, 8.5, 6.6, 4.7, 5, 5.7, 4.3, 2.3, 7.6, 5.2, 
          6, 8.7, 7.5, 7.7, 11, 9, 6.5, 8.7, 5, 2.2, 7.5, 7.5, 3.5)

wind <- c("Present", "Absent", "Absent", "Absent", "Absent", "Absent", 
  "Absent", "Absent", "Absent", "Present", "Absent", "Absent", 
  "Present", "Present", "Absent", "Absent", "Absent", "Absent", 
  "Absent", "Present", "Absent", "Present", "Absent", "Absent", 
  "Absent", "Present")

df <- data.frame(dust,wind)

plot <- ggplot(data=df,aes(x=wind,y=dust))+geom_boxplot()+stat_smooth()

plot

geom_boxplot can take more detailed args like middle (i.e., the average value), ymin, lower, etc. check out ?geom_boxplot — Chris, May 01 '18 at 06:10
`geom_boxplot(middle = mean(dust))`. Also, why `stat_smooth`? What does it do? — Rui Barradas, May 01 '18 at 06:11
@RuiBarradas don't forget the `aes`: `geom_boxplot(aes(middle = mean(dust)))` — neilfws, May 01 '18 at 06:15
@Chris With two upvotes so far, this seems to be a question that deserves an answer for future users to refer to. Wouldn't you post your comment as an answer? — Rui Barradas, May 01 '18 at 06:21

Marcus Campbell · Accepted Answer · 2018-05-01T06:43:25.260

There are a few ways to do this:

1. Using middle

The easiest is to simply call:

plot <- ggplot(data = df, aes(y = dust, x = wind)) + 
        geom_boxplot(aes(middle = mean(dust))

2. Using fatten = NULL

You can also take advantage of the fatten parameter in geom_boxplot(). This controls the thickness of the median line. If we set it to NULL, then it will not plot a median line, and we can insert a line for the mean using stat_summary.

plot <- ggplot(data = df, aes(y = dust, x = wind)) + 
  geom_boxplot(fatten = NULL) +
  stat_summary(fun.y = mean, geom = "errorbar", aes(ymax = ..y.., ymin = ..y..),
               width = 0.75, size = 1, linetype = "solid")
print(plot)

Output using fatten = NULL

As you can see, the above method plots just fine, but when you evaluate the code it will output some warning messages because fatten is not really expected to take a NULL value.

The upside is that this method is possibly a bit more flexible, as we are essentially "erasing" the median line and adding in whatever we want. For example, we could also choose to keep the median, and add the mean as a dashed line.

This answer is not correct. aes middle does not have the desired effect. — treetopdewdrop, Apr 13 '22 at 17:54
Ah, I just checked and it looks like you are right! It looks like this solution no longer works with current versions of ggplot2. I'll edit this post with a correct solution later today. — Marcus Campbell, Apr 13 '22 at 18:55

score 5 · Answer 2 · answered May 01 '18 at 06:28

5

ggplot(df) + 
  geom_boxplot(
    aes(
      x = wind, ymin = min(dust), lower = quantile(dust, .25), 
      middle = mean(dust), upper = quantile(dust, .75),
      ymax = max(dust)
    ), stat = "identity"
  )

answered May 01 '18 at 06:28

Chris

1,575
13
20

I receive the following error when I want to rn the code: Can't draw more than one boxplot per group. Did you forget aes(group = ...)? – Abbas Nov 13 '22 at 14:13

How do I plot the mean instead of the median with geom_boxplot?

2 Answers2

1. Using middle

2. Using fatten = NULL

Linked