ggplot2 : How to reduce the width AND the space between bars with geom_bar

Question

I understand that one can change the width of a bar in geom_bar using the width argument. That does work, but then it creates a larger gap between the bars. Is there a way to manually push the bars closer to together? Should I be manipulating the axis somehow instead?

Here is an example, changing width to 0.3 on the right to get the desired bar width.

library(tidyverse)
library(gridExtra)

p1 <- ggplot(iris, aes(Species, Petal.Length)) + geom_bar(stat="summary")
p2 <- ggplot(iris, aes(Species, Petal.Length)) + geom_bar(stat="summary", width=0.3)
grid.arrange(p1,p2,nrow=1)

Note: I know this question is similar to this one, but the answer to closing the gap was not apparent.

How to change the space between bars in geom_bar?

Magisterial

What have you tried so far? Perhaps, you want to look at this post to learn how to provide a reproducible example. – Herat

I tried changing the width, as described. Is a reproducible example really necessary here? This is a question about a basic feature of ggplot2, not something idiosyncratic to whatever code I'm using. – Magisterial

Sounds like you want to reduce the overall width of the x-axis. Maybe consider something like p1 + coord_fixed(ratio = 0.02)? – Raised

@Alex, have you seen this post? It mentions about position_dodge() – Herat

I don't think this is a bad question - in fact, I haven't seen this answered to my own satisfaction yet. Also, with the addition of the reproducible code, there's no call for a negative rating on the question. – Cybill

Answer 1

I would adjust the plot's aspect ratio, and have ggplot automatically assign the right width for the bars and the gap between them:

  ggplot(iris, aes(Species, Petal.Length)) + 
      geom_bar(stat="summary", width=0.4) +
      theme(aspect.ratio = 2/1)

Produces this:

Inequitable

Answer 2

Rather than taking the width smaller, which narrows the bars but increases the inter-bar space, set width = 1 to remove all space between.*

ggplot(iris, aes(Species, Petal.Length, fill=Species)) + geom_bar(stat="summary", width=1)

The default value is 0.9, so you can get very small spaces by setting width = 0.95

ggplot(iris, aes(Species, Petal.Length, fill=Species)) + geom_bar(stat="summary", width=0.95)

• With fill=Species, I took the liberty of adding color to help see the different bars when there is no space between.

Cybill

This does not really answer the question as OP want to specifically reduce the gap size, not just mask them with wider bars. – Adrea

This helped me today, Thank you width=1 is quite handy – Harms

@Adrea The OP seemed to be asking for a way to bring the bars closer together. One interpretation is that they wanted skinnier bars. I provided an alternative interpretation of just removing the space in between. I haven't found a way to do both at the same time, however (skinny bars with no gap in between). Maybe some scaling to the final plot? – Cybill

Answer 3

I ended up trying to figure out this problem while attempting to compare two plots, one with many groups of bars and the other with only one group, so instead of presenting two graphs with bars of different width produced by

library(patchwork)

ds <- iris %>% dplyr::mutate(Has.Large.Sepal = Sepal.Length > 5)

p1 <- ggplot(ds, aes(x = Species, y = Petal.Length, fill = Has.Large.Sepal)) +
      geom_bar(stat = "summary", position = "dodge")
p2 <- ggplot(ds %>% filter(Has.Large.Sepal == TRUE), aes(x = Species, y = Petal.Width)) + 
      geom_bar(stat = "summary", position = "dodge", fill = "#00BFC4")

p1 + p2

Which yields (note that the blue color on the legend apply to both graphs because I am filtering ds in the second plot to only include individuals with Has.Large.Sepal == TRUE)

I shrinked the second bar with the help of patchwork to make the bars' width similar in size to those of my previous plot

p2 <- ggplot(ds %>% filter(Has.Large.Sepal == TRUE), aes(x = Species, y = Petal.Width)) + 
  geom_bar(stat = "summary", position = "dodge", fill = "#00BFC4", width = .95) + 
  theme(axis.text.x = element_text(angle = 90))

p1 + (p2 + plot_spacer())

Obtaining a more comparable pair of plots

Dissimilitude