18

[Data and code to produce plot at the end of the post]

Using ggplot, I plotted a bargraph with error bars, bars are grouped by two factors (one on X axis, one with fill). I would like to increase the green distance between the groups on the xaxis to make the plot easier to read: example

The closest thing to a solution on stackoverflow I found here (where someone asked my question in an unanswered comment), here, here, but I didn't manage to apply these without massing up the error bars. Can someone point me to the right parameter to tweak?

Data:

structure(list(Condition = c("Difficult", "Easy", "Difficult", 
"Easy", "Difficult", "Easy", "Difficult", "Easy", "Easy", "Difficult", 
"Easy", "Difficult"), Measure = c("Competence", "Competence", 
"Value", "Value", "Interest", "Interest", "JOL", "JOL", "Difficulty", 
"Difficulty", "Effort", "Effort"), mean = c(5.5, 4.72, 4.04, 
5.39, 3.51, 3.77, 4.34, 4.61, 3.51, 1.51, 3.44, 1.73), sd = c(1.26, 
1.62, 1.94, 1.34, 1.46, 1.46, 1.73, 1.68, 1.5, 0.86, 1.53, 1.1
), se = c(0.14, 0.18, 0.22, 0.15, 0.16, 0.16, 0.19, 0.19, 0.17, 
0.1, 0.17, 0.12), s.size = c(80, 80, 80, 80, 80, 80, 80, 80, 
80, 80, 80, 80)), .Names = c("Condition", "Measure", "mean", 
"sd", "se", "s.size"), row.names = c(NA, -12L), class = "data.frame")

which is:

   Condition    Measure mean   sd   se s.size
1  Difficult Competence 5.50 1.26 0.14     80
2       Easy Competence 4.72 1.62 0.18     80
3  Difficult      Value 4.04 1.94 0.22     80
4       Easy      Value 5.39 1.34 0.15     80
5  Difficult   Interest 3.51 1.46 0.16     80
6       Easy   Interest 3.77 1.46 0.16     80
7  Difficult        JOL 4.34 1.73 0.19     80
8       Easy        JOL 4.61 1.68 0.19     80
9       Easy Difficulty 3.51 1.50 0.17     80
10 Difficult Difficulty 1.51 0.86 0.10     80
11      Easy     Effort 3.44 1.53 0.17     80
12 Difficult     Effort 1.73 1.10 0.12     80

Code that I used to make the plot (excuse the comments, I'm learning how to use ggplot and find it helpful to take notes)

library(ggplot2)
ggplot(DF, aes(x=Measure, y=mean,fill=Condition)) + 
  geom_bar(stat="identity",
           colour="black",    # Black outline for all
           position=position_dodge())+# Put bars side-by-side instead of stacked
  geom_errorbar(aes(ymin=mean-se, ymax=mean+se),
                position=position_dodge(.9), 
                width=.25)+
  #order the groups on the xaxis
  scale_x_discrete(limits = c("Interest", "Value","Effort","Difficulty","Competence","JOL"))+
  coord_cartesian(ylim=c(0,7)) +
  #change color of bars
  scale_fill_manual(values=c("#ffcc00ff","#ffffff"), name = "Condition") + 
  #change ticks on yaxis
  scale_y_continuous(breaks=seq(0,7,by =1)) + 
  geom_hline(yintercept=0) +
  geom_vline(xintercept=0)+
  theme_bw()+
  labs(x="", y = "Rating (0-7)")+
  theme(axis.line.y = element_line(color="black"),
        axis.title.y = element_text(margin = margin(r=8)),
        axis.title.x = element_text(margin = margin(r=25)),
        panel.background = element_rect(fill = NA),
        panel.grid.major = element_blank(),
        panel.border = element_blank())
Kastany
  • 427
  • 1
  • 5
  • 16
  • 1
    Possibly useful info here: https://stackoverflow.com/questions/50077342/ggplot2-how-to-reduce-the-width-and-the-space-between-bars-with-geom-bar – AntoniosK Aug 17 '18 at 10:05
  • You can use `position = position_dodge(0.5), width = 0.25` in your `geom_bar` statement and adjust the `position = position_dodge(0.5)` in the `geom_errorbar`as well... – kath Aug 17 '18 at 10:13
  • 1
    Thank you, but position_dodge does not work. That also inserts a whitespace between the yellow and white bars, which I do not want. – Kastany Aug 17 '18 at 10:21

2 Answers2

17

What's about? 1. Use geom_col instead of geom_bar as recommended. 2. specify suitable position_dodge(0.5) and width=0.5 and 3. remove unnecessary code. 

ggplot(d, aes(x=Measure, y=mean, fill=Condition)) + 
  geom_col(colour="black",width=0.5,    
           position=position_dodge(0.5)) +
  geom_errorbar(aes(ymin=mean-se, ymax=mean+se),
                position=position_dodge(0.5), width=.25)+
  scale_x_discrete(limits = c("Interest", "Value","Effort","Difficulty","Competence","JOL")) +
  scale_y_continuous(breaks=seq(0,7,by =1),limits = c(0,7), expand = c(0,0))+
  scale_fill_manual(values=c("#ffcc00ff","#ffffff"), name = "Condition") + 
  labs(x="", y = "Rating (0-7)")+
  theme_minimal() +
  theme(axis.line = element_line(color="black"),
        axis.ticks = element_line(color="black"),
        panel.border = element_blank())

enter image description here

Roman
  • 17,008
  • 3
  • 36
  • 49
  • 1
    Could you perhaps point me to an explanation why geom_col is better than geom_bar? And thank you so much for taking the time to clean up my code!! Very helpful in the struggle to understand better which part of the code does what, amongst all the examples that I'm taking apart and recycling while learning :) – Kastany Aug 17 '18 at 15:04
  • Take a look here: https://mzuer.github.io/blog/2020/05/30/r-geom_bar-and-geom_col In short: geom_col is the same as geom_bar(stat="identity") – Bouncner Apr 26 '22 at 10:09
5

Thanks all for thinking along and AntoniosK for the link to this question which helped me find a solution which works for me (though it feels a bit hacky): manually change the width of the bars, tweak the aspect ratio (width to height of plot) in theme, adjust position_dodge in error bars to match width of bars:

geom_bar(width = 0.7)
theme(aspect.ratio = 3/5)
geom_errorbars(position=position_dodge(.7))

(and I also moved the legend to the top of the plot, not visible in the screenshot)

solution

Kastany
  • 427
  • 1
  • 5
  • 16