Why doesn't the error bar show up when using ggplot?

Question

The error bar doesn show up when plotting.

Here is my original code:

aa<-structure(list(sample = c("L-FF", "L-FF", "L-FF", "L-FF", "L-FF", 
"L-FF", "L-SS", "L-SS", "L-SS", "L-SS", "L-SS", "L-SS", "L-MM", 
"L-MM", "L-MM", "L-MM", "L-MM", "L-MM", "L-MF", "L-MF", "L-MF", 
"L-MF", "L-MF", "L-MF", "L-MS", "L-MS", "L-MS", "L-MS", "L-MS", 
"L-MS"), group = c("FF", "FF", "FF", "FF", "FF", "FF", "SS", 
"SS", "SS", "SS", "SS", "SS", "MM", "MM", "MM", "MM", "MM", "MM", 
"MF", "MF", "MF", "MF", "MF", "MF", "MS", "MS", "MS", "MS", "MS", 
"MS"), group2 = c("P0N0", "P0N1", "P1N0", "P1N1", "P2N2", "P3N0", 
"P0N0", "P0N1", "P1N0", "P1N1", "P2N2", "P3N0", "P0N0", "P0N1", 
"P1N0", "P1N1", "P2N2", "P3N0", "P0N0", "P0N1", "P1N0", "P1N1", 
"P2N2", "P3N0", "P0N0", "P0N1", "P1N0", "P1N1", "P2N2", "P3N0"
), depth = c("Low", "Low", "Low", "Low", "Low", "Low", "Low", 
"Low", "Low", "Low", "Low", "Low", "Low", "Low", "Low", "Low", 
"Low", "Low", "Low", "Low", "Low", "Low", "Low", "Low", "Low", 
"Low", "Low", "Low", "Low", "Low"), mean = c(8.43666666666667, 
8.41666666666667, 8.42333333333333, 8.50333333333333, 8.34, 8.37, 
8.19333333333333, 8.32, 8.34333333333333, 8.39, 8.25666666666667, 
8.48, 8.34666666666667, 8.24666666666667, 8.31, 8.39333333333333, 
8.24, 8.49333333333333, 8.35666666666667, 8.35, 8.51333333333333, 
8.48, 8.34333333333333, 8.52666666666667, 8.34333333333333, 8.20666666666667, 
8.30666666666667, 8.4, 8.12666666666667, 8.47666666666667), se = c(0.0696020433927371, 
0.0926162932629988, 0.0829323686974982, 0.0545690184791495, 0.0435889894354068, 
0.0802080627701062, 0.202182535788277, 0.0550757054728609, 0.0744610263456291, 
0.0529150262212925, 0.0584047182264505, 0.0458257569495582, 0.0669161996662826, 
0.0811035004039766, 0.0702376916856848, 0.061734197258174, 0.056862407030773, 
0.0523874454850056, 0.0545690184791497, 0.138924439894498, 0.0120185042515464, 
0.0568624070307733, 0.0392994204085055, 0.0762306441735284, 0.066916199666282, 
0.170522074165715, 0.12387269450708, 0.0360555127546394, 0.0484194634877803, 
0.0463081466314994)), class = c("tbl_df", "tbl", "data.frame"
), row.names = c(NA, -30L))

And here is my code:

library(ggplot2)
library(ggpubr)
library(readxl)
library(ggsci)

p<-ggplot(aa, aes(group,mean,fill=group2))+
  geom_bar(stat="identity",width = 0.7,size=0.2,alpha=1,color='black',position="stack")+
  geom_errorbar(aes(ymin =mean-se, ymax =mean + se),width=0.2,position="identity") +
  theme(legend.text=element_text(family="Times",size=10))+
  labs(x ='' , y = 'Organic matter (mg/kg)', fill = 'sample') +
  theme_bw()+
  scale_y_continuous(expand = c(0, 0), limit = c(0,60),breaks = seq(0,60,10))+
  guides(fill=guide_legend(title = ''))+
  theme(panel.grid = element_blank(),  panel.background = element_blank())+
  theme(
    text = element_text(family="Times",size=14), 
    plot.title = element_text(family="Times",size = 14),
    axis.text.x = element_text(family="Times",size = 14, colour= 'black'),
    axis.text.y = element_text(family="Times",size = 14, colour= 'black'),
    legend.text = element_text(family="Times",size = 14, colour= 'black'))+
  theme(axis.ticks.length = unit(0.1,'cm'),
        plot.margin = unit(c(0.5, 0.5, 0.5, 0.5), "cm")) 
p + scale_fill_nejm()

Here is the figre I got with the code: But the error bar doesn't show up. Here is the effect I expected:

How strange can it be?

Any suggestions are welcome! Thank you advance!

Allan Cameron · Accepted Answer · 2022-09-17 20:35:02Z

There are two problems here. The first is that your standard errors are very small compared to the bars themselves. The size of each standard error ranges from 0.012 to 0.20, meaning your error bar height varies from 0.024 to 0.40, but the y axis ranges from 0 to 60. That means that your standard errors vary between 1/2500 and 1/150 the height of your panel. Even with a relatively large panel, this means the error bar will be between 1 and 3 pixels tall, effectively appearing as a flat line. The effect can be improved a little by dodging instead of stacking, since you can effectively make everything 6 times taller, but the error bars remain very small:

ggplot(aa, aes(group, mean, fill = group2))+
  geom_col(width = 0.7, size = 0.2, alpha = 1, color = 'black', 
           position = position_dodge(width = 0.8)) +
  geom_errorbar(aes(ymin =mean-se, ymax =mean + se, group = group2),
                width = 0.2, position = position_dodge(width = 0.8)) +
  theme(legend.text=element_text(family="Times",size=10))+
  labs(x ='' , y = 'Organic matter (mg/kg)', fill = 'sample') +
  theme_bw()+
  scale_y_continuous(expand = c(0, 0), limit = c(0,10),breaks = seq(0,60,10))+
  guides(fill=guide_legend(title = ''))+
  theme(panel.grid = element_blank(),  panel.background = element_blank())+
  theme(
    text = element_text(family="Times",size=14), 
    plot.title = element_text(family="Times",size = 14),
    axis.text.x = element_text(family="Times",size = 14, colour= 'black'),
    axis.text.y = element_text(family="Times",size = 14, colour= 'black'),
    legend.text = element_text(family="Times",size = 14, colour= 'black'))+
  theme(axis.ticks.length = unit(0.1,'cm'),
        plot.margin = unit(c(0.5, 0.5, 0.5, 0.5), "cm")) + 
  scale_fill_nejm()

The other issue is that the errorbars have position = "identity", which means they all sit at the same height instead of stacking. You get strange results when you try to stack errorbars to stack however. Although there are ways round this, the height issue means that these are not really worth attempting, because the errors are effectively too small to be shown on the plot.

If the size of the errors is an important thing for you to display, your best option might be to use coord_cartesian to zoom in to the relevant part of the y axis. The bars still needs to be dodged for this to work, though:

ggplot(aa, aes(group, mean, fill = group2))+
  geom_col(width = 0.7, size = 0.2, alpha = 1, color = 'black', 
           position = position_dodge(width = 0.8)) +
  geom_errorbar(aes(ymin =mean-se, ymax =mean + se, group = group2),
                width = 0.2, position = position_dodge(width = 0.8)) +
  theme(legend.text=element_text(family="Times",size=10))+
  labs(x ='' , y = 'Organic matter (mg/kg)', fill = 'sample') +
  theme_bw()+
  scale_y_continuous(expand = c(0, 0), limit = c(0,9))+
  coord_cartesian(ylim = c(8, 9)) +
  guides(fill=guide_legend(title = ''))+
  theme(panel.grid = element_blank(),  panel.background = element_blank())+
  theme(
    text = element_text(family="Times",size=14), 
    plot.title = element_text(family="Times",size = 14),
    axis.text.x = element_text(family="Times",size = 14, colour= 'black'),
    axis.text.y = element_text(family="Times",size = 14, colour= 'black'),
    legend.text = element_text(family="Times",size = 14, colour= 'black'))+
  theme(axis.ticks.length = unit(0.1,'cm'),
        plot.margin = unit(c(0.5, 0.5, 0.5, 0.5), "cm")) + 
  scale_fill_nejm()

If possible, could you please show me the error bar in the stack effect when position = "identity"? I put one new figure above. Not sure if it's that kind of "strange results" you mean... @Allan Cameron
@Chouette the point is that you can't zoom in to a stacked plot, since the stacked bars stretch from 0 to 60, so even if you got the error bars to stack, they would just look like flat lines between the bars on your plot. It would be tricky and pointless.

Collectives™ on Stack Overflow

Why doesn't the error bar show up when using ggplot?

1 Answer 1

3 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

3 Comments

Your Answer

Sign up or log in

Post as a guest

Related