CODEWITHSUNDEEP
rggplot2bar-chart
meclark
meclark

Reputation: 13

Add ±SE label with Mean labels on top of bar plots?

I am trying to add the mean ± standard error value in text on top of my bars, in my bar-chart. Like Excel does, sorta... I have the mean, but I can't manage to find a way to add the standard error value as well (and even less a "±" symbol).

Is there a way to do so?

Here's my data (nh1), the code, and the graph I have for now.

dput(nh1)
traitement  N nombre_moules         sd         se         ci
1           canard 12    33.7500000 53.0696369 15.3198846 33.7188386
2       canard_lam  9    37.2222222 48.8102904 16.2700968 37.5189105
3              lam  9     0.4444444  0.5270463  0.1756821  0.4051236
4            nouee 12     0.8333333  1.4034589  0.4051437  0.8917153
5     nouee_canard  9   119.2222222 37.1274083 12.3758028 28.5386523
6 nouee_canard_lam 12    83.3333333 58.3225098 16.8362584 37.0563548
7        nouee_lam  9     4.3333333  4.3011626  1.4337209  3.3061663
8           temoin 12     0.5833333  1.2401124  0.3579896  0.7879298

Thank you for your help!

    nh1 <- summarySE(data = my_data_nombremoules, measurevar="nombre_moules", groupvars="traitement")
nh1

nh1$traitement <- factor(nh1$traitemeent)

ggplot(nh1, aes(x = traitement, y=nombre_moules
                                , fill=factor(traitement))) + 
  geom_bar(position=position_dodge(), stat="identity") +
  geom_errorbar(aes(ymin=nombre_moules-se, ymax=nombre_moules+se),
                width=.2,                    
                position=position_dodge(.9))

nh2 <- ggplot(nh1, aes(x = traitement, y=nombre_moules,
                        fill=factor(traitement))) + 
  geom_bar(position=position_dodge(), stat="identity") +
  geom_errorbar(aes(ymin=nombre_moules-se, ymax=nombre_moules+se),
                width=.2, position=position_dodge(.9)) + 
  stat_summary(fun=mean, colour="black", geom="point", size=2,show.legend = FALSE) +
  stat_summary(fun=mean, colour="black", geom="text",size=4, show.legend = FALSE, 
               vjust=-0.7, aes( label=round(nombre_moules, digits=1)))

Graph I have for now

(On a side note, I tried using geom_text instead of stat_summary for my mean labels, but I couldn't reduce the number of digits with digits=1, it gave me the "aesthetic unknown" error...)

Upvotes: 0

Views: 1404

Answers (1)

Allan Cameron
Allan Cameron

Reputation: 173858

It's not clear why you're using stat_summary here, since your data are already summarised. You can use geom_point and geom_text directly.

I have had to try to reclaim your data from your plot since you did not provide raw data in your question (see data used at the end of this answer)

I have to agree with DJJ in the comments: this looks messier with the SE labels attached. I have chosen geom_label here instead of geom_text to stop the numbers being obscured by the error bars.

library(ggplot2)

ggplot(nh1, aes(x = traitement, y=nombre_moules, fill = factor(traitement))) + 
  geom_col() +
  geom_errorbar(aes(ymin = nombre_moules - se, ymax = nombre_moules + se), width =.2) + 
  geom_label(aes(label = paste(nombre_moules, "\ub1", se)), nudge_y = 5, size = 4,
             label.size = 0, label.r = unit(0, "pt"), fill = "#ebebeb") +
  geom_point(size = 3) +
  theme(legend.position = "none", axis.text.x = element_blank()) +
  labs(x = NULL, y = NULL)

Data used

nh1 <- structure(list(traitement = c("a", "b", "c", "d", "e", "f", "g", "h"), 
                      nombre_moules = c(33.8, 37.2, 0.4, 0.8, 119.2, 83.3, 4.3, 0.6), 
                      se = c(16, 16, 0.05, 0.05, 13, 17, 1, 0.1)), 
                 class = "data.frame", row.names = c(NA, -8L))

Created on 2020-07-02 by the reprex package (v0.3.0)

Upvotes: 2

Related Questions