Possible to use stat_pvalue_manual() with dodged bar chart?

[emcnerny]

Is there any way to define the group names (columns 'group1' and 'group2' in the df) so that the new stat_pvalue_manual function can map the comparisons to specific bars on a dodged bar chart with grouping factors for both x and color?

[kassambara]

Please, try this:

library(rstatix)  # https://github.com/kassambara/rstatix
stat.test <- ToothGrowth %>%
  group_by(dose) %>%
  t_test(len ~ supp) %>%
  adjust_pvalue() %>%
  add_significance("p.adj")


library("ggpubr")
ggbarplot(ToothGrowth, x = "dose", y = "len", add = "mean_se",
          color = "supp", palette = "jco", 
          position = position_dodge(0.8)) +
  stat_pvalue_manual(stat.test, x= "dose", y.position = 30, label = "p = {p.adj}")

[Adam-JJJJJ]

Please, try this:

library(rstatix)  # https://github.com/kassambara/rstatix
stat.test <- ToothGrowth %>%
  group_by(dose) %>%
  t_test(len ~ supp) %>%
  adjust_pvalue() %>%
  add_significance("p.adj")


library("ggpubr")
ggbarplot(ToothGrowth, x = "dose", y = "len", add = "mean_se",
          color = "supp", palette = "jco", 
          position = position_dodge(0.8)) +
  stat_pvalue_manual(stat.test, x= "dose", y.position = 30, label = "p = {p.adj}")

Hi，Doc, thanks for your codes. May i ask that if i have three columns（three "supp"）per "dose", for instance, i compared treatment 1 and 2, 1 and 3, respectively, how can i separate the two different t.test results? In my results, two t.test results was stacked in the treatment 2 (because the treatment 2 was the middle column), how can i separate two results above the two columns (treatment 2 and 3)?

I've tried position=position_dodge(0.9）in stat_pvalue_manual, but it did't seem to work. How to set position in stat_pvalue_manual?
Thank for ur time.

[kassambara]

fixed now, thanks!

Please install the latest dev version and try the R code below:

df <- ToothGrowth
df$dose <- as.factor(df$dose)
# t - test
library(rstatix)  # https://github.com/kassambara/rstatix
stat.test <- df %>%
  group_by(supp) %>%
  t_test(len ~ dose, ref.group = "0.5") %>%
  adjust_pvalue() %>%
  add_significance("p.adj")
stat.test

# Plot
bp <- ggbarplot(df, x = "supp", y = "len",
                fill = "dose",
                add = "mean_sd", add.params = list(group = "dose"),
                position = position_dodge(0.8)) 
bp
# Add pvalue
bp + stat_pvalue_manual(
  stat.test, x = "supp", y.position = 33,
  label = "p.adj.signif",
  position = position_dodge(0.8)
)