By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement . We’ll occasionally send you account related emails.

I am using stat_compare_means() to carry out an anova. I need to also carry out the post-hoc Tukey test and would like to add p value comparisons to the figure, as is possible with the Kruskal-Wallis test.

So currently Kruskal-Wallis, followed by the post-hoc Dunn test can be implemented as:

my_comparisons <- list( c("0.5","1"),c("1","2"),c("0.5","2"))
df <- ToothGrowth
ggbarplot(data=df,"dose","len", add = c("mean_se")) +
stat_compare_means(comparisons = my_comparisons)

I am picturing the code for anova followed by Tukey test looking something like this:

my_comparisons <- list( c("0.5","1"),c("1","2"),c("0.5","2"))
df <- ToothGrowth
ggbarplot(data=df,"dose","len", add = c("mean_se")) +
stat_compare_means(method="anova", comparisons = my_comparisons)

Best wishes

for pairwise-comparison between groups, you need to specify method = "t.test" (Instead of anova) or method = "wilcox.test"

my_comparisons <- list( c("0.5","1"),c("1","2"),c("0.5","2"))
df <- ToothGrowth
ggbarplot(data=df,"dose","len", add = c("mean_se")) +
    stat_compare_means(method="t.test", comparisons = my_comparisons)

# A tibble: 3 x 8
   term group1 group2 comparison estimate  conf.low conf.high   p.adj
1  dose    0.5      1      1-0.5    9.130  5.901805 12.358195 2.0e-08
2  dose    0.5      2      2-0.5   15.495 12.266805 18.723195 1.1e-11
3  dose      1      2        2-1    6.365  3.136805  9.593195 4.2e-05
# Visualize
library(ggpubr)
ggboxplot(df, x = "dose", y = "len") +
  stat_pvalue_manual(
    stat.test, label = "p.adj", 
    y.position = c(29, 35, 39)
  ropolomx, EmilioEm, RosarioIacono, Ryxicon, guatavita, millerh1, MichaelPeibo, leowill01, biobibibi, angelolimeta, and 4 more reacted with thumbs up emoji
  bernardojoconnor and Minh-AnhHuynh reacted with hooray emoji
    All reactions
          
I found a way around it but this is a simpler way to it. Thank you!
On Fri, Aug 10, 2018 at 6:41 PM Alboukadel KASSAMBARA < ***@***.***> wrote:
 Closed #102 <#102>.
 You are receiving this because you commented.
 Reply to this email directly, view it on GitHub
 <#102 (comment)>, or mute
 the thread
 <https://github.com/notifications/unsubscribe-auth/AoMCywGkn8KekASsXTSY_knjx8XSlxUBks5uPcXkgaJpZM4VgUnA>
          
Thank you kassambara for a nice explanation. I have a similar problem, but with stat_compare_means. How could I include a Tukey test in this function, instead of a t.test, wilcox, kruskal anova?
I tried your code with stat.test. However, I get the following error:

Warning messages:

1: Computation failed in stat_compare_means():

Non-supported method specified. Allowed methods are one of: t.test, t.test, t.test, wilcox.test, wilcox.test, wilcox.test, anova, anova, kruskal.test, kruskal.test
          
Is there a way to do the same thing with the outputs of glht like so:
stat.test <- aov(len ~ dose, data = df) %>%
  summary(glht(., linfct=mcp(dose="Dunnett")))
          
Hi guys,
Just a silly question, was Dunn's test finally implemented? @BlueCeramicRipples structured an answer, but I guess this way applied only ANOVA.
Any suggestions in this regard?
Best wished and be safe!
          
As a follow up on the question from @Rhodnius : Which test is performed when doing pairwise comparisons + global p value as stated in your docs?
"p + stat_compare_means(comparisons = my_comparisons)+ # Add pairwise comparisons p-value

stat_compare_means(label.y = 50)                   # Add global p-value"
My plan is to implement Kruskal-Wallis + Dunn's for a faceted plot, each facet containing 3 groups. Is it automatically the case that Dunn's is used for the pairwise comparisons if you chose Kruskal?
  tukey_hsd()

@kassambara Could you please tell me how we could use ANOVA-Post Hoc Test for grouped data?

I want to get something similar to this picture but people always just show t.test.

Thank you very much!