Specify the number of comparaison in a "holm" p.adjust test on ggplot

Hello, I'm currently trying to ad significant stars (on a within comparaison analysis) on my ggplot using this code:

>df <- data.frame(x = GOL5$Tache, y = GOL5$value)

p <- ggplot(GOL5, aes(Tache, value)) + 
  geom_dotplot(binaxis = "y", dotsize= .01)+
  geom_boxplot(alpha=.5 )+
  ggtitle("GOL5 ")+
  labs(y="Valeur d'asymétrie")+

df_pair <- ggstatsplot::pairwise_p(df, x, y,paired = TRUE, p.adjust.method = "holm")

  plot = p,
  df_pairwise = df_pair,
  data = df)

So i get this graph with this data :

But, i would love to able to specify the of comparaison i want for the p-value. Because when i do:

KK8<-pairwise.t.test(GOL5$value, GOL5$Tache, paired = TRUE)
KK8=p.adjust(KK8$p.value, method = "holm", n = 54)

I dont get the same significance for the data, like the comparaison between LECTM and SEM should not be significant as we see in this table on the post column (column of the pvalue post adjustment with 54 comparaisons) :
Signif table post adjust

So my goal is to have the value from KK8 for the significative value on my plot, because the geom_signif doesn't allow me to specify I want to correct with 54 comparaison on 'holm'.
I hope I'm clear...

1 Like

So, i send a message to the creator of the ggstatsplot, to have an answer, and he say me that we can't specify the number of comparaison we want in the Holm correction so you have to do your comparaison out of the boxplot and put the stats of signifiance by you self on the graph.

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.