geom_tukey() in ggplot facet_wrap

[MikkelZ]

Is there a way to treat the plots in a facet as individuals, so that the significance letters reset from one plot to another?

[ethanbass]

I'm not sure I understand what you mean. Can you give an example?
When using geom_tukey with facet_wrap, the call to facet_wrap has to come first as in the vignette.

For example if you have a plot "p":

p <- penguins |> tidyr::drop_na(sex, species) |> 
  ggplot(aes(x=species, y=body_mass_g)) +
  geom_boxplot() + 
  geom_jitter(col="slategray", alpha=0.6) +
  ylab("Body mass (g)") +
  xlab("Species") + egg::theme_article()

this first example will produce the correct plot:

p+ facet_wrap(~sex) +
  geom_tukey(where="whisker")

whereas this second plot will come at wonky because geom_tukey doesn't know you have facets when you call it.

p +   geom_tukey(where="whisker") +
  facet_wrap(~sex)

[MikkelZ]

Thanks so much for the reply Ethan! Sorry I didn't make myself clear, let me give you an example here: Imagine that there are multiple observations (values) from three groups under three conditions. I want to plot three plots for the conditions, each plot will have the values by groups. And I want to test if the values are different between groups in each plot. So I did: ggplot(data, aes(x=group, y=values))+ boxplot()+ facet_grid(~condition)+ geom_tukey() Assuming the observed values by group and condition are all different from each other, we will end up with three plots, the Tukey significance letters should be: plot 1: a, b, c plot 2: a, b, c plot 3: a, b, c This is what I meant by 'reset the significance letters from one plot to another other'. However, it seems like the significance letters continued across plots: plot 1: a, b, c plot 2: d, e, f plot 3: g, h, i It would be ok if I only have 3 groups by 3 conditions. But in reality, I have 14 groups by 12 conditions. Thus, the final plots look very messy with multi-alphabet significance letters. I hope this makes my question clearer :) otherwise I can make some graph when I am back in the office tomorrow. Thanks so much for your help! Mike

…

On Wed, Dec 13, 2023, 4:29 PM Ethan Bass ***@***.***> wrote: I'm not sure I understand what you mean. Can you give an example? — Reply to this email directly, view it on GitHub <#2 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BEU544I3UU6XTEEW22XBIW3YJHCTZAVCNFSM6AAAAABATFLCWGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQNJUGE2DKNJUGU> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[ethanbass]

Ah I understand what you mean now. Thanks for the clarification. You can control this behavior with the type argument. The default is the two-way ANOVA. In this case, ggtukey will do a global test across all facets. To just test for differences within the facetted groups, you can select type = "one-way". I think this is the option you want.

[MikkelZ]

Perfect! It is working exactly how I wanted it. Thanks so much for the help, Ethan! Cheers, Mike

…

On Wed, Dec 13, 2023 at 5:54 PM Ethan Bass ***@***.***> wrote: Ah I understand what you mean now. Thanks for the clarification. You can control this behavior with the type argument. The default is the two-way ANOVA. In this case, ggtukey will do a global test across all facets. To just test for differences within the facetted groups, you can use type = "one-way". I think this is the option you want. — Reply to this email directly, view it on GitHub <#2 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BEU544PSKM7DJJHXJM7ZFCTYJHMWDAVCNFSM6AAAAABATFLCWGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQNJUGM2DCOBZGM> . You are receiving this because you authored the thread.Message ID: ***@***.***>

-- Dr Yiou ZHU, Mike 216, Nordnesborder 3, Havforskningsinstituttet 5005 Bergen, Norway ***@***.*** ***@***.*** --------------------------------------------------------------------------

[ethanbass]

Excellent!