Is it possible to compare between facets? about ggsignif HOT 6 CLOSED

A valid question. I was thinking the same.

Update:

I have three levels and have only used one level (0.2) in this example. Using all would required group-wise / fill-wise comparisons (??).

subs5_freq %>%
group_by(breed, study_area, level ,veg_class) %>%
filter(level == 0.2) %>%
ggplot(aes(x=factor(veg_class), y=freq, fill=factor(level))) +
geom_boxplot(width = 0.75, position = position_dodge(width = .9)) +
facet_grid(study_area~breed) +
geom_signif(comparisons = list(c("1_lessgood", "2_good"),
c("2_good", "3_verygood"),
c("1_lessgood", "3_verygood")),
test = "wilcox.test", map_signif_level=TRUE)

In this case however, it would be nice if the somparisons would be stacked

from ggsignif.

@G-Thomson

Something like this?

ggplot(mpg, aes(class, hwy)) +
geom_boxplot() +
facet_grid(.~type) +
geom_signif(comparisons = list(c("2seater", "compact")), map_signif_level=TRUE)

from ggsignif.

@vestlink

This is what I'm aiming for:

Update:

Note that while this is possible, its not really what I want.

ggplot(mpg, aes(type, hwy)) +
  geom_boxplot() +
  facet_grid(.~class) +
  geom_signif(comparisons = list(c("auto", "manual")), map_signif_level=TRUE)

`

from ggsignif.

Hi @G-Thomson,

I am afraid it not possible to draw anything across facets, as long as you stay within the ggplot2 framework. But ggplot is build upon gtables, so you can use that package to draw across facet. Here is an answer on stackoverflow demonstrating the basics. But this is something I cannot facilitate with my package.

@vestlink, if I understand you correctly you want to do comparison of groups that are drawn with geom_boxplot(position = "dodge"). Unfortunately this is not automatically because the dodging draws at some to geom_signif unknown location. So it would be necessary to draw the comparisons manually as I did in the advanced example in the README.md. Now you have the further complication that you probably want to draw different brackets per facet. The best code I found to do this looks like this:

ggplot(mpg, aes(type, hwy)) +
    geom_boxplot() +
    geom_signif(data=data.frame(class=c("2seater", "midsize"), type=0, hwy=0, group=1), aes(group=group), annotations="abc", y_position = 46, xmin=1, xmax=2)+
    geom_signif(data=data.frame(class=c("minivan"), type=0, hwy=0, group=1), aes(group=group), annotations="123", y_position = 46, xmin=1, xmax=2)+
    facet_grid(.~class)

As this is all pretty complicated I would suggest to just create a new variable that combines the veg_class and level. So the new code would like this:

iris %>%
    mutate(Variable=paste0(Species, "-", Petal.Width > 1)) %>%
    ggplot(aes(x=Variable, y=Sepal.Width, fill=Petal.Width > 1)) +
    geom_boxplot() +
    geom_signif(comparisons = list(c("setosa-FALSE", "versicolor-FALSE"))) +
    facet_wrap(~ I(Sepal.Length > 5.0))

from ggsignif.

@const-ae Thank you for your comments. It is true that things get complicated across facets. I will study the code and suggestions further...

from ggsignif.

@const-ae Thanks for getting back to me.

from ggsignif.