Guidance for using statsExpressions to customize plots?

[madebyafox]

Are there any vingnettes/examples of how to use statsExpressions to create labels for customized plots? I am trying to transition to this approach (as is recommended in ggstatsplots) to have more control over aesthetics.

I have tried to follow the example provided https://indrajeetpatil.github.io/statsExpressions/ to use centrality_description() for example.

My custom raincloud plot:

ggplot(df_s, aes(x = test_percent, y = pretty_condition, fill=pretty_condition)) +
  stat_slab(
            geom = "slab",
            position = position_nudge(y = .20), # nudged up by .20
            scale = .5) + 
  geom_boxplot(aes(fill = NULL),
               width = .20,
               size = 0.5,
               show.legend = FALSE,
               alpha = .25) + 
  geom_point(aes(color = pretty_condition), 
             position = position_jitter(width = .01, height = .05), 
             size = 1.0,
             shape = 19, # circles
             alpha = .5) + 
  labs(title = "Distribution of Test Phase Accuracy",
       x = "Test Phase Proportion Correct", y = "Condition") +
  theme(legend.position = "blank") 

renders

However, when I try to adapt this to use statsExpressions via the example:

centrality_description(df_s, y = test_percent, x = pretty_condition) |>
  ggplot(aes(y = test_percent, x = pretty_condition)) +
  geom_point() +
  geom_label(aes(label = expression), parse = TRUE, nudge_y = -0.05) + 
  coord_flip()+
  stat_slab(
            geom = "slab",
            position = position_nudge(y = .20), # nudged up by .20
            scale = .5) + 
  geom_boxplot(aes(fill = NULL),
               width = .20,
               size = 0.5,
               show.legend = FALSE,
               alpha = .25) + 
  geom_point(aes(color = pretty_condition), 
             position = position_jitter(width = .01, height = .05), 
             size = 1.0,
             shape = 19, # circles
             alpha = .5) + 
  labs(title = "Distribution of Test Phase Accuracy",
       x = "Test Phase Proportion Correct", y = "Condition") +
  theme(legend.position = "blank") 

the custom plot geoms are not rendered. It is as if the mappings declared in centrality_description() are not passed along to them?
.

My goal is to create something like this:

BUT without having to manually hack the layers

p <-   ggbetweenstats(data = df_s, x = pretty_condition, y = test_percent,
               plot.type = "box", type = "nonparametric",
               centrality.type = "parametric",
               package = "RColorBrewer",
               palette = "PRGn",
              #boxplot.args = DOESN'T EXIST
               centrality.point.args = list(color="black", size = 3, shape = 1),
               point.args = list(alpha=0), #suppress points
               ggplot.component = ## modify further with `{ggplot2}` functions
                list(
                  aes(color = pretty_condition, fill = pretty_condition),
                  scale_colour_manual(values = paletteer::paletteer_c("viridis::viridis", 3)),
                  scale_fill_manual(values = paletteer::paletteer_c("viridis::viridis", 3)))
               ) +
  ggdist::stat_halfeye(
    alpha = 0.7, 
    point_colour = NA,
    adjust = .5, 
    width = .5, .width = 0, 
    justification = -.5) +
  geom_boxplot(
    alpha = 0.1,
    width = .2, 
    outlier.shape = NA
  ) +
  geom_point(
    size = 2,
    alpha = .5,
    position = position_jitter(
      seed = 1, width = .05, height = .02
    )
  )  +
coord_flip() + theme_clean() + theme(legend.position = "blank")
p$layers[[3]]=NULL #remove ggstatsplot boxplot; not customizable via args
e <- statsExpressions::oneway_anova(data = df_s, x = pretty_condition, y = test_percent,
               type = "nonparametric") #gen stats test
p + labs(title = "Distribution of Test Phase Accuracy",
         y = "Proportion of correct responses in test phase", x = "",
         subtitle = "Impasse condition yields higher scores and greater variance",
         caption=e$expression[[1]])

[edit] I see there is a mistake in the code above, I should be using two_sample_test rather than oneway_anova in the statsExpressions statement, above.

Thanks for the support, and for this awesome package!