I seem to have an error in the way my distribution looks. The bottom ridges of each of the facetted graphs are not at the same scale as the other ridges above, or relative to the number of counts (i.e. scale dots shown).
Is there a way to scale all distributions relative to one another?
season_names <- c(`0` = "COOL", `1` = "HOT-DRY",`2` = "HOT-WET")
dCLEAN %>% ggplot(aes(x = tdb, y = as.factor(tsv), fill = as.factor(season))) + ggdist::stat_halfeye( adjust = 0.9, justification = -0.15, .width = 0, point_colour = NA) + geom_boxplot( width = 0.2, outlier.colour = NA, alpha = 0.5)+ ggdist::stat_dots( side = "left", justification = 1.18, binwidth = 0.1) + facet_wrap(~ season, labeller = as_labeller(season_names)) + theme_bw() + theme(strip.background = element_rect(fill="white")) + theme(legend.position = "none") + scale_color_grey()+ scale_fill_grey()1 Answer
By default, the densities are scaled to have equal area regardless of the number of observations. If you wish to scale the areas according to the number of observations, you can set aes(thickness = stat(pdf*n)) in stat_halfeye(). This sets the thickness of the slab according to the product of two computed variables generated by stat_halfeye(): the density (pdf) and the number of observations per group (n).
Here's an example from the raincloud plots section of the dotsinterval vignette:
set.seed(12345) # for reproducibility
data.frame( abc = c("a", "b", "b", "c"), value = rnorm(200, c(1, 8, 8, 3), c(1, 1.5, 1.5, 1))
) %>% ggplot(aes(y = abc, x = value, fill = abc)) + stat_slab(aes(thickness = stat(pdf*n)), scale = 0.7) + stat_dotsinterval(side = "bottom", scale = 0.7, slab_size = NA) + scale_fill_brewer(palette = "Set2") This example uses stat_slab() in place of stat_halfeye(). stat_slab() is a shortcut stat equivalent to stat_halfeye() but without the point and the interval (so it saves you passing .width = 0 and point_color = NA to stat_halfeye()).
