I have a 2x2 study design (i.e. 4 groups) to test the interaction of two different interventions, and obtain multiple measurements for each subject, measured across different sites. I can nicely plot these data using a facet_grid for the different sites and each of the measures, showing each of the four treatment combinations. An example of this...
site <- rep(c(1:3), times = 16)
treatmentX <- rep(c("x", "x", "y", "y"), times=12)
treatmentY <- rep(c("a", "b", "a", "b"), times=12)
reading1 <- runif(48)
reading2 <- runif(48)
reading3 <- runif(48)
d <- data.frame(site, treatmentX, treatmentY, reading1, reading2, reading3)
d1 <- gather(d, reading.type, measure, reading1:reading3)
head(d1)
ggplot(d1, aes(x=treatmentX, y=measure)) +
geom_boxplot(aes(color = as.factor(treatmentY)),
position = position_dodge2(preserve = 'single')) +
facet_grid(reading.type~site, scales = "free")
What I would like to do is now annotate this facet_grid with the results of a 2-way ANOVA and post-hoc tests for comparisons of interest. I could run a separate 2-way ANOVA for the treatmentX:treatmentY interactions and posthoc test and use this to annotate an individual graph using ggpubr::stat_pvalue_manual(), but I'm not sure how to annotate in the context of a facet_grid, let alone automate this for each panel.
My question is whether it is possible to automatically perform this 2-way ANOVA and post-hoc analysis for each of the reading.type and site combinations and then label/paste these results onto each of the panels in a facet_grid, in order to indicate significance from a post-hoc test between groups?
Any help would be greatly appreciated!
Thanks!
