How to not have percentage labels overlap (repel percentage labels) in Likert bar plot?

Question

I'm still relatively new to R, so any help here is very appreciated.

I created a Likert bar plot, but some of the percentages shown on the plot overlap. I am wondering how to prevent this overlap from occurring? I read about ggrepel, and tried to incorporate ggrepel code but have been unsuccessful. Does anyone know if code exists to ask the percentages shown on the plot (i.e. from plot.percents=TRUE) to not overlap?

Here is my code without an attempt to incorporate ggrepel:

likert.bar.plot(likert_Prevention.ALL, plot.percents=TRUE, plot.percent.high= FALSE, plot.percent.low=FALSE, 
            center = 3, low.color = "grey72", high.color = "red3", neutral.color = "lightsalmon",
            legend.position = "bottom", text.size=3.5)

and my attempt to incorporate ggrepel:

likert.bar.plot(likert_Prevention.ALL, plot.percents=TRUE, plot.percent.high= FALSE, plot.percent.low=FALSE, 
            center = 3, low.color = "grey72", high.color = "red3", neutral.color = "lightsalmon",
            legend.position = "bottom", text.size=3.5) + labs(geom_label_repel()) + labs(geom_text_repel())

Below is a cropped image of the overlapping percentages currently:

Thank you in advance for your help and time! Please let me know if any clarification is needed, as this is my first post to stackoverflow.

Answer 1

Not the easiest thing to do and I don't know if the flying arrows make it more complicated. Here is something using an example data:

library(likert)
library(ggplot2)
library(plyr)
library(dplyr)
data(pisaitems)

items29 <- pisaitems[,substr(names(pisaitems), 1,5) == 'ST25Q']
names(items29) <- c("Magazines", "Comic books", "Fiction", 
                        "Non-fiction books", "Newspapers")
l29 <- likert(items29)

The default looks like this:

likert.bar.plot(l29, plot.percents=TRUE, plot.percent.high= TRUE, 
plot.percent.low=TRUE, center = 3, low.color = "grey72", high.color = "red3", 
neutral.color = "lightsalmon",legend.position = "bottom", text.size=3.5)

You make a plot without the labels:

p = likert.bar.plot(l29, plot.percent.neutral=FALSE, plot.percent.high= FALSE, 
plot.percent.low=FALSE, center = 3, low.color = "grey72", high.color = "red3", 
neutral.color = "lightsalmon",legend.position = "bottom", text.size=3.5)

Pull out the neg, positive and neutral coordinates:

dat_pos = g$data %>% filter(value>0) %>% 
group_by(Item) %>% arrange(variable) %>%
mutate(ycoord=cumsum(value)-0.5*value) %>%
slice(2:n())

dat_neg = g$data %>% filter(value<0) %>% 
group_by(Item) %>% arrange(desc(variable)) %>%
mutate(ycoord=cumsum(value)-0.5*value) %>%
mutate(value=abs(value)) %>% slice(2:n())

dat_neu = g$data %>% group_by(Item) %>% 
filter(duplicated(variable)) %>% 
mutate(value=2*abs(value),ycoord=0) 

And add that to the plot:

p + geom_label_repel(data=rbind(dat_neu,dat_pos,dat_neg),
aes(x=Item,y=ycoord,label=paste0(round(value),"%")),size=3,force=10)