Creating 95% confidence band with ggplot - r

Question

I'm having trouble plotting a 95% confidence band with ggplot.
Here is my code:
> my.data
scar response.rate
1 HTS 0.88
2 HTS 0.56
3 HTS 0.56
4 HTS 0.82
5 HTS 0.10
6 HTS 0.47
7 HTS 0.83
8 HTS 0.60
9 Linear 0.83
10 Linear 0.56
11 Linear 0.79
12 Linear 0.55
13 Linear 0.70
14 Linear 0.50
15 Keloid 1.00
16 Keloid 0.83
17 Keloid 1.00
18 Striae Distensae 0.33
19 Striae Distensae 0.33
ggplot(my.data, aes(scar, response.rate))+geom_point()+geom_smooth()
The output it produces:
When I use numbers for the scars, I am able to produce the following:
Any way I can produce the same graph with the scar type instead of numbers?

Answer 1

This method you create the CI's manually, then plot them using geom_errorbar()
library(ggplot2)
# Creating some data:
my.data = data.frame(c(rep("H",5),rep("L",5),rep("K",5),rep("S",5)), rnorm(20,1,.5) )
names(my.data) = c("scar", "response.rate")
# Standard error function
foo = function(x){sd(x)/sqrt(length(x))}
# Creating CI's manually
my.aggs = cbind(aggregate(response.rate ~ scar, data = my.data, FUN = foo),
aggregate(response.rate ~ scar, data = my.data, FUN = mean))
names(my.aggs) = c("scar","se","","means")
# Plotting
ggplot()+
geom_point(data = my.data, aes(as.factor(scar), response.rate)) +
geom_errorbar(data = my.aggs, aes(scar, ymin=means-1.96*se, ymax=means+1.96*se), width=.1)
# Alternative method that doesn't include points
library(gplots)
plotmeans(response.rate ~ scar, data = my.data)