How to compute SE of an observed variable and plot errorbars in R using plyr and ggplot.

errorbars-se-ddply.R

require(plyr)
require(ggplot2)

# say dat is a data.frame with the predictor x, grouping variables z1, z2, and z3 (factors or discrete numeric values)
# and the observed values are in the variable y.  then I would do:

dat.summ <- ddply(dat, .(x, z1, z2, z3), summarise,
                  y.mean=mean(y), y.se=sd(y)/sqrt(length(y)))

# this is equivalent to
dat.summ <- ddply(dat, .(x, z1, z2, z3), function(d) {
  with(d, data.frame(y.mean=mean(y), y.se=sd(y)/sqrt(length(y))))
})
       
# summarise just provides a nice shortcut.

# then to plot w/ errorbars:
ggplot(dat.summ, aes(x=x, group=z1:z2:z3, y=y.mean)) +
  geom_line() +
  geom_errorbar(aes(ymin=y.mean-1.96*y.se, ymax=y.mean+1.96*y.se))