ggplot2 shifting bars to only overlap in groups

-----Original Message-----
From: hyiltiz at gmail.com
Sent: Thu, 12 Feb 2015 22:08:36 +0800
To: r-help at r-project.org
Subject: [R] ggplot2 shifting bars to only overlap in groups

Hi all,

I have four factors for a continuous time variable along with its
confidence interval. I would like to produce a publication quality error
bar chart that is clear to understand. For now, I used colors, x axis
position, facets and alpha level to distinguish them.

I would like to overlap each pairs of bars with the same color a bit as a
group, but not overlap each and every bars with each other.

Here is a minimal example:

N = 32
df<- data.frame(gender=gl(2,1,N, c("male","female")),
          direction=gl(2,2,N, c("up","down")),
          condition=gl(4,4,N, c("c1","c2","c3","c4")),
          location=gl(2,16,N, c("east","west")),
          t=rnorm(N, 1, 0.5),
          ci=abs(rnorm(N, 0, 0.2)))
pp <-
  ggplot(df, aes(x=gender, y=t, fill=condition, alpha=direction)) +
  facet_grid(location~.) +
  geom_bar(position=position_dodge(.9), stat="identity", color="black") +
  geom_errorbar(aes(ymin=t-ci, ymax=t+ci),
                width=.2,                    # Width of the error bars
                position=position_dodge(.9)) +
  scale_alpha_discrete(range= c(0.4, 1))
pp



In the attachment, I have added the output figure, while manually editing
the SVG file to make the lower-left group of bars to make them as I
wanted.
(The spacing in between each pair is not necessarily required.)


?Best
?
========================
He who is worthy to receive his days and nights is worthy to receive* all
else* from you (and me).
                                                 The Prophet, Gibran
Kahlil