错误条无法在条形图上正确绘制

Question

I am having trouble getting my error bars to plot correctly. Here is my data frame:

> d
         Date Plate.1 Plate.2 Plate.3 Plate.4 Plate.5 Plate.6 Plate.7 Plate.8 Sum Average        SE Treatment
1  2014-10-15       2       2       0       5       8      11      11       0  39   4.875 1.6304852         1
2  2014-11-12       5       4      11       6       7       2       9       6  50   6.250 0.9955257         1
3  2014-12-11       1       0       0       0       0       0       0       1   2   0.250 0.1636634         1
4  2015-02-11       0       0       4       0       2       1       0       1   8   1.000 0.5000000         1
5  2015-03-09       0       0       0       0       0       0       0       0   0   0.000 0.0000000         1
6  2014-10-15      30      22      24      19      10      16      19      42 182  22.750 3.4369318         2
7  2014-11-12       8       5       4       9      12      23      14      10  85  10.625 2.1207942         2
8  2014-12-11       0       5      21      19       2       9       4       0  60   7.500 2.9215945         2
9  2015-02-11      16      11       5       8       5      17       0       0  62   7.750 2.3126207         2
10 2015-03-09       0       0       0       1       0       0       0       0   1   0.125 0.1250000         2

I am using this code to produce a grouped barplot:

ggplot(d, aes(x=Date, y=Average, fill = Treatment)) +
  geom_bar(position="dodge", stat="identity") +
  geom_errorbar(aes(ymin=Average-SE, ymax=Average+SE),
                width=.2,                    
                position = position_dodge(.9))

For some reason the error bars are not plotting correctly and the graph looks like this:

Hopefully the solution is a simple one, however it is eluding me. Thanks in advance for any help!

Answer 1

Updated in light of your comment - for this type of plot, you can assign the value for position_dodge to a variable and use that for all subsequent plots. I hit on the value of 20 with a little trial and error; if the other data you're working with is very different numerically, you may need to fine tune this.

d <- data.frame(Date=c("2014-10-15", "2014-11-12", "2014-12-11",
                "2015-02-11", "2015-03-09", "2014-10-15",
                "2014-11-12", "2014-12-11", "2015-02-11",
                "2015-03-09"),
                Average=c(4.875, 6.250, 0.25, 1, 0,
                22.75, 10.625, 7.5, 7.75, 0.125),
                Treatment=c(rep(1, 5), rep(2, 5)),
                SE=c(1.6304852, 0.9955257, 0.1636634, 0.5, 0,
                3.4369318, 2.1207942, 2.9215945, 2.3126207, 0.125))
d$Date <- as.Date(d$Date)
d$Treatment <- factor(d$Treatment)
library(ggplot2)
pd1 <- 20
ggplot(d, aes(x=Date, y=Average, fill=Treatment)) + 
    geom_bar(width=pd1,
             position=position_dodge(pd1),
             stat="identity") +
    geom_errorbar(aes(ymin=Average-SE, ymax=Average+SE),
                  width=pd1 / 2,                    
                  position=position_dodge(pd1))

giving:

I think the problem you were having is that using position="dodge" allows ggplot2 to calculate an 'optimal' dodge. However you need to see/state the value explicitly to pass it as an argument to geom_errorbar .

Setting the width in geom_bar to the same value as position_dodge in both geom_bar and geom_errorbar (here pd1 <- 20 ) overcomes this.