R help - Apr 2010 - Error bar

Hi there,
Using the code below I generated the plot attached. The error bars at day -3, 2
and 4 appear are larger. I was thinking if there is a way I could make
all the error bars to be of the same size. I don't know if that makes sense.
If not, then, is there a way I can plot only these 3 largest errorbars at
day axis (day = -3, 2 and 4) and omit the rest of the error bars.
I would be much obliged for any help
Warmest
Ogbos
temp<-split(y,x)
 means<-sapply(temp,mean)
 stderr<-function(x)sqrt(var(x)/length(x))
 p<-sapply(temp,stderr)
 n<-sapply(temp,length)
 ciw<-qt(0.975,n)*p/sqrt(n)
  library(gplots)


 day = -5:5
 par(mfrow = c(3,1))

 plotCI(day,means, ciw,gap = 0.05, sfrac= 0.005, type = "l",cex = 1.0)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: plot.png
Type: image/png
Size: 4972 bytes
Desc: not available
URL:
<https://stat.ethz.ch/pipermail/r-help/attachments/20100406/e117e880/attachment.png>

On 04/07/2010 03:49 AM, ogbos okike wrote:
> Hi there,
> Using the code below I generated the plot attached. The error bars at day
> -3, 2 and 4 appear are larger. I was thinking if there is a way I could
make
> all the error bars to be of the same size. I don't know if that makes
sense.
> If not, then, is there a way I can plot only these 3 largest errorbars at
> day axis (day = -3, 2 and 4) and omit the rest of the error bars.
> I would be much obliged for any help
Hi Ogbos,
It doesn't make much sense to force the error bars to the same size, for 
if you do, they don't tell you anything useful. Obviously days -3, 2 and 
4 had more widely varying temperatures than the others on the plot. You 
could plot the daily means with:

plot(day,means,ylim=c(0,20),type="b",...)

and then set all of the small standard errors to NA:

ciw[ciw<5]<-NA

and use one of the error bar functions:

library(plotrix)
dispersion(day,means,ciw)

I use dispersion because I know that it ignores NA values. Other error 
bar functions probably do this as well.

Jim