[R] plotting (a) confidence intervals (b) standard error

[Ben Bolker]

  This is a FAQ (although you've gotten farther than most in figuring out
how to do it for yourself):

http://www.R-project.org/nocvs/mail/r-help/2000/0451.html
http://www.R-project.org/nocvs/mail/r-help/2000/2289.html
http://www.R-project.org/nocvs/mail/r-help/2000/3326.html
http://www.r-project.org/nocvs/mail/r-help/2001/2050.html

  In short, you can either use arrows() with angle=90 to construct your
own error bars, or look at the code for plotCI contained within

http://www.zoo.ufl.edu/bolker/R/{windows,linux}/bbmisc.{tar.gz,zip}
(pick your platform).

 Ben Bolker

On Thu, 29 Nov
2001, Shravan Vasishth wrote:

> Hi all,
>
> I'm building a plot of the values in tmeant (below) against positions 1 to
> 5, using matplot.
>
> tmeant looks like this:
>
>         case1     case2
> pos1 861.8466  818.5909
> pos2 961.2841  976.3466
> pos3 878.6080 1262.8523
> pos4 950.8011 1129.6080
> pos5 968.1080 1063.3920
>
> I also have lower (object tl) and upper (object tu) bounds on the
> confidence intervals as follows:
>
> tl:
>
>         pos1     pos2      pos3     pos4     pos5
> case1 761.4777 872.8367  756.7253 806.7503 849.9667
> case2 735.1803 878.2557 1058.0583 991.7346 895.0864
>
> tu:
>
>          pos1     pos2     pos3     pos4     pos5
> case1 962.2155 1049.731 1000.491 1094.852 1086.249
> case2 902.0015 1074.437 1467.646 1267.481 1231.698
>
> My R code is given towards the end of this message; it plots the values in
> tmeantmatrix against positions 1-5, and puts the confidence interval (CI)
> bars at the right points using tl and tu.
>
> However, I can't figure out how to get the CI bars to have those
> delimiters (a shape like "I"), all I get is a straight line through the
> relevant point.
>
> A related question is: I also have standard errors in a matrix called tse:
>
>           pos1     pos2     pos3     pos4     pos5
> case1 49.76907 43.85765  60.4369 71.42927 58.58170
> case2 41.36010 48.63950 101.5495 68.36609 83.45631
>
> Is there some way to plot these (instead of CI bars) without having to
> first compute lower and upper bounds (i.e., without having to specify x0,
> y0, and x1, y1 in "segments") for each standard error? I guess this would
> amount to having a way to refer to each point already plotted, and saying
> something like: take each point as the dimension of an error bar, and put
> it at the relevant point (I'm thinking of something like gnuplot, which
> has a nice semantics for this: if you give it three columns to plot and
> specify that you want error bars, it will use the third column as the
> dimensions of the error bars).
>
>
> #----------------The code for the graph with CI bars------------------------
>
> tmeantmatrix<- matrix(tmeant,ncol=2)   # make tmeant a 2-column matrix
>
> # plot using matplot:
>
> matplot(c(1:5), tmeantmatrix, xlab = "Position", ylab = "Mean Reading Time
> (msecs)", main = "Test")
>
> # add lines connecting the points:
>
> matplot(tmeantmatrix, pch = 1:2, type = "o", lty=1:2, lwd=3, col =
> rainbow(ncol(tmeantmatrix)),add=TRUE)
>
> # add CI bars:
>
> xzero <- t(matrix(rep(1:5,2),ncol=2))
>
> xone <- xzero
>
> segments(x0=xzero, y0=tl, x1=xone, y1=tu)
>
> #-------------------end of code------------------------------
>
> Many thanks in advance,
>
>

-- 
318 Carr Hall                                bolker at zoo.ufl.edu
Zoology Department, University of Florida    http://www.zoo.ufl.edu/bolker
Box 118525                                   (ph)  352-392-5697
Gainesville, FL 32611-8525                   (fax) 352-392-3704

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._