[R] graphics

Peter Dalgaard BSA p.dalgaard at biostat.ku.dk
Thu Dec 2 14:48:33 CET 1999

Bill Simpson <wsi at gcal.ac.uk> writes:

> As I understand it, all this discussion is referring to how the graphics
> look on the screen. Maybe it is pointless to mention the following, but
> here I go:
> - what is on the screen (X11) should agree with what is printed
> (dev.print)

This is a bit problematic. I'd say that it should be readable before
anything else. If one wants exact preview, use ghostview & friends.
Also for a pleasant working environment (at least with X11), I like to
use unscaled fonts as far as possible.

> - the default sizes of letters and symbols are too small in
> ouput produced by dev.print. Should be double the size. I always use
> cex=2, but numerous things get buggered up. In my opinion, the current
> behaviour is a bad legacy from S. Default S plots use labels and symbols
> that are too small. Remember that plots submitted to journals
> and book publishers always get scaled down, so in the original version
> they have to be quite large. Take a look at V&R, pp16-17 for plots whose
> axis labels are smaller than surrounding main text.

(As Brian pointed out, pointsize is better than cex for this purpose.
Pardon the pun...)

I do agree that the S defaults are a bit small, but it all depends on
what they are to be used for: 

 - as an intermediate tool in model development &c
 - full-page illustration in book/journal
 - half-page, in-text illustration
 - slide

(with increasing text-to-plot ratios). And it also depends on personal
tastes - e.g. Jim Lindsey pretty squarely want to maximize the plot
area, i.e. minimum legible fonts, if I interpret him correctly.

I think I'm with Brian in saying that the plots in V&R are certainly
not seriously bad, and there's not really any cause for assuming that
the sizes of text on a plot are directly related to those in the
running text (unless one has been foolish enough to use the same
font!) They're readable enough and the whole thing is in reasonable
balance. The labels are a bit too close to the axes and tick marks

Incidentally, I reached out for the nearest stats book from the pre-S
era (Breslow&Day, 1980) and found some pretty ugly clashes between
plots and the surrounding text. It isn't easy...

   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907
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

More information about the R-help mailing list