[R] BoxPlots, 1 Way ANOVA and Non-Statisticians.

[Chuck Cleland]

Carlos:
   You could just give the same values for the ylim argument in each 
boxplot

  par(mfrow=c(1,2), las=1)
  boxplot(rnorm(10), rnorm(10), rnorm(10), ylim=c(-3,3))
  boxplot(rnorm(10, 0.5, 1), rnorm(10, -.5, 1), rnorm(10), ylim=c(-3,3))

Carlos J. Gil Bellosta wrote:
> I am working in a project and I have a number of observations belonging 
> to several classes. Using a 1 Way ANOVA, I have rejected the equality of 
> means hypothesis with a very small p-value. However, the people I have 
> to present my results to are not statisticians and they are not very 
> likely to be much impressed by a 1.32434e-12 like number/thing.
> 
> Therefore I have decided to make to boxplots, one for the actual data 
> and the second one for simulated data where the equality of the means 
> holds so that the difference in the distributions can be visually 
> appreciated.
> 
> The problem is that, for the simulated values, being more regular, the 
> range of variation is smaller and, therefore, the heigth of the window 
> where their boxplot is drawn is also smaller. As a result, the scales of 
> the two boxplots are not the same and part of the appeal of the visual 
> approach is lost in the way.
> 
> My question is, is there a way to make two different boxplots within a 
> "common window"? (Or rather, a common size window or, more concretely, 
> so that it spans over the same range on the vertical axis).

-- 
Chuck Cleland, Ph.D.
NDRI, Inc.
71 West 23rd Street, 8th floor
New York, NY 10010
tel: (212) 845-4495 (Tu, Th)
tel: (732) 452-1424 (M, W, F)
fax: (917) 438-0894