[EMAIL PROTECTED] (Simon, Steve, PhD) wrote in message news:<[EMAIL PROTECTED]>...
> Have you graphed the data? If the times for one condition are clearly
> way better than all the others, then perhaps a graph is all that you
> need. Boxplots should work nicely, or a simple scatterplot might also be
> good.

Given Pingu's description of the nature of his problem (trying
to motivate choice of one criterion to use), this may well be
all that is necessary. Simple pictures often have the advantage
of being readily understood. 

I'll second this advice, with the comment that you need to
be careful not to summarise out the potential problems
mentioned by the other posters, at least until you have
checked for their presence. (For example, avoid using a
display that hides a learning effect unless there isn't one.)

Glen
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to