On 7 Sep 2001 20:39:09 -0700, [EMAIL PROTECTED] (Melvin Yap) wrote:
> I'm running a 2 x 5 fully-within ANOVA design, where A has 2 levels
> and B has 5 levels. After finding a significant interaction, I looked
> at the simple main effect of A at each level of B. My question is: how
> do you compute the effect size of each of the simple main effects? I
> was wondering if I could simply do a paired sample t-test of each of
> the significant simple effects and then compute d from the
> t-statistic.
Paired t-tests are the usual followup for repeated measures,
and you can get an effect size from one. If you need to draw
distinctions based on p-values, that would be the 'consistent'
choice. That is about the only reason for that choice.
Even if your standard deviations and correlations are in a
narrow range, you might want to use a second measure of
effect: something that will be exactly the same between all
your followups. You could use your raw scores if they have
a familiar unit, such as 'seconds'. Or use, say, the a 'd'
based on the between-subject standard deviation, which
would be a single, overall standard.
--
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
