The best measure of reliability is the standard error of measurement. It's
really the same as the within-subject standard deviation (SD you expect to
get when retesting a subject many times), but you take out any change in
the mean between trials. For any reasonable sample size and two trials,
Pearson correlation plus between-subject SD gives the within-subject SD via
a simple formula. ANOVA is better. Mixed modeling is best of all, and
the only way to go when you have several sources of error. Don't forget
precision of your estimates: this is one place where p values are
unquestionably useless. I have written heaps at
http://newstats.org. Click on the various links to reliability.
Will
At 08:06 22/03/01 +0300, Awahab El-Naggar wrote:
>Dear Colleagues
>I have been using "test-retest" method for calculating reliability by
>applying the Pearson Product Moment (PPM) analysis...
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
