Having problems with message search?
Hello all,
I'm wondering if anyone knows of good resource or method for determining sample size needed to detect a given effect with a given power level in a repeated measures design (with single factor or multiple factors and levels). For instance, I would like to be able to make a statement such as: "a sample size of N = ? is sufficient to detect a 5% difference in variable x with power level of 0.8." The only relevant message I found at the Sport Science website was from Nov. 5, 2004, but I did not see anything in the message regarding how to determine sample size in a repeated measures analysis a priori. I have also consulted the text, Statistical Power Analysis by Murphy & Myors, 2nd Ed, 2004, but it only mentions information regarding how and why repeated measures analyses are more powerful than between-groups designs, with no mention of how to actually calculate sample size. Finally, Design Sensitivity, by Lipsey, 1990, provides an equation for determining the effect size with paired samples (which includes the mean difference b/t treatment and control populations, correlation between paired observations in those populations, and the common standard deviation of the two populations). This effect size is then used with a Power chart (intended for independent samples) to determine the sample size. I am not satisfied with this method, especially for the small samples (N = 8 or 10) that we typically use for our research.
Any help would be greatly appreciated.
------------------------------------------------
Jonathan Wingo
Department of Exercise Science
University of Georgia, USA
Send Email
