 |
 |
Hi friends, I have a question to you on a specific sample size calculation. In one of our projects in a sample (as one group) a mean value on a dependent variable is being expected of 10 with a sd of 20, where the values represent percentages. It appears a sample size calculation already has been done on those data, a minimal number of observations which is needed to show, that the 10% is significantly different from 0, is 62. So I tried to reproduce that number and have been using the One sample test on means with nonparametric correction (Wilcoxon) of the program NCSS/PASS; I yielded a size of 28. Now our sample very likely has a skewed distribution (as it concerns a percentage) and the applied method assumes a symmetrical distribution at least. So my feeling says the used method is not quite correct, but I could not find a better alternative in PASS. How would you do such a sample size calculation and what software would you use? Regards - Jim. -- . . . . . . . . . . . . . . . . Jim Groeneveld, MSc. Biostatistician Science Team Vitatron B.V. Meander 1051 6825 MJ Arnhem Tel: +31/0 26 376 7365 Fax: +31/0 26 376 7305 [log in to unmask] www.vitatron.com [common disclaimer]