192013-03-08 10:17:30localhostconverted to 1.6 markup182008-02-28 11:50:50PeterWatson172008-02-28 11:50:39PeterWatson162008-02-28 11:29:36PeterWatson152007-03-02 10:42:07PeterWatson142007-03-02 10:41:49PeterWatson132007-03-02 10:40:25PeterWatson122007-03-02 10:38:48PeterWatson112007-03-02 10:38:28PeterWatson102007-03-02 10:32:57PeterWatson92007-03-02 10:32:33PeterWatson82007-03-01 16:28:20PeterWatson72007-03-01 16:27:41PeterWatson62007-03-01 16:25:00PeterWatson52007-03-01 16:24:15PeterWatson42007-03-01 16:23:52PeterWatson32007-03-01 16:23:19PeterWatson22007-03-01 16:20:14PeterWatson12007-03-01 16:16:41PeterWatsonThe spreadsheet gives the power for inputted group proportions (which should sum to 1), Type I error and a given total sample size. The expected proportions assume equal group sizes. Power may also be obtained using R. The expected effect size, $$\omega$$, defined below is from Cohen (1977,1988, Chap 7). $$\omega = \sqrt{ \sum_text{1}text{no. groups} \frac{\mbox{(observed proportion - expected proportion)}text{2}}{ \mbox{ expected proportion}}}$$ The thresholds 0.1, 0.3, 0.5 correspond to small, medium and large $$\omega$$ effect sizes.