FAQ/dunp62013-03-08 10:17:39localhostconverted to 1.6 markup52010-07-22 13:12:35PeterWatson42010-07-22 13:07:45PeterWatson32010-07-22 13:07:08PeterWatson22010-07-22 12:52:46PeterWatson12010-07-22 12:47:37PeterWatsonSample sizes for testing differences in an unpaired t-testA popular effect size for t-tests (unpaired or paired) is Cohen's d which represents the standardized difference between two group means. The highlighted rows in the table below represent small, medium and large effect sizes as suggested by Cohen (1988). This table gives sample sizes required in each of the two groups. 90% 80%
d 1-tailed 2-tailed 1-tailed 2-tailed
0.1 1708 2096 1234 1568
0.2 428 525 310 393
0.3 191 234 138 175
0.4 108 133 78 100
0.5 70 85 51 64
0.6 49 60 36 45
0.7 36 45 27 34
0.8 28 34 21 26
0.9 23 27 17 21
1.0 19 23 14 17
1.1 16 19 12 15
1.2 13 16 10 13
1.3 12 14 9 11