[monkellipse]

40 subjects.

[saas_sam]

Sample size can be small if the controls are good and the effect is large. 40 could be way too many for some studies, way too few for others. "40 subjects" means nothing by itself.

[monkellipse]

Fair enough! My intuition would be the more subjects the better. I realize there’s probably a whole field of study on this subject, and intuition is often wrong :)

[Cenk]

For an 18 month trial 40 subjects is not bad

[claytonjy]

And an effect size this large shouldn't need many subjects...though 28% strikes me as too good to be true.

[cosentiyes]

and published in American Journal of Geriatric Psychiatry with an impact factor of 4...