Hacker News Reader

Honest question: does 'N' usually exclude the control group? Total human participants was 18, 6 for the control group.

I agree that N is too small to warrant such hubbub.

Normally, you just report them separately as in “six subjects were assigned to a control arm, while twelve received treatment.” The groups are often about the same size, though you might put more subjects in the treatment group if you suspect some heterogeneity in the response to treatment.

A better design would test all subjects in both conditions—take a baseline sample, treat everyone and take a second sample, and then take a final sample after you think the treatment has washed out. An even better design would be a cross-over experiment, where half of the subjects receive the treatment first, then are “crossed over” to the control group, while the rather half get a mirror-image treatment.

Reply on news.ycombinator.com