Context : UsabilityStuff

JakobNielsen points out you only need to test with a small number of users ( ~5) to find the most egregious usability bugs

The original only need 5 :

15 for card sorting

for InformationArchitecture :

: NB : assumes you are getting users to organize a hierarchy. (Contrast AWebsiteIsNotATree)

Larger number because of the difference between evaluation method (extracting info. about the faults of the already existing) and generative method (asking for more informational input from those being tested.) The latter requires more users (more information) ... I wonder why? (See also : GenerateAndTestInParallel)

Hmmm, but this is lousy science : Thus, if you see a single quantitative study that contradicts all that's known from qualitative studies, it's prudent to disregard the new study and assume that it's likely to be bogus. But when a quantitative study confirms what's already known, it's likely to be correct, and you can use the new numbers as decent estimates, even if they're based on less data than you would ideally like.

(WarpLink) CreativeNetwork. Clearly the number you need to test with isn't somehow related to any social group size factors. But it might be fun to speculate about deeper possible commonalities eg. what size of group is big enough to bring a sufficient sampling of a data-set, but small enough to share a model without synchronization costs spiraling. OTOH this may be nuts ...