You're reading indeed too much into that ^^.
What I meant is that as rule of thumb 20-25 samples for a non-parametric test should often give more than sufficient results (don't ask me where I got that from though).
The small differenc between 2nd+3rd are probably unlikely. The main point is the difference between 1st+4th, and I guess the chance there is higher.
EDIT: Should be mentioned that I've always worked with more tightly controlled stuff, and that the normal sample size in the lab was often 3 per group. Didn't work with any population cohorts etc, and also nothing medical, so was never a problem lol.
What I meant is that as rule of thumb 20-25 samples for a non-parametric test should often give more than sufficient results (don't ask me where I got that from though).
The small differenc between 2nd+3rd are probably unlikely. The main point is the difference between 1st+4th, and I guess the chance there is higher.
EDIT: Should be mentioned that I've always worked with more tightly controlled stuff, and that the normal sample size in the lab was often 3 per group. Didn't work with any population cohorts etc, and also nothing medical, so was never a problem lol.