3 ms·

> N=42 This is noise.

• p_j_w 5y ago
  1. There's more to determining if an effect is in the statistical noise than just sample size. Effect size, for example, also comes into play. 2. Not every study will, or even should, try to be the final word on some matter. If you come up with some novel hypothesis in social science, you're not going to jump into a 2,000 person study.
• FabHK 5y ago
  It might be noise, but you can't tell from N=42 alone. It is possible to find "significant" results with small groups (my back-of-the-envelope calculation suggests that the proportion of success between the groups must differ by more than 30 percentage points to be p<0.05). They might well have done the maths and might have highly significant p-values. Note the supplementary materials [1]. I don't understand them without the paper, but they sure do have a lot of "*** p < 0.001" results. (Entirely unrelated side note: A shame that Sci-Hub currently does not add papers due to litigation...) [1] https://journals.sagepub.com/doi/suppl/10.1177/0956797621993111 https://journals.sagepub.com/doi/suppl/10.1177/0956797621993...
• NotSwift 5y ago
  42 happens to be the answer to the "Universal Question About Life, The Universe and Everything". [0] [0] https://en.wikipedia.org/wiki/The_Hitchhiker%27s_Guide_to_the_Galaxy https://en.wikipedia.org/wiki/The_Hitchhiker%27s_Guide_to_th... PS: I get it that some people will down-vote this because jokes like this do not really stimulate an intellectual discussion. But reading something by Douglas Adams is a really good idea.