LogFAQs > #891755150

Anteaterking posted...

Also just curious CovXY, but have you heard of the GRIM test?

No I hadn't, that's a pretty interesting idea though. Seemingly going to catch minor errors in reporting, but occasionally if someone is really lazy at fabricating data.

Bok_Choi posted...

sample size is a real thing
but also who you're sampling is important

9/10 samping technique is much more relevant to talk about than sample size.

But there's 2 points I want to mention, which all revolve around valid criticism in science:
1. Complaints about sample size must be justified statistically. Sampling is a counter intuitive procedure, and the required sample size for most studies is much lower than most people would intuit.
2. Valid criticisms always come in the form of: "this was incorrect, which could lead to your conclusion x even when the true underlying result is y."

"This was incorrect," is a meaningless criticism, unless you can link it directly to the manner in which it'll influence the conclusion. So if the study is sampling from Americans, you need a pretty good reason that the result won't generalize to el salvadorians.

The Admiral posted...

Complaining about sample size is how you dismiss the daily examples of far-left stupidity on college campuses.

No, I complain about biased sampling, not sample size. If you sample for a quality, you are guaranteed to only have that quality in your sample, given the quality exists within your population.
---

=E[(x-E[x])(y-E[y])]