Aella on Twitter: "@Jsevillamol @kushie0 Isn't that just p-hacking? For lots of correlation testing I typically have significantly more stringent standards (like p<0.00001 or something, depending on amount of tests I do) for what p values I consider relevant" / Twitter

Ok my sample size is n=a billion, what random questions from my Date Me survey do you want me to check correlations between form.typeform.com/to/o5sqkyg8

11

3

33

Replying to

can you punch it into something that will analyze strongest correlations and weakest correlations?

1

Replying to

Probably you can do it automatically, I'm an unskilled pleb so I can do it all manually, it'll just take me a while.

2

1

Replying to

and

There are some reasons you might not want to do that. If you systematically pick out the stronger correlations you are bound to pick up noise. en.m.wikipedia.org/wiki/Multiple_.

2

Replying to

and

Isn't that just p-hacking? For lots of correlation testing I typically have significantly more stringent standards (like p<0.00001 or something, depending on amount of tests I do) for what p values I consider relevant

5:25 PM · Apr 24, 2021

Replying to

and

Yep, that is pretty much equivalent what multiple hypothesis comparison adjustment amounts to.

1

Conversation