/edu/ - Education

>what is a controlled study
Where you have two research subjects, the test group and the control group

You change something from the default state for the test group, and leave the control group the same

>when is the correlation significant enough to be taken seriously,
Depends on what you're looking for, whether you want to minimize false positives or negatives on the likelihood of the effect being real etc

Sometimes spotting spotty research is just a matter of looking for tells

If what the abstract claims is different or hyping the results the research deserves extra scrutiny in some cases for example

It's all shit, or at least most of it is. Google "the null ritual".

>>682
Do you mean this? I started reading it but gave up after a few pages because I don't know anything about the statistics it speaks of.

>>721
Yes, that is exactly what I was referencing. There is an episode of a podcast called "Not Related" that goes into it as well, although the actual demonstration of the issue is in the paper.

The methods that it talks about in there are exactly the kinds of statistics you will need to learn if you are interested in experimental design, or at least if you want to understand the quantitative methods used in academia rn. I was taught null hypothesis testing in my undergrad, but fortunately I also studied enough math to see that it is largely bullshit. I mean, it does have an application but the way it is applied is quite literally cargo cult behavior.

You may also want to look into the replicability crisis in science, as well as the new paradigms in theoretical physics that will hopefully displace the dead-end trend towards quantizement that we've seen taken to ridiculous extrememes in the past few decades. I don't have much faith that this will actually be realized, bourgeois academia is in decay, but the theoretical groundwork is there (quantum gravity is one such theory which is kinda controversial but there are others as well).