3 ms·
Looks like that to me too. The interesting thing is that, with 40 variables the expectation is that they would find 2 spurious correlations. They got unlucky a
by marcosdumay 3y ago
Looks like that to me too.
The interesting thing is that, with 40 variables the expectation is that they would find 2 spurious correlations. They got unlucky and only got to publish 1.
Also, with about 50 control variables that nobody has any idea how related they are to the outcome, I'm sure I could mess with the numbers until I find some 10 spurious correlations on the set. They quit too early.
Still, those studies are important to lead future research into the actual relevant questions. It's journalists that shouldn't touch them.