Comment by lutusp
19 hours ago
> they say people felt 23pc less animosity. Assuming their measurements are okay, what would the statistical power of this experiment ? I dont think they report a null hypothesis.
It's a psychology study, a study from a field whose results famously fail to be replicated roughly 2/3 of the time, even when they meet the 0.05 P-factor criterion that assures publication.
Also, many modern psychology studies don't have control groups, and don't consider the null hypothesis. Too much trouble.
Also also, a paywalled study funded by taxpayers. Wasn't this practice supposed to have been stopped?
The problem is with 0.05 criterion. In particle physics, the hypothesis tests are done two ways with the alternate hypothesis flipped to be the null hypothesis and it has a lower threshold of passing as in - we believe when theres very strong evidence else we dont. Atleast thats my read from the Higgs boson paper - https://higgsml.lal.in2p3.fr/files/2014/04/documentation_v1....
> The problem is with 0.05 criterion.
That's certainly one problem -- many have argued that it's too easy to meet this evidentiary standard, which explains why so many weak, non-replicable psychology papers get published.
You comment comparing psychology to physics is apt -- the evidentiary standard in hard science fields is much higher.