Oh no! They screwed up both p-value and p-hacking. (Usually p-hacking is done by cherry-picking data or cherry-picking the method for analyzing the data, not by manipulating data.)https://twitter.com/JennCVeilleux/status/1188973832633233408 …
-
which both falls under removing data from the entirety - and thus manipulation of the data IMHO. designing the tests so that no bad result data is created at all would be different... but then again, they had way less space than a tweet to "define" it :)
I think we'll have to disagree on the meaning of "manipulation of data". Of course it's wrong to misrepresent the context of an experiment by, for example, failing to mention the 19 similar experiments one conducted with inconclusive results, but I wouldn't call that manip of dat