"On behalf of fellow statisticians, I would like to thank you for all that you have done for the advancement of statistics."
to which Gosset replied: "Oh, that's nothing - Fisher would have discovered it all anyway." Stella Cunliffe (1976)
If you use "spurious correlation" synonymously with "correlation that is not causal", you equate "correlation" with "causation". #statsmisconception#LitTip wrt #spuriouscorrelation: https://t.co/J12St9j3cu
"Statistical Literacy – it’s all in the communication” - our new review published today explores what statistical literacy is and what it means for how statistics are communicated to the public https://t.co/JhAla2y7WH
@ajordannafa@edelmann_domi @gtallen89 2/ That being said, "modern social science" is using Fisherian testing, too (often mixed with NP-testing). See e.g. the use of p-values here: https://t.co/m32FtJ0JzX
The use is Fisherian because the p-values serve as a measure of strength of evidence against the null.
@ajordannafa@edelmann_domi @gtallen89 1/ You made general claims about a particular statistical measure (paraphrased: "p-values are tied to NP-testing“, „p-values are tied to causal claims") Both claims are wrong, as shown above, regardless what a particular field ("modern social science") does in practice.
I will be teaching an introduction to Philosophy of Statistics this spring, ed 4-6:30. Follow us on https://t.co/aUJ6CM4m5d where all the materials will be available. Some classes may be zoomable midway through.
@lakens@smreeteemehta Many mathematicians do something even more inefficient: They write down their talks and then write it down *again* and *in front of the audience*. However, with the level of precision required, I think that's not a bad idea. I reckon it might be similar in philosophy.
You might want to argue that these causal models are incorrect. You are probably right! The point is not that they are “correct”. The point is that causal models help to clarify the conclusions we can draw. Having a clear causal definition of “bias” helps in this discussion.
... but it is certainly influenced by considerations of the relative convenience and the empirical facts."
Source: Neyman, 1937, Outline of a Theory of Statistical Estimation Based on the Classical Theory of Probability, p. 336 #LitTip
"In theoretical work the choice between several equally legitimate theories is a matter of personal taste only. In problems of application the personal taste is again the decisive moment, ...
This is a very good thread. A very easy thing we can do as senior researchers is to send good work to journals that do not play this game. I sent my Sample Size Justification paper - and the many citations the preprint was getting - to @CollabraOA for example