You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more
25 out of 28 introductory psychology books define or explain statistical significance wrong 
https://journals.sagepub.com/doi/full/10.1177/2515245919858072 …. I formatted and re-arranged their data to show which books commit the most frequent mistakes. Raw data: https://osf.io/cnd2v/ pic.twitter.com/Zw9st5oJGH
The authors provide teaching materials with a very nice illustration of what statistical significance is: https://osf.io/qg9t2/ . It is a procedure to control errors if you were to do the same exp + analysis many times, assuming each time that H0 is true no matter prior data.pic.twitter.com/aoCnpqMUn3
P-values and statistical significance is from the frequentist tradition. This is what it's like to be a frequentist:https://twitter.com/jonaslindeloev/status/1143937661889077248 …
What if we tallied the number of statistical fallacies committed in intro-sections in scientific articles?pic.twitter.com/eFmuKsKsiy
Personally, I find that Bayesian inference + focus on parameter estimation (rather than binary model comparison) is a better way to go. It is harder (and not without it's own fallacies), but there seem to be no easy fixes.
Isn't what you call the "inverse-probability error" fallacy the same as the "validity" fallacy? Also, I don't understand why you call the latter "validity".
Sorry, thought that you were a co-author. s/you/they.
Yeah, I was a bit confused about the naming of the fallacies as well. They had them from another paper, so the blame goes way back, it seems :-)
Maybe they were victims of the fallacy that P(A) + P(not A) ≠ 1? 
Haha, now you only need to give it a confusing name! The "non-unitary universal probability fallacy"?
Or the "holism fallacy"? The whole is unequal to the sum of its parts.
That's actually pretty good.
And check out http://psycnet.apa.org/fulltext/2017-24635-002.html … "A Systematic Review of Bayesian Articles in Psychology: The Last 25 Years", many don't even discuss priors, have enough info to reproduce, or do sensitivity analysis. Seems like a crisis in psychology than in frequentism or bayesianism.
Saving this for after the weekend. Thanks for sharing! Will take a look at the frequency with which these errors occur (that number is sure to rise for Bayes as popularity grows)
The two first columns say the same
One is definition of significant. The other is implied definition in explanation of result.
I don't get it. Wikipedia says the p-value is the probability that a given result would be obtained if the null hypothesis is true. The H0 is always that there is no relation between the things. Isn't that the same thing as saying it's due to chance?
Oh, nevermind, I just realized my own mistake. If a result obtains when the H0 is true, that means it is not caused by the relation in the alternative hypothesis. But it could still be due to something other than chance.
Yes! It's convoluted making inference about models when you assume one to be true a priori.
Neat although I am having trouble finding the breakdown of book by fallacy on the OSF link.
It's data_updated.csv
Thanks. I had found that but don't know what the columns (e.g., df.f1) represent.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
