4 ms·
The p-value (a function used for testing a statistical hypothesis) corresponding to these results was less than 0.001, indicating that the probability that the
by BoardsOfCanada 11y ago
The p-value (a function used for testing a statistical hypothesis) corresponding to these results was less than 0.001, indicating that the probability that the findings were due to chance alone (or that the null hypothesis is true) is less than 0.1%.
Actually, I think it means that IF the null hypothesis is true, there is only a 0.1% chance that we would get this much deviation (or more) from it due to chance. Which isn't the same thing. Still, seems like a significant result of course, but let's be correct about these things.
- axonopathy 11y agoIt looks like the author took this comment to heart.
- eru 11y agoaxonopathy, it seems you are hell-banned for some reason..
- solve 11y agoThe whole science industry should follow the lead of those scientific journals who've banned p-values from their publications. P-values just lead to so much confusion and nonsense. And that's without even considering the scientists who cheat the p-values, which is actually just a tiny part of the problem with p-values.
- _-__--- 11y agoI think that the outright banning of p-values is not the right way to go. The researchers that misuse them and the reviewers that apparently don't understand them are a much bigger problem than the p-value metric itself. What's needed is a better understanding of empirical conditional statistics and the application of that understanding to the degign of experiments and the analysis of the results produced therefrom.
- solve 11y agoAnd the best way to do that is to eliminate the crutch. Many of the more data-centric fields did away with p-values long ago in favor of far more robust approaches to establishing the value of a paper. Look at papers in the data science field. Using p-values in a paper there would get you laughed out of the industry fast. Only problem that p-values ever solved was allowing journal reviewers to be lazier, helping scientists to be lazier, and making writing more sensational press releases easier.
- abandonliberty 11y agoCould you point me to a starting point for understanding the new more robust approaches?
- ds222 11y agoAre there journals for data science? I am not aware of any...
- gwern 11y agoI left a comment on OP several days ago pointing that out; still hasn't appeared nor has it been fixed. Apparently they don't care about being correct.
- bandris 11y agoThe correction seems to be there now.