Opher Donchin
@opherdonchin
Followers
727
Following
2K
Statuses
3K
Neuroscientist studying motor control and the physiology of the motor system.
beer sheva, israel
Joined January 2011
@kagererlab Sure. CIs are HDIs with a flat prior. p values are an inverse posterior predictive p values with a delta prior. Still, I'm not sure about chi squared tests or ANOVA and having slides or lecture notes or a Jupyter notebookwould be lovely.
0
0
1
Hitting the nail on the head!
@LNuzhna The diagram from Roger Penrose I think expresses the circular dependence conundrum of science nicely. The question isn't "is X real?" but "what does 'real' mean?"
0
0
1
So Bayesians: I'm giving one lecture on Bayes in a full course on advanced regression. What is your go-to example of easy to do with Bayesian stats but hard or impossible in frequentist?
0
0
3
RT @NiskyLab: 1/🚀 Excited to share our new research! 🎉 This study, led by the talented Hanna Kossowsky Lev, explores the fascinating inter…
0
1
0
RT @reziliusReza: Spontaneous recovery is a fundamental feature of motor adaptation, implying at least two timescales of learning. A new pa…
0
10
0
@neuro_data @beenwrekt @learnfromerror I still want to know what is the thing for which it was designed where it works. Can you tell me, please?
1
0
0
@learnfromerror @neuro_data @beenwrekt This seems somewhat close to the idea of severe testing (which I don't pretend to understand in any depth):
@EikoFried P values dichotomize results but in a way that is unhelpful. It might be useful to have "almost certainly just noise" and "potentially interesting". Or maybe "quite probably meaningful" and "more data needed". We've got something unhelpful in the middle.
0
0
0
0
0
0
@neuro_data @beenwrekt @learnfromerror I'm not. My personal breaking point was when I learned there were 5 different Fs defined for unbalanced mixed model ANOVAs each with a different abstruse logic. Frequentist stats felt like theories of angels dancing on the heads of pins.
0
0
0
0
0
0
@neuro_data @beenwrekt @learnfromerror It's worse than that. I've been teaching it for 20 years. You'd think I might have gotten curious. The mumbo jumbo is always too much for me. Still, you believe it helps you. That's not nothing!
1
0
1
@neuro_data Nope. Happy to admit that I either never learned or didn't understand. I always thought that they are supposed to reveal if your results are "real". I don't think they do that. If it's something else, I'm interested. Really.
1
0
0
@neuro_data I'll bite. What is that thing it was designed for that it works at? Is it maybe like a phlogiston meter?
1
0
0
0
0
0
@ryancbriggs I'm the chair of our IRB. This article was unsatisfying. It confuses anecdote and data. It presents (perhaps legitimate) complaints as a cost/benefit analysis. These issues deserve a more serious treatment. Does one exist?
0
0
0