Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Bayes Factors to Get the Most Out of Linear Regression: Practical Guide Using R (thewinnower.com)
62 points by mindcrime on Oct 16, 2015 | hide | past | favorite | 2 comments


I disagree with this approach. It does not prove that the null hypothesis is true, only that it is better than an alternative model.

Rather than trying to prove that a quantity is identical to zero, I would rephrase the problem as trying to determining what the quantity is, and giving a confidence interval for it.

When the confidence interval has become sufficiently small, and includes zero, then we may say that any difference is neglible.


This looks really great. It seems just about as easy to use as R's built-in lm functionality.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: