Skip to content
Prev 326 / 7420 Next

Are likelihood approaches frequentist?

Ben Bolker escreveu:
If the concern here is about the subjectivity in choosing a reference 
value (or distribution) to make a decision, it is not different from 
p-values for sampling-distribution inference (thanks Rub?n) or from the 
a priori distributions for Bayesian inference.

I think the central issue is to distinguish between the evidence value 
of a given data set, which is only a piece of our decision(1), and the 
decision itself. I made this point in a previous msg: 
https://stat.ethz.ch/pipermail/r-sig-ecology/2008-May/000148.html

1- Decision = adoption of a given model or hypothesis, for explicative, 
predictive or descriptive purposes.


Cheers

Paulo  Prado

--