Under what conditions do the Bayesian and frequentist hypothesis testing lead to the same conclusion (rejection or acceptance of a hypothesis)? For example, if the test concerns statistical independence between two categorical variables?
$\begingroup$
$\endgroup$
4
-
$\begingroup$ In large samples they will agree approximately and this will also be the case if the prior is uninformative regarding the null hypothesis. $\endgroup$ – Michael R. Chernick May 31 '17 at 19:35
-
$\begingroup$ So, if I have no prior beliefs, I can use either one and end up with the same result? $\endgroup$ – whamalai May 31 '17 at 20:44
-
1$\begingroup$ If you have no prior belief why bother with the Bayesian approach? $\endgroup$ – Michael R. Chernick May 31 '17 at 20:54
-
$\begingroup$ That was exactly the point! I have just begun to study the Bayesian approach and the papers always advertise how much better it is. I wanted to check if a non-believer would ever benefit from the Bayesian approach in hypothesis testing. I understood the situation is different in parameter estimation. $\endgroup$ – whamalai Jun 1 '17 at 16:06
Add a comment
|
