Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.



How to construct a frequency based test without a sufficient statistic
Posted:
Oct 24, 2012 1:07 PM


If I have a distribution that has no sufficient or ancillary statistics, is there a way to construct valid tests from the maximum likelihood estimate other than using Bayesian methods and testing over the posterior distribution?



