On the preservation of local asymptotic normality under information loss (Q1113230)

This paper considers a situation where there are unobservable random variables, and where what is actually seen are other variables that are less informative than the unobservable ones. It is shown that if the unobservable random variables satisfy certain conditions, such as the LAN conditions, then the observable random variables will also satisfy the same requirements. The results are proved under negligibility conditions imposed on the case where they are independent. For the proofs the paper applies the asymptotic decision theory of the first author. Several interacting examples are studied in detail.