Moments in mixture models with non-identically distributed observations
Wilfried Seidel · Communication in Statistics- Theory and Methods · 1998
For each be a known family of distributions and X i an observation of the mixture of with an unknown mixing distribution F. By using a standard approach, an estimator [mcirc]l(n) for the lth moment of F can be constructed from If each is a location family, simple conditions can be given for the asymptotic normality of [mcirc]l(n). The variance of [mcirc]l(n) can be estimated in (at least) two different ways; the estimators coincide in the case of identically distributed observations.