The topological foundation and some properties of the mixed estimator (Q1584186)

A common phenomenon in statistical inference is the absence of a sufficient amount of information to estimate a parameter or to test a hypothesis with a desirable level of accuracy. In these cases a synthesis of different sources of information on the same subject can provide sharp estimates and powerful tests. Focusing on estimation, there are two basic categories of information concerning the parameter of interest which lead to the sample and non-sample estimator. A sample estimator is a result of a typical statistical design or a well presented study in the literature. A non-sample estimator can be provided from another study on the subject where no information on the size and the distribution of the sample was recorded. Such an estimator can also be formed as an expert opinion whose knowledge on the subject is based upon experience rather some data analysis results. Combining information from different sources can lead to sharper estimates than the ones produced by each source separately. A sample with one or more non-sample estimators will be considered in the process of estimating a real parameter. In this paper, the mixed estimator is derived and its topological foundations are examined.