Statistical Properties of Allocation Averages
Abstract
The purpose of this paper is to describe a simulation experiment which was performed to verify empirically Brogden's result concerning the unbiased nature of allocation averages. The theoretical proof for equality of allocation averages, computed from measures of actual performance and from least squares estimates of this performance, requires that optimal assignment be performed over least squares performance estimates for an infinite number of individuals. In the many simulation experiments conducted in BESRL's Statistical Research and Analysis Division, the allocation average computed from performance estimates, rather than from actual performance values, is often the basic statistic from which other experimental results are evaluated. The number of observations on which these results are based is, however, far from infinite. There was thus need to demonstrate that the two kinds of allocation average remain essentially equal when based on relatively small numbers of observations, as is typical of actual experiments.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1968
- Accession Number
- ADA079341
Entities
People
- Elizabeth N. Abbe
Organizations
- United States Army Research Laboratory