Using Stein's Estimator to Predict Universe Scores from Obtained Scores
Abstract
The purpose of this paper is to introduce and apply a recently developed statistical method for estimating true (population) scores from observed (sample) scores. Provided that at least three scores are available, this method overall will give more accurate true score estimates than the individual maximum likelihood estimates (MLE), regardless of the true abilities of the examinees (Efron and Morris, 1977). The method can be used without knowledge of the (Bayesian) prior distribution, and normality of the true scores being estimated need not be assumed. The theoretical and practical implications of the method extend beyond psychological measurement to the very foundations of statistical inference and have caused some tumult in that discipline during the past decade.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1978
- Accession Number
- ADA077967
Entities
People
- Frederick H. Steinheiser Jr.
- Stephen L. Hirshfeld
Organizations
- U.S. Army Research Institute for the Behavioral and Social Sciences