Foundations for the Mathematical Notion of Information in Item Response Theory and Robust Ability Estimation,
Abstract
Item response theory is an attempt to apply statistical theory and methdology to inferences about a subject's ability. This article describes an attempt to develop the concept of influence in item response theory that parallels a similar concept in robust theory in order that the methods of robust theory be better applied through the results may be ad hoc. Some of the progress in this direction has resulted in a new class (h-estimators) of robust estimators of ability and a systematic approach to the derivation of bias and variance formulas. Not only may these formulas be used for constructing confidence intervals for a subject's ability, they may be used to compare the efficiency of the h-estimators to that of the standard maximum likelihood estimator.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1981
- Accession Number
- ADP001396
Entities
People
- Douglas H. Jones
Organizations
- Educational Testing Service