Modeling Incorrect Responses to Multiple-Choice Items with Multilinear Formula Score Theory.
Abstract
Multilinear Formula Score theory provides powerful methods for solving psychological measurement problems of long standing. In this paper the question of information in incorrect option selection on multiple choice items is addressed. Multilinear formula scoring (MFS) is first used to estimate option characteristic curves for the Armed Services Vocational Aptitude Battery Arithmetic Reasoning test. Accurately estimated curves are obtained for real and simulated data. Then the statistical information about ability is computed for dichotomous and polychotomous scorings of the items. Moderate gains in information are obtained for low to slightly above average abilities. The dichotomous and polychotomous models are then compared for their relative performances in appropriateness measurement. The rates of detection of some types of aberrance responding were more than 100% higher for optimal polychotomous model index. Consequently the MFS polychotomous model provides opportunities for better testing by allowing more accurate ability estimates, improvements in the theory and practice of item writing, and more powerful appropriateness measurement.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1987
- Accession Number
- ADA187887
Entities
People
- Bruce Williams
- Fritz Drasgow
- Gregory Candell
- Mary E. Mclaughlin
- Michael V. Levine
Organizations
- University of Illinois Urbana–Champaign