Individual Test Patterns: Are We Correcting for Guessing in the Wrong Direction?

Abstract

Certainly cogent arguments can be made for regressing the ability estimates obtained from unusual response patterns inward from their observed values. We have tried to elucidate some of them here. Yet experience has taught us that guessing is an activity that occurs frequently, and accounts for much (though by no means all) of the anomalous responses observed. Therefore we would like to propose a middle way. It would seem that a proper scheme to corrent for unusual response patterns would be in-between the traditional correction (that assumes that all anomalies are from guessing) and one that professes complete ignorance of the cause of such unusual response patterns. The method that we prefer regresses the 'corrected' ability estimates inward, although not usually as far as using an uncorrected raw score.

Document Details

Document Type
Technical Report
Publication Date
Jan 01, 1981
Accession Number
ADP001398

Entities

People

  • Howard Wainer

Organizations

  • Educational Testing Service

Tags

DTIC Thesaurus Topics

  • Continents
  • Geographic Regions
  • North America
  • Virginia

Readers

  • Applied Combinatorial Optimization and Logic Circuit Design.
  • Artificial Intelligence
  • Regression Analysis.