Enhancing Ability Estimation With Test-Taking Behavior

• In Event: The Use of Process Data in Educational Assessment

Abstract

Many large-scale assessments are low-stakes assessment, but some examinees do not take these tests seriously and display suboptimal testing behavior such as rapidly guessing and omitting items. This study evaluates a method to incorporate test-taking behavior into estimation: using a summary of test-taking behavior as a covariate. Two studies explore the approach, a real data application and a simulation study. The results indicate that when covariates representing response time and/or response propensity are used, information and accuracy increase for individuals of higher-than-average ability. The approach applies existing measurement theory, models, and methods in a novel manner, but may offer significant gains in the accuracy, fairness, and validity of ability estimates.

Author

• Ann Arthur, University of Nebraska