Modeling item revisiting behavior in computer-based testing: Exploring the effect of item revisitations as collateral...

Area of Science:

• Educational Measurement
• Psychometrics
• Cognitive Psychology

Background:

• Item revisiting is a frequent test-taking strategy linked to reduced test anxiety and enhanced validity.
• Revisiting patterns offer insights into examinee motivation, effort, and engagement, informing educational policy.
• Understanding item revisiting is crucial for accurate statistical inference in educational assessments.

Purpose of the Study:

• To propose a novel item revisiting model incorporating a monotonicity assumption.
• To evaluate the proposed Bayesian estimation algorithm and assess model fit.
• To demonstrate the impact of ignoring item revisiting on parameter estimates.

Main Methods:

• Development of a new item revisiting model with a monotonicity assumption.
• Three simulation studies to test the Bayesian estimation algorithm, parameter bias, and model fit.
• Analysis of item revisiting sequences alongside response and response time data.

Main Results:

• The proposed Bayesian estimation algorithm performs effectively.
• Ignoring item revisiting sequences leads to biased parameter estimates.
• The novel model demonstrates good fit and effectively utilizes revisiting behavior.

Conclusions:

• Item revisiting behavior provides valuable data for improving parameter estimation precision.
• The proposed model offers a robust framework for analyzing item revisiting in assessments.
• This approach enhances the validity and interpretability of test results.