Using Response Times for Joint Modeling of Response and Omission Behavior

Area of Science:

• Psychometrics
• Educational Measurement
• Data Science

Background:

• Adequate modeling of missing responses requires understanding nonresponse mechanisms.
• Computer-based assessments provide rich examinee behavior data, including response times and omissions.
• Existing research often models response and omission behaviors separately.

Purpose of the Study:

• To propose a novel framework for simultaneously modeling response and omission behaviors.
• To leverage timing information for both response and omission behaviors.
• To enhance the understanding of nonresponse mechanisms in large-scale assessments (LSAs).

Main Methods:

• Integrating research on item omissions with response time data modeling.
• Developing a unified model for simultaneous analysis of response and omission behaviors.
• Utilizing timing data for both response generation and item omission processes.

Main Results:

• The proposed model offers a deeper understanding of general response and nonresponse behavior.
• It models the time required for examinees to respond or omit an item.
• The framework can account for nonignorable item omissions, a significant advancement.

Conclusions:

• The developed model provides a comprehensive approach to analyzing complex examinee behaviors in computer-based testing.
• It enhances the accuracy of statistical models by incorporating timing and omission data.
• This framework is valuable for improving the validity and fairness of large-scale assessments.