A Multiprocess Item Response Model for Not-Reached Items due to Time Limits and Quitting

Area of Science:

• Psychometrics
• Educational Measurement
• Data Science

Background:

• Existing models for not-reached items assume a single underlying process.
• Missing data in tests can stem from various reasons, including time constraints or reduced motivation.
• Distinguishing these reasons is crucial for accurate performance assessment.

Purpose of the Study:

• To develop a novel model that disentangles different missing data mechanisms for not-reached items.
• To utilize response time data from computerized tests to differentiate between speed-related and quitting-related missingness.
• To provide a more fine-grained understanding of test-taking behaviors.

Main Methods:

• Utilizing response time data from computerized testing.
• Developing a statistical model to simultaneously model different missing data mechanisms.
• Conducting a simulation study to evaluate model estimation.
• Applying the model in an empirical study to analyze test-taking behavior.

Main Results:

• The proposed model successfully distinguishes between missing data due to lack of speed and missingness due to quitting.
• Simulation results demonstrate accurate estimation of the model parameters.
• The empirical study reveals insights into diverse test-taking behaviors.

Conclusions:

• The new model offers a more sophisticated approach to understanding not-reached items in computerized tests.
• Differentiating missing data sources enhances the interpretation of examinee performance.
• This approach provides valuable insights into the complex factors influencing test-taking behavior.