A Bayesian Generalized Explanatory Item Response Model to Account for Learning During the Test.

Area of Science:

• Psychometrics
• Educational Psychology
• Cognitive Science

Background:

• Existing item response models do not fully capture learning effects during testing.
• Repeated engagement with test items can influence subsequent performance.

Purpose of the Study:

• Introduce a novel explanatory item response model.
• Account for learning occurring during psychometric testing due to item operation repetition.
• Investigate different learning contingencies (non-contingent, contingent, differential contingent).

Main Methods:

• Extension of the operation-specific learning model.
• Development of a general model formulation with special cases.
• Bayesian framework for model estimation and evaluation.
• Simulation study to assess parameter recovery and model selection.
• Empirical study for real-data application.

Main Results:

• The proposed model successfully accounts for learning effects in psychometric testing.
• Simulation results demonstrate the reliability of estimation and evaluation methods.
• Empirical data analysis validates the model's applicability in detecting learning.

Conclusions:

• The new model provides a robust framework for understanding and quantifying learning within psychometric assessments.
• The model's flexibility in handling different learning types enhances its practical utility.
• This research contributes to a deeper understanding of test-taking as a learning process.