On Interim Cognitive Diagnostic Computerized Adaptive Testing in Learning Context.

Area of Science:

• Educational Measurement and Psychometrics
• Learning Analytics
• Computerized Adaptive Testing

Background:

• Traditional cognitive diagnostic computerized adaptive testing (CD-CAT) assesses students at a single time point.
• The learning context requires assessments that adapt to changing student mastery profiles and evolving attribute sets.
• Existing CD-CAT designs do not adequately address the dynamic nature of learning environments.

Purpose of the Study:

• To propose and evaluate interim CD-CAT designs suitable for educational settings.
• To investigate the impact of learning history and attribute hierarchies on interim CD-CAT performance.
• To enhance the application of CD-CAT for continuous learning monitoring.

Main Methods:

• Developed two interim CD-CAT designs: one with static prior attributes, another allowing attribute change.
• Proposed an individual prior incorporating learning history and population models for interim CD-CAT.
• Extended the generalized deterministic inputs, noisy, "and" gate (GDINA) diagnostic index (GDI) to handle item hierarchies.

Main Results:

• Simulations demonstrated that interim CD-CAT employing an individual prior outperforms methods using population priors.
• The extended GDI effectively accommodates item hierarchies, providing insights into item selection.
• The proposed methods show significant promise for monitoring student learning progress dynamically.

Conclusions:

• Interim CD-CAT designs offer a viable approach for continuous assessment within learning environments.
• Individualized priors are crucial for optimizing interim CD-CAT accuracy in dynamic learning contexts.
• This research pioneers the use of CD-CAT for effective learning monitoring.