Combining Cognitive Diagnostic Computerized Adaptive Testing With Multidimensional Item Response Theory.

Area of Science:

• Educational Measurement
• Psychometrics
• Computerized Adaptive Testing (CAT)

Background:

• Current Dual-CAT models use unidimensional IRT, limiting application to multidimensional tests.
• There's a need for adaptive testing that assesses both general ability and finer-grained skills simultaneously.
• Existing methods struggle with the complexity of multidimensional assessment within adaptive testing frameworks.

Purpose of the Study:

• To develop and validate a multidimensional Dual-CAT (Dual-CAT) for improved applicability in assessments.
• To propose and evaluate novel item selection methods tailored for multidimensional adaptive testing.
• To enhance the accuracy of ability estimation and control item exposure rates in multidimensional CAT.

Main Methods:

• Development of new item selection algorithms for multidimensional Dual-CAT.
• Conducting simulation studies to assess the performance of proposed methods.
• Validation using a real-world item bank study to evaluate estimation accuracy and exposure rates.

Main Results:

• The developed multidimensional Dual-CAT framework is effective for complex assessments.
• Proposed item selection methods demonstrate superior performance compared to traditional approaches.
• The study confirmed improved estimation accuracy and controlled item exposure rates.

Conclusions:

• The multidimensional Dual-CAT offers a more applicable and accurate approach to ability testing.
• Novel item selection strategies are crucial for optimizing multidimensional adaptive testing.
• Future research should explore further advancements in multidimensional CAT methodologies.