Application of Binary Searching for Item Exposure Control in Cognitive Diagnostic Computerized Adaptive Testing

Area of Science:

• Educational Measurement
• Computer Science
• Psychometrics

Background:

• Cognitive diagnosis represents a new era in educational assessment, distinct from item response theory (IRT) by modeling discrete latent traits.
• Item selection in cognitive diagnostic computerized adaptive testing (CD-CAT) is framed as a dynamic searching problem.

Purpose of the Study:

• To adapt the binary searching algorithm for efficient item selection in CD-CAT.
• To introduce novel methods that overcome the limitations of previous CD-CAT item selection algorithms, such as low efficiency and user-dependent weighting.

Main Methods:

• Adapted the binary searching algorithm to address item selection in CD-CAT.
• Developed two new algorithms: Stratified Dynamic Binary Searching (SDBS) for fixed-length CD-CAT and Dynamic Binary Searching (DBS) for variable-length CD-CAT.

Main Results:

• The proposed SDBS and DBS algorithms effectively achieve multiple item selection goals simultaneously.
• Simulation studies demonstrated that these new methods perform comparably or better than existing CD-CAT algorithms.

Conclusions:

• Binary searching algorithms offer a more efficient and user-friendly approach to item selection in CD-CAT.
• The SDBS and DBS algorithms represent a significant advancement in cognitive diagnostic assessment, enhancing the adaptive testing process.