Stochastic Curtailment of Questionnaires for Three-Level Classification: Shortening the CES-D for Assessing Low,...

Area of Science:

• Psychological assessment
• Psychometric methods
• Clinical screening

Background:

• Efficient clinical assessment tools are crucial to minimize respondent burden.
• Computerized adaptive testing methods require efficient classification for practical application.
• The Center for Epidemiologic Studies-Depression Scale (CES-D) is a widely used depression screener.

Purpose of the Study:

• To extend Stochastic Curtailment (SC) for three-class classification.
• To evaluate the efficiency and accuracy of three SC methods (SC-EP, SC-SOR, SC-MOR) for clinical screening.
• To compare performance under regular and optimal item ordering.

Main Methods:

• A post hoc simulation study using CES-D item scores from a large sample.
• Comparison of SC via Empirical Proportions (SC-EP), SC via Simple Ordinal Regression (SC-SOR), and SC via Multiple Ordinal Regression (SC-MOR).
• Evaluation of respondent burden (number of items) and classification accuracy.

Main Results:

• Under regular item ordering, SC-SOR and SC-MOR required fewer items than SC-EP with comparable accuracy.
• Under optimal item ordering, SC-SOR demonstrated efficiency gains without significant capitalization on chance.
• SC-MOR showed substantial capitalization on chance under optimal ordering.

Conclusions:

• SC-SOR is an efficient and accurate method for three-class clinical screening.
• SC-SOR offers a favorable balance between respondent burden and classification accuracy.
• The choice of SC method and item ordering impacts screening efficiency and reliability.