Scholion.ai

Optimal Diagnostic Questionnaires

George T. Duncan

Operations Research · 1975 · Volume 23 · Issue 1

DOI: 10.1287/opre.23.1.22

Abstract

This paper develops dynamic-programming and other shortest-route algorithms for optimizing sequentially patterned questioning procedures, called diagnostic questionnaires. It uses graph-theoretic methods and solves certain combinatorial problems associated with the Catalan numbers. The paper obtains general properties of optimal questionnaires, including an ordering relation, a convexity result, and the determination of essentially complete classes of questionnaires. It explores a logarithmic charging scheme in detail and obtains a minimal result. Finally, the paper examines the relation with Huffman coding of information theory, and a simple method of Shannon efficiency.

Enable JavaScript for citation tools and additional paper interactions.