Current browse context:
stat.AP
Change to browse by:
References & Citations
Statistics > Applications
Title: Bayesian decision theory for tree-based adaptive screening tests with an application to youth delinquency
(Submitted on 18 Jun 2021 (v1), last revised 28 Jun 2022 (this version, v3))
Abstract: Crime prevention strategies based on early intervention depend on accurate risk assessment instruments for identifying high risk youth. It is important in this context that the instruments be convenient to administer, which means, in particular, that they should also be reasonably brief; adaptive screening tests are useful for this purpose. Adaptive tests constructed using classification and regression trees are becoming a popular alternative to traditional Item Response Theory (IRT) approaches for adaptive testing. However, tree-based adaptive tests lack a principled criterion for terminating the test. This paper develops a Bayesian decision theory framework for measuring the trade-off between brevity and accuracy, when considering tree-based adaptive screening tests of different lengths. We also present a novel method for designing tree-based adaptive tests, motivated by this framework. The framework and associated adaptive test method are demonstrated through an application to youth delinquency risk assessment in Honduras; it is shown that an adaptive test requiring a subject to answer fewer than 10 questions can identify high risk youth nearly as accurately as an unabridged survey containing 173 items.
Submission history
From: Chelsea Krantsevich [view email][v1] Fri, 18 Jun 2021 21:40:58 GMT (482kb,D)
[v2] Sat, 23 Oct 2021 06:15:56 GMT (1324kb,D)
[v3] Tue, 28 Jun 2022 01:49:28 GMT (1388kb,D)
Link back to: arXiv, form interface, contact.