TitleEstimating ability and item-selection strategy in self-adapted testing: A latent class approach
Publication TypeJournal Article
Year of Publication2004
AuthorsRevuelta, J
JournalJournal of Educational and Behavioral Statistics
Volume29
Number4
Pagination379-396
Publication Languageeng
ISBN Number1076-9986 (Print)
Accession Number2005-00264-002
Keywordsestimating ability, item-selection strategies, psychometric model, self-adapted testing
Abstract

This article presents a psychometric model for estimating ability and item-selection strategies in self-adapted testing. In contrast to computer adaptive testing, in self-adapted testing the examinees are allowed to select the difficulty of the items. The item-selection strategy is defined as the distribution of difficulty conditional on the responses given to previous items. The article shows that missing responses in self-adapted testing are missing at random and can be ignored in the estimation of ability. However, the item-selection strategy cannot always be ignored in such an estimation. An EM algorithm is presented to estimate an examinee's ability and strategies, and a model fit is evaluated using Akaike's information criterion. The article includes an application with real data to illustrate how the model can be used in practice for evaluating hypotheses, estimating ability, and identifying strategies. In the example, four strategies were identified and related to examinees' ability. It was shown that individual examinees tended not to follow a consistent strategy throughout the test. (PsycINFO Database Record (c) 2007 APA, all rights reserved)