Files in this item



application/pdf111004-IACAT.pdf (2MB)
Slides and Presentation NotesPDF


Title:Rapid CAT Prototyping with OSCATS: The Open-Source Computerized Adaptive Testing System
Author(s):Culbertson, Michael J.
Subject(s):Computerized Adaptive Testing
Item Response Theory
Abstract:One obstacle to the widespread, rapid development and adoption of CAT is the relative lack of readily available CAT software. Most current CAT systems are either home-grown or proprietary, requiring new users either to be strong programmers or pay licensing fees. A free, off-the-shelf CAT library would allow more developers to enter the CAT field without the redundant burden of writing and debugging computer code for common CAT functions. This presentation describes how to use the new Open-Source Computerized Adaptive Testing System (OSCATS) for rapid prototyping and testing of CAT systems. OSCATS provides code for common CAT routines, which can be mixed-and-matched to customize a CAT with different algorithms and item models for evaluation of CAT systems via simulation. Moreover, OSCATS is extensible, allowing developers to incorporate novel or custom routines easily. OSCATS can be controlled from C, Python, Perl, PHP, Java, and MATLAB. Although OSCATS is still under development, it already has the capability to implement some of the most popular CAT algorithms and models for Item Response Theory and Diagnostic Classification. As a demonstration, an OSCATS-based program will be described to investigate the effect on ability estimation error of answering the first few items correctly or incorrectly under maximum Fisher information, Kullback-Leibler, and a-Stratified item selection.
Issue Date:2011-10-04
Citation Info:Paper presented at the Second annual meeting of the International Association for Computerized Adaptive Testing, Pacific Grove, Ca, Oct 3-5, 2011.
Genre:Conference Paper / Presentation
Publication Status:unpublished
Peer Reviewed:not peer reviewed
Date Available in IDEALS:2011-10-12

This item appears in the following Collection(s)

Item Statistics