ISOQOL 2004 Symposium
"Stating the Art: Advancing Outcomes Research Methodology and Clinical Applications"
June 27-29, 2004
Boston Park Plaza Hotel
Boston, MA, USA

Monday, June 28, 2004

10:45 am - 12:45 pm
Concurrent Sessions
Session 1: Modern Measurement Theory and Applications
Presenters: Ronald K. Hambleton and Chih-Hung Chang
Chair: Kathy Wyrwich
Proffered Papers: TBD

Item Banking and Computerized Adaptive Testing in Health Outcomes Assessment
Chih-Hung Chang, PhD
Assistant Professor, Northwestern University, Chicago, IL

In the management of chronic diseases, the increased interest in patient-reported outcomes (PROs) assessment has been encouraging. The unmet need is a clinically meaningful, psychometrically robust, user-friendly, dynamic and real-time PROs assessment system. Recent advances in modern test theory (i.e., item response theory; IRT) and computer technology make it possible to deliver PROs questions to arrive at an accurate estimate of patient health status with fewer targeted questions. This presentation will address issues related to item bank and computerized adaptive testing development, including 1) methodology to link or equate commonly used PROs instruments; 2) methods to discern and model the structure of higher-order dimensions of self-reported health; 3) steps and challenges to establish an item bank; and 4) technical considerations of developing a computerized adaptive testing (CAT) platform for real-time health status assessment and reporting.

Choosing Amoung Item Response Theory Models
Ronald K. Hambleton, PhD
Professor and Co-Director, Center for Educational Assessment, University of Massachusetts, Amherst, MA

Item response models have been applied to educational tests and credentialing exams for nearly 30 years, and are receiving wide use today in the Quality of Life research area. These applications are wide-spread because of shortcomings with classical approaches to instrument development (e.g., item statistics that are sample dependent, respondent scores that are test dependent) as well as the attractiveness of many features of item response theory (IRT) models that include (1) item parameter invariance (over samples of respondents), (2) ability parameter invariance (over samples of questions), (3) flexibility in instrument development and score equating, (4) unique estimates of measurement error for respondents and (5) enhanced score reporting options. IRT models come in many varieties (over a 100) to handle (1) unidimensional as well as multidimensional data; (2) binary, polytomous, and continuous response data; and (3) ordered as well as unordered categorical responses. The purposes of this presentation will be to (1) provide a comprehensive review of several of the popular IRT models, and (2) offer criteria that can be used in choosing IRT models for particular applications.