Welcome to the UIUC IRT Modeling Lab website!
The IRT Modeling Lab investigates the development and application of IRT models to data across:
- Cognitive ability tests
- Licensing exams
- Job attitude scales
- Personality inventories
- Vocational interest measures
Data from such diverse sources requires a corresponding diversity in IRT models. To this end, we utilize models with varying numbers
of parameters to fit dichotomously and polytomously scored item responses. We also use parametric and nonparametric models characterizing
dominance and ideal-point response processes.
This website will help you learn about IRT models and their use. We begin with a general introduction and then describe data preparation
that is needed prior to analysis. IRT models assume that a single dominant latent trait underlies responses to all items on a test
or scale; it is important to investigate dimensionality prior to IRT analysis. If the item set is found to be sufficiently unidimensional
for IRT, various models should be considered for use. After selecting a model and estimating its parameters, it is important to check
model-data fit. In this way, potential inadequacies in the fitted model can be examined and the data analyst can determine whether
a more general model is needed.
IRT helps researchers and practitioners address many measurement problems. One of the most important is differential item functioning
and differential test functioning (DIF/DTF). Here we examine whether items function the same way for two or more groups. DIF/DTF methodology
was originally developed to assess whether cognitive ability test items functioned similarly for racial, ethnic, and gender groups.
Many other important questions can be addressed with this analysis, such as
- Do the measurement properties of personality items for samples of research participants generalize to job applicants?
- How rapidly do the measurement properties of cognitive ability test items change over time?
- Do the measurement properties of test or scale items change when the items are adapted for administration in other languages?
IRT also facilitates test construction. For example, IRT provides tools to enable assembly of parallel forms of tests or the creation
of a test the maximizes measurement precision in a specific ability range (e.g., near the cutting score for a licensing exam).
Faculty
Fritz Drasgow received his Ph.D. in quantitative psychology from the University of Illinois at Urbana-Champaign in 1978. He was an
Assistant Professor at Yale University's School of Organization and Management and returned to the University of Illinois in 1982,
where he has been an Assistant Professor, Associate Professor, and Professor of Psychology and of Labor and Industrial Relations. He
is currently a Director of the University of Illinois Center for Human Resources Management, which is a joint University/Industry
venture designed to address important HRM problems.
Drasgow's research focuses on psychological measurement and the application of quantitative methods to important practical problems.
Much of his research has been concentrated on the identification of individuals who may have been mismeasured by an assessment tool.
He has also studied methods for the detection of items with different measurement properties across various groups of people and languages.
Computerized tests and assessments have also been the focus of much of Drasgow's research. For example, he has used multimedia computer
technology to assess social and interpersonal skills not easily measured by paper-and-pencil tests. Drasgow has also worked on a
large project developing and testing a comprehensive model of sexual harassment in the workplace.
Drasgow is a former chairperson of the American Psychological Association's Committee on Psychological Tests and Assessments and the U.S. Department
of Defense's Advisory Committee on Military Personnel Testing. He is currently the chair of the Department of Defense and Department
of Labor's Armed Services Vocational Aptitude Battery Norming Advisory Group. Drasgow has also provided consultation on testing and measurement
issues to a variety of organizations in the private sector. Drasgow is a member of the editorial review board of nine journals, including
Applied Psychological Measurement, Journal of Applied Psychology, and Personnel Psychology.
Alumni
Sasha Chernyshenko (University of Canterbury)
Siang Chee (David) Chuah (AICPA)
Stephen Stark (University of South Florida)
Wayne Lee (Valtera)
Krista Mattern (College Board)
Patrick Wadlington (Hogan Assessment)
Students
Ben-Roy Do
Scott Bedwell
Bradley Brummel
We hope you find the website interesting and helpful. Please e-mail IRT tutorial web site related
questions to Ben-Roy Do.
IRT Tutorial
|