|
Current Lab
Members |
|
William Stout, Founder |
 |
Professor Stout's research interests are
cognitive diagnosis, test fairness, and nonparametric methods for determining
latent trait dimensionality. His work in cognitive diagnosis concerns applications of the Fusion model for cognitive and skills diagnosis. In the area of test fairness he has developed a latent trait model for explaining the phenomenon of differential item functioning, and has worked with others to construct SIBTEST and its variations to test for DIF in a variety of situations. Professor Stout has also constructed a test of latent trait unidimensionality, known as DIMTEST. In addition, he has studied the properties of a weaker version of unidimensionality known as "essential unidimensionality" and has supervised the development software to implement several tools for dimensionality assessment that were created by Professor Stout or other lab memebers. |
|
|
|
|
|
|
Jeff Douglas, Co-director |
 |
Professor Douglas' interests include large sample theory for latent variable models, cognitive diagnosis, and item response models for longitudinal data. His work in asymptotics concerns joint estimation of structural and incidental parameters, and how this might be affected by considerations of unidentifiability. Current work with Xiaodong Li focuses on joint consistency of maximum likelihood estimation. Professor Douglas also collaborates with other present and former lab members on cognitive diagnsosis. His work in that area focuses on test construction, goodness-of-fit, model selection, polytomous attributes, higher-order latent traits, computerized adaptive testing and applications of Markov chain Monte Carlo. He has also applied item response modeling to analysis of quality-of-life assessment in clinical trials, and worked with Dr. Chen Wang to develop methods for analyzing quality-of-life in clinicals trials having survival as the primary endpoint. |
|
|
|
|
|
|
Louis Roussos, Co-director |
 |
Text of Louis Roussos' interests... |
|
|
|
|
|
|
Bob Henson, Research
Assistant |
 |
Bob's primary research interests are in educational and psychological testing. He is currently conducting research on several topics in skills diagnosis. These include utilization of Kullback-Liebler information for test construction, as well as K-L based indices of reliability. In addition, he consults on software development and applications of Markov chain Monte Carlo in this are. Bob also has an interest in the psychology of gambling. He has taken part in an NIH funded study using advanced psychometric techniques to
reveal different characteristics and patterns of behavior among gamblers. |
|
|
|
|
|
|
Tzur Karelitz, Research
Assistant |
 |
Text of Tzur Karelitz's interests... |
|
|
|
|
|
|
Xiaodong Li, Affiliated
Student |
 |
Xiaodong Li conducts research in the areas of quantile regression, large sample theory, and spatial data analysis.
Among other topics in quatile regression, Xiaodong is interested in developing quantile regression methods for time series data.
His work in psychometrics concerns large sample theory, including current work on the consistency of nonparametric maximum likelihood estimation in item response models. Xiaodong also consults on contamination of groundwater, which requires techniques of spatial data analysis and prediction in time and space. |
|
|
|
|
|
|
Juan Moran, Research
Assistant |
 |
Text of Juan Moran's interests... |
|
|
|
|
|
|
Jonathan Templin, Research
Assistant |
 |
Jon's research interests focus on psychometrics as applied to
educational measurement. Currently, he is working on expanding general
models for skills diagnosis to include polytomous items and polytomous
attributes. He is also involved in the development of efficient methods
for estimation of latent attribute vectors with MCMC. Jon has worked on
and is interested in models for social networks, judgment and decision
making, and robust estimation in structural equation modeling. |
|
|
|
|
|
|
Xueli (Shirley) Xu, Research
Assistant |
 |
Xueli's research in psychometrics includes methods for computerized adaptive testing and methods for equating using nonparametrically estimated item response models. In the area of computerized adaptive testing, Xueli has studied methods for efficient attribute classification in skills diagnosis. She has also developed methods and accompanying theory for conducting CAT with nonparametrically estimated item characteristic curves. The distinguishing feature in that setting is that derivatives of the ICCs are generally unavailable, which implies that the maximum information criterion may not be used. Xueli has also developed methods of nonparametrically estimating equating functions by using software that fits splines with monotonicity constraints. |
|
|
|
|
|
