Colloquiums, Lectures, and Seminars

Upcoming events

Helen Huston - Statistics 501 Presentation

Helen Huston will give a talk on the following statistical literature and problems project in partial fulfillment of requirements for the Master of…
Add to my Calendar 2021-03-08 23:00:00 2021-03-04 08:03:55 Helen Huston - Statistics 501 Presentation Helen Huston will give a talk on the following statistical literature and problems project in partial fulfillment of requirements for the Master of Science in Statistics     “Bayesian Item Response Theory Modeling” Abstract: Anomia, or word finding difficulty, is a primary diagnostic feature of stroke-induced aphasia, a language disorder resulting from damage to the speech centers of the brain. The cognitive processes that are typically impaired reflect a patient’s ability to correctly name objects. Naming deficits often persist after other symptoms subside while naming accuracy is a good indicator of overall severity. Item response theory uses statistical modeling to make inferences about underlying latent traits like naming ability on the basis of observed patterns of behaviors. The data used must meet strong assumptions to ensure model validity. In 2015, Dr. Fergadiotis, Dr. Kellough and Dr. Hula used item response theory to model the Philadelphia Naming Test within a frequentist framework. We refit these models within a Bayesian framework in STAN and analyzed their performance. Follow-up analysis used posterior predictive model checking methods to assess model assumptions. This talk focuses on the methodology used to build the STAN models, the tools available to assess overall model performance, and the tenability of the three key item response theory assumptions. Keywords: Aphasia, Philadelphia Naming Test, probabilistic program, STAN, Bayesian inference, identifiability, posterior predictive model checking.  Under the direction of Dr. Daniel Taylor Rodriguez      Time: Mar 8, 2021 03:00 PM Pacific Time (US and Canada)   Join Zoom Meetinghttps://pdx.zoom.us/j/88497515804   Meeting ID: 884 9751 5804, Password: 219789   Join Zoom Meeting https://pdx.zoom.us/j/88497515804 Kathie Leck, Fariborz Maseeh Department of Mathematics + Statistics, mathstatgrad@pdx.edu Kathie Leck, Fariborz Maseeh Department of Mathematics + Statistics, mathstatgrad@pdx.edu America/Los_Angeles public