Speaker: Thomas Metzger, Department of Statistics, OSU
Title: Bayesian Model Selection with Latent Group-Based Effects and Variances with the R Package slgf
Abstract: In the first part of my talk, I will present the R package slgf which enables the user to easily implement my linear modeling approach to detect latent group-based regression effects, interactions, and/or heteroscedastic error variance through Bayesian model selection. I will focus on the scenario in which the levels of a categorical predictor exhibit two latent groups, treating the detection of this grouping structure as an unsupervised learning problem by searching the space of possible groupings of factor levels.
In the second part, I will discuss approaches to integrating statistical consulting into a well-rounded graduate-level statistics curriculum. Such programs often emphasize technical instruction in theory and methodology but can fail to provide adequate practical training in applications and collaboration skills. I argue that a statistical collaboration center (“stat lab”) is an effective mechanism for providing graduate students with the necessary training in technical, non-technical, and job-related skills. I provide evidence of its positive impact on students via analyses of a survey completed by 123 collaborators who worked in the Laboratory for Interdisciplinary Statistical Analysis (LISA) between 2008–15 while it was housed at Virginia Tech.
