Speaker: Professor Jarad Niemi

Title: Bayesian Mixed Effect Logistic Regression for Virtual Reality Experiment

Abstract: Confined spaces have limited means for entry and exit and thus have additional hazards compared to non-confined spaces. Safety professionals train utility workers to identify hazards in both confined and unconfined spaces. A virtually reality (VR) experiment was conducted where safety professionals and utility workers were placed in confined and non-confined spaces and asked to perform a task. Upon completion of the task, individuals were asked to recall hazards identified in the space. Participants were then allowed to re-entry the VR experiment with the purpose to explicitly identify hazards. A Bayesian mixed effect logistic regression model was estimated to identify differences between safety professionals and utility workers ability to identify hazards between confined and non-confined spaces as well as during the initial task and the subsequent hazard identification. Contrasts are utilized to compare estimated probabilities of detection between safety professionals vs utility workers, confined vs non-confined spaces, and initial task vs subsequent hazard identification.