Statistics Colloquium by William Ruggiero '19, Wednesday, April 3

Not All Grades are Created Equal: Using Bayesian Latent Trait Formulation and Item-Response Modeling to Accurately Weight GPA by William Ruggiero ’19, Wednesday, April 3, 1:10-1:50 pm, Stetson Court Classroom 105, Statistics Colloquium

Abstract:  When comparing students’ academic achievement, it is difficult to determine who performed better as GPA is heavily biased towards courses taken and individual professor grading. In a 1997 paper, Duke University professor Valen E. Johnson used Bayesian Latent Trait Formulation and Item Response Modeling as a way to eliminate the inequities associated with GPA-based measures. This talk will examine Johnson’s methods and results as well as discuss other possible problems this method could be applied to.