Online Program

Saturday, February 20
CS20 Statistical Modeling and Bootstrapping Sat, Feb 20, 9:15 AM - 10:45 AM

RealVAMS: An R Package for Fitting a Multivariate Value-Added Model (VAM) (303179)

*Jennifer Broatch, Arizona State University 
Jennifer Green, Montana State University 
Andrew Karl, Adsurgo, LLC 

Keywords: R, accountability, mixed models, evaluation, value-added models

This presentation will illustrate an R package, RealVAMS, for fitting a multivariate value-added model (VAM). RealVAMS is an evaluation tool that can be used to assess teams, teachers, doctors, management, or employees. In education, RealVAMS may be used to estimate a teacher's effect on student achievement beyond what is expected given the student's and their peers' prior performance and demographics. Similarly, RealVAMS can be used to estimate the contributions of hospitals to patient care, players to teams, or employees to a company. In each context, the desired estimates cannot be directly measured, so they are inferred indirectly. The RealVAMS package uses a multivariate generalized linear mixed model to show the relationships among the contributions of the estimate toward different outcomes. The simultaneous modeling of joint continuous and binary outcomes for VAMS was not available prior to the development of the RealVAMS package. This presentation will include a brief discussion of the contributions of the modeling options within RealVAMS and a detailed application of the package to educational data for immediate practitioner use.