Contingent and Predictive Analytics
Robert H. Meyer, Jeff Dominitz, & Soobin Kim
This memo provides an introduction into contingent analytics, a methodology and framework for evaluating the short and long-term consequences of student outcomes such as student achievement, course selection and performance, and attendance. The contingent analytics framework attempts to bridge two methodologies that are quite different, but related: (1) predictive analytics and early warning systems and (2) evaluation methods. This memo shows how the two frameworks are different but connected. The proposed research builds on our work with a consortium of school districts in a single state where we have developed predictive analytics models to produce forecasts of high school and postsecondary outcomes for students currently in high school, middle school, and elementary school, as early as third grade. Practitioners and policymakers may desire to use predictions such as these to identify students at-risk of dropping out and to trigger interventions and supports designed to increase the likelihood of staying in school through high school graduation. As such, this type of predictive model could be incorporated in Early Warning Systems (EWS) that have been adopted in school districts across the nation.