Assess students' ability to apply quantitative models of voter behavior to predict turnout in U.S. midterm elections using demographic and socioeconomic data. Tasks include identifying appropriate data sources (e.g., ACS, Census, voter files), selecting and justifying a model form (logistic regression, multilevel models, or supervised ML), performing feature selection and diagnostics, evaluating predictive performance and calibration across geographic units (county/precinct), interpreting coefficients and marginal effects, and discussing model limitations, bias, and ethical considerations. Require a brief report with methodology, key results, and policy-relevant interpretation.