PROJECT SUMMARY The proper evaluation of public policies on health and other outcomes is vital for determining whether a policy was effective and also whether it should be continued or implemented in other regions. For example, Philadelphia, along with six other U.S. cities, has implemented excise taxes on sweetened beverages with the goal of decreasing consumption of added sugars which have been associated with obesity and other serious health conditions. However, what if consumers try to avoid the tax by purchasing beverages in neighboring regions that did not implement the tax? Consumers may also be encouraged by the higher price of sweetened beverages to purchase other high-sugar snacks and drinks rather than purchase the taxed beverages. Without properly understanding and accounting for these complex but common situations (called interference), researchers may grossly overestimate the health impact of public policies. Further, policy evaluations often rely on survey data for more granular insights on the causal mechanisms of a policy, like individual-level changes in consumption. However, these data may not represent populations of interest, such as those that are underserved. To address these challenges we propose new causal estimands, specify unique identification assumptions, construct doubly robust semi-parametric estimators, and apply them to multiple studies of the Philadelphia beverage tax. This proposal addresses critical gaps in the evaluation of public policies. Our work is organized under the following aims: Aim 1 will develop innovative causal methods to estimate heterogeneous policy effects under diverse direct and indirect (e.g., spillover) exposures and generalize and transport these heterogeneous policy effects to settings with different exposure, sociodemographic, and geographic contexts. Aim 2 will develop novel causal methods to examine substitution effects using a potential outcomes framework under a set of newly estab- lished identification conditions. Aim 3 will develop a novel synthetic control approach that allows the assessment of individual-level outcomes obtained via surveys. The first two aims will apply these new methods to evaluate the effects of the Philadelphia beverage tax on purchasing behaviors using volume sales data and transport these effects to other U.S. cities. Here, we will also estimate beverage tax effects in regions that implemented similar taxes (e.g., San Francisco, CA; Seattle, WA) to assess effect generalizability. The third aim will analyze the effect of the tax on consumption and obesity outcomes using data from the Youth Risk Behavior Surveillance System. User-friendly software will be produced and made publicly available for all proposed methods. This project will provide robust and flexible tools for policymakers to evaluate, transport, and generalize public policies under complex real-world settings.