Project Summary There is growing consensus that place affects health. A wide range of community-level contextual factors including poverty, walkability, food environment, social cohesion, social networks, and crime, among others, have been conceptualized as drivers of a broad range of individual health outcomes. However, the extent to which these physical and social contextual influences are causal and contribute to geographic disparities in health remains an open question. That is because individuals can self-select into communities based on observed (e.g. socioeconomic background) and unobserved (e.g. tastes, preferences) characteristics that can also influence health, making it difficult to isolate causal effects from correlations using observational data. The extent to which causal pathways versus self-selection contribute to the link between county contextual environment and individuals' health is a critical question in the development of effective public health policies. This study proposes to investigate the causal mechanisms and pathways through which contextual factors influence cardiometabolic (CM) outcomes in adults, including BMI/obesity, diabetes, and hypertension. To disentangle causal pathways from self-selection, we will leverage a unique natural experiment created by the periodic relocation of military service-members to different counties, thereby exposing them to geographic areas with varying burdens of CM risk for reasons and durations outside the individuals' control. This offers a unique opportunity to provide evidence on causality with respect to three main research questions. First, whether exposure to counties with higher CM risk (proxied by a higher obesity rate) increases individuals' risk for these health conditions. Second, which specific county- and neighborhood-level physical and social contextual factors mediate the relationship between individuals' CM outcomes and the county's obesity rate. And third, whether contextual effects on individuals' CM outcomes are moderated by sex and race-ethnicity. To address these aims, we will link individual-level longitudinal data from the Millennium Cohort Study to county- and neighborhood-level contextual data on the physical and social environment and estimate multi- level longitudinal models. We will use structural equation modeling to assess the role of a rich set of county and neighborhood level contextual factors as mediators. Also, the demographically diverse sample will allow us to examine how these relationships vary across sex and race/ethnicity – a critical contribution given the substantial disparities in CM conditions. Concerns about generalizability are limited given the large and diverse sample, their substantial exposure to civilian communities, and similarities between their health behaviors and outcomes and those among civilians. The study is likely to have a high impact given that the combination of a natural experiment design with longitudinal data methods...