ABSTRACT
Although empirical work has shown that personal and spousal education are both related to health, the nature of these associations has been harder to establish. People select into marriages on the basis of observed and hard-to-observe characteristics, complicating the job of the researcher who wishes to make causal inferences. In this article, we implement a within-sibling-pair design that exploits variation within pairs in spousal education to generate estimates of spousal crossover effects. Results-based on a long-term study of siblings and their spouses-suggest that spousal education is positively related to health, but to a greater degree for women than men. Sensitivity analyses show that these patterns are unlikely to derive from measured differences between individuals or unmeasured characteristics that sort them into unions. These results are consistent with network-based theories of social capital, which view education as a resource that can be mobilized by network ties to enhance health.