We analyze data from the Minnesota Twin Registry (MTR), combined with the Socioeconomic Survey of Twins (SST), and new mortality data, and contribute to two bodies of literature. First, we demonstrate a beneficial causal effect of education on health and longevity in contrast to other twin-based studies of the US population, which show little or no effect of education on health. Second, we present evidence that parents compensate for differences in their children's health endowments through education, but find no evidence that parents reinforce differences in skill endowments. We argue that there is a bias towards detecting reinforcement both in this paper and in the literature. Despite this bias, we still find statistical evidence of compensating behavior. We account for observed and unobserved confounding factors, sample selection bias, and measurement error in education.