Question: In an ankle on how exercise improves health, the New York nines reported on an observational study that found that each hour spent running added two hours to a person's life expectancy [Brady. 2006]. A week later, a letter to the editor questioned whether the results really proved anything about the impact of exercise on health, and suggested that the study could just as well be showing that "those with a strong heart and good health are otherwise more likely to enjoy running and do it more regularly." How does this challenge to the exercise study relate to the problems faced by economists trying to assess the causal effects of economic policy? How could you design an experimental study to estimate the impact of running on life expectancy?