The whole tension with studying the effects of insurance is, you have to wonder why some people have insurance and other people don't, and whether those reasons could be related to the outcomes you are studying, Finkelstein explains, like the possibility that people who are sicker seek out insurance more.