I am reading Robert Sampson’s Great American City: Chicago and the Enduring Neighborhood Effect, which William Julius Wilson calls “one of the most comprehensive and sophisticated studies ever conducted by a social scientist.” Sampson makes a strong case for considering neighborhoods (which are examples of settings or contexts) separately from individual people, and then looking empirically at how the two are related. That is a different approach from assuming that people are the only real or important social entities, whereas things like neighborhoods are mere amalgams of variables. His method also challenges the widespread application of medical treatment models to social issues.

Consider, for example, broccoli. Eating broccoli is a behavior of individuals. Its probability is influenced by an individual’s tastes, desires and beliefs; the availability, market demand, and cost of the vegetable; the alternatives; and prevailing social meanings. We observe a positive correlation between eating broccoli and health. The question is whether the broccoli causes the good health outcomes, when lots of other factors might be implicated. For instance, maybe people who eat broccoli are also people who smoke less. So we control for as many other factors as we can. Ideally, we experiment by giving a randomly selected treatment group broccoli and comparing them to a control group.

Neighborhoods are also observed to correlate very strongly with individuals’ outcomes. People are much healthier, wealthier, and safer in some neighborhoods than in others–in Stockholm as well as Chicago. So it is tempting to treat a neighborhood like broccoli, and examine its impact on residents while holding other factors constant. There are even randomized experiments in which people are given vouchers to move into new neighborhoods. But here are several reasons why neighborhoods are not like broccoli:

• Neighborhoods are defined relatively. Everyone could eat broccoli, but everyone cannot live in a given city’s most fashionable neighborhood. If people were randomly assigned to move into the most fashionable (or hippest, or most dangerous) neighborhood, it would lose its character.
• Neighborhoods are connected. If you switch from French fries to broccoli, no connection is created between the two vegetables. But if lots of people move from Chinatown to Flushing (Queens), the two communities are changed by means of the important new links between them.
• Neighborhoods are spatially related. If Neighborhood B lies between A and C, then residents of B are probably affected by conditions in A and C–unless, for example, a gigantic Interstate separates B from C. Also, A and B may both belong to a larger district that excludes C. One cannot tell the relationships among neighborhoods without mapping them spatially (whereas one can see which vegetables people eat by examining individuals’ choices).
• Neighborhoods are influenced by the reasons people live in them. Broccoli is broccoli whether your mother makes you eat it or you love it stir-fried with chillies. But some neighborhoods are open destinations of choice–everyone who can afford the rent is welcome, and that is part of their charm. Others are racially segregated or deeply traditional, so that an outsider would feel uncomfortable or unsafe moving in. The “effects” of a neighborhood on its residents are often related to the way that people come to live in it.
• Neighborhoods matter intrinsically. Regardless of its effect on residents, we might like to preserve Greenwich Village (or The Mission, or Koreatown). The neighborhood’s survival does not trump human concerns, but it adds an extra concern. In contrast, nutritionists don’t care about broccoli except as means to promote health.
• Perceptions of neighborhoods linger even when their populations or features change, creating lasting effects. Sampson finds, for example, that perceptions of neighborhoods’ disorder in Chicago are weakly related to actual disorder yet predict poverty rates years later, as people choose to relocate and invest in neighborhoods they perceive as orderly. (Unfortunately, both Black and White observers use the proportion of the population that is Black, rather than any objective signs of disorder, to decide which neighborhoods are disorderly.)

One of Sampson’s great contributions is to help develop ways of directly assessing contexts, not merely individuals’ perceptions or experiences of their environments. He and his colleagues call that science ecometrics, although I find that the same word has also been claimed for a branch of ecological economics.

(Those thoughts are loosely inspired by Sampson, not meant as a paraphrase.  See also “more to life than individual attributes.”)