The paper's maintained hypothesis, that social interaction is stronger in denser areas, arose from the conjecture that high densities facilitate interaction by putting people in close proximity. The results, however, show the opposite effect, and a key question is why. One possibility is that the crowding associated with a dense environment might instead spur a need for privacy, causing people to draw inward. Such behavior could reflect the old saying: "good fences make good neighbors." Alternatively, higher interaction in low-density suburbs could be a consequence of the spatial layout of residences. Outdoor activities like gardening and mowing the lawn could provide opportunities for relaxed, unplanned encounters with neighbors involved in similar activities. By contrast, neighbor contact for apartment dwellers must rely on fleeting encounters in building hallways or elevators, which may be less fruitful. Another possibility is that dense environments offer residents more sources of entertainment (museums, theaters, etc.), lessening the need to interact with others in the pursuit of stimulation.28 Finally, even though the overall MSA murder rate is included as a covariate, high densities at the individual tract level may be associated with higher-than-average criminal activity, making people suspicious of one another and more reluctant to interact.The paper is well worth a read, although it's technical so be warned!
Sprawl promotes social interaction
Our colleague Joel Schwartz called our attention to a forthcoming article in the Journal of Urban Economics that fnds strong, robust empirical evidence that sprawl promotes social interaction. Jan Brueckner and Ann Largey examine metropolitan level data to determine whether people living in lower density urban areas interact less with neighbors, have fewer friends, participate in fewer activities, etc. They drew on the survey responses from 29,000 people in the Social Capital Benchmark Survey at Harvard University's Kennedy School of Government. What is remarkable is that their results are robust and consistent, regardless of how they specified their empirical model.