Quantifying cities’ emotional effects

Phil Salesses, Katja Schectner, Talia Kaufmann and Cesar A. Hidalgo

MIT Press: “A color-coded map of the perceived safety of New York City neighborhoods, based on Web volunteers’ comparisons of images extracted from Google Maps’ “street view” archive.
Image: Macro Connections Group
July 24, 2013

The “broken-windows theory,” which was propounded by two Harvard University researchers in the early 1980s, holds that urban “disorder” — visible signs of neglect, such as broken windows — actually promotes crime, initiating a vicious feedback loop. The theory was the basis for former New York mayor Rudy Giuliani’s crackdown on petty crime, but it’s come under sharp criticism from some social scientists. One of the difficulties in evaluating the theory is that it’s hard to quantify something as subjective as visible disorder.
In the latest issue of the journal PLoS One, researchers from MIT’s Media Lab present a new online tool that they hope will help social scientists take a more rigorous look at city dwellers’ emotional responses to their environments. The tool presents online volunteers with pairs of images randomly drawn from Google Maps’ compendium of street-level photographs; each volunteer selects the image that better represents some qualitative attribute. Algorithms use the results of the pairwise comparisons to assign geographical areas scores, from one to 10, on each attribute.
In the experiments reported in the PLoS One paper, volunteers ranked the neighborhoods depicted in the images according to how safe they looked, how “upper-class,” and how “unique” — an attribute selected in the hope that it would not be strongly correlated with the other two. The researchers found that the scores for the U.S. cities selected for the study — New York and Boston — showed greater disparity between the extremes for both class and safety than did those for the two Austrian cities selected, Linz and Salzburg.
They also found that, controlled for income, area, and population, the perceived-safety scores for neighborhoods in New York correlated very well with incidence of violent crime”.