Why do people feel safer in some cities than others? Is it a function of actual crime or of other factors that shape the way we perceive cities?
Crime, especially violent crime, varies considerably across U.S. cities and metro areas (as well as within them), with some posting record lows last year and other surging to record or near-record highs. But, a wide range of studies suggest that perceptions of safety are conditioned by other obvious and less obvious factors that effectively signal community safety. The infamous "broken windows" theory outlined by James Q. Wilson and George L. Kelling (in a 1982 issue of The Atlantic) suggests a close connection between neighborhood conditions and actual crime as well as perceptions of it. "Social psychologists and police officers tend to agree that if a window in a building is broken and is left unrepaired, all the rest of the windows will soon be broken," they wrote. "This is as true in nice neighborhoods as in rundown ones."
A recent survey by the Gallup Organization provides a comparative gauge of resident perceptions of safety across large U.S. metros. Based on data from the ongoing Gallup-Healthways Well-Being Index survey, it assesses the percentage of residents who say they "feel safe" walking alone at night in America's 50 largest metro areas. The table below (based on the Gallup-Healthways poll) lists the metros where residents report feeling the most and the least safe walking alone at night.
|Safest Metro Areas Among 50 Largest MSAs|
|Metro Area||Yes, safe %|
|Minneapolis-St. Paul-Bloomington, MN-WI||80|
|Salt Lake City, UT||77|
|Austin-Round Rock, TX||77|
|Least Safe Metro Areas Among 50 Largest MSAs|
|Metro area||Yes, safe %|
|New Orleans-Metairie-Kenner, LA||59|
|Riverside-San Bernardino-Ontario, CA||61|
|Houston-Sugar Land-Baytown, TX||63|
Table adapted from Gallup
Minneapolis-St. Paul has the highest percentage of residents who say they feel very safe. Denver, Raleigh, Boston, Salt Lake City, and Austin also rank high in perceived safety. We should note of course that this data was compiled before Monday's horrific attacks at the Boston Marathon. Conversely, Memphis, New Orleans, Riverside, Houston, and Jacksonville have the lowest percentages of residents reporting that they feel safe. Even in these metros, however, more than half of residents say they feel safe.
• • • • •
What is the connection between perceived safety and actual crime? And what are the factors in addition to crime that might act on the perception of safety?
With the help of my colleague Charlotta Mellander, I decided to take a look. Mellander ran statistical correlations between the Gallup-Healthways measure of safety and various measures of crime, as well as social, demographic, and economic conditions of metros. As usual, I point out that correlation does not imply causation, but points only to associations between variables. It's also worth remembering that the Gallup-Healthways data cover only America's 50 largest metros. The results might differ substantially if smaller metros were included, and perhaps more using data at the more fine-grained neighborhood level.
Interestingly, while Gallup finds a substantial connection between crime and safety, the results of Mellander's analysis are mixed on this score. According to Gallup: "Though crime statistics are not available for all MSAs, there is a strong negative correlation (-.64) between the FBI's 2010 violent crime rate for an MSA and the percentage of the MSA residents who report feeling safe -- with the cities where residents feeling safest having lower crime rates, and vice versa." Mellander's analysis picks up a similar negative correlation (-.63) between perceptions of safety and gun murders (homicide only, not including suicide) per capita. But she finds no statistically significant association between perceived safety and a range of crime per capita measures based on the FBI Uniform Crime Reports, including for violent crime and property crime.
Resident perceptions of safety, however, are strongly connected to the wealth of metros according to her analysis. The share of metro area residents who feel safe is positively correlated with income levels (.38).
Perceptions of safety also turn on education levels. The correlation between perceived safety and educational levels (measured as the share of adults that are college graduates) is among the highest in her analysis (.55). The scatter-graph below plots the connection between perceived safety and the share of adults that are college grads across metro areas.
The structure of the local economy and labor market also plays a role. There is a substantial correlation between perceived safety and the share of workers in knowledge economy jobs (.46). Conversely, perceived safety is lower in more blue-collar working class metros (with a negative correlation of -.33).
Artists and cultural creatives have long been perceived as markers of urban revitalization and gentrification: There is also a positive correlation between the share of workers in artistic and creative jobs (.40) and perceived safety, though that may reflect the overall affluence of a place as well.
On the flip side and not surprisingly, feeling unsafe is closely associated with race and poverty. Resident perceptions of safety across metros are negatively correlated with the share of the population that is non-white (-.55), the share of residents below the poverty line (-.53), and the unemployment rate (-.47). Perceived safety is also negatively associated with homelessness (-.43), in line with both poverty-based explanations and the broken windows theory.
It's commonly suggested that people are more anxious and fearful in bigger cities. But Mellander's analysis picks up no statistical association between either population density or population size. Of course, this finding may be affected by the fact that the data and analysis is limited to just the 50 largest metros. That said, there is substantial variation in population — from roughly one million to about 20 million across these metros — and considerable variation in density across these metros as well. Still, it's likely that density would play a bigger role in perceptions of safety across say urban versus suburban neighborhoods.
Growing inequality across America is not just an economic problem; according to a growing chorus of social scientists, it is also is a key contributor to America's fraying social cohesion. Based on this, one might expect perceptions of safety to turn on levels of inequality across metros. But, Mellander's analysis finds no statistical connection between inequality (measured by the Gini coefficient) and perceived safety.
It's also commonly assumed that home-ownership gives people a greater stake in communities, promoting greater social cohesion and bolstering the feeling of safety. But Mellander's analysis finds no statistical association between perceived safety and the share of the population that rents or monthly housing costs at the metro level. While this finding might differ if smaller metros were included, or if the analysis were expanded to the neighborhood level, it's again worth noting that there is substantial variation in renting and housing costs across the 50 largest metros.
Religion is seen to be a key factor in the social capital and cohesion of places. Social conservatives would add that religion is a marker of the moral fabric of a community. One would thus expect people to feel safer in more religious communities. But that is not the case at the metro level, according to Mellander's analysis. In fact, the share of the population that identifies as very religious is negatively correlated with perceived safety at the metro level (-.33).
Climate plays a role as well, according to Mellander's analysis. Perceptions of safety are also negatively associated with mean high temperatures in January (-.57) and July (-.38). In other words, warmer metros are generally perceived as being less safe.
• • • • •
Simply put, metro areas where people feel less safe have higher rates of poverty, higher rates of unemployment, higher levels of homelessness, higher shares of population that are non-white, lower levels of education (measured as the share of adults who are college grads), more working-class economic structures, and higher levels of religiosity. Metros where people feel safer have higher levels of income and education, less unemployment, less homelessness, and more knowledge-based and creative economic structures.
When all is said and done, the perception of safety seems to reflect the affluence, education levels, and economic circumstance of places at least as much, or perhaps more so, than the underlying reality of crime.
Top image: A Massachusetts State Police officer stands guard at the scene after explosions reportedly interrupted the running of the 117th Boston Marathon in Boston, Massachusetts April 15, 2013. (REUTERS)