I have been saying for a long time that looking at gun-violence numbers at the state or even the county and city level doesn’t really tell us much about gun violence and certainly doesn’t give us much of a roadmap for figuring out what to do about gun violence. This is because while anyone in America is ‘free’ to live wherever they want to live, the truth is that our residential neighborhoods tend to be very segregated by race and by income even down to the square-block level, or what the U.S. Census refers to as a ‘tract.’

             Now a tract, according to the Census, is actually an area which holds somewhere between 2,500 and 8,000 people which is roughly 900 to 3,000 households per tract.  Census tracts often cross town or city borders and the space they cover tends to reflect the degree to which residents in that particular tract share basic social and civic amenities such as schools, shopping and parks or open space.

Given the granularity of mapping tracts, using this information to identify levels of gun violence opens up all kinds of new perspectives that otherwise would remain lost from analysis and view. For example, this year the national gun violence rate (computed by taking the number of gun homicides and dividing per 100,000 persons) will end up somewhere between 3 and 4.  The city of Springfield, MA, will have a gun-violence rate of 11 or 12, but a neighborhood like Indian Orchard (which covers two census tracts) may not have a single gun homicide at all.

And this is an important distinction, knowing the difference between city gun-violence rates and tract gun-violence rates, because people tend to live and stay around their own neighborhoods, and thus the quality of life in their neighborhood is a much different issue than the quality of life in the city as a whole.  It also happens to be the case, according to the BJS, that most homicides take place in, in front of or down the street from the victim’s home. So, for example, if you live in Census tract 67000 of Springfield, there were 5 murders committed in that neighborhood in 2015, which gives the neighborhood a homicide rate of 61(!), but if you walk a half-mile into the adjacent census tract of Longmeadow, that place hasn’t had a homicide in years. Obviously, the strategies we might develop to reduce gun violence in Springfield would have no relevance to Longmeadow even though the two tracts lie side by side.

Using census tracts to better understand gun violence has produced a remarkable collaboration between our friends at the Gun Violence Archive (GVA) and the folks who write about gun violence issues for The Guardian, whose work in this area is really superb.  What the Guardian has done is to take the 13,039 instances (actually victims) of gun violence found by the GVA for 2015 and plotted each one against the census tract in which it occurred.  And guess what? It turns out that more than one-fourth of all gun homicides took place in census tracts containing 1.5% of the total population living in the United States.

That’s an extraordinary concentration of violence in what totals a fairly small amount of space. But even more interesting is the fact that the homicide rate in all the tracts where homicides occurred was 32.9.  In other words, where gun violence actually takes place the result is a gun homicide rate which is ten times higher than the national gun violence rate as a whole.

What this tells us is that yes, we have a national gun violence problem if only because thousands of Americans are gunned down every year. But while the problem is national in scope and size, it is local in terms of where it actually occurs. Thank you GVA for giving us this data, thank you Guardian for giving this data a new and important look.