The Problem with Averages in Understanding Guns, Violence, and Crime: No One Lives in “The United States”

The problem with averages is that there is no “United States of America” when it comes to guns, violence, and crime, but many Americas. Some of these Americas – like my neighborhood in Winston-Salem – are more like our first world counterparts in the OECD, and some of them are more like the third world politically, economically, and socially.

Back in January, a video called “Number One with a Bullet” by someone named Bill Whittle – apparently a conservative blogger — got a lot of attention on the various internet gun sites I survey. People loved it. As of today, it has over 700,000 views on YouTube, and thumbs up are nearly 7,000 to just over 300 thumbs down.

In addition to these cheers, the video also drew jeers, notably in a response by the consistently anti-gun founder of “Armed with Reason,” Evan DeFillipis, on the ever faithful(ly anti-gun) Huffington Post (“Better Than Somalia – How to Feel Good About Gun Violence”).

I watched the video, saw some good and some bad in it, and moved on. Or at least I thought I had. But I couldn’t stop thinking about it, especially about how Whittle and DeFillipis basically talked past each other and so, as happens most of the time in these gun debates, no mutual understanding is gained.

For my part, I think Whittle is completely unhelpful on one point, and extremely insightful on another; DeFillipis is the mirror image, being very helpful on one point, and completely blind on another.

Round 1: The U.S. and Per Capita Homicide Rates Inter-Nationally

Whittle spends the first three minutes or so of the six minute video showing the ranking of the countries of the world according to per capita homicide rates. #1 Honduras, #2 Venezuela, etc. Whittle smugly notes that the United States is not even in the Top 5 or the Top 10, 20, 30, etc. The United States with 4.7 murders per 100,000 population in 2012 ranks #111 in the world, just behind Yemen and Niger and just ahead of Latvia and Micronesia.

Here, Whittle’s data is not incorrect, but his interpretation is questionable. Do I care that the United States’ per capita homicide rate is better than Venezuela’s or Mozambique’s or Turkmenistan’s? Or, as DeFillipis correctly observes, better than Somalia’s? Absolutely not. I care how the United States compares to comparable nations – advanced, (post-)industrial, democratic nations. DeFillipis looks at Organisation for Economic Co-operation and Development (OECD) member nations ranked as high-income by the World Bank (those with a gross national income > $12,616). Among these 31 countries, the United States has the highest per capita homicide rate. That is significant.

Not even knowing what the other 30 high income OECD nations are, I could just watch the list as Whittle scrolled through the countries and know that his argument was ridiculous. Not a single country I think is comparable to the United States politically (democratic) and economically (rich) ranks higher than the U.S. No country I would want the U.S. to emulate ranks higher. That is significant and here Whittle’s rhetoric is simply unconvincing.

My judgement: DeFillipis 1, Whittle 0.

Round 2: The U.S. and Per Capita Homicide Rates Intra-Nationally

In the second half of the video, Whittle looks more in depth at per capita homicide rates in different U.S. cities. Here he is onto something extremely important in terms of the problem with averages. Aggregating data for the entire United States helps us see some things, but blinds us from other things. Most importantly as concerns exposure to homicidal violence, no one lives in “the United States,” per se. We live in 50 different states (and the District of Columbia). But we don’t just live in one of 50 states, we live in one of over 3,000 particular counties or county-equivalents. But we don’t just live in one of 3,000+ counties, we live in one of thousands of cities, towns, municipalities, unincorporated areas, and so on. My risk of being a victim of homicide in my home town of Winston-Salem, is different from my risk in the next city over, Greensboro, or the state’s capital, Raleigh.

Whittle recognizes this, and begins by observing the homicide rate in Detroit (54.6 per 100,000) is almost 12 times the average for the United States. Were Detroit ranks alongside the world’s countries, it would rank #2, just behind Honduras and just ahead of Venezuela. Whittle lists other extremely violent cities like New Orleans, St. Louis, Baltimore, Newark, Oakland, etc. – all of which inflate the average murder rate for the United States.

By contrast, there are cities which have extremely low homicide rates, like Henderson (Nevada) at 1.5 per 100,000, Lincoln (Nebraska) at 1.1, and Plano (Texas) at 0.4. Whittle observes that if the entire country had Plano’s homicide rate, the United States would rank #211 out of 218 countries, including a number of those OECD nations we ought to be comparing ourselves to like France, Italy, Denmark, Spain, and Germany.

Moreover, even city-wide averages can obscure the realities of relative risk. We don’t even live in particular cities, but in particular neighborhoods. I don’t have data on the geographic distribution of homicides in Winston-Salem, but the following graphic shows the geographic concentration of assaults in particular neighborhoods in my home town. The ring shows my neighborhood.

Whittle mentions Chicago, and as I have written previously, Chicago is very instructive. Chicago is the 13^th most murderous city in the United States with 18.5 homicides per 100,000 inhabitants, 4 times the national average. But as Andrew Papachristos has shown in his research there are vastly different rates of homicide and gunshot injury according to where one lives in Chicago.

Papachristos takes these distinctions even further, because even in a high-crime neighborhood in Chicago not every person is equally exposed to homicide risk. In a study of gun homicides in one high-crime neighborhood in Chicago (82,000 people living in a 6-mile area), Papachristos and his co-author Christopher Wildeman found that “41% of all gun homicides occurred in a network component consisting of approximately 4% of the population of the community.”

Papachristos, Anthony Braga, and David Hureau have also studied the risk of gunshot injury in Boston’s Cape Verdean community. Again, the risk of gunshot injury is highly concentrated in certain social networks. 85% of gunshot victims are found in the network, as depicted in the graphic below.

An implication of Papachristos’s research accords with Whittle’s ultimate conclusion: “Maybe it’s not the guns. Maybe it’s the people holding the guns.” Maybe more guns in the hands of the wrong people leads to more crime, and more guns in the hands of the right people leads to less crime? I wrote previously about a promising study I saw presented at the American Society of Criminology which looked at homicide in New Orleans. The authors set out to move the guns and crime debate forward by distinguishing between the effect of legal and illegal guns on homicide. They hypothesized that presence of legal and illegal guns affect homicide rates, but in different ways. Legal guns reduce gun homicide rates (supporting Lott’s more guns, less crime argument), while illegal guns increase gun homicide rates (supporting Cook’s more guns, more crime argument).

The problem with averages is that there is no “United States of America” when it comes to guns, violence, and crime, but many Americas. Some of these Americas – like my neighborhood in Winston-Salem – are more like our first world counterparts in the OECD, and some of them are more like the third world politically, economically, and socially. DeFillipis unintentionally recognizes this in making another point using a U.S. State Department travel warning for Guatemala: “Violent crime is a serious concern due to endemic poverty, an abundance of weapons, a legacy of societal violence, and weak law enforcement and judicial systems.” This could very well be issued as a travel warning to U.S. citizens traveling to certain parts of certain cities here.

DeFillipis is largely silent on distinctions like this, and dismisses any talk of racial disparities in homicidal violence and violence due to gang involvement.

Hence my judgement for round 2: DeFillipis 0, Whittle 1.

Conclusion: A Lose-Lose for Understanding

In the end, it’s a tie: DeFillipis 1, Whittle 1. But rather than seeing it as a “win-win,” I see this as yet another “lose-lose.” No dialogue, no understanding, not even any attempt at understanding. But of course, I already knew that.