A Problem with National Homicide Statistics

Too much aggregation: It's not just a problem for economic data.

Donald Trump has been using and abusing national homicide statistics to suit his own political ends. But, using nationwide statistics — even correct ones — has long been problematic.

We’ve learned, for example, that when comparing the homicide rate in the US, comparisons with other countries of vastly different size and demographics are not particularly useful. Because of the vast geography of the United States, it is important to break down the homicide rate into smaller pieces, and this gives us insights into how statewide and nationwide gun regulations fail to account for large disparities in rates of violence.

One such method in breaking down the homicide numbers is to look at individual States. However, even this is problematic. If one is to look at a more detailed level, one finds that even neighboring jurisdictions can have wildly different homicide rates. One great example is the comparison of adjacent counties on opposite sides of the Mexico-US border. Further, if one were to visit the Trulia Crime Map, one can find that violent crime can vary depending on which block of a city you currently find yourself. This becomes evident in this map of Chicago —a city famous in the media for being a chronically violent place. Of course, if one goes to, say, the Avondale neighborhood or along Lake Michigan, the chances one will find himself a victim of a homicide is slim.

Breaking It Down to Metro Area and Beyond

So, just how many people live in these high homicide rate areas in the United States? The FBI keeps a record of the number of homicides and homicide rates of major Metropolitan Statistical Areas (MSA) around the country. By using the 2014 MSA Homicide Statistics, we can find out just how many people live in relatively dangerous areas. For some context, we can then draw some comparisons with other parts of the world. 

To do this, I’ve broken down the homicide rate in metro areas into eight separate classifications starting at 0 to 1 homicides per 100,000 people and increasing by one until reaching 6 to 7 with the remainder over 7. What this generated is the following table, including a comparable foreign nation that fits that homicide-rate classification.

What this shows is that a solid 46% of the United States population lives in an area that is similar to Liechtenstein, a European nation that does not have much of a reputation for being a particularly murderous place. If we add Latvia, the nation classified as Northern European with the highest homicide rate, the level of homicide safety in the United States jumps to 57% comparable to EU Member States.

An additional 22% are in what gun control proponents would consider marginal homicide rates, while 21% are living in areas much worse.

However, as noted above with Chicago, this level of detail is still insufficient to identify how homicide rates and population exposure really is. In fact, if we dig deeper still, we find that of all known relationships reported to the FBI between the culprit and victim, approximately 79% are people who live in the same residence or near the victim, such as family members, significant others, or neighbors. This indicates that it is particularly rare that someone will cross town to kill another person and murders of complete strangers are also relatively unusual.

Unfortunately, we don’t have a neighborhood-by-neighborhood homicide database nationwide to test this. However, we can get an idea by looking at the above breakout by NYPD Police Precincts. The NYPD is broken down into 76 precinct locations. These locations are fairly small geographically compared to the above MSA. Because of the much closer proximity, we can get a stronger measure of homicide distribution using the above methodology. The following are the results:

What this shows is that the City of New York has over half of its population within low-level homicide rate levels in 2014 and 65% is comparable to EU Member States. We would likely find similar results in most MSA areas around the country since crime tends to concentrate in small areas, such as a small band between Compton and East Hollywood in Los Angeles County.

Because of this severe fragmentation of homicides across the United States, it is difficult to argue that firearms are the culprit since, if they were, we would expect a more uniform homicide rate at least across MSAs and not highly concentrated into specific neighborhoods while the majority of the population is insulated from such violence. Because of the nature of the State, which tends to apply blunt one-size-fits-all solutions to problems, the answer is unlikely to lie with any kind of legislative action since the problem is far more nuanced than broad, nationwide statistics can identify. In fact, the State may also be contributing to the problem.

Whatever the underlying cause to these problems, the evidence makes it far more difficult to conclude the presence of a State-led firearms control regime is the reason behind low homicide rates in nations with firearm control laws when, at minimum, 57% of the United States already enjoys those levels without having to pass restrictions or bans.