US Homicide Rates

The pressure's on the screen
To sell you things that you don't need
It's too much information for me
--Duran Duran

In the wake of last week's school shooting tragedy in Newtown CT, myriad statistics are being tossed about by both anti-gun and pro-gun proponents. The numbers get confusing in a hurry, particularly when it is possible to cherry pick data that fit an agenda.

Stated differently, there is a partisan tendency to build one's case around information that, while not necessarily wrong, is incomplete. Indeed, the tendency of politicians and media to select 'authoritative sources' that fit a particular agenda was the basis for Groseclose's (2011) work.

As such, I am skeptical of many figures that I read and hear. When possible, I prefer 'source' data'--basic data upon which more sophisticated (and often biased) analyses can be built. Of course, the source data themselves can be biased. For example, these pages have spilled much ink on the bias and manipulation embedded in US government econometric series (e.g., inflation, unemployment, GDP).

Nonetheless, source data expressed as simple time series provide a useful starting point, as they permit scrutiny of fundamental biases of data collection. If the biases are deemed small, then time series can provide useful information on long term levels, trends, and variation. They also may allow for some (emphasis!) degree of causal analysis by inserting events and looking at changes in 'before' and 'after' periods.

Let's start with a basic one here: US Homicide Rates. The US Department of Justice is the gatekeeper, and the data can be found here. These data are likely more trustworthy than federal econometric data because they are based on simple counts rather than surveys and/or formulaic manipulation.

Figure 1 shows total homicides in the US from 1950 thru 2010, expressed as homicides per 100,000 people.

The homicide rate has been trending down for about 20 years and it at its lowest level in 50 years.

Figure 2 breaks down homicides by weapon used. Note that the time horizon here is shorter than in Figure 1. Data were only available from 1980 thru 2008.

More homicides are committed with handguns and other guns (rifles, shotguns) than any other weapon. Knives are next. Over this 30 year horizon, handguns and knives have generally declined as homicide weapons of choice. 'Unknown' weapons have seen the biggest increase.

Figure 3 combines the handguns + other guns categories from Figure 2 to present a cleaner view of gun-related homicides.

Just under half of all homicides employ guns. Over the 30 year period observed, the overall trend is slightly lower.

Combining declining overall homicide rates since 1980 with data on weapons usage indicates fewer gun-related homicides per capita than a few decades back. When one considers that gun ownership in America has markedly increased during this period as have the number of citizens carrying firearms (still looking for source data on both), then it appears that gun control proponents have a difficult case to make w.r.t. gun-related homicides in the US.

Simply put, more guns have not corresponded to more homicides. In fact, the opposite is more likely to be true.

Of course, guns can be employed for other crimes besides murder. We'll explore violent crime in a future missive.