This is Aaron Rester's blog:

Field Notes from the Digital Prairie

Wednesday, February 20, 2013

Design Lab: Concealed Carry vs. Violent Crime

A few years back, after a one-day class with information-design rock star Edward Tufte, I decided to try my hand at creating a complex infographic by charting violent crime rates in different states against a number of variables, including strictness of gun control laws, poverty rate, and diversity. The biggest problem with this attempt was that it contained no temporal index. My father, an ardent gun-rights supporter, suggested that I look instead at how crime rates change after laws prohibiting the carrying of concealed weapons are repealed; the argument from gun-rights advocates, of course, is that if criminals are unsure who might be carrying a weapon then everyone is less likely to be the victim of crime. I decided to take him up on the proposition, and have spent the last year or so working on the graphics displayed below. Little did I know that by the time I was finished with them, the Seventh Circuit would have struck down Illinois' status as the last state with a total ban on concealed carry or that the nation would be embroiled in a passionate debate over the place of guns within our society after Newtown.

I should state at the outset that my interest in creating these images is less in influencing social policy -- I am not naive enough to think that I have much chance in doing that -- than in testing my ability to communicate information about large amounts of data. I also recognize that, not being a statistician, there are certainly more sophisticated ways of analyzing the same data that I have examined here; should this post happen to inspire someone with more statistical chops to check my findings, so much the better. All of that said, I think the data I've accumulated here (also available in this spreadsheet) speaks pretty powerfully to the question of whether permitting citzens to carry concealed weapons has a deterrent effect on violent crime.

A few notes on sources and terminology: stats on violent crime are easy to find via the FBI, but it is surprisingly difficult to find official information about when changes in gun laws have been made. The closest I could find to a comprehensive survey of changes in concealed carry laws was the cheekily named pro-gun site Radical Gun Nuttery!, which provides a list of dates and a (partial) list of citations for those dates. States are ranked as having concealed carry laws that fall into one of four classifications:
  • no-issue (allows no private citizen to carry a concealed weapon)
  • may-issue (allows concealed carry with a permit that may be granted at the discretion of some local government office)
  • shall-issue (acquiring a permit only requires meeting a predetermined set of criteria)
  • unrestricted (no permit required)
For the first part of this project, I looked at 27 states whose laws changed from either no- or may-issue to shall-issue or unrestricted between 1982 and 2009. Since crime rates can fluctuate from year to year for any number of reasons, I looked at the five years before and the five years after the laws were changed. The graph below shows those changes, along with the national average and, for perspective, four states (Illinois, New York, California, and Vermont) that underwent no changes over that time period.

Aside from a few anomalies (such as Alaska's increase after going unrestricted or Vermont's consistent rate over many years), the graph would appear to largely bear out my initial hypothesis that gun laws have little effect one way or the other on violent crime rates -- nearly all of the pre- and post-change violent crime rates track closely with the national average, as do the rates of states that underwent no change.

However, I decided to look more closely at the data and compared the five-year average before each state's change and after with the changes in the national average over the same times. This turns out to be a clearer picture of possible effects of the change in gun laws, and the picture is striking. Only four states appear to have had positive outcomes in relation to the national crime rate, while 23 had negative outcomes.

  • National average increased, state average decreased: 1 (OR)
  • National average decreased, state average decreased more: 2 (KY, TX)
  • National average increased, state average increased less: 1 (PA)
  • National average decreased, state average increased: 6 (AK, OH, CO, MN, TN, UT)
  • National average increased, state average increased more: 6 (MS, ID, FL, GA, MT, WV)
  • National average decreased, state average decreased less: 11 (AR, AZ, OK, MO, NC, SC,  NM, LA, MI, NV, VA)
Overall, states that switched from no-issue to shall-issue laws had an average change 10.13 percentage points worse than the national average, while those going from may-issue to shall-issue fared an average of 7.68 percentage points worse. So, on average, if the national violent crime rate was going up over a given time period, the average state that transitioned to more permissive concealed carry laws saw its violent crime rate grow faster than the national rate; if the national rate was going down, transitioning states saw their crime rate fall more slowly than the national average. This suggests that not only does greater ability to carry concealed weapons not deter violent crime, it may in fact exacerbate it.

Finally, just to make sure limiting my data to five-year averages wasn't hampering my view in some way, I looked at how no-issue states' violent crime rates in comparison to the national average over more than two decades. If, as the argument for concealed carry goes, a citizenry without the ability to legally carry concealed weapons is easy prey for criminals, then one would expect no-issue states to have, on average, violent crime rates higher than the national average. In fact, however, as the graph below indicates, the average violent crime rate for no-issue states was consistently below the national average (the weight of the red line indicates the number of no-issue states, while the red dots indicate the distribution of states). Until the sample size shrank to two, the majority of no-issue states had violent crime rates below the national average. Once again, we're left with evidence that, at the very least, the inability for citizens to legally carry concealed weapons does not necessarily encourage violent crime.

So... what do you think of these graphics? How could they, or the data communicated in them, be improved?