This provocatively-titled book is exactly what it purports to be-- a study of the effect of nondiscretionary ("shall-issue") concealed carry permits on the crime rate. (Nondiscretionary means that once a person has met various well-specified criteria for receiving the permit, it must be issued). John Lott is an economist who used statistical analyses of crime data to arrive at his conclusions. One key conclusion is that even after controlling for many potentially influential variables, such as poverty, population density, arrest rate, conviction rate, other gun laws, and the overall cyclic nature of violent crime in America, there are statistically significant reductions in crime after the passage of nondiscretionary gun laws.
I would call this book thorough. Lott patiently explains everything he did in his analysis. He includes tables and charts showing all the correlations he came up with, even many that aren't statistically significant. Put simply, he tried to include all available crime data from the FBI, from state and local police forces, and from earlier private studies. This is the most comprehensive study that has been done on this subject so far. Of course, all the data are not quite consistent, and furthermore, analyzing on a state level actually gives a different view of the results than analyzing on the county level. Lott explains these differences, and how he accounted for them in this research. In cases where it is possible, he reanalyzes the same data under different conditions (say, excluding certain states or looking at a different level of fineness), and comments on the various results.
All of this thorougness at least eased my fundamental fear of brute-force statistical analysis. The complex calculations typical of economic, social science, and particle physics research-- heavy number-crunching where the results cannot be intuited from the raw data-- suffer from the phenomenon of GIGO (garbage in, garbage out). If you take a totally bizarre set of random data and start massaging it with some of these statistical techniques, it is possible to find all sorts of correlations, and come up with some kind of unifying explanation that really has nothing to do with reality. This is not fear of math or distrust of hard science. It is simply what can happen when a computer is blindly crunching and the human brain intervenes only at the end. But as I said, Lott's care in testing for significance and for double-checking results at least gave me confidence in him as a researcher.
That does not mean the data are perfect. In some cases, Lott had to analyze results from just two or three states because the others didn't fit the set of criteria he was looking at or didn't collect the data he needed to do the analysis. In other cases, the data spanned only a short time period-- say, if a state had passed a nondiscretionary law just a couple of years earlier so the effect of the law wasn't totally clear yet. But these problems aren't Lott's fault; I think he did the best he could have with limited data. I'll rate the book a solid "+" for its honesty and thoroughness.
I've already mentioned the key result of finding (in msot cases) lower violent crime rates after the passage of shall-issue concealed carry permits. By itself, this result has not surprisingly provoked a lot of outrage and brought in many challenges. I was interested to see that Lott also had a bit of supporting evidence that was sometimes overlooked by challengers and was harder to discount. Namely, when a certain county adopts a nondiscretionary law, violent crime in neighboring counties without the law rises. This could result from criminals fleeing areas with more concealed weapons, or it could result from law-abiding citizens with legal guns protecting themselves from crime. Whatever the reason, it has proved difficult to dismiss Lott's main result as error and simultaneously account for this other result.
Do I believe Lott's results? I suppose I do, within the realm of the GIGO effect. To the degree that we have crime data and have tested the effect of nondiscretionary laws in various states and counties across the US, the results seem consistent. This research should certainly not be dismissed (or suppressed).
But the game's not over. We need more data, and more studies like Lott's. A few other research groups have analyzed the same data in order to check Lott, and most came up with similar results, but of course, there were conflicts and questions. And some researchers have claimed to get rather different results. This is a good thing, by the way! It happens all the time in research, particularly in "cutting-edge" cases where the data are still a bit thin and the assumptions made in the analysis are not yet agreed-upon and affect the results strongly. The disagreement over Lott's results imply that we need more research. I hope we will keep collecting detailed crime data so that another study like this can be done in 5-10 years, when the results can be even more accurate.
Of course, not everyone is interested in accuracy. Because gun control is a political issue, all kinds of troubles get tangled in with the research. There are problems with funding, for instance. Obviously, the groups who care most about gun control are the ones willing to fund research, but then there are questions about whether the research will be unbiased. Even worse, however, is that the belief that research will be biased means that gun-advocacy and gun-control groups can never have their research unchallenged. I suppose this is good-- we must be vigilant, after all-- but I am left wondering... what if a study really was unbiased? Would anyone believe that it was? How could the researchers prove it?
Other troubles arise from overapplying or misapplying the results of research to determine policy. It is common to take a single result out of a complex study and build it up into an ideological mountain, then use it to pass legislation. Sometimes this is completely inappropriate, such as if the result were taken out of context and didn't apply across the board. Gun control is an area where this sort of thing can happen easily. Lott is fairly careful to state the limitations on his results. He doesn't make sweeping conclusions about what policies we ought to pursue as a result of finding that crime drops by a few percent after concealed carry laws are implemented. (That's right-- it's only a few percent. Hard to believe people get so bent out of shape about that; more on this below). But still, I hope we can be careful with gun and crime research results. (More on this below too, after I present a couple of things that bothered me).
Nonetheless, there were a few parts that I found unsatisfying. First, Lott makes some questionable comparisons of gun laws and crime in other countries to those in the United States. He points out that some countries with tougher gun restrictions (such as Russia) have higher crime rates than the US, while some with laxer gun laws and high gun ownership (such as Switzerland) have much lower crime rates than the US. I worry that such comparisons are spurious due to two factors: (1) the vast number of other variables that affect crime rates. These variables (poverty, population density, police coverage, etc) of course differ by country-- a lot. I don't even know how to think about the difference between Switzerland and the US in terms of these variables. Lott accounted for them in his US studies, but not in his international ones. And (2) what about the countries with tougher gun laws, lower citizen gun ownership, and lower crime (such as Japan)?
I'm not saying you can't make rigorous comparisons across different countries-- and I'd love to see the results-- but I'm not sure Lott achieved that in his book.
Second, I was unsatisfied with the relative magnitudes of some effects. Lott also compares the change in crime rate that occurred in other countries after the passage of various gun laws. This seems reasonable, at least as a top-level view. For instance, both England and Australia experienced higher crime after the passage of laws restricting citizen ownership of firearms, sometimes by staggering amounts (like 70+%) in certain crime categories. The implication in the foreign country cases is that toughening the laws had a hand in raising the crime rate. This supports Lott's thesis by supporting the contrapositive (fewer guns, more crime)
Likewise, Lott documents the severe increases in gun restrictions that have occurred in the US over the past few decades. For example, public high schools used to have "shooting clubs," and the kids could bring their weapons to school on the day the club met. In other cases, kids could store guns in their cars if they had been hunting before school. Today, kids have been expelled for carrying water pistols, and many schools do not allow even pictures of guns or knives on campus.
This supports Lott's thesis by debunking the gun-control side's counterclaim that fewer guns means less crime, since there is clearly more crime today, including school violence, than a few decades ago.
That's fine, but I am disturbed by the relative magnitude of these effects compared to those found in Lott's research. Double-digit increases in crime rate caused by disarming good citizens seem out of balance with single-digit (and often less than 5%) reductions in crime associated with arming good citizens. Lott does not argue that the large increases seen in England and Australia are entirely due to the effect of tougher gun laws, but the implication is that they played a role and the exact effect is unspecified. This felt unsatisfying to me. Why bring in these vague warnings of 70% increase in crime after a gun law was passed without rigorous results saying how much of the 70% was due to the law? That doesn't stack up relative to the rigorous results of a few percent decrease with the passage of nondiscretionary concealed carry laws.
Similarly, in the US examples, Lott paints a picture of enormous restrictions being enacted over the recent decades-- a veritable tidal wave of disarmament-- and yet, passage of a few nondiscretionary concealed carry laws has a rapid and noticeable effect in reducing crime (at least by a few statistically significant percent).
I'm not sure that all these results add up in my head. I am supposed to believe that tough gun laws in some foreign countries have contributed to immediate, dire increases in crime. But in the US, we have been building up such laws for decades without quite such dire increases. And then, their effect can be noticeably reduced within a few years of letting citizens get concealed carry permits. How robust is the relationship between laws and crime? It is strong and immediate in England and Australia with regard to tough gun laws, as well as in the US with regard to concealed carry laws. And simultaneously, it is weak in the US with regard to gun control laws, since huge restrictions were passed over decades and yet crime trends could be immediately reversed. Hmmm.
This leads to another comment, which is really a criticism of the strong reaction to the book. Across the board, Lott finds very small (but statistically significant) decreases in crime associated with the arming of citizens. Mostly, the results are 1.5-3%, with the occasional result over 5%. Of course, any reduction in crime is good-- especially horrible crimes like murder and rape-- but really, these results are not earth-shattering. Looking at the data, my reaction was, "OK..... this is good, but can we think of more effective ways to reduce crime?"
I suppose that the real issue is the follow-up implication of the data: if the laws we have now managed to reduce crime by a few percent in the places they were enacted, imagine if we made even less restrictive laws, and passed them in even more places! We could reduce crime by a lot.
Maybe, maybe not. That's a very speculative conclusion. It assumes a linear extrapolation from the situation we are in now. More permits, less crime -----> even more permits, even less crime. But we don't know, for example, whether Lott's results occurred because there are relatively few places issuing nondiscretionary permits. When the map fills up with places criminals don't want to live, what happens? When every other citizen might be armed, what happens? Criminals are smart; presumably the arms race will go on with them developing new capabilities to counter the defensive tactics of law-abiding citizens. Of course, that is no reason not to pass more concealed-carry laws. Arms races happen, and the best we can do is try to keep them controlled. Lott's results probably suggest that the good guys can win for a little longer, and to a little greater degree, if we allow more guns. But I think we must be very cautious about extrapolating too far in the future about the long-term results. The future is often more interesting than a linear extrapolation.
So overall, I found this book to be intriguing. Given the hodgepodge nature of most crime data, I had despaired of finding any real correlations between guns and crime from a casual analysis I performed on my own. I concluded (in another essay on this site) that there may actually be no correlation; the effects of economics, local history, and a million other variables might simply swamp the effects of guns, be they positive or negative. But Lott has done the hard work of trying to digest all the crime data. Since he gets some consistent results, perhaps he succeeded. As I said above, I think it's worthy of more research.
Copyright © Kim Allen 2001
