How Not To Use Statistics: Gun Violence

I had a brief mini-debate with a couple of co-workers yesterday, one of whom was appalled that people would advance any pro-gun argument in response to the Virginia Tech shootings. Ironically, I - who have never fired or even held a gun - was defending gun rights to two guys who have both fired weapons. They also expressed disbelief over the recent D.C. Circuit decision to overthrow a thirty-year-old gun ban in the District of Columbia.

Here are my basic thoughts: I don't for a minute buy the argument that the 2nd Amendment applies only to militias. I defy any liberal to tell me that David Koresh can declare himself and his followers to be a "militia" and magically have a right to possess weapons that I cannot legally possess. The Amendment mentions militias as part of the justification for the right to bear arms, but the right itself is clearly stated to belong to the people at large; "...the right of the people to keep and bear arms, shall not be abridged." It is equally clear that the militias referred to are to be self-regulated, not regulated by the government at any level. They are intended to safeguard the people from the threat of a police state.

Secondly, I generally find the argument that "if guns are banned, then only criminals will have guns" persuasive (ignoring, of course, the tautological interpretation of that statement). If someone breaks into my house, I think he should be in the fear of God that I'm going to catch him and potentially blow his brains out. I think it's absolutely appropriate for a criminal to have to exercise some risk, and if he knows for sure that I won't have a gun, doesn't that embolden him? (Note that this argument really only applies in the case of random criminal behavior. If someone actually targets you, specifically and intentionally, then there just may not be much you're going to be able to do about it.) So it would seem to me from this line of reasoning that gun bans would probably have little effect on "targeted" crime but could increase the frequency of "random" criminal behavior, as I'm using the terms here.

So, from a purely principled standpoint, I'm a 2nd Amendment supporter. But I'd like to be surer that these principles actually have the desired effects, and people tend to throw around the arguments pro and con without really looking seriously at the statistics. So I did some looking around for statistics on the D.C. crime rates related to the passing of the gun law in 1976, and I found some commentary on National Review Online from a few weeks back, just after the ban was overturned. I read with particular interest the section by John Lott Jr., who is the author of a couple of pro-gun-rights books:

For several decades, D.C.’s gun ban has served an important educational purpose. With the nation’s strictest gun-control laws, gun-control advocates have been embarrassed that the city has frequently had the highest murder rate of any large city in the U.S. This was hardly the case prior to the ban. Yet, the D.C. Circuit Court striking down the ban will prove just as embarrassing because the long predicted surge in violent crime will not occur.

Surely the ban cannot be blamed for all the District’s crime problems. The police department has had severe problems over hiring standards as well as management and morale issues.

But the long-term changes in crime rates before and after the ban are difficult to ignore. In the five years before Washington’s ban in 1976, the murder rate fell from 37 to 27 per 100,000. In the five years after it went into effect, the murder rate rose back up to 35. During this same time, robberies fell from 1,514 to 1,003 per 100,000 and then rose by over 63 percent, up to 1,635. The five-year trends are not some aberration. In fact, while murder rates have varied over time, during the 30 years since the ban, the murder rate has only once fallen below what it was in 1976.

D.C.’s experience strongly suggests that gun bans disarm only law-abiding citizens while leaving criminals free to prey on the populace.

The four links in the post are all the same, and point to a spreadsheet of D.C. crime statistics. I think we all know that statistics can lie, but let's consider how bad this particular post is.

(Let me first point out that over the thirty years, the rates for all crimes have gone up considerably, and this part may support his final sentence. But the five-year trends he cites are a gross misuse of statistics, and that's what I want to focus on here. My reason? NEVER use bad statistics to support a good idea. The ends do not justify the means, and besides, it gives your critics ammunition. Er, no pun intended.)

Let's examine the murder rate per 100,000 citizens from 1966 to 1986:

• 1966: 17.5
  
• 1967: 22.0
  
• 1968: 24.1
  
• 1969: 36.0
  
• 1970: 29.2
  
• 1971: 37.1
• 1972: 32.8
• 1973: 35.9
• 1974: 38.3
• 1975: 32.8
  
• 1976: 26.8
• 1977: 27.8
• 1978: 28.0
• 1979: 27.4
• 1980: 31.5
  
• 1981: 35.1
• 1982: 30.7
• 1983: 29.4
• 1984: 28.1
• 1985: 23.5

The way Lott cites trends is entirely misleading. He makes it sound as if crime was steadily decreasing between '71 and '76. This is clearly false; it's obvious from one glance at the data that 1976 was abnormally low. Now look at the four years after: the rate stayed more or less consistent, and then abnormally jumped to 35 in 1981. Far from being a representative measure, the five-year interval has been cherry-picked to make it sound like murders steadily increased in frequency. Again, a quick look at the data shows that 1981 was abnormally high for the span from 1976 to 1985. If he had arbitrarily chosen that span, he could make it look like the gun ban had a moderately beneficial effect. More to the point, a much better (though still probably misleading to some degree) metric for a "starting point" for the gun ban would have been the average of the years '71-'76, which was 34. By that metric, the five-year period after the ban was indeed lower, with an average from '76-'81 of 29.4.

The real point is that all of these fluctuations are more or less in the noise. You can't really tell much from them at all, especially with the number of uncontrollable and unaccountable factors that go into things like murder rates. Just look at '69-'71: a ~20% drop followed by a ~20% increase. Why? Who knows? But that fluctuation was totally independent of the gun ban.

Now lest I be accused of cherry-picking, I'll just point out that there was a significant crime rise starting in1986, when the level skyrocketed, reaching a peak in 1991 of 80 murders per 100,000 residents. Could the gun ban have contributed to this? Sure. But it's evidently not so cut-and-dried as Lott would have you believe, and I should think that the increase would be primarily due to some developments in the mid-1980s. Maybe Lott is more thorough in his books, but this particular example makes me reticent to accept his arguments at face value.

I'm wide-open to rebuttal by you Second-Amendment types out there, so fire away. (Metaphorically speaking, of course.)