Anatomy Of An Error, And Another Error

It begins with Matthew Yglesias noting on his blog a strange looking statistic in a Foreign Affairs piece:

This Foreign Policy article on Steig Larsson and the decline of the Swedish utopia concludes with what seems to me to be a whopper of a factual error:

If you look at the statistics, Sweden is not a particularly violent country, nor a particularly lenient one to criminals. It is in about the middle of the European averages for both figures. There were 230 homicides in Sweden in 2009, compared with 143 in Washington, D.C., which has a population a bit more than half Sweden’s size. But compare these figures to what they were in the years when Sweden looked like a utopia. In 1990, there were 120 homicides in Sweden, and 472 in Washington. There is a convergence here that doesn’t flatter Sweden.

In fact, there are 9.3 million people in Sweden and only 600,000 in Sweden. In other words, Sweden has about fifteen times the population of Washington DC and less than half the murders.

Wow, the author of the Foreign Affairs piece, Andrew Brown, got the population of Washington wrong by a factor of fifteen?!? What a maroon, right?

Well, wrong. The fact the numbers Brown presents (something over 4.65 million for DC) are so different from the actual population of the District of Columbia should have been a clue that we has citing another statistic, and, of course, that is exactly what Brown was doing. Brown used some version of the Metropolitan Statistical Area, which in 2009 was estimated at a little over 5.4 million for Washington DC. As a result Yglesias, more interested in playing "gotcha!" than doing a little thinking and research, really doesn't help us understand what is going on here.

Picking up on Yglesias, Balloon Juice manages to be even more unhelpful:

If I had written or edited this article, I would right now be deciding between suicide and a career change. I know that sounds harsh, but it’s true.

You mean such as writing something about demography without, seemingly, knowledge of a basic demographic measure (i.e. the Metropolitan Statistical Area)? Good luck with your suicide decision Mr. Balloon Juice.

Now, to be fair to Yglesias he does acknowledge his error, at least to a point. At some point after he wrote his initial posting he added:

It seems clear that the author of the piece meant to refer to the population of the overall Washington DC metropolitan area but if that's what he wants to talk about then he needs to add up the total number of murders in that area.

Gee, if it is so "clear" why didn't you mention that in your initial writing on the matter?

Now, to be sure, Brown did make an error in his piece in using the murder statistics for the city of Washington DC instead of for the MSA. From what I can tell, the F.B.I. has not released the relevant statistics for 2009 yet, so I cannot find out what the actual murder numbers should be for the Washington MSA for 2009. In 2008, the rate was 7.5 per 100,000. Using the 2009 population estimate that would give us roughly 405 murders in the DC MSA for 2009.

But let us remember, Brown's original point was that the murders numbers were converging over time. Are they?

Using the HUD USER demographic data I can see that in 1992 the murder rate in the Washington MSA was 15.7. Alright, using the MSA 1990 census data the population of the DC MSA was 4.1 million. This would make the numbers of murders for the MSA in 1990 be roughly, 644, while in 2009 there were 405. The homicide rate decreased 37%. In 1990 Sweden had a population of right around 8 million and 120 homicides were committed. In 2009 the numbers were 9.3 million and 230. This means the homicide rate increased 167%. So, the convergence is real, if not of the magnitude that Brown implies.

So, yes, Brown screwed up, but so did Yglesias.