“One study of a high-crime community in Boston found that 85% of gunshot victims came from a network of just 763 young men—or less than 2% of the local population.” So says The Economist in a review of the causes of a spike in the murder rate and potential solutions to the problem entitled “Midsummer Murder”. It’s an exciting finding – if we can identify this 2% of men and hit them with a targeted intervention, we can essentially solve America’s horrific homicide problem.
But a friend of mine pointed out an odd discrepancy within the article: it states that only 14% of murder victims are gang members. How are there all these socially-connected young men committing almost all the violence, if they are not in a gang? And it is oddly unspecific about the details of the Boston study – who are these men? Are they black, like most of the other young men discussed in the piece? To sort these things out, I googled the entire sentence to see if I could find the original study.
What I found instead was the original sentence. Papachristos and Wildeman (2014) write:
For example, a recent study of a high-crime community in Boston found that 85% of all gunshot injuries occurred entirely within a network of 763 young minority men (< 2% of the community population)
The Economist just took this sentence and did synonym replacement. Their version is quite clearly a weakly paraphrased version of the original.
Interested in the details of what was going on, I dug up the actual study being cited here, which is also by Papachristos (with Braga and Hureau).* And it turns out that the The Economist’s quoted-without-attribution summary doesn’t even appear to be accurate. The study only looked at 763 people – not the whole community. And the 85% figure is the share of gunshot victims within the 763-person sample who are socially-connected to each other somehow. I see no mention in the article of the 2% figure that is cited by Papachristos and Wildeman (2014) and The Economist. To be sure I didn’t overlook it, I used ctrl-F to look through the entire article for “two”, “2”, “%”, and “percent”. It appears to be a mischaracterization of the statement in the Papachristos et al. (2012) abstract that “The findings demonstrate that 85 % all of the gunshot injuries in the sample occur within a single social network”.
It might also be a true statement that is just not supported by the underlying article. Papachristos presumably has the data, and sometimes people cite their own previous papers in order to reference a dataset. Usually when you do that, you say something like “authors calculations using data from
” but maybe it got edited out.
It turns out that Papachristos and Wildeman (2014) – the other article where the unattributed quote was pulled from – actually addresses the question we are interested in: how socially-concentrated are murders. They study a poor, black neighborhood in Chicago, and find that 41% of all murders happened in a network of just 4% of the sample. The rest of the murders are actually even more spread out than you might guess – they don’t happen within the “co-offending network” – the set of people who are tied together by having been arrested together. Instead, 59% of murders are are spread throughout the 70% of the community that is not linked together by crime. While there are patches of highly-concentrated homicides the overall problem is not very concentrated at all.
That is a much less sexy finding, that lends itself much more poorly to simple solutions to problems of violence. If homicide is highly-concentrated then we can rely entirely on targeted interventions like Richmond’s program to pay high-risk men to stay out of crime – a program directly motivated by evidence that a small number of men are responsible for almost all murders.
What The Economist did here is not exactly plagiarism, but it wouldn’t fly in an undergraduate writing course either. At the same time, I can definitely empathize with the (unnamed per editorial policy) author. They might have fallen into the trap of reading the original sentence and then having trouble coming up with a truly distinct way of saying the same thing – that’s definitely happened to me before. Maybe they were worried about screwing up the meaning of the sentence by changing it too much. Or perhaps the editor cut out a citation that was in there.
Whatever the reason for the lack of attribution, it has consequences: because the quote was just lifted from another piece, the author didn’t catch the fact that it was an erroneous description of the study’s findings. If the original source of the quote had been cited, other people might have noticed that it contradicts what the quote says, or thought it was odd that they cited the source paper’s description of another study but not its own findings. Lots of people read The Economist. Policymakers thinking about how to combat homicide could be misled by this false but eye-catching statistic into investing in approaches that probably won’t work, instead of things like cognitive behavioral therapy that actually make a difference.