« Support our troops | Main | Department of awful statistics »

Suckers for a good story

14 Jan 2008 10:33 am

In the comments to the previous post, Brooksfoe, who I respect a great deal, points out that the narrative "Soldiers exposed to combat come home to kill" does not flunk the gut check test.

No, indeed, it doesn't. I can build a quite plausible story where combat makes people into crazy killers.

The problem is, it's just a story. History is full of those stories that turned out not to be true. As my commenter points out, I can also build a plausible story where combat makes you realize the sanctity of human life and makes you less likely to kill. Or where the amazing human capacity for compartmentalization makes it have no overall effect. A look at one piece of gross evidence--the massive return of combat troops post World War II does not seem to show evidence of a killing spree; homicide rates fall during the war (not surprising; we shipped our prime homicide age overseas by the millions, plus national emergencies tend to surpress both crime and suicide); rates return to their 1940 level in 1946, then fall rapidly, which is not what I would expect if combat is really so brutalizing. But lots of things changed, yadda yadda. The point is, there is usually more than one plausible story. To check whether your story is true, you need data.

The Times article purports to provide data. But it does not. It provides raw numbers without reference to the size of the relevant population. This is statistical garbage. I'm sure that some Ivy League graduates committed crimes during the same period, and some of them probably blamed it on their high stress upbringing. But this does not tell us anything without knowing how many Ivy League graduates there are.

Humans are designed by evolution to make some bad cognitive snap judgements. We are very prone to the post hoc, ergo propter hoc fallacy, because that was the safest way to bet when you're a naked, vulnerable east african plains ape. These heuristics are the best we could do on the veldt, but with time for cooler reflection, the problems become obvious.

In any group of a couple of million people between the ages of 18 and 24, some will be mentally unstable. Some will snap without warning--"He was such a nice boy". Some will commit spectacular, inhuman, surprising crimes which no one will ever hear about outside of their local area because unfortunately, young people snapping is not rare enough to make the national news. Some will drive drunk and be charged with vehicular homicide. Some will abuse their children. Obviously the people in their lives who are shocked by this behavior will associate the fact that they went to Iraq with the change, in the same way that parents of children with autism associate vaccines or other random events with the change that occurred in their children around age two. But that doesn't mean the event caused the change.

To know whether one event caused another, you would first want to know that the event you are seeking a cause for was actually unusual. That's why you need to know whether the homicide rate among veterans actually spikes when compared to a) soldiers who didn't go to war b) similar demographic groups among the population. The New York Times doesn't even try. Their work is not merely anecdotal silliness; it's anecdotal silliness masquerading as actual information. This puts my hackles up.

TrackBack

TrackBack URL for this entry:
http://meganmcardle.theatlantic.com/cgi-bin/mt/mt-tb.cgi/18686

Comments (92)

I'm having trouble squaring all of this with your attack on the homeless veterans number (which I'm very confident is correct) on the basis that it flies in the face of common sense.

To clarify: I'm confident that your criticism is correct, not that the original claim is.

The horrors of WWII made the people of Europe, unlike us sheltered folks in the USA, recognize the and reject the madness of war, right? Or so I hear.

I didn't bother to read the article, but I saw the headline with all the photos and remember thinking to myself "Well, there have to be several hundred thousand -- maybe a half million or more -- veterans who served in Iraq and are now stateside -- and you'd expect a decent number of homicides to have been committed by such a large group of (mostly) youngish males." Well, thanks for pointing out, Megan, that apparently I was wise not to have bothered reading the piece.

The NYTimes article is an example of extremely lazy journalism. Their methodology appears to have been the absolute minimum that could have been done to avoid the label of "purely anecdotal evidence". Indeed, the spokesperson from the Pentagon points out the obvious flaw in even this minimal statistical treatment- that the pre-9/11 media reports of homocide might have been less likely to focus on the enlistment status of the accused.

Just because it is difficult work to adjust for demographics in statistics like these does not absolve one from having to do so in order to establish the facts that support the overall assertion- that the wars in Iraq and Afghanistan have increased the homicide rates of its direct participants. Without this work, and this work is definitely not in Part I, where is should be, then the entire article really has no basis other than to report that some veterans of the wars have been killers.

Megan, I understand why this raises your hackles. But I think you are discounting the fact that it is impossible, at this moment, to get the data you want in order to write about this topic.

To speak to the question of whether exposure to combat leads to higher rates of violence, you need to look to studies of populations that have been around for long enough for studies to have been conducted -- i.e. studies of violence committed by Vietnam vets. There is wide agreement among studies of Vietnam vets that vets have higher rates of PTSD than non-vets and that combat vets have higher rates of PTSD than non-combat vets. There is dispute about whether Vietnam combat vets exhibit greater levels of violence than non-combat vets or non-vets. The Times article refers to this controversy in the literature.

News organs are not social scientists. They have to report on things that are going on right now, often before scientific literature is available to analyze the subject, and often in the absence of sufficient statistical information. The Times can't get stats on how many vets in total have committed violent crimes. It doesn't know whether the crimes it can identify are an undercount of 2 in 3 or an undercount of 1 in 4. What it can do is something analogous to what a social scientist might do in similarly limited circumstances: repeat identical methodologies during combat and non-combat periods. That's what it did. One problem with this method is the one the Pentagon spokesperson IDs, which is that the news media is now more sensitive to these vet-crime stories and may be picking them up more often. And the Times story includes this critique.

News organs are also storytellers. In situations where statistical data are not available, they often approach the truth through powerful stories. Then they do their best to try and determine whether those stories speak to something larger that's going on in society. The Times could have simply written a story about three particularly heartbreaking cases of Iraq vets whose service seems to have contributed to their subsequent violent episodes. Instead, it did the best job it could of trying to identify every such case it could find, and comparing the incidence of such cases to that in peacetime. It found a lot of stories, and a good deal more than in peacetime. What you're seeing here is not innumeracy. It's absence of data, and a news organization trying to tell a powerful and true story with the best information they could get.

Yikes! That should have been "homicide" in the first paragraph.

Incidentally, the relevant stat is probably not how many people have been deployed to Iraq, but how many people have seen combat in Iraq. In Vietnam about 15% of those deployed to the theater were in combat situations. I don't know what the stats are in Iraq or Afghanistan.

What if the U.S. started innoculating 13% of children
in the U.S. with a vaccine called BTS and then ramped it
up to 65% in 1999 and to 76% in 2006? But, the BTS Vaccine
doesn't contain mercury or Thimiserol. And what if the
negative effects of this drug were the following on babies
between the ages of 0 and 6 months:
- statistically significant social skills delays at 6 months
- statistically significant motor skills delays at 6 months
- decreased sleep duration by 6%-8% during the first 6 months of life
- increased rates of sleep apnea (lack of oxygent to the brain)
- 500% to 2500% increase in deformational plagiocephaly
- increased rates of torticollis and shoulder retraction
- 43% increase in sleep awakenings (this causes sleep stage
fragmentation of the four NREM and 1 REM stages of sleep)
- 40% increase in the duration of sleep awakenings

Considering that both plagiocephaly is associated (I'm not suggesting
causality) with neurodevelopmental delays and that motor skills delays
are associated (again, I'm not suggesting causality) with speech and
language disorders one would think the BTS vaccine would be investigated.

Why isn't the BTS Vaccine investigated??????????

What if it was a drug with all these negative effects? WOuld it be
investigated then? Certainly.

Actually, there is not BTS Vaccine or BTS Drug. What I am actually
referring to is the negative effects reported in Peer Reviewed Journals
(cited in the articles below) which are associated with the SIDS Prevention
"Back to Sleep" Campaign. The above negative physiological effects occur
when a babie is put to sleep on their back and not their stomach.
Unfortunately, because the "Back to Sleep" campaign is not a drug or a
vaccine and just a change in sleep position it seems that the scientific
community cannot grasp the enormity of the damage this is doing to
a generation of children.

Yes, there is a genetic basis for many types of autism.

But, I believe the Autism Epidemic is primarily caused by the SIDS prevention
"Back to Sleep" campaign regardless of it's positive intentions and results.
How much of the SIDS reduction is actually due to a reduction in reporting,
a decrease in maternal smoking, increased developments in technolgoy, decreased
sleep duration of back sleeping infants, and increased safe bedding
awareness?

If the the negative effects of the SIDS prevention "Back to Sleep" Campaign
were caused by the SIDS Prevention BTS Vaccine you bet your bottom dollar they
would be investigated?

If the negative effects of the SIDS prevention "Back to Sleep" Campaign were
caused by the SIDS Prevention BTS drug you bet your bottom dollar they would
be investigated?

But, since these negative effects are caused by everyones favorite feel good
program they cannot be investigated. This year another 3 million plus babies
will be subjected to this campaign. When will it be validated? When will it
be investigated?

http://www.thescientificworld.com/SCIENTIFICWORLDJOURNAL/toc/TSWJ_ArticleLanding.asp?jid=141&FromPage=Toc&ArticleId=1785&navFrom=Toc&From=Result

http://www.oandp.com/edge/issues/articles/2006-12_02.asp

http://www.oandp.org/jpo/library/2004_04S_005.asp

http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1595182

1. You know what would be awesome? Rather than merely saying that the New York Times should do this research, you should do the research they failed to do! But that would, eep, require actual journalism, which as a member of the so-called new journalism you are unwilling to do. You've said it would be easy for the NYT to do the necessary legwork. What's good for the goose....

2. In the comments section of the last post, you again said that the NYTs should compare the homicide rate of recent vets to the general population. As I said before, though, that comparison is meaningless. They/you/someone needs to compare the homicide rate of recent vets to people of similar demographical composition, in terms of age, income, education, etc. Otherwise it's all hot air.

3. I don't understand why there can't be value to the story simply in telling the stories of men who have come back from war to commit terrible crimes.

4. You have offered no defense for the fact that you're saying that reporters at the New York Times don't know many members of the armed forces. So look: even if that were true, it's meaningless ad hominem; you don't know it's true, and you've offered no evidence to suggest that it is; and it begs the question of precisely how many vets you have through your door in a given week. The idea that only people who regularly interact with the armed forces can meaningfully comment is stupid; the idea that people at the Times don't interact with members of the armed forces is unsupported; and the casual, guilt-by-association smear of the Times and anti-war people in general as being clueless about and unaffected by the war is far beneath someone of your intelligence.

Ms. Mcardle, you write:

"The Times article purports to provide data. But it does not. It provides raw numbers without reference to the size of the relevant population."

But what about this?

"The Times used the same methods to research homicides involving all active-duty military personnel and new veterans for the six years before and after the present wartime period began with the invasion of Afghanistan in 2001.
"This showed an 89 percent increase during the present wartime period, to 349 cases from 184, about three-quarters of which involved Iraq and Afghanistan war veterans. The increase occurred even though there have been fewer troops stationed in the United States in the last six years and the American homicide rate has been, on average, lower."

Well at least Megan is no longer demanding that the NYT compares the homicide rate of Iraq veterans to that of the general population. Now she asks that they compare the vets to "similar demographic groups among the general population." OK, a slight improvement.

Like brooksfoe, I found only one statistical claim in the article that involves rates: the comparison of the number of homicides involving vets during the Iraq war with the number of homicides involving vets during the six years prior to the war. And even that comparison is presented with the caveat that the methods used in the research are weak due to lack of data, etc. The Times makes no other statistical claims in the article. It is explicitly presented as a story, not as a scientific study. Why then is the Times taken to task for being insufficiently rigorous? Now, one could certainly take ideological issue with the "war is hell" slant of the article, just as conservatives would like to see fewer "All Quiet on the Western Front"s or "Deerhunter"s and more John Wayne heroics and Hogan's Heroes. But criticizing the story on the basis of statistics is really bizarre.

Freddie,

Your first point is a valid one- someone could do a better, much better job. Indeed, I think it would be great if Megan did just that.

Your third point would be a good one, too, but that is not what the writers did in this case:

Individually, these are stories of local crimes, gut-wrenching postscripts to the war for the military men, their victims and their communities. Taken together, they paint the patchwork picture of a quiet phenomenon, tracing a cross-country trail of death and heartbreak.

The writers did not limit themselves to just telling these individual stories. They wanted to assign a distinctive cause to this particular subset of homicides, but then failed to do the necessary work to support the hypothesis.

One minor quibble is that comparing WW-2 with Vietnam/Kuwait/Iraq, etc. is kind of an apples-to-oranges comparison for two reasons:

1.) WW-2 was a necessary 'Just War' whereas Vietnam and Iraq were/are not. It seems plausible that teaching young men to kill a real, legitimate enemy is likely to do less damage to their internal moral compasses than teaching them to kill an 'enemy' who really isn't.

2.) I don't have my copy of Lt. Col. Dave Grossman's book "On Killing" available right now, but if I recall correctly, he points out that, even as late as WW-2, the vast majority of U.S. combat soldiers never even fired their rifles, and those who did mostly fired over the heads of the enemy. But by the time Vietnam came along, the Armed Forces had gotten *MUCH* better at training (brainwashing?) soldiers to overcome their natural reluctance to kill.

So, the fact that WW-2 veterans came home and immediately beat their swords into ploughshares doesn't say anything one way or another about veterans of more recent wars (e.g. Tim McVeigh.)

Yancey, one point I think we can all agree on is that this:

"Taken together, they paint the patchwork picture of a quiet phenomenon, tracing a cross-country trail of death and heartbreak."

is one of the worst-written sentences in the annals of twenty-first century journalism -- and the competition is fierce.

Yancey, one point I think we can all agree on is that this:

"Taken together, they paint the patchwork picture of a quiet phenomenon, tracing a cross-country trail of death and heartbreak."

is one of the worst-written sentences in the annals of twenty-first century journalism -- and the competition is fierce.

I agree with both of you. What the Times has done here is very irresponsible. I just think it says more about contemporary journalism than it does about the political mores of journalists who write about the military.

Freddie wrote: You know what would be awesome? Rather than merely saying that the New York Times should do this research, you should do the research they failed to do! But that would, eep, require actual journalism, which as a member of the so-called new journalism you are unwilling to do. You've said it would be easy for the NYT to do the necessary legwork. What's good for the goose....

Given your penchant for turning context sideways, I'll bet that you snore N's at night.

It may have come to your attention that when our host does journalism, it is published in The Atlantic's print space. When she engages in open-discussion punditry, it appears on the blog space.

Sure, it would be interesting if she put together a more in-depth counteranalysis of the statistical issues at stake here. But on the other hand, her real journalsim responsibilities, which she evidently does engage in as evidenced by her print piece from two weeks ago and ongoing salary from her employer, possibly don't allow time for a twenty-hour delve into returning veteran statistical arcania.

Your jilted-lover verbal spew notwithstanding.

1.) WW-2 was a necessary 'Just War' whereas Vietnam and Iraq were/are not. It seems plausible that teaching young men to kill a real, legitimate enemy is likely to do less damage to their internal moral compasses than teaching them to kill an 'enemy' who really isn't.

"Just war/legitimate enemy" -- by whose definitioneering? This is unhelpful goalpost hologramming at best. A mentality of believing that you get to define when the enemy is legitimate based on personal prejudices is, if anything, a lot more likely to produce a homicidal psychopath (such as a Unabomber or a McVeigh) than for someone simply acting under well-defined orders.

2.) I don't have my copy of Lt. Col. Dave Grossman's book "On Killing" available right now, but if I recall correctly, he points out that, even as late as WW-2, the vast majority of U.S. combat soldiers never even fired their rifles, and those who did mostly fired over the heads of the enemy. But by the time Vietnam came along, the Armed Forces had gotten *MUCH* better at training (brainwashing?) soldiers to overcome their natural reluctance to kill.

This is more helpful, but also lacks a lot of context. Perhaps most importantly, the modern solider may be trained much more effectively at killing, but in the context of the US at least, they are also trained a lot more rigorously in how to identify and avoid civilians while also being required to engage in a lot of humanitarian work after the initial combat phase is over. IMO that would be more likely to promote compartmentalization skills.

Can anyone reconcile this passage from the first web page:

The New York Times found 121 cases in which veterans of Iraq and Afghanistan committed a killing in this country, or were charged with one, after their return from war. In many of those cases, combat trauma and the stress of deployment — along with alcohol abuse, family discord and other attendant problems — appear to have set the stage for a tragedy that was part destruction, part self-destruction.

Three-quarters of these veterans were still in the military at the time of the killing. More than half the killings involved guns, and the rest were stabbings, beatings, strangulations and bathtub drownings. Twenty-five offenders faced murder, manslaughter or homicide charges for fatal car crashes resulting from drunken, reckless or suicidal driving.

With this passage from the second:

This reporting most likely uncovered only the minimum number of such cases, given that not all killings, especially in big cities and on military bases, are reported publicly or in detail. Also, it was often not possible to determine the deployment history of other service members arrested on homicide charges.

The Times used the same methods to research homicides involving all active-duty military personnel and new veterans for the six years before and after the present wartime period began with the invasion of Afghanistan in 2001.

This showed an 89 percent increase during the present wartime period, to 349 cases from 184, about three-quarters of which involved Iraq and Afghanistan war veterans. The increase occurred even though there have been fewer troops stationed in the United States in the last six years and the American homicide rate has been, on average, lower.

Of what were there 349 and of what were there 121?

why isn't anyone linking the NYT emotion-fest with the bill, winding its way through Congress, to disarm(bar from gun ownership) returning Iraq-War veterans (?)

http://www.veteransnetwork.net/archives.php?edition=18

Of what were there 349 and of what were there 121?

And isn't the relevant number "about three-quarters of" 349 -- i.e. 262?

Their work is not merely anecdotal silliness; it's anecdotal silliness masquerading as actual information. This puts my hackles up.

Sort of like a professional political blogger building a post around an urban legend about political correctness?

Like David Cohen, I am puzzled by the two sets of numbers purportedly describing the same subset of homicides, however, I did notice a grammatical difference in the two descriptions which may hold the key:

The first number, 121, refers explicitly to discovered reports of Iraq and Afghanistan veterans that committed or were charged with homicide. The second number, 349, is described as the number of reported homicides "involving military personnel.

Does the word "involve" mean the 349 committed or were charged with homicide, or does it mean the 349 were involved in some other way, such as a witness, victim etc.?

... and some will snap because of PTSD. I think maybe being involved with a war increases one's chances of developing that particular instability, but I don't have any proof and I'm a liberal so I'm biased.

Yancey: That's a problem, too, but they seem to use "killing," "homicide" and "murder" as interchangeable terms. If they're not -- if they are comparing "homicides" from 1995 to 2001 with killings from 2001 to 2007 -- then they've drifted into out-and-out lying. Obviously, since homicides are a subset of killings (and in the 2001-2007 period they are explicitly including negligent, reckless and justified killings), there are always going to be more homicides then killings and saying that there were fewer homicides from 1995 to 2001 then there were killings from 2001 to 2007 would be completely meaningless.

I should note that they claim that the methodology was identical across the periods, so presumably the types of incidents they are counting are the same.

On the other hand, I wonder if they corrected for the likelihood that more information is available on-line post-2001.

I'm bookmarking this post so that I can point every glibertarian who writes a post hyping the horrors of socialized medecine/the nanny state/immigration/gun grabbers/etc. etc. etc. using the latest anecdote back here to Megan for a hearty spanking.

"But I think you are discounting the fact that it is impossible, at this moment, to get the data you want in order to write about this topic."-brooksfoe

Yeah, why let the absence of data stop you from offering opinions confidently?...

And let me just add, parenthetically, that like Megan, I have the utmost respect for Brooksfoe. He had dazzling SAT scores (as he assures us), after all.

Humans are designed by evolution to make some bad cognitive snap judgements. We are very prone to the post hoc, ergo propter hoc fallacy, because that was the safest way to bet when you're a naked, vulnerable east african plains ape.

So... Megan's rebuttal to a plausible pop sociology story that may or may not be true is to tell a plausible pop evolutionary psychology story that may or may not be true. Bonus wanker points for the all-too-common but nonetheless dismal teleological word choices in the phrase "humans are designed by evolution..."

Why do you respect brooksfoe? I haven't seen him/her make one insightful comment ever. Just being nice?

121 cases in which veterans of Iraq and Afghanistan committed a killing in this country, or were charged with one

What happened to the presumption of innocence?

"Why do you respect brooksfoe? I haven't seen him/her make one insightful comment ever. Just being nice?"-Person

Person, I for one am deeply angered by your unjust attack on that great thinker. Never mind that he thinks one doesn't actually need data to make empirical generalizations and that he fails to understand the nature of a market economy, he's a master of standardized tests--if his own reports are to be trusted.

Now, you might say that Megan is easily impressed, but I too have nothing but admiration for brooksfoe. It takes real talent to make as much a hash of economics as he does on a regular basis.

Come, come. Anyone who watches a crime drama on tv knows that the great majority of crimes seem to be committed by businessmen or wacked out druggies.

Megan,
By these strange standards you've set for what is "actual information" and what is merely a "story" or "anecdotal silliness"—(Teeheehee! These stories of death and despair and violence are SOOOOOO SILLY!)—can we assume that you also did not think that the Abu Ghraib story was worth reporting? After all, wouldn't we first need to determine how the incidences of homicide, torture, and sexual abuse in that prison compared to incidences among the general population of Iraq before we get all bent out of shape? I mean, isn't it just a story about a few primates who went haywire out on the veldt, or something? Why should we single out that place or those people, when somewhere in Iraq right now there's a murder being committed and a man being sexually assaulted? And wouldn't those guys probably have suffered similar fates at the hands of their countrymen or Al Qaeda or the Iranian Revolutionary Guard, anyhow? I mean, how can we say anything for sure about Abu Ghraib until we can get some REAL DATA and do some serious statistical analysis!


Why do you respect brooksfoe? I haven't seen him/her make one insightful comment ever.

That carries no more weight than the usual detritus spilled by the claptrapped gaggle of mouth-breathers who drop by to snidely explain MM's multitude of failings. When translated out of partisanese, it means, "Has never made a comment that strongly conformed to my library of well-polished prejudices."

And keep in mind that I've argued with brooksfoe at least as much as anyone else here. He plays hardball and sometimes he loses, but he always comes back the next morning with a smile and no grudges.

tinisoli, are you really as obtuse as you pretend to be?

You cite what appear to be statistically valid studies for points which are somewhat related to the article under discussion. Nobody is arguing that these studies do not exist; we are arguing that the lack of data makes the NYT's point in the current article at best questionable and unsupported.

Then you bring up Abu Ghraib. But the point of that story was not that soldierly misbehavior was becoming more common due to the war, it was that such misbehavior occurred at all--a qualitative rather than quantitative point. We generally start from an assumption that our troops will not stack prisoners up naked and then take pictures of them; that they did so at all is news. On the other hand, the fact that some veterans commit more ordinary crimes is not particularly surprising; there have been military laws, military lawyers and courts martial for as long as there have been armed forces. Soldiers are not angels. We are (or we should be) aware that such things are possible. That is, the qualitative point is not news; it is the quantitative one that is interesting.

The NYT made a quantitative assertion based on both a flawed methodology (searching of news reports) and numbers that, even had they been gleaned in a proper manner, were missing key components such as the denominators. Those of us with a rudimentary knowledge of statistics (and my knowledge is, indeed, rudimentary) object to this practice, regardless of the SAT scores of the practitioners.

This is not difficult to understand.

The most famous Black Economist, Walter Williams endorses Ron Paul.

http://en.wikipedia.org/wiki/Walter_E._Williams

brooksfoe,

First, you cannot "approach the truth" on the question of whether military or combat experience increases the probability of subsequent violent behavior by "telling powerful stories" about a few individuals. Anecdotes are not evidence of general patterns or trends. Anecdotes may be useful for illustrating or describing particular kinds of event or experience, but they are not a statistical tool.

Second, even in the unlikely event that the New York Times really was doing "the best job it could" to answer the question by looking for "stories" it should have realized that its methods were utterly inadequate to the task.

I have a generally high opinion of the New York Times. It still is the paper of record for me, and I think its overall quality and comprehensiveness put other papers to shame. But its liberal bias sometimes gets the better of it even beyond the editorial pages, and this seems to be one such case.

anony-mouse: I can call (and have called) comments insightful, even when I disagree with the poster's ultimate conclusion. I call a comment "insightful" when it raises an important, non-obvious point, or phrases a known point in a more convincing way. "Agrees with Person" is not part of the criteria. Bias inevitably creeps into all of what we do, but your derision is unwarranted.

"brooksfoe" has never made a comment meeting the condition listed above. Megan_McArdle is devoting an entire blog post to addressing a brooksfoe comment, indicating that it's one of his better ones else she probably wouldn't feel the need to feature it. That comment makes a pretty appalling error -- which should tell you something.

I ask about insightful brooksfoe comments so I can update my estimation of him, and learn something if the comment is in fact insightful, NOT because I like insulting people (although I like that too).

Rob,

Where does the Times article state that it is making a grand quantitative statement or that it is not a predominantly anecdotal article? IT DOESN'T. It uses some numbers to suggest a point, with several caveats, and it fills most of the space with real-life stories to give us "actual information" on some of the cases. That's all it does, and that's all it purports to do. The attack that Megan has launched is premised on her own rigid concept of what is interesting or meaningful information, not to mention her own conception of what is normal or unexceptional behavior among military personnel. She is simply dressing it up under the guise of statistical standards. Furthermore, she is rejecting the possibility that other people may be interested in the article even though it is largely anecdotal or inadequately scientific according to her standards. Given the glossy, simplistic caricature that so many right-wingers have concocted of the American soldier——anyone remember what they thought of Pat Tillman before he turned out to be an atheist who questioned the legality of the Iraq invasion?——it is interesting to actually read true-life accounts of what has happened to some
servicemen upon returning home. And the connection between PTSD and both substance abuse and violent crime has already been taking shape for decades in the scientific literature, even if that connection remains unproven or was given short shrift in this article.

The Abu Ghraib analogy is legit. Why is the surprise or interest that you felt after Abu Ghraib more legitimate than the surprise or interest that a Times reader may have felt while reading Sunday's article? If we can expect servicemen to drive drunk and shoot their girlfriends and hide out in the woods after shooting a stranger at a party, why is anything at Abu Ghraib so anomalous? And if you reject the anecdotal nature of the Sunday article, why did you not dismiss Abu Ghraib as mere anecdotal silliness? Why did you not cry out for quantitative data in that case?

Mixner,

Anecdotes may be useful for illustrating or describing particular kinds of event or experience, but they are not a statistical tool.

MMMMMM. That's DEEP. Now please point to the line in the Times article in which the reporters stated that their research methodology and assembly of stories serves as a "statistical tool".

You guys have simply turned an article in a goddamn daily newspaper into some kind of bad-science-paper straw man. And you are denying the possibility that the stories in the article could be interesting and at the same time not lead us to conclude anything about the causal relationship between war and crime at home. Also, we'd miss out on quite a few stories if reporters waited for data to emerge that pointed to a clear trend. Shall we just forget about this unpleasant PTSD stuff until we've let the academics sort it all out sometime down the road?

'Sorry, Private Pyle, but we can't give a shit about your story just yet because Megan McArdle needs more data. For now we'll just throw your silly stories into the Unserious anecdote pile. But hang in there, kid. We'll be with you in a few decades. Keep an eye on the econ blogs for updates. And thanks for your service to the nation.'

In short, he doesn't meet your definition of insightful, which in turn is based on noble-sounding criteria that, at the root, still break down to the subjectivity of your own sculpted views.

How is that different than what I said?

Where does the Times article state that it is making a grand quantitative statement or that it is not a predominantly anecdotal article? IT DOESN'T.

Let's go to the tape. Or more precisely, let's go to a quote that Yancy has already posted upthread for your convenience:

Individually, these are stories of local crimes, gut-wrenching postscripts to the war for the military men, their victims and their communities. Taken together, they paint the patchwork picture of a quiet phenomenon, tracing a cross-country trail of death and heartbreak.

A "postscript to the war" suggests a causal connection, and a "quiet phenomenon" suggests the same thing. Given that they haven't shown a correlation, causal claims are rather strong.

In addition the choice to publish this article at all suggests that the NYT's editors believe there is a causal connection here. They don't publish stories about all the murders in the country, or about all the drunk-driving accidents; the only plausible reason to publish those committed by veterans is a belief that they are somehow connected to military service.

We may simply disagree about Abu Ghraib. The weird semi-sexual abuse (or sexual semi-abuse, or whatever) that occurred there was...weird. It would be weird if committed in an US prison, whether by guards or by inmates. It would be weird, although not criminal, if it was the regular practice of some fetish club somewhere. It's even weirder in that it obviously involved the participation (or at least tolerance) of a large number of people; most of us would like to think that if we stumbled on an obvious crime, we'd say something.

A guy shooting his girlfriend or hiding from the law after doing so is not all that weird. It's somewhat depressingly routine even among those with no combat experience. The only thing that makes these cases special is the alleged connection to military service--a connection which has not been established.

I said precisely the same thing two days ago on my blog. Nice job, Megan; great minds think alike.

anony-mouse: Because your previous comment disallowed the possibility that I would deem as insightful a comment whose conclusion contradicted my worldview, while the criteria I outlined demonstrably allows for this.

Now, your turn to substantiate your ramblings: if the bias of everyone eradictes the merit of their remarks, what is the point of pursuing the truth, let alone discussing any matter in the comments section of a blog?

Person just doesn't see brooksfoe's profundity as I do. Look at this, for example:

"Meanwhile, it is false to claim that tax cuts are a good way of using ficscal policy to fight recessions."-brooksfoe

Aren't you dazzled by the brilliance? Brooksfoe has discovered something about economics that even Mankiw and Bernanke have never understood.

Brooksfoe ought to write his own textbook.

I wonder about the timing. It seems like a week or two since the story about the missing/murdered pregnant Marine and her alleged killer surfaced (both of whom were personnel clerks a Camp LeJeune, no overseas service.

I can picture someone in the NYT office wondering about this new cross-country trail of death and heartbreak, going to LexisNexis or whatever and searching ((homicide OR murder OR manslaughter) AND (veteran OR military OR PTSD)).

Track down a few of the alleged perpetrators of the more lurid articles for quotes, get a pro forma denial from the Pentagon, sprinkle liberally with tired cliches and alarmist prose and BAM there's your front page story.

rwe: Thanks for the support, but the last one isn't bad provided it has justification.

The insightful comment question is still open. brooksfoe, you're not disqualified from nominating.

Megan is absolutely right about the New York Times' editorial staff's lack of familiarity with the military.

Remember, this is the same publication that referred to a medal called "The Purple Star," and said that the Congressional Medal of Honor was an award given out to songwriters, and that bolt-on armor for humvees could stop a 155mm artillery shell. That's just the howlers I can name off the top of my head.

Incidentally, of the lowest ten zip codes for military recruiting per capita, seven of them are immediately around the New York City area.

I would be extremely surprised if any editorial staff employees of the New York Times are currently drilling reservists. They may have some on the physical plant, business or printing/distribution staff, but I see no evidence that there is any recent military experience in their news room at all.

It uses some numbers to suggest a point, with several caveats, and it fills most of the space with real-life stories to give us "actual information" on some of the cases. That's all it does, and that's all it purports to do.

That may be a mostly-accurate description of what it purports to do, but that definitely ain't what it does. Have you read it?

Granted: It appears the contributors have probably identified someone who suffered from PTSD or was inadequately screened before entering and/or exiting the military. However, they then take a series of events that may or may not have been PTSD driven and which readily occur for similar circumstances among the civilian population, and try to build the case on that. Why? What "point" are they trying to "suggest" here, especially when (as others have noted) no effort is made to show whether this is a statistically relevant problem, or merely examples of news that normally doesn't even make page 10 these days unless it contains either extraorinary circumstances or a (pardon the expression) white-hot blonde under the age of thirty? To review their case studies:

1. Matthew Sepi: 130lb man confronted at night in a bad neighborhood by two 200lb armed gang members, one of whom was higher than a kite. Killed one and wounded the other. Happened to use an illegally-concealed assault rifle in the event, but could have achieved the same effect, completely legally, with any decent-sized handgun and a CCW permit.

2. Noah Gomez: Confronted and shot a car thief. Committed suicide afterward. Could just as easily have been a civilian who was sick of being victimized by low-lifes, then had immediate regrets after the deed was done.

3. Archie O'Neill: Acquired a mistress at some point after acquiring a wife, shot mistress after she threatened to kill his family while he was away. This tragic story is as old as mankind.

4. Stephen Sherwood: enlisted to get health insurance after his wife got pregnant. His wife later tells him she wants to end the relationship -- after he has made an enormous personal sacrifice for her very sake -- and he then commits a murder-suicide. This tragic tale is quite old, too.

5. Seth Strasburg: This one looks like a clear and unquestionably legitimate example of the article's thesis.

6. Lucas Borges: Started out with a drug history, stepped up to a more power agent, drove recklessly while high and killed a young woman in another vehicle. Another tragic story that occurs daily.

In short, of six accounts the article cites, PTSD and/or supervision problems very probably contributed to the outcome; but only one gives clear support for the thesis. The events that took place in four others are extremely routine in terms of actors, tools, and surrounding circumstances. A fifth one was merely self-defense, save for judgment errors by the defender. And all of this, intermingled with a liberal serving of negative scare language that implies an enormous problem, falls under a heading of "War Torn Part 1: Across America, Deadly Echoes of Foreign Battles."

Mmmffff. For the sake of those portrayed, I hope they (or their families, for those who committed suicide) get the help they need. But the article isn't exploring the questions that really matter; it is presenting a sensationalized narrative with poor contexualization. Y'know, like Fox News.

Incidentally, of the lowest ten zip codes for military recruiting per capita, seven of them are immediately around the New York City area. -Jason Van Steenwyk

Per capita stats recruitment stats don't tell you much when you're looking at the most densely populated metropolitan area on the continent. Implying that New York isn't doing its part in the military because its per capita involvement is low is like saying that New York didn't suffer all that much on 9/11 because its per capita losses weren't all that bad.

Per capita stats recruitment stats don't tell you much when you're looking at the most densely populated metropolitan area on the continent.

Now this is just plain bizarre. Per capita numbers (unlike the numbers in the NYT article under discussion) are the best way to talk about recruiting success; it would make no sense whatsoever to talk about the absolute number of recruits from NYC because of the density. But the per capita number (assuming no odd demographic biases, such as a lack of young people) is exactly what you want to gauge whether, say, a recruiter's office in Manhattan was pulling its weight. If there's a lot of people, there should be a lot of recruits to make up for that.

tinisoli, you seem to be at war with conventional statistical assumptions: 121 "killings" is newsworthy even if we don't know the denominator, but low per capita recruitment numbers in the NYC area are expected and normal because the denominator is so unfairly big. Huh?

anony-mouse,
That's really nice that you've summarized the Times article so that the events are now retold in your own nonchalant, clinical voice. But you're still glossing over all those pesky-if-unquantifiable details from friends, family, law enforcement personnel, and the still-living perpetrators of some of those crimes; and many of those details do indeed suggest that the experiences those men had in Iraq and the PTSD that some of them displayed and/or were diagnosed with upon returning did indeed play a role in the tragedies that unfolded. You also gloss over the fact that several of the men were A) carrying deadly weapons, B) deliberately seeking out dangerous situations or finding comfort in combat-like environments, C) behaving in ways that were unrecognizable to their friends and family and utterly inconsistent with their pre-service criminal histories. You are essentially telling those people, No, sorry, you're all just deluding yourselves because you don't want to admit that maybe your loved one was just another wacko. Real nice.

Rob,
Jason pointed to the NY metro area's low per capita military recruitment as evidence of its indifference toward military service. If he looks at per capita recruitment as a measurement of military involvement, it's always always always going to make huge megalopolises like New York look bad. There's always going to be a bias against big cities, yes, because the denominator is so big and so full of people who cannot serve. Military recruiters do pretty well in urban areas, regardless of this bias, but in all regions the driving force is economic opportunity, or lack thereof. Period. It's not about what the local paper of record writes, despite what Jason would have us believe. Recruiters target young people based on proven demographic criteria. Here's a good read on how this gets done in one region:
http://www.boston.com/news/nation/articles/2004/11/29/military_recruiters_pursue_target_schools_carefully/

However we hash this story, Megan... Thanks for taking it on in this way. Like a lot of other category stories about military vets [suicide, etc.] it is easy to find a truism that fits, for a reporter and run with it. Having a healthy discussion that is new and improved "Now WITH numbers!" would certainly be helpful. Even though statistics can be made to fit any idea... at least having some gives a common frame for us to reference...

tinisoli wrote:

Jason pointed to the NY metro area's low per capita military recruitment as evidence of its indifference toward military service.

Oh, jeebus. No I didn't. It may be evidence of that, it may not be. You'd have to look at some other factors, such as age distribution, to support that idea.

But you DON'T have to control for age distribution to support the idea that reporters drawn from that demographic are less likely than the population at large to be closely associated with an active duty military member.

Megan's point was limited to the notion that the New York Times reporters are unlikely to know many active duty military personally, as was mine.

Rob Lyman wrote, "In addition the choice to publish this article at all suggests that the NYT's editors believe there is a causal connection here."

This is the point I've been turning over in my head. The story is a feature, not breaking news. It could have been written last year or next. It could have waited until an academic study had actually provided some evidence.

I don't know why the NYT decided to pursue this story; I suspect a reporter or editor ran across some of the individual stories and thought they had the tail of something.

It's incredibly saddening. Like all accounts of senseless death and wasted lives, these are tragic tales.

But how, exactly, is a special series on "veterans ... who have commited killings" any more news than a special series on "veterans who have saved lives" or "veterans who are good people."

Sound dumb? In the 1982 Rick Atkinson won a Pulitzer prize for (in part) a series of articles on the West Point class of 1966. He extended that into the brilliant book The Long Gray Line.

The book is harrowing, esp. its accounts of small unit tactics in Vietnam, but it is impossible to read without coming away with immense respect for the young officers portrayed.

'samatter, rwe, you got bad grades?

I'll leave aside the problems with using tax cuts as fiscal stimulus

(http://www.cbo.gov/ftpdoc.cfm?index=3251)

and just say I understand where Mixner is coming from, but that I think the overall connection between combat vets' PTSD and a propensity for violence is one of those gut instincts that has held up in most (not all) of the scientific literature:

"Studies have consistently shown that male veterans with posttraumatic stress disorder (PTSD) evidence higher rates of violent outbursts and aggressive behavior than those without the disorder, and exhibit more hostility expression
and poorer anger control (Beckham, Feldman,
Kirby, Hertzberg, & Moore, 1997; Kulka et al., 1990; Lasko, Gurvits, Kuhne, Orr, & Pittman, 1994; McFall, Fontana, Raskind, & Rosenheck, 1999)."

http://www.amydmarshall.net/files/PTSD%20and%20aggression%20JTS%202007.pdf?PHPSESSID=8921e17d0693ac7499c918f65517dca5

Obviously, combat vets have PTSD a lot more often than non-veterans do. I know a number of Vietnam vets. It's a self-selecting sample because I meet the ones who were strongly enough affected to return to Vietnam, but if I ever had any doubts that PTSD is a real diagnosis, I no longer do. It's not that these guys have been transformed into emotional wrecks or haven't gone on to lead happy and successful lives. But you can see the traces of how behavior alters in the aftermath of trauma: in extreme cases, obsessive jokey storytelling punctuated by sudden brief weeping (former Navy pilot/POW), or stubborn hewing to an ideological line about one's experiences while sighing and rocking back and forth (former Vietnamese interrogator). It's not necessarily different from behavior you'd see in other people exposed to very hostile, violent environments -- Holocaust survivors, ex-political prisoners, people with addiction or suicide in their families. It doesn't mean that everyone or even most people exposed to these environments will be severely damaged by them. It doesn't mean that pitying or coddling them is the best response; for many people, the best treatment is to forget about the trauma entirely and get a job in a supportive and demanding social environment. But when you do see the stories of people for whom it's gone wrong, to deny the importance of the time they spent in hell seems perverse.

And Megan's point was pulled right out of her bum, Jason.

She offered no proof or support to her "notion" that Times reporters don't know military folk (as though that would prove anything in and of itself), and neither did you. (The irony of Megan struggling so mightily to produce valid stats in two different posts about the absence of valid stats in the Times article is overwhelming.) Your application of per capita military recruitment by zip code was plainly inappropriate and misleading. You have to look at population size and density before you blindly compare two regions, especially if the factor you're looking at is at all density-dependent. That's why statisticians at the FBI or DoJ do not include per capita murder rates of small towns when they look at per capita murder rates of big cities: because a string of three murders in a town of 300 could make New Orleans look downright peaceful. And that would be really dumb and counterproductive.

It would also be really dumb to assume that the reporters who worked on the Times article all hail from New York in the first place. So there goes your theory about reporters who are "drawn from that demographic" and therefore ignorant of the plight of troops and their families, which was itself an assertion based on fallacy. Do you realize how many professionals in New York were not raised there? And do you understand that just because you work for a newspaper you may in fact live and work in another location?

tinisoli, I actually don't think it's inaccurate to claim that NY Times reporters do not tend to hail from that part of the US populace that provides military recruits. What is inaccurate is to claim that they don't KNOW any soldiers. They know a lot more soldiers than do most of the readers who are criticizing such an article, because they report on them.

Deborah Sontag, one of the two reporters on the story, was the Times' Jerusalem bureau chief from 1998 through early 2001 -- i.e. the second Intifada. She had young kids in school as reports of bombings came in. She watched the Twin Towers burn from the hotel where her family was living on their return to NYC. In the years since, she has written dozens of stories on the VA health system and the difficulties faced by returning Iraq and Afghanistan vets. She clearly knows plenty of soldiers, and she clearly knows plenty about PTSD. She may come at these issues from a viewpoint Jason Van Steenwyk disagrees with, but it's a viewpoint held by plenty of people involved -- including soldiers.

brooksfoe,

If there is a correlation or causal relationship between military/combat experience and an increased propensity for subsequent violent behavior, it can only be established by proper scientific study, not by popular journalism. The criticism of the New York Times here is not that there is no such relationship but that arguing for it on the basis of an informal survey of news stories is irresponsible.

However, reading the piece again I do notice that the writer mentions that they consulted police and military reports as well as media sources, so their investigation does seem to be a bit more serious and thorough than I previously thought. But it's still no substitute for proper scientific research.

However, reading the piece again I do notice that the writer mentions that they consulted police and military reports as well as media sources, so their investigation does seem to be a bit more serious and thorough than I previously thought.

You didn't notice this the first time you read the article? Or was that in fact the first time you read the article. Be honest, Mixner. Santa is listening.

But it's still no substitute for proper scientific research.

Arrrrrggggh. Who ever said it was!? It is still a mystery to me why Megan and some of you guys reacted to this article as though it was a fraudulent academic paper that somehow slipped through a review process and found its way onto the pages of a scientific journal. But maybe I'm just new to this here scene, and I don't yet comprehend the anti-NYTimes bent of this here blogger and her cohorts.


Henry,

with this: "The story is a feature, not breaking news. It could have been written last year or next. It could have waited until an academic study had actually provided some evidence.

I don't know why the NYT decided to pursue this story; I suspect a reporter or editor ran across some of the individual stories and thought they had the tail of something.

It's incredibly saddening. Like all accounts of senseless death and wasted lives, these are tragic tales.

But how, exactly, is a special series on "veterans ... who have commited killings" any more news than a special series on "veterans who have saved lives" or "veterans who are good people.""

you ask good questions.

see if this: http://forums.military.com/eve/forums/a/tpc/f/672198221/m/4130054051001/inc/-1
and: http://clusty.com/search?v%3afile=viv_878%4029%3aaWStVG&v%3aframe=list&v%3astate=root%7cN469&id=N469&action=list&sw=%7cH.R.%202640%7c&sec=1200378270&

help provide an answer to your Q's..

You have to look at population size and density before you blindly compare two regions, especially if the factor you're looking at is at all density-dependent.

I can see why you'd look at demographic distribution--a place with a lot of retirees is a bad one to recruit in--but what the hell does "density" have to do with recruitment success? Should dense areas have more recruits per capita because the average driving distance to the recruiting station is smaller?

Please explain this to me, because I don't get it. We routinely use percentages and per-capita rates when comparing areas of different densities: income in NYC is higher than rural Georgia, but the murder rate is probably higher too.

Why is recruiting somehow special, such that per capita rates are unfair?

I'll take it as a given that soldiers have PTSD more than civilians and that PTSD makes one prone to violence. The NYT article should have been able to show that the rate of murder and/or killing by soldiers returning from Iraq/Afghanistan is higher than that of civilians of the same demographic. If it is not, and I have seen some evidence that it is not, then THAT is an unusual finding worthy of investigation.

A more accurate article might have been: In spite of high rates of PTSD, soldiers have a lower average muder rate than civilians.

But the NYT didn't finish the article. They implied that, overall, soldiers returning from combat have a higher murder rate than civilians but never actually showed their work (assuming they did it).

As was mentioned above, they lumped murder, manslaughter, accidents, and reports of arrests (without convictions) together to get a number that, over a 6 year period, gives a very low "murder" rate when compared to the civilian population.

EI

What I find ironic is that whenever an article like this comes out, a few bloggers will expand upon the article by finding additional information and/or find errors/omissions, all within hours. And the MSM claims that they have an advantage over bloggers in their professionalism and access to fact-checking and better research resources. Apparently, they just don't use them.

EI

tinisoli,

I haven't encountered someone so eager to make bad arguments in a long time:

She offered no proof or support to her "notion" that Times reporters don't know military folk (as though that would prove anything in and of itself), and neither did you.

Yes, I did. I offered a number of gross factual errors that saw print in the Times that no one who had done even a single enlisted stint, or had an immediate family member who had done so would make: Purple Stars, a congressional medal of honor for songwriting, etc. Which means those howlers passed muster with the reporter, at least two layers of editors, a fact checker (what a concept!) and the copy editor who writes headlines.

I further buttressed the anecdotal evidence with the statistical evidence that New Yorkers are less likely on a per capita basis to join the military. In fact, they are statistical outliers in that regard, way out on the left end of the bell curve. That's statistically significant.

Your application of per capita military recruitment by zip code was plainly inappropriate and misleading.

No it isn't. When looking at whether a given set of persons (reporters) in a given area (NYC) personally know any active duty soldiers or drilling reservists on a personal level, the per capita rate of active duty soldiers is exactly the right demographic to use.

Further, it's not misleading in the least. It is what it is. New Yorkers are less likely to serve in the military. New Yorkers are therefore less likely to closely associate with someone in the military, to the very same extent that New Yorkers are more likely to closely associate with other New Yorkers than with the rest of the population at large.

You have to look at population size and density before you blindly compare two regions,

On a per capita basis? That's idiotic. Besides. I'm not comparing two regions. We're just talking about New York. You know, home of Pauline Kael, famous for her apocryphal exclamation upon hearing of Nixon's victory in 1972, "How can that be? Nobody I know voted for him!"

It would also be really dumb to assume that the reporters who worked on the Times article all hail from New York in the first place.

What would be even dumber than that is to assume that either Megan or I made that argument to begin with. But that doesn't seem to have stopped you.

I would posit that New York Times reporters are somewhat more likely to emerge from New York than from, say, Hopkinsville, Kentucky. Where people DO tend to know active duty military.

What is inaccurate is to claim that they don't KNOW any soldiers. They know a lot more soldiers than do most of the readers who are criticizing such an article, because they report on them.

That's hilarious. Reporting on someone != knowing them on a personal level. If you will notice, many of the people most critical of the article are themselves veterans, including myself.

Deborah Sontag, one of the two reporters on the story, was the Times' Jerusalem bureau chief from 1998 through early 2001 -- i.e. the second Intifada. She had young kids in school as reports of bombings came in. She watched the Twin Towers burn from the hotel where her family was living on their return to NYC.

Not relevant. You can do that for years and not know a single active duty US soldier or marine on a personal level.

In the years since, she has written dozens of stories on the VA health system and the difficulties faced by returning Iraq and Afghanistan vets.

That MIGHT be relevant. You would think she would write a better article then, instead of this crap.

She clearly knows plenty of soldiers

Again, reporting on someone != understanding them or knowing them well. Further, the soldiers a reporter on this beat would normally encounter - the ones wrapped up in the creaking military health care system, are not necessarily going to be representative.

and she clearly knows plenty about PTSD.

I don't see any evidence of that.


Rob (and to some extent, Jason)

We routinely use percentages and per-capita rates when comparing areas of different densities: income in NYC is higher than rural Georgia, but the murder rate is probably higher too.

But comparing the murder rate of New York with that of rural Georgia would not be very useful precisely because the population densities are so different. This is why every year people look at stats such as Cities (>100,000) Ranked By Murder Rate and not Zipcodes Ranked By Murder Rate. Nor do we find much use in comparing the per capita income of a Manhattan zip code with the per capita income of an Alaskan zip code. It has little value, because that population density affects so many factors.

Here are some real numbers:

The Manhattan zip code 10021 has about 100,0000 people of whom 7000+ are civilian veterans, or about 7%——well below the national average of 12.7%.

The Georgia zip code 30206 has 2258 residents, with 211 civilian veterans, or about 12.5%——right at the national average.

Now, in which zip code would you prefer to work in if you had to rely on veterans to provide information for human interest stories about war? The one with 7000 veterans, or the one with 211? I guess Jason would prefer the one with fewer people, since he is more interested in the place that has the higher percentage of vets. Which would you choose?

Jason,
I'm afraid that I'll have to let someone else respond to some of your recent comments, as it was their comments that you were attacking in the latter half of your last post. (I think you're looking for brooksfoe, IIRC.)
Now then...
When looking at whether a given set of persons (reporters) in a given area (NYC) personally know any active duty soldiers or drilling reservists on a personal level, the per capita rate of active duty soldiers is exactly the right demographic to use.
And how exactly, Mr. Wizard, did you reach this conclusion? Are you basing this on the assumption that New York Reporter knows 1000 people in her neighborhood, and so does Joe Blow in Kentucky; and because Kentucky has higher per capita military recruitment it's therefore more likely that Joe Blow's acquaintances will be military than it is likely that NY Reporter's will be ? Well, that's some neat and tidy arithmetic, Jason