Tuesday, December 27, 2005

Military Service and Socioeconomic Status.

Citizen Cain is a big fan of both Jane Hamsher and Digby. Both bloggers are irritated with Washington Post ombudsman Deborah Howell for her criticisms of a November 4 Ann Scott Tyson front pager about the socioeconomic profile of Army recruiting. Tyson's article was based on an analysis of military recruiting data by the National Priorities Project (NPP).

I agree with Hamsher and Digby that Howell's critique is weak, and that it seems to be based on not much more than noting that some people disagree with the NPP analysis, without providing much context for judging who is right. But the dispute caused me to look more closely at the Washington Post article, the NPP analysis and a critique of it by the Heritage Foundation. Based on my review of these sources, I would say that the Tyson article was misleading.

Tyson's article states:
Many of today's recruits are financially strapped, with nearly half coming from lower-middle-class to poor households, according to new Pentagon data based on Zip codes and census estimates of mean household income. Nearly two-thirds of Army recruits in 2004 came from counties in which median household income is below the U.S. median.

Such patterns are pronounced in such counties as Martinsville, Va., that supply the greatest number of enlistees in proportion to their youth populations. All of the Army's top 20 counties for recruiting had lower-than-national median incomes, 12 had higher poverty rates, and 16 were non-metropolitan, according to the National Priorities Project, a nonpartisan research group that analyzed 2004 recruiting data by Zip code.

I have several problems with this. First, despite Tyson's assertion that nearly half of recruits come from "lower-middle-class to poor households," the available data don't actually tell us the socioeconomic status of the households recruits come from. The data tell us what zip codes recruits come from, so at most we can say that nearly half of recruits come from lower-middle-class to poor geographic areas. Tyson alludes to this, but in a way that isn't at all clear. While common sense and experience tells us that many recruits are "financially strapped," the article doesn't add a lot to our knowledge of this topic.

Second problem is that "lower-middle-class to poor," depending on how you define, could be most of the country. But it seems obvious that Tyson is trying to say that recruits come disproportionately from lower-middle-class to poor" regions. It isn't clear that this assertion is true. You would never know from Tyson's article that the NPP data show that the poorest regions of the country (those with median incomes below $20,000 per year) are actually underrepresented among recruits. The very wealthiest regions are the least reprented among military recruits. Zip codes with median incomes between $25,000 and $55,000 are overrepresented. The Heritage Foundation, drawing on the same Department of Defense data, reports the data by income quintile, and show that in 2003 the poorest quintile areas are the least represented among recruits, the second richest quintile is the most represented, and the richest quintile is the second most represented. Taking the NPP and the Heritage data together produces a picture where the poorest regions and the very wealthiest regions are underrepresented among recruits, but middle class regions, including at least middle-middle (and upper-middle-middle?) classes are well represented. A somewhat different picture than what Tyson reported.

Third problem is that while Tyson doesn't directly say it, she implies that the military is relying increasingly on poor areas. Here's her lead:
As sustained combat in Iraq makes it harder than ever to fill the ranks of the all-volunteer force, newly released Pentagon demographic data show that the military is leaning heavily for recruits on economically depressed, rural areas where youths' need for jobs may outweigh the risks of going to war.
But in fact Heritage shows that the poorest and second quintile zip codes become less well represented among recruits between 1999 and 2003.

Finally, Tyson is mislead by NPP's analysis of the "top 20 counties for recruiting." In fact these are the 20 counties with the highest recruitment rate, not the highest number of recruits. All of these counties are counties with very small populations, where the addition of one or two recruits raises the recruitment rate significantly. Therefore, the recruitment rates for these counties isn't statistically meaningful, any more than it is statistically meaningful that there were many poor rural counties with zero recruits. Digby says that Heritage reports "a bunch of arcane gobblodygook that I defy Howell or anyone else to interpret." Actually, it isn't that complicated. A high recruitment rate from a county that had only a few recruits doesn't mean much.

Note: this post has been modified slightly since original posting.