Creative Destruction

October 31, 2006

US Military Recruits From All Classes, Races

Filed under: Science,War — Robert @ 1:54 pm

Some conventional tropes of critics of the American military:

* the military draws mostly or disproportionately from the poorest Americans

* the military draws disproportionately from minority populations

* the military draws disproportionately from the poorly educated

All wrong.

The median household income for recruits is slightly higher than the national median.

The top income quintile provides 22.85% of the recruits; the bottom quintile provides 13.66%. (Suggestion to the critics: switch your argument to an inquiry into why the poorest Americans feel so alienated from their country, they won’t even fight for it.)

Racial statistics are a little harder to pithily summarize, but basically the military is ethnically representative of the country. Blacks, who were over-represented by about 17% a few years ago, are now under-represented by about 4%. Pacific Islanders are the most over-represented group, with a whopping 649% over-representation. Asians are the most under-represented, at 69% of proportionality. Those groups are relatively tiny; the Big Three are all close to 1.0.

Educationally, the military considers a “high quality” recruit to be a high school graduate who scored above the 50% on the Armed Forces standardized test. The proportion of high-quality recruits has gone from 57 percent in 2001 to 64% last year (down slightly from 67% in 2004). Category IV recruits (essentially the “let’s give them a chance but not have high hopes” cohort) are 4.4% of total recruits.

The one piece of conventional wisdom that’s accurate: more recruits come from the South.

(H/T: NRO)


  1. You declare a carefully researched paper published in a highly respected journal meaningless because of errors that no one in the field of epidemiology consider errors and you believe a Heritage Foundation report implicitly? I guess your skepticism of data comes and goes with the political implications of the results.

    Was the HF report published in ANY peer reviewed journal? Did you notice that they claim that the lowest socioeconomic quintile is underrepresented based on zip code data, not information about the financial situation of each recruit? They also give no confidence intervals on any of their data. How much of what they claim as trend is just random variation? Impossible to tell, since they didn’t give CIs.

    Oops, got to go. More later.

    Comment by Dianne — October 31, 2006 @ 6:55 pm | Reply

  2. (Suggestion to the critics: switch your argument to an inquiry into why the poorest Americans feel so alienated from their country, they won’t even fight for it.)

    It’s more likely that the poorest Americans are not accepted into the military because they do not score well enough on the aptitude tests.

    The Armed Forces Standardized Test is essentially a proxy for an IQ test. It doesn’t let people from the bottom quintile (Category V) of the country join up, and limits the number of people from the second lowest quintile (Category IV).

    As income correlates with IQ, it is likely that the bottom quintile economically are disproportionately likely to be in the bottom two quintiles and so not to be accepted for service.

    This is also one reason why blacks are not disproportionately over-represented in the service; black Americans tend to score lower on IQ tests and are therefore less likely to be accepted into the Army.

    Comment by Glaivester — October 31, 2006 @ 6:59 pm | Reply

  3. Re-reading the third paragraph, what I should have written was:

    The Armed Forces Standardized Test is essentially a proxy for an IQ test. The military doesn’t let people from the bottom quintile (Category V) of the country join up, and limits the number of people from the second lowest quintile (Category IV).

    Comment by Glaivester — October 31, 2006 @ 7:01 pm | Reply

  4. Dianne, the HF report isn’t a sample survey. It doesn’t have confidence intervals.

    Comment by Robert — October 31, 2006 @ 7:33 pm | Reply

  5. Further thoughts on the HF report:
    1. To elaborate on their method of establishing socioeconomic status: They use the recruit’s home zip code to establish their SES, based on average income in that zip code. But zip codes can contain multiple neighborhoods and extraordinary variability in income. For example, in zip code 10037, which was specifically mentioned in the report, one can find both public housing projects with residents with median incomes near zero and million dollar condos. I also note that they use the income data from 1999 for that zip code. The neighborhood has changed signficantly since 1999 and using income data from that era is extremely misleading. In general, the use of the 1999 data for determining income levels in various zip codes is a very poor choice, as many neighborhoods have changed drastically in the past 7 years. They do mention that the complete zip codes for all recruits were not available and that these were excluded from analysis. So, no, it was not a complete analysis of every recruit and not an unbiased look at those recruits who are examined.

    2. They compare two data sets and conclude that they are different without giving p-value or CI to define whether the differences are random or signficiant. For example, they state that blacks are less overrepresented now than in previous years. If their data can be trusted, this is true, but whether the difference is significant and representing a trend is not explored and can not be without proper statistical analysis, which they fail to include.

    3. Some zip codes were excluded because they were listed as 3 digit codes, indicating very sparsely populated regions, some of which had median incomes listed as zero. In other words, they made the exact same exclusion that you accused the Roberts group of making. Why doesn’t it bother you in this context?

    4. Data from Reserves or National Guard units were not included. This may or may not affect results.

    5. No peer-reviewed literature was cited in this article, as would be expected for a scholarly work. Again, this doesn’t necessarily affect the quality of the data or the conclusions but it does indicate a lack of connection to the academic world and therefore lack of feedback on their work that would normally come with such connections.

    6. The HF is not an unbiased source. Their statements on conditions in the military should be treated with the same skepticism as one might treat the results of a drug company trial of a drug that company makes. I’d say the same about a report published by the “Michael Moore foundation” or similar group.

    7. A significant percentage of recruits declined to answer the question of whether they were of Hispanic origin or not. Again, they need confidence intervals to estimate the likely true range of Hispanic ethnicity in the military. A disproportionately large percentage of either Hispanics or non-Hispanics might have declined to answer. If so, this effect might significantly change their data.

    8. This quote from the report: “The 100 three-digit[11] ZCTAs with the highest proportion of blacks (in any combination of other races) according to Census 2000 contained 14.63 percent of the adult population. The recruits from these areas represent 14.09 percent of the 2003 cohort, 14.14 percent of the 2004 cohort, and 13.37 percent of the 2005 cohort. This indicates that these areas are not being overtly targeted to enlist large numbers of black recruits” makes an erroneous conclusion from the data*. The finding that fewer recruits are being obtained from the zip codes with the highest proportion of blacks does not say anything about whether or not these areas are being overtly targeted for recruitment: It may simply be that the targeting is ineffective. Data on the level of effort devoted to these areas (for example, data on how much money is spent in recruiting efforts in these zip codes versus zip codes with predominantly white residents) is needed before one can come to any conclusion about how much recruiting is going on in these areas.
    *I’ll leave aside the problem of including conclusions in the results section of a paper for now, but it is extremely poor form.

    Comment by Dianne — October 31, 2006 @ 8:32 pm | Reply

  6. The enlisted ranks of the military have consistently been drawn from the “middle 50%” of high school graduates, predominantly male graduates, disproportionately Southern. Very few enlisted soldiers ever go on to become officers.

    There are about 3,152,000 high school graduates each year.

    About 85% of all active duty military troops are men. This Navy and Army are both close to this average, in the Marines 94% are men, in the Air Force about 80% are men.

    About 103,000 men join the enlisted ranks of the military each year (based on double the number of E-1 ranked active duty military people, typically a 6 months stint time 85%). There are a little over 1,000,000 male high school graduates who aren’t in the top third of their graduating class each year. The military successfully recruits about 10% of this demographic. It gets a much larger percentage of this demographic in the South (which has relatively low numbers of Asians and Hispanics, and relatively high numbers of blacks, compared to the national averages), around 20%, and a lower percentage elsewhere. They are also disproportionately rural.

    There are about 19,000 women who join the enlisted ranks each year, there is good reason, given the relatively recent history of women in military service and the much smaller percentage of female high school graduates who enlist, about 1.2%, to believe that they may not be typical demographically of male enlistees.

    Declining African-American recruitment during the Iraq War has been significantly influenced by opposition to the war and active community efforts to discourage enlistment. Historically, African-Americans have been overrepresented in support specialties relative to their total numbers in the force, and have done so to further post-service careers, while whites have been overrepresentated relative to their total numbers in the force, in combat specialties.

    These trends are consistent with the fact that recruiting pressure has been greatest during the Iraq War not in combat specialties, but in support specialties. People who signed up to shoot rifles aren’t discouraged by knowing that they may actually end up doing so. People who signed up to learn to be electricians and mechanics are more likely to be discouraged by the prospect of combat encounters in a war with no front lines.

    Very low quotas on high school dropouts and poor performers on standardized tests were instituted after Vietnam, although these limits are being relaxed somewhat during the current war. These very low performers are usually not even allowed to choose military occupational specialties such as “Infantry” instead being shunted into becoming cooks and truck drivers. This keeps out about a quarter of people in the pertinent age group.

    Those in the top quarter academically overwhelmingly choose college over military service. Even those who do choose military service, generally do so by becoming officers, mostly via ROTC, which helps pay their way through college. About 17,655 newly minted commissioned officers (grade O-1) join the military each year (two-thirds of the total number of people in that rank which the average person spends 18 months in). About 1.4 million people graduate from college with a four year degree each year. So, about 1.2% of college graduates have either done ROTC or attended a service academy, and among men this is close to 3% (men are less likely than women to graduate from college), while about 0.3% of female college gradates go onto military service as commissioned officers. This also has a strong regional bias.

    I agree that the socio-economic status and median income methodology used by the Heritage Foundation, while suggestive for further study, is ultimately worthless. It’s suggestive findings, however, are consistent with the general conclusion that the military recruits largely from the middle 50% of the population. Its figure indicate “underrep­resentation for both the high and low ends of the income distribution.”

    Contrary to the Heritage Foundation’s assertion, none of this undermines the claim that this is a war being fought by the poor for the economic benefit of the rich (although it doesn’t prove that either). The rich, comprised really of an “upper middle class” of college educated managers and professionals, and a monied class made up of wealthy individuals (people who make more money from investments than from labor) are both drastically underrepresented in the military, and have received the lion’s share of the rewards of the economic growth seen in the U.S. since the 1970s. The rest make up a mix of the poor, the working class, and the true middle class. While the poor, by and large, hasn’t been fighting this war, the working class and true middle class have been fighting it.

    Comment by ohwilleke — November 1, 2006 @ 6:57 pm | Reply

  7. I adjusted the comment above, with some editing, at my blog, Wash Park Prophet, which is in the blogroll.

    Comment by ohwilleke — November 1, 2006 @ 7:48 pm | Reply

  8. The military attracts all who are willing.

    Comment by Jack Napiare — June 17, 2011 @ 9:16 am | Reply

RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Blog at

%d bloggers like this: