January 05, 2006
Military Times Survey: Update
A quick update on yesterday's post on the Military Times poll of its readers on active duty military service. I noted that the raw data was available to anyone with the time and inclination to import it into their statistical software. Reader MC took me up on my challenge and ran some cross-tabulations by rank on party, ideology and some other questions. Within the Military Times subscriber base, at least, the the differences by rank (comissioned officer vs. enlisted) are quite small.
My rendering of RC's cross-tabs follow below. I left out the result for warrant officers that RC included because they were only about 2% of the sample and thus too small for a reliable tabulation. Also, I do not have the exact sample sizes, but comissioned officers were reported as 43% of the sample of 1,215 respondents and enlisted personnel were 48% of the sample. [UPDATE: the sames sizes are n=581 for enlisted, n=536 for commissioned officers. Thanks to reader BW].
The differences? Enlisted personnel were slightly less likely than commissioned officers to describe themselves as conservative or Republican, but the difference appears to result from a greater number enlisted personnel who declined to answer the question. Their responses are nearly identical on the Bush job rating and the question of whether the US "should have gone to war with Iraq.
This issue is important because, as noted yesterday, the Military Times readership includes a far greater share of officers (45%) than the overall population of active military (15-20%). While we should be careful not to assume that enlisted Military Times subscribers represent all enlisted personnel, within this sample at least, the result would have been essentially the same had officers been only 15% rather than 45%.
Now MP makes no claims of expertise when it comes to the military, so he strongly recommends reading the helpful comments from readers on yesterday's post. In particular, "ex Navy in Sea" suggests that Military Times subscribers may include more college educated enlisted personnel than typically found among all enlisted. "Jimbo" also suggests that the results should be cross-tabbed by branch, as the Iraq War "impacts" most members of the Army and Marines.
RC - or anyone else -- Do you feel like running more crosstabs? The raw data can be found here.
ANOTHER UPDATE: RC and others continue the conversation in the comments below. Also, Gordon Towbridge from Military Times emailed with these comments:
Very interesting stuff -- I'm glad people have taken the data set and run
with it. Confirms that our decision to make the data available is a good
one. My thanks to you and your readers for the thoughtful discussion.
I should have mentioned in our conversations that I'd run quite a few
crosstabs and found, as in previous years, that there was surprisingly
little difference in opinions of officers and enlisted, and between the
ground forces (Army and Marines) and the rest of the military. One of the
common objections you see in other emails to us, blog writeups, etc., is
that we don't provide crosstabs by service, rank, or between those who've
been to Iraq and those who haven't, but there just wasn't much interesting
there this year -- the numbers don't move. On the officer/enlisted
question, I suspect this is probably a function of the fact that even our
enlisted troops tend to be older and higher in rank, which might make
their mindset closer to officers than we might expect. The similarity
between the ground and air/naval forces is more interesting, and I'm not
sure I can explain that one.
Your readers are right to point out that our sample is much more educated
that the military as a whole (though people would be surprised how many
enlisted troops have college degrees these days; I know two NCOs with
Ph.D's). "Ex-Navy in SEA" asks about reserve forces -- we do have
reservists and Guard troops who are mobilized to active duty in the
sample; it's our best guess that we simply don't have enough reservists
among our subscribers to do as good measuring attitudes in that
population. It'd be nice to get at though; the strains they're facing are
much different, in some ways more severe, than those in the active force.
Thanks to Towbridge and the Military Times publications for sharing the data that enables this discussion!
Related Entries - Polls in the News
With regards to Jimbo's comment, the crosstabs on Army/Marines versus other branches of service aren't too different. Same with attitudes between military personnel deployed to Iraq/Afghanistan and those not deployed. There's only about a 2-3% difference in any of those categories.
One difference that caught my eye - the response to Question 24 regarding overall Presidential approval differs by little over 7% between those in the Army or Marines versus another branch of service. While the approval rate is lower among those in the Army or Marines, the disapproval rate is about the same. The difference is translated into more responses of "No Opinion" and "Decline to Respond". I ran a Chi-Square test on the observed distribution of Army/Marines responses against expected value computed by the frequency of the responses in the Other Branch subset. The Chi-Square test was statistically significant.
Recognizing the limitations to the sample, is there is a cultural element to the Army or Marines that may have resulted in this difference?
If anyone wants the dataset for SPSS with variable labels entered - please let me know.
Posted by: MC | Jan 5, 2006 8:12:08 PM
Hmmm. I'm quite surprised that the numbers were more-or-less the same for officers and enlisted. My recollection is that we rarely agreed on much of anything. :-)
I may not have been clear in my comment in yesterday's post. The military has a very formal hierarchical structure, and there is a very definite class culture to the ranks. Generally speaking, nearly all officers are college educated. With few exceptions (like battlefield commissions), a college degree is one of the basic requirements for becoming an officer. I found one Dept of Defence report from 2001 stating 96% of officers have at least a 4-year degree. Likewise, almost all enlisted personnel have only a high-school (or equivalent) education. An enlisted person with a college degree is an extreme rarity. I was able to find one recent Air Force statistic indicating 3.1% of enlisted with a degree (note that is just one branch of the military, and generally the highest educated one as well; other branches would likely have a significantly lower percentage).
A side effect of this is that the average age is quite different as well. A vast majority of enlisted personnel join the military soon after high school, and are under the age of 20 when they start. Because they have to get a degree first, most officers are in their mid-20s when they start. The same 2001 DOD report stated the mean age for enlisted recruits is 19, while the mean age for officer recruits is 27.
So, because enlisted personnel are generally both younger and less educated than officers, it is not surprising that enlisteds decline to answer for general political views and/or party affiliation, or have no opinion. When looking back, at age 18 and freshly enlisted in the Navy, I had strong opinions about a few issues of the day, but had not yet developed a strong lasting political overview or party affiliation.
Posted by: ex-Navy in SEA | Jan 5, 2006 9:51:43 PM
ex-Navy in SEA:
In the poll sample, 102 of the 581 enlisted personnel reported a BA/BS, while 174 of 536 officers reported a BA/BS. About 65% of the officers reported holding a graduate degree.
The sample also shows the age/rank pattern you describe, with about 45% of the enlisted respondents under the age of 33, while the officers are more evenly distributed, including 41% over the age of 41.
Looking at the sample, 20% of the respondents in the Army or Marines are under the age of 28, while the respondents from the other branches tend to be older.
As you suggest, youth rather than a cultural element may be the culprit of the "No Opinion" response. A regression analysis may prove more conclusive; however, while this has been a fun diversion, I must return to my own research...
I have a difficult nut to crack and while I left it alone, to my surprise, it didn't solve itself. :)
Posted by: MC | Jan 5, 2006 10:46:49 PM
I have run some crosstabs by enlisted/officer and by military branch. The results echo the comments made above by MC - that respondents in the Marines and Army tend to younger and that 20% of enlisted respondents have a BA/BS degree or higher and that less than 3% of officers do not have at least a BA/BS degree.
I have also looked at several other survey questions by these two variables (enlisted/officer and military branch). If anyone would like the crosstabs (SPSS Output saved as a PDF file) or the SPSS dataset, please feel free to contact me.
Posted by: Brett | Jan 5, 2006 11:34:39 PM
Thanks for this...very interesting. As if the guys at Duke need a publicity agent, I would point out that Feaver and Holsti at Duke have done some basic research based on historical data that might be of interest to your readers.
Here are two links:
-an article by Ole Holsti:
-a major book on military attitudes co-edited by Peter Feaver (the book specifically compares military attitudes to a nationally representative survey):
Posted by: Ike | Jan 6, 2006 9:04:03 AM
I just looked at the available data, and was surprised to see that THREE YEARS of raw data is posted! I'm grabbing it all--and linking to this blog--for a project in one of my spring statistical methods courses. Thanks for the pointer.
Posted by: Mike Anderson | Jan 6, 2006 9:59:23 AM
Great to see a constructive dialog with the Military Times on this.
Posted by: Thomas Allen | Jan 6, 2006 2:47:57 PM
MC and Brett:
Your data that shows that roughly 20% of enlisted respondents have a BA/BS degree is interesting. This indicates that the enlisted respondents to the poll are likely older and much more educated than the average enlisted person (according to the Air Force's own numbers, 3.1% of overall enlisted members have a BA/BS degree, per http://www.afpc.randolph.af.mil/pubaffairs/release/2005/04/demographics.htm from April 2005; and Air Force personnel tend to be the highest educated of the different branches of service).
I agree with Gordon Towbridge that their enlisted respondents tend to be older and more senior than average, and that probably explains why there isn't as great a difference between the enlisted and officer numbers as I would have guessed.
This coincides with my own recollection. As a young, newly enlisted sailor right out of high-school, I remember when new issues of the Navy Times (part of the Military Times) came out, there would be stacks of them in the barracks lounge, in squadron common spaces, and various places on base. Many of us would read them, but I'm pretty sure very few of the younger enlisted guys actually subscribed. I would guess as we grew older, rose in the ranks, moved off base and started families, we would have less access and time to read the paper on base. Plus, as we raise in seniority, we would view the military more seriously (or more maturity?) over time, and would likely take more interest in the military papers.
Anecdotally, I would also guess there is a difference in attitudes during wartime. I served 8 years during peace time. When I was in, there was a pretty strong inter-service rivalry, particularly between Navy and Marines. But I think that was largely surface joking. I remember older guys in my squadron who had served during Vietnam said that most of that joking got set aside, and the services rallied together in times of combat. This may explain why there isn't much difference between the branches of service at this time. I would be curious to see if this type of inter-service similarity carries over in peacetime.
I read over my previous comments, which in hind-sight seem overly negative. I do find this survey quite interesting. And I think the Military Times has done the best job possible trying to gather their data. As you originally point out, active duty military is an extremely difficult group to survey accurately. Almost all junior enlisted personnel live on bases in barracks and have no outside phone (other than cell phones), usually for the first couple years of service. This makes them almost impossible to poll.
Ultimately, however, I still think that the poll skews to senior and older personnel, and probably misses a huge chunk of younger, junior enlisted people. Whether or not their opinions would be significantly different is anybody's guess.
Posted by: ex-Navy in SEA | Jan 6, 2006 7:57:21 PM
In addition to the respondent skew in college education noted by others, the sample frame of subscribers to Military Times publications is skewed toward the middle and upper pay grades for both officers and enlisted:
Rank N Survey % Total Service %
E1 1 .2% 3.4%
E2 1 .2% 6.6%
E3 16 2.8% 18.8%
E4 43 7.5% 22.5%
E5 104 18.1% 21.7%
E6 167 29.1% 15.0%
E7 151 26.3% 8.6%
E8 58 10.1% 2.4%
E9 33 5.7% 0.9%
Rank N Survey % Total Service %
O1 26 5.0% 11.6%
O2 38 7.2% 14.4%
O3 134 25.5% 33.5%
O4 116 22.1% 21.0%
O5 124 23.6% 13.6%
O6 84 16.0% 5.5%
O7 2 .4% 0.2%
O8 1 .2% 0.1%
O9 0 .0% 0.1%
Data for total service personnel by rank come from http://www.factmonster.com/ipka/A0004604.html and are based on totals across all services except the Coast Guard as of April 30, 2005. It is not clear whether these data are for active duty personnel only, or if they include active reserves, National Guard, and/or inactive reserves.
As these tables show, lower-ranking enlisted personnel (E1-E4, corresponding in the Army as private through specialist fourth class) are severely underrepresented in the poll, constituting about 11% of enlisted survey respondents but 51% of all enlisted service personnel. The lower tier of non-commissioned officers (E5-E7, corresponding in the Army as sergeant through sergeant first class), which includes most squad leaders through platoon sergeants, constitutes 74% of enlisted survey respondents but 45% of enlisted service personnel. The highest enlisted ranks (E8 and E9, in the Army sergeant first class and sergeant major) are also highly overrepresented, making up 16% of enlisted survey respondents but only 3% of enlisted service personnel.
Among officers, company-level commissioned personnel (O1-O3, in the Army second lieutenant through captain) make up just 38% of commissioned survey respondents but 59.5% of all officers in the four services. Battalion-level commissioned officers (O4 and O5, which are majors and lieutenant colonels in the Army) are 46% of commissioned survey respondents but 35% of total commissioned personnel. Field-grade commissioned officers (06 and up, corresponding in the Army to full colonel through the general ranks) are also somewhat overrepresented as about 17% of commissioned poll respondents but just 6% of commissioned personnel.
Thus it is true that the pay grades most likely to experience combat are underrepresented in this survey. However, my analysis of rank-related differences in support for the Iraq war suggests that there is little difference in support for the war by pay grade within the enlisted and commissioned ranks. Responses to the “Should the U.S. have gone to war in Iraq” question run 74% for lower-ranking enlisted (E1-E4), 68% for lower-tier noncoms (E5-E7) and 79% for upper-tier noncoms (E8 and E9); among officers, support runs 68% for company-level officers (O1-O3), 68% for battalion-level officers (O4-O5), and 64% for field-grade officers (O6 and up).
This may be a case where sampling bias doesn’t translate into response bias.
Posted by: Scott Althaus | Jan 10, 2006 11:43:09 AM
The comments to this entry are closed.