December 14, 2007
Manufacturing Dissent: An Odd Poll Raises QuestionsBy Ray Robison
A recent Los Angeles Times/ Bloomberg poll which has been carried in most major media outlets claims that President Bush has lost the support of military veterans and their families over the war in Iraq. The LA Times writes,
A similar article based on the poll is running at the website Military.com. It has an accompanying unscientific online poll which asks: "Has the war in Iraq been worth it?"
At the time of this writing the results are:
To be sure, self-selection by respondents yields unscientific results, Still, it's a little odd that a website that has a predominately military readership would have results that are completely opposite the LA Times/Bloomberg poll results for the question of whether or not the war was worth it when put to military members and families. And before the conspiracy theorists get going we should note that this website is not affiliated with the military and nobody is forced to answer these polls.
Which should raise some serious questions about the methodology of the LA Times/Bloomberg poll. Some of it is described in the reporting but I have requested more information from Los Angeles Times poll director, Susan Pinkus and I hope to hear back from her about how this poll was conducted. But until I can gain more detail, there is some information available about the poll in the reporting.
The LA Times report states:
This poll which is being billed as a rebuke to the president by military families includes only 10% of respondents who actually claim to have a family member who is serving or has served in Iraq or Afghanistan. Therefore, the number of respondents who had family involved in Iraq specifically is even less than 10%. So the people who are supporting family members in Iraq actually had very little to do with these conclusions as a whole.
But look again at the claims from Bloomberg and the LA Times. They claim the respondents were critical over Iraq, not Afghanistan. Let's be generous and guess that two-thirds or 100 of these respondents claimed to have a family member who served in Iraq. According to a recent USA Today article, over 1.5 million troops have served in Iraq. To try to take a poll with 100 or so respondents out of a pool of over 1.5 million is absurd.
There is no statistical validity and no way to assign any confidence to the conclusion that military family members of those who have fought in Iraq are turning against the President because of Iraq. The methodology statement notes that the margin of error for this subgroup is 8%. I would suggest it is quite a bit larger when you consider other factors.
How did they confirm that these respondents are actually military family members? What was the verification process, if any?
The polling company, Interviewing Service of America, Inc. defines "family members" as people who are considered part of the immediate family. But how many people answered that they were immediate family members but really were not? We all know that there are plenty of people in the general population angry with the President over the war. How do we know that uncle Joe, who hates President Bush, didn't get one of these calls and say "sure, I have a family member in Iraq" because some nephew he hasn't seen in ten years went to Iraq and he just wanted to bash Bush?
The poll also claims the respondents were weighted by census portions for "national region". If anything, the calls should have been weighted to the regions by rate of service in the military, such as the rural South, not by the general population. Weighing by overall census population means that military relatives in highly populated areas like New York City and LA, places where President Bush and the Iraq war are especially unpopular, are overrepresented compared to military families from the rural South. If this is going to be touted as a military poll it should have been conducted weighting for military families' distribution regionally.
And let's look a little closer at the respondents identified in the article. The methodology statement claims that respondents were chosen randomly. But I am skeptical. Much as the recent revelations that the CNN Republican debate was loaded with liberal activists called "plants" the pollsters here may have done some gardening of their own.
The LA Times quotes Mary Meneely of Arco, Minnesota who said
and then compared Iraq to Vietnam. Her son, who was an Air Force reservist wasn't even in Iraq, but Afghanistan.
Her husband, Tom Meneely is a liberal activist who writes anti-Bush tirades to the media. He wrote to TIME magazine:
And in the Star-Tribune:
Tom Meneely essentially accused Gen. Petraeus of betraying his country to serve President Bush. Sound familiar? This statement makes it quite likely that he is reciting from the MoveOn.org talking points that would surface a few days after his letter in that reprehensible ad in the New York Times.
The LA Times also cites
As an army sergeant, he would be required to have an army email account through the Army Knowledge Online portal. I have searched for "Datta" in the email address database. I found four people with the last name Datta and none matched the rank of SSG or held a rank even close to that. The most plausible match is an army Specialist which is several ranks and years junior to a staff sergeant.
It is odd that an extensive search failed to turn up any information on either Datta's son -- not even a news article about his "serious wounding" -- or the Meneely's son on the Internet. But the article doesn't actually name either son which makes it difficult to verify their stories. I have requested confirmation from the LA Times of the identity of these military members.
But at least the LA Times bothered to talk to respondents of the poll unlike the Bloomberg reporter who just went to anti-war groups and plastered quotes from them among the poll data to give the appearance they were involved in the polling in order to support what looks like a junk poll.
Everything about this looks fishy. I have yet to see one spouse quoted as a respondent in these articles. I have seen no web postings from any of the people discussed here to indicate that before this article, they were concerned about a family member in Iraq. The polling method does not match the conclusions.
If you conduct a study, you have to identify the questions you want to answer first and set the conditions to answer them. You can't skim data of the top without a real understanding of what went into it.
As they say "garbage in, garbage out".
Ray Robison is proprietor of Ray Robison: Pointing out the Obvious to the Oblivious and the co-author of the just-published book Both in One Trench.