Monday, May 28, 2012

Polling Vets

Remember this Reuters headline from just two weeks ago?

"Obama Leading Mitt Romney Among Veterans"

According to the article, a poll commissioned by the wire service gave Mr. Obama a seven-point lead among voters who have served in the armed forces, "higher than his margin in the general population."  Reuters even found a number of veterans in South Carolina (a red state with a large military population) who planned to pull the lever for the president in November.  Here's a sample pull-quote from the puff piece:

Terry Seawright, a Navy reservist who drives a Fedex truck, voted for Obama in 2008 and plans to do so again in 2012. "I like the coolness and calmness of him," said Seawright, 46. "I like the way he handled Egypt and Libya. He said, ‘No troops on the ground.'"

Memo to the editorial team at Reuters: you might want to get your polling partners at Ipsos (which conducted the survey) to re-tabulate the results, or retract the original article.  Turns out that Reuters may be a bit off in gauging support among veterans for Mr. Obama's re-election.  

On this Memorial Day, Gallup is out with a new poll that shows Mitt Romney is an overwhelming favorite among voters who describe themselves as veterans.  In fact, Mr. Romney's lead is so lopsided, it clearly refutes the veracity of the earlier Reuters survey.  From Gallup's analysis:
U.S. veterans, about 13% of the adult population and consisting mostly of older men, support Mitt Romney over Barack Obama for president by 58% to 34%, while nonveterans give Obama a four-percentage-point edge.

These data, from an analysis of Gallup Daily tracking interviews conducted April 11-May 24, show that 24% of all adult men are veterans, compared with 2% of adult women.
Obama and Romney are tied overall at 46% apiece among all registered voters in this sample. Men give Romney an eight-point edge, while women opt for Obama over Romney by seven points. It turns out that the male skew for Romney is driven almost entirely by veterans. Romney leads by one point among nonveteran men, contrasted with the 28-point edge Romney receives among male veterans.
Put another way, if the previous Reuters poll had any degree of accuracy, then there has been a 31-point swing among veteran voters in recent weeks.  Of course, that scenario is highly unlikely, at best.  Such swings among a particular demographic group are virtually unheard of; in fact, Gallup did some of its polling during the same time that Ipsos was coming up with very different results--among the same voting group.  The laws of probability suggest that both polls should reflect similar results, not the wide differential between the Gallup survey and the work done by Ipsos.  
So, what's the real story?  In political polling (and other forms of opinion survey), there are the occasional outliers, surveys that simply contradict previous findings on the same subject.  But once again, it's hard to envision an outlier--in this case, the Ipsos poll--that is so far off the mark.  
It's also worth noting that the internals for the Reuters survey were never published (or if they were, we couldn't find them).  By comparison, Gallup explains its sampling very clearly.  Their results are based on interviews with more than 3,000 individuals who are both veterans and likely voters, individuals who generally turn out on election day.  
Still, there is a cautionary tale in the Gallup findings.  When you remove veterans from the mix, Romney's advantage among male voters essentially vanishes.  Additionally, the percentage of veterans is the highest among the oldest age groups in the survey.  As the size of the veterans block continues to shrink, the GOP will find it harder to maintain its long-time lead among male voters.  
But in this election cycle, veterans will be a crucial segment of the electorate.  If you don't believe that, consider this recent development: last week, the Obama campaign began targeting military voters in Virginia, a key battleground state that the President carried in 2008.  But not surprisingly, most media outlets didn't understand the reason for the sudden push.  The Wall Street Journal claimed that the Obama campaign sensed an opportunity, citing a recent poll that showed Mitt Romney with only an eight-point lead among veterans.    
Clearly, the WSJ and its broadcast partner, NBC News, might want to conduct another survey of veterans in the coming weeks.  While they identified a GOP preference among that group, there is a sizable difference between their results, and those of the Gallup organization.  Indeed, that latter survey explains the real reason  for Mr. Obama's campaign targeting military voters and veterans.  In a nation with more than 20 million veterans, no incumbent president can given his opponent a 24-point lead, even among a group that represents less than 10% of the American population.  Five months ahead of the November election, the President is trying to make up lost ground, among a constituency where he's supposed to be "competitive."                          



No comments: