09/25/2012 04:54 pm ET Updated Nov 25, 2012

Vets Retreating From Obama? Not So Fast

It's a provocative headline from Politico: "Veterans Retreating from President Obama."

Provocative, but not exactly true.

First, in the spirit of disclosure, I talked with the reporter who wrote the piece for 45 minutes, and walked him through the many reasons that veterans (especially younger veterans) are supporting President Obama, and how their voice can sway key voters that progressives usually have trouble reaching. And believe me, from expanding funding for veterans care, to constructing new vets centers under the stimulus, to passing legislation to get veterans hired, to doing more to help vets with PTSD and traumatic brain injuries, to getting us out of Iraq, to killing Osama bin Laden, there's a lot that veterans like about the President.

I also explained that where older veterans tend to be white males and more conservative, the newer generation of veterans contains more Latinos, African Americans, and women -- demographics that heavily tilts towards President Obama. That likely means that we're in the midst of a real paradigm shift, as far as the veterans' community goes, and the political leanings of its members. I do not know why I wasn't quoted in the story, nor why none of the facts or stats or poll numbers I passed on were used.

Now on to the facts.

The writer says, "Back in May, Obama had the lead among Afghanistan and Iraq veterans. But a Reuters/Ipsos poll from September says that's evaporated, with Romney now up 48 percent to 34 percent."

What doesn't he say? That a Zogby poll from Sept 21-22 shows that President Obama leads Mitt Romney among veterans and military families, 54%-40%, and that lead expands to 56%-32%, when you include third party candidates, among all voters.

That May Reuters-Ipsos poll referenced above found a 44-37 lead for the President. Considering that nearly every single poll has the President expanding his lead since then, it'd be reasonable to expect that the President's May 44-37 lead among veterans has expanded with the rest of the country, or at least remained about the same. Yet Reuters now says Mitt Romney has had an amazing 11 point bounce and the President has crashed by 10 points in a few months. Is it more likely that has happened, or veterans are more or less trending more towards the President with the rest of the polls, which would put him in line for that 54-40 lead among vets and military families that Zogby shows today?

Of course, this isn't the first time that there have been different poll findings regarding veterans. In May, when Reuters-Ipsos found that 44-37 lead among veterans for the President, here was Politico's headline:

"Mitt Romney has huge lead with vets, poll says"

Yep, that's right. Politico ran with a Gallup poll, which showed Mitt Romney with a lead among veterans.

Polling today also shows President Obama is pummeling Mitt Romney on questions of who is a stronger leader and better on foreign policy and terrorism issues. In some polling among all voters, the gap between the President and Mitt Romney is as wide as 15 percent. But I'll do something that I rarely ever do. Quote a FOX News poll. It found:

  • Who is better on Foreign policy: Obama 54 / Romney 39
  • Who can better protect the US from terrorist attacks: Obama 49 / Romney 41
  • Who would be a stronger negotiator with foreign leaders: Obama 49 / Romney 43

These are issues that military troops, veterans, and their families care deeply about. While those numbers don't break down veterans, I find it extremely hard to believe that the veterans community is so wildly out of line with the American public, at large, that they're flooding over to Mitt Romney, as the Politico article would have you believe.

Here's my point. The "polling" of veterans is incredibly inconsistent and unreliable. It's impossible to use any existing polls to show definitive veteran support for President Obama or Mitt Romney. The best we can do is guess where President Obama started with veterans, and presume veterans' opinions don't move in complete and utter contrast to the rest of the population. But guessing isn't good enough to write a story on.

Worse, when you start to pull veterans out of a larger pool of polling subjects, you're going to get wildly varying results. Because a small fraction of the population has served in war (and less than 1 percent in Iraq and Afghanistan), your poll sample will likely reflect that. Simply put, you're not polling strictly veterans as the full respondent sample, you're polling Americans and then yanking the veterans out, for a very tiny and unrepresentative sampling with incredibly unreliable results. And, if you're pulling just Iraq and Afghanistan veterans from that, the sample size will be even smaller and even more unreliable.

Further, who knows, of this tiny sample, where the veterans are coming from or who they are? Vets, largely, are like the general population. If you have a Christian Conservative veteran from Alabama, he or she will probably be for Romney. But a union worker in Pittsburgh? Probably for the President. Even within a single state, you could just happen to have a small handful more veterans in conservative areas of the state, which swings the veterans' findings one way or the other.

If your sample size is small, and slightly tilts towards vets from conservative areas, then your results will swing wildly in that direction. On the flip side, if you poll a few more progressive veterans in a small sample, the results will swing in that direction.

That's why, in the same month, one poll can find huge support for the President among veterans, and another find huge support for Mitt Romney.

So, Politico's story just isn't true. It's coming up with a thesis, and jamming a bunch of unreliable numbers in there to back it up, while completely ignoring contradicting numbers. I would be extremely interested to see polling done only of veterans and military families, with a responsible sample size. That would lead to a reliable thesis and more responsible reporting.