The Third-Party Factor
Since this week seems to be "Numbers Week" here at CW.com, I'd like to take a very close look at two recent polls, because they show something astonishing which everyone appears to be either ignoring or missing: When pollsters ask about four candidates instead of two, Obama's lead goes up. When Bob Barr and Ralph Nader are added into the mix, they seem to both draw votes mostly away from McCain, widening the gap between McCain and Obama.
This seems strange, since the conventional wisdom would have it that Barr will draw voters away from McCain (he's already being spoken of as a "spoiler" for McCain in one particular state -- Georgia -- where Barr hails from); and that Nader will draw voters from Obama. But, as these two polls show, that doesn't seem to be the case.
Both of these polls were taken about a week and a half ago. Zogby/Reuters polled people from July 9-13, and ABC/Washington Post polled people from July 10-13. Their results were strikingly similar.
Taking the Zobgy/Reuters poll first, we see that Barack Obama is leading John McCain 47 to 40. Furthermore, 10 percent were "not sure" and three percent said "someone else." But if you read a little deeper, the poll also restated the question as a four-way contest between Obama, McCain, Barr, and Nader. From the Zogby site: "Adding minor presidential candidates, including Independent Ralph Nader and Libertarian Bob Barr to the equation, both minor candidates garner three percent support and Obama's lead over McCain jumps to 10 points, 46% to 36%."
So, when third-party candidates are actually named, Obama loses one point (47-46) but McCain loses four points (40-36), even though Barr only gets three. Undecided goes up to 12 from 10 too, mysteriously.
The ABC/Post poll mirrors this (see a quick overview of the data if you don't want to read the whole poll). Looking at Registered Voters, we find that when the matchup is presented as a McCain v. Obama match, Obama leads by an impressive eight points, 50-42, with seven percent falling into "other" in one way or another. But when you add in Nader and Barr, this climbs to an even more impressive ten point lead, at 49-39, with Barr pulling two percent and Nader five. Once again, Obama loses only a single point, while McCain loses three.
The ABC/Post poll is interesting for another reason as well, as it provides numbers for Eligible Voters (listed as "Adult" or "General Population"), Registered Voters, and Likely Voters. Likely Voters were a lot closer than the other two samplings, putting the race very close, at Obama 49 and McCain 46. Likely Voters were also not asked (apparently) about the third-party choices. Among Eligible Voters, the numbers are even more astonishing, going from Obama 51, McCain 39 (with just the two of them named) to Obama 49, McCain 36, Barr 2, Nader 5. Obama loses two points, while McCain still loses three, but Obama's lead actually grows from twelve points to an enormous thirteen points.
But the inescapable conclusion is, however you add it up, Barr and Nader on the ticket is drawing more support from McCain than Obama, and widening the gap between them. This even shows up in a Fox poll (which I, as a general rule, usually ignore) taken July 22-23. They show Obama only up over McCain by a single point (41-40) but when Barr and Nader are added in, Obama widens his lead two points (40-37). Once again, Obama loses one, McCain loses three. Interestingly enough, in this poll, Nader only pulls two percent and Barr zero. Meaning that some McCain voters will vote Nader over Obama or McCain. Perhaps these are the disgruntled Hillary voters we've been hearing so much about?
Now, as with any look at polling, there are a lot of reasons why it's foolish to read too much into these snapshots. In the first place, there are all the inherent inaccuracies with any poll. Secondly, conventional wisdom in this particular case says that third-party candidates poll a lot higher than they usually reap in actual votes on Election Day. But the trend seems pretty clear to me: when all four candidates for president are mentioned by name to the public, McCain's support is soft and about three percent drift off to other candidates. Obama's support is a lot stronger, and he only bleeds off one percent.
While it's too early to draw anything conclusive from these numbers this early in the race, it is an interesting bit of data to ponder. And one to keep an eye on, as the election season progresses. Nader and Barr seem to be actually helping Obama out, by showing the weakness of John McCain's support. And remember, there are more than a few states where a one or two percent advantage may just decide the race, so it's not just an academic exercise.
[New! I've added to the site, over on the left side about halfway down, a collection of links to 2008 Election Tools, which I will doubtless be adding to before November. To keep an eye on this data, follow the "Pollster: Obama v. McCain" and the "Pollster: Obama v. McCain v. Barr v. Nader" links.]
-- Chris Weigant
"But the inescapable conclusion is, however you add it up, Barr and Nader on the ticket is drawing more support from McCain than Obama, and widening the gap between them."
I've heard a lot of people saying this, but the conclusion is not exactly accurate in its explanation. They aren't "drawing support from McCain" because he didn't really have it to begin with. It only appears that way because of a question that has no basis in reality. They aren't "adding in" the other candidates to the second question; they are actually "taking away" candidates in the first. While it seems like the same thing, it really isn't, because one is based in reality, the other is a hypothetical.
Asking "If the election were today; who would you vote for?" and "If the election was today, and you could only vote for Obama or McCain, who would you pick?" are two totally different questions. One is an accurate assessment of how things are now if the election was today, the other is an assessment of a hypothetical election. I feel the two person polls should never even be asked. They give nothing but a false representation of the vote.
Nader isn't pulling votes from McCain, because McCain never had them. Most Nader votes are anti-Obama votes. Democrats that are angry at Obama (PUMA's, etc) that won't vote for Obama. So when restricted to an Obama/McCain choice they are anti-Obama so "vote" for McCain. But the ballot isn't restricted, so McCain really doesn't have their support, it's just an appearance of support because of the faulty question.
Barr votes are mostly right wing anti-McCain votes but are more anti-Obama than anti-McCain. So again, when given the faulty question of just Obama/McCain they go with McCain. But again, the ballot isn't going to be that faulty two person match up, it will be all four in many states (and even other parties too). The multi-person polls are far more accurate picture of the support because they resemble the actual ballots.
The fault in the media's argument that McCain is losing support when adding in the other two candidates is that you can't look at it in a bubble like that. They are using a hypothetical match up to give him unrealistic support that he really doesn’t have, then trying to say that he’s losing that support by those candidates being there. It gives the impression that if those third party candidates lose support towards election day that it all goes towards McCain. That is the real fault... because they are suggesting that the Nader "anti-Obama vote", if it diminishes, is going to go with McCain. Some might, but some of those disaffected Dems may return to the flock and vote Obama... or some may just stay home.
The 2% increase in undecided is also "anti-Obama" votes. When given the choice between Obama/McCain they go against Obama, but when given 3 choices and Obama they become undecided. They are still anti-Obama, but aren't "for" anyone when given a choice. They are clearly not pro-McCain at this time.
You can’t say what is going to happen with these people, so trying to give that support to anyone is just a guess at best. But the media is trying to put that support in McCain’s corner with these poor polling questions to make it a tighter race.
So the most accurate situation you have is:
46% Obama
36% McCain
3% Barr
3% Nader
10% Truly Undecided
2% Undecided Anti-Obama
Compare those numbers of support to other polls that ask different questions on issues, and they line up a lot better than what the faulty two person polls display.