Tue April 17, 2012
The Obama-Romney Poll-A-Palooza: What's It Mean?
Originally published on Wed April 18, 2012 6:25 am
President Obama is leading presumed GOP nominee Mitt Romney big in recent national polls.
No, wait. Polls show he's trailing Romney by a couple of percentage points.
Oh — this just in: Obama is actually leading Romney, but the race is tightening.
It's a general election poll-a-palooza out there, people.
But what do all the numbers mean?
"I have friends who support Obama, and friends who support Mitt Romney," says Scott Keeter, survey research director at Pew Research Center. "I tell them not to get too excited or too depressed at this point."
His advice: Wait until September to let poll-based enthusiasm, or depression, take hold.
He'll get no argument from Mark Blumenthal, the former editor of pollster.com and now senior polling editor at the Huffington Post.
The current national polls, Blumenthal says, "do give us a general sense of where the race stands, and it looks structurally like it will be a very close race."
At this point, however, the national surveys are no more predictive of what will happen in November than a coin toss, he says.
Well, maybe not quite a coin toss — but close, says Chris Wlezien, a Temple University political scientist and author, with Bob Erikson at Columbia University, of The Timeline of Presidential Elections: How Campaigns Do (And Do Not) Matter.
"Right now, you have about half a story, and Obama is a slight favorite," Wlezien says. "He has a probability of winning of maybe 2 out of 3, or higher."
If the election were run now 100 times, Obama would win 67 to 70 times, he says. That's "hardly close to a sure thing," but not a coin toss.
The authors, based on their analysis of data going back to 1952, suggest that historically polls won't be highly predictive of the election outcome until late October. But this year's campaign feels further along and more polarized than some in the past, Wlezien says. So consider the past, he advises, a "conservative indicator."
Still, it's 200-plus days until voters mark their ballots, and the current predictive accuracy of polling is not all that high.
That doesn't suggest, however, that politicos, campaigns, news junkies and those of us in the media won't continue to obsess over the river of numbers.
To that end, let's take a look at the mixed messages various polls have been sending us over the past several days:
1. Obama has a comfortable early lead nationally.
A CNN poll has Obama up by 9 percentage points over Romney. An ABC/Washington Post survey gives the president an 8-point lead, while also suggesting that Romney is on pace to become the least popular presidential nominee since 1984.
2. Obama has a slim lead nationally.
The president holds a 4-point lead in an Ipsos/Reuters poll and is up by 3 points in a YouGov/Polimetrix online poll.
3. Romney, consolidating Republican support, has inched ahead nationally.
The first day of Gallup's daily phone-call general-election tracking poll gives the Republican a 47-45 percent edge over the incumbent president. An automated poll by Rasmussen Reports has Romney up 47-44 percent.
4. Obama is leading, but the race is tightening.
A Pew poll released Tuesday has Obama's 12-point edge last month shrinking to 4 points, 49-45 percent. The poll found that with Rick Santorum out of the Republican primary race, nearly two-thirds of party members say the GOP will unite solidly behind Romney, up from 57 percent who held that belief in February.
"I don't know what value there is to watching this day to day," Blumenthal says. "Month to month? Maybe."
"It's always interesting to look back over the campaign, but usually presidential campaigns don't move that quickly except around the conventions or major events."
That being said, Blumenthal adds a caveat: "But as a data geek, I love having this data."
Nate Silver, whose FiveThirtyEight blog for The New York Times routinely digs into politics and polls, on Monday issued tips for reading polls at this point in the campaign season. Topping his list is "be patient," pointing to early "statistical noise" that doesn't mean much. Coming in second was "take the poll average," meaning look for trends at sites that average national poll results.
The current polling may not tell us who is going to win, Keeter says, but it does tell us something.
"The electorate is pretty evenly divided," he says. "That's not news, but it suggests that both candidates have a good bit of upside potential, in terms of persuadable voters."
There are disappointed Obama supporters from 2008 who may be in the persuadable category, and also those who don't yet know a whole lot yet about Romney.
One issue with the polls being done right now is that they are typically surveying registered voters instead of the more predictive "likely voters," Keeter says, simply because it's too early to reliably identify those likely to vote.
Variances in results of polls taken at the same time, Wlezien says, are more a reflection of pollsters' methods — including how they weight answers — and sampling errors.
Keeter suggests that observers keep an eye on the national polls because they reflect the larger media environment and the state of commentary on the race. As the days tick down and more polling is done in battleground states where the race will be decided, look for those state-based surveys to reflect not only a broad zeitgeist, Keeter says, but also how the campaign is being waged via candidate visits, on the air, by micro-targeting and via social media.
And the best strategy for a poll-deluged civilian? Check back in September.