Obama up by 13? Explaining the Bloomberg poll
By Aaron Blake,
A Bloomberg poll released Wednesday shows President Obama leading Mitt Romney by 13 points.
If you read this blog with any regularity, this will shock you. And it should; after all, it’s the only national poll since March to show either candidate leading by double digits, and most other recent polling shows Romney neck and neck with the president.
So what gives?
It could simply be outlier, but a review of the Bloomberg poll, which was conducted by Des Moines-based Selzer and Co., shows the poll could skew Democratic for two reasons:
1) The sample appears to be far more highly educated than the overall population, with nearly half of voters having completed college.
2) The questions that come before the Obama vs. Romney question may create a sort of “priming of the pump” effect that favors the incumbent.
First, education. Most polls adjust their samples for education levels, so that more than 40 percent of respondents have a high school diploma or less. But the Bloomberg poll adjusts for only age and race, so its sample features only 27 percent of respondents with no college education.
What this means is that the polling sample is much more educated than others, which generally means it’s more liberal.
Still, the poll contains very normal ratios of Democrats to Republicans. So there may be something else at work, which is what we’re going to call “Priming of the pump.”
While this poll, like others, asks the head-to-head Obama vs. Romney question, it also asks a number of questions before the horserace. Among them:
* Do you have a favorable or unfavorable opinion of Obama, Romney, former presidents Bill Clinton and George W. Bush, and Donald Trump?
* How would you rate Obama’s job performance in several different areas?
* Which party would you vote for in a generic Republican vs. Democrat congressional race?
The problem with polling is that some people don’t have strong opinions, and their responses can be nudged — ever so slightly — by the preceding questions. And while the questions above don’t have any inherent bias, you could definitely make the argument that they might incline people to support the incumbent in a head-to-head matchup.
Obama and Clinton, after all, are popular (and very popular, in Clinton’s case, with a 69 percent favorable rating), while Romney, Bush and especially Trump are not so personally popular. Already, the poll is reminding voters of popular Democrats and unpopular (or less-popular) Republicans.
Right after that, the poll asks about Obama’s approval rating, and he scores higher than in any other recent poll: 53 percent. He also gets pretty decent approval numbers on a number of other measures too, including “creating jobs.”
And then, the generic congressional ballot, which has favored Democrats for a while now and this time favored them 45 percent to 38 percent. Republicans in Congress are unpopular right now, which is why Obama is running against them.
So in other words, respondents went into the Obama vs. Romney question with a few things in their minds that included George W. Bush, Donald Trump and Republicans in Congress.
Obama can only hope that voters will walk into the polls on Nov. 6 thinking about the same three things.
We reached out pollster Ann Selzer, who conducted the poll (and who has a sterling track record of predicting the outcome of the Iowa caucuses in her capacity as the pollster of choice for the Des Moines Register) to get her take on why the Bloomberg poll appears to be such an outlier.
Here’s her take on our take:
On education: Obama leads with all four education subgroups we capture. So, weighting for education, if we were so inclined, is not likely to change the result by much.
On question order: We ask the horserace after asking about right/wrong track (in the toilet) and Obama’s job approval on economy, healthcare, relations with China (all not so good), so I think it’s hard to argue that the order produced a halo effect. We asked Obama and Romney favorability (in a rotated list) along with the other names and Obama was much higher than Romney’s. So, if you think of that as a clean read on the two candidates, it’s hard to argue this affected the horse race.
We’re happy to entertain possible explanations, but these do not seem viable.
What do you make of it all?