Skip to main contentSkip to navigationSkip to navigation
voting
A Pew Research survey shows that President Obama has a 6pt net favorable rating among adults. Photograph: Rui Vieira/PA Wire/Press Association Images
A Pew Research survey shows that President Obama has a 6pt net favorable rating among adults. Photograph: Rui Vieira/PA Wire/Press Association Images

Could recent political polls be painting a picture of what's to come for Obama?

This article is more than 10 years old
The approval among registered voters is far lower than among adults – likely showing how the 2014 midterms will play out

The mainstream media panic over the CNN/ORC poll showing President Obama's net approval dropping 17pt to -9pt is over. Pew Research came out with a survey giving Obama a 6pt net favorable rating among adults – only down 2pt since its prior survey. The truth is that the drop is likely in between the two polls, as I noted on Wednesday. I, however, want to focus on an overlooked split from the Pew survey: Obama's net approval rating among all adults at +6 versus registered voters at -2.

It's not difficult to see how these two data points present different pictures. One paints a president who enjoys relatively broad support among the public, while the other one says he's actually not popular. The question is whether one point is "righter" than the other one and whether it means anything. It's part of a long standing debate that Mark Blumenthal covered well in 2010.

Most national surveys start with an adult sample. The reason is quite simple: it's very easy to know what the adult population looks like thanks to the census. So when you randomly call up all households, as most public polls do, you have a known parameter.

This leaves less room for error. There is no set registered voter parameter that everyone can agree upon*.

Polling among all adults has the additional advantage of allowing us to know how all Americans feel. This may sound corny to some, yet it's important to point out that the government represents all Americans, not just those who vote. If a pollster's job is to allow politicians and us to know what's important to the people, then polling adults is the way to go.

The problem, of course, is that elections are not fought among all adults. Only registered voters can cast a ballot, and only some of them do. The politicians that I know are most interested in getting reelected. Historically that might not have been a problem for displaying results among adults, but adults who are not registered these days tend to be more favorably inclined toward the Democratic party.

You saw this split in action last year. Because young voters and Hispanics were the most likely to not be registered, President Obama enjoyed a 10pt greater advantage among all adults than the smaller pool of registered voters. That's quite a wide split and very similar to the 8pt difference in Obama's approval rating among the two groups right now per Pew.

Other pollsters don't have nearly as wide of a contrast between the two groups, but it's apparent**. YouGov's latest survey shows a 3pt difference, while the Washington Post last month had it at 4pt. The reason could be that Pew does a better job than most of reaching farthest corners of the electorate who are the least likely to respond to a survey or vote.

Still, 3 to 4pt would be enough of a difference for the following adult surveys – Bloomberg, CBS/New York Times, Gallup, National Journal and NBC/Wall Street Journal – to turn from a positive net approval to an even or negative net approval among a registered voter subsample. You can also probably subtract a few more points if you project a likely electorate for a midterm electorate, which is even whiter and older. These are important distinctions for electoral purposes in the upcoming 2014 elections.

More so than any time in recent history, the president's party's fate in midterm elections seems to be linked to president's own standing. In 2006, President Bush's approval was only 43% in the exit polls. His party got 44% of the House vote. What about in 2010?

In an average of adult polls with undecideds allocated in the prior two weeks before the 2010 election, 49.5% of Americans approved of Obama's job. That's still a negative net approval rating of -1pt, yet a 1pt deficit in the national House race would have been small enough for the Democrats to hold onto the House given incumbency advantage.

So then how did Republicans win the House? The electorate that voted wasn't all adults.

In 2010, President Obama's approval was 44% in the exit polls. His party got 45% of the House vote. Indeed, in every instance from CNN to Gallup to Pew to the Washington Post to YouGov, the gap in approval among voters and all adults in pre-election polls was mirrored in the ballot test between the Democratic and Republican candidate.

Thus, we're back where we started. Neither approval of registered voters or adults is perfect. Adults are easier to poll, and we know how the entire country feels. Of course, registered voters are the ones who can vote, and polls of them have tended to be accurate even if individuals lie about their registration. It also seems a registered voter sample of Obama's approval rating, which paints a worse picture for the president than approval of all adults, probably gives us a better idea on how the 2014 midterms will play out.

*One way to know that a respondent is a registered voter is to call a list of known registered voters like most campaign pollsters and Public Policy Polling. The disadvantage is that these lists tend to not to cover all voters. The advantages of lists versus randomly calling households is a hotly debated topic.

**It should also be pointed out that it's not rare to see in any one poll that adults and registered voters have the same approval for Obama, though, it is unusual and likely because of sampling error.

Most viewed

Most viewed