Gallup vs. Rasmussen

« Factcheck.org: 'Blame Dodd' Attacks Ignore Facts | truthseeker77's Blog | Jon Taplin: An angry, mediocre, Krugman-hating blogger who does not back up his claims »

March 21, 2009, 10:08PM

As I compose this entry, President Obama has an approval rating of 64% in the Gallup tracking poll, in contrast to the 55% he receives in the Rasmussen daily tracking poll.

Even more striking is the difference in the disapproval figures: 26% in Gallup versus 44% in Rasmussen.

Is there a good scientific explanation based on the methodology or wording used by these pollstesr that can explain this discrepancy? Or is one of them fudging or inventing these stats in order to favor a certain group?

http://www.gallup.com/poll/113980/Gallup-Daily-Obama-Job-Approval.aspx

http://www.rasmussenreports.com/public_content/politics/obama_administration/obama_approval_index_history

8 Comments

| Leave a comment

Who funds each of them, directly or indirectly, the most?

Posted by Mike7Woodson
March 21, 2009 10:47 PM | Reply | Permalink

I was told that Rasmussen right wing prone.

Posted by Aunt Sam
March 21, 2009 11:35 PM | Reply | Permalink

Have we considered the precise wording of the questions? And do we know the nature of sampling?

It's normal for polls to vary some, especially on the vague stuff like this.

Posted by Tom Wright
March 22, 2009 1:05 AM | Reply | Permalink

Rasmussen is a decidedly right leaning poll. Gallup tends to run a bit more neutral, while Research 2000 runs to the left, hence the the DKos/Research 2000 poll as of 3/19 showing Obama with a 69% approval rating. My suggestion combine all three and average and you probably have something closer to the actual numbers or 63% (62.67%)

Posted by jsfox
March 22, 2009 10:57 AM | Reply | Permalink

Whatever it currently is, it will be going down and precipitously so with all the Obama bashing from the so-called progressives in the blogosphere and on TV.

Posted by Candide
March 22, 2009 3:36 PM | Reply | Permalink

Which one rhymes with Rasputin?

Posted by TM
March 22, 2009 7:12 PM | Reply | Permalink

This may add some light to the question . . .

Nate Silver over at 538 has this explanation, although it is related to polls that were conducted relative to the stimulus package back in February. But the heart of the reason there's a noted difference in outcomes still holds true with these Obama approval ratings.:

All polling of the stimulus to date in fact, has shown at least a plurality and usually a majority of Americans in support, with margins varying depending on question wording, how the pollster constructs its sample, and so forth.

The most pessimistic result for the stimulus package comes from a poll released last week by Rasmussen Reports, which pegs the stimulus as being supported by 42 percent of likely voters with 39 percent in opposition. This represented, moreover, a modest decline for the stimulus from a week prior, when Rasmussen found 45 percent of likely
voters in favor and 34 percent opposed.

Rasmussen, however, has shown considerably narrower margins for the stimulus than other polling conducted during the same period. Gallup, for example, showed the stimulus preferred by a 52-37 margin, with essentially no change in support from a similarly-worded poll conducted in early January. And a Democracy Corps poll shows the stimulus with wide support, with 69 percent in favor of the initiative against just 24 percent opposed.

In short, there is some evidence -- the trendline in the Rasmussen poll -- that the stimulus has become less popular. There is no evidence, on the other hand, that the stimulus has become unpopular; on the contrary, the preponderance of polling evidence suggests it remains a course of action that most of the public likes.

Why the stark difference between Rasmussen and, say, Democracy Corps? Both are applying a likely voter methodology, whereas Gallup and most of the other polls on the stimulus are surveying all adults. I understand the argument that likely voters are the ones who matter from the standpoint of electoral politics -- but determining who is and who isn't a likely voter when there is no election at hand is a little abstract for my tastes. (Are these people who are likely to vote 2010? In 2012? People who were likely to have voted in 2008? Or what?) This is one situation where I'd trend to prefer the major news orgs' polling, as they tend to place more emphasis on just this sort of issue-based polling as opposed to the horse race stuff.

fivethirtyeight.com/2009/02/stimulus-still-popular-after-all-these.html

I tend to take Nate's opinions quite seriously in light of the fact that he knows what he's talking about. Even if he did screw up on his Oscar Awards picks.

~OGD~

Posted by OldenGoldenDecoy
March 23, 2009 12:26 AM | Reply | Permalink

The following was copied from the Rasmussen web page.

"Generally speaking, polls of all adults will show a somewhat higher rating for President Obama than polls of likely voters. Why is this? Primarily because some demographic groups such as young adults are less likely to vote than others. These same groups also happen to be segments of the population where the current president gets rave reviews. So if a poll of all adults shows the president’s approval rating at 60%, you’d expect a comparable poll of likely voters to show a rating of roughly 57%."

Personally, I think the difference between Adults and Likely Voters is substantially more than 3%, especially for President Obama. As Nate notes, the president enjoys wide support among young people. It's also worth noting that -- notwithstanding the last election -- blacks are less likely to vote than whites. As a result, they will be disproportionately excluded from Rasmussen's ratings.

I think Scott Rasmussen would be doing himself a favor by publishing his results for Adults AND likely voters.

Posted by poli sci prof
March 31, 2009 3:53 PM | Reply | Permalink