« No evidence for Mellman's "magic" 50% approval | Main | Twitter roundup »

June 25, 2010


Pew was of course reporting on a different poll and a different time period. The headline used by Pew would have been inaccurate if used to describe the recent NBC/WSJ poll.

While Brendan is certainly correct that the shift shown by that poll is within the margin of error, it's also true that given the margin of error there may have been a major shift in Obama's approval rating. To characterize the results assuming one extreme of the margin of error and ignoring the other extreme is the worst kind of example of imposing a narrative on the facts.

If Brendan wants a more appropriate headline for the NBC/WSJ poll, how about "Effect of Recent Turmoil on Obama's Ratings Unclear"? But make no mistake. The headline used for the Pew poll wouldn't have been more appropriate for the NBC/WSJ poll than the headline MSNBC used, it would have been less appropriate.

For what it's worth, the polls were conducted over almost the same time period (Pew: June 16-20, NBC/WSJ: June 17-21).

My point about the headline was about coverage of Obama's poll numbers generally, not the MSNBC story specifically. I'm updating the post to clarify that point.

As far as Rob's larger point, while it's of course possible that there has been a major shift in Obama's approval rating, the combination of all the polls that are out there can help us create a more precise estimate than a single poll with a given margin of error. Given the current range of results, it seems unlikely that a major shift has taken place.

What I meant about time period is that Pew was considering the shift, if any, between January and June, whereas MSNBC was discussing the shift between early May and mid-June.

I'm happy to see Media Matters and others throwing out polls that gives results Dems don't like. There's no statistical justification for doing so. If Dems want to ignore their true problems, it's fine with me. That's how they managed to unexpectedly lose Congress to the Gingrich revolution.

Seriously, polls' accuracy can only be checked by comparing election results to polls shortly before the election. AFAIK Rasmussen has done as well as other polls by this measure. The fact that Rasmussen generaly gives somewhat different results is all the more reason to include it in the average IMHO.

Discussion of the margin of error is not that important IMHO. The best measure of the change is the change, regardless of the margin of error. If Obama's popularity declined by 3%, then 3% is the best estimate we have of that decline.

As Rob pointed out, the MOE says the "real" change may be larger or smaller than 3%, but so what? If the measured change were 3% down and the MOE were 3% at the starting and ending time point, then we could conclude that the "real" change was between +3% and -9%. Still, -3% would be the best estimate.

BTW the statistical MOE is not the only error. Polling structure can bias a poll. After all, the difference between Rasmussen most other polls is not due to statistical sampling error. if one added polling error to statistical margin of error, one would likely get a total MOE of 6% or more. Then we'd be seldom be able to draw any conclusions at all.

Media Matters For America is purely a leftist advocacy organization. Its conclusions and even its data are as reliable as that of, say, Michael Savage and should be treated similarly.

To understand who produces Media Matters' content, please check these links :



A related question is: Should President Obama's approval ratings be heavily damaged by the oil spill? That is, has he failed badly in Presidential his responsibilities?

A few conservative and local news sources say he has failed badly. E.g., a recent article from AoL News Opinion: Here's How Obama Can Turn the Oil Spill Cleanup Mess Around dispassionately lays out steps the government could or should have taken -- steps they should certainly take now, such as:
-- provide effective organization
-- make all skmmers available
-- accept help from foreign countries
-- waive the Jones Act, which impedes foreign help

If this article is accurate, then a huge amount of shore damage was avoidable, even though the leak continues. If articles with these same facts and more passion start routinely to lead the network news, then Obama's popularity would plummet. However, because the media don't want to attack Obama, I think he'll be spared that kind of coverage.

Addendum to the prior post:

This article contends that we should have immediately burned the oil off:

For some bureaucratic reason we could not burn the oil off as we corralled it with boom when the spill first happened. This response would have generated a fraction of the pollutants into our vast atmosphere than your average volcano does (let alone a Mt St Helens class eruption). The 2010 Iceland volcano eruption (pictured below) dwarfs anything a single burning oil well can produce. Yet burning the oil was deemed a ecological no-no.

AFAIK this criticism is not a staple of network news.

The comments to this entry are closed.