Letter to CNN regarding coverage of polls


October 22, 2004

Tom Hannon
Executive Director, Political Coverage
The CNN Building
820 First Street, NE
Washington, DC 20002

Dear Mr. Hannon:

I'm writing out of great concern -- and confusion -- over CNN's recent reporting of national and state polling on the presidential race.

Media Matters for America has noticed a distinct trend in the way your network reports polls -- specifically, polls that show favorable results for President George W. Bush have received greater coverage than polls that show less favorable results for him or more favorable results for Senator John Kerry. In several instances, CNN has reported a poll that shows a relatively positive result for Bush when more recent polling shows less favorable numbers.

Examples include:

  • On October 18, Inside Politics host Judy Woodruff gave an assessment of the presidential race by combining the results of five recent polls -- CNN/USA Today/Gallup, Newsweek, TIME magazine, ABC/Washington Post, and IBD/TIPP -- into a "poll of polls" of likely voters that showed Bush with a five-point lead. But, inexplicably, the "poll of polls" didn't include two fresh polls that showed the race tied: one by Zogby, one by Rasmussen.

    That same day, a CNN graphic read "Polls: Bush's favorability rating rose from 51% to 55%." But that was a misuse of the plural, and a misleading claim: only one recent poll showed Bush's approval at 55 percent, while three others showed it between 49 and 51 percent.

  • On October 19, CNN host Rick Sanchez referred to a weeks-old poll that showed Bush with a seven-point lead in Ohio -- despite the fact that there were no fewer than five more recent Ohio polls, all of which showed a closer race and the most recent of which gave Kerry a two-point lead.
  • Also on October 19, CNN anchor Daryn Kagan claimed that a "comprehensive overview of five post-debate polls shows the Bush campaign having a bit more breathing room; it shows Bush with a four-percentage-point lead, just beyond the margin of error." But the "overview" wasn't "comprehensive" -- it omitted the most recent poll, as well as three other post-debate polls. Coincidentally, all four of the polls omitted from the "comprehensive overview" showed better results for Kerry than the polls used in the "overview" indicated.

    Kagan was also wrong about that "four-percentage-point lead" -- it wasn't "just beyond the margin of error," it was well within the margin of error. The on-screen graphic indicated that the margin of error was four points; what Kagan (or her CNN writers and producers) should know by now is that margins of error apply independently to both numbers. Thus, for a lead to be "outside" a four-point margin of error, the lead must be more than eight points.

  • On October 21, Woodruff read a series of state polls, all of which were quite favorable for Bush. But for at least three of the states -- New Hampshire, Wisconsin, and Michigan -- there was more recent polling that showed better results for Kerry.

I am perplexed about this reporting. I find it difficult to imagine that CNN is intentionally presenting only polling favorable to Bush, even when more recent polling exists. Yet I can discern no other pattern to CNN's reporting; for example, this isn't a situation where CNN is simply ignoring all polling except CNN/Gallup polls (a practice I would object to, but at least it would be clear what CNN is doing).

I look forward to an explanation for CNN's editorial choices regarding polling, and in any case, I hope and expect CNN will either begin covering all credible publicly released polls equally and fairly, or refrain from any further on-air discussion of polls.


David Brock
President and CEO
Media Matters for America

Polling, 2004 Elections
We've changed our commenting system to Disqus.
Instructions for signing up and claiming your comment history are located here.
Updated rules for commenting are here.