Susquehanna Prof Finds Political Poll Sponsors Give Own Polls Favored Coverage

As long as opinion polls are conducted, newspapers will always have self-made stories to print on the latest political campaign -- like the ones that will end next Tuesday. But in a research paper titled "'And they're off!' How newspaper-owned polls handicap public opinion in horse-race campaign coverage," J. Patrick McGrail, assistant professor of communications at theatre arts at Susquehanna University in Selinsgrove, Pa., found that newspapers favor the inclusion of their own polls over those conducted by other media outlets.

I thought you might find this particularly of interest with another election day just a week away.

In his study, which he presented at to the International Communications Association (ICA) during its 2001 conference in Washington, D.C., McGrail examined the popularity polls of the Republican and Democratic presidential candidates in 1999, and how prominently and frequently they were reported in four national daily newspapers -- USA Today, The New York Times, The Washington Post and The Los Angeles Times. He entered the search word "poll" into Lexis-Nexis, and the 10 polls that emerged most frequently among presidential candidates were studied from those papers during the calendar year 1999. The polls that emerged from this search were, in descending order of frequency:

1. the CNN -- Gallup -USA Today Poll

2. the Los Angeles Times Poll

3. the Washington Post Poll

4. the New York Times Poll

5. the Pew Research Center Poll

6. the Mason-Dixon Poll

7. the Reuters Poll

8. the Zogby Poll

9. the American Research Group Poll

10. the Quinnipiac Poll

After identifying the polls, McGrail determined if the newspaper reporting the results had sponsored the poll and if these articles reported on newspaper or non-newspaper polls. He then assessed these articles' prominence in the newspaper itself based on a five-point scale that examined the placement of the story, the section the story appeared in, whether a graphic accompanied the story, whether a photo accompanied the story, and the word count of the story.

Of the 158 articles studied, 102 (64.6%) came from USA Today. The Los Angeles Times had 26 (16.5%), The Washington Post had 19 (12%), and The New York Times had 11 (7%).

"The fact that USA Today had more poll articles than all other papers combined creates an inherent caveat for extrapolation and interpretation of data. I found that the other papers usually polled in conjunction with a running story on a particular candidate, whereas USA Today commissions polls at regular intervals that then form the basis for feature stories," says McGrail.

Of the 102 poll stories that USA Today reported on, McGrail found that 90 of them (88.2%) were generated by USA Today. Of the 26 poll stories reported on by the LA Times, 20 (76.9%) were sponsored by the paper. However, only six (31.6%) of the 19 poll stories that the Washington Post ran were generated by the Post, and only four of the 11 (36.4%) stories that the New York Times ran were generated by it. Because of the strength of the relationship where the larger numbers are concerned, McGrail concluded that papers favor the inclusion of their own polls over someone else's.

"It (the newspaper's polls) allows them (the reporters and editors of the newspaper) to literally manufacture news, instead of the more expensive gathering of it," McGrail says. "The poll becomes a kind of brand for them, something else for them to be known for in a crowded marketplace.

"The most significant change in polls since the Roper and Gallup halcyon days of the 30's and 40's is the fact that most of them are now owned by media companies. And therein lies a problem. Despite cursory attention paid to methodology, my own research shows that the media have a marked tendency to over-report their own polls and under-report anyone else's regarding the status of a candidate. This can be a distinct problem when the race is very tight."

McGrail's study is particularly appropriate with another election just a week away, and newspapers now inundating readers with political polls -- particularly in the closest races.

"Polls emphasize the 'horserace' aspect of campaign journalism, with its singular obsession with who's ahead and who's behind," he says. "This feeds into the overall dichotomous narrative of the news, which is always seeking to simplify and clarify ambiguous truth."

That has not been the case in the Pennsylvania Governor's race, where Democrat Ed Rendell has a double-figure lead on Republican Mike Fisher -- regardless of which poll is used.

"As for polls and the current Pennsylvania governor's race, I believe that the polls accurately reflect a significant pull-ahead for Rendell," says McGrail, "One reason I believe is that this is the conclusion of the Keystone Poll, run by academics out of Millersville University, who are painstaking in their methodology."

"But polls, even of likely voters, cannot accurately reflect fast moving tides at the conclusion of campaigns. There is as much art as science to it."