A poll in 20 swing states published on Tuesday showed John Kerry still clinging to a narrow lead over President George Bush in the key election battlegrounds, but it raised questions about the reliability of such polls at a volatile point in the campaign.
While some other surveys have shown the election as a tight race, a string of recent polls has given the president a double-digit lead, provoking controversy about the various techniques used by the survey groups.
Tuesday’s poll by Zogby International suggested that although Kerry was losing ground, his support was better distributed in the battleground states, giving him a potential edge in the electoral college, which reflects the way each state votes and ultimately selects the president.
If the election was held today, according to Zogby, Kerry would have 264 votes in the electoral college (six short of the number needed to win), and the president would have 241 votes.
Florida, which accounts for 27 votes, and Arkansas, with six votes, would be tied.
This is a significant erosion of Kerry’s margin since the last Zogby poll two weeks ago, when he had a 51-vote lead in the electoral college.
But it contrasts dramatically with the the latest national poll by Gallup, published last week, which gave Bush a 14-point lead over Kerry in a three-way race with Ralph Nader, the leftwing independent candidate, who drew 3% support.
That in turn differed sharply from another national survey last week by an equally respected body, the Pew Research Centre in Washington, which gave Bush a single-point lead.
Some experts have pointed to differences in methodology. All three polls asked the preferences of likely voters, but each used a different questionnaire to determine who counted as a likely voter.
More controversially, Zogby International weights its results according to party identification, adjusting them so that the proportion of its sample that identifies with each party conforms to the population as a whole.
Other polling organisations adjust for height, race, and gender, but argue vehemently against weighting by party identification, saying that it is subject to change.
John Zogby, the head of Zogby International, disagrees, arguing that party identification changes only glacially and that the unusually high number of Republicans in the Gallup sample suggested it was skewed.
Maureen Michaels, an independent pollster in New York, said there was merit in both approaches, and argued that the wide variation in poll results in past weeks has more to do with the volatility of the electorate than the techniques of the pollsters.
”We tend to have a lot of volatility at this time of the year in the polls,” she said. ”Things are going to bump around. But there’s no slam-dunk in this election, and no one is going to win by a landslide.”
The polls do agree on one overall trend though: President Bush enjoyed a surge of support after the Republican convention, and has held on to some of that momentum.
An average of the polls of the past few weeks gives Bush a six percentage point lead.
Almost all the polls show a significant erosion of support for Kerry in a key Democratic constituency: women.
In 2000 Al Gore would have beaten Bush by 11 points if the only voters had been women. Kerry’s advantage is now much slimmer, and some polls have even shown Bush winning the female vote.
The Bush campaign has particularly targeted the female electorate, under the banner ”W is for women”, derived from his middle initial. He has stressed that Afghan women have benefited particularly from the fall of the Taliban, and emphasised the impor tance of the ”war on terror” to keeping American families safe.
The tactic appears to have been so successful that commentators are now talking about the pivotal role of ”security moms”, replacing the ”soccer moms” of the 1990s whose support won Bill Clinton two elections. – Guardian Unlimited Â