Could the oh-so-close presidential polls be skewed because they don’t account for many young voters who use cell phones only?
“That’s the biggest bunch of horse . . . I’ve ever heard,” says pollster Brad Coker, who operates the widely used Mason-Dixon Opinion Research poll.
About six weeks before the presidential election, most polls in Virginia give one candidate or the other an edge of 2 to 4 percentage points.
Some supporters of Sen. Barack Obama, puzzled by why he can’t mount a big lead over Sen. John McCain in a difficult environment for Republicans, say the pollsters are missing younger, pro-Obama voters who have cell phones only.
Even some pollsters raise another uncertainty about the plethora of Virginia polls — whether all of the respondents who say they back Obama will vote in November for the nation’s first black major-party nominee.
Nearly 250,000 first-time voters have registered in Virginia this year, and 42 percent are under the age of 25. The overall gain has pushed Virginia’s voter rolls to 4.8 million people.
Coker said if the sample for a poll includes the number of young voters in proportion to the population, the absence of cell-phone users doesn’t matter. He said exit polls taken during the 2004 presidential election showed no difference in voting behavior between landline and cell-phone users.
Polls are weighted to match the demographic composition of the electorate, Coker said.
The Pew Research Center’s Scott Keeter, a former pollster at Virginia Commonwealth University, found that cell-only respondents are significantly more likely to support Obama. But he said they also are substantially less likely to be registered to vote and, if registered, less likely to go to the polls.
A Pew survey in June found that Obama held a 48 percent to 40 percent advantage over McCain among cell-phone users and a 46 percent to 41 percent advantage among landline users.
The Gallup organization, one of the oldest and most respected polls, says it does account for cell-phone users. About 15 percent of households now use cell phones only.
Residents of those households tend to be younger, more minorities and more transient, the Gallup organization’s Web site says.
Those would be more likely to be Obama supporters.
Since Jan. 2, Gallup has been including cell phone-only households in all of its telephone surveys, the Web site says. The most recent national Gallup poll, taken Friday, shows Obama leading by 5 percentage points.
Coker said the Obama campaign should be more worried about the so-called “Wilder effect” or “Bradley effect.”
The phenomenon was named for Virginia’s L. Douglas Wilder and California’s Tom Bradley, black office holders who saw substantial poll leads disappear on Election Day. This resulted in a theory that some voters are embarrassed to tell pollsters that they will not support a black candidate.
In 1982, Bradley, the mayor of Los Angeles, led in the polls but lost California’s election for governor.
Two days before Virginia’s 1989 election for governor, Wilder led his Republican opponent, J. Marshall Coleman, by 15 percentage points, according to one poll. Wilder won the election, but it was so close there was a recount.
In an interview last week, Wilder, now Richmond’s mayor and an Obama backer, said the public polls were wrong in 1989. Wilder said his own campaign’s internal polling showed the contest was much closer.
He said he expects only minor slippage for Obama, “de minimus” in Wilder’s words.
If there is slippage, it would be among independents, who already are more reluctant than partisans to tell their views, Coker said.
Quentin Kidd, who directs a poll at Christopher Newport University in Newport News, said many pollsters have dismissed the Wilder effect this year, because Obama did well among whites in the Iowa caucuses and the New Hampshire primary.
In Virginia’s Feb. 12 Democratic primary, Obama won handily with 63.6 percent of the vote. Obama outperformed his tally in all of the public polls leading up to the primary.
But Kidd said he still expects some slippage for Obama in the South on Nov. 4.
“I know most people will say they want to vote for the better candidate, but when they get in the voting booth, they might think otherwise,” he said.
Kidd pointed to another polling obstacle. Polling on weekends, particularly on Sundays, tends to miss likely Republican voters, he said.
Two Virginia polls released Wednesday showed dramatically different results, pointing to the importance of how pollsters estimate turnout.
McCain led Obama 53.8 percent to 46.2 percent in a Christopher Newport poll. But in a survey by Public Policy Polling, Obama led McCain 48 percent to 46 percent.
Tom Jensen, communications director for Public Policy Polling, says Christopher Newport underestimated the turnout of blacks and young voters in Virginia, so it undercounted Obama’s support.