The Wall Street Journal-20080130-Campaign -08- In Campaign 2008- Pollsters Are Biggest Losers- A Heavy Influx Of New Voters Adds to Difficulty

Return to: The_Wall_Street_Journal-20080130

Campaign '08: In Campaign 2008, Pollsters Are Biggest Losers; A Heavy Influx Of New Voters Adds to Difficulty

Full Text (868  words)

The expected huge voter turnout in next week's Super Tuesday primaries is likely to cause red faces among pollsters -- again.

The polling industry missed Hillary Clinton's comeback in New Hampshire, vastly underestimated Barack Obama's victories in Iowa and South Carolina, and overestimated John Edwards's strength in Nevada by a factor of four or five. On the Republican side, John McCain did better in South Carolina than most polls predicted, Mike Huckabee did better in Iowa, and Mitt Romney shredded the predictions in the Nevada caucuses.

That track record isn't likely to improve as voters place their ballots in 23 states Tuesday. "My advice [to fellow pollsters] is take two aspirins and wake up Wednesday morning," said Peter Hart of Hart Research, which conducts The Wall Street Journal/NBC News poll.

Pollsters face a double problem this year. Huge numbers of unpredictable first-time voters are turning out, and the campaign's fast pace is causing voters to change their minds -- often more than once and at the last minute.

In South Carolina's Democratic primary, where Mr. Obama took 55.4% of the vote Saturday, 532,000 people cast ballots, up from 290,000 four years ago, according to state records. In exit polls, 21% of a sample of those voters said they made up their minds within the previous three days, and 11% said they had decided only that day.

Voter turnout is typically low in primary elections, and turnout among the young, the poor and nonwhites is typically lower in all elections. Pollsters use past voting behavior to predict who will turn out in an election, and base their predictions on interviews with those people. But this year's races are generating such interest that they are turning out people who don't usually vote and whose preferences may not have been factored into polling calculations.

In the South Carolina Democratic exit poll, one-third of voters said they had a high-school education or less, 29% said they had a family income of less than $30,000, and 55% were African-American.

All of those groups, who typically are underrepresented in polls, went for Mr. Obama by wide margins.

In New Hampshire, exit polls of Democratic voters suggested that 11% were under age 25 and 19% were casting a primary-election ballot for the first time. Mrs. Clinton won New Hampshire with the apparent last- minute help of middle-aged women, but young and first-time voters went decisively for Mr. Obama.

Polling accuracy is "put at some risk . . . when there's a sudden doubling of turnout and a surge of participation by groups that have not participated before," said Scott Keeter, polling director for the nonprofit Pew Research Center in Washington.

David Woodard, who conducts Clemson University's Palmetto poll, said 40% of the South Carolina voters he sampled said they were undecided three days before balloting, when his polling stopped. Another 30% who were pressed to give an answer said they still could change their minds. His poll showed Mr. Obama ahead, but with 27% of the vote. "I never saw that Obama tidal waving coming," he said.

In the compressed schedule, a small event also can have an outsized impact. Pollsters believe Mrs. Clinton might have turned the New Hampshire results in her favor with a brief, emotional moment the day before balloting there.

Pew's Mr. Keeter, among others, believes former President Clinton's comments about race just before the South Carolina primary may have given a last-minute boost to Mr. Obama.

Once the two party nominees are selected and the country heads into the November general election, polls should be more reliable, pollsters said. Voters typically are more familiar with the candidates and not as likely to switch between them in a general election, said Mr. Keeter.

Turnout will remain unpredictable -- particularly if Democrats capture voter interest by nominating either the first woman or the first African-American at the head of a major-party ticket. But it likely won't double, as it did in South Carolina, and pollsters will have more time to identify and interview likely voters.

Still, changing habits are making it harder to find voters. Telephone answering machines, caller ID and call-blocking devices make it easier for households to avoid pollsters.

Mr. Keeter said Pew's response rate -- the percentage of people who will pick up the phone and respond to questions -- is about 20% to 25%, down from 36% a decade ago. Hardest to reach are less-educated, lower-income and minority voters.

Younger voters are hard to find, too, because of their preference for cellphones. Cellphone numbers aren't listed and often have area codes that don't reflect where their users live; a pollster calling a California area code could reach a student in Texas, for example.

People with only a cellphone tend to be younger, more Democratic and more liberal than those who also have a landline. But they also tend to be less politically active, which means they haven't skewed poll results so far, pollsters said.

"It doesn't mean the relationship will hold forever," said Michael Traugott, president of the World Association of Public Opinion Research, who sits on a pollsters' committee on cellphone use. By some estimates, cell-only households could increase to about 25% of all households this year.