Redlawsk: Late Decisions Explain NH Polling

Redlawsk: Late Decisions Explain NH Polling

University of Iowa political science professor David Redlawsk told Iowa Independent that late decision making and an unexpected shift in independent voting patterns help explain Hillary Clinton's unanticipated New Hampshire victory.

"A, people moved towards Clinton, no matter what they had told the pollsters a couple days before," said Redlawsk, who conducts his own polling research. "B, independents went more over to the Republican side than the pollsters had anticipated. And those independents who did vote on the Democratic side were extremely Obama. You don't have to take a whole lot of those away plus see some movement to Clinton to change the numbers pretty substantially."

Redlawsk said the compressed campaign schedule and late shifts in opinion made it harder to predict Clinton's victory. "While candidates campaigned in New Hampshire, for all practical purposes the campaigns didn't really ramp up hard core until those five days" after Obama's Iowa win. "We know that people made decisions very late in the process overall. Clinton campaigned aggressively and intensely during that time, and seemed to people to be much more human than she had come across before."

"The polling pretty much ended Sunday. One of the pollsters was running on Monday but they were doing three-day averages," Redlawsk said. "So there were changes, but they really didn't end up affecting the three-day average enough to make a dramatic difference. There were hints of it, but most of the polls were pretty much done. I think there's a very good possibility that enough independents decided at the last minute essentially they were going to go vote for McCain on the Republican side, that it lost Obama serious points.

Redlawsk isn't sure if the polls themselves steered people toward the GOP race. "The numbers seemed to have spiked up so high for Obama with everyone saying he was going to win that it could have given some people permission in effect to say, let's go over to the Republican side. I think part of it is some people looking at it and going, well, (Obama) doesn't really need our votes. I also think there really was a move to Clinton that just didn't get picked up, independently of the independents. I think it just happened late."

"Remember, too, that every poll is a snapshot at a point in time," he added. "Even if you can average the polls and get a sense of what might be, they're all subject to error, plus or minus X percent. I'm not suggesting the methodologies were bad, but polling is always a probabilistic thing. The media tends to take polls as if they're election returns, when what they really are is estimates subject to error."

Redlawsk said he was "annoyed" by the way the New Hampshire result has been reported as a come-from-behind Clinton triumph. "While Clinton certainly did win it, of course, prior to Iowa her lead was routinely double digits. Obama did, in fact, surge substantially in NH coming out of the win in Iowa with Clinton third place," he said. "For him to come within three points considering where he was coming from, in any other circumstances other than this frenzy of polls saying 'Obama's winning' would have been considered a victory or at least one of those 'exceeds expectations.'"

The late decision making and independent shift is a more plausible explanation than charges of voting machine fraud that have led to a recount request from Dennis Kucinich. "While I recognize the reality that voting machines in some places are not as secure as we might like, I tend to look for the simpler answers."

Another argument that's been made has been variously called the "Wilder effect" or the "Tom Bradley effect," referring to two African-American candidates for governor in the 1980s whose pre-election polling was higher than their Election Day vote totals. Redlawsk has done some polling on this question, most recently in a March University of Iowa study. "Trying to get at the question is extraordinarily difficult," he said

You can't ask people directly, 'would you vote for a black man.' They're going to say of course, because that's the social pressure. But you can ask questions like 'the fact Obama is an African-American will be a problem for him.' In other words it's not me, it's other people. We asked that last question in March, and a substantial percentage, 20 to 30 percent, said it will be a problem for him. On the other hand across our full sample, both Republicans and Democrats, less than 10 percent said it's important that the candidate be of the same race that they are. That number is substantially smaller on the Democratic side than on the Republican side but it’s tiny on both sides.

Redlawsk also says exit poll evidence makes this polling effect less likely. "Why would they lie to one pollster the day before the vote, and then tell the truth to the exit polls after they vote?" he said.

While Redlawsk acknowledged that racism in the privacy of the voting booth may be an issue, he said such voters are less likely to be Democrats. "I would see that effect being more likely in a general election than in a Democratic primary."

Iowa, of course, does not have a private vote in the caucuses, and Redlawsk, a John Edwards supporter, thinks this may have been a factor in Obama's favor. "I've actually though all along that in places where Obama was likely to be extremely strong, like Johnson County, his being an African-American would be a substantial benefit," he said. "And it would be a benefit because it was public, because in a good lefty progressive place like (Iowa City), there was some pressure of how could you not stand up for the first African American with a reasonable opportunity. It's a factor, whether you like it or not. I happen to think Edwards is more progressive, and I had some frustration coming from the Edwards perspective."