Can You Trust the Polls?

On the following day, the Gallup Poll said the president led his rival by 55% to 42% among likely voters, a huge difference from the Pew poll. An ensuing CBS-New York Times survey showed Kerry trailing Bush by nine percentage points. Democratic hand-wringing and calls for upheavals in the Kerry campaign followed.

On Thursday, however, a Wall Street Journal/NBC poll showed Kerry has pulled nearly even with the president. The poll of 1,006 registered voters, taken over the weekend, showed Bush leading Kerry by just 48% to 45%. The president lost ground on Iraq, the poll found, with 52% of respondents saying the war wasn't worth the military-casualty and financial costs.

Why such divergent poll results? There are several factors, say experts, including the lightning-speed news cycles of 24/7 networks like CNN and Fox News, as well as the Internet. But Andrew Kohut, president of the Pew Research Center, says it's really not that unusual to see rapidly shifting public opinion at this stage of an election season.

"In September 2000, Gallup's tracking poll went up and down and up and down," says Kohut, who used to work with the Gallup organization. "In other election years, we've seen the same thing. As voters come to have more of a definitive point of view as we approach the election, these polls, as they have in every election in recent memory, will come together and provide a pretty good estimate of the outcome of the election."

SmartMoney.com asked Kohut for his views on how the situation in Iraq might influence the outcome of the race, and how important the presidential debates will be.

SmartMoney.com: As one might expect, polls have been in the news a lot lately. But one thing is different this time around — the disparity between different poll figures. Why do you think one poll says Bush is leading in the double digits, and others such as yours say the race is far closer?

Andrew Kohut: The polls now are dissimilar after having been similar for a very long time because public opinion is unsettled. There are small methodological differences in the ways these polls are being conducted that did not create differences in their results from March through much of August. But now, voters are not as committed to their points of view as they were earlier in the race, since the Republican convention in the month of August unsettled views. There's a lot of voter volatility. That's not very different from most elections. In September 2000, Gallup's tracking poll went up and down and up and down. In other elections we've seen the same thing. As voters come to have more of a definitive point of view as we approach the election, these polls, as they have in every election in recent memory, will come together and provide a pretty good estimate of the outcome of the election.

SM: How much do the disparities have to do with the timing of the polls?

AK: Timing is something, sure. The polls that were conducted right after the Republican convention had some of the highest margins. The polls ever since, if you took the average, have been drifting toward a closer margin.

SM: Your latest poll shows that the size of the swing vote has actually increased, rather than decrease as it normally would. What does that tell you?

AK: It's a puzzle. There's a lot of discontent about the president's handling of the economy and his handling of the war in Iraq. The war in Iraq seems to get worse day by day. On the other hand, he has a good image as a strong leader, terrorism continues to be a top issue, and Senator Kerry has not made a good showing as voter attention has shifted to him since [the Democratic] convention.

SM: Bush gets a 42% leadership rating, while Kerry's is pretty low at 16%. Is that fairly common for a sitting president to be that far ahead of the challenger?

AK: A sitting president almost always has the advantage on leadership because he's played that roll and there's a stature gap, so to speak. But in this election, this issue of leader rather than manager of the government, which might have been the best description of the job in the 1990s when we didn't face a foreign threat, is pretty important, and the Republicans have been very successful in raising questions about Kerry's leadership abilities.

SM: One issue that has gained attention is defining the term "likely voter." How do you separate out a likely voter from an unlikely voter?

AK: I was president of the Gallup organization for 10 years and worked there for more than 20, so I basically use the Gallup system, which is a combination of questions that ask people how likely they are to vote in this election and that look at their past behavior and their attentiveness to the election. But I don't put as much emphasis on this issue this early in the season, because we know that the composition of the vote can change. The likely voters of September can be different from the likely voters of late October and early November. It's a tough, tricky thing to do.

SM: A major factor in this race, you've pointed out, is the criteria middle-class women are using to judge the candidates. Bush has a bit of a lead in the strength and ability to fight terrorism categories among women, you've found. Previously, middle-class women were more likely to vote for Democrats. How important is that?

AK: I think it's crucial. Men are going to support Bush on balance. The issue for men is how big the margin is. Women have been more variable, and that's because they are cross-pressured. In some polls they've been stronger backers of Kerry, in others it has been pretty close. So we have to keep an eye on them. That's one of the battlegrounds.

SM: How is Bush doing on Iraq in your polls?

AK: Not very good. But when we ask people if they have more confidence in Bush than Kerry, he leads Kerry by a lot of points. This is even in polls where Kerry has done reasonably well. Kerry just hasn't been able to make any progress in instilling confidence, even among people who are critical of Bush.

SM: The focus of Kerry's campaign seems to be swinging back and forth between terrorism, the economy and Iraq. What are your polls telling you about which issue is going to be the central one come Nov. 2?

AK: All of those issues are going to be central, but from my own point of view I think Iraq is going to be the most important. If this were only about terrorism, Bush would win. If it were only about the economy, Kerry would win. So Iraq is the trump card.

SM: Is Nader having any significant impact?

AK: Not much, but if it gets really close, a tiny impact could make a difference.

SM: Democrats seem pretty pessimistic, since 61% of the respondents in your latest poll said Bush is more likely to win, and only 23% said Kerry is most likely to win.

AK: Yes, that's very different from what we found a month and a half ago. Maybe with the more recent polls getting closer the Democrats will begin to gain hope. But they certainly lost hope in early September.

SM: How important are the debates going to be?

AK: I think they're going to be very big. The indecision that people have could be resolved in these head-to-head showings.