Six Questions for Nate Silver on Polls and the Election
Nate Silver is the founder of the popular website FiveThirtyEight.com and a writer, analyst, and partner at a sports media company called Baseball Prospectus. He developed a system called PECOTA, which predicts future athletic performance (accurately calling, among other things, the Chicago White Sox’s dismal 72–90 record in
2005, two years after they were world champions), and his political projections have been equally impressive. As a Newsweek story on Silver noted, “He came within 20 delegates of the final split on Super Tuesday (out of nearly 1,700) and 2.5 percent, on average, in the other six post-March primaries.” Silver is an Obama supporter, which he says does not influence his methodology. I recently spoke to Silver by phone and asked him six questions about how the election is shaping up as we head into the last stretch of the campaign. This interview was edited for length and clarity.
1. The polls have been jumping around a lot, with McCain getting a big bounce after the G.O.P. convention, but Obama seems to be back on top now. Are long-term trends really emerging, or should we just assume the polls will keep shifting from one day to the next?
Last week was difficult to interpret because we had a few things going on at once: Wall Street melted down, and McCain’s convention bounce began to wear off. But now there seems to be a sharp uptick for Obama. He’s got a two- or three-point lead, which is not robust, but it matters, and it’s starting to get close to the election. If you look at the broad arc of the campaign, Obama has been ahead by three, five, six points, except for when Jeremiah Wright was front-page news and, later, immediately after the G.O.P. convention. At the point of equilibrium Obama wins by a few points, so McCain needs to do something to move away from that equilibrium–say, raise significant doubts about Obama or win the foreign-policy debate by a big margin. If he loses the debate, he doesn’t have a lot of cards to play; if he wins, it’s probably back to a rough tie. But we give Obama about a 75 percent chance of winning the election. He’s stronger than his slim lead in the polls suggests.
2. In a recent post on your site, you criticized an AP poll and related news story that suggested that Obama’s support “would be as much as 6 percentage points higher if there were no white racial prejudice.” What’s the gist of your critique? And is it possible that there will be some sort of “Bradley Effect on Election Day?
There is no doubt that Obama’s race hurts him with some voters, and that some people will not vote for him because he’s black, but race probably helps him with other voters. The AP didn’t really open up and disclose how they got that 6 percent, but it looks like their survey reflected the views of all adults–very different from the views of likely voters. As to the Bradley Effect, there was no evidence during the primaries that white voters are lying to pollsters about their willingness to vote for Obama. In fact, Obama frequently outperformed his polling numbers. In North Carolina, the primary polls showed it might be a tight race, but he beat Hillary Clinton by fourteen points. So there and in some states in the Midwest the pre-election polls apparently underestimated the black turnout. There may even be a reverse Bradley Effect, where African-American voters are reluctant to tell pollsters that they are going to vote for Obama. In the abstract, the AP argument may make sense, but only in the abstract (for example, if Obama were a white candidate, he wouldn’t be Barack Obama). And you have to remember that Hillary Clinton was never going to run away with the election if she had won the nomination. When she exited the race in June, she was only three points ahead of McCain. That’s where Obama is now. (Note–Silver has written a detailed web post on this topic)
3. So why is the race so close?
McCain doesn’t get enough credit. This is the guy Kerry almost picked as his running mate in 2004. He has a lot of crossover appeal; he’s perceived as a maverick and a moderate; and he has a good story–he’s a war hero. A lot of Democrats kind of like him. All along he was the only Republican candidate who was in range of Obama or Clinton. Any other Republican would have been running ten or fifteen points behind at this point.
4. In the end, will McCain’s choice of Sarah Palin help or hurt the G.O.P. ticket?
Her favorability ratings have declined by ten or fifteen points in the span of a week. She’ll play well among “Perot independents” out West, and she obviously took Alaska out of play, but in a lot of places her résumé may seem thin. McCain had to take a risk because he was so far behind, but it was a big risk. It violated the first rule of picking a vice presidential candidate, which is “Do no harm”–she’ll turn off a lot of voters. Vice presidential nominees are usually very likable. People forget now, but even Cheney had high favorability scores when he was picked in 2000.
5. What are the key states that are likely to determine the election?
The battlegrounds change on a daily basis. Obama has to win Michigan and Pennsylvania. There are scenarios where he could lose one of them, but it would make things tough. Obama’s easiest path to victory is to hold the Kerry states and also win Iowa, Colorado, and New Mexico. That would give him 273 electoral votes. Right now he’s ahead in Iowa and New Mexico by seven or eight points, but Colorado is closer. The only Kerry states where McCain has a chance are Michigan and Pennsylvania, and maybe New Hampshire, but I’m not sure New Hampshire’s four electoral votes will matter. McCain thought he might win in Wisconsin or Minnesota, but that doesn’t look likely now. And he has to hold on to Ohio, Colorado, Florida, and Virginia, and even Indiana and North Carolina could be close. So a lot of strange things could happen, but Obama only has two “must-win” states while McCain has half a dozen.
6. And what about the congressional races? A few months ago there was talk of huge Democratic pick-ups in both houses.
On the Senate side, Democrats will probably gain five or six seats but probably won’t win a filibuster-proof majority. They should win seats in Virginia and New Mexico, and probably New Hampshire. Oregon and Minnesota are toss-ups. On the House side, the Democrats should win another fifteen or twenty seats. Here’s where picking Palin helped, because she’ll help bring out the base in the congressional elections. It should be a good day for the Democrats but not a landslide.