Economist has Seven questions for Nate Silver
NATE SILVER has been called a "prodigy", a "wunderkind", and a "genius". Using his unique methodology, he was able to correctly predict every state but Indiana in the 2008 presidential election. He also got every Senate race right. That is why we here at The Economist find ourselves visiting his website, FiveThirtyEight, several times a day. This week Democracy in America talked with Mr Silver about polling, predictions and politics.
DIA: Since you were able to predict so many 2008 races months ahead of time based on immutable factors, how much do political campaigns really matter?
Mr Silver: There are two separate questions here: how much campaigns matter and how much candidates matter. Candidates clearly do matter, in the sense that it's not uncommon to get results in individual races that significantly diverge from the national trends. For instance, right now we have two Republican senators in Maine, which voted for Barack Obama by 18 points, and two Democratic senators in North Dakota, which hasn't voted a Democrat for President since 1964. Likewise, we have a Republican governor in Vermont and a Democratic governor in Kansas. When you see gravity-defying results like those, it usually boils down to a very talented candidate.
But how much do campaigns matter? They clearly matter some. Hillary Clinton, who proved to be a much stronger candidate than I expected but ran a very poor campaign, is probably Exhibit A in this regard. On the other hand, John McCain’s campaign basically went bankrupt during the Republican primaries and he managed to win the nomination.
DIA: Was there one campaign from 2008 that stood out as best at beating the mathematical odds? If so, how did they do it?
Mr Silver: The Republican Joseph Cao defeating Democratic incumbent William Jefferson in Louisiana's 2nd congressional district, in which probably something like 70% or 75% of the voters are Democratic. It's clear enough how Mr Cao managed to do this-Mr Jefferson was exceptionally corrupt and ran a complacent campaign-but it was nevertheless an impressive win.
DIA: Despite all of the focus on Pennsylvania in the last days of the 2008 presidential campaign, it fell so easily to Barack Obama that people asked why it was even considered a "swing state". What, in your mind, are the swing states to watch in 2012? (Which blue state is most obtainable for Republicans? Which red state for Democrats?)
Mr Silver: Generally speaking, the same states that were swing states in 2008. But I would keep an eye on Georgia for the Democrats, which is rapidly beginning to urbanise and is becoming part of the "New South". For the Republicans-maybe New Jersey? One would expect that their anti-tax message would play well among wealthy voters, and there are a lot of wealthy voters in Jersey.
The other thing to watch is whether Barack Obama performs better in some parts of Appalachia, where I think his race hurt him in 2008. It wouldn’t surprise me if, after watching him on television for four years, some voters who had previously harboured some degree of racial animus begin to feel more comfortable with him. If so, Mr Obama could be more competitive in states like West Virginia.
DIA: You wrote that the McCain campaign "radically overestimated the importance of appealing to the base" in 2008. What type of political message would best serve Republicans trying to form a winning electoral coalition in 2010, 2012 and beyond.
Mr Silver: Well, that’s the $64,000 question. Clearly if the economy has not recovered, the Republicans will have the script somewhat written for them, and they can rail against the expansion of government or deficits or taxes or what have you. But I think they'll have to do it in an affirmative, pragmatic way-the "party of no" critique is a little hackneyed, but is probably fairly effective counter-messaging. And if the economy recovers? I don’t think it matters much what they do.
If they’re spending a lot of time talking about gay marriage, by the way, I think that's a sign that they're in trouble-unless perhaps the Supreme Court has ruled on it, which could reset the whole debate.
DIA: I've heard you say that baseball analysts put too much emphasis on what just happened. Is the same true of political analysts?
Mr Silver: For sure. And in baseball, at least, they're playing every year, whereas you only have a presidential election once every four years. The McCain campaign operated under the assumption that the political world hadn't changed since 2004-that Mr Obama couldn't turn out black voters or young voters, that swift-boating would work, that Mr Obama couldn't possibly win states like North Carolina and Indiana-and they paid a price for it. On the other hand, I think some Democrats might be a little bit complacent right now. There are a lot of things that can go wrong-both known unknowns and unknown unknowns. What if Afghanistan turns into the next Iraq? What if swine flu winds up killing several hundred thousand Americans? What if there's a nuclear exchange in Kashmir? What if there's a significant, unpredicted increase in the crime rate? Some of those things might hurt the Democrats and others might not, but there's a pretty decent chance that the core issues in 2012 will be things that we haven't even thought about yet.
DIA: You reach some surprising conclusions using your statistical techniques. How good is your average reader's understanding of your work? Do people ever say, "I don't care about your fancy 'mathematical' methods, that can't be right so you must be doing something wrong."
Mr Silver: More often I get the opposite-people who tell me they don’t really follow the analysis, but trust my conclusion. That's more worrying in some ways, I think.
DIA: What went wrong with the Oscar predictions?
Mr Silver: Hah! Well, you know, we got two different predictions wrong-we missed on best supporting actress and we missed on best actor. The best actor prediction I don't feel that bad about. It's very rare for someone to win best actor twice within a short period (Sean Penn had won for "Mystic River" in 2003), and it was probably only the fact that Mickey Rourke had been a colossal jerk to his peers that prevented him from winning. Forecasting deals in probabilities, not certainties, and sometimes you can make a good forecast and wind up with an incorrect result, or vice versa. If I had told you at the start of last year that the Arizona Cardinals were going to make the Super Bowl, would that have been a good prediction? No-if I'd told you that you would have had me institutionalised.
On the other hand, with the best supporting actress category, I did some sloppy modeling and clearly screwed up-that might have been the worst prediction since "Dow 36,000". My apologies to anyone who bet their mortgage on Taraji P. Henson.
source