Crowds vs. Experts

To be honest, I’ve never been a big fan of Intrade, the Iowa Electronic Markets, or any of the other online prediction markets. People go to these sites and trade contracts or positions (read: take bets) on whether a given event will occur, such as which party will control the House or Senate (to the extent that anyone can control the Senate) after the next election, or which candidate will win the presidential race. The theory of collective intelligence—the rationale for paying attention to sites such as Intrade—was made by James Surowiecki in his 2004 book, The Wisdom of Crowds: Why the Many Are Smarter Than the Few and How Collective Wisdom Shapes Business, Economies, Societies, and Nations.

My bias is to listen to the experts. I am more likely to pay attention to a crowd if the crowd is composed of political analyst Stu Rothenberg, NBC Political Director Chuck Todd, and about a dozen other folks whose opinions I respect rather than a random mass of people who have their own opinions, often times with little to back them up.

Pros have enormous amounts of experience and background knowledge about a subject. Rothenberg, Todd, and other professionals constantly work their contacts and scrutinize data, including polling data, that are not necessarily available to the public. What they say and think is worth considering. When we agree, I feel reassured; when we disagree, I go back and test my assumptions, just in case they are right and I am wrong. The system has worked well for me.

Having said all of that, I’ve finally found a use for Intrade. The website is a very convenient way to quantify conventional wisdom. According to Intrade on a recent (Monday) night, President Obama had precisely a 58.8 percent chance of getting reelected. Apparently, his odds had dropped about six-tenths of 1 percentage point in the previous day. This decline is presumably a result of the crowd’s assessment of the impact of Obama’s endorsement of same-sex marriage (as if that will really have an impact six months from now). On that Monday, the Intraders saw Republicans as having a 74.9 percent chance of keeping their House majority. Democrats had a 29.8 percent chance of regaining the chamber. These predictions strain credibility a bit, as the odds add up to more to than 100 percent, but that’s another matter.

On this one wager, the numbers are not too far off The Cook Political Report’s prediction that Republicans have a 75 percent chance of holding the House (and, yes, Democrats have a 25 percent chance of taking it). In the Senate, Intrade says that Republicans have a 56 percent chance of taking control (its phrase, not mine). Democrats have a 27.9 percent chance. My hunch is that the odds of neither side controlling the Senate are 100 percent. At The Cook Political Report, we see the Senate as purely a 50-50 proposition.

But it’s the 58.8 percent chance of Obama winning that interests me today, because that prediction stands in stark contrast to what most pollsters, Democrats and Republicans alike, whom I talked with privately, believe. The number crunchers who conduct and analyze polls, and others who study these things closely, see a lot of metrics pointing to a very close contest that could go either way. They don’t see an election in which either Obama, or Mitt Romney, is likely to have an almost six-in-10 chance of winning.

Take the polls, for example. The averages of all major national polls show the race as extremely close. Pollster.com gives Obama a 1.2-percentage-point lead over Romney, 46.3 percent to 45.1 percent. Realclearpolitics.com pegs Obama’s lead at 2 points, 47 percent to 45 percent. Gallup’s seven-day tracking poll puts the president’s lead at 1 point, 46 percent to 45 percent. Undecided voters, particularly, often break away from well-known, well-defined incumbents (the “what you see is what you get” rule for those in office). Does this really translate into a strong advantage for the president?

Obama’s job-approval ratings are often “upside down” in pollster parlance, with disapprovals running higher than approvals in both Pollster.com (46.9 percent approve; 48.4 percent disapprove) and Realclearpolitics (48 percent approve; 48.5 percent disapprove). Gallup also shows 47 percent approve and disapprove numbers for the week of May 7-13. Is that really a decisive edge?

In terms of the Electoral College, seven states—Colorado, Florida, Iowa, Nevada, Ohio, Pennsylvania, and Virginia—are likely to be extremely close. New Hampshire might also be tight. (I am increasingly skeptical that Obama can win North Carolina.) I pay a lot of attention to the top-dollar surveys by the Obama and Romney campaigns—and, for that matter, what highly regarded pollsters doing surveys for various senatorial and gubernatorial candidates and for ballot initiatives in the states say. I don’t put a lot of stock in the dime-store polls, which bloggers and Internet armchair analysts so avidly follow (ask them about calling cell phones; that separates the top-notch pollsters from the cut-rate crowd).

Don’t get me wrong: I’m not predicting that Obama will lose. I’m only pointing out that the discrepancy is real between what the pros on the sidelines and those in the press box are seeing, versus those with the view from the cheap seats. Just sayin’.