In Blinded By the Right, David Brock talks about conservative operatives buying huge numbers of conservative books in New York City. They did this (and presumably still do), in order to game the New York Times Best Seller List. If a book makes this list, the author will be invited on talk shows and the (mostly vile) ideas in the book will get more attention. Well, it appears the conservatives are doing this with polling data.
Other than Nate Silver’s Five Thirty Eight predictions, no polling data source gets more attention than Intrade, the political betting market. Last night, Silver wrote that Intrade’s prediction for Mitt Romney was way out of line with not only models like his, but with other political betting markets. In general, Silver’s model has given Obama a chance of winning in the high sixties. Markets other than Intrade put the number in the mid-sixties. Intrade is predicting a 55% change of an Obama win.
Silver suggests that people may be trying to game the system, although like a good statistics nerd, he is very careful not to make any concrete claims. This idea is bolstered by an $18,000 purchase of Romney stock Tuesday morning. This single purchase caused Romney’s Intrade predicted chance of winning to rise from 41% to 49%.
As Silver says, there is no way to know what is behind this $18,000 bet. But we do know a few things. First, if the bettor had real information that indicates Romney is more likely to win than the models predict, he could have made far more money investing in regular stocks (e.g. coal). Second, the return on a Romney bet would have been substantially more if he had placed his wager on one of the other betting markets where Romney was going for 10 percentage points less. To me, this says that the bettor was either an idiot or a political operative.
Combine all of this information with what we know about how conservatives approach winning elections. It isn’t just gaming the Best Seller List. Republicans have a long and disgusting history of political dirty tricks. Think about it: would a party that suppresses votes in legal and illegal ways really shy away from this kind of behavior? Of course not. I think there is almost no question that Republican operatives are gaming Intrade.
Up to now, this kind of thing has worked pretty well for them. The mainstream media are reporting that Romney has “momentum”—a term that has no meaning in this context. (Nor is Romney surging in the polls, which I guess is what they mean to imply.) And all of this is coming from the Romney campaign claiming that they are, like Charlie Sheen, “Winning!” They are even claiming that they will win Nevada. So what is a poor reporter to do when he wants to back up this narrative? He can’t look at the polls: they’re a mess. He can’t look at the models: they all show Obama winning handily. But Intrade? The only betting market they know about? It supports the narrative!
The fact that the Romney campaign’s claims and the Intrade betting all come from the same guys reminds me of the lead up to the Iraq War. Cheney tells the New York Times that Iraq was WMDs. The New York Times reports it. Cheney goes on TV and quotes the New York Times. Luckily elections are a little more fact based than justifications for war. A little more.
 Silver notes that other models like his predict a substantially higher likelihood of an Obama win. So if anything, Silver’s model is tilted toward Romney.
 The Intrade prediction was way off before the $18,000 bet, so I suspect that Romney supports have been gaming it for a while.
Update (25 October 2012 9:31 am)
All good nerds think alike. Ezra Klein on the “momentum” question:
Update (25 October 2012 12:25 pm)
I hope it is obvious why I’m focusing on electoral “momentum.” If not: it is this meme that the Romney campaign is pushing because they are hoping for a bandwagon effect. Here is Jonathan Bernstein on The ‘Momentum’ Myth:
Update (25 October 2012 7:27 pm)
As you can see in the comment below from Intrade Exchange Operations Manager, Carl Wolfenden, the $18,000 bet was allegedly not from a single trader. (It’s not that I doubt him, but the truth is that I don’t know.) He claims it was a statistical fluke that happened when 40 different traders all bet on Romney when there were very few bets on Obama.
My thinking on this is that it doesn’t matter to the case that I’m making here. Intrade still needs to explain why it is that their market is so much more bullish on Romney than other markets. What’s more, those 40 traders could have been working together. It hardly matters: that spike in Romney’s numbers lasted a short period of time. But in the end, his numbers settled back to the usual unreasonable place. Frankly, if Republicans were trying to game Intrade, they wouldn’t do it at one time; they would do it constantly. I never meant to suggest anything else. I always figured the $18,000 trade was just part of a wider effort. How else could Romney’s chance of winning been 41% before the bet?