It’s main point: prediction markets are more accurate than polls; the farther away from an election the ‘more’ more accurate they are.

And for those who are skeptical about polling, ask yourself this — if the polls are so inaccurate, then why do campaigns at all levels spend so much money having them done?

I don’t think anyone here is arguing that polls are generally inaccurate. Polls can be highly accurate and very useful if they are representative of the population of interest. What people here are wondering about is whether traditional phone polling is representative of the population of interest. I submit that land-line phone polling of “likely Republican voters” is not representative of those who will vote in primaries because it does not include those who a) do not use landlines, b) do not answer unknown numbers, c) do not stick around to answer pollsters’ questions when the phone is picked up, and d) are not considered to be a “likely Republican voter”.

Just listen to kids argue about who’s faster, or smarter, or stronger. They all talk big until someone says lets put some money on it, and then all the charlatans shut up.

The Intrade predictions market is good for measuring momentum at the moment. The reason why Ron Paul is doing so well lately is because he has been some positive media coverage with media appearances and debate performance.

I would not use this to predict the outcome of the Republican nomination until about a week or so before the Iowa Caucuses.

I think you’ll find that it’s often used incorrectly and in the wrong situations. An example of this would be polling the general public 12-18 months ahead of an election that only a very small subset are paying close attention to.

At least with Intrade, you know the sample group is a group that is self-selected to be the ones paying close attention to the situation. That doesn’t free them of bias, of course, but it’s a very useful data point.

I’m not saying that we should automatically accept the InTrade poll because it’s new, but I am saying that you’ve no real cause to automatically dismiss it simply because it’s not what everybody else uses. Innovation often flies in the face of conventional wisdom and proves it wrong…InTrade may very well be inaccurate, but the premise behind it (financial self-interest) should at the very least earn it some benefit of the doubt because the people who follow it have a vested interest in insuring its accuracy. If it turns out to be a load of bunk, I’ll be happy to write it off as just an example of wishful thinking, but until then I think it merits some consideration.

“I think you’ll find that, for the most part, political polling does a pretty good job when its used for what its intended.”

I’ve also found that, for the most part, polls often tend to come up short and are highly susceptible to the bias of the people creating and compiling them. They’re not nearly accurate enough for me to accept them as Bible or to cause me to close my mind to other possibilities that may hold more promise.

Such tests exist… For example, this paper appears to be exactly what you’re looking for.

I might have to buy this one…

I think you’ll find that, for the most part, political polling does a pretty good job when its used for what its intended.

Your argument isn’t an argument in support of polls, Doug, it’s an argument against finding more effective ways to analyze and map support, which is frankly a rather Luddite position for you to take.

Again, has anyone actually tested the predictive ability of these betting websites vs. opinion polling in a real-world situation ?

Your average Republican voter is a middle-class white man working full-time. He’s older than 30. He’s probably married and has children. He’s never heard of MeetUp and probably never YouTube, either. He spends little to no time on the internet.

And, of course, straw polls are basically meaningless. My home state, NJ, had about 240,000 Republican voters in the 2000 primary, despite the fact that Bush had already locked up the nomination by then. The recent Jersey-wide GOP straw poll attracted about 200 attendents paying $18 each and meeting in a central location. Primary voting, OTOH, is easy and free.