Monday, February 07, 2005

The Wisdom of Krauts

The Super Bowl ranks slightly above a Wayne Newton Christmas special on my list of entertainment choices, but far below all the other things I could think of doing last night, including editing PowerPoint slides for a pro bono project on tax reform. Which is what I was doing when I received an email from Jim, an old college friend who was asking me to send him my prediction of the Super Bowl champion before the game started.

No, this wasn't a football pool. Jim's a epidemiologist, and he was "having a little fun with our anonymous internet Longitudinal Survey Engine (aiLSE), testing something called Prelec's Bayesian Truth Serum (BTS)."

Based on my limited exposure to epidemiology, it's a great career track for someone who's interested in the drug culture but is too smart to become a user, a dealer or a cop.

He went on to explain: "We developed the aiLSE to help us in longitudinal research so that users can remain anonymous and report sensitive stuff. Someone in our MSU Psychiatry department works with the frail elderly, got concerned about evidence of elder abuse, and now is using it in Michigan as an anonymous hot-line so that nursing home staff can report elder abuse
without fear of reprisal. It really cuts down on our research costs, and adds a level of anonymity that can't be beat."

If you're interested how the "truth serum" is designed to work, read the article linked above. Suffice it to say, the method was not devised for handicapping the Super Bowl. (Only one respondent out of 20 picked the Eagles to win.) Rather, it's designed to ferret truth out of situations in which respondents might be tempted to lie — or obscure their real opinion.

Does BTS have application to political polling? Unlike polls, which ask voters (or potential voters) how particular issues sway them, BTS would throw in another layer of questions about how the respondent believes others' votes will be influenced by the same issue. Jim is starting to get there with his next experiment:

"Right now we are working on the Saturday Spartans-Wolverines basketball game, with 20 Spartan partisans and 20 Wolverine partisans, plus anyone else who weighs in.

"I'm interested to see if the truth serum works no matter what the skew of the sample. Prelec claims that it works, because partisanship shows up in the difference between one's own probability score and one's prediction about how the rest of the world will vote. If so, the Spartan bias will be counterbalanced by the Wolverine bias (in theory)."

But how do dominant political parties get similar counterbalancing? They certainly don't get it from within their own ranks, unless they take pains to find it.

(All you epidemiologists out there, be sure to chime in with comments and clarifications. It's time for me to segue to my dilettante's point.)

Which was going to be what, exactly?

Well, I'm also simultaneously engaged in four different projects — a strategic plan, a branding consultation, a community engagement process and a tax reform proposal — that all involve gathering input and insight from a diverse group of people. Not just the stakeholders and true believers, but from the bystanders and potential opponents. It's not scientific, but in each case, we recognize the value of hearing different perspectives and contradictory voices. In fact, we seek them out, because we know we'll reach a better result.

The political process often makes a show of collecting the peoples' views, but then uses the information selectively to batter the opposition. That's not the way all these other organizations go about solving important problems.

Jim's email also got me thinking about two books that have made it into my reading stack, but not yet into my lap: Malcom Gladwell's Blink: The Power of Thinking Without Thinking and James Surowiecki's The Wisdom of Crowds. How might we apply the concepts of "rapid cognition" and group decision-making (think Apollo 13) to the political realm, I wondered?

As I looked for links on these two books, I found Surowiecki's and Gladwell's exchange in the Slate Book Club, which will continue throughout the week.

Can you spell Gestalt?

I'll save my thoughts for another post. Meanwhile, dip in.

2 Comments:

Blogger bob said...

I got the same email, but I didn't see it until the day after.
And obviously anyone who thought Michigan had a snowball's chance in hell, is a major Wolverine partisan.

Krauts?

5:10 PM  
Blogger bob said...

I checked out the Blink link.

When I was a bartender I used to say that I could see trouble coming through the door. I seemed to have a very accurate sixth sense about people who were going to cause problems. And that was in the seventies when everyone looked wierd!

5:18 PM  

Post a Comment

Links to this post:

Create a Link

<< Home