| Sun | Mon | Tue | Wed | Thu | Fri | Sat |
|---|---|---|---|---|---|---|
| 1 | 2 | |||||
| 3 | 4 | 5 | 6 | 7 | 8 | 9 |
| 10 | 11 | 12 | 13 | 14 | 15 | 16 |
| 17 | 18 | 19 | 20 | 21 | 22 | 23 |
| 24 | 25 | 26 | 27 | 28 | 29 | 30 |
| 31 |
Authors' Committee
Chair:
Andy Eggers (Gov)
Members:
Weihua An (Soc)Kevin Bartz (Stats)
Sebastian Bauhoff (HealthPol)
John Graves (HealthPol)
Justin Grimmer (Gov)
Jens Hainmueller (Gov)
Mike Kellermann (Gov)
Ellie Powell (Gov)
Gary King (Gov)
Weekly Research Workshop Sponsors
Alberto Abadie, Lee Fleming, Adam Glynn, Guido Imbens, Gary King, Kevin Quinn, Jamie Robins, Don Rubin, Chris Winship
Recent Comments
Recent Entries
- Gans and Leigh on the "Baby Bump"
- Exxon-tainted research?
- Awards for IQSS faculty
- The Tree-Friendly Academic, Part II: The Editing Process, and Getting Off the Monitor
- The Tree-Friendly Academic: Whither A Useful Free PDF Editor?
- Nicholas and James are Featured in the NYT again
- Matching Portfolios
- Harvard Program on Survey Research (on Youtube)
- Placebo effects and the probability of assignment to active treatment
- IN, NC Rehash; WV Prediction
Categories
Blogroll
Brad DeLongCognitive Daily
Complexity & Social Networks
Developing Intelligence
EconLog
The Education Wonks
Empirical Legal Studies
Free Exchange
Freakonomics
Health Care Economist
Junk Charts
Language Log
Law & Econ Prof Blog
Machine Learning (Theory)
Marginal Revolution
Mixing Memory
Mystery Pollster
New Economist
Political Arithmetik
Political Science Methods
Pure Pedantry
Science & Law Blog
Simon Jackman
Social Science++
Statistical modeling, causal inference, and social science
Archives
- August 2008
- June 2008
- May 2008
- April 2008
- March 2008
- February 2008
- January 2008
- December 2007
- November 2007
- October 2007
- September 2007
- June 2007
- May 2007
- April 2007
- March 2007
- February 2007
- January 2007
- December 2006
- November 2006
- October 2006
- September 2006
- August 2006
- July 2006
- June 2006
- May 2006
- April 2006
- March 2006
- February 2006
- January 2006
- December 2005
- November 2005
- October 2005
- September 2005
Notification
« March Madness | Main | Applied Statistics - Ken Kleinman »
18 March 2007
Three-way ties and Jeopardy: Or, Drew questions the odds
It's been in the news that a three-way tie happened on Jeopardy on Friday night. From the AP article:
The show contacted a mathematician who calculated the odds of such a three-way tie happening — one in 25 million.
I have to believe that the mathematician contacted didn't have all the facts (and the AP rushed to meet deadline), because once you're in Final Jeopardy there's little randomness about it. It's all down to game theory.
Suppose we first estimate the odds that all three players are tied at the end of Double Jeopardy.The total dollar value shared by all three is around $30000, or about $10000 each. Since questions have dollar values which are multiples of $200, we could reasonably assume that there are 100 dollar values, between 0 and 20000, where each player can end up. So the odds of a tie at this stage should be no more than one in a million - and this is a very conservative guess, since I assume that the probabilities are all equal (whereas they would likely have a central mode around 10000.)
Breaking a three way tie with a Final Jeopardy question would then require that all three players bet the same amount, and I think the odds are considerably less than 1 in 20 that they'd all bet the farm no matter the category.
But it shouldn't even get that far. The scenario on Friday night had two players tied behind the leader who didn't have a runaway. So we have somewhere around 1 in 20,000 odds that this would happen (the factor of two because the third player could be ahead or behind the tied players.)
The runners-up would both be highly likely to bet everything in order to get past the leader. And the leader, in this case, placed a tying bet for great strategic reasons - getting one more day against known opposition rather than taking the chance of a new superstar appearing the next day - as well as a true demonstration of giving away someone else's money to appear magnanimous.
Even if the leader only had a 10% chance of making that call, and given that the other two players were pressured to bet high, that's still 1 in 200,000 - over 100 times more likely with a fairly conservative estimation process.
Posted by Andrew C. Thomas at March 18, 2007 11:14 PM