Saturday, November 21, 2009

Can it smell baby poo?

Statfight: Draft Position and Per-Play QB Performance

A few of my students frequently want sports examples about statistics. Here is a good one.

Survey Design Bias

Another example of biased survey design (here is a link to the full survey results).
Ninety-four percent of Obama voters correctly identified Palin as the candidate with a pregnant teenage daughter, 86% correctly identified Palin as the candidate associated with a $150,000 wardrobe purchased by her political party, and 81% chose McCain as the candidate who was unable to identify the number of houses he owned. When asked which candidate said they could "see Russia from their house," 87% chose Palin, although the quote actually is attributed to Saturday Night Live's Tina Fey during her portrayal of Palin during the campaign. An answer of "none" or "Palin" was counted as a correct answer on the test, given that the statement was associated with a characterization of Palin.
Obama voters did not fare nearly as well overall when asked to answer questions about statements or stories associated with Obama or Biden -- 83% failed to correctly answer that Obama had won his first election by getting all of his opponents removed from the ballot, and 88% did not correctly associate Obama with his statement that his energy policies would likely bankrupt the coal industry. Most (56%) were also not able to correctly answer that Obama started his political career at the home of two former members of the Weather Underground.
I'm thinking if Zogby had asked questions about Obama that were of a similar nature to the questions about McCain/Palin, they would have gotten more of their negatives correct. For example, "which candidate has been tied to radical preacher Jeremiah Wright?" or "which candidate has a muslim half-brother?" probably would have gotten a higher response rate, and in similar proportion to the McCain questions. On the other hand asking about McCain, "which candidate said that auto industry jobs aren't coming back?" would have gotten a lower response rate.
I also wonder what the survey results look like controlling for education and income. Democratic voters tend to be poorer and less educated (and not because liberal policies are dumb - they are just more compassionate towards people who are less well-off). That didn't make the report either, but it was part of the survey.

Thursday, November 19, 2009

Remember This?

One day, I think, Bill Bellichick will be respected for making a bold and well-calculated decision, even if the game ends up impacting the Pats' run this year.

Monday, November 16, 2009

Hail, Bellichek

I have never liked Bill Belichick... Until last night. I wasn't sure what the numbers on last night's failed attempt on 4th and 2 are, but I had read this paper by David Romer, which basically says that teams punt WAY less than the optimal amount. Then, there's this analysis by Advanced NFL Stats:
With 2:00 left and the Colts with only one timeout, a successful conversion wins the game for all practical purposes. A 4th and 2 conversion would be successful 60% of the time. Historically, in a situation with 2:00 left and needing a TD to either win or tie, teams get the TD 53% of the time from that field position. The total WP for the 4th down conversion attempt would therefore be:

(0.60 * 1) + (0.40 * (1-0.53)) = 0.79 WP

A punt from the 28 typically nets 38 yards, starting the Colts at their own 34. Teams historically get the TD 30% of the time in that situation. So the punt gives the Pats about a 0.70 WP.

Surely, Belichick would not have been criticized if he had punted and the Colts had driven down the field, but based on historical stats, he made the right choice. That would have been the conventional approach, and would not have been controversial.
P.S. Matt Millen said "well, those stats don't take into account the fact that the Colts have a better-than-average quarterback." True, but they also don't take into account that the Pats do too, not to mention the fact that, if having Peyton Manning increases the probability of a Colts touchdown on a 29 yard field, it also increases the probability of a Colts touchdown on a 70 yard field. To me, that's close to a wash.