The following is a guest post by @sunset_shazz.
The Philadelphia Eagles currently have 8 former members of the Oregon Ducks on their roster at large:
Brandon Bair, Taylor Hart, Josh Huff, Jeff Maehl and Casey Mathews on the 53-man, Wade Keliikipi and Will Murphy on the practice squad and Kenjon Barner on injured reserve.
This concentration of Ducks on former Oregon head coach Chip Kelly’s NFL team has been a source of much hand-wringing. The accusations of Duck Bias from fans and the media have reached sufficient volume for General Manager Howie Roseman to publicly deny any bias in the evaluation process:
"He lets us make our own evaluations and talk to him about it, and then he gives us his opinions," Roseman said. "He doesn’t do anything different than he does for any player from any other school. It’s really not fair that he gets hit on some of this stuff when the guys that we’re keeping, we’re keeping because they’re good players and they can contribute to our football team. That’s for all of us.
"He wants the best players. He wants to win games. He’s very selfish in that regard. I don’t think it’s fair that he gets accused of anything other than that. To me, the question is: Does Coach want anything other than to win football games and to have the best players on the team? I think there’s no question to anyone that’s around him that that’s his sole function and his sole desire."
Other Chip Kelly acolytes dismiss accusations of Duck bias as the result of “hatred”:
Today, we will attempt to answer two questions: Is there an Eagles Duck bias? And should we care?
What is Bias?
Part of the problem with this issue is that the word “bias” subsumes a portfolio of different effects, described in fields as varying as psychology, economics, statistics and decision theory. Daniel Kahneman and Amos Tversky described psychological biases over 40 years ago, resulting in a Nobel Prize for Economics. Nate Silver has noted that various statistical biases in selection and sampling affect areas such as political polling. Economists have long held that certain biases can be rational.
We have chosen a purely quantitative method to estimate the degree of bias within the Eagles player evaluation process. We examined the NFL roster data compiled by ourlads.com (all data as of 9:30 PM EST August 31, 2014). There are presently 32 former Ducks on NFL rosters: 14 NFL teams have no Oregon alumni; 12 have one Duck on their rosters; 3 teams (the Bears, Broncos and Niners) have two; 2 teams (the Panthers and Giants) have three, and, as everyone by now is acutely aware, the Eagles have eight. Sharp readers will note that the Eagles have exactly one quarter of all NFL-caliber Ducks on their team.
Including members of 53-man rosters, practice squad players who have been identified, players on Injured Reserve, the Physically Unable to Perform list, and suspended players, we estimate that 1.50% of NFL players are Oregon Ducks. Importantly, the Eagles (12.3% Ducks) skew the average significantly – excluding the Birds, Ducks make up only 1.16% of NFL rosters.
We asked ourselves, given the pool of NFL players currently on rosters league-wide, what are the odds that, by chance, a given team would have at least 8 Ducks on its overall roster? We used the cumulative distribution function of the binomial distribution; this is a statistical technique that allows one to estimate the odds of achieving a given result over time, assuming the underlying probability is well understood. For example, the binomial distribution tells us the odds of flipping heads on a fair coin 3 times in a row is 7:1 against (0.125 probability).
By looking at the total proportion of Ducks on NFL rosters (1.5%), we infer that this is the “base rate”, or statistical likelihood, that any given player in the NFL is an Oregon Duck. Then, by using the binomial distribution, we can compute the likelihood that, if one was ignoring the players’ alma maters, one would get 8 (or more) Oregon Ducks on the Eagles' 67-man roster at large. The answer: 0.000753%, or 132,724:1. For comparison, the odds of being struck by lightning any time in one’s lifetime are estimated at 3,000:1. Howie Roseman doth protest too much.
Does it Matter?
Our friend Sam Lynch points out that none of the Ducks in Philly are likely to achieve any meaningful playing time. And far from being “haters”, we are avid fans of Chip Kelly and his evidence-based approach to roster construction, player development, pace, play-calling and #sportscience.
There are plenty of good reasons why Chip could have a “rational bias” for Duck players, particularly for those players toward the bottom of the roster (at “replacement level”). NFL roster construction is, after all, an uncertain endeavor, full of wide confidence intervals surrounding the various dimensions of player evaluation — talent, work ethic, coachability, etc. To the extent that Chip has detailed, intimate knowledge of his former players, some of these “known unknowns” can be resolved, and for players with near-equivalent evaluations, the “Duck knowledge” could prove decisive.
[Brian's Note: The "perfect information" Chip may have in regard to many of his former players is a benefit. But that also means he should understand the limited upside of certain players. He stands to reap larger rewards from spending a roster or practice squad spot on someone with potential whom he and his staff can evaluate and coach up. Higher ceiling vs. higher floor.]
If Chip and Howie were to defend their approach as rational decision-making under uncertain conditions, we would be inclined to agree — or at least listen. But to suggest that there is no bias at all beggars belief.
@sunset_shazz is a Philadelphia Eagles fan who lives in Marin County, California. He is a huge fan of Chip Kelly, but isn’t fond of Kool-Aid.