Jonah Lehrer writes about how bad NFL teams are at drafting talented players, particularly at the quarterback position.
Despite this advantage, however, sports teams are impressively amateurish when it comes to the science of human capital. Time and time again, they place huge bets on the wrong players. What makes these mistakes even more surprising is that teams have a big incentive to pick the right players, since a good QB (or pitcher or point guard) is often the difference between a middling team and a contender. (Not to mention, the player contracts are worth tens of millions of dollars.) In the ESPN article, I focus on quarterbacks, since the position is a perfect example of how teams make player selection errors when they focus on the wrong metrics of performance. And the reason teams do that is because they misunderstand the human mind.
He talks about a test that is given to college quarterbacks eligible for the NFL draft to test their ability to make good decisions on the field. Evidently this test is considered important by NFL scouts and indeed scores on this test are good predictors of whether and when a QB will be selected in the draft.
However,
Consider a recent study by economists David Berri and Rob Simmons. While they found that Wonderlic scores play a large role in determining when QBs are selected in the draft — the only equally important variables are height and the 40-yard dash — the metric proved all but useless in predicting performance. The only correlation the researchers could find suggested that higher Wonderlic scores actually led to slightly worse QB performance, at least during rookie years. In other words, intelligence (or, rather, measured intelligence), which has long been viewed as a prerequisite for playing QB, would seem to be a disadvantage for some guys. Although it’s true that signal-callers must grapple with staggering amounts of complexity, they don’t make sense of questions on an intelligence test the same way they make sense of the football field. The Wonderlic measures a specific kind of thought process, but the best QBs can’t think like that in the pocket. There isn’t time.
I have not read the Berri-Simmons paper but inferences like this raise alarm bells. For comparison, consider the following observation. Among NBA basketball players, height is a poor predictor of whether a player will be an All-Star. Therefore, height does not matter for success in basketball.
The problem is that, both in the case of IQ tests for QBs and height for NBA players, we are measuring performance conditional on being good enough to compete with the very best. We don’t have the data to compare the QBs who are drafted to the QBs who are not and how their IQ factors into the difference in performance.
The observable characteristic (IQ scores, height) is just one of many important characteristics, some of which are not quantifiable in data. Given that the player is selected into the elite, if his observable score is low we can infer that his unobservable scores must be very high to compensate. But if we omit those intangibles in the analysis, it will look like people with low scores are about as good as people with high scores and we would mistakenly conclude that they don’t matter.
13 comments
Comments feed for this article
April 10, 2011 at 11:58 pm
crack
You should do yourself a favor and avoid reading anymore of berri’s stuff.
April 11, 2011 at 2:25 am
Nick Rowe
Sample selection bias?
April 11, 2011 at 3:56 am
Anonymous coward
I think it is survivorship bias.
I see your point (and it is a good one), But I don’t think it completely invalidates the analysis, he’s just answering a slightly different question.
If it is a good predictor and the relationship is linear, shouldn’t the relationship hold even in a sub sample?
April 11, 2011 at 9:07 am
jeff
yes but dampened, and if there is any negative correlation between the two characteristics (plausible for IQ and physical ability) then it can reverse. for example, suppose that you have two scores x_1 and x_2 and the sum must be greater than 10 to be selected. And suppose x_1 and x_2 are independent.
In the overall population people with x_1 = 0 are worse than people with x_1 = 10 on average by [E(x_2) + 10 ] – E(x_2) = 10. In the selected population the difference is [E(x_2) + 10] – E(x_2 | x_2 > 10) which is smaller.
April 11, 2011 at 7:47 am
Rajiv Sethi
Jeff, I wouldn’t be so quick to dismiss this. In your post you say “scores on this test are good predictors of whether and when a QB will be selected in the draft.” The post you link to simply says when. It’s a key distinction: if almost all of those taking the test end up being drafted in some round then survivorship bias will be negligible, and the claim that the test is not predictive of performance (conditional on other observables) would be sound. And I have to say that the examples in the post you link to are quite compelling:
“Many of the most successful quarterbacks in NFL history reportedly had subpar Wonderlic results. Donovan McNabb scored a 14 and Brett Favre a 22, while Randall Cunningham, Dan Marino and Terry Bradshaw each scored 15. What’s more, several QBs who had unusually high marks — guys like Alex Smith and Matt Leinart, who scored 40 and 35, and were top-10 picks in their respective drafts — have struggled in the NFL, largely because they make poor decisions on the field. “People obsess over the stuff they can measure,” says former NFL quarterback and current ESPN analyst Tim Hasselbeck(Wonderlic score: 23).”
This also makes theoretical sense to me. I think that the relationship between performance on this test and on the field is probably nonlinear. At some point being too “book smart” can start to hurt you. Not just in football; I think this is true also for trading in financial markets.
April 11, 2011 at 7:58 am
Peter Klein
I’d call it sampling on the dependent variable. You see this in management research all the time. “We examined a sample of high-performance firms and found that size (or leverage or diversification or whatever) is unrelated to profitability.” Well, duh. Without knowing anything about firms of similar size etc. that failed, you can’t draw any inferences about the causal effect of those characteristics. The “Good to Great” genre of management literature commits this fallacy on a consistent basis.
April 11, 2011 at 10:47 am
ajkl
I think this fallacy is called “dynamic selection bias”, e.g in Heckman and Singer. There, you understate the effects of observables on staying in school because having “survived” previous schooling decisions induces a negative dependence between observables and unobservables conditional on making it so far.
April 11, 2011 at 11:30 am
Dan
I haven’t read this Wonderlic paper, but previous research by Berri & Simmons on drafting quarterbacks has suffered from a massive case of selection bias. Berri previously claimed that there was virtually no relationship between when a quarterback was drafted and how well he performed, and argued that this shows that NFL teams do a terrible job of drafting quarterbacks. But he only compared quarterbacks who were drafted and got enough playing time to have meaningful data, making no attempt to correct for the large selection bias created by the fact that decisions about which quarterbacks get playing time are not made by random assignment.
It’s true that quarterbacks who are drafted in the 6th round and then get significant playing time end up playing about as well, on average, as quarterbacks who are drafted in the 1st round and then get significant playing time. But that’s comparing most 1st rounders (since most 1st rounders do get to start a few games) with only a small fraction of 6th rounders (since only a small fraction of 6th rounders ever get to start a few games). A much smaller proportion of quarterbacks drafted in the 6th round turn into successful NFL quarterbacks, compared to quarterbacks drafted in the 1st round.
April 11, 2011 at 11:58 pm
Charles Talleyrand
The comment about height and propensity for All-Star selection is misguided. The NBA selects a fixed number of guards, forwards, and centers. In practice, this means it selects a fixed number for tall, very tall, and extremely tall players.
If you follow the NBA, you might find this explanation over-simplified. But I believe the point is clear.
April 12, 2011 at 12:09 pm
SirWellington
Atomistic Fallacy. Could be ecological fallacy, which would be the opposite depending on how you look at it/ how the recruiters came to their conclusions. It’s common when dealing with data relating to humans. Human characteristics don’t match data very well.
“The fallacy one commits when making inferences about groups or aggregates from individuals (see Ecological Fallacy).”
April 12, 2011 at 12:37 pm
Michael
I’m not sure I follow that having a low observed score implies a high unobserved score. By definition under classical test theory, X (observed score) = T (unobserved or true score) E (error value can be /-).
What you can infer are other, possibly observable, factors compensating (e.g., letters of recommendation compensating for low GRE scores).
April 20, 2011 at 4:44 pm
Links 4/20/11 [Mike the Mad Biologist] | iPhone 2 die 4
[…] on the mega lab Bats are worth billions What Is This Fallacy Called? Hold Onto Your Floppy Disks, […]
April 11, 2012 at 9:00 pm
Dan
The fallacy is correlation proves causation.