Site Meter

On Brains and Football

Dave Hoffman

Dave Hoffman is the Murray Shusterman Professor of Transactional and Business Law at Temple Law School. He specializes in law and psychology, contracts, and quantitative analysis of civil procedure. He currently teaches contracts, civil procedure, corporations, and law and economics.

You may also like...

8 Responses

  1. Sarah Lawsky says:

    Your criteria seem to imply that there could never be a good visual representation of a bad underlying study, and your post criticizes the study, not the graph. As you analyze it, this graphic is misleading only because it is based on a misleading study. In fact, based on your criteria, any “good” (eye-catching, simple, accurate, whatever) graphic of a bad study will constitute a bad graphic, because it will make clear and communicate well the underlying data.

    I believe Tufte’s suggestions and criticism are more along the lines of what Andrew Gelman regularly does on his blog ( — see the posts in the category “Statistical Graphics”), in which the analysis of the graphic relies in no way on the quality of the study, but only on how the graphic communicates the study.

  2. A.J. Sutter says:

    Sarah read my mind before I had even finished reading your piece (or begun it, for that matter). I second her comment.

  3. Dave Hoffman says:

    Sarah and AJ,
    I think you’ve misread me – or I’ve been unclear.

    The graph is bad because the data are bad.

    But it’s also bad because it doesn’t convey the uncertainty that we ought to have about the data (i.e., it makes us think that we can be confident that O-Linemen have a wonderlic of X, but we don’t know what the variance between the population is, or whether differences in the sizes of the bubbles are statistically significant). In that way, the graph makes the data (arguably) worse.

    You could have a good graph of bad data – it would make clear for the reader how uncertain and contingent the data are. And you can have a bad graph of good data — the traditional tufte problem.

  4. Managing Board says:

    I may be misremembering Zimmerman’s argument from when I read that book about 15 years ago, but I think he says offensive linemen have to be smart because they play every single offensive down and thus have to know the team’s entire repertoire of blocking schemes. These can be quite complex and you have to know what the other linemen are doing. The backs and ends just have to know their particular plays, which are some variation on “run this way and do your thing.” So even if the offensive players are basically running scripts–and most of the time they are–the linemen have to know ten times as many scripts as the backs and ends.

    Teams could rotate offensive linemen in and out like they do backs and ends. But teams don’t do this, apparently because it’s useful to have the same five guys play as many downs together as possible. The other offensive player who plays every possible down is the quarterback, who also has a high score.

    So it’s no so much a “close to the ball” theory as one based on expected playing time and the sheer number of plays that have to be memorized and executed.

  5. Dave Hoffman says:

    Yes, I’ve heard that argument. But the wonderlic doesn’t really measure memory, it measures (if anything) how fast you can make relatively easy decisions.

  6. Adam Benforado says:

    As Dave suggested, I think it’s important to distinguish what the Wonderlic measures and what is needed to be a successful football player. It makes intuitive sense that there are certain “thinking” positions in a game like football and that people who are better “thinkers” (measured by how well they do on timed written problem-solving tests) will do better in these positions. Unfortunately, that intuition does not stand up to scrutiny.

    Jonah Lehrer has a nice summary of the latest research in the mind sciences about game performance in his book How We Decide as well as on his blog.

    Here is an excerpt from one of Lehrer’s blog posts on the topic (

    “Why is the Wonderlic so useless at predicting which quarterbacks will succeed? The reason is that finding the open man during an NFL game involves a very different set of decision-making skills than solving a problem on an IQ test. While quarterbacks need to grapple with complexity – the typical offensive playbook is several inches thick – they don’t make sense of the football field the way they make sense of questions on a multiple-choice exam. The Wonderlic measures a specific kind of thought process, but the best quarterbacks don’t think in the pocket. There isn’t time.”

    “So how do quarterbacks do it? How do they make a decision? It’s like asking a baseball player why he decided to swing the bat: the velocity of the game makes thought impossible. What recent research in neuroscience suggests is that quarterbacks choose where to throw the ball by relying on their unconscious brain. Just as a baseball player will decide to swing at a pitch for reasons he can’t explain (he’ is acting on subliminal cues from the hand of the pitcher), an experienced quarterback picks up defensive details he’s not even aware of. Although he doesn’t consciously perceive the lurking cornerback, or the blitzing linebacker, the quarterback’s unconscious is still able to monitor the movement of these players. And then, when he glances at his receivers, his brain automatically converts these details into a set of fast emotional signals, so that a receiver in tight coverage gets associated with a twinge of fear, while an open man triggers a burst of positive feeling. It’s these inarticulate emotions, and not some elaborate set of rational calculations, that tell the best quarterbacks when to let the ball fly. The pocket, it turns out, is too dangerous a place to think.”

  7. dave hoffman says:


    It strikes me that the lack of correlation between the wonderlic and QB success doesn’t necessarily mean that QBs don’t need decision making skills (or the bigger claim that their decisions are subconscious – a “situation sense” in the pocket). That’s a plausible interpretation, but it’s not required by the data. Rather, variance in outcomes just means that it is quite hard to predict QB success based on individual, as opposed to team and systemic, characteristics.

  8. Adam Benforado says:

    Hey Dave, I agree that the lack of a correlation alone doesn’t tell you much; I think it’s other research on the prefrontal cortex and subconscious processes that help make Lehrer’s assertions more robust. On a somewhat different note, it’s interesting to think about whether the Wonderlic might be put to better use with respect to coaches. As a Redskins fan, I’ve wondered all season about coach Jim Zorn’s decision-making skills at critical moments . . . Could the Wonderlic have saved the team from last night’s beating at the hands of the Giants?