When QBs decide to take three knees!


Statistical misdirection: Lamar Jackson is off to a good start this year. He's off to a good start both running and passing the ball, but also in terms of trying to earn too much money.

Yesterday, how good was his running? With one minute left in the game, Jackson had run the ball eight times for a total of 110 yards. 

That was 13.8 yards per rushing attempt, a rather gaudy number.

One minute later, the whistle blew—and his numbers looked substantially different. As you can see, this is Jackson's official line for yesterday's rushing performance:

11 rushes, 107 yards
9.7 yards per rushing attempt

What had happened to Jackson's performance? Simple! With Baltimore comfortably in the lead, he "took a knee" at the end of the game, thereby running out the clock.

In fact, he took a knee on three consecutive plays—and under NFL statistical procedures, taking a knee is counted as a rushing attempt. The quarterback is charged with a rushing attempt, and he's charged with losing a yard, or sometimes with losing two.

Should "taking a knee" be counted as a rushing attempt? We would say that it pretty much shouldn't.

For starters, it obviously isn't a rushing attempt. It also isn't a rushing attempt when a quarterback gets sacked for a loss of yards—but sacks, and the yardage lost, are treated as a separate statistical category. It seems to us that "taking a knee" should be recorded separately too.

Our question:

Did Jackson run for 13.8 yards per rushing attempt, or was it just 9.7? As you can see, the difference is large.

In this most fully informed of all possible worlds, what should the record books tell us?

For professional skeptics only: You can track the eleven rushes here.  That includes the three "rushing attempts"-that-weren't right at the end of the contest.


  1. Somerby is not asking a technical question. He is talking about what the stat should measure and how it should be interpreted.

    1. This depends on the purposes of the measurement which Somerby does not discuss.

  2. Point:

    You're not getting a truly accurate representation of a QB's rushing ability by counting kneels.


    Stats are supposed to be a totally mathematical/quantified representation of what happened during plays. With no exceptions. Making exceptions for some plays is imposing subjectivity and qualitative aspects of the game onto that mathematical representation, which defeats the purpose of the stats in the first place.

    1. The stats have to add up. It's an accounting requirement.

    2. It is too early in the season for these stats to mean much. With more data points, the inclusion of an aberrant measurement will have less impact on the average than it does at the beginning of the season. The assumption may be that such knee-taking may occur at some point for each of the people being measured, is a random event, and thus is noise. The more data is collected, the less such events will matter. So, a career average will be more stable than one based on only 2 games at the beginning of a season. So this is nothing to get excited about and not a reason to change what is included. I agree with the comment about not introducing subjectivity into what the numbers are measuring.

      This is why you are not supposed to go through a data set and exclude low values based on their not being a true representation of a player's ability or some such nonsense.

  3. Great analysis of the political press
    here. Is Bob trying to show us he’s
    a regular guy since he’s into sports?
    He’s ready for Morning Joe!!

  4. Nichole Wallace is crushing it in the ratings, demolishing
    the five. Also, Fair to NOTE, soundly beating all the other
    MSNBC shows. This certainly indicates that America is
    concerned with accountability for Trump's rape of the

    1. Her Mother must be very proud of her.

    2. We who can view Trump as the menace he is are.

  5. ____________________________________

  6. Somerby doubles down on meritocracy.

    What a sad lost soul he is.

    Where's that confounded moral compass?

    Monetizing sports is a weird (and harmful) way to organize a society; sports should be about friends and family having fun in the backyard - you might as well start monetizing eating, breathing, and pooping, too! Somerby is such an asshole, he'd be drafted in the first round.

  7. Taking a knee could be easily counted in a separate category, but what about walking around slowly or just standing around, to run out the clock? What about less obvious but equally time-consuming acts? Do those each get separate categories?

    And how does one guess the intention of the team by watching what is going on? Is it always clear that the intent is to run out the clock or might there be some confusion or error that should be counted toward a rushing play? The more this becomes part of observer judgment, the more problematic having such a separate category will be.

    Does Somerby believe that no one has discussed these possibilities and issues before? It can be assumed there are reasons for the way things are currently done. People who appoint themselves critics of the world frequently spend little time guessing what the reasons for the status quo might be. They tend to assume others are stupid and have never thought about what seems obvious to them. But they may be overlooking things that are obvious to others, such as what do you do with a separate category of knee-taking. Would anyone consider it important enough to analyze? Yet energy would have to be expended in counting and reporting it. It would fatten the pages of the record-book, expanding numbers, and making statistics more complex, but to what purpose? Because Somerby thinks it is unfair to count knee-taking as rushing? Is that a good enough reason? And what if the purpose of counting it that way is to discourage it from being used as a tactic. Penalizing players stats would tend to discourage it.