I have seen several tweets over the last week about an NFL team fumbling too much and the steps the coach took to remedy the problem. I rarely watch football and do not follow it at all, and the comments on the coaching strategies were equally uninteresting, simply boasting of the importance of ecological psychology. I have no idea if the team improved or not this week.
We imagine coaching to be an active profession: We expect coaches to do something. When a team or player has a weakness, we expect the coach to address and fix this weakness. To a certain extent, we view this as the primary role of a coach; if the coach cannot fix this weakness, what is he doing? Why are we paying him?
The NFL team broke a record through two games. Almost inevitably, through nothing but probability, their performance was likely to improve in the next game. Small sample sizes are unreliable for that reason; there is not sufficient time for luck and other variables to even out. One or two games are performances; they do not indicate permanence.
Imagine a 70% free-throw shooter. Through the first two games of the new season, he makes only one of his first 10 free throws. Maybe this breaks a record for futility. Is he suddenly a 10% shooter? Did he forget how to shoot during the offseason?
Like the NFL team, the coach of the 1/10 free-throw shooter will feel compelled to do something to fix the player’s free-throw shooting. Standing around not doing or saying anything is viewed as incompetence. We want action from our coaches. Typically, the coach demands more practice, more repetitions, extra free-throw attempts for the player or the team after practice often with some consequence (running) for missed shots.
Of course, in the next game, the player, based solely on probabilities, likely will shoot better.
Coaches have two possible responses when players or teams struggle: First, they can focus intently on this problem, devoting more and more time and concentration to fixing the issue; second, they can carry on with their normal practices and instructions.
Again, based on probabilities, the performance is likely to improve either way. The 70% free-throw shooter is unlikely to continue shooting 10%, just as the fumble-prone team is unlikely to continue losing so many fumbles.
Now, when the coach acts, changing practice to fix the problem, we attribute the improvement to the coaching, to whatever actions the coach took. When the coach ignores the problem and continues with normal practices, we invent explanations: The players concentrated more or played tougher or whatever fits our mental models for players’ improvements. We struggle to attribute short-term performance to luck and other uncontrolled variables; regression to the mean is not a satisfying explanation for a performance improvement even if the improvements amount to a good hop bouncing favorably as opposed to the opponent.
When the coach acts and the player shoots extra free throws, running for missed shots, we have evidence extra practice and running works. Of course, as I wrote last week, without the control group, we do not know if it was the extra practice and running or simple regression to the mean. However, because the extra practice was designed to improve performance, and performance improved, we attribute the improvement to the extra practice.
Many coaching behaviors gain credibility in this manner, whether within a person or throughout society. A team plays poorly in the first half, and the coach yells and screams at half time. The team plays better in the second half. See, yelling and screaming work.
Our coaching behaviors should not be based on small sample sizes. Yelling and screaming every time a team plays poorly is unsustainable. For a good team that has one bad game a season, the tactic may work. However, for a team that is more up and down, the yelling loses its effect; short-term performance may improve the first time, but each time the coach loses it, the effect lessens until it becomes the coach’s expected, normal behavior, and players tune out the coach.
Similarly, if the extra free-throw practice works, and the player shoots 8/8 next game, moving to 9/18 for the season, is additional extra practice required to reach 70%? How much extra practice? What if the improvement is less drastic, maybe from 1/10 to 4/8? The extra practice worked, as the player improved, but he still is shooting far below his usual shooting percentage. Is more intervention required?
What if, instead of extra practice, the coach fixed the player’s shot through some instructions or modifications of the player’s technique? The player improves slightly, from 1/10 to 4/8, but what if the improvements stop? What if his free-throw shooting hovers around 50% after 10, 15 or 25 games? The change worked, right? He did not continue to shoot 10%. However, he shoots 20% worse than expected. What if the changes caused the decreased shooting performance, either due to an inappropriate physical change or psychologically from a lack of confidence due to the changes?
The coach may be buoyed by his ability to improve players’ shooting and believe in the instructions and drills that caused this 40% improvement from 10% to 50%, but the reality is a 20% decrement from 70% to 50%. The initial two bad games easily could have been a small sample size and improvements expected as the player regressed positively toward the mean. However, whether due to hubris, unawareness of the players’ previous shooting success, or a lack of appreciation for sample sizes, the coach believes he improved the player’s shooting, and moves on to other shooters, attempting to replicate his success with 60%, 70% or 80% shooters because, as they say, players can always improve their shooting.
Even more unfortunate, if the other players decrease their shooting percentages, moving closer to 50% like the original shooter, the coach is unlikely to be swayed due to the anchoring bias: Our first impression (the intervention works) creates an anchoring point that can incorrectly influence subsequent observations. The performance decrements by the other players are not evidence the coach’s intervention does not work because clearly the initial evidence was his intervention worked. Therefore, the coach questions the players; they are not listening or trying hard enough. They need more practice, more repetitions. They need greater consequences for missed shots. They need to lean in to the intervention even more.
When this occurs with an otherwise successful (i.e. winning) coach, the interventions are passed to other coaches. The coach may win due to talent or defense or post play or some other aspects of the game, but we associate the winning coach with knowledge of all subjects. Of course, reading this, I imagine most shake their heads or do not agree. However, in the last year, hundreds if not thousands of coaches liked and retweeted the shooting advice and instructions of college coaches whose teams shot 26% from the three-point line and 59% from the free-throw line and 29% from the three-point line and 71% from the free-throw line respectively despite recruiting every player on their teams. We live in a post-fact world. We listen to the loudest voices, not the most competent. We learn from vibes, not specific documented successes.
My friend Oscar, the first coach I assisted when I started my AAU coaching career, said his first rule was do no harm (hippocratic oath). Unfortunately, our expectation for coaches is to do; we want action, even when inaction is the best course. We expect coaches to take timeouts, to yell at players, to instruct, to draw up a new play, to do things. In our rush to action, we often act hastily, and we often draw incorrect conclusions. Also, in our desire to act, and change things, we often do not allow time for luck and other variables to even out. We overcorrect, and often the overcorrecting has negative consequences for players who end up confused or lose trust and confidence in the coach or the style of play. Often, continuing along the same course is the correct response, even after a bad game or performance by the team or a player. Performance is temporary, but constant changing (action) can lead to permanent underperformance. Slow and steady wins the race.