Commentary invited by editors of Scientific American

# Cognitive Biases in Sports: The Irrationality of Coaches, Commentators and Fans

Phil Jackson coaching LA Lakers. Wikimedia Commons.

The Minnesota Twins are my favorite baseball team. So I was ecstatic when, back in June, they went on a nine game winning streak to come within a few games of first place. But I was also confused: they were plagued by injuries, AAA players filled half their roster, they weren’t scoring very many runs and they weren’t pitching particularly well. How was it, I asked, that they almost pulled off the magical “worst-to-first”?

ESPN and Sports Illustrated commentators had their explanations as usual, ranging from better coaching, to more focus by the players, to warmer June weather. But being the skeptic I am, I wasn’t buying any of this nonsense. Instead, I recalled the famous “hot hand” study (pdf) done by psychologists Tom Gilovich, Amos Tversky and Robert Vallone. After pouring through shooting records of the Philadelphia 76ers 1980-1981 season, they found that the chance a basketball player has of making a shot is actually unrelated to the outcome of his previous shot. In Gilovich’s words, “a player’s performance on a given shot is independent of his performance on previous shots.”

Similar results were found with baseball teams and baseball players. Michigan State University psychologist Gordon Wood demonstrated that the probability of an MLB team winning after a win or losing after a loss was fifty percent after analyzing the outcomes of all 1988 Major League Baseball games. Likewise, Indiana University statistician Christian Albright found (pdf) that batters, “[do] not differ significantly from what would be expected under a model of randomness,” after analyzing statistics from Major League players through four seasons. In short, an MLB game and at bat are, like the outcome of a basketball shot, unaffected by past performance. This means that the Twins won nine in a row because, well, inherent in a string of 162 baseball games are streaks. The problem is that we do a really bad job of interpreting these streaks.

Our tendency to “see” streaks plays to a bigger point: our intuitions about sports are typically way off the mark.

This is highlighted in Scorecasting, a wonderful new book by University of Chicago finance professor Tobias Moskowitz and Sport Illustrated writer Jon Wertheim. The book does to sports what Freakonomics did to economics. That is, it “confronts conventional sports wisdom.” They ask, for example: Is it really preferable to punt on fourth down rather than go for it? To try to achieve the highest available spot in the draft? Is there an I in team? Does defense truly win championships? And, of course, is it smart to keep feeding the teammate with the hot hand?

We know that the hot hand is a myth, so let’s take fourth downs.

For that, consider a study (pdf) done a few years ago by an economist David Romer. Romer analyzed every fourth down that occurred in the first quarter of every NFL game from 1998 to 2000. After considering several variables Romer developed a mathematical model that gave the probability of successful converting fourth downs and kicking a field goal depending on position and circumstance. Then, he compared his model with what actually happened. He concluded that, “the behavior of National Football League teams on fourth downs departs systematically from the behavior that would maximize their chances of winning.” In other words, it is usually better to go for it on fourth down, and coaches do a really good job of not doing this.

Joe Mauer swings at Metrodome. Wikimedia Commons.

Studies like Romer’s abound. It turns out that if you run the numbers, coaches and managers are the kings and queens of making non-optimal decisions. With the exception of Kevin Kelley who categorically opposes punts and usually goes for the onside kick (by the way, he has won multiple state championships as head football coach of Pulaski Academy in Little Rock Arkansas), Billy Beane of Moneyball fame and those who consider sabermetrics, coaches and managers are slow to adapt studies like Romer’s into their play books.

Why would anyone not adjust their game plans in accordance with data that would allow them to choose optimally? Why would you ignore information that could help your team succeed? Why is it, in other words, that when Red Auerbach, long time coach of the Boston Celtics, heard about Gilovich’s “hot hand” study he responded, “so he made a study… I couldn’t care less.”

The short answer is that coaches and managers are like all of us: they only look for what confirms their intuitions and ignore what contradictions their intuitions, what psychologists call confirmation bias – a vastly interesting cognitive tendency I have written about before. Confirmation bias helps explain a lot. It’s the Republican who only watches Fox or the Democrat who only watches MSNBC; it’s the lady who sees Mary in toast or the creationist who sees fossils as evidence of God; it’s the grad student who can’t see any flaws in his work or the professor who only looks for evidence that supports her study. We automatically see the world as we want to, not how it is.

In addition to confirmation bias, there’s motivated reasoning, confirmation biases evil twin. Motivated reasoning describes our propensity to scrutinize ideas that oppose our intuitions less than ideas that support our intuitions. My favorite study (pdf) that illustrates motivated reasoning comes from Ziva Kunda. It goes like this. Kunda brought participants into a room and had them play a game. Before they started they watched two other people play the game, one who was far superior (and actually a confederate). Here was the catch: Kunda told half the participants that the expert player would be their teammate and the other half that he would be their opponent. Kunda found that the participants lined up to play with the expert praised his skills while those lined up to play against the expert dismissed his skills and labeled him as lucky.

I hope you can see where I am going with this. When you consider confirmation bias and motivated reasoning, it is easy to understand how a coach or manager would stick to their play books even in the face of data that overwhelmingly shows their strategies to be non-optimal. Being bound by the tendency to only notice what fits with our beliefs (confirmation bias), and having the propensity to scrutinize ideas more when they run contrary to our beliefs (motivated reasoning) is a deadly poison for optimal decision making.

Don’t get too comfortable. You – the fan – aren’t exempt. According to NYU professor of psychology Gary Marcus, who I recently spoke with:

“Fans are subject to motivated reasoning, too. An executive at a major sports league once told me about a study his league did in which fans watched instant replays from controversial plays, and had to judge whether the refs had made valid calls. People were vastly more likely to think that the ref made the right call when the call favored the home team – even though the games were from an earlier season, and it was obviously too late to change anything. Human beings are built to enjoy being “right”, and it often distorts our perception.”

This helps explain why fans can look at the same statistics, the same standings and the same plays and come to entirely differently conclusions.

And then there are the commentators.

SportsCenter studios. Wikimedia Commons.

When it comes to analyzing plays and calling games, they are cognitive disasters. Along with confirmation bias and motivated reasoning, they exemplify what Nassim Taleb calls the narrative fallacy. The narrative fallacy describes our “limited ability to look at sequences of facts without weaving an explanation into them, or, equivalently, forcing a logical link, and arrow of relationship, upon them.” For example, if the Twins win a game, commentators may point to good pitching, good hitting or good managerial decisions to explain why they won. When they do this they are creating the sense that they understand why the Twins won. This is an illusion. Their explanations are mere post hoc stories that add silly irrelevant facts just to garner more views and increase entertainment. The only fact of the matter about the Twins winning is that they scored more runs than the other team.

They also epitomize the conjunction fallacy, another cognitive bias that describes our tendency to think that, “the probability of any event A and any other event B occurring together has to be less likely than (or equal to) the probability of event A by itself.” Here is the classic Kahneman and Tversky example:

Linda is 31 years old, single, outspoken, and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in anti-nuclear demonstrations.

Which is more probable?

1) Linda is a bank teller.

2) Linda is a bank teller and is active in the feminist movement.

If you are like 85% of people, you chose option two even though there is a higher probability that option one is more likely because there are fewer conjunctions; counter-intuitive but it’s true. How do sports-casters commit this error?

During big games or plays they often declare something like this: “Tonight’s game saw the first no-hitter thrown by a lefty pitcher during the month of April.” Statements like this give the viewers a sense that what they are watching is rare and unique. This is also an illusion. Again, it’s commentators adding silly irrelevant pieces of information (e.g., lefty or April) to garner more views and increase entertainment. All the while, it’s just another game.

Don’t waste your time calling them out on their errors; confirmation bias and motivated reasoning are robust biases that do an excellent job of narrowing people’s minds; once the narrowing begins, it’s nearly impossible to change people’s minds.

I’m making a simple point. When it comes to sports, data > intuition. Yet, coaches, managers, fans and commentators alike continue to go with their guts, especially when it comes to a team or player that is close to their hearts. If they want to decide optimally or speak about their beloved team or player with slightly more intelligence they should turn off their cognitive biases and look at the data. Trying to persuade someone to change a strategy, root for another team or consider why the sports team from your area is superior to the sports team from their area is not unlike trying to persuade a Republican that Obama is a good President or an atheist that God exists. It’s just not going to happen.

I’m sad to report that the Twins are currently 29 games out of first place. If manager Ron Gardenhire is smart, he will ignore his intuitions to try and understand how data can help his team in 2012. If I’m smart, I’ll mind my cognitive biases so I can actually have an informed discussion about sports.

About the Author: Sam McNerney graduated from the greatest school on Earth, Hamilton College, where he earned a bachelors in Philosophy. After reading too much Descartes and Nietzsche, he realized that his true passion is reading and writing about cognitive science. Now, he is working as a science journalist writing about philosophy, psychology, and neuroscience. He has a column at CreativityPost.com and a blog at BigThink.com called "Moments of Genius". He spends his free time listening to Lady Gaga, dreaming about writing bestsellers, and tweeting @SamMcNerney. Follow on Twitter @SamMcNerney.

The views expressed are those of the author and are not necessarily those of Scientific American.

 Previous: Space – It’s not just for Astronauts: Personal reflections from a NASATweetup MoreGuest Blog Next: A physicist flirts with philosophy (and lives to tell the tale)

1. 1. thack86 1:30 pm 09/22/2011

How does the “hot hand” study relate to those in the “zone”?

2. 2. jtdwyer 1:42 pm 09/22/2011

Unless I’m confused, didn’t you mean: “worst-to-first”?

Also, doesn’t everyone know that there’s a strong correlation between consecutive wins and underwear streaks?

3. 3. smcnerne 2:10 pm 09/22/2011

@thack86 people who have a “hot hand” are said to be “in the zone.” They are two sayings that describe the same thing (the “hot hand”, however, is usually describes basketball players while “in the zone” describes all athletes).

@Jtdwyer thanks for the tip, fixed.

And Yes! obvious correlation….

4. 4. smcnerne 2:47 pm 09/22/2011

Just found a nice article that compliments some of the things I mentioned. It’s over at the almost thought-provoking Freakonomics blog.

http://www.freakonomics.com/2011/09/22/picking-the-nfl-playoffs-how-the-experts-fumble-the-snap/

5. 5. ndd412 3:14 pm 09/22/2011

I like the idea of this article and I think it does well in explaining viewer/commentator reactions to sports. My one problem is in stating that commentators can’t rightfully point to good pitching/hitting as an influence in the win.

There are a lot of individual factors that go into sports, especially baseball. When you look at a pitcher, there are influences which can be accounted for when you compare situational statistics, such as the influence of weather on the amount pitches break. If a pitcher is pitching in a situtation that will enhance/diminish the effectiveness of a pitch, it will certainly affect the mindset and strategy of that pitcher. If this is the case, you can certainly point to that pitcher as having significant effect on the chances of that team to win the game.

6. 6. showme 3:42 pm 09/22/2011

Love the article. I fall victim to a few of these irrationalities occasionally, but I have realized for a long time how ridiculous some of them are, especially when it comes to commentators. Almost every day you will hear some sportscaster saying something like “This is the first time since [current year - 2 or 3] that [team x] has done [some slightly unusual event].” I would bet that if you tried hard enough you could come up with some event for almost every team for almost every game, yet they still talk about them as something exciting. I chalk it up to a reason I don’t believe Mr. McNerney brought up in the article: Sportscaster [current sportscaster] attempting to sound interesting in order to keep his job, which is a very rational thing to do, more rational than talking intelligently, I’m afraid.

7. 7. smcnerne 3:50 pm 09/22/2011

Saying that weather affected, either in a positive or negative way, the chances a team has of winning a game is completely weightless. It says nothing. Everything affects the chances of a team winning a game. There’s weather, time of day, time of year, health of players, health of managers, number of fans in attendance etc. Baseball is such a complication game that it would take a fool to say that one person or one thing significantly affected the game. It’s mere speculation.

Yet we continue to explain baseball games, or any athletic game, in simple terms. This is a delusional and it unfortunately happens all the time. At the start and end of a game commentators and “experts” give their keys to the game. They are usually two or three overly general bullet points. They say nothing. Baseball is far too complex to have “keys” or bullet points. If you watch sports with even a hint of a critical lens you know this.

8. 8. smcnerne 4:49 pm 09/22/2011

@showme I’m afraid you’re right. When it comes to rationality and entertainment, entertainment wins.

9. 9. tsardoz 12:51 am 09/24/2011

As an engineer with a PhD in time series analysis I see signals everywhere. There are mathematical tools to analyse trends and quantify them – its called system identification. I use these tools to forecast match outcomes. And guess what? I bet on them and make money. My average winning margin on football is 10% of bet on spread bets after several hundred matches so far. This is a very high margin. I have read many of the papers on “hot hands” and just shake my head that this is taken so literally by many people. To say “hot hands” dont exist as the author does here is just its own form of confirmation bias. It ignores the possibility that trends can exist in sport and are measurable. Is it possible that training can improve performance? Of course it is! Do I have my own confirmation bias? Maybe but I put my money where my mouth is and after several hundred bets am well in front so I think my bias is less than those who just read hot-hands literature and denounce time series trends in sport off-hand.

10. 10. smcnerne 2:43 pm 09/24/2011

@tsardoz Using your PhD and experience as reasons to refute the hand hot tells me that you think you know a lot more than you do. I didn’t say that hot hands don’t exist, I said that people do a poor job of interpreting them. There are trends in sports, they are measurable, and you can use them to inform your opinion. No doubt.

If you want to refute the hot hand studies do so constructively. Don’t just tell me how much money you made, that makes you sound really foolish.

11. 11. sterge 12:52 pm 09/26/2011

The generalization and biased idea that an Atheist presented with factual evidence of the existence of a “God” or “Gods”, would refuse to “believe” in said evidence, is humorous to me. The very lack of any such evidence is what made me drop my existing bias of my Circumstantial (i.e. – born into and raised in) Christianity. I did, in fact, look at the actual stats to come to a conclusion. I would agree that there may be some that are so angry that they may refuse to see the evidence, but it’s a funny and strange analogy to use in this article; using the reverse is much more relevant in our current society. The majority of people to still cling to silly, unfounded, tribal religions because of their biases. Moving to atheism is actually an examining of the evidence and would actually be a position that supports these studies.

12. 12. smcnerne 2:52 pm 09/26/2011

I think confirmation bias is still at work when an atheist points to evidence against the existence of a God or Gods. Keep in mind that confirmation bias, motivated reasoning and the like are just as helpful as they are harmful.

In other words, atheist have a much more accurate outlook, but they are still bound by CB like everyone else.

13. 13. tsardoz 12:47 pm 10/1/2011

smcnerne you are wildly extrapolating the “hot hands” studies to make assumptions about Minnesota Twins winning streak as just being a random event. You have no evidence that this is true at all. Try telling Tiger Woods that he is just going through a random bad run or that he just got lucky when he was younger. The fact is that “hot hands” must exist with all sportsmen to some extent. Tiger could not shoot sub par when he was a two year old and I doubt that he can when he is 80. It is all a matter of choosing the right model and working out the right parameters for any given time series. You can do this when you get enough data. The “hot hands” models are generally very simplistic. They say that streaks just represent random deviations from an average. But what do you take as average? What is the shooting percentage of your favourite player? What is Tiger Woods’ round average? Do you take it over the past season, the past 10 seasons or the past week? This is where system identification techniques (as used by engineers and mathematicians who understand time series) come into play. They can work out for you what the relevant time period is to determine these trends and averages based on the data, provided there is enough of it. Psychologists tend to lack the mathematical background to appreciate these models.

You say I don’t know as much as I think I do. I suspect I know much more about this than you, given I studied time series data analysis for my PhD. I notice you have a bachelor’s degree in philosophy so I somehow doubt you have the mathematical skills to judge whether I am writing truth or fiction. So my guess is you’ll just revert to your confirmation bias which you stated pretty clearly in the first two paragraphs.

14. 14. tsardoz 12:32 am 10/2/2011

Just to add a bit more evidence.
See

http://www.dartmouth.edu/~stats/rendleman.pdf

This paper provides a model of how golf skill changes with time. Now I do not like their spline model much as it just an arbitrary function fitting algorithm with no foundations in system identification time series. System identifcation (I have mentioned this term several times now – have you googled it yet?) provides a theoretical foundation based on engineering signal theory of how variables naturally change with time. Much of modern electrical engineering is based on these models. However I do not think that using SI models would necessarily lead to different conclusions to those drawn by the authors of this paper. You can clearly see the skill levels of golfers changing with time.

Now I am not attempting to debunk the hot hands papers. Indeed I largely agree with them when you look at very short time scales. I do raise the question however, how does one determine an average skill level? Over what period does one measure such averages? eg. See figure 3 in my reference.

In your argument you are extrapolating from the hot hands papers which looked at runs of individual shots to team winning streaks. It doesnt work. My reference clearly shows skill levels change with time. This would happen with the Minnesota Twins as well. The problem is that you have no way of knowing how to differentiate a lucky streak from a genuine shift of skill level. This could be determined from system identification analysis of all baseball matches in order to determine the appropriate model (usually either first or second order is sufficient) and estimate values of the model parameters. Only then would one be able to say if a nine game streak represented a genuine skill change or just a random fluctuation. Now I have not done this for baseball but have for several other sports. I have found an appropriate “time constant” (google that too) is often around 4 games for other sports so a good guess would be the same for baseball.

You must sign in or register as a ScientificAmerican.com member to submit a comment.
Click one of the buttons below to register using an existing Social Account.