Wednesday, October 10, 2012

Of Cognitive Fallacies, Streaks and the "Hot Hand" in Volleyball

Human Frailty

For many years psychologists have studied the human tendency to see patterns where none exist.  We look for patterns in random data and we see patterns in random events.  In studies of probability guessing where colors are randomly presented, but weighted to appear with different probabilities, humans guess the next color by trying to guess the pattern and rats just go with the color that appears most often in the series, and with that, humans allow themselves to be outperformed by rats.[1]

We don't do so well analyzing situations involving chance.  We do even worse when we feel uncertain or out of control.  Randomness fools us because we don't recognize it.  We like to make sense of the world by seeing patterns and assigning to every event a distinct and definite cause.  So our thinking leads us to misinterpret all kinds of events around us and to make decisions that sometimes are less than optimal and other times are flat out wrong.[2]

The world is full of "patterns of randomness" - streaks - that are routinely misinterpreted and then relied upon as if they portended a trend.  In our oceans of data we struggle to interpret correctly and sometimes see meaning where none exists.  Indeed, that's why most readers will see (or suspect) a pattern in this random sequence:      

OXXXOXXXOXXOOOXOOXXOO[3]

We see patterns everywhere, but perhaps in no context do we do so more fervently than in sports.  So let's look at our bias toward patterns as it manifests itself in the popular belief in the "hot hand."
 
The "Hot Hand"

The "hot hand" in sports has been debated for years.  Is a basketball player more likely to continue shooting well after making his first four shots?  Will a volleyball player more likely get a kill after putting away the last three balls?  Athletes, coaches and fans nearly universally accept that players running a streak of success are more likely to continue succeeding on subsequent attempts until they "cool off."

But is this really true?  Is there any such thing as a "hot hand" in sports or are we just seeing random streaks of success within a much larger performance range and body of work?  Is each hit, goal and kill independent of the one before (as in flipping a coin) or does the streak of success drive further success? 

Fallacy or Not?

In the seminal study of the hot hand in sports, psychologist Thomas Gilovich and his colleagues looked at streak shooting in basketball.[4]  Extensive shot analysis revealed no evidence that a player's chance of hitting a shot was greater following a hit than following a miss.  Indeed, Gilovich found that for nearly 90 percent of players studied, the probability of making a shot was actually slightly lower following a make than following a miss.  Accordingly, it was concluded that belief in the "hot hand" was merely "an erroneous perception of a positive correlation between successive shots" - in other words, a cognitive fallacy.

Research since then has by and large reached a similar conclusion, leading some experts to summarily state that empirical evidence of the existence of the hot hand in sports is "considerably limited."[5]

Though the hot hand may be a myth, research has spawned analysis of some interesting questions on the periphery.  Researchers have questioned what impact opponent's defensive changes have on detection of the "hot hand."  Others have examined whether belief in the hot hand, even if fallacious, can beneficially drive in-game allocation decisions and result in the optimal player getting the ball[6].

"Hot Hand" Revisited - Attacking Streaks in Volleyball

In the last six months a group of German researchers re-visited the hot hand phenomenon to determine whether it exists in the sport of volleyball and whether belief in it drives players' and coaches' decisions to distribute the ball to particular players.  The research is published in the Journal of Experimental Psychology: Applied[7].

Defining the Hot Hand

When we speak of the "hot hand" we mean more than just a performance streak.  Streaks in sports occur all of the time and have been shown in a variety of settings.  We all know that baseball players can get three or four hits in a row and volleyball players can get three or four straight kills.  A belief in the hot hand, however, refers to a belief that the streak portends a greater likelihood of success on each successive attempt.  Thus researchers in the present study defined a hot hand as a "higher probability  . . . to score again after two or more [kills] compared with two or three [hitting errors]."

What they found was rather interesting:  the hot hand exists in volleyball; athletes and coaches overwhelmingly believe in it; and both rely on it to make allocation decisions in the game.

Summary of Results

To determine whether athletes believe in the hot hand in volleyball the authors asked 94 sports science students with athletic experience at the German Sport University Cologne to respond to a questionnaire on the subject.  Eighty-six of the ninety-four athletes (91%) reported believing in the hot hand in volleyball.   In other samples taken, 90% of 21 other athletes reported and 92% of coaches also reported believing in the hot hand.

The authors next sought to determine whether that belief was founded. To determine whether a hot hand does in fact exist the researchers analyzed game performances of 26 male volleyball players from the top volleyball league in Germany.  They found that half of the players experienced a hot hand phenomenon and half did not.

Given that the hot hand appeared to exist for some players but not everyone, the authors examined how setters use their belief in the hot hand to distribute the ball.  Setters in an experiment were sensitive to both a hot hand belief and overall hitting percentages, but the "hot hand" belief was the stronger cue in actual set distribution.

Significantly, the researchers found that setters more often distributed the ball to players experiencing streaks of success and this lead to better performance than when setting the ball to the player with the higher average hitting percentage.  The authors cautioned, however, that a distribution strategy that ignores hitting percentage would be less than optimal where the differences in hitting percentage between the "hot" and the "non-hot" player compensate for the advantage of setting the streak. 

The German authors are not the first to posit that optimal play may result from setting the player with the perceived "hot hand."  A study published in 2001 by a psychologist at the University of Michigan showed that distributing the ball to the player perceived to be "hot" in basketball positively impacts the team regardless whether the hot hand phenomenon actually exists.

The studies' author demonstrated that basketball players with higher shooting percentages experience more streaks of made shots, so when streak shooting produces a perception of "hotness," distributing the ball to the "hot hand" actually results in the best shooting player getting the ball more often[8].

Why Do We Care?

So what does all this mean?  As a volleyball coach I really don't know whether the "hot hand" phenomenon exists in volleyball or not.  I do know that collectively I've played in, watched and coached thousands of volleyball games and sometimes players hit hot streaks and sometimes players hit cold streaks.  What to do with that information - I'm still learning.

But uncertainty should not shroud the more fundamental point of research on the subject - that people, coaches included, tend erroneously to detect patterns in random data and often make poorer judgments because of it.  Recognition of this fact - of the frailty of our own thinking - should inform and improve the process by which we all make decisions.

As coaches we make thousands of decisions every season.  We make personnel decisions, we decide tactics, strategies, and systems of play.  We choose whether, how and when to give feedback.  We decide whether to run practice, how long it will be, and what its design will consist of.  We even decide who will be on our teams and whether and how often they will play.  Our thinking and our choices permeate our programs and touch the lives of many, many people.  With research continuously showing our impact on the lives of other people we should care deeply about how we make decisions and at least be aware of our own frailties.  We owe it to our players and the profession to constantly search for ways to improve.

NOTES

[1]  Mlodinow, L., (2009).  The drunkard's walk: How randomness rules our lives.  New York: Vintage Books.

[2]  For two excellent books on how we make decisions when faced with uncertainty I recommend Leonard Mlodinow's The Drunkard's Walk: How Randomness Rules Our Lives and Thomas Gilovich's How We know What Isn't So: The Fallacy of Human Reasoning in Everyday Life.

[3]  Gilovich, T. (1991).  How we know what isn’t so: The fallibility of human reason in everyday life.
New York: The Free Press.

[4]  Gilovich, T., Vallone, R., Tversky, A. (1985).  The hot hand in basketball: On the misperception of random sequences. Cognitive Psychology 17: 295–314.

[5]  Bar-eli, M., Avugos, S., Raab, M., (2006). Twenty years of hot hand research: Review and critique.  Psychology of Sport and Exercise 7(6): 525-553. 

[6]  Burns, B. D. (2001). The hot hand in basketball: Fallacy or adaptive thinking? In J. D. Moore, & K. Stenning (Eds.), Proceedings of the Twenty-third Annual Meeting of the Cognitive Science Society (pp. 152-157). Hillsdale, NJ; Lawrence Erlbaum; Raab, M., Gula, B., Gigerenzer, G. (2012).  The hot hand exists in volleyball and is used for allocation decisions.  Journal of Experimental Psychology: Applied 18(1): 81-94.

[7]  Raab, M., Gula, B., Gigerenzer, G. (2012).  The hot hand exists in volleyball and is used for allocation decisions.  Journal of Experimental Psychology: Applied 18(1): 81-94 (available here).

[8] Burns, B. D. (2001). The hot hand in basketball: Fallacy or adaptive thinking? In J. D. Moore, & K. Stenning (Eds.), Proceedings of the Twenty-third Annual Meeting of the Cognitive Science Society (pp. 152-157). Hillsdale, NJ.