The Mystery of the Hot Hand
By Dave Nussbaum
If you’re a basketball fan you’ve probably seen the hot hand with your own eyes. A player hits one shot, then another, and he seems to have found his rhythm. Pretty soon he seems unstoppable – it’s like he can’t miss. I know I’ve seen it – I’ve even experienced it firsthand.
You can imagine my confusion, then, when the professor in my college Intro to Social Psychology class told me that there’s no such thing as the hot hand. I’ve been seeing a pattern where none actually exists – it’s all just chance. Based on research by Tom Gilovich, Robert Vallone, and Amos Tversky, the best predictor of the likelihood that a player will hit his next shot is his average shooting percentage; whether or not he’s made his last shot, or even his last several shots in a row, doesn’t make much difference.
My hand shot up immediately. Surely there has to be an effect of the confidence you get from hitting a few shots in row, I protested. This was not the professor’s first experience with a student challenging her pronouncements based on personal experience, so she had an answer ready: Apparently not, according to the data. Undeterred, I perseverated. How about if you miss a bunch of shots in a row and your confidence is shot, that must make a difference, right? A little exasperated, the professor tried to move on politely. My friend Tom, sitting next to me, was by this point thoroughly embarrassed by my persistence and had slunk down several inches in his chair. Still, I was convinced – if they hadn’t been able to find evidence for the hot hand, then they probably just weren’t looking hard enough.
So you would think, then, that 15 years later I would feel vindicated. New data suggests that when you use much richer data than was available previously, the mythical hot hand effect appears to exist. But I don’t feel all that vindicated. In the meantime I’ve come to realize that while the hot hand effect may be real (in that it can be detected with mountains of very precise data) most of the time that I thought I had detected it, chances are it was just a pattern that looked like the hot hand, but was in fact just a statistical fluke. Because I’m human, I’m excellent at detecting patterns, whether or not they’re really there.
To learn more, I reached out to Michael Kraus (@mwkraus), a sports fan and a social psychologist at the University of Illinois at Urbana Champaign, who has an interest in the hot hand research and was willing to share his reactions to the new findings. Here’s my interview with him, conducted recently over email:
DN: Let’s start by laying out the background: what did the original research on the hot hand effect actually say and how did they reach their conclusions?
MK: In the original research, Gilovich and colleagues observed that there is a widespread belief in the hotness of streak shooters. Among the basketball fans surveyed in the original, for example, 91% believed that a shooter has an increased likelihood of making the next shot if s/he has made the last couple prior shots. However, when examining actual data from professional basketball players and college players at Cornell, making prior shots did not predict subsequent increases in shooting performance. The research concludes that the hot hand is an illusion based on people ascribing something special to random streaks when in fact, that is all they are: random.
DN: What are the new findings?
MK: The new findings examine the same question of the hot hand, but using new motion tracking software in all professional basketball arenas (called SportVu). The software allows the researchers to control all the variables that influence shot accuracy in live basketball settings (e.g., closeness of defenders, shot difficulty by location, defender height, time point in the game). In the new study, the researchers can ask: is a player on a shooting streak shooting at a higher percentage than they typically do from this exact location on the basketball court. This innovation allows for a more precise test of the hot hand question. The results support the existence of the hot hand: Players shoot (very) slightly better than expected (i.e., about 1%) when on a streak than when not on a streak.
Interestingly, the researchers find evidence suggesting that NBA players (unsurprisingly) believe in the hot hand. That is, when on a hot streak, players take shots from greater distance, are more likely to take their teams next shot, and are more likely to be closely covered by a defender–all behaviors indicating that players are reacting to the apparent hotness of a teammate or opponent.
DN: Does this new work suggest original was wrong?
MK: No way. I interpret the findings from the new evidence as mostly consistent with the original work. In both papers we see evidence of players/fans overestimating the extent that prior streaks matter for future performance–taking shots from greater distance with closer defenders suggests the players in the new study are victims of the hot hand fallacy just like the fans were in the original research. That players show a 1% increase in accuracy following streaks is probably not enough of an increase in shooting performance to justify this sort of “gun-slinging” of deep/contested jump shots.
DN: Is the study convincing?
I’m not entirely sure that a 1% increase in shooting accuracy following streaks is what, phenomenologically, we are talking about when we think of the hot hand. When a person is hot in basketball, players and fans want the hot player to take more shots and play more selfishly. If the actual hot hand is only a tiny increase in shooting accuracy, I think it would lead fans and players alike to abandon this idea that we need to feed shooters on streaks.
DN: Is there other research that supports the hot hand?
I actually think that the idea of a hot hand psychological state is not all that controversial. In psychology research we have work on choking by Sian Beilock (PDF) suggesting that people enter psychological states in high pressure situations that impede performance. In the challenge/threat work of Wendy Mendes (PDF), and before that, Jim Blascovich (PDF), we have consistent evidence suggesting that challenge physiological states–characterized by higher blood volume and greater sympathetic nervous system activation–boost performance on physical and cognitive tasks. If we think of challenge and threat as a continuum, we can think of the hot hand as existing at the extremes of the challenge spectrum and choking as existing at the extremes of the threat side.
Putting aside the question of whether or not the hot hand exists, what is still clear though, is that we fundamentally overestimate the extent that our sports heroes enter into hot streaks when we watch their performances in games. That was true in the 1970s and it’s still true today. I’d actually tie this sort of bias to Jost’s motivated social cognition perspective (PDF) – we have a desire to see meaning and purpose in the randomness of the world in order to satisfy some of our basic needs (perhaps in this case, to understand the world as predictable and controllable).
Michael W. Kraus is an Assistant Professor at the University of Illinois, Urbana-Champaign and director of the Champaign Social Interaction Laboratory (www.krauslab.com) where he studies the hierarchical and emotional dynamics of social life. Michael blogs regularly about social-personality psychology for Psych Your Mind (http://psych-your-mind.