Evidence-Based Basketball (Go Big Red!)

The Cornell men’s basketball team is on a tear this season, at 21-4 so far, and with wins (or near-wins) over nationally ranked teams. Sell-outs here used to be limited to hockey, but now the Newman Arena is packed for the home games and there’s talk of the Big Red going further in the NCAA tournament than ever before. The big media are even reporting on them. I admit that as sports go, I’m basically a fair-weather fan, so our increasingly winning team has me in the bleachers every time they’re at home.

At a recent game, I happened to sit next to Tom Gilovich, professor of social psychology at Cornell. And seeing Tom reminded me that one of his studies is a great example of the difference between science and “common sense.” In this blog, we try to look at what science has to offer, and how it may conflict with popular opinion, media reports, or political viewpoints. Tom and colleagues used the basketball court to look at why people erroneously see patterns in random occurrences, despite evidence to the contrary.

They looked at the idea of a basketball player having the “hot hand” in a 1985 article. That is, a player is thought to hit baskets in streaks, so when the player has made a shot or two, everyone wants to get the ball in his or her hands. Gilovich and colleagues surveyed 100 basketball fans, who overwhelmingly believed in shooting streaks like this. Then, they analyzed shooting data for the Philadelphia 76ers and conducted an experiment using the men’s and women’s basketball teams.

The result? No hot hand. Players aren’t more likely to hit the next shot if they hit the one before. Instead, the chance that someone would make a shot was approximately the same whether or not he or she had made or missed  similar shots previously.

It’s just a misattribution we make. But we all deeply believe it’s true, and probably would continue to do so even when presented with hard data that no hot hand exists. It’s analogous to the persistence of erroneous beliefs in a number of more important arenas (like health, for example).

I now stand corrected about this particular belief. However, if he hits a couple three-pointers in a row with a minute left, it’s going to be hard to keep myself from standing up and screaming “Get it to Wittman!”

Comments

  1. Mike C says:

    Wow! This brings my entire sports-viewing and opining career to a halt! It’s interesting that sports video games actually employ intangible ideas such as a Momentum Meter, where a team’s “momentum” (and thus subsequent chances at success) improve with repeated scores or big plays.

    I wonder what the results would be when examining the so-called “streaky hitter” in baseball?

  2. Ugo says:

    I’m not so sure that this study correctly defined ‘Hot Hand’. Correct me if i’m wrong, but it sounds as if the study says that if any person hit a (one) three pointer at any point in a game, they were deemed the hot shooter, then the investigators noted whether or not the next three pointer that player shot went in.

    A couple points:

    1) I’d argue that any subsequent shot taken by the player (regardless if its from 5ft or 23ft) should be considered in this study. What if a player hits a three, then makes 3 consecutive 2 point field goals, and then 5 minutes later, misses his next three pointer. Under this study design, we would ignore the 3 two point field goals. Can we confidently then say that the 4 consecutive shots were not the result of a hot hand?

    2) I think a better definition of a hot hand would be to follow a player that makes 2 consecutive 3 pointers or 3 consecutive 2 point field goals from a distance of at least 12 ft. Then the next shot, regardless of location should be considered.

    3) Also, what is the time difference between shots? Streaky players normally cool off between halves, or even as quickly as after a time out. That would also be pretty interesting to take a look at.

    In all, i think its a pretty cool study. I’m a bit of a nerd and bball junky, so i’d like to read more of the science behind the study before i further criticize. Sounds pretty cool though. Nice job!

    • Karl says:

      Hi Ugo,

      Great points! I’m going to forward your comment to Tom Gilovich, and perhaps he’ll share his thoughts.

    • Tom Gilovich says:

      HI Ugo,

      You’re correct to point out that specifying when a player should be considered “hot” is not a simple matter. As a result, we examined a number of definitions in our research and found the same result each time: Although players’ hits and misses do come in streaks, the length and frequency of those streaks do not exceed what you would expect if consecutive shots were independent (that is, if they have no influence on one another). So, players sometimes hit 4, 5, or 6 shots in a row, but they are not more likely to hit their next shot regardless of their recent history of success.

      Two analyses are particularly relevant to your questions. First, we had members of Cornell’s varsity basketball teams take 100 shots from various points along an arc an equal distance from the basket. We asked the players to tell us if they felt hot at any time during this 100-shot shooting experiment. Thus, the definition of what constituted being “hot” was determined by the players themselves; no definition was imposed on them. What we found was that:

      (1) we could predict when the players would be likely to say they were hot. If they hit several shots in a row, they usually told us they were hot.

      (2) The players did not shoot a higher percentage on the shot (or 2 shots, or 3 shots, and so on) immediately following their statement that they were hot. Thus, players sometimes make several shots in a row and when that happens, they often feel hot (what one might call a “descriptive” hot hand). Nevertheless, widespread beliefs to the contrary, that feeling of being hot doesn’t correlate with future shooting success (there is no “predictive” hot hand).

      The other analysis comes from the NBA’s 3-point shooting contest conducted during the All-star weekend. As in our study, the length and frequency of various streaks (3, 4, 5, etc. hits in a row) are just what you’d expect if the shots were independent of one another. And players are not more likely to make a shot after having hit 1, 2, 3, etc. shots in a row than after having missed 1, 2, 3, etc. shots in a row.

Trackbacks

  1. [...] world’s attention (except for, unfortunately, much of the United States). As we did earlier with basketball, we at Evidence-Based Living couldn’t help asking the question: Is there evidence-based [...]

Speak Your Mind

*