March 24, 2008
Prof. Suzanne Shu
Everyone wants to pick a winner, but figuring out who’s on a streak and who’s not is tricky. Three in a row seems to be the key. But is Hillary or Barack on a streak? And what about all those men’s and women’s basketball teams in the NCAA tournaments?
SMU Cox Marketing Professor Suzanne Shu and co-author Kurt Carlson of Duke University got to thinking about perceived streaks and who – or what – is hot, studied the problem and wrote a paper on it for Organizational Behavior and Decision Processes magazine.
Read about their findings…
People often perceive streaks where they do not exist. But, little is known about what constitutes a streak in the mind of an observer. SMU Cox Marketing Professor Suzanne Shu and co-author Kurt Carlson of Duke University propose that the third repeat event in a sequence is pivotal to the belief that a streak is emerging. In "The Rule of three: How the third event signals the emergence of a streak" published in September's Organizational Behavior and Decision Processes, five studies find evidence that "3" holds significance in the minds of observers across many domains.
Prof. Kurt Carlson
The idea behind the research came from an impromptu conversation between the two authors. "We were talking about the idea of 'hot hands' in basketball, when a player that's on a hot streak makes sequential baskets, and is given the ball more often than the player not on a streak," Shu recalls. "And so we thought about what makes people think there's a streak. My co-author noticed that when teams won three weeks in a row, that the betting got interesting. So this perception of streak occurs in sports, but what about in coin flips, and other domains?" The authors then merged the idea of the perception of a streak with when that idea emerges: what is the magic number which makes a streak 'real,' apparently?In five studies, the authors note direct and indirect evidence that perceived streakiness plateaus with the third repeat outcome in a sequence. The rule of three also draws support from human learning and cognition, such as the optimal number in advertising exposures, and perception and judgment in the role of patterns. The evidence to support this rule of three comes from various domains, including: observation of randomly determined probabilistic outcomes (coin flips), investment decisions in response to performance histories, and basketball shooting percentages.
Of the five studies, Shu notes the Google search for the incidence of terms like "two-game winning streak" or "three-game", "four-game" and so on. She mentions, "The fact that reporters and people who are observing things are likely to call a string of three a streak is notable. It was not two events or wins which caused mention of a streak, but three. Additionally, this is not a lab observation. So, it is a much broader phenomenon than just coin flips, for example." The study found a substantial jump in the occurrence of three-event runs versus two-event runs, thus supporting the rule of three. This suggests streaks do not begin in earnest until the third time.
The findings reveal a difference in behavior when the rule of three is applied. Another study considered how much money students would invest in a derivative after seeing recent performance. As the rule of three predicts, behavior modified after the change from a two-day rise in value to three. The participants were betting (investing) on the emergence of a streak, with about 40% mentioning this trigger point.
A third study in the research looked at strategic reactions in basketball. There was evidence to support the rule of three in the shooting percentages of every player of the team. That is, when a sequence of three is possible, e.g., misses or makes, likelihoods changed after those three misses or makes. The player shooting three makes in a row had a greater likelihood of missing, whereas the player with a three-time miss had a greater chance to make.
Consider the case of predicting which team is likely to win an upcoming game. If the rule of three operates in team competitions, as the Google search data suggest, then gamblers will tend to bet for (against) teams that have won (lost) their last three games. If so, the odds, which adjust to account for the amount of money bet on each team, will be too far in favor of the team that has won three games in a row. In other words, a team that has won three games in a row will be overpriced. In the face of this bias, a savvy gambler (who is aware of the rule of three) might do well to bet against teams that have won three games in a row and bet for teams that have lost three games in a row. The extension to predicting the success of any entity (a student, a firm, a security) follows naturally, with similar opportunities to take advantage of the misevaluation of those entities that have had three successive failures or successes.
Given the likelihoods judgments, such as a player making the next shot, are there downstream consequences of these judgments? Will it impact how resources are directed to or away from the player? These inferences of a streak may be found to drive behavior, such as the basketball manager judging performance or assigning tasks to players given their streak perception. This too can play out in the world of work.
The employee with three consecutive positive events might be tagged as a high performer. The opposite could be thought of an underperformer, who may in fact simply be on a bad streak, set to reverse the next time. Shu surmises, "It comes back to coin flips or gambling to predict the next outcome like in the case of stocks. If something happens three times in a row, you have to step back and look at what's behind the third event. Am I reacting to the fact that it happened three times or is there a greater truth in the situation? Is it statistically possible?" She mentions that with coin flips, you can't predict after three really, and with stocks it is even harder. "If a hot streak happens after three stock price increases, people will jump in to buy, and so it will go up, she continues. "So what is the best way to view these phenomenon? At some point the stock will come down."
The power of three is seen across many domains as diverse as religion, mythology, architecture and law. The idea of three is used by comedians, ministers, and salespeople. The number three may hold a special significance for human judgment, the authors conclude.
"The Rule of three: How the third event signals the emergence of a streak" by Suzanne Shu of SMU Cox School of Business and co-author Kurt Carlson of Duke University was published in the September 2007 issue of Organizational Behavior and Decision Processes.
# # #