When Losing Leads to Winning 1 When Losing Leads to Winning

In general, the further individuals, groups, and teams are ahead of their opponents in competition, the more likely they are to win. However, we show that through increasing motivation, being slightly behind can actually increase success. Analysis of over 6,000 collegiate basketball games illustrates that being slightly behind increases a team’s chance of winning. Teams behind by a point at halftime, for example, actually win more often than teams ahead by one. This increase is between 5.5 and 7.7 percentage points, or half the proverbial home team advantage. Ancillary results underscore the motivational effects of being slightly behind. Most of the scoring boost for teams slightly behind occurs right away. An experiment further demonstrates that being slightly behind increases effort, while also casting doubt on the alternative explanation that being slightly ahead induces complacency. Taken together, these results illustrate that losing can sometimes lead to winning. When Losing Leads to Winning 3 Intuition suggests that being ahead in everything from scientific competitions to sports should increase the likelihood of winning. Teams which are ahead in football, groups who are further along towards sequencing a genome, or horses ahead in a race should be more likely to win. Indeed, basketball, baseball, football and hockey teams which are ahead early in the game win over two thirds of the time (Cooper, DeNeve, & Mosteller 1992) and teams that are further ahead tend to win more (Stern 1994). But could being slightly behind actually increase success? Stated more strongly, could losing sometimes lead to winning? We suggest this possibility based on research regarding motivation. People have a drive to achieve that is driven, in part, by goals (Atkinson 1957; Lewin 1951; Miller 1944). Though scoring a touchdown or sequencing part of the genome requires the same amount of effort whether a team or person is ahead or behind, research suggests there is a goal gradient, whereby humans and animals expend more effort the closer they get to achieving a goal (Hull 1932; Brown 1948; see Helizer 1977 for a review). Animals run faster when they are closer to a food reward (Hull 1934), and people work harder when a task is almost complete (Kivetz, Urminsky, & Zheng 2006). Goals can act as reference points (Heath, Larrick, & Wu 1999), and due to the diminishing sensitivity of the value function in prospect theory (Kahneman and Tversky 1979), people should exert more effort when they are close to their goals. Though most work in this area has looked at individual goals and motivation, these ideas may have important implications for competition. In many competitive situations, individuals or teams may use their opponent as a benchmark. If a competitor’s goal is to win, or beat their opponent, then being slightly behind during a moment of When Losing Leads to Winning 4 reflection may lead them to work harder to achieve the desired outcome. Consequently, we argue that relative to being slightly ahead, being slightly behind an opponent can increase the likelihood of winning. Teams that are losing by a little, for example, may actually be more likely than their opponents to win. We test these possibilities using field data from actual competitions and a laboratory experiment. The studies demonstrate that being slightly behind increases success: It boosts effort and leads teams to be more likely to win. The experiment casts doubt on an alternative explanation based on complacency, and illustrates the moderating role of self-efficacy in these effects. FIELD STUDY: COLLEGIATE ATHLETICS Our first test uses NCAA basketball data to examine how a team’s likelihood of winning varies based on the point differential at halftime. We expect that being slightly behind should increase a team’s chance of winning. If this boost is large enough, teams behind by a point at halftime may actually be more likely than their opponents to win. Finally, if this effect is driven by goal proximity, as we suggest, then we should observe such teams scoring more points than their opponents right after halftime. Method We used ESPN.com to collect data on all NCAA basketball games from 20052008 for which play-by-play data was available. We focus on all games where the point differential at halftime was +/-10, which resulted in 6,572 games. We focus on halftime When Losing Leads to Winning 5 because all players may not know the exact score at other points in the game, and the halftime break gives them time to recognize the score. There are no quarters in college basketball and the endogeneity of timeouts make them less useful to examine. We collected which team was at home, the time and point value of each basket scored throughout the game, and team records throughout the season. Results We first examine the relationship between score differential at halftime and winning percentage. Not surprisingly, in general, the further teams are ahead, the more likely they are to win (Figure 1A). Teams up by six, for example, win about 80% of the time. Furthermore, this relationship is quite linear. Every two points better a team is doing relative to its opponent at halftime is associated with an approximately 8 percentage point increase in the probability of winning. There is a strong discontinuity, however, around zero. On a descriptive level, rather than having a winning percentage that is 8 percentage points less than teams ahead by a point (i.e., 46% vs. 54%), teams that are behind by one point are actually more likely than their opponents to win (triumphing in 51.3% of games, see Figure 1B for a close-up with error bars). This is somewhat noteworthy given the many reasons why teams losing by one should be less likely to win. They are not as good as their opponents (lower winning percentage), for example, and mechanically, have to score two more points than their opponent in order to emerge victorious. To formally test whether the difference in winning percentage is statistically different from “expected,” we perform a more structured analysis using a regression1 On average, teams score approximately 69 points in a game. When Losing Leads to Winning 6 discontinuity model (Thistlethwaite & Campbell, 1960). This method fits a quintic regression line through all the data to control for the expected effects of halftime score difference on winning percentage. A dummy variable for being behind by one point or more is also included which allows for a discrete jump in the polynomial function for being behind. Such analyses are useful because they allow us to make causal inferences about the effect of being behind, but such claims can only be made about points near the discontinuity (i.e., losing by one). As one moves further from the discontinuity, other potential factors make such inferences less tenable. Regression results indicate that being slightly behind significantly increases a team’s chance of winning, and does so by 5.5 to 7.7 percentage points depending on the specification (Table 1). The overall result holds, and in fact gets larger, when controlling for team quality (winning percentage that year), home-team advantage, second half possession arrow (who gets the ball to start the second half), and team fixed effects. Teams behind by one, for example, win more often than expected (51.3% vs. 46%, t(367) = 2.04, p < .05) Furthermore, consistent with our hypothesis, the increased probability of winning is driven largely by the losing team exerting more effort immediately after they are slightly behind. Teams that were down by one point at the half, for example, scored almost 1.2 points more than their opponents during the second half, but nearly half this difference occurred in the first four minutes (Figure 2). When Losing Leads to Winning 7 Discussion Can losing lead to winning? Analysis of NCAA basketball games suggests it can. Not only did being behind at halftime increase a team’s chance of winning (between 5.5 and 7.7 percent), but compared to their opponents, teams that were behind by one were actually more likely to win. The boost is comparable to the increase in winning percentage from being ahead by two points at the half, or about half the size of the proverbial home team advantage (Cooper et al. 1992). Ancillary results further indicate that the increased winning percentage is driven by the losing team exerting additional effort directly following the detriment. Teams that were slightly behind at halftime scored more in the second half, but did so most strongly right after the break. We can also cast doubt on some alternative explanations. Though one could argue that being behind might induce a strategy change, such changes are unlikely if a team is only down by a point. Further, any change in strategy should not necessarily lead teams to be more likely to win. Similarly, though one could argue that being behind might encourage people to take risks, it is unclear that this should increase success. If anything, riskier play could just as easily lead teams to lose. As with most field data, however, it is difficult to rule out all alternatives. Though unlikely, one could argue that teams which are slightly behind win more because referees treat them differently. Alternatively, one might suggest that the second half scoring differential is driven by winning teams becoming complacent. Because one team’s effort directly affects the other’s output, however, it is difficult to tease apart these mechanisms in the field data. There is no evidence of reduced scoring by the winning 2 Analyses also indicate that teams that are slightly behind are no more likely to win if their coach is more experienced, casting doubt on the possibility that the effect is driven by strength of the coach’s speech or halftime strategy adjustment. When Losing Leads to Winning 8 team, for example, but even if there was, it would be difficult to tease apart complacency from better defense by the team down by one. Consequently, we conducted an experiment that directly tests how competitive feedback influences effort. EXPERIMENT: COMPETITIVE FEEDBACK AND EFFORT We examined motivational effects of competitive feedback more closely by having participants engage in a short competition task. The task was divided into two periods. Between periods, some participants were given competitive feedback: that they were slightly behind, tied, or slightly ahead of their opponent. Control participants were not given competitive feedback. This design allowed us to examine both whether being slightly behind increases effort and whether being slightly ahead induces complacency. We also measured self-efficacy to further examine the mechanism behind these effects. Self-efficacy (Bandura, 1977; 1997) describes the belief that one can achieve desired outcomes. Consequently, we argue that the motivating effects of being slightly behind should be more likely to occur among individuals with high self-efficacy. Method Participants (N= 114) were paid $10 to complete a set of experiments. First, they filled out an 8-item general self-efficacy scale (e.g., “In general, I think that I can obtain outcomes that are important to me,” 1 = strongly disagree, 5 = strongly agree, Chen, Gully, & Eden 2001). Then, after completing some filler tasks, they were told they would engage in a short competition with another participant and whoever had the most When Losing Leads to Winning 9 points would win an additional $3. The game involved pressing the ‘a’ and ‘b’ keys in succession as quickly as possible. Pressing the combination in the correct order scored a point. The game had two, twenty-second periods divided by a short break. The only difference between conditions was the information participants received between the periods. In the competitive feedback conditions they were told they were slightly behind (-1 point), tied, or slightly ahead (+1 point) of their opponent. In the control condition, they were given no information about their relative performance and were just told that the second period was about to begin. Results and Discussion We examined how competitive feedback influenced effort by taking the difference in participants’ output across the periods and comparing it across conditions. Responses to the self-efficacy scale were averaged across items, and we performed a median split on participants’ scores for ease of explication (results are the same when a continuous measure is used). Can being slightly behind an opponent lead people to exert more effort? Results suggest it can (Figure 3). As predicted, competitive feedback influenced effort, F(3, 106) = 2.56, p < .06, but was moderated by self-efficacy, F(3, 106) = 2.82, p < .05. Planned contrasts revealed that being slightly behind opponents led individuals with high selfefficacy to exert more effort. Participants informed that they were slightly behind, and who had high self-efficacy, exerted more effort than participants in any of the other 3 Centering self-efficacy and regressing increased effort on dummies for each feedback condition and those dummies interacted with self-efficacy revealed that being behind increased performance, t(106) = 2.10, p < .04, and that this effect was stronger for individuals with higher self-efficacy, t(106) = 1.77, p = .08. No other effects reached significance. When Losing Leads to Winning 10 conditions, all ts > 2.19, ps < .03. All other conditions were equivalent, all ts < 1, ps > .40, indicating that being slightly ahead did not decrease effort. These findings further demonstrate the motivating impact of being slightly behind; merely telling people they were slightly behind an opponent can lead them to exert more effort. The results also cast doubt on the possibility that the field data results are driven by winning team complacency. Though people may get complacent when they are far ahead, this study shows that being slightly ahead did not decrease effort. Finally, the results demonstrate the important role of self-efficacy, or beliefs in one’s ability to achieve goals, in driving people who are behind to work harder.