The target in recurve archery is 70 meters away. It has 10 concentric rings, with the center scoring 10 points and the outside circles scoring less (Figure 1). No points are awarded if an arrow misses the target completely.

Tournaments comprise two phases: the “ranking round” and the “elimination round.” During the ranking round, athletes shot a total of 72 arrows in order to determine their rankings. During the elimination rounds, athletes of comparable rankings compete against one another. The highest-ranked athlete competes against the lowest-ranked athlete; the second-highest-ranked athlete competes against the second-to-last competitor, and so on. Each match ends with the loser falling out and the winner advancing. This continues until just two competitors compete in the final. Semifinal losers compete for third place.

In a match, athletes compete in a best-of-five set format. Each set contains three arrows for each participant. The winner of a set receives two points, while a tie awards both competitors one point. The first player to reach six points wins. Each set begins with the higher-ranked athlete shooting first, and then the athlete with fewer points shoots first in the next set.

In current study, elite athletes are those who compete in international (London 2012, Rio 2016, Tokyo 2020 Olympics, Yankton 2021 Hyundai World Archery Championships) and continental championships (Antalya European Grand Prix 2021). Athletes who were unable to participate in the above events were excluded due to the unavailability of competition data. We have collected arrow-by-arrow data from various archery events such as the London 2012, Rio 2016, Tokyo 2020 Olympics, the Yankton 2021 Hyundai World Archery Championships, and the Antalya European Grand Prix 2021. This data was obtained from the official World Archery Federation website.¹ For the London 2012 Olympics archery game, we obtained the data from the archived website. ² After collecting the data, we examined it thoroughly for 10 years from 2012 to 2021. We discovered 6,374 instances of three consecutive shots, totaling 19,122 shots. For each set of three shots, we recorded the player’s identity, the points achieved, the shooting order, whether the player participated in the Olympics or other games, and the set points in the game when the shots were taken.

We thoroughly examined the arrow-by-arrow data spanning 10 years from 2012 to 2021. In total, we discovered 6,374 instances of three consecutive shots, totaling 19,122 shots. For each set of three shots, we recorded the points achieved, the player’s identity, the shooting order, whether the player participated in the Olympics or other games, and the set points in the game when the shots were taken.

Our analysis is aimed at understanding whether the previous poor results or unfavorable game status affect subsequent players’ performance and whether this effect is heterogeneous according to gender. Our empirical strategy is quite simple: We first analyze the respective scores of archers in their third arrow in relation to the outcome of the previous two shootings. Specifically, we compared the points scored on the third shot under four scenarios: after two consecutive 10s (Momentum A, Hit-Hit), after a first shot of 10 and a second miss of 10s (Momentum B, Hit–Miss), after a first miss of 10 and a second shot of 10 points (Momentum C, Miss-Hit), and after two consecutive missing 10s (Momentum A, Miss–Miss). If a cold-hand effect exists, we expect to observe lower scores on the third shot following two consecutive 10s than following the other combinations. If an archer’s performance significantly decreases after missing 10 points compared to those hitting 10 points, we consider it to be the existence of a cold-hand effect (Table 1). To investigate whether men and women respond differently to the outcome of the first two arrows, we conducted a regression analysis. We used the performance of the third arrow as the dependent variable and the outcome of the previous two shootings (momentum type) as independent variables. The momentum type is crucial for understanding the cold-hand effect, with “Hit–Hit” scenarios potentially reinforcing performance, while “Miss–Miss” scenarios are expected to exacerbate performance declines. A deeper exploration of how these scenarios interact with gender could offer richer insights. We also included the gender of the player as an interaction term. Next, we performed the same analysis using game status as an additional independent variable. To account for differences in players’ abilities and heterogeneity, we used both players’ scores in ranking rounds.

Since the dependent variable is a count of integers, we used a Poisson general linear model (GLM), which detected under-dispersion (Zuur et al., 2009). To correct for this, we used a quasi-GLM with the variance given by φ × μ, where μ is the mean and φ is the dispersion parameter estimated at 0.231. This adjustment means that all standard errors were multiplied by 0.481 (the square root of 0.231). It should be mentioned that under-dispersed count data can be handled using the quasi-Poisson approach (Hostetler et al., 2012; Otterbeck et al., 2019). While a negative binomial regression model could address potential overdispersion, the quasi-Poisson model was chosen due to its better fit for the specific data characteristics observed in this study.

For all GLM analyses conducted in this paper, the “Multicomp” package was used to conduct pairwise comparisons of means. The statistical software package R (R development core team 2018) was utilized for statistical analyses and graphing. All tests were two-tailed, and a p-value below 0.05 was considered significant.

Our final method is a Poisson regression-based test for the cold hand that examines how the performance of the last arrow per set depends on the performance of the previous two shots.

We also interact the player’s gender with performance on the previous two arrows and game status to test for gender differences in response to past performance and intermediate game status.

Our Poisson regression takes the following form:

Arrow 3 i , with the number of points on the last arrow per set i , is Poisson distributed with a mean of μ i .

The statistically significant evidence of a cold-hand effect is presented in Table 2. This is supported by the negative and significant coefficients that are linked to previous worst performances [t = −12.862 e ^β = 0.878 (0.86–0.90), p < 0.001], specifically missing the bullseye twice in a row. It indicates significantly lower scores (12.2%) after two missing 10-point shots compared to the case after two 10-point shots.

Moreover, the conditional score of the third shot after two missing 10-point shots is significantly higher than the average score following the other cases (p < 0.001, Figure 2A).

In conclusion, Table 2 shows that archers tend to shoot worse after missing the bullseye twice, regardless of the player’s gender, general skill, game status, shooting order, or game type. The cold-hand effect, while statistically significant, also presents practical implications for coaching strategies, suggesting that interventions should focus on enhancing resilience after consecutive poor performances.

In Poisson regression, we see that the player’s general skill has a highly significant effect on their conditional score for the third shot (Figure 2F). This takes into account the size and statistical importance of heterogeneity. And this control variable is the most important thing in these regressions. Also, the “Game Type” binary variable shows a higher score on the third shot per set in Olympic games than in Championship games (Figure 2C). Both the player heterogeneity and game type effect are significant at the 0.0001 level.

Furthermore, when in game status C1 (losing the current set will result in the loss of the entire game while winning the set will have no impact on the outcome of the game), archers perform worse on the third arrow than when it is in game status C0 (winning or losing the current set has no impact on the outcome of the game, Figure 2D). However, none of these differences among statuses were statistically significant (Table 2). Similarly, the order of archery had no significant effect on the performance of the third arrow (t = 0.017, p = 0.986).

By including these variables in our study, we improve the reliability of our findings and reinforce the conclusion that the cold-hand phenomenon is present in repeated archery attempts.

Although male archers scored slightly higher than female archers on the last arrow of each set (Figure 2B), this difference was not significant (t = 1.214, p = 0.225). Table 2 shows that the coefficient (0.035) for the interaction between gender and the previous worst performance (missing bullseye twice) is positive. This information indicates that when archers get equal bad scores on the first two arrows, male athletes do much better on the third arrow than female athletes, with a 3.5% difference (p < 0.001, Figure 3A).

Another significant aspect of the interaction effect is between gender and game status. In a situation where losing can only be avoided by winning, the third arrow performance gap between male and female athletes significantly increased by 2.4% (Table 2). This suggests that the decline in the performance of female athletes on the third arrow is considerably higher than that of male athletes when on the verge of failure (game status C1, Figure 3B).

Published reports suggest that men generally have higher self-confidence than females in competitive situations (Lirgg, 1991; O’Connor et al., 2020). Research has demonstrated that receiving negative feedback can diminish self-assurance, especially in women (Roberts and Nolen-Hoeksema, 1989), perhaps leading to decreased performance (Woodman and Hardy, 2003). Lirgg (1991) emphasized the possibility that males tend to overestimate their performance, whereas females may underestimate their performance, which can lead to differences in confidence ratings (Lirgg, 1991).

Undoubtedly, self-confidence must play a role in archers’ ability to bounce back from adversity. Self-confidence will impact archers’ subsequent motivation and performance after an abysmal performance or facing near failure. According to prior theory and research, males may be more resilient to such a loss due to a higher level of self-confidence. On the other hand, because females’ self-confidence may be more sensitive to feedback, a setback may be more likely to reduce females’ self-confidence and interfere with subsequent performance.

Male athletes typically display lower levels of anxiety than female athletes (Krane and Williams, 1994; Correia and Rosado, 2019). Female athletes exhibited higher levels of competitive trait anxiety (Kristjánsdóttir et al., 2018) and higher levels of worries (O’Donoghue and Neil, 2015).

The intensity and direction of competitive trait anxiety were investigated by Perry and Williams (1998) based on gender disparities (Perry and Williams, 1998). While cognitive anxiety levels were not different between males and females, females were more likely to have a debilitating interpretation of it.

Females reported that cognitive anxiety was a hindrance to tennis performance. In addition, males more often interpreted somatic and cognitive anxiety as facilitative to competitive performance. The female archers exhibited higher levels of felt arousal and cognitive anxiety than the male archers.

Gender differences in levels of competitive trait anxiety might contribute to the understanding of potential gender differences in archers’ ability to overcome a disadvantageous situation.

Studies have shown that males tend to employ problem-focused coping strategies more frequently, while females rely more on emotion-focused coping strategies (Tamres et al., 2002; Hammermeister and Burton, 2004; Anshel et al., 2009). Psychological factors of competitive anxiety are related to coping strategies (Hoar et al., 2010). Active coping, planning, effort, and suppression of competing activities are examples of problem-focused coping methods that have a positive correlation with positive affect (Watson et al., 1988), which is a reflection of enjoyable engagement. Positive affect was also found to be inversely correlated with behavioural disengagement and wishful thinking (Gaudreau et al., 2002).

These findings have implications for archers’ responses to frustration in a match. If females are more likely to resign after a failure and males are more likely to engage in problem-focused strategies, males are more equipped to overcome a loss.

Extensive literature confirms the presence of a significant and enduring disparity between genders in terms of competition (Frick, 2011; Wozniak, 2012; Deaner et al., 2016; Birk et al., 2019). Trait competitiveness measures have shown that males are more competitive than females. Studies showing females as more competitive have had small samples of high-level athletes, which are not typical of females. In contrast, the male samples were more prominent and more representative. Although the results are inconsistent, there seems to be reasonable support for this gender difference. A high level of competitiveness would likely contribute to an archer’s motivation to overcome a loss.

Summing up, differences exist between the two gender groups in a variety of fields such as confidence, anxiety, coping strategy, and competitiveness. Such variation may explain the causes of sex differences in response to failures.

From a physiological perspective, poor performance in recurve archery was correlated with high real-time heart rate (Lu and Zhong, 2023). Some heart rate variability (HRV) studies have indicated that females may exhibit a higher overall complexity of heart rate dynamics than males (Ryan et al., 1994). Women show a greater mean heart rate (Koenig and Thayer, 2016). It should be mentioned that a higher heart rate—which indicates an increase in psychological stress—is associated with lower scores. Archery is a sport that requires fine movement control, and postural stability is considered an important variable in achieving high performance (Sarro et al., 2021). An increased heart rate can accelerate the imbalance of the human body posture and reduce the ability of people to maintain body balance posture. Subsequently, it affects shooting performance.

Furthermore, the gender-specific autonomic differences may contribute to the weaker resilience to setbacks in females. For example, during adolescence, girls are 3 times more likely to develop post-traumatic stress disorder (PTSD) than boys (McLaughlin et al., 2013). Sex or gender differences in cognitive styles contribute to resilience for post-traumatic stress disorder (PTSD) and other mood disorders.

The absence of qualitative data on psychological factors and athletes’ backgrounds limits the depth of the conclusions that can be drawn. Unfortunately, the constraints of our data prevent a thorough investigation into the disparities in how men and women respond to failure across various sports, the labour market, and marital outcomes. The results may not be fully applicable to other sports or competitive contexts where the dynamics of pressure and feedback differ significantly. Additionally, the dataset lacks detailed information on psychological interventions or external support factors, which could play a significant role in shaping athletes’ responses to pressure. Furthermore, our analysis is limited in its capacity to elucidate the underlying causes for the observed gender differences in the archery context. The observed gender differences in response to consecutive poor performances may be influenced by other factors such as experience level, access to psychological support, and prior exposure to high-pressure situations, which should be considered in future research. It is recommended that future studies, equipped with more extensive data, delve deeper into these issues to provide a more comprehensive understanding of high-level performance under stress.