Introduction

It is frequently conjectured that sports participation does not only improve individuals’ physical and mental health, but also positively affects other domains of life. One prominent example is the labor market where physically active individuals are reported to perform better than inactive individuals regarding both income and job quality (e.g., Cornelißen & Pfeifer, 2009; Lechner, 2009; Cabane, 2011; Hyytinen & Lahtonen, 2013; Lechner & Sari, 2015; Lechner, 2015; Cabane & Lechner, 2016). Besides higher levels of productivity due to improved health, the setting in which sports participation is embedded may also foster qualities which prove advantageous beyond sports, e.g., perseverance, social interactions, or coping with stress. Although participants in the present study are several years away from entering the labor market, there is broad evidence on the importance of skills acquired during childhood and adolescence for outcomes in later life, including labor market success (see, e.g., Almlund, Duckworth, Heckman, & Kautz, 2011, for an exhaustive review of the literature). Moreover, sports participation in adolescence can prove advantageous when applying for an apprenticeship after finishing school (Piopiunik, Schwerdt, Simon, & Woessmann, 2020) and is also found to be positively related with labor market participation and job characteristics in the late 20 s and early 30 s, respectively (Cabane & Clark, 2015).Footnote 1

Sports, in particular at a professional level, is frequently associated with competition. At the same time, cooperation of individuals regularly poses a crucial determinant for success—not only in team sports. At first glance, this may appear contradictory: While competition is a zero-sum game with mutually exclusive goals from the participants’ point of view as the prize is fixed ex-ante, cooperation is usually achieved by overcoming a social dilemma, i.e., individuals sacrifice their personal benefit to augment the outcome for the entire group (Sethi, 2010). Nevertheless, both competition or cooperation are vital elements of success in sports and other domains of life, although not simultaneously. This notion is reflected by the portmanteau “coopetition” which has been introduced to management literature by Nalebuff and Brandenburger (1996; see also Nickisch & Nalebuff, 2020). In terms of individual labor market outcomes, both competitiveness and cooperativeness may be valuable—depending on the specific situation—as (a) firms increasingly rely on team work and (b) promotions are frequently based on tournament-like relative performance evaluations (Lazear & Shaw, 2007).

The present paper focuses on the institutional setting of an elite school of sports (ESS) rather than on individual levels of sports participation. These institutions are part of the German sport system and provide the opportunity to reconcile compulsory school attendance with extensive practicing necessary to pursue a career in professional sports and participation in (inter-)national competitions (DOSB, 2021). Although this track is demanding, there is no evidence that individuals attending an ESS perform academically worse than students at regular schools or suffer from mental conditions due to stress (Güllich & Richartz, 2016; Breithecker & Brandl-Bredenbeck, 2017). To contribute to the literature studying the particular group of pupils attending an ESS, we examine competitive and cooperative behavior in children attending an ESS as compared to their peers attending regular upper secondary schools. Our research is guided by the following directed hypotheses:

(H1).

ESS students are more cooperative than students attending regular schools.

(H2).

ESS students are more competitive than students attending regular schools.

There are previous studies utilizing methods from experimental economics which report differences in behavior between individuals playing sports and those not playing sports regarding, in particular, cooperativeness (Celse et al., 2017) and competitiveness (Eber & Willinger, 2004; Eber, 2006; Fallucchi et al., 2020). It is important to emphasize that these studies rely on correlations and do not draw causal conclusions on the effect of sports participation; the same applies to our approach which focuses on the institutional setting of the ESS in the first place. More precisely, the goal of the present paper is to investigate competitive and cooperative behavior among students attending an ESS and regular upper secondary schools, respectively. Our initial point of view is that there are no differences between both groups.

Procedures

The results presented below are based on data gathered as baseline for a study which examines the capacity of a sports-based intervention in fostering personal and social skills. Participants were children aged 10–13 years (N = 184; meanRegular schools = 11.7 years (standard deviation [SD]Regular schools = 0.46 years); meanESS = 11.7 years (SDESS = 0.28 years)); difference statistically not significantly different from zero (U(nRegular schools = 158, nESS = 26) = 2246, p > 0.1, two-sided Mann–Whitney U test) attending four regular upper secondary schools and one ESS in the same city in Germany. The experiments utilized to gather data were conducted on class level. At the time of the study (December 2016), the composition of the classes had been in effect for approximately one and a quarter years. Although the outcomes were based on both individuals’ decisions and decisions by other students within the same class, it has to be noted that participants were guaranteed anonymity, i.e., individual decisions were not disclosed at any time. This paper reports the results only of the first wave of a larger study, prior to the intervention in the treatment group (for details, see Haas et al., 2021b, Woll et al., 2018; see also our companion paper Haas et al., 2021a, which reports results on the longitudinal assessment of the intervention on cooperative behavior on students at regular schools). The study was approved by the superintendent of the local school district and the Board of Ethics of the Karlsruhe Institute of Technology. To obtain measures of various preferences and traits, we utilize experimental methods which are established in economics (see, e.g., Sutter, Zoller, & Glätzle-Rützler, 2019, for a comprehensive review on results of economic experiments conducted with children and adolescents). The prominent feature of this approach is the use of monetary incentives which induce participants to respond truthfully (Bardsley et al., 2010). In the present paper, we limit the scope to competitive and cooperative behavior.

To study individuals’ competitiveness, they were asked to carry out a real-effort task which consisted in replacing letters of a batch of words by numbers for 2 min (adopted from Erkal, Gangadharan, & Nikiforakis, 2011). The number of correctly enciphered letters provided a signal sk of individual k’s effort. Prior to the working stage, they were asked to choose between two compensation schemes, either a piece-rate or a tournament prize. The piece-rate scheme yielded a fixed compensation of 3 tokens for each correctly enciphered letter; thus, an individual’s final payoff was 3 × sk tokens (capped at 300 tokens). The payoff under the tournament scheme, however, was based on the comparison of individual k’s number of correctly enciphered letters, sk, and the number of correctly enciphered letters sl of another individual l who had also chosen the tournament scheme. The individual with the higher number of correctly enciphered letters received the winner prize of 600 tokens and the other individual received the loser prize of 0 tokens; ties were broken randomly. Since the payoff in the tournament scheme was based on the relative performance of two individuals and features a prize spread which imposes strong incentives for winning due to the substantial difference between the winner and loser prize, individuals who chose the tournament scheme were deemed competitive.

A linear public good game was utilized to study individuals’ willingness to cooperate. Each group consisted of four individuals (indexed i = 1, …, 4) who received an endowment of 120 tokens each. They decided independently and simultaneously how to allocate their individual endowment to a private and a collective account, respectively (in increments of 24 tokens). Contributions to the collective account by individual were denoted by ci ∈ {0, 24, 48, 72, 96, 120}. Tokens which were not contributed to the collective account, i.e., 120 − ci, were transferred to the individual’s private account. Each token in the private account yielded a payoff of 2 tokens to each individual. For each token allocated to the collective account, every individual of the group received a payoff of 1 token. Thus, individual i’s payoff πi (in tokens) was denoted by

$$\pi _{i}=2\times (120-c_{i})+(c_{1}+c_{2}+c_{3}+c_{4})$$

From an individual perspective, a contribution of 1 token to the collective account yielded a payoff on 1 token, while the same contribution to the private account yielded a payoff of 2 tokens. From the entire group’s perspective, however, a contribution to the collective account yielded a payoff of 4 tokens. Hence, the individually optimal strategy—following the standard assumption in neoclassical economic theory of individuals striving to maximize their own material gain—was to allocate the entire endowment to the private account (i.e., ci = 0 tokens), whereas the total payoff of the group was maximized when all individuals contributed their entire endowment to the collective account (i.e., ci = 120 tokens)—which raised a social dilemma. Individuals who refrained from taking advantage of individual incentives and contribute to the collective account instead were, therefore, deemed cooperative.

At the end of the session, one experiment was chosen for payoff. Tokens earned in this particular experiment were converted into euros (€) at an exchange rate of 75 tokens = 1 €; additionally, each individual received a fixed participation fee of 2 €.

Results

Individuals’ competitiveness is indicated by their choice of the tournament scheme in a real-effort task. At regular schools, 68 of 158 individuals (43%; 95% confidence interval [95% CI]: [36%, 51%]) are deemed competitive, whereas this characterization applies to 18 of 26 individuals (69%; 95% CI: [50%, 83%]) at the ESS. The difference in the propensity to choose the tournament scheme between the ESS and regular schools of 26%-points (95% CI: [6%-points, 44%-points]) is small (Cramér’s V = 0.18; 95% CI: [0.04, 0.32]) and statistically significant (χ2(1, N = 184) = 5.15, p = 0.02). Yet, it is important to note that difference in the number of correctly enciphered letters between both groups is statistically not significantly different from zero (meanRegular schools = 17.03 letters (SDRegular schools = 8.71 letters); meanESS = 18.96 letters (SDESS = 7.37 letters); U(nRegular schools = 158, nESS = 26) = 1682, p > 0.1, two-sided Mann-Whitney U test). Parametric results obtained by a univariate Probit regression (see specification (1) of Table 1) confirm this result as the coefficient of the binary variable representing the ESS is positive and statistically significant. Adding a dummy variable for female individuals in specification (2) yields a negative and statistically significant coefficient. In line with previous findings reported in the literature (Niederle & Vesterlund, 2007; Niederle, 2017), this indicates that female individuals are substantially less inclined to engage in competitions. Looking at the willingness to compete by gender and group in Fig. 1 shows an interesting result: The share of female and male individuals who choose the tournament compensation scheme and, hence, are deemed competitive correspond by and large to observations frequently reported in comparable studies. It is worth noting, however, that in the ESS—even though a gender gap in competitiveness occurs—the overall propensity to compete is substantially higher than in regular schools. A visual inspection of Fig. 1 reveals that female individuals at the ESS are about as competitive as male individuals at regular schools. The design of the current study, however, does not allow to disentangle the influence of environmental circumstances from selection effects.

Table 1 Competitiveness: choice of tournament scheme
Fig. 1
figure 1

Frequency of tournament scheme chosen by group and gender. ESS elite school of sports

Contributions to the collective account in the public good game indicate individuals’ willingness to cooperate. For simplicity, we denote contributions in this section not in tokens but in coins (1 coin = 24 tokens), i.e., contributions range from 0 to 5 coins. An overview on average contributions in both groups is provided in Fig. 2. At regular schools, individuals contribute on average 2.41 coins (95% CI: [2.22 coins, 2.84 coins]) which corresponds to almost half of their initial endowment. This magnitude is in line with results from other experiments which frequently report contributions ranging from 40% to 60% of the initial endowment in one-shot public good games and the first round of repeated public good games, respectively (Ledyard, 1995). Contributions to the collective account in the ESS amount to 3.31 coins (95% CI: [2.59 coins, 3.78 coins]) or about two-thirds of the initial endowment on average. This indicates that the individuals at the ESS are on average more inclined to cooperate than their counterparts at regular schools. A one-sided Mann–Whitney U test indicates that the mean contributions to the collective account at the ESS are statistically significantly higher than at regular schools (U(nRegular schools = 158, nESS = 26) = 1211.5, p < 0.001). According to the measure by Algina, Keselman, and Penfield (2005), this effect is large (δt = 1.07; 95% CI: [0.42, 1.47]). This finding is supported by a univariate linear regression reported in Table 2 as the coefficient of the binary variable representing the ESS is positive and statistically significant. Unlike in the case of competitiveness, the coefficient of the binary variable representing female participants is not statistically significantly different from zero. Utilizing a Tobit model as robustness check yields the same result with regard to the propensity to cooperate between individuals attending the regular schools and the ESS.

Fig. 2
figure 2

Contributions to the collective account by group (with number of observations). Error bars indicate the 95% confidence interval. ESS elite school of sports

Table 2 Cooperativeness: contributions to the collective account

Discussion

Our results show that individuals who attend the ESS behave differently from their counterparts at regular schools. ESS students in our study are both more competitive and more cooperative. Although this finding may seem contradictory at first glance, it can be justified by the fact that both properties provide a crucial prerequisite for success in sports—depending on the specific situation. Yet, it has to be noted that this study focuses on the institution of the ESS, not on individual levels of sports participation. From a methodological perspective, borrowing an approach from experimental economics links individuals’ decision to real consequences by using variable payment. This alleviates the potential issue of demand effects which are difficult to avoid in self-answered questionnaires. Our results are subject to limitations: First, the cross-sectional nature of the data set does not allow us to infer causal conclusions. Second, only one ESS was part of the study; generalizations are, therefore, to be treated with caution. Third, other aspects such as more shared time in the ESS because of additional trainings could have an influence on the results. Fourth, since this paper is a brief research communication, we refrained from a comprehensive discussion of existing literature that could shed light on further aspects of interest. Nevertheless, our study using incentivized economic experiments provides highly interesting results that should be further analyzed in future studies.