Do people who prefer competition perform better in a competitive environment?

Master’s Thesis

Do People Who Prefer Competition Perform Better in a

Competitive Environment?

Name: Xiran Ma

Student Number: 11826118

Program: Economics, track BE & GT

Supervisor: Ailko van der Veen

Date: 2018-8-10

15 ECTS

Statement of Originality

This document is written by Student Xiran Ma, who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the content.

Do People Who Prefer Competition Perform Better in a Competitive Environment?

ABSTRACT

There is no uniform conclusion in the literature on the relationship between competition and performance, in some cases, competition can promote people's performance, but sometimes competition has no effect on performance. The impact of competition on performance may changes due to the difference of gender and achievement goals in each experiment. This paper examines whether an individual’s environmental preference can affect people's performance in a competitive environment. In a laboratory experiment, 40 participants performed a real task under non-competitive piece rate and a competitive tournament environment, after that they were asked about their environment preference. Experiment results show that the performance gap between people with different preference is not significant. Individual’s performance in a competitive environment is rarely affected by preference, especially for those who prefer a non-competitive environment. To conclude, people’s preference about environment has little effect on performance.

Table of Content

Introduction

Competition is an important part of our lives, most people have encountered a certain amount of competition since the beginning of school, and it appears not only in education but also in employment. During competition participants maximize their own personal interests, and their aim is to pursue attractive goals and strive to outperform each other. On the one hand, competition is the full realization of human potential. People will work and learn more intensely in a competitive environment and, at the same time, they can obtain more realistic evaluations. More importantly, competition is considered to be beneficial to the growth of talents, enabling them to be discovered and recognized by society. However, on the other hand ,it cannot be ignored that competition also produces some negative effects. Failure in competition reduces participants’ enthusiasm a lot, and mental stress from competition is also a very important negative result. This view has been proved by studies, when people in the face of stress, especially chronic stress, more resources are allocated to the brain to solve problems, at the expense of the immune system, digestive system, and body function repair system, so people in occupations with high work pressure are more likely to get sick (Dinan, 2001). Therefore, people have different opinions and preference on competition. Some people enjoy working and studying in a competitive environment, but there are also many people who avoid competition in life and prefer a non-competitive environment. Because the environment has a very strong impact on people, todays job-seekers will take the work environment into consideration when looking for a job. From a company perspective, managers always pay great attention to employee productivity. The interests of the enterprises are always greater than those of the employees, so the efficiency of the employees is equally important for the enterprise. What companies need to consider is that in which environment employees can perform well, whether a fully competitive environment can bring the greatest benefits to the company, or whether in the environment that employees like can improve their efficiency. If the employees’ environmental preferences are not related to performance, then there would be no need for companies to understand employees'

preferences for job competition, and the budget could be spent on other employee benefits projects to encourage employees and generate greater returns. However, if the employees’ preferences have an impact on performance, then the companies have to consider whether the profit that may be generated in the future is great enough to cover the cost, which was spent to meet employee's preferences before.

Thus, my research question is: do people perform better in the environment they prefer? To exam whether preference can affect people’s performance I studied their choice of preferred work environment in an environment where all other affective factors had been eliminated and used a controlled experiment to create this environment. Participants were asked to do a real task in both competitive and non-competitive environments. Furthermore, time commitments were the same under both environments. Under these conditions I could determine whether people performed better in the type of environment they prefer.

In general, the experimental results are consistent with the hypothesis. The differences between two environments between people who prefer competitive environment and non-competitive environment are insignificantly. Although there is a positive correlation between the preferences and in which environment performance is better, the individuals' achievements in different environments are not significantly affected by their preferences, even in the case of controlling other possible influencing factors. Especially for those who prefer non-competition environment, the performances are the same in both environments. People who like the competitive environment perform a little bit better than those who like non-competitive environment, whether they are in a competitive or non-competitive environment. Overall, participants perform better in a competitive environment than in a non-competitive one, and the performance gap is greater when competition is involved. Moreover, participants ’ gender significantly impacts the environmental preference, with most men expressing predilection for a competitive environment, while women inclined towards a non-competitive environment. Last but not least, a strong negative correlation occurs between the performance of the competitive

environment and the educational level of participants.

Since competition is a topic has been studied from various aspects, I describe the most important related literature and discuss my paper’s relation to the literature in Section 2. The rest of the paper’s structure is as follows: Section 3 presents the experimental design and procedure based on this research question, and proposes my predictions and assumptions about the experimental results. Section 4 analyzes the results gathered in this experiment, and a series of explanations are given for the results of the experiment. What’s more, Section 4 also lists the limitations of the experiment. Section 5 explores some possible extensions of relative research and concludes.

Literature review

According to the literature, competition has different effects on performance in different situations. There are many studies confirming the positive effects of competition on the performance of students and employees (e.g. Lam et al. 2004, Worm and Buch 2014, Marinovic and Povel 2017, Gneezy et al. 2003), but Murayama and Elliot 2012, and Eccles and Midgley 1989 produce different answers. Murayama and Elliot (2012) defined three types of competition, which are trait competition, perceived competition and structural competition. No matter which type the situation is, their analysis shows that the correlation between competition and performance is very limited. Generally, there is no significant relationship between the competitive environment and performance. However, a large selection of literature on competition proved that competition can affect people’s performance. Eccles and Midgley (1989) came to a different conclusion while examining a secondary school classroom: when the competitiveness of the class increases, students are found to be associated with a decrease in academic performance. The opposite results have been stated by many other papers (e.g. Lam et al. 2004, Worm and Bush 2014, Marinovic and Povel 2017). Students in a competitive situation perform better than students in a non-competitive

situation when solving an easy task, but it cannot be ignored, competitiveness induced students’ performance goals so that more students opt for an easy test after setback (Lam et al. 2004). Worm and Bush (2014) found competitive work can be a motivating factor in E-learning, because experimental results show that people from competing groups not only were significantly better at the test but also had significantly better learning efficiency, moreover, they also generated improved self-awareness regarding the test result. And if misreporting disappears at work, competition for talent has unambiguously positive effect on efficiency (Marinovic and Povel, 2017).

Based on existing research on competition issues, the impacts of competition on performance vary among people due to the following factors: gender, goals, motivations, etc. Abundant literature has made in-depth and detailed discussions on the relationship between gender with competitive preferences and competition performance (e.g. Niederle and Vesterlund 2005, Niederle and Vesterlund 2011, Booth and Nolen 2010, Gneezy et al. 2003), and the motivational factor also has been tested in the classrooms and job markets (Heidemeier and Bittner 2012, Bergin 1995). A series of papers stated the effects of competition on performance have gender differences, for example, Gneezy et al. (2003) conducted a laboratory experiment and found that when the competitiveness of the experiment was increasing, there were significant improvements in performance for men but not for women. In some tasks when competition was imposed exogenously by the experimenter, the same conclusion was drawn (Price, 2012). Both laboratory and field studies provide adequate evidence and confirm that few women enter and win competitions, and even controlling other factors such as discrimination and preferences for work hours. Because of this significant difference, some articles have been researched further and the reasons given. One of the reasons is that mens’ performance tends to respond more positively to increasing competition (Niederle and Vesterlund 2011). Niederle & Vesterlund found a large gender difference in tournament entry willingness, and they attributed this phenomenon to gender difference in confidence and in attitude towards

competition. Furthermore, Lee et al. (2016) gave another reason why men and women have different competition experiences from the perspective of relationships. The existence of competition breaks the female peer culture values of harmony and the appearance of equality, but not males’, as a result, males have a greater preference for competition than females. In addition to identifying possible causes, Booth and Nolen's (2012) experiments also rule out the nurture causes of differences in gender preferences. Through the experimental results that girls from single-sex schools behave more like boys, they indicated nurture is not an explanation for the fact that women shy away from competition. When studying gender differences in competitive preferences, a series of articles involved real tasks in experiments to obtain more reliable data and conclusions when conducting laboratory experiments. Some of these experiments ask the experimenter to add up sets of five 2- digit numbers, and some ask participants to solve mazes. The experiment in this paper also uses mazes as a game to measure the participants’ performance.

Another competitive topic that is widely discussed is the motivation of people. The relationship between motivation and competition has been analyzed in different experimental backgrounds and different experimental groups. (e.g. Lam et al. 2014, Heidemeier and Bittner 2012, and Bergin 1995). There are two kinds of typical goals, performance achievement goals and learning achievement. Performance achievement goals mean that individuals focus on their own normative or social achievements and try to seek positive assessments of their abilities, and learning achievement goals imply that the individual not only pays attention to the standard of compliance and strives towards it, but also pays attention to improving his abilities in the process (Dweck & Leggett, 1988). Competition only has a positive effect on performance in a few cases, and often reduces the performance goal and worsens people’s self-evaluation after failure (Lam et al., 2014). Heidemeier and Bittner (2012) examined how competition within teams influences which type of achievement goals employees choose, and their survey showed competition has a positive relationship with performance goal in highly competitive teams. Bergin (1995) investigated the

influence of a mastery goal situation versus a competitive goal situation, concluding that the former resulted in higher achievements and greater interests.

The existing literature on the issue of competition concentrates on several topics, especially the gender differences in competition. Laboratory experiments through real tasks to reflect the different performance between men and women, field research gave evidences for a gender performance gap not only in sports competitions, but also selective educational programs or fellowships' competition and the labor market in general competition. Due to the large amount of experimental and field evidence, the impact of gender on competitive performance has been determined. So, further research is being conducted on the causes of these differences. The influence of motivation on competition has also been discussed extensively. Papers have researched participants' motivation and final performance by competing in different contexts such as the classroom and the workplace. But in addition to these two highly important aspects, are there other factors that may affect people’s performance in a competitive environment? If so, do these factors, as omitted variables, have a bias and conflict with existing research conclusions? So, my thesis raises the question of whether people's preferences are a new variable that affects performance in a competitive environment, and if so, how does people’s environment preference determine their performance. The exploration of this issue can provide not only reference for the recruitment of the company's human resources department, but also find other possible omitted variables in competition and performance problems.

To answer this research question, I propose and experimentally test whether people’s preference can be an explanation. In the present study, I placed participants in two situations, one competitive environment and the other non-competitive environment, and asked them to do a maze test. They were told they would be paid according to their performance. What is different with other experiments before, in this experiment participants' rewards are not related to their preferences. All performances in both environments determine the ultimate reward for the participants, which avoids that people’s choices are inconsistent with their real ideas in order to get

higher returns.

Experimental design

An experiment was conducted in which participants solved a real task. Standing research on competition issues has usually opted for mazes as experimental games, for instance, Gneezy 2003, Hoff and Pandey 2004. In this experiment, mazes also were utilized as a real task in part two. The experiment was divided into three parts: part one is a competition awareness test, which can reflect their attitude towards competition, part two is a two rounds maze test in different environments, and part three is questionnaire about personal preference and other personal information. After the experiment, 8 participants were randomly selected to be paid, and the payoffs were according to their performances in two rounds of maze tests.

The competition awareness test in part one is composed of five multiple choice questions, which came from Su Xuan and Ge Ming(2007). They used this test to explore the relationship between competition attitude and test anxiety of senior school students. These five questions are: 1. Do you want to exceed others in your daily work and study? 2. What do you think is the correct attitude towards competition? 3. In your spare time, what type of book is your favorite? 4. Do you like tasks that are heavy workload, high risks, and high incomes? 5. In the choice of career, do you want to find a lifelong secure job? To confirm participants ’ competition awareness, every A option they choose scored 2 points, B option scored 1 point, and C scored 0, the higher the score, the more intense the participants ’ sense of competition. The maze test includes two rounds, competitive environment round and non-competitive environment round. Each round is composed of 8 mazes, progressing from simple to complex. The non-competitive environment was a piece-rate of 50 cents per correctly solved maze in 3 minutes, the competitive environment required participants to perform in tournaments of 4 participants, in which the one with the highest number of correctly solved mazes was paid 5 euros

and the others got nothing. The only difference between two environments is how payoffs are calculated. To solve the mazes, participants needed to connect the two exits of the maze by pen and they were forced to complete the maze problems in order. Every participant took part in both environments and were asked to try their best in each round, but participants could accumulate experience in the first round and perform better in the second round. In order to decrease the learning effect, some participants started with the non-competitive environment and the others started with the competitive environment. The mazes of the two rounds were completely different, all participants were given the same maze questions in the first round and second round, regardless of which environment they started with. Finally, part three, a questionnaire, asked for their preference for these two environments, and some basic personal information such as gender, educational background and family situation.

Competition awareness test: 5 multiple choice questions

PartⅠ

PartⅡ

PartⅢ

Maze test: Two rounds, participants are asked to solve as many maze questions as possible in 3 minutes each round.

Questionnaire: Ask participants environment preference in this experiment,and some other personal information.

Figure Ⅰ Experimental process

First of all, the experimenter distributed the instructions and participants’ questions were answered in private. After 10 minutes, when all participants had read and understood the instructions, part one started. Then after 3 minutes, participants finished five single choice questions about competition awareness, and were allowed to read the first background of part two. When the bell rang, they could start the first

round of maze tests, and participants needed to stop answering immediately when the bell rang again, and move to the second round of part two. For part three, around 5 minutes were provided to fill in the questionnaire and leave personal information so that the experimenter could keep in contact with them, then the experiment ended. During this experiment, participants had no idea about the results of the first part (competition attitude test) and maze test, only the participants who were picked knew their own payoff and the performance of the other three group members in the tournament.

The hypothesis about this experiment is that there is no difference in the relative score (score in the competitive/ score in the non-competitive environment) between people who chose a competitive environment and people who choose a non-competitive environment. At the same time, gender differences are also predicted to show in the results. Compared to men, female participants prefer piece-rate instead of tournament.

40 participants took part in this experiment, 20 of which are undergraduates and master students from the University of Amsterdam and have different majors and study background. These forty participants consisted of 12 males and 28 females, some of them have worked for many years while some still study.

Results

The experimental results show that of 40 participants, 20 prefer the tournament competitive environment and the remaining 20 people prefer the piece-rate non-competitive environment. FigureⅡpresents the performance gaps ( performance in the competitive minus performance in the non-competitive environment) of all participants, for tournament preferring participants, the average performance gap is 0.05, and for piece-rate preferring participants, the average gap is 0, which means there is no difference in their two environment performance. It can be clearly seen in Figure Ⅱ that the effect of preference for the performance is not only positive, but

some participants prefer the environment in which they performed relatively bad. This also proves that participants were not affected by the performance in part two when deciding their own environmental preferences. Thus, the gap of people who prefer competitive with the gap of people who prefer non-competitive is not significantly different, only 0.05. To further analysis these data, I start with a non-parametric test,

Figure Ⅱ All Participants’ Performance Gap

The performance gaps of all participants, which indicate the difference between the performance of participants in the environment they like and they don't like. The vertical axis corresponds to the number of solved mazes’ gap between Tournament environment and Piece-rate environment, and the horizontal axis is the number of participants. “Tournament” means tournament preferring participants, “Piece-Rate” is the participants who like piece-rate.

Kruskal-Wallis test to verify normality. Test result shows that the probability of this data is 0.2645, which is larger than 0.05, so we can conclude this data is a normal distribution. Based on the normal distribution characteristics, I regressed two preference groups' performance gap and their environment preference to test whether the preference is a potential factor for people’s performance. The result shows that the p-value of this regression is 0.678, respectively, the hypothesis cannot be rejected. So the relative score (score in the competitive minus score in the non-competitive environment) between people who chose a competitive environment and people who

choose a non-competitive environment is not different. In another word, the change of environment has no effect on participants' performance no matter what kinds of environment participants like. In addition, the regression between the participants’ performance in tournament and their preferences can also provide evidence for this conclusion. Tournament performance and preference have a slightly positive relationship with p-value = 0.541, therefore, participants’ performance in competitive environment are not affected by their personal preference, and this also leads the performance gap for two preference group are the same .

In general, piece-rate and tournament performance are close, the average solved maze number in tournament is 5.7, and in piece-rate environment is 5.625. Figure Ⅲ presents the average number of mazes solved by all participants with different preference for tournament and piece-rate environment. When participants were in the environment they preferred, the average performance was 5.7, and when they were in the environment they did not favor, the average performance was 5.625. In further detail, for the participants who preferred tournament, the average number of solved mazes in tournament was 5.8, and the average number in piece rate was 5.65. On the other hand, for participants who liked the piece-rate environment, the average number in piece-rate was 5.6, and the performance in tournament was 5.6. There is no difference for participants when the working environment is changed, but compared to participants who like tournaments, their performances are slightly worse in both situations, especially in tournament environment.

Figure Ⅲ Average performance

The average number of solved mazes under the tournament and piece-rate environment for all participants with a different preference. The vertical axis corresponds to the number of mazes solved, and the horizontal axis is the types of environment.

The potential interpretations for this phenomenon may be that, first of all, people who like tournaments are more likely to hold a performance achievement goal. This can be confirmed by the option they choose in the first round, The results from the first part of the experiment show that 55% of participants who prefer tournament think they want to perform better than others in their daily learning and work. Related research on motivation has pointed out that individuals who have a performance achievement goal are eager to outperform others and enjoy the feeling of winning. During the tournament round participants cannot know the accomplishments of other competitors, to win the competition they must try their best to solve as many mazes as possible. So the wish to outpace others in the experiment was very strong, which prompted them to answer more questions in time. In contrast, they did not need to compete with others in the piece-rate round, which means there was no winning or losing, and such an environment reduced their motivation. On the other hand, among 20 participants who preferred piece-rate, only 20% participants showed a desire to exceed others in the first round test. They may have comparably less of a performance-goal orientation, regardless of the competitive environment or

non-competitive environment, preference does not matter to their motivations and performance. As a result, environmental changes cannot effectively stimulate participants’ output in the experiment. However, it must be acknowledged that the gender ratio of participants had a certain impact on the results. Of 40 participants, 28 were women and only 12 were men. In the following analysis, most of the women chose the piece-rate environment, only 3 men chose piece-rate. Therefore, the number of women in piece-rate preference far exceeds that of men, making women dominate the two rounds.

In addition to excluding the effects of preferences, the experimental results also provided other information. As it is possible to gather from the information found so far, participants perform slightly better in a competitive environment than in a non-competitive one, and this can be explained by several reasons. Firstly, because the number of participants who like tournament and piece-rate is the same, considering nevertheless that preference does not significantly affect piece-rate oriented people, for participants who like tournament there is still a positive correlation between performance and preference. This group of participants has a relatively good performance in the tournament, so the small gap between the performance of the tournament and the piece-rate can be interpreted as the tournament preferring participants have a better performance in the competitive environment. For those who like to compete, the pressure in a competitive environment may help them to achieve more. The competitive pressure in the experiment is very high, it only comes from the competition with the three players in the same group. A lot of research shows that an appropriate amount of stress can focus participants ’ attention and improve the efficiency of problem solving, thus when the participants were in the tournament round, they feel pressure and their speed of solving mazes increased compared to the piece-rate environment. Moreover, this improvement may be due to the fact that the group of participants who took tournament as round 2 had a higher learning effect. The environment order of part two is a dummy variable in the regression with p-value 0.02, thus the order has a certain effect on performance. It can

be interpreted as such because this group of participants quickly gained experience in the first round of the process, which made their thinking clearer in the second round, so they responded more quickly, producing better results.

But the design of the mazes could be the reason for the small gaps. In the two round maze games, every round included 8 mazes, which consisted of 2 low difficulty mazes, 4 medium difficulty mazes and 2 high difficulty mazes. Due to the limited time and the difficulty of the problems, only a small number of participants could successfully solve the last two problems. Most of the participants could only proceed to the 5th and 6th mazes, therefore the performance gap between the two environments is not obvious. If the number of medium difficulty mazes were increased, most of the participants could continue to solve the problems rather than being trapped by difficult tasks, then the performance in the two environments may have been different. However, this doubt can be overturned by the following arguments. Firstly, even if there are few participants who can complete the last two problems, the difficulty of the maze test is not the main reason. The data shows that more than half of the participants did not complete the last moderate difficult question, which is the 6th maze. Thus it is not the difficult problem that traps most people. If preferences can affect participants ’ performance, they should complete more moderate difficult questions in their interested environment of choice. Secondly, there are difficult mazes in both environments, if the participants’ preferences affect their performance, then they should work harder to solve difficult mazes in their preferred environment. In this case, the number of people in a preference group that solved 6th maze should be bigger. Especially for participants trapped by a difficult maze, the preferred environment should give them more incentives. However, the evidence suggests that this is not the case, there are 13 participants who like tournament completed the 6th maze, and 11 participants who like piece-rate successfully solved the 6th maze. Left 14 participants did not get motivated in the preferred environment, resulting in significantly better output in this experiment. Through the analysis of the two reasons above, it is determined that there is no obvious inevitable connection

between the individual's preferences and performance.

Through regression analysis, gender and education do not show a significant influence on participants’ performance gap, but there is a noteworthy difference in the preference of male and female in the competitive environment. This experimental result is in line with the results of many previous papers (e.g. Niederle and Vesterlund 2007, Price 2012). The result is immediately obvious by looking at Figure Ⅳ in the succeeding test. As Figure Ⅳ reveals the ratio of male participants and female participants preferences, male subjects gravitate towards the tournament, while female participants are attracted to the piece-rate environment. In particular, 9 out of 12 (75%) male subjects think tournament is more attractive to them, while 11 out of 28 (39%) female subjects express more enjoyment of piece-rate environment. This difference has a p-value of 0.031, and t-value is -2.24, which indicate the attitudes toward competition between male and female are considerably different, meaning males are more competitive than females. This result was also confirmed in the first part of the experiment. The test results showed that 8 of the 12 male participants were eager to surpass others in work or study (accounted for approximately 66.7%), while only 7 of the 28 female participants expressed a desire to perform better than others, accounting 25%.

But there are no significant gender differences in the performance with tournament or piece-rate environment. In the tournament environment, the average number of mazes solved by female subjects was 5.68, and that of male participants was 5.75, which was slightly higher than that of females. In the piece-rate round, the average number of mazes solved by females did not change compared with the tournament, which was still 5.68, but the average number of male participants decreased, to only 5.5. This result is similar to some existing research, as Gneezy et al. (2003) mentioned that the performance of women did not change significantly in all the experimental treatments, and Niessen & Ruenzi (2008) also shows that compared to males, female fundraisers were less likely to have an extremely good or poor performance. Although most female participants preferred the piece-rate environment, their performance did not change with changes in the environment. Male participants have a strong preference for the tournament environment, and their performance is consistent with their preference. Therefore, male subjects have a better result in the tournament environment. The differences in male and female performance in tournament and piece-rate environments is not significant with p-values of the two sided t-test are 0.865 and 0.678, respectively, the somewhat higher performance of men is not significant. Previous research has shown that male and female participants have different tastes for tournament and suggested some possible explanations. Firstly, men and women differ in their beliefs about their relative performance, women are less confident than men about their relative ability (Niederle & Vesterlund 2007, Balafoutas & Sutter 2010, Niederle et al. 2010, Wozniak et al. 2010). Secondly, it might be that women have the cost of effort, which can improve their output only when the costs are very high (Gneezy et al. 2003). Furthermore, women’s dislike for competition is also a possible reason.

Another interesting fact is that in the environment of the participants' respective preferences, as the results show in Table Ι , participants with a Master's degree performed far worse than those with a Bachelor's degree (p-value = 0.003). But this conclusion is not true in an environment that participants were not interested in. Since

most of the experimental participants are currently undergraduate or master's degree students, the difference in academic qualifications can also reflect the difference in their ages. In other words, this result can also be understood as the relatively younger participants performed better when they are in their preferred environment.

Table Ι

OLS Performance In preferred Environment

（1） （2） Education -0.2934 -0.9230769** -0.3578 -0.2929 Siblings -1*** -0.1639 N 40 40

The regression between participants’ performance in their interested environment with sibling number and education level. Standard errors are in parentheses; *p<0.10, **p<0.05,***P<0.001.

This result can be explained by the fact that participants with relatively young ages will have more environmental preferences than those who are older. In the solution of the maze problem, young people have faster reaction and judgment. When they enter the wrong route, they can quickly return and find the correct route. At the same time, they can effectively accumulate experience when they complete the mazes, especially in their preferred environment. They may be more focused in their favorite environment, which makes them perform well. Although the impact of education on the performance gap is not significant, the degree of education and the gap in performance between two environments are negatively correlated with a coefficient of

-0.3447. The higher the degree of education, the smaller the performance difference, so Master students are more stable in both environments compared to Bachelor students. Another aspect of the data is that the number of siblings has a great influence on the performance of participants in different environments. The results show in Table Ι clearly, the number of siblings and the performance of the participants showed a significant negative correlation trend. The two sides t-test confirmed the significance of this relationship with p-value = 0.002 in both environments. Therefore, the more siblings a participant had, the fewer mazes he or she completed. Similarly, the participants' performance in the liked and disliked environment also shows the same trend. However the influence of siblings has few effect on people’s performance gap, so this characteristic has little meaning for the experimental results.

The above research results may have some deviations, because there are still some potential problems and practical reasons that cannot be overcome in the experiment. The most important point is the selection of samples. Due to the limited conditions, a large number of participants are from the school students with little social experience, so the experimental sample’s understanding and contact with the workplace may not be as good as actual job-seekers. This may have some impact on the external validity of the experimental results. And female participants account for the vast majority, which is not very conducive to investigating gender differences in competition issues. Second, the rewards of the experiment may also bring some deviations. According to Vernon Smith’s precepts, the more rewards in the experiment the better. Participants react differently to the amount of compensation, but if the amount of compensation in each round of experiments is high enough, participants will be more motivated and we can get more accurate results. Or the maze games should pay according to the difficulty of the maze, the harder the maze solved, the more rewards there are. As a result, the participants can be motivated, and it is possible to get more obvious differences. In addition, the external validity of the maze game as a real task is yet to be verified. Although there have been a lot of experiments using maze games before, whether the conclusions obtained through the maze can be

effectively verified in real life is still not certain. Finally, the omitted variable bias can also be a problem, and there may be other related factors that affect people's performance that this paper did not include.

Conclusion

The findings of the present study lead to several conclusions. First, the performance gap is not different comparing the preference groups. Individual ’ s performances in competitive and non-competitive environment are only slightly affected by their preference. People who prefer a competitive environment perform a little bit better in a competitive environment than in a non-competitive one, but the gap is not obvious. Secondly, these always have better performances compared to those who prefer non-competitive environment, which is particularly evident in the competitive environment. On the other hand, people who like non-competitive environments are basically not affected by environmental preferences, and their performance is stable regardless of competition in the environment. Another particularly striking result is that there is a big difference between men and women in the preference for competition, which is primarily consistent with previous studies. Men showed a strong willingness to compete with others in both the competition awareness questionnaires at the beginning of the experiment, and in the environmental preferences one at the end. Compared with men, women are more inclined to choose a non-competitive environment. In terms of performance, the influence of male and female preferences is not significant. Women’ s performance is completely unaffected by the presence or absence of competition, but male performance is more affected by the competition element. Male performance is better in a competitive environment, but it declined in a non-competitive environment, where women have a relatively good performance.

This study also considers the number of siblings people have and their educational background, and although these variables have no obvious connection

with people's competitive attitude, they seriously affect their performance. The fewer siblings people have, the better their performance. This phenomenon has also been confirmed in other studies (eg. Belmont and Marolla 1973, Zajonc 1976, Falbo 1978, Claudy 1979), showing that people with fewer siblings obtain higher academic achievements and learning skills. The degree of education also determines the performance to a certain extent, as undergraduate students perform better than master students. In this experiment in particular, this result then also reflects the relationship between age and performance, because most of the participants are currently college students.

The present study was an attempt to investigate the relationship between people’s preference and performance. The results of this study can give some guidance to company managers. The conclusions show that the performance gap is not different comparing the preference groups, and there is not a obvious correlation between people's environmental preferences and performance. No matter how competitive the company's environment is, employees' performance will not fluctuate greatly. This means that even if the company puts a lot of effort into understanding the employee's environmental preferences and trying to improve the company's competition environment accordingly to bring greater profits to the company, the performance of the employees will not increase significantly. At the same time, although people in the competitive environment will have relatively good performances, the impact is not very significant in a laboratory environment, so this conclusion needs to be verified in further field experiments.

The results shown in this paper are influenced by the samples and cannot fully reflect reality, especially on the issue of gender. The external validity of the experimental conclusions also requires experimentation or investigation of job seekers in real life to be confirmed. Perhaps the relationship between competitive attitudes and working performance can get more accurate answers in field experiments and further research. Although the experiment in this paper shows that preferences cannot strongly change people's preferences, this may also be related to the number of people

competing and the intensity of competition. This article do not change the intensity of the competitive environment and therefore no analysis can illustrate this question. So this possibility needs to be verified after further research.

Appendix

Table Ⅱ

OLS Performance Gap in Two Environment

（1） （2） Preference 0.35 0.3485 0.3537 0.3548 Round 1 Tournament -0.0152** 0.3548 N 40 40

Dependent variable is participants’ performance gap between tournament and piece-rate, variable Tournament shows the maze test order, order is 1 means the participant took tournament as first round. Education indicates the educational level of Participants. Standard errors are in parentheses; *p<0.10, **p<0.05,***P<0.001.

Table Ⅲ OLS Preference （1） （2） (3) (4) Female -0.3571** -0.3616** -0.3661** -0.3774** 0.1594 0.1601 0.1619 0.1663 Siblings 0.0417 0.0716 0.7341 0.1211 0.1258 Education level 0.1063 0.6724 0.1784 0.1958 Major -0.3759 0.0811 N 40 40 40 40

Dependent variable is participants’ environmental preference, *p<0.10, **p<0.05,***P<0.001.

Reference

Balafoutas L, Sutter M. 2010. Gender, competition and the efficiency of policy interventions. IZAWork. Pap. 4955, Bonn, Ger.

Bergin, D. (1995). Effects of a Mastery Versus Competitive Motivation Situation on Learning. The Journal of Experimental Education, 63(4), pp.303-314.

Blake, J. (1981). The Only Child in America: Prejudice versus Performance. Population and Development Review, [online] 7(1), p.43.

Dinan, T. (2001). Guest Editorial: stress, depression and cardiovascular disease. Stress and Health, [online] 17(2), pp.65-66.

Dweck, C. S., & Leggett, E. L. (1988). A social-cognitive approach to motivation and personality. Psychological Review,95, 256–273.

Gill, D. and Prowse, V. (2014). Gender differences and dynamics in competition: The role of luck. Quantitative Economics, 5(2), pp.351-376.

Gneezy, U., Niederle, M. and Rustichini, A. (2003). Performance in Competitive Environments: Gender Differences. The Quarterly Journal of Economics, 118(3), pp.1049-1074.

Heidemeier, H. and Bittner, J. (2012). Competition and Achievement Goals in Work Teams. Human Performance, 25(2), pp.138-158.

A. Davis, General Social Surveys, 1972-1978: Cumulative Data (Chicago: National Opinion Research Center, 1978).

G. Claudy, W. S. Farrell, Jr., and C. W. Dayton, The Consequences of Being an Only Child: An Analysis of Project Talent Data (Palo Alto, California: American Institutes for Research, 1979).

Johnson, D., Johnson, R. and Roseth, C. (2012). Competition and performance: More facts, more understanding? Comment on Murayama and Elliot (2012). Psychological Bulletin, 138(6), pp.1071-1078.

Lam, S., Yim, P., Law, J. and Cheung, R. (2004). The effects of competition on achievement motivation in Chinese classrooms. British Journal of Educational Psychology, 74(2), pp.281-296.

Belmont and F. A. Marolla, “ Birth order, family size, and intelligence, ” Science182 (14 December 1973): 1096-1101.

Lee, S., Kesebir, S. and Pillutla, M. (2016). Gender differences in response to competition with same-gender coworkers: A relational perspective. Journal of Personality and Social Psychology, 110(6), pp.869-886.

Marinovic, I. and Povel, P. (2017). Competition for talent under performance manipulation. Journal of Accounting and Economics, 64(1), pp.1-14.

Niederle, M. and Vesterlund, L. (2007). Do Women Shy Away From Competition? Do Men Compete Too Much?. The Quarterly Journal of Economics, 122(3), pp.1067-1101.

Niederle M, Vesterlund L. 2010. Explaining the gender gap in math test scores: the role of competition. J. Econ. Perspect. 24(2):129–44

Price, C. (2010). Does the Gender Preference for Competition Affect Job Performance? Evidence from a Real Effort Experiment. SSRN Electronic Journal.

Zajonc and G. B. Markus, "Birth order and intellectual development," Psychological Review 82, no. 1 (1975):74-88; and R. B. Zajonc, "Family configu- ration and intelligence," Science 192 (16 April 1976): 227-236.

Su Xuan and Ge Ming, "The relationship between competition attitude and test anxiety of senior middle school student, " Chinese Primary Health Care, 2007, 21(9): 37-38.

Falbo, "Sibling tutoring and other explanations for intelligence discontinuities of only and last borns," Journal of Population 1, no. 4 (Winter 1978): 349-363.

Worm, B. and Buch, S. (2014). Does Competition Work as a Motivating Factor in E-Learning? A Randomized Controlled Trial. PLoS ONE, 9(1), p.e85434.

Wozniak D, Harbaugh WT, Mayr U. 2010. Choices about competition: differences by gender and hormonal fluctuations, and the role of relative performance feedback. Work. Pap., Univ. Oregon.