Showing off effect : does being public affect the choice between hard or easy tasks and how does it depend on gender?

(1)

FACULTY OF ECONOMICS AND BUSINESS

MASTER THESIS

Showing off effect: Does being public affect the choice

between hard or easy tasks and how does it depend on

gender?

Author: Venera Danielyan 11085045

Program: MSc Business Economics

Track: Managerial Economics and Strategy

Supervisor: Thomas Buser

(2)

Does being public affect the choice between hard or easy

tasks and how does it depend on gender?

July 2016

ABSTRACT

This paper investigates whether the decision between easy or hard task differs when the results and choices for the task are announced publicly and if gender plays a role in those decisions. An experiment in high school was conducted, where 17 similar1 classes from grades 9 to 12 were randomly assigned to one of the three treatments: Public Results, Public Choices or Non-public. Two levels of difficulty for same real-effort task were presented to participants and they had to choose which one they preferred to solve before the start of the experiment. After the main experiment subjects filled in a questionnaire on risk attitude, self-confidence, challenge tolerance, comparative abilities and on importance that they give to classmates’ opinions. The main results show that there is a significant difference when the results of the hard task are announced compared to when only task choices are announced or when nothing is announced. Furthermore, the effect is more dramatic for women than for men, as women choose more often the hard task when the results are announced. Overall, the data confirms that there is some showing off effect going on and especially among females.

Keywords: decision making, task difficulty, experiment, gender, ability

1_{The high school chosen for the experiment was a typical school in Yerevan with no outstanding features or} student pool, the number of students in each class varied from 15 to 25, but because of coming summer holidays some of the students were absent on the days of the experiment so the number of students in each class was from 9 to 17

(4)

2

1. INTRODUCTION

“IT’S IN YOUR MOMENTS OF DECISION THAT YOUR DESTINY IS SHAPED”- Anthony

Robbins

Decision making is a crucial element of our everyday life and hugely affects the future we will have. Starting from a university to attend, career to pursue or a specific task to accomplish, we are confronted with various decisions to make. Nowadays economists show growing interest towards behavioral motives of economic agents. Many recent studies focus on decision making process: what drives specific decisions and how those decisions are affected by external factors? Being part of the society implies that we are influenced by others, hence looking at how public decision making differs from private is highly relevant. In many experimental studies, the decision to donate or not is observed for public or private situations. However, an interesting area of study that is under investigated is whether there is also a difference in the decisions between easy and hard tasks, when we compare public versus private conditions. To empirically investigate whether making the results and choices public affects the decisions between easy and hard task, I conducted an experiment in high school, where participants had to choose their preferred task difficulty level under different degrees of public announcement. Hence, this paper will examine the effects of public announcement and choice of task difficulty.

As in real life most of the decisions are made in public and it’s in human nature to always compare to others, it will be interesting to see how we accept challenges under public pressure. Are we scared of giving bad information about our abilities by taking up the hard task or are we trying to show-off that we are better than others by choosing the hard task? To understand this behavioral aspects of decision making I will answer to the following research question: Does being public affect the choice between hard and easy tasks and how does it depend on gender?

The paper will also discuss potential gender differences in the decision among easy and hard tasks, as a growing body of literature confirms, there is a significant difference in competitiveness and challenge seeking behavior between men and women. Niederle and Yestrumskas (2008) through their lab experiments found that men choose the hard task 50% more often than women, independent on their previously tested performance level. Hence, this will be an interesting comparison to see whether this percentage difference changes when the public announcement is added. Women are considered more vulnerable to public image and it might be the case that they feel the public pressure while making a decision, for example

(5)

3

choosing a career to pursue. It might explain why many women do not run for a president as everyone views it as a hard duty to realize.

I will conduct an experiment in high school to test whether there is a difference between the decisions made anonymously and publicly, when participants have to choose between easy or hard tasks. Standard economic theory suggests that if participants are rational and are able to anticipate their ability, there should not be a major difference whether decisions are made privately or publicly, because they should always choose the level of difficulty that they are capable of solving.

The rest of the paper is structured in the following way: Section 2 provides a brief overview of existing related literature, Section 3 explains the experimental design, Section 4 derives the main hypotheses, Section 5 presents the results and outcomes from the experimental data and Section 6 gives final remarks and concludes.

2. LITERATURE REVIEW

Papers such as Soetevent (2004) show the effect of anonymity in giving with a field experiment. When the decisions of subjects to donate are kept in private they tend to give less compared to when the donations are made public. Hence, the fact that other people are watching changes the decisions made. The author explains these findings with the fact that when the donations are made public, they give additional information to others about the decision maker. Therefore, wanting to appeal to the public the decision maker might reconsider his/her choices. In regards to the decision to donate, usually there is no much controversy involved; if the cause is considered to be good then people are expected to donate, otherwise they transfer bad information about their own charitability and care for the cause. However, it is not very obvious how those decisions will vary in a different context.

Public versus private decision making is tested by Dan Ariely et al. (2007) in the charity context. For their experiment participants are randomly allocated either to public or private donations treatment, where the main task was to click two keys on the computer keyboard to donate money for charity. Hence, there was some effort involved from the participant to increase the donations for the charity and the results show that when the donations were public the effort was higher compared to when they were private. This is mostly explained by image motivation analysis, hence when the donations are publicly visible, people want to appeal prosocial and they click more for the good cause. Therefore, when the public component is introduced we see more effort is being put to behave prosaically even though the cause is the same. However, this paper does

(6)

4

not look at what will happen if the decisions made involved some sort of consequences for the decision maker.

Decisions can also be different based on our attitude towards challenges. In her book Mindset Dweck points out that there are different types of attitudes towards challenge. Some people like to take on challenging tasks, while others try to stay more in their comfort zone. While the author looks at different types of people with a different mindset to understand what drives to ultimate success and long term achievements, the fact that most of the decisions are made in the social environment is highly ignored. Whether we choose an easy or a hard path to pursue is highly dependent on our mindset, however there might also be image concerns involved. For example, a student choosing his/her courses might be willing to take the challenging ones, however if he/she fails then the classmates might know about that during the resit, so fearing from leaving a bad image among classmates, the student might choose the easy courses.

In this context another important aspect to consider is confidence. It is more likely that confident people take more difficult tasks, as they feel they can handle them. Paper by Benabou and Tirole (2001) examines the effects of confidence in one’s abilities on the motivation and willpower to perform. With their formal model of self-confidence and motivation the authors depict why people choose to remain ignorant about their abilities and why sometimes they take on ambitious tasks in which they are sure to fail. While, the literature categorizes different types of individuals who have differing preferences in their choices and effort, it will be interesting to observe how taking out the anonymity will affect self-confidence and performance of participants.

On the other hand, Köszegi (2006) suggests that people care about ego and self-image, which predicts that besides pure economic benefits people get self-image utility from their decisions. The model discusses a simple scenario when people are faced with a decision problem between two financial choices and there is a possibility to collect free information on personal ability to be able to later decide whether to take on the ambitious or unambitious financial task. It turns out that people with ego utility might prefer to acquire free information in smaller pieces, which is explained with self-image protection motive. People are satisfied with their beliefs about their ability, hence they do not want to get disappointed by collecting more information. Therefore, this model is quite relevant for my studies as it depicts how people might make decisions between ambitious and unambitious tasks. However, the paper does not test whether the perception of self-image and reputation changes when decisions are presented publicly. Therefore, with only self-image concerns, there should still be no difference from public and private treatments, but

(7)

5

if there is an additional component involved when the decisions and results are portrayed publicly then the utility might change accordingly and the decisions can be different2_.

Köszegi (2006) also discusses potential applications of the ego utility, overconfidence and task choice model with the prospective entrepreneur’s decision making process. Managing an enterprise is a challenging task and is more informative about person’s business skills compared to just being employed, hence again for self-image protection reasons an individual might choose to forgo his chance of owning a business, fearing from getting negative information about his abilities. While the self-image factor plays a role in decision making, one should also take into consideration the public image effect. If an employee fails a specific task not that many people will know about his failure and there is also a chance to put the blame on others. However, if an entrepreneur fails the business then he is solely responsible for that failure, which gives bad information to public. Therefore, while the model by Köszegi does not address this aspect of decision making, it is important to question how public image affects our decision making. A psychology paper by Rind and Benjamin (2001) reveals the effects of public image concerns and how they differ from self-image on compliance. An experiment was conducted with male shoppers who were sitting in the shopping mall either alone or with their female companion when a confederate approached them to sell tickets. The presence of the companion strengthened the public image concerns hence males bought more tickets as a results.

3. EXPERIMENTAL DESIGN

The experiment is designed in high school, where compared to lab experiments students are set in a more natural setting and the announcement of the results and choices is feasible. The age of the participants ranges from 14 to 17, so they have already experienced some sort of decision making and they understand the consequences. Students have to complete a real effort task which consists of finding the two numbers in a nine number matrix that add up to the required sum; 10 for the easy task and 9.38 for the hard task. The experiment starts with a choice of the task difficulty level by the participant, afterwards the instructor distributes the chosen task level and participants have five minutes regardless of the difficulty level to finish the exercise. To pass the task students have to complete correctly at least ten matrices out of fifteen for both difficulty levels.

The Participants were incentivized by the presence of the school director and their teacher, who were making sure that students are working on the task. Also as subjects are in a school

2_{In this paper I will only look at empirical results through experiment and will not build a model for new type of} utility

(8)

6

environment where every day they get new assignments to complete in class, the task did not require special effort from them. Other external incentives were not used during the experiment, because giving something additional to students who have passed the task will have the same effect as announcing the results, because everyone can observe whether a student gets a gift or not. When subjects have completed all parts of the experiment and the results are announced for the hard task if public results treatment is implemented, the instructor gives private feedback to all other participants on whether they have passed the task or not.

There are three treatments according to the level of public component involved. In the first treatment, the choice of the difficulty level and the results for the hard task are announced in class after the experiment. In this treatment the results are public only for the hard task and it is announced in a pass or fail format. Hence, students who want to avoid their results to get public can choose the easy task. In the second treatment, only the choice of the task difficulty level is announced in class, but nobody gets to know the results, hence the public component is present only for the choice of the difficulty level. For the third treatment, all information is kept private and only students themselves know what difficulty level they have chosen and how well did they do. Each class is randomly allocated to one of the three treatments. For instance, the first class passed the public result treatment, while the second randomly selected class had the public choices and for the next class the control treatment was tested and the same logic continues for the upcoming classes by making sure that the number of participants from the three treatments is around the same.

After the main experimental part, students are given a questionnaire to complete. To measure the level of self-confidence, I asked to rate their self-confidence level in general on a scale from 1 (very low) to 7 (very high). To understand the risk attitudes of different participants, I asked “How do you see yourself: Are you generally a person who is fully prepared to take risks or do you try to avoid taking risks?” (Thomas Buser 2016). The answer is on a scale from 0 (“unwilling to take risks”) to 10 (“fully prepared to take risk”). The questionnaire includes a question on perceived abilities, which allows the student to rank himself/herself among the classmates from top 1% to bottom 5% of the class. Students attend all their classes with the same group, hence they already have an understanding of their abilities compared to others. As the number of students in each class is on average only fifteen, I have chosen a simplistic scale to categorize student abilities into outstanding, good, average or below average. To get a brief measure on how students handle unknown situations, I also asked questions about their attitude towards challenges, learning and success3. As the literature confirms (Dweck 2006) people who love

(9)

7

challenges and have a growth mindset are more likely to go for the hard path, hence eliciting their general mindset with the questionnaire and comparing those answers with their actual choice of task difficulty is of high interest.

The experiment was conducted in May 2016 at N65 high school in Yerevan, the capital of Armenia. A total of 229 high school students participated in the experiment over three different treatments. The subject pool consisted of 110 females and 119 males, divided into 17 classes in total. The treatment variation was randomly assigned to different classes, where the number of students in each class varied from 9 to 17. Overall the experiment lasted around 35 minutes in each class.

4. HYPOTHESES

It is possible to have several hypothesis for the treatments effect and for the gender difference: Hypothesis 1: The students will choose less often the hard task when the results are public compared to when they are not, as choosing the hard task will increase their probability of failing so they will give bad information to classmates about their abilities.

The paper by Schlenker (1991) defines self-identification as the process of showing oneself to be a particular type of person through self-disclosure and public self-presentation. When people face threats to their identity, for example when they might experience failure, they try to change the timing, terms or the outcomes of the event. Hence, if faced with a challenging task, when the chances of failure are high, people will try to avoid the task or work less hard. This is consistent with my experimental design, as choosing the hard task has higher probability of failure and is a potential threat to the students’ established identity among classmates. Thus, participants are more likely to choose the easy task to avoid threats to their identity and image.

The study conducted by Curley et al. (1986) showed that publicizing subjects’ decisions, which increased the likelihood of being evaluated by others in the group, raised the level of ambiguity avoidance. If we apply this finding to my study, when the results are announced publicly students are exposed to evaluation by others in class, hence the decision makers’ reaction should be to decrease the chances of ambiguity in task completion by choosing the easy task.

Hypothesis 2: If the results are not announced, but choices can be observed by classmates, the students will choose more often the hard task compared to public results and non-public treatment.

This might be explained by the fact that when the results are not announced students have nothing to lose by choosing the hard task. They are not afraid that when they choose the hard task they

(10)

8

might fail and everyone will know about it. Therefore, this treatment gives the chance for “free” image creation, as they can show-off everyone that they are smart by choosing the hard task and nobody will ever know if they failed.

Hypothesis 3: When the results are not announced publicly and nobody can know the choice between hard and easy tasks, one could assume that the students will choose less often the hard task compared to when choices are announced. As in this case, the results do not create a public image the sole driver for the choice should be the perceived ability of the student by her/himself. It is in human nature to always compare to others and Blanton et al. (1999) confirms that comparison affects our decisions. Depending on how our own abilities compare to others’ abilities we judge whether we are more or less talented than the people around us. The unit of comparison for my experiment will be the class and to understand how students view themselves compared to their classmates I will use the questionnaire results where students will rank themselves subjectively.

To sum up, we expect public choices treatment to have the highest number of hard tasks chosen, followed by public results, as this treatment allows for showing off when the choices are announced however it comes with a cost of bad reputation in case of failure when the results are announced. We expect to see the least number of hard tasks chosen in non-public treatment as there are no reputational concerns, hence less motivation to choose the hard task.

Hypothesis 4: Overall, females will choose less times the hard task compared to males, as the studies show females are less willing to enter competition4_{and are more risk-averse.}

However, one should also expect gender differences between the different treatments.

Hypothesis 4.1: Females will choose less times the hard task compared to males for both the public results and public choices treatments. Meta-analysis from 150 studies realized by Byrnes et al. (1999) concluded that on average men are more risk taking than women, hence in the public results treatment when the risk of public failure is higher for the hard task, men are more likely to take the risk and choose the hard task than women. Moreover, another study by Böhm (2013) found that under public signal, where the relative rank of the performer is visible to others, the performance increase is higher for men than for women. Which means that when others can see the performance men are more likely to show off by increasing their effort level. Hence, I also expect to see more males than females choosing the hard task when the results or choices are announced publicly.

4_{Niederle and Vesterlund 2007 “Do women shy away from competition? Do men compete too much?”} Thomas Buser 2016, “How does the gender difference in willingness to compete evolve with experience?”

(11)

9

Hypothesis 4.2: Males and females on average will choose the hard task with around the same

frequencies when nothing is announced publicly. As there is no social image attached to the

decision and the literature confirms that abilities of men and women are not significantly different, we should expect to see no significant difference between the choices for control treatment.

Overall, standard economic theory assumes for rational decision makers there should be no difference between the treatments, as one should choose the task he/she is capable of handling, however as social aspects have a huge influence on our behavior it will be interesting to see how much is this influence present in our decision making and towards what direction it biases our choices.

5. RESULTS

5.1 Public announcement effect on choice between hard or easy task

In this part I will present the results between the three treatments to see whether public announcement affects the decisions between hard or easy task. The data collected shows that there is a significant difference between the treatments. Table 1 presents descriptive statistics of the averages during all the three treatments.

Table 1: Descriptive statistics

Note: The table shows averages for all three treatments with standard deviations in the parentheses. Task difficulty refers to the choice of the participant between easy or hard task at the beginning of the experiment. Easy task was given the value of 0 and the hard task the value of 1. The other three variables are measured with questionnaire. The last three columns show p-values from t-tests between different treatments.

Scale Public Results (1) Public Choices (2) Non-public (3) p-val (1)&(2) p-val (1)&(3) p-val (2)&(3) Task Difficulty binary 0.60

(0.49) 0.38 (0.49) 0.41 (0.49) 0.0083 0.0313 0.6726 Self-confidence 1-7 5.19 (1.37) 5.50 (1.12) 5.32 (1.08) 0.1214 0.5129 0.3175 Risk attitude 0-10 7.26 (2.21) 7.65 (1.93) 7.58 (1.92) 0.2476 0.3558 0.8217 Perceived abilities 0-1 0.67 (0.24) 0.71 (0.24) 0 .72 (0 .23) 0.3255 0.1751 0.6947 N 72 83 74

(12)

10

Overall, when the results are announced publicly for the hard task, participants tend to choose more often the hard task compared to other treatments. Looking at the averages, we see that in public results treatment subjects chose more the hard task than in public choices treatment and this difference is significant (p=0.009, chi-squared test). If we compare public results with the control treatment, we again see that it was more likely to observe the hard task chosen in the public results treatment and the difference is significant (p=0.031). Subjects are less likely to choose the hard task when only choices are public compared to control group, but the result is not significant with p-value 0.670.

Students have rated their general self-confidence level and public choices treatment has the highest average score of 5.50, however the p-values show that the difference is not significant between all the three treatments. We see similar findings for risk attitude, which shows that there was no major difference between the subjects’ confidence level and risk attitude. Hence, we can assume that the difference between treatments is not derived from confidence or risk taking tendency. However, we might still need to control for these variables to measure the true effect of public announcement on task choice.

Existing literature argues that one of the potential factors affecting whether students choose to go for the hard task or for the easy one is ability. In the experiment the ability is measured through questionnaire, where students have to compare their own abilities with that of their classmates and they had to rank themselves among classmates. The class is a good measure for perceived abilities, as in real life as well people usually work in teams. Be it at school, at university or at workplace, we are always good or bad in comparison with others. Therefore, understanding how students compare themselves among classmates and how this affects their choices is relevant for depicting their choice behavior. One can assume that the students who perceive themselves of high ability will choose more often the hard task compared to lower ability students. The figure below illustrates the relationship between perceived ability and the likelihood of choosing the hard task.

(13)

11

The vertical axis shows all four possible ability levels that students could rank themselves. The horizontal axis measures the mean for choosing the hard task. Here again the easy task is mark as 0 and hard is marked as 1, hence the mean shows the likelihood of choosing the hard task for each perceived ability level.

As figure 1 presents our prediction is correct, because students ranked at bottom 5% never chose the hard task, while the likelihood of choosing the hard task increases with higher perceived ability reaching to 0,6 for students who rank themselves at the top 1% of the class. This is an important factor to consider while analyzing for differences in the public and non-public treatments. Although students were randomly selected to different treatments, it is possible that they were allocated in a way that some treatments had more high ability students than other treatment, which might jeopardize the results. Therefore, I will include a control variable for ability in my regression analyses.

Table 2 continues the analyses using OLS regressions to measure the impact of different treatment effects on choices. The coefficients from the first regression confirm my initial findings as public results has higher rates of hard task being chosen than the control group. The positive coefficient means that when the results for hard task were announced publicly students were 0,178 more likely to choose the hard task compared to non-public group. This difference is significant at 5% level. If we look at the public choices treatment, there we see the opposite effect as compared to control group, when the choices were made public students chose less often the hard task, however this difference is not significant as the p-value is 0.673.

0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 Bottom 5%

Average Top 5% Top 1%

Figure1:Liklihood of choosing the hard task based on

perceived ability

(14)

12

Table 2: OLS estimates of public effect on task choice

Task Difficulty (1) Task Difficulty (2) Coefficient (SE) p-value Coefficient (SE) p-value Intercept 0.419*** (0.058) 0.000 -0.119 (0.181) 0.509 Public Results 0,178** (0.081) 0.031 0.233*** (0.081) 0.004 Public Choices -0.033 (0.079) 0.673 -0.022 (0.078) 0.778 Perceived Ability 0.276** (0.133) 0.039 Self-confidence 0.016 (0.028) 0.573 Risk-Attitude 0.036** (0.017) 0.039 Challenge -0.111 (0.181) 0.111 Observations 229 226 R2 _0.034 _0.109 Adjusted R2 _0.025 _0.085 * p <0.10, ** p < 0.05, *** p < 0.01

Note: The table shows coefficients from OLS regressions of a treatment dummy on a binary indicator for choosing between hard or easy task. The dependent variable is non-public treatment. Robust standard errors are presented in parentheses. Two separate regressions are presented with same dependent variable; one without the other with control variables. The p-values from a t-test are presented in a separate column.

The first part of table 2 presents the results of a simple regression; however, there might be other factors, which are not included in initial regression, but are likely to influence the outcome. As discussed previously perceived ability is positively related to likelihood of choosing the hard task, hence it might affect the results from the regression analyses, therefore we should include a control variable for perceived ability.

Lenney et al. (1983) discuss potential effects of self-confidence on individual’s choice between task difficulty levels. While the main analyses in their studies concentrate on gender differences,

(15)

13

they prove that self-confident people are more likely to choose the harder task. Therefore, controlling for self-confidence, I will make sure that the difference between treatments is not merely driven by the variation in the number of self-confident people in each treatment. Also other factors such as risk attitude, challenge tolerance might affect the choice between hard or easy task and if between treatments these factors are different then they might bias the results. Hence, I conducted further regressions with perceived ability, self-confidence, risk attitude and challenge tolerance5_{as control variables.}

The results from second part of table 2 are still consistent with original regression analyses. The coefficient for public results has increased from 0.178 to 0.233, but it again shows that with announcing the results publicly participants were more likely to choose the hard task, hence adding the control variables increases the effect of public announcement and the coefficient is still significant. What regards to public choices treatment adding control variables decreases the difference between the control treatment and public choices;- however, the sign of the coefficient is still negative and insignificant.

Looking at the coefficients for control variables one can see that only perceived ability and risk-attitude are significant. As the figure in earlier analysis has shown and the regression outcome also confirms, high ability students are more likely to choose the hard task. The same applies for risk-attitude as students with higher willingness to take risks are 0.036 percent more likely to choose the hard task. Self-confidence has similar direction as risk-attitude, however this effect is not significant. What refers to challenge, based on the questionnaire results we get a negative coefficient for challenge, however this is not significant. By adding the control variables we can also see that the explanatory power of our variables has increased a bit as the adjusted R-squared is higher that the initial one.

5.2 Gender effect on task choice by treatment type

In this section I will try to analyze whether gender plays a role in deciding between hard or easy tasks. As mentioned in the hypotheses we should see females choosing less times the hard task compared to males;- however, it is also interesting to see the differences in patterns for different treatments. Hence, I will compare and investigate the gender differences in overall decision making and between the treatments.

(16)

14

Figure 2 shows the proportion of participants who chose the hard task in each treatment presented by gender. Visually we can see that there is no much difference between public choices and non-public treatments as they have almost the same average number of subjects who have chosen the hard task. However, there is a more noticeable difference between the treatment with public results and the two other treatments, with averages of 0.61 for male and 0.58 for Female participants. This shows that on average men are more likely to choose the hard task;- however, for women the difference between public results and other treatments is bigger than for men. Women are 2.23 times more likely to choose the hard task when the results are announced publicly compared to when only the choices are announced, while men are only 1.20 times more likely to go for hard task when public results are compared to public choices treatment.

Note: The graph shows averages for choosing the hard task for each treatment separated by gender. The vertical axis shows the likelihood of choosing the hard task, while the horizontal axis is the three treatments by gender.

The number of males and females participating was around the same, however 65 males took the hard task while only 41 females and this difference is statistically significant with a p-value of 0.0083. Although overall men chose more often the hard task, the figure shows that the difference between public results treatment and other two treatments is more prevalent for women than for men. A brief look at the bars shows that women were at least two times more likely to choose the hard task when the results were announced compared to when they were not. This difference in preferences is quite striking. Hence, it would be interesting to see the statistical significance tests for differences between treatments by gender.

Public Results Public Results Public Choices Public Choices Non-public Non-public 0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 Male Female Av era ge n u m b er o f h ar d t asks ch o se n

Figure 2: Willingness to take the hard task by gender and

treatment type

(17)

15

Table 3: Descriptive statistics by gender

Male Female Difference

p-value(1)&(2) 0.3891 0.0036** 0.8133(1) p-value(1)&(3) 0.4441 0.0118** 0.0192**(2) p-value(2)&(3) 0.9170 0.8557 0.0342**(3)

Note; The table presents p-values of t-test results for difference in treatments by gender. The notion (1) presents public results, (2) public choices and (3) non-public treatments. Hence the first column shows the difference in two respective treatments for male participants and the second column shows the same for females. The last column compares the difference between genders in choosing the hard task for each treatment accordingly.

As table 3 shows for men the difference between all three treatments is not significant, thus in all of the treatments around the same number of males chose the hard task. However, for women the difference between public results and public choices and the difference between public results and non-public are significant. Therefore, women are significantly more likely to choose the hard task if the results are announced rather than not.

To continue the analysis for gender difference I will run OLS regressions controlling for perceived ability6 as previous regressions showed ability plays a role in decision making. I will not include risk-attitude as a control variable for this regression, because risk taking behavior might itself be one of the explanations why women choose differently than men, hence it can be an endogenous variable. Charness & Gneezy (2012) and other large part of gender based literature states that there might be a difference in risk taking between men and women, which might explain why women choose less competition and in my experiment this might explain why women choose less times the hard task. Although some authors (Booth and Nelon 2009) argue that observed gender differences in risk-attitude might reflect social learning rather than inherit gender traits, generally it is still highly believed in economics that there is some difference between male and female risk taking.

6_{In the regression analysis for gender effect I want to look only at perceived ability and risk attitude as previous} regressions showed that only these factors had significant influence on outcome variables. Hence, I have excluded analyzing self-confidence and challenge in further analysis as they did nothave important contribution.

(18)

16

Table 4: Public announcement effect on task choice by gender

Task Difficulty Male (1) Task Difficulty Female (2) Task Difficulty Interaction (3) Coefficient (SE) p-value Coefficient (SE) p-value Coefficient (SE) Intercept 0.231 (0.161) 0.162 0.055 (0.151) 0.716 0.246* (0.127) Public Results 0.147 (0.117) 0.210 0.281** (0.118) 0.019 0.143 (0.111) Public Choices 0.009 (0.110) 0.934 -0.032 (0.104) 0.758 0.008 (0.105) Perceived Ability 0.380** (0.182) 0.046 0.337* (0.195) 0.089 0.360*** (0.135) Female -0.207* (0.113) Public Results x Female 0.136 (0.162) Public Choices x Female -0.041 (0.154) Observations 119 110 229 R2 _0.041 _0.117 _0.105 Adjusted R2 _0.016 _0.093 _0.081

Note: The table shows regression results based on gender controlling for perceived ability. The dependent variable is non-public treatment. All columns are separate regressions and the first two present results for male and female participants, while the first column includes interaction term for treatments and gender. Robust standard errors are presented in parentheses.

As we can see from Table 4 regression results, for men both the coefficients for public results and choices are positive, which means that male are more likely to choose the hard task when the results or the choices are announced in class, however this difference is not significant. On the other hand looking at the results for female, we see that for public results they are also more likely to choose the hard task compared to non-public and this difference is significant. Moreover, the negative coefficient for public choices shows that surprisingly women are less likely to choose the hard task when only choices are made public, however the result is not

(19)

17

significant, hence we cannot derive a solid conclusion from this part. Perceived ability is positive and significant for both male and female subjects. Furthermore, to test the significance of gender difference in treatment effects I have included interaction terms in the third column of table 4. The coefficient from interacting public results treatment with female is positive, but not significant. This means that while women react twice as strongly to public results treatment compared to men, this gender difference between treatments is not significant.

At last it might be interesting to see how well the subjects completed their chosen task difficulty level and whether this varies between the treatments.

Table5: Average scores by gender and treatment type

All Public Results Public choices Non-public

Male 6.848 (2.889) 8.361 (3.457) 5.171 (1.745) 7.19 (2.432) Female 6.854 (3.139) 7.61 (3.306) 6.285 (3.366) 6.75 (2.489)

The standard errors are presented in the parentheses.

As the table shows overall there is no much difference between males and females as on average they both completed almost 7 out of 15 matrices. However, between the treatments there is more difference, as males did better in public results treatment, while females did better in public choices and non-public treatments. Hence, we can also conclude that while women choose more often the hard task for public results treatment compared to men they scored less on average. Overall, analysis lead to a conclusion that there is a significant effect from publicly announcing the results, as participants chose more often the hard task when the results were announced compared to when they were not. The treatment effect was stronger for women than for men, but not significantly, hence the effect is not solely gender based. These findings, do not follow my initial hypotheses, hence in the next section I will discuss several possible explanations for observed behavior from participants.

(20)

18

5.3 Possible explanations

My first hypothesis that students should choose more often the easy task when results are announced publicly is rejected, as students chose more often the hard task in the public results treatment. This might be partially explained with the theory about social status concerns suggested by Veblen (1915). He proposed that people might behave in a certain way to signal their social status to others. A model by Rege (2006), also confirms my findings, proving that people may care for social status because it serves as a signal about their abilities, hence in my experiment as well students chose the hard task in the public results treatment to signal their classmates about their abilities. The results show that high ability students were more likely to choose the hard task compared to lower ability ones, therefore we can assume that when top students decided to choose the hard task, they were more confident that they will pass the test, hence announcing the results was a motivation instead of discouragement.

Secondly, from the data we also rejected the hypothesis that students should choose more often the hard task when only choices are announced. This is a strange finding as with public choices treatment there was a “free” possibility to show-off by choosing the hard task, without any fear of failure as the results will not be announced publicly, hence nobody will know in case of failing. In contrast, what the findings reveal is that students chose less often the hard task with public choices treatment compared to public results. One possible explanation might be that students do not want to choose the hard task if they can’t show off their scores. Hence, exerting additional effort on the hard task they want to make sure that other students can also see their high achievements.

The hypothesis regarding the overall gender effect is confirmed as the analyses showed that on average men chose the hard task more often compared to women. Both males and females were more likely to choose the hard task under public announcement of results. However, according to my hypothesis 4.1 we should have seen females choosing less often the hard task compared to men for public results, which did not happen as a result of the experiment. Instead what the results showed there was a significant difference between public results and other two treatments for women, as surprisingly women preferred to choose more the hard task when the results were announced. One potential explanation can be that females wanted to show that they are capable of solving a hard problem and as only public results treatment allowed to prove everyone that they can choose the hard task and get high results they preferred to choose more often the hard task only for this treatment. The findings might be a result of an unequal roles of men and women in society, especially in developing countries, where majority of women are still believed to be

(21)

19

able to solve only the less challenging tasks, while men are in charge of complex decisions. Hence, women are still trying to show everyone that they also might be good at solving hard problems and apply for high positions in the job market.

6. DISCUSSION AND CONCLUSION

The public announcement effect is highly investigated in charity and donations context, where the findings confirm that people are hugely affected by the fact that others can observe their behavior and when public component is introduced people donate more. My study partially coincides with these findings as when results are announced publicly we see more students going for the hard task compared to when there is no public announcement. For men the showing off effect for the public treatment is not new, as huge amount of literature including Böhm (2013) confirm that men are very sensitive to social cues and in front of others they try to look better. However, it is interesting to see this effect, although not significant, for public results treatment among females, who are traditionally believed to be less capable then their male counterparts. Although women have achieved significant results in many fields, they are still viewed as less competitive and are less represented in the labor market in Armenia as the report of Asian bank on country gender assessment reveals7.

From the educational perspective also there are still some disciplines that are viewed as more feminine or masculine. Young women tend to choose the traditionally female subjects, which are humanities and languages, while men choose the technical fields which are believed to be harder and later are better paid8_{. This problem is present also in other countries as the studies by}

Inter-American development bank reveal there is still a gender gap, as women are

7 _{Asian Development Bank “Armenia- Country Gender Assessment Report 2015”}_._{Quote from the report:}

“Gender stereotypes contribute to women’s lower levels of representation in politics, in formal employment, and as business leaders. Men canre also be negatively impacted by stereotypes, especially those that portray men as solely responsible for providing for their families financially—norms that are often increasingly difficult to fulfill given the realities of the labor market in Armenia today.”

(22)

20

underrepresented in STEM (Science, Technology, Engineering and Mathematics) disciplines9. Hence, starting from schooling years there is a gender difference in the choices that women and men make for their future. This might explain why when men and women with almost equal abilities are presented to choose between hard and easy task, men are more likely to go for the hard task. Therefore, women are more inclined to show the society that they are capable of handling the hard tasks and hence their choices are more differentiated among the public results and other treatments.

Overall, the study was conducted only in one high school to control for school specific differences, however this fact also acts as a limitation for external validity, because the experiment does not allow to predict whether we will have the same results in different schools for different countries. Moreover, the students were aware that they are participating in an experiment, which might also bias the results. Even though the experiment was conducted in a more natural setting for students and the format was familiarized to their usually ways of completing written tasks in class, still the fact that they knew they are passing an experiment might change their behavior.

Throughout the analyses ability was measured with questionnaire results, which is what students ranked themselves according to their beliefs about their comparable abilities. However, this might not portray real abilities. Unfortunately, due to confidentiality and ethics issues, I could not collect data on their actual ability compared to other students and the only other measure I could compare was the scores from the task. Therefore, for further development of the study, it is beneficial to compare GPA or other academic results and task choices to understand whether real ability plays a role in decision making and whether only high achieving students chose more often the hard task when the results were public.

Even though there are many factors that still could be further investigated, what this study shows is that public announcement not only affects the decision for donations, but it also affects the choice between hard or easy tasks. This can be applied to work situations as it turns out people are more motivated to take challenging tasks when they are watched by others and everyone can observe their end result. Overall, once again it confirms that public image concerns play a role in the decision making process, especially for women who are traditionally underestimated.

9 Rafael Castillo, Matteo Grazzi and Ezequiel Tacsir (2014). What does the literature say? Inter-American Development Bank No. IDB-TN-637

(23)

21

7. REFERENCES

Adriaan R. Soetevent 2004. “Anonymity in giving in a natural context—a field experiment in 30 churches” Journal of Public Economics89 (2005) 2301–2323

Agnieszka Tymula and Jackson Whitehair 2015 “Watched by a Stranger: Influence of Observation on Individual Decision Making Under Risk and Ambiguity” Social Science

Research Network Working Papers 2015-08, University of Sydney, School of Economics.

Alison L. Booth and Patrick J. Nolen February 2009. “Gender Differences in Risk Behaviour: Does Nurture Matter?” Institute for the Study of Labor IZA DP No. 4026

Asian Development Bank “Armenia- Country Gender Assessment Report 2015”

Barry R. Schlenker, Michael F. Weigold and Kevin Doherty 1991. “Coping with accountability: Self-identification and evaluative reckonings.” Pergamon general

psychology series, Vol. 162, (pp. 96-115)

Billie Hughes, Howard J. Sullivan and Mary Lou Mosley 2016 “External Evaluation, Task Difficulty, and Continuing Motivation” The Journal of Educational Research, Vol. 78, No.

4 (Mar. - Apr., 1985), pp. 210-215

Billie J. Hughes, Howard J. Sullivan and James Beaird 2016 “Continuing Motivation of Boys and Girls under Differing Evaluation Conditions and Achievement Levels” American

Educational Research Association Volume 52 (10) – Dec 1

Blanton, Hart; Buunk, Bram P.; Gibbons, Frederick X.; Kuyper, Hans 1999 “When better-than-others compare upward: Choice of comparison and comparative evaluation as independent predictors of academic performance.” Journal of Personality and Social

Psychology, Vol 76(3), 420-430

Botond Köszegi 2006, “Ego Utility, Overconfidence, and Task Choice” Journal of the

European Economic Association, Volume 4, Issue 4, pages 673–707

Bruce Rind and Daniel Benjamin 2001. “Effects of public image concerns and Self-image on compliance.” The Journal of Social Psychology, 134(1), 19-25

(24)

22

Dan Ariely, Anat Bracha, Stephan Meier 2007 August “Doing Good or Doing Well? Image Motivation and Monetary Incentives in Behaving Prosocially” Institute for the Study of

Labor IZA DP No. 2968

Ellen Lenney, Joel Gold and Chris Browning 1983. “Sex differences in self-confidence: The influence of comparison to others' ability level”. Sex Roles September 1983, Volume 9, Issue 9, pp 925-942

Gary Charness and Uri Gneezy 2012. “Strong Evidence for Gender Differences in Risk Taking”. Journal of Economic Behavior & Organization 83 (2012) 50–58

Harter, S. (1978). Pleasure derived from challenge and the effects of receiving grades on children's difficulty level choices. Child Development, 49, 778-798

James P. Byrnes, David C. Miller and William D. Schafe 1999. “Gender Differences in Risk Taking: A Meta-Analysis”. American Psychological Association Vol. 125, No. 3, 367-383 Johnmarshall Reeve 1978 “Intrinsic Motivation in Competition: The Intervening Role of Four Individual Differences following Objective Competence Information” Journal of

Research in Personality 21, 148-170 (1987)

Lisa R. Anderson and Sarah L. Stafford, The College of William and Mary “An Experimental

Study of the Effect of Announcements on Public Goods Contributions”

Mary F. Luce, James R. Bettman and John W. Payne 1997 “Choice Processing in Emotionally Difficult Decisions” Journal of Experimental Psychology Vol. 23, No. 2, 384-405

Michael F. Weigold and Barry R. Schlenker 1991. “Accountability and Risk Taking.”

Personality and Social Psychology bulletin vol. 17 no. 1 25-29

Muriel Niederle and Alexandra H. Yestrumskas 2008 “Gender differences in seeking challenges: the role of institutions” National Bureau Of Economic Research Working Paper No. 13922

Paul E. Levy and Ann H. Baumgardner 2016 “Effects of Self-Esteem and Gender on Goal Choice” Journal of Organizational Behavior, Vol. 12, No. 6 (Nov., 1991), pp. 529-541 Rachel F. Adler, Deena Rubin, Abdul Rahman Mohammad, Amna Irfan, Haridu Senadeera, Timothy Nguyen 2015 “That’s easy! The effects of objective and subjective task difficulty when multitasking” Science Direct

(25)

23

Rege, M. (2008). Why Do People Care about Social Status? Journal of Economic Behavior

& Organization 66(2), 233–242.

Robert Böhm and Tobias Regner 2013. “Charitable giving among females and males: an empirical test of the competitive altruism hypothesis”. Springer Science and Business Media

New York J. Bioecon (2013) 15:251–267

Roland Bénabou and Jean Tirole 1999 “Self-confidence and Intrapersonal Strategies” Roland Bénabou and Jean Tirole 2001 “Self-confidence and Personal Motivation” The

Quarterly Journal of Economics (2002) 117 (3): 871-915

Veblen, T. (1915). The Theory of the Leisure Class: An Economic Study of Institutions. Macmillan.

Shawn P Curley, J. Frank Yates and Richard A. Abrams 1986. “Psychological Sources of Ambiguity Avoidance”. Organizational Behavior and Human Decision Processes 38, 230-256 (1986)

Thomas Buser 2016 “How does the gender difference in willingness to compete evolve with experience?” Tinbergen Institute Discussion Paper TI 2016-017/I

Weidong Li 2004 “Examining the relationships between ability conceptions, intrinsic motivation, persistence, and performance” Louisiana State University

(26)

24

APPENDIX: Experiment and instructions

Main experiment

Easy Task Number

Please, circle the two numbers that add up to 10 in each matrix. You cannot use a calculator. 3,73 5,19 1,69 7,23 8,41 4,63 5,37 6,4 2,37 6,37 7,33 1,95 2,77 5,67 3,63 3,13 3,72 8,25 6,84 7,29 4,44 5,76 2,71 3,51 8,36 6,29 2,61 6,36 8,46 3,66 1,54 6,74 4,71 3,54 5,49 8,67 7,35 1,15 5,18 3,65 5,78 8,95 4,82 7,45 4,42 7,37 2,64 4,56 3,25 6,63 5,64 4,26 4,73 6,75 7,61 3,49 6,51 4,54 5,36 7,61 3,99 5,16 2,39 6,43 3,67 7,54 2,46 2,13 5,43 7,97 5,73 4,37 2,78 3,78 7,86 5,94 5,28 2,24 2,78 3,78 4,72 5,43 3,96 2,83 5,37 6,13 3,64 7,17 4,47 2,93 5,64 6,27 7,64 4,46 4,37 5,23 6,13 2,56 3,87 6,38 4,48 5,27 5,48 4,78 3,82 3,22 5,52 7,38 6,91 5,29 2,61 5,59 3,19 4,31 7,39 7,41 6,51 6,65 7,15 3,85 4,84 5,16 6,74 7,45 3,96 4,26 6,75 4,75 8,21 3,85 5,49 5,71 6,39 4,29 6,25

(27)

25

Hard Task Number

Please, circle the two numbers that add up to 9,38 in each matrix. You cannot

use a calculator.

6,75 4,75 8,21 2,83 5,49 2,99 6,39 4,69 6,25 6,91 5,29 2,61 5,53 3,14 2,31 6,37 5,44 3,85 6,65 6,15 4,85 3,96 2,13 6,25 4,45 4,84 4,54 3,57 3,96 2,83 6,47 3,96 4,21 4,84 5,81 5,72 5,62 6,27 3,74 2,46 6,26 5,22 5,16 2,56 3,12 6,28 3,31 5,07 6,43 4,78 3,05 2,95 5,52 7,38 6,43 3,67 2,54 2,96 2,13 4,23 6,97 5,15 4,37 7,61 3,49 6,64 2,74 5,36 1,67 3,92 5,16 2,39 2,76 3,73 2,85 5,94 5,22 2,24 3,95 3,78 5,43 7,35 1,13 5,18 3,65 5,78 1,43 5,83 7,45 7,95 6,36 3,42 3,63 1,54 5,74 7,74 2,52 5,49 5,96 6,37 2,64 4,56 2,83 6,55 4,62 4,26 4,73 5,75 6,31 7,33 1,95 2,77 5,67 3,61 4,19 3,61 5,19 3,73 5,19 1,69 7,23 1,45 1,65 5,37 5,63 3,75 2,81 2,29 6,16 5,76 2,72 3,52 3,22 6,29 5,68

(28)

26

Questionnaire Number 1. Gender

o Male o Female

2. Please circle your self-confidence level in general ( 1- very low 7- very high) 1 2 3 4 5 6 7

3. How do you see yourself: Are you generally a person who is fully prepared to take risks or do you try to avoid taking risks? ( 0= unwilling to take risks 10=fully prepared to take risks)

0 1 2 3 4 5 6 7 8 9 10

4. How would you classify yourself among your classmates based on your abilities? o I am among top 1% of the class

o I am among top 5% of the class o I am an average student

o I am among bottom 5% of the class

5. Please, explain why did you choose the specific task difficulty level?

6. How is your attitude towards challenges: o I love challenges

o I try to avoid challenges

7. You can learn new things, but you can’t really change your basic intelligence o Agree

o Mostly agree o Mostly disagree o Disagree

8. I consider success is about o Learning

o Proving everyone that you are smart

9. How important is it for you what your classmates think about you? o Very Important

o Slightly important o Not important

(29)

27

INSTRUCTIONS10

General

Hello everyone, today you are going to pass a short experiment, which measures your abilities and effort on a specific task. The experiment will consist of two types of tasks: easy and hard, and you have to choose which one you prefer to do before the start of the experiment. Afterwards you will be given 5 minutes to complete the task. If you have answered correctly to 10 out of 15 questions you have passed the task, if less than 10 were correct you have failed the task.

Treatment 1

For students who have chosen the difficult task the results will be announced in class after the experiment. Pass or fail will be announced only for the difficult task. For the students who have chosen the easy task the results will not be announced, hence this will only be a good exercise for them to know their own capabilities. Are there any questions? Now I will come to each of you and you have to tell out loud your name and which task you prefer and I will give you your preferred task difficulty level. You cannot start the task until instructed to do so, so please after receiving the paper do not flip it over yet. Now you can start! And you have 5 minutes to complete the task.

The time is up, please, stop! Now I will collect your answer sheets and will give you a short questionnaire to fill in. Please, write your number on the questionnaire and also your class code.

(Collecting questionnaires and announcing pass/fail for the difficult task)

This is the end of the experiment, thank you for your participation and wish you success in your future studies.

Treatment 2

The results of the tasks will not be announced and hence nobody will know how you did on the test, so this is just a self-check exercise to know your own abilities. Now I will come to each of you and you have to tell out loud your name and which task you prefer and I will give you your preferred task difficulty level. You cannot start the task until instructed to do so, so please after receiving the paper do not flip it over yet.

Now you can start! And you have 5 minutes to complete the task.

10_{All experiment including instructions were translated in Armenian to make sure there are no language barriers} for understanding the requirements of the task.

(30)

28

Treatment 3

The results of the tasks will not be announced and hence nobody will know how you did on the test, so this is just a self-check exercise to know your own abilities. Now I will approach you and you have to choose which task you want to complete, only you will know the level of difficulty for your task.

Now you can start! And you have 5 minutes to complete the task.

Showing off effect : does being public affect the choice between hard or easy tasks and how does it depend on gender?

FACULTY OF ECONOMICS AND BUSINESS

MASTER THESIS