The revenge : does an unpleasant treatment lead to the preference for a larger outcome discrepancy?

(1)

The

The revenge:

revenge:

Does

Does aaaan

n

n unpleasant

unpleasant

unpleasant treatment

treatment

treatment lead

lead

lead to

to

to the

the

the preference

preference

preference for

for

for aaaa

larger

larger outcome

outcome

outcome d

d

discrepancy

iscrepancy

iscrepancy????

Yingqi Miao 10712747

MSc ECO

Behavioral Economics and Game Theory Prof. Frans Van Winden

(2)

Statement

StatementStatementStatement ofofofof OriginalityOriginalityOriginalityOriginality

This document is written by Student Yingqi Miao who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is 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 contents.

(3)

Abstract Abstract AbstractAbstract

Studies have been done on social relationships and various aspects of their behavioral consequences. However, current studies often ignore how, under the effects of social relationships, one would perceive and behave given various differences between his own outcome and that of his opponent. This paper applies an experimental method to investigate two questions. First, how does a negative social relationship develop from an unpleasant treatment? Second, how does the negative social relationship influence one's preference in a lottery for options with various levels of outcome discrepancy? Results of the experiment showed that the non-positive emotional reaction towards an unpleasant treatment from the interaction has an impact in the development of a negative social relationship. Also, the negative social relationship promotes the preference for a large outcome discrepancy with the other.

Content Content ContentContent 1. 1.

1.1. IntroductionIntroductionIntroductionIntroduction············3333 2.

2.

2.2. MethodologyMethodologyMethodologyMethodology············5555 2.1

2.1 2.1

2.1 ExperimentalExperimentalExperimentalExperimental designdesigndesigndesign············5555 2.2

2.2 2.2

2.2 ExperimentExperimentExperimentExperimentalalalal procedureprocedureprocedureproceduressss············8888 2.3

2.3 2.3

2.3 HypothesesHypothesesHypothesesHypotheses············10101010 3.

3.

3.3. ResultsResultsResultsResults············11113333 3.1

3.1 3.1

3.1 DescriptiveDescriptiveDescriptiveDescriptive analysisanalysisanalysisanalysis············11113333 3.2

3.2 3.2

3.2 RegressionRegressionRegressionRegression analysisanalysisanalysisanalysis············11117777 3.2.1

3.2.1

3.2.13.2.1 VariablesVariablesVariablesVariables············11117777 3.2.2

3.2.2

3.2.23.2.2 RegressionRegressionRegressionRegression resultsresultsresultsresults andand analysisandandanalysisanalysisanalysis ············11118888 4.

4.

4.4. ConclusionConclusionConclusionConclusion············22222222 Reference Reference ReferenceReference············22225555 Appendix Appendix AppendixAppendix············22227777 Appendix Appendix Appendix

Appendix A.A.A.A. ExperimentExperimentExperimentExperiment instructionsinstructionsinstructionsinstructions············22227777 Appendix

Appendix Appendix

Appendix B.B.B.B. QuestionnaireQuestionnaireQuestionnaireQuestionnaire············22229999 Appendix

(4)

1. 1.

1.1. IntroductionIntroductionIntroductionIntroduction

Economic behavior is often affected by various factors such as decisions of other people, comparing payoff with others or with one’s expectation. One reason behind this is that people are connected to each other during economic interactions where one treats others and receives a treatment from others. The feature of an interaction generates different effects. Think of theories like reciprocity, social reference, and the counterfactual effect. At detailed inspection, social relationships develop from interactions and change the way people perceive each other. For instance, in a two-person interaction such as a centipede game or a Bertrand game, a player’s choice between cooperation and noncooperation is not only a decision of his own but also a treatment to his opponent. The results caused by the decision affect the way people feel about each other and their emotions. Furthermore, in this very social context, a certain relationship between two players develops. The relationship between players who chose to cooperate is different from that between players who did not cooperate. Therefore, people’s behavioral reactions differ from each other based on the feature of the social relationship.

Before discussing the effects of social relationships on people’s behavior, it is essential to make clear the definition of a social relationship and how it develops. There is no formal and universal concept of a social relationship in related literature. However, a social tie is often defined and used. According to Sonnemans, van Dijk and van Winden (2004), individuals have neutral feelings towards strangers, thus, how much an individual cares about the well-being of the other person is a key in determining the social tie. They also considered the critical component of a social tie to be sentiments as disclosed in a survey on interpersonal attachments in the paper of Baumeister and Leary (1995). Attanasi, Hopfensitz, Lorini and Moisan (2014) gave a definition of a social tie in their paper based on the psychological concept of social identity and they also named specific examples of positive social ties like family relatives and close friends. In this paper, the benchmark of a social relationship is the neutral relationship between two strangers. But after an interaction, they attach weights to the utility of their opponent, either positive or negative. A positive or negative social relationship develops based on the characteristic of the treatment in an interaction, which could also trigger different emotional reactions.

With a clear understanding of a social relationship, attention should be transferred to its effects on people’s behavior. Research has been done investigating behavioral consequences of social relationships on various dimensions. Van Dijk, Sonnemans and van Winden (2002) examined the formation of social ties in a public good experiment where a circle-test is designed for the measurement of social ties. Linde and Sonnemans (2012) focused on how social ties affect people's risk preference. Another example stresses the effect of positive social relationships upon cooperation (Attanasi, Hopfensitz, Lorini and Moisan, 2014). However, there is little research focusing on the aspect of comparing outcomes with others. Therefore, one of the focus of the paper concerns the question that when one compare himself with

(5)

another person, whether he prefers a large or a small difference between the outcomes. In the remainder of this paper, the difference between the outcome of oneself and that of his opponent will be referred as the outcome discrepancy.

It is not uncommon for a decision maker to evaluate his outcome by comparing with others. During comparison, two issues will be noticed - the valence of the outcome discrepancy and how big the discrepancy is. Firstly, speaking of the valence of an outcome discrepancy, one can find that his outcome could be either larger (positive) or smaller (negative) than that of his opponent. Thus in this paper, the valence of the outcome discrepancy will be defined in the form of social situations - a social gain situation where a person earns more than his opponent and a social loss situation where he earns less. A neutral situation is not included and discussed here since no outcome discrepancy is observed when the outcomes are equal. Secondly, the preference of how big the discrepancy is reflects a person’s attitude towards inequity in some degree. People who prefer a large outcome discrepancy are likely to have a different inequity attitude from those who prefer a small outcome discrepancy. Moreover, to a decision maker, a large discrepancy does not always mean the same under different social situations. Rational individual theory obviously suggests that a person prefers a large discrepancy in a social gain situation rather than that in a social loss situation.

A combination of these two issues discussed in the previous paragraph - the social situation and the outcome discrepancy - naturally intrigues the following question. What type of outcome discrepancy does a decision maker prefer in a certain social situation? Trying to answer it, one can think of theories like prospect theory, inequity aversion and loss aversion. As per individual decision maker, the preference for avoiding losses over acquiring gains indicates “losses loom larger than gains” (Kahneman and Tversky, 1979). Moreover, efforts have been made on extending the theories into a social context. Fehr and Schmidt (1999), for example, proposed that people pay more attention to being worse off than being better off than others. Bault, Coricelli, and Rustichini (2008) investigated experimentally and found that in the social domain, gain seems loom larger than loss, which turns out to be an opposite result compared to that in the private situation. Linde and Sonnemans (2012) concluded from their experiment about risky choices in a social context that “participants are more risk averse” in a loss situation (where they could earn no more than their opponent) than in a gain situation. In most of the studies, the type of social situation is explicit to decision makers. However, there is little research exploring the behavior when the type of the social situation is unknown during the decision-making process. As a result, this paper investigated the preference for the outcome discrepancy in an uncertain social situation. That is, participants do not know what type of social situation they would end up in when they are asked to choose among options with various outcome discrepancies.

In previous paragraphs, the valence of social relationship and the preference for outcome discrepancy are mentioned respectively. Studies have been done on both

(6)

but a combination of these two topics is often ignored. As a result, what I am interested in is that whether the valence of social relationship has any influence on the preference for outcome discrepancy. To answer this question, it should be made clear about the establishment and development of a social relationship from a certain interaction. As a result, I conducted a two-part experiment to find answers to the research questions. The first part of the experiment allows the development of social relationships and the second part investigates the behavioral effect of the social relationship upon the preference for outcome discrepancy. In order to distinguish the influence of social relationships from other possible factors such as inequity attitudes and risk attitudes, two groups with different treatments are designed and set up in the experiment. To be specific, in the treatment group, social relationships are developed in Part 1 so that they could generate influence on the following behavior. However, social relationships are established in the control group as well but they cannot have any effects on how participants choose in Part 2. This method ensures that the difference between the behavior in two groups could be only caused by the valence of social relationships. The design and conduction of the experiment will be explained in detail in section 2.

The remainder of this paper is structured as below. A description of the methodology will be given in section 2, including the experimental design, the conduction of the experiment and the hypotheses. Section 3 introduces the main findings of the experiment and provides the statistical analysis via a regression model. Section 4 concludes.

2. 2.

2.2. MethodologyMethodologyMethodologyMethodology

In order to find answers to the research questions discussed before, an experiment was designed to observe behavior. This section explains the methodology consisting of the design and procedures of the experiment, followed by the main hypotheses of the behavior.

2.1 2.1

2.12.1 ExperimentalExperimentalExperimentalExperimental designdesigndesigndesign

The experiment starts with a dictator game, allowing the establishment of social relationships. The dictator game is followed by a lottery game in order to observe the effects of social relationships. Two treatments are applied in this experiment - a treatment group for investigating the behavioral consequences of social relationships and a group where social relationships are developed but cannot have any effects on behavior as a control. The sequence of the experimental events is presented in a chronological order as shown in table 1. A detailed description of the conduction of the experiment will follow in section 2.2. During the experiment, participants are given instructions and related information sequentially, which means they are unable to have any knowledge of what will happen next. The instructions, along with the

(7)

questionnaire, a formal communication model with the participants will be given in the appendix. Now different parts of the experiment will be discussed in more detail.

Table Table

TableTable 1111 The sequence of experimental events

Event Treatment group Control group

Random determination of identities ID number, role and group

Part 1. Dictator game Random paired with a partner

Part 2. Lottery game Paired with the same

partner as Part 1

Paired with a different partner from Part 1

Questionnaire Based on the role of the participant

Random determination of lottery The result of Part 2

Part Part

PartPart 1111 ---- thethethethe dictatordictatordictatordictator gamegamegamegame

The experiment starts with a dictator game to allow participants to develop social relationships with their randomly determined partner for this part. At this stage, participants will be only given the instruction of Part 1, without any knowledge of what will happen next. In a dictator game, the dictator is sometimes referred as “the proposer”, and he is asked to split an amount of money and offer to “the recipient”. (e.g. Forsythe et al., 1994; Hoffman et al., 1994; Bolton & Zwick, 1995). Slightly different, in Instruction Part 1 of this experiment, a dictator is referred as a “giver” instead. Similarly, the amount of money for the dictator to split is in the form of points which could be possibly converted to euros. Moreover, the distribution of the fixed amount of the points between the giver and the recipient is described as a “gift”. To my knowledge, some experiments show that the gesture of giving or withholding a gift has strong positive or negative influence on a recipient’s behavior (Malmendier and Schmidt, 2012). Based on that, it is reasonable to believe that a gift-giving context would promote the emotional reactions during the establishment of social relationships to some degree and, therefore, enable the behavioral effects of social relationships to be more significant. To be specific, in this experiment, a giver is given 50 points as the initial endowment. Then he needs to split it by giving a gift to the recipient, ranging from 0 to 50 points. This dictator game is a one-off, during which the participants do not have knowledge of what is coming next. Also, participants are informed of the independence of points earned from two parts of the experiment. This makes sure that possible strategic decisions are minimized because the givers will not be worried about the points from Part 2 being influenced by the decisions of their partners.

(8)

a two-person interaction. Therefore in order to make the recipients more relevant in this part, they are asked to write down their expectations of the gift they would get. As a result, the difference between the gift and the recipient’s expectation can be easily observed and used to judge the characteristic of the treatment.

Part Part

PartPart 2222 ---- thethethethe lotterylotterylotterylottery gamegamegamegame

A lottery game is then played where the recipients need to choose one out of five options. Each option has two possible outcomes that both can happen with the probability of 50%, and all the options have an expected value of 50 points. Table 2 below presents these options as in a social context to all participants (see also Appendix A - Instruction Part 2). In this way, after the result of the lottery is determined, both participants in a pair learn how many points they get from their own choice as well as the points earned by the other. Therefore, it is easy and straight forward for a participant to observe the difference between the outcome of himself and that of his partner. Since each option has two possible outcomes, two situations from the viewpoint of the recipient are therefore defined. The situation where the recipient earns more points than the giver is defined as the social gain situation (referred as “+” situation). On the other hand, the situation where the recipient gets fewer points than his partner is called the social loss situation (referred as “-” situation).

For each pair of participants, only one situation will come true. The result of the lottery will be determined later via flipping a virtual coin online. The head represents the “+” situation while the tail means that the “-” situation holds. Once the situation is determined, the points earned by each participant will be determined as well.

Table Table

TableTable 2222 The lottery game

Situatio n Probabil ity The giver's points

The recipient's points

A B C D E

+ 50% 60 70 75 80 85 90

- 50% 40 30 25 20 15 10

Now we will have a detailed inspection of the lottery game in the perspective of outcomes as shown in table 2. In this part, the outcome of the giver only relies on the randomly determined situation. To be specific, a giver will get either 60 points in “+” situation or 40 points in “-” situation. On the other hand, the recipient’s outcome depends on the situation as well as his own choice of the option. From the viewpoint of the recipient, the five options have different outcome discrepancies with the giver. For instance, among all five options, option A has the smallest absolute discrepancy

(9)

in outcomes - in the “+” situation 10 points more than the giver or in the “-” situation 10 points fewer than the giver. However, option E has the largest absolute outcome discrepancy - in the “+” situation 30 points more than the giver or in the “-” situation 30 points fewer than the giver. To sum up, these options differ from each other in both situations. This lottery game combines the uncertainty of social situation type and the outcome discrepancy level in the simplest way. While the recipient is choosing, the giver needs to write down what he expects his partner will choose. The choice of the recipient could be influenced by various factors during the decision-making process. Take risk attitudes for example. A person who is high risk seeking probably has a preference for option E. Therefore, in order to eliminate other possible factors and furthermore emphasize the significant effects of social relationships, a control group is set where social relationships are developed but cannot have any effects on the choices. As mentioned in table 1, participants will be coupled randomly with a partner in Part 1. However, slightly different in Part 2, a recipient in the treatment group will not change the partner while participants from the control group will have a new matched partner. In this way, if there are any differences in the choices between the two groups in Part 2, it could be concluded that it is affected by social relationships rather than other factors.

This main task of the experiment is followed by a questionnaire with different questions respectively for the giver and the recipient. In the questionnaire, basic information about the participants such as age and gender is collected. Two questions regarding their opinions should be paid attention to. First, recipients are asked to report their feelings about the results of Part 1. To be specific, they are supposed to choose one emotion out of five categories. Five emotional reaction categories are available to choose from - satisfied, surprised, felt nothing, unhappy but not angry and angry. The emotion category ‘surprised’ indicates positively surprised when a recipient received more than his expectation. Second, recipients are asked to give opinions of the type of the social relationship with their partner. To be concrete, they can choose among three options positive, neutral and negative -to define the social relationship. These two questions are highlighted here because the answers are used as the measures of emotional reactions and social relationships. After this questionnaire, the results of the lottery game for each pair are determined via an online tool called “Coin Flipper”.

2.2 2.2

2.22.2 ExperimentalExperimentalExperimentalExperimental proceduresproceduresproceduresprocedures

A total number of 40 students took part in this experiment. In the beginning, the identities of the participant, including a unique ID number and his role throughout the entire experiment are determined randomly. Also, there are two groups in the experiment - a treatment and a control. Thus, the experimenter assigned him in one of the groups randomly. All the preparation is conducted with the help of an online tool named “True Random Number Generator” (“https://www.random.org/”). The ID

(10)

number is an integer number between 1 and 40 representing every participant and each of the participants will be coupled with a partner. The two opposing roles - giver and recipient - are represented by the number 1 and 2, respectively. The experiment was conducted pair by pair thus, every two subjects participating in the experiment at the same time makes a pair. Once two participants are paired with each other, the experimenter uses the True Random Number Generator to generate a number determining the role for one of them, and naturally the other participant will have the other role. The determination of the group for participants is slightly different. To be specific, the first 10 pairs of subjects participating in the experiment will compose the treatment group while the other 10 couples are in the control group. From all the information about identities, a participant will be only informed of his own ID number and his role throughout the experiment. Also, the entire experiment is anonymous, thus, none of the participants will have access to any personal information of others.

Due to time and place restrictions, the experiment was conducted online with 20 pairs of participants. Each participant was contacted individually by the experimenter via the social media websiteFacebook. This website supports instant communication

and various types of file transportation including doc and pdf. The messages could be seen by two people who are communicating with each other. This feature ensures that all the information a participant could know comes only from the experimenter. This also makes sure that participants cannot communicate with each other during the experiment. In order to maintain neutrality, the communication with all participants was conducted following an identical message model, which could be seen in Appendix C. As shown in table 3, participants followed steps in terms of different roles.

Table Table

TableTable 3333 The steps of the experimental procedures for each pair of participants

Event Giver (G) Recipient (R)

Start Informed of ID number.

Step 1 Read the General Instruction.

Send a message to confirm once finished.

Step 2

Read Instruction Part 1.

Send a message of the gift. Send a message of the expectation.

Informed of the expectation. Informed of the gift.

(11)

The experiment was conducted first with the treatment group followed by the control group. In the treatment group, 10 pairs of subjects participated in the experiment one by one following the steps as mentioned in table 3. Two participants in a pair were contacted by the experimenter simultaneously but individually because the Facebook website communicate page allows multiple chat windows at

the same time. However, for the control group, where participants need to be recoupled, two pairs of participants took part in the experiment at the same time so each participant could switch partner for Part 2 during the experiment.

In order to give the right motivation to all participants, the payoff procedure is clarified in the instructions available to every participant. It is known that participants can earn points for themselves during the experiment, depends on the decisions that they will make. Points from the two parts are independent of each other and they are accumulated to a total number. It is highlighted that at the end of the experiment only one participant will be paid out due to a budget constraint. To avoid misunderstanding, every participant was informed of the payoff procedure twice - at the beginning as well as the end of the experiment. One ID number was selected randomly via the True Random Number Generator after all participants finished the experiment. This very participant had his earned 110 points converted to money (in euros) according to the exchange rate “100 points = 10 euros” and he was paid 11 euros by bank transfer.

2.3 2.3

2.32.3 HypothesesHypothesesHypothesesHypotheses

Based on the motivation and design of the experiment, four hypotheses of participants' behavior are raised as presented below.

Hypothesis Hypothesis HypothesisHypothesis 1111

In the dictator game, most of the gifts from the givers will be smaller than what the Send a message to confirm once finished.

Send a message of the expectation. Send a message of the choice. Informed of the choice. Informed of the expectation.

Step 4 Coin Flipping and calculation of the points by the experimenter. Informed of the results of coin-flipping and earned points for both.

Step 5 Fill in questionnaires.

(12)

recipients expect. The other gifts may be equal or bigger than the expectations. Studies have been conducted on the dictator game (e.g. Hoffman, McCabe and Smith,

1996; Bolton, Katok and Zwick, 1998; Cason and Mui, 1998; Bardsley, 2008). Rational individual theory predicts that the dictator would not share and keep all the amount. On the other hand, some laboratory experiments reported results which are different from the predictions of standard economic theories. For instance, Forsythe et al. (1994) conducted several experiments and the result of the dictator game with payment shows that not all dictators are unwilling to share the amount. Engel (2011) complemented a meta-study summarizing an amount of 129 experiments of the dictator game and concluded that in reality only 36% dictators follow the profit maximization hypothesis.

In this experiment, participants are aware that they are paired with a randomly determined partner. Assuming the selfishness of participants, the givers will not care about the other’s benefits and focus only on their own profits. Thus, the gifts will be given in favor of the giver’s interests. Moreover, it is known to all participants that there are no correlations between the payoff of everyone from each part of the experiment. Thus, a giver could give whatever gift without worrying about his total points being influenced. These thoughts promote the profit-maximization motivation. Naturally, I expect that most of the givers would give as few points as possible or even nothing to the recipients in order to get more points for themselves which could possibly converted to euros.

However, there is little research discussing the expectation of the recipients. Following the rational individual theory, a recipient does expect his partner to propose a proper gift. It is expected that participants have a preference to have more points which could be converted to money with a certain probability. And this preference may be strengthened by learning the unequal fact that the giver will be given 50 points unconditionally at this stage. In this case, recipients form their own expectations in favor of their own interests.

In summary, the first hypothesis about the experiment is that what a giver proposes and what a recipient expects will be different. Furthermore, most of the gifts are smaller than the expectations. And others may be equal or bigger than the expectations.

An unpleasant gift triggers negative emotions of the recipient.

The gift is seen as a treatment from the giver to the recipient. Different aspects of the gift will be focused on. In a dictator game, the fairness of a proposal (in this case the fairness of the gift) is often mentioned and used. Speaking of the concept and foundation of fairness, discussions are often seen in competitive markets. For instance, Rabin (1993) defined fairness according to the kind/hostile intention behind

(13)

the action, which depends on whether the payoff distribution is equitable. Fehr and Schmidt (1999) defined fairness as a self-centered aversion to inequity. Applying Fehr and Schmidt's model, for example, a 25-point gift seems fair and reasonable because a gift with half of the initial endowment (50 points) ensures that both participants in a pair get the equal amount of points.

However, instead of fairness, I think the expectation influences the formation of the emotion and furthermore the determination of the social relationships. The difference between a gift and its expectation can be used as a measure of the feature of the treatment in an interaction. To be specific, a gift which is smaller than the expectation represents an unpleasant treatment towards the recipient by the giver. As mentioned before, the emotion is measured according to the answers provided by the participants in the questionnaire. I think one’s emotion depends on how he sees and feels himself being treated by others. When a person receives an unpleasant treatment his feeling is negative.

To sum up, it is expected that an unpleasant gift is a trigger of negative emotions. That is, the deviation of the gift from the expectation reflects how the recipient feels.

Negative emotions generate a negative social relationship.

Emotions measure how people feel and feelings may influence the way people evaluate the relationship with the other. It is expected that personal emotions play a role in determining the social relationship. In this case, a recipient’s negative feelings about the treatment from his partner can lead to the development of a negative social relationship in his opinion. As mentioned before, social relationships are measured by asking the recipients about their opinions upon the type of the relationship between themselves and their partners. In the questionnaire, the recipients are provided with three options - positive, neutral and negative social relationship. In this case, social relationships are measured independently. A detailed measurement method will be given later (see p.17).

Negative social relationships promote a preference for the option with a larger outcome discrepancy in the lottery.

As discussed before, the effects of social relationships involve various aspects. The focus of this paper is its impact on people’s preference for outcome discrepancy, which usually involves a social situation. That is, in a social situation, the outcome discrepancy is often presented as a social gain or a social loss. Speaking of which aspect a decision maker would attach more weight to, the gain or the loss,

(14)

researchers have different opinions. According to Fehr and Schmidt (1999), in a social context, loss looms larger than gain, which is a straight extension of the perspective of individual decision makers (Kahneman and Tversky, 1991). However, Bault, Coricelli and Rustichini (2008) conducted experiments and found that in the social domain, gains loom larger compared with losses. I am convinced by the study of Bault et al (2008) thus I believe that social gain seems more attractive to individuals compared to social loss. That means, although the lottery in the experiment gives the same probability of ending up in a social gain situation and a social loss situation, a participant cares more about the possible social gain situation rather than the possible social loss situation. In that case, influenced by a negative social relationship, a participant desires to be better off than the other person. Furthermore, I believe the unpleasant treatment he just received triggers the motivation to take a revenge. As a result, in order to hurt the other's feelings, he chooses option E, for example, to enlarge the discrepancy in outcomes. The combination of a social gain situation and a large outcome discrepancy makes the participant believe that he can be much better off than his opponent and thus make his opponent feels negative.

On the other hand, recipients in the control group are aware that they have a new partner for Part 2. One can expect that the social relationship which developed in Part 1 would have no impact on their choices at all. Rational participants would make the choice according to their risk attitudes or their opinions towards inequity level. To be concrete, a loss-averse person would insist the belief of “loss looms larger than gain” where people assign larger value to possible loss and smaller value to possible gains. For instance, a risk-averse or inequity-averse person would like to choose option A while a risk-seeking person would probably prefer option E. Thus the fourth hypothesis is that social relationship affects the preference for outcome discrepancy and a negative social relationship promotes the preference for a larger outcome discrepancy.

3. 3.

3.3. ResultsResultsResultsResults

This section presents the results of the paper in the following order. Firstly, an overview of all decisions participants made in the experiment will be presented and analyzed. This is followed by a detailed statistical regression analysis.

3.1 3.1

3.13.1 DescriptiveDescriptiveDescriptiveDescriptive analysisanalysisanalysisanalysis

Table 4 shows participants’ decisions in the experiment. Each row of information consists of four elements: a gift by the giver (Gift G), a gift expectation by the recipient (Expectation R), a choice by the recipient (Choice R) and a choice expectation by the giver (Expectation G).

(15)

Table Table

TableTable 4444 Experimental results

Treatment group Control group

Part1 Part 2

(same partner) Part1

Part 2 (same partner) Gift G Expectati on R Choice R Expectati on G Gift G Expectati on R Choice R Expectati on G 25 30 B B 25 20 A E 0 25 E A 30 20 E A 10 15 C C 10 19 C A 25 1 B C 20 25 A B 25 1 B A 5 20 A C 25 12 A C 30 35 A A 25 29 D C 10 35 A A 0 10 E D 25 45 C C 10 20 C E 5 10 A A 0 10 E C 0 25 A C 25 30 B B 25 20 A E 0 25 E A 30 20 E A 10 15 C C 10 19 C A 25 1 B C 20 25 A B

In order to make sure that the only variable between the treatment and the control is the effect of social relationships, a test is done to compare the data in two groups to see if there is any significant difference. The means of two variables - Gift G and the difference between Gift G and Expectation R (referred as G-E) - in two groups have been compared under an independent sample t-test. Table 5-1 and table 5-2 present the results. With a detailed inspection, it could be found that for the means of Gift G in two groups, there is no significant difference (p=.77>.05). Similarly, the difference between the means of G-E in two groups is not significant either (p=.19>.05). Provided that there are limited observations in the experiment, it could be concluded the sole difference between two groups lies in that whether the social relationship has effects on the choice of the participants.

Table Table

TableTable 5555-1-1-1-1 Group statistics: Treatment and Control

Variable Group N Mean Std. Deviation Std. Error Mean

Gift G Treatment 10 14.50 11.66 3.69 Control 10 16.00 11.26 3.56 G-E* Treatment 10 -.80 16.01 5.11 Control 10 -9.40 11.89 3.76 Table Table

(16)

Variable t df Sig. (2-tailed) Mean Difference Std. Error Difference

Gift G -.29 18.00 .77** -1.50 5.12

G-E* 1.36 18.00 .19** 8.60 6.31

* G-E stands for the difference between Gift G and Expectation R. ** Significant at a 5%-level.

Table 4 discloses directly the gift and the expectation for each pair of participants. A further inspection shows that the gifts and the expectations vary from each other in different degrees. Also, it could be noticed that there is no expectations equal the real gift exactly. More detailed data about how much a Gift G deviates from the Expectation R is given below. In Table 6, 15 out of 20 recipients (75%) expected more than what they actually got from the givers, supporting Hypothesis 1 that most of the gifts are smaller than the expectations. Figure 1 provides a direct view of the difference between Gift G and Expectation R.

Table Table

TableTable 6666 Experimental results, realistic (expected) gifts and emotions*

Treatment group Control group

E G G-E Emotion E G G-E Emotion

25 30 -5 Felt nothing 25 20 5 Surprised

0 25 -25 Angry 30 20 10 Surprised

10 15 -5 Unhappy but not angry 10 19 -9 Unhappy but not angry

25 1 24 Surprised 20 25 -5 Felt nothing

25 1 24 Satisfied 5 20 -15 Unhappy but not angry

25 12 13 Surprised 30 35 -5 Felt nothing

10 20 -10 Unhappy but not angry 5 10 -5 Felt nothing

* E stands for Expectation R; G stands for Gift G; G-E stands for the difference between Gift G and Expectation R.

(17)

Figure Figure Figure

Figure 1111 Gift G vs Expectation R

Then, support for Hypothesis 2 requires the correlation between unpleasant gifts and negative emotions. In table 6, recipients’ self-reported emotions are listed next to the difference between Gift G and Expectation R. As discussed before, an unpleasant gift has a higher expectation thus the difference is negative (marked red). According to this criteria, 15 out of 20 gifts are defined as unpleasant treatments. However, the corresponding emotions are either felt nothing, unhappy but not angry or simply angry. In this case, negative and neutral emotional reactions coincide with unpleasant gifts. Moreover, recipients who received a higher-than-expectation gift report the positive emotion like surprised or satisfied. That is, the self-reported emotion reflects how a recipient feels about the gift. Therefore, Hypothesis 2 is partially supported because an unpleasant gift triggers negative or neutral feelings. In order to find evidence to support Hypotheses 3, a detailed regression analysis will follow in section 3.2. Now I will present a descriptive result of the experiment in order to support Hypothesis 4. In the lottery game, choices from the treatment group vary much more than those from the control group. Figure 2 below shows that option A is the most popular choice (70%) in the control group. On the other hand, in the treatment group, choices vary much more. This result provides a direct support to Hypothesis 4 that social relationships play a critical role in affecting the preference for outcome discrepancy.

(18)

Figure 2222 Choices: Treatment group vs Control group

3.2 3.2

3.23.2 RegressionRegressionRegressionRegression analysisanalysisanalysisanalysis

The following part will present a regression model and its results. The regression model consists of three components - a recipient’s emotional reaction for Part 1, the social relationship in his opinion and his choice in the lottery game - and the relationships among them. As Hypotheses 3 and 4 assumed in section 2.3, these three variables are closely linked to each other. The reason for running a regression analysis is that statistical evidence would provide more convincing support besides descriptive information. Hence, all the information collected in the experiment is translated into numbers for the convenience of statistical analysis. Section 3.2.1 introduces the definition of variables which are included in the regression model and section 3.2.2 gives the results of regression and analyzes.

3.2.1 3.2.1

3.2.13.2.1 VariablesVariablesVariablesVariables a.

a.

a.a. LotteryLotteryLotteryLottery choicechoicechoicechoice ((((ChoiceChoiceChoiceChoice))))

The first variable is the dependent variable. The focus of this paper is how negative social relationships affect a recipient's choice in the lottery game. The choice could be one out of five options. As shown below, it is represented in the form of numbers for simplification.

(19)

Table Table

TableTable 7777 Numbers representing the variable Choice

Choice A B C D E

Number 1 2 3 4 5

b. b.

b.b. Self-reportedSelf-reportedSelf-reportedSelf-reported emotionsemotionsemotionsemotions ((((EmotionEmotionEmotionEmotion))))

This variable is one of the independent variables. Emotions about the results of Part 1 is measured according to recipients’ answers in the questionnaire. Emotion is assumed to have direct effects on both social relationship and choice. Available emotion categories for recipients to choose from and the corresponding numbers are presented as below.

Table Table

TableTable 8888 Number representing the variable Emotion

Emotion Satisfied Surprised* Felt nothing Unhappy but not angry Angry

Number 1 2 3 4 5

* “Surprised” indicates the situation where the gift is higher than the expectation of the positively surprised recipient. “Surprised” is ranked behind “Satisfied” due to the reason that it has a corresponding smaller difference between Gift G and Expectation R. There were no reversed cases.

c. c.

c.c. SocialSocialSocialSocial relationshipsrelationshipsrelationshipsrelationships ((((RelationshipRelationshipRelationshipRelationship))))

In the multiple regression, the social relationship is the mediator. As discussed before, a social relationship is caused by different emotion categories. On the other hand, Hypothesis 4 mentioned that social relationship is the reason why participants choose a certain option, which means that the social relationship is supposed to have a direct effect on choice. In this experiment, the question that “how do you evaluate the relationship between you and your partner” in the questionnaire is used as a measure of social relationships.

Table Table

TableTable 9999 Number representing the variable Relationship

Relationship Positive Neutral Negative

Number 1 2 3

3.2.2 3.2.2

3.2.23.2.2 RegressionRegressionRegressionRegression resultsresultsresultsresults andandandand analysisanalysisanalysisanalysis

With a clear knowledge of all the variables involved, an overview of the regression model and its results will be presented in this section. First, the effects of certain

(20)

variables on another are tested with leaner regressions to find evidence for Hypotheses 3 and 4 respectively. The effects include the effect of Emotion on Relationship and the effect of Relationship on Choice. Then, as Hypotheses 3 and 4

suggested, a negative social relationship is supposed to play a role in the process of negative emotional reactions affecting behavior. In order to explore whether the influence of Emotion on Choice is mediated by the development of Relationship, a

mediation analysis is established according to the steps proposed by Baron and Kenny (1986) and will be explained in detail next. Three variables are presented in blue circles in Figure 3. The arrows indicate the relationship between each two variables. To be concrete, a, b and c represent the direct effect of one variable on the other. On the other hand b’ and c’ represent the indirect effect of one variable on the second variable controlling for the third variable. For instance, effect c’ indicates the effects ofEmotion on Choice controlling for Relationship.

Figure 3333 The mediation analysis

Table 10-1 and table 10-2 present the results of linear regressions of the treatment group and the control group respectively.

Table Table

TableTable 10101010-1-1-1-1 Regression results - Treatment group (N=10)

Set 1 - Dependent variable:Relationship

Variable Coefficient Standard Error t value p value

Emotion .31 .12 2.57 .03*

Constant .89 .42 2.15 .06*

Set 2 - Dependent variable:Choice

(21)

Emotion .95 .25 3.85 .01*

Constant .06 .87 .07 .94*

Relationship 1.79 .67 2.69 .03*

Constant -.21 1.31 -.16 .88*

Emotion .74 .34 2.20 .06** Relationship .70 .74 .95 .38* Constant -.56 1.09 -.52 .62* * Significant at a 5%-level. ** Significant at a 10%-level. Table Table

TableTable 10101010----2222 Regression results - Control group (N=10)

* Significant at a 5%-level. ** Significant at a 10%-level.

Two tables share the same format. Each table consists of several sets of data and each set of data consists of a headline which indicates the dependent variable and the corresponding data such as coefficient and p-value. To be concrete, table 10-1 has four sets of data while table 10-2 has two. Also, the first column lists the independent variables included in the regression and their coefficients are shown in the column next to it. The standard errors are listed in the third columns. The fourth and fifth column respectively show the t-values and p-values indicating whether the regression is significant.

Set 1 - Dependent variable:Relationship

Emotion .66 .17 3.90 .01*

Constant -.16 .57 -.29 .78*

Emotion -.39 .58 -.67 .52*

(22)

Comparing the data of set 1 in both groups, it could be seen that the estimated coefficients of variableEmotion are correspondingly .31 (p=.03) and .66 (p=.01) in the

treatment and the control group, which are both significantly different from zero at a 5%-level. This suggests that the feature of the emotion (positive, neutral and negative) reflects and affects the feature of the social relationship (positive, neutral and negative). For instance, a self-reported negative emotional reaction could lead to a negative social relationship in one's opinion. This result supports Hypothesis 3. As mentioned before, effect c shows thatChoice depends on Emotion and results of

the regression on effect c are demonstrated by data set 2 in both groups. A comparison of the p values for Emotion in table 10-1 and 10-2 shows that in the

treatment group, the relationship betweenChoice and Emotion is significant (p=.01)

at a 5%-level. However, Choice is no longer a reflection of Emotion in the control

group (p=.52).

Effect b represents the effect of Relationship upon Choice and this effect only exists

in the treatment group due to the reason that social relationships were developed in the control group but cannot have any effects. According to Table 10-1 data set 3, the coefficient forRelationship is 1.79, which is significantly (p=.03) different from zero at

a 5%-level. This means thatRelationship has a positive effect on Choice. For instance,

negative emotions such as angry (number 5) could lead to a choice like option E (number 5).

As mentioned before, the mediation effected will be tested following Baron and Kenny's proposal (1986). The first step is to show the correlation between Emotion

andChoice and this is done with the support of data set 2 in table 10-1. Then with

the help of data set 1 in the same table, it is proved thatEmotion is correlated with

the mediator - Relationship. Afterward, a multiple regression with Choice as the

dependent variable and Relationship and Emotion as independent variables is done

and shown in table 10-1 data set 4. The coefficients for each variable are controlled for the influence of the other variable. The results show that controllingRelationship

removes some effects of Emotion but the coefficient for Emotion is still significant

(p=.06) at a 10%-level. This suggests thatRelationship partially mediates the effect of Emotion on Choice. On the other hand, Relationship seems to lose its effect on Choice when the influence of Emotion is controlled (p=.38). In this case, Emotion

seems to have a stronger mediation effect compared to Relationship, which means

that the effect of Relationship on Choice is caused by Emotion. In this opinion

disproves the effect of social relationships discussed in Hypothesis 4. However, it seems unlikely due to a specific motivation analysis which is given below.

As mentioned before, a large-outcome-discrepancy option (i.e. option E) reflects the recipient's desire to revenge the giver who just gave him an unpleasant treatment, via the method of being better off. There are three possible reasons explaining the motivation behind the revenge of trying to hurt the other's feelings. Firstly, it is caused by the mood. If a recipient is unhappy or even angry at the situation that he gets an unpleasant gift which does not satisfy his expectation, he would like to take a

(23)

revenge by choosing an option that could provide a larger outcome discrepancy between himself and his partner. In this case, instead of social relationships, particular emotional reactions towards the specific partner influence the choices in both groups. Moreover, the results of the multiple regression seem to support this conclusion. However, it seems unlikely because one can observe clearly from figure 2 that there is a difference between choices in the two groups. Also, data set 2 in table 10-2 shows thatEmotion and Choice are no longer significantly related, which means

emotions do not influence choices in the control group. As a result, the mediation effect of Emotion found in the regression is probably due to limited observations of

data.

Secondly, the competitive context plays a role. If the recipient sees the lottery as a competition, which means he is eager to be better off than the other person in a social context, it could also happen that he likes taking risks and prefers the option with a larger outcome discrepancy. Again, if this is the case, recipients in two groups would have the same incentive and behavior. Thus, the choices in both groups would not differ much from each other. But Figure 2 disproves that. Last but not least, it is the social relationship that motivates the revenge. As tested before, the effect of social relationship is the sole significant difference between the treatment and the control (see table 5-1 and 5-2). A social relationship has an influence on people's behavior. When a participant believes that there is a negative social relationship between himself and his opponent based on the unpleasant treatment, he would like to take the revenge to hurt the other's feelings. Thus, when it comes to the possible difference between his own outcome and that of the other, he prefers a larger outcome discrepancy in order to be better off. This is motivated by the belief of "gain looms larger than loss in a social context". This recipient puts more weight for the social gain situation than the social loss situation. As a result, in the treatment group where the social relationship has an effect, the choice is motivated to move in the direction of a larger outcome discrepancy.

In summary, there is a clear relationship among components throughout the whole experiment which could be seen as below. In an interaction, a giver provides a gift to a recipient. If the recipient believes he gets an unpleasant treatment, which is in the form of a gift smaller than the expectation, he feels negatively such as unhappy or even angry. Moreover, he believes the social relationship between himself and the giver is negative. Influenced by the negative social relationship, the recipient prefers the option with a large outcome discrepancy in a lottery as a revenge.

4. 4.

4.4. ConclusionConclusionConclusionConclusion

The paper has two main focuses. The first is the relationship among three elements: treatment in the form of a gift, emotional reaction, and social relationship. The other focus is whether a negative social relationship promotes the preference of a large-outcome-discrepancy option in a lottery game. As mentioned before, current

(24)

literature studies not only the establishment and development of social ties but also its behavioral consequences on various aspects. However, there is little research on how social relationship affects attitudes and behavior towards possible outcome discrepancies with others. The paper, therefore, designed an experiment in which first it was detected how a social relationship develops after an interaction and then explored the behavioral influence of social relationships upon outcome difference. The results show that an unpleasant treatment (in the form of a gift which is smaller than the expectation in the experiment) triggers a non-positive emotion. Furthermore, a negative social relationship develops which is correlated with the emotion. Moreover, in the lottery, participants who have a negative social relationship are more likely to choose an option with a larger outcome discrepancy compared to those who have a positive social relationship or those who do not have any social relationships. Neither emotion nor competitive could lead to the difference in choices between two groups. The results of the regression give an affirmative answer to the research question that does social relationship affects behavior upon outcome discrepancy. To be specific, the development of the negative social relationship leads to the preference of a larger outcome discrepancy.

Strictly speaking, there are limitations and biases in this paper in two main aspects. First of all, with respect to the measurements of variables, improvement could be made to the design of the experiment. The interaction in Part 1 is designed as a dictator game with a gift-giving setting. Although it makes sense that a person compares with a specific object, it is more common that economic interaction happens among a larger group of people. Therefore social relationships would become more complicated and realistic after several repetitions of interactions. Also, personal feeling is a complicated variable which has various aspects and it is difficult to describe and measure. In the experiment, the emotional reaction was measured by the report from participants. Also, five categorizations of emotions for recipients to choose from are ambiguous to some degree. Although these emotion categories belong to either positive, neutral and negative range, the measurement can still be improved.

Secondly, speaking of the participant pool, the sample size is not big enough. The insufficient observations do not provide enough credibility to the regression analysis. As mentioned before, approximately 83% of the participants studied Economics and Business at universities. This could lead to a bias in the database because these students basically have learned certain economic knowledge in different degree and, therefore, they may have a clue about the purpose of the experiment and behave unnaturally without even realizing it. However, efforts have been taken on the experimental setting in order to avoid strategic decisions as much as possible. For example, it is stressed to the participants about the independence of points earned from other participants’ decisions as well as from each part of the experiment.

Nevertheless, the findings of the paper could inspire studies on the effect of social relationship upon attitudes and behavior towards outcome discrepancy in at least

(25)

two aspects. First, in this paper, only the effect of negative social relationships is observed and studied. It might be interesting to study the effect of positive social relationships upon the preference of outcome discrepancy and distinguish it from the effect of other factors such as risk attitudes. Second, it is observed that negative social relationships correlate the choice with large outcome discrepancy. But how much does the social relationship affect the preference is still unknown.

(26)

Reference Reference ReferenceReference

Ariely, D., Huber, J., & Wertenbroch, K. (2005). When do losses loom larger than gains?. Journal of Marketing Research, 42(2), 134-138.

Attanasi, G., Hopfensitz, A., Lorini, E., & Moisan, F. (2014). The effects of social ties on coordination: conceptual foundations for an empirical analysis.Phenomenology and the cognitive sciences, 13(1), 47-73.

Bardsley, N. (2008). Dictator game giving: altruism or artefact?.Experimental Economics, 11(2), 122-133.

Bault, N., Coricelli, G., & Rustichini, A. (2008). Interdependent utilities: how social ranking affects choice behavior. PloS one, 3(10), e3477.

Baumeister, R. F., & Leary, M. R. (1995). The need to belong: desire for interpersonal attachments as a fundamental human motivation.Psychological bulletin, 117(3), 497. Bandura, A., & Jourden, F. J. (1991). Self-regulatory mechanisms governing the impact of social comparison on complex decision making. Journal of Personality and Social Psychology, 60(6), 941.

Baron, Reuben M., and David A. Kenny. "The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations." Journal of personality and social psychology 51.6 (1986): 1173. Becker, G. (1974). A theory of social interactions.

Bolton, G. E., & Zwick, R. (1995). Anonymity versus punishment in ultimatum bargaining. Games and Economic behavior, 10(1), 95-121.

Bolton, Gary E., Elena Katok, and Rami Zwick. "Dictator game giving: Rules of fairness versus acts of kindness." International journal of game theory, 27.2 (1998): 269-299. Fehr, E., & Schmidt, K. M. (1999). A theory of fairness, competition, and cooperation. Quarterly journal of Economics, 817-868.

Feld, S. L. (1981). The focused organization of social ties. American journal of sociology, 1015-1035.

Forsythe, R., Horowitz, J. L., Savin, N. E., & Sefton, M. (1994). Fairness in simple bargaining experiments. Games and Economic behavior, 6(3), 347-369.

Granovetter, M. (1985). Economic action and social structure: the problem of embeddedness. American journal of sociology, 481-510.

Hoffman, E., McCabe, K., Shachat, K., & Smith, V. (1994). Preferences, property rights, and anonymity in bargaining games. Games and Economic Behavior, 7(3), 346-380. Isen, A. M. (2001). An influence of positive affect on decision making in complex

(27)

situations: Theoretical issues with practical implications. Journal of consumer psychology, 11(2), 75-85.

Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica: Journal of the Econometric Society, 263-291.

Linde, J. S. J.(2012). Social comparison and risky choices. Risk Uncertain,44(1), 45-72. Malmendier, U., & Schmidt, K. (2012). You owe me (No. w18543). National Bureau of Economic Research.

Medvec, V. H., Madey, S. F., & Gilovich, T. (1995). When less is more: counterfactual thinking and satisfaction among Olympic medalists. Journal of personality and social psychology, 69(4), 603.

Murphy, R. O., Ackermann, K. A., & Handgraaf, M. (2011). Measuring social value orientation. Judgment and Decision Making, 6(8), 771-781.

Sonnemans, J., Van Dijk, F., & Van Winden, F. (2006). On the dynamics of social ties structures in groups. Journal of Economic Psychology, 27(2), 187-204.

Rabin, M. (1993). Incorporating fairness into game theory and economics.The American economic review, 1281-1302.

Reuben, E., & Van Winden, F. (2010). Fairness perceptions and prosocial emotions in the power to take. Journal of Economic Psychology, 31(6), 908-922.

Van Dijk, F., Sonnemans, J., & van Winden, F. (2002). Social ties in a public good experiment. Journal of Public Economics, 85(2), 275-299.

Van Dijk, F., & Van Winden, F. (1997). Dynamics of social ties and local public good provision. Journal of Public Economics, 64(3), 323-341.

(28)

Appendix Appendix AppendixAppendix Appendix Appendix

AppendixAppendix A.A.A.A. Experiment instructions

The experiment instruction consists of three parts: General Instruction, Instructions Part I and Instruction Part II. Participants receive different versions of instructions depending on their roles and the groups in the experiment correspondingly. For instance, the instructions for the givers in the treatment group are presented as below. All versions of instructions are available upon request.

1. The General Instructions

Welcome to the experiment. This decision-making experiment consists of two parts. You will get instructions at the beginning of each part. Read and follow carefully the instructions which contain everything you need to know. The experiment is anonymous. At the beginning, you will be given a unique ID number and you will be recognized only according to the ID number during the experiment, which means anonymity is guaranteed. You must refrain communication with other participants during the experiment and make sure that your cellphone is switched off.

According to your performance and the decision you've made, you have a chance to win some cash. In this experiment, you may earn points for yourself, depending on the decisions that you and other participants will make. Points you earned from each part are independent from each other. At the end of the experiment, one ID number

will be selected randomly via an online number generator tool

(https://www.random.org/) and the corresponding participate will have his/her earned points converted to euros according to the following exchange rate:

100 points = 10 euros and he/she will be paid by cash.

The experiment consists of two parts, and in each part you may need to make a decision or not depending on your role in the experiment. The points you get from two parts will be accumulated to a final one, which counts towards your payoff if your ID number is selected to be paid. The main part of the experiment is followed by a short questionnaire you need to fill in. The payment procedure will take place after the entire experiment finished. One ID number will be selected via an online number generator (https://www.random.org/) then the result will be announced individually.

If you have any questions during the experiment at any point, feel free to ask the experimenter and you will get the answer in private.

2. Instruction Part 1

We will now outline the instructions for the first part of the experiment. In this part of the experiment, you will be matched randomly with another participant who will

(29)

be referred to you as "your partner". You will be defined as "the Giver" or "the Recipient" in this part. The Giver will be given 50 points and then need to make a decision to give a gift to his/her partner (the Recipient) which values from 0 to 50 points. The Recipient will receive nothing at the beginning but will be informed of the gift from the Giver and can only accept the gift.

You are "the Giver".

Your gift: __________

When everyone has finished making the decisions, you will be informed of the result of this part and then we will advance to the second part of the experiment.

3. Instruction Part 2

We will now outline the instructions for the second part of the experiment. In this part of the experiment, you will be matched with the same person you met in Part 1 as your partner.

In this part of the experiment, there is a 50-50 lottery determining you and your partner's points. Your lottery is given as: you can either get 60 points or 40 points, with the chance of 50%, which is independent of your partner's choice. Your partner needs to define his/her own lottery by making a choice among five options: A, B, C, D and E. After your partner made a choice, a coin will be flipped to determine which situation of the lottery will be paid. The head means "+" situation is chosen while the tail means "-" situation is chosen.

Situation Probability

The Giver’s

points

The Recipient’s points

A B C D E

+ 50% 60 70 75 80 85 90

- 50% 40 30 25 20 15 10

The way to read the table above is as follows: If your partner chooses "A"

� under "+" situation, your partner earns 70 points while you earn 60. � under "-" situation, your partner earns 30 points while you earn 40. If your partner chooses "B"

� under "+" situation, your partner earns 75 points while you earn 60. � under "-" situation, your partner earns 25 points while you earn 40. If your partner chooses "C"

(30)

� under "+" situation, your partner earns 80 points while you earn 60. � under "-" situation, your partner earns 20 points while you earn 40. If your partner chooses "D"

� under "+" situation, your partner earns 85 points while you earn 60. � under "-" situation, your partner earns 15 points while you earn 40. If your partner chooses "E"

� under "+" situation, your partner earns 90 points while you earn 60. � under "-" situation, your partner earns 10 points while you earn 40.

As your partner need to make the choice, you need to write down your expectation of his/her choice.

Your expectation: ______

When everyone has finished making the decisions, a coin will be flipped and you will be informed of the result of this part. Then we will advance to a short questionnaire.

AppendixAppendix B.B.B.B. Questionnaire

Questionnaire - For the Giver

1. Your gender: □ Male □ Female 2. Your age: ______

3. Your major: □ Business and Economics □ Others

4. What is your strategy to provide the gift to your partner in Part 1? □ Maximized my points;

□ Cared about joint benefits; □ Tried to be fair;

□ Otherwise, specify ________________

5. Do you think your gift is fair to your partner? □ Yes □ No

6. How do you evaluate the tie (the relationship) between you and your partner after the first part? □ Positive □ Negative □ Neutral

7. In Part 2, do you think your partner would have made a different decision if you had given him/her a different gift? □ Yes (Briefly explain ________________ ) □ No

(31)

□Not applicable, I have a different partner

8. How do you evaluate yourself in risk-taking: are you generally a person who is fully prepared to take risks or do you try to avoid taking risks? Please cross a circle on the slide below, where the value 0 means: "not at all willing to take risks" and the value 10 means "very willing to take risks".

0 10

9. How do you evaluate yourself in inequity-accepting: are you generally a person who is fully agreed with inequity or do you try to avoid inequity? Please cross a circle on the slide below, where the value 0 means: "fully agree with inequity" and the value 10 means "fully disagree with inequity".

0 10

Questionnaire - For the Recipient

1. Your gender: □ Male □ Female 2. Your age: ______

3. Your major: □ Business and Economics □ Others 4. Please answer the following questions in terms of Part 1. (1) What amount of gift (in points) did you expect? ______

(2) How many points did you actually receive from the giver? ______ (3) Do you think the gift is fair? □ Yes □ No

(4) How do you feel about the results?

□ Surprised □ Satisfied □ Felt nothing □ Unhappy but not angry □ Angry □ Otherwise, specify ________

5. How do you evaluate the social relationship between you and your partner after the first part?

□ Positive □ Negative □ Neutral

6. Please answer the following questions in terms of Part 2. (1) What is your choice? □ A □ B □ C

(32)

(2) What is your strategy for making this decision? __________________

(3) Do you think you would have made another choice if you would have got a different gift from the giver in Part 1? □ Yes (Briefly explain ____________ ) □ No 7. How do you evaluate yourself in risk-taking: are you generally a person who is fully prepared to take risks or do you try to avoid taking risks? Please cross a circle on the slide below, where the value 0 means: "not at all willing to take risks" and the value 10 means "very willing to take risks".

0 10

8. How do you evaluate yourself in inequity-accepting: are you generally a person who is fully agreed with inequity or do you try to avoid inequity? Please cross a circle on the slide below, where the value 0 means: "fully agree with inequity" and the value 10 means "fully disagree with inequity".

0 10

AppendixAppendix C.C.C.C. Communication model - messages for participants

For the Givers

Dear participant, welcome to this experiment. Your ID number is ______, which is generated randomly on this websitehttps://www.random.org/. In this experiment, you will be only recognized according to your ID number. Please follow the steps and send back messages containing required information.

Step 1.

a. Read "General Instruction", send a message to confirm after you finished reading Step 2.

a. Read "Instruction Part 1", send a message to confirm after you finished reading. b. Send a message of your gift to your partner.

c. Your gift is ______. Your partner expects your gift would be ______. Step 3.

a. Read "Instruction Part 2", send a message to confirm after you finished reading. b. Send a message of your expectation of the choice.