Effects of Manipulating Individual Identity

Published Date: 21 Feb 2018

Abstract:

This study examined how changing the perception of social distance changed the way in which subjects interacted in the Trust Game specifically looking into the social preferences they displayed. A discussion of both economic and sociological research demonstrates the inherent variability of social identity and social distance and the interconnected nature these concepts have with social preferences. A two-stage experiment involving subjects of different nationalities and genders was carried out with a different identity focus in each stage. Findings show that subjects exhibit greater social preferences when interacting with those of a similar focused identity yet the degrees of these preferences displayed vary in accordance to the identity focus. Results suggest that social distance, although easily manipulated, is a powerful force in interactions. The results are consistent with previous studies into group membership, identity, nationality, gender and social preferences. Given this study's far-reaching implications it should be viewed as the premise for future study of this topic.

Introduction

The concept of 'social distance' as elucidated by Akerlof (1997) is profoundly linked with social identity and social preferences. It is the differences in individuals' social identities, including for example, race, gender, class and status, that determine the benefits of interaction and to what extent of social preferences are displayed. Yet when individuals are not entirely familiar with each other, the perception of social identities is based on very few observations and is therefore not entirely accurate. The purpose of this dissertation is to examine whether or not variations in the perception of social identity, and thus social distance, will affect the extent to which social preferences are manifested.

Studies on the effect of identity are in no way new. Akerlof and Kranton (2000) studied its relevance and magnitude with regards to gender discrimination in the workplace, poverty and social exclusion, and the household division of labour. Akerlof (1997) also studied the effect of social distance, asserting that the social identity of individuals can substantially affect their interactions. What has not been tested, however, is to what extent identity is merely a perception. Individuals can and do possess multiple identities varying in dominance from situation to situation. The hypothesis of this paper is that social preferences are relevant in experimental games, but these are dependent on how each player perceives the other's social identity, a perception that is inherently variable. Players react with varying degrees of self-interest according to the identity they perceive, even when identical character profiles are involved. Recent economic studies fail to sufficiently incorporate current sociological thinking as to the nature of social identity and although this study will give evidence for the relevance of social distance in determining degrees of social preferences, it will attempt to shed light on individuals' inconsistent perception of the distance between them.

In order to add weight to this position, an experiment was organised in which participants of various profiles were invited to play the Trust Game. Each game was played face-to-face with no verbal communication. The experiment was divided into two stages, a nationality focused stage and a gender focused stage. Each participant played the game twice and never with the same opposing player. Before each game, participants were assigned to certain rooms in which video clips were played depending on the identity focused on. In the first stage, participants were split into three rooms, one for each nationality present: Scottish, English and Chinese. After one play of the Trust Game, the experiment commenced its second stage and participants were split into two rooms, one for each gender, and again asked to play the game.

The results show that there is significant variation in the degrees of social preferences displayed depending on the identity being focused on hereafter referred to as the 'focused identity'. When nationality was the focused identity, those with this common identity exhibited greater social preferences than those of differing nationalities, regardless of gender. When gender was the focused identity, those with this common identity exhibited greater social preferences than those of differing gender, regardless, to some extent, of nationality. This lack of consistency shows that the perception of social distance is not a constant through both stages and yet it is still proves to be a determining factor in how players interact.

The following content of this paper is arranged into four sections. Section 2 will address the relevant literature with regards to social preferences and identity, showing their relevance, interconnected nature and inherent variability, discussing the application of various theories in the context of this study. Section 3 will describe the design and implementation of the experiment used in this study with the results, analysis and comparisons to similar studies presented in Section 4. An interpretation of these results and further conclusions will be offered in Section 5.

Theories of Preferences and Identity

To clearly understand the argument presented, a holistic discussion of current theories must be embarked upon. The relevance of social preferences and the forms of their analysis through experimental games will be examined in this section to create the context in which this study is placed. The concepts of identity and how they shape perceptions and interactions will then be considered before discussing their relationship with social preferences.

Social Preferences

The assumption that man is motivated by self-interest is one that has dominated economic theory and is indeed fundamental to the very ideology of the vast majority of economists. As Adam Smith (1910:13) suggests, "It is not from the benevolence of the butcher, the brewer, or the baker, that we expect our dinner, but from their regard to their own interest." Inherent in this analogy is the implication that self-interest is the primary motivator in our economic interactions, and it is through this impulse that equilibria are formed. In Game Theory, the Nash Equilibrium is obtained by each player assuming the other possesses purely selfish motives. This assumption, however, is not always consistent with reality. Many factors influence an individual's decisions and it is evident that the maximisation of material gain is not always a dominating motivation. Some of the factors influencing an individual's decisions can relate to social status, group membership, inequity aversion, reciprocity, spite, envy, altruism and identity. These factors form a preference set, referred to in this study as 'social preferences'.

Although the theory of self-interest has been ingrained in traditional economics, modern economists have increasingly questioned this supposition. In fact the traditional authors of this theory were also wary of its realism. Even though Edgeworth (1881:16) writes that "the first principle of Economics is that every agent is actuated only by self-interest", he concedes that Man is, in reality, "for the most part an impure egoist, a mixed utilitarian," (Edgeworth 1881:104). It is evident that we are influenced by unselfish motives in our interactions with others and it follows that these interactions will in many circumstances involve material profit and loss. This is the realm of social preferences. Amartya Sen (1977:336), in his seminal paper on the irrationality of traditional preference theory states that "the purely economic man is indeed close to being a social moron." He argues that theories that do not encompass the influence of sympathy and our commitment to it are incomplete. It is the result of this basic emotion that our social norms, laws and codes of conduct are formed and without which no society would be feasible (Johansen 1977).

In various studies into social preferences, it is seen that they are comprised of multiple aspects. Reciprocity, inequity aversion, altruism and trust are all basic components of this preference set (Andreoni et al. 2002). When kindness or animosity is shown, the impulse exists to reciprocate in like manner. When inequity exists, the urge exists to rebel against it. Altruism is the virtue that depends on no form of expected gain or reciprocity, the purest form of good will. Trust is the confidence in the future actions of another and is arguably the most variable of the social preferences and one that will be revisited in this study.

Each of these attributes are far from abstract philosophical concepts, exert great influence on our economic interactions. Agell and Lundberg (1995) discovered that as a result of workers being influenced by fairness and equity customs, wage cuts were often unprofitable, with workers objecting to an 'unfair' action. This can have a direct impact on the extent that a company vertically integrates, with out-sourcing often being far more viable given the reciprocal attitude of workers. Bewley (1999) also noted that a firm's policy can affect worker morale and as such companies must take into account the perceived fairness of their policies. The extent of tax evasion has also been correlated to how fair it is perceived, and in fact the entire structure of tax systems are thoroughly debated and altered according to the principles of equity and merit (Seidl and Traub 2002). The perception of unreciprocated generosity has been one of the causes of the general reduction in support from the US welfare state. People are disinclined to support welfare structures that give the impression that they are helping a poor segment of society that refuse to help themselves, content to live off the goodwill of others (Bowles and Gintis 2000). Fukuyama (1995) also correlates economic prosperity with higher levels of trust, suggesting that social preferences are a very powerful force even in the current global economic system.

Sequential games

Although the examples given are wide-ranging in scope, involving relatively large economic issues, social preferences have also been proven to have a significant effect on individual interactions. Previously experimental games have been used to prove the dominating forces of self-interest, the equilibria of these experiments being calculated by assuming that all agents were exclusively self-interested (Fehr and Schmict, 2001). Recently, economists have carried out several experiments involving non-cooperative games that contradict this presumption. Guth, Schmittberger and Schwarze (1982) were among the first to create a game that did just this. It is called the Ultimatum Game.

There are two players in the game, one a Proposer and one a Responder. The Proposer must divide an amount of money X between the two players, offering the Responder any amount Y = X. If Responder accepts, the Proposer receives the remaining money X – Y, if they reject, both receive nothing. Under self-interested preferences, the efficient equilibrium is the one in which the Proposer gives the least amount possible to the Responder, who will accept any amount. In reality however, offers of less than a fifth of X are rejected about half time and Proposers anticipating this generally offer around 30 to 50 percent of X (Hoffman et al. 1996). This result clearly shows that factors other than self-interest are at play. It would be reasonable to assume that altruism and reciprocity both play a part in the decision-making process of both players. The Proposer may be influenced by a code of morals and a concept of fairness to offer more than the standard equilibrium distribution. The Proposer must also take into account the Responder's sense of reciprocity and animosity towards a seemingly unfair distribution. Although the Responder is under no monetary incentive to reject a low offer, yet his social preferences mean that he is able to achieve some utility by spiting the Proposer, thereby valuing a certain amount of reciprocity over monetary value.

Given the fact that the Proposer's actions may be driven only by the fear of reciprocity and no sense of altruism, it is worthwhile to look into the Dictator's Game first introduced by Kahneman et al. (1986) and refined by Forsythe et al. (1994). In this game, the Responder, now called the Recipient, is not given the option to accept or refuse the amount given by the Proposer. If the Proposer is motivated by self-interested alone, they will offer nothing to the Recipient but as many experiments have shown, this is not always the case. Henrich et al. (2001) find that in most dictator game experiments there is a primary mode offer of zero percent of the Proposer's total wealth and a secondary mode offer of 50 percent. Some groups show a primary mode offer of 20 percent and a secondary mode of 50 percent providing strong evidence of inequity aversion. In addition to supporting the notion that man is not exclusively self-interested, studies also confirm that fear of reciprocity is present in the Ultimatum Game and that Proposers apply backwards induction with average offers being lower in the Dictator Game (Roth et al. 1991).

The Trust Game, developed by Berg et al. (1995) is a game that can be used to test the presence of altruism, inequity aversion, reciprocity and its namesake, trust. The game is played with an Investor and a Trustee, with the former being given an initial endowment of X and the latter given nothing. The Investor is then able to give any amount Y between 0 and X. The amount the Trustee receives will be tripled, amounting to 3Y. The Trustee is then given the option to give any amount Z between 0 and 3Y back to the Investor thereby making the payoffs of the Investor and the Trustee X – Y + Z and 3Y – Z respectively. The Trustee is under no monetary incentive to return any amount and as such, under strictly self-interested preferences the Investor will predict this and give the Trustee nothing but, as with the Ultimatum and Dictator Games, studies show that many players of the Trust Game deviate from this equilibrium. Berg et al. (1995) find that almost all Investors give some amount of money to the Trustee and that a substantial number of Trustees return at least the same amount and that a third even returned more than they received. The amount returned also increases with the amount given thus supporting the theory that reciprocity is an integral part of many preference sets. Investors and Trustees are able to display inequity aversion by choosing to give or return amounts that will equalise final payoffs. Trustees can also display altruism by returning anything over and above the amount needed to equalise payoffs. It is interesting to note that there is substantial variation in the amounts given, with no clear average amount entrusted. The variation is not unsurprising, however, given the inherent inconsistency in levels of trust that individuals demonstrate in their interactions with various individuals. In society, trust placed in an individual is dependent on who that individual is or, in other terms, trust placed is dependent on the perceived identity of the individual in question.

Identity

Identity, at its most fundamental level, is at the base of all human interaction. For an individual to interact with another, the individual must have a clear concept of both himself and of the other. It is in the consideration of these two concepts that decisions are made. Descartes (1912:167) famously stated "…I think, therefore I am," and in doing so sparked off the philosophical debate on what truly directs our thoughts and actions. Hume (1888) further develops this by exploring our perception of ourselves, our identity. It was his belief that we can only perceive ourselves, and build our identity, by categorisation in the light of selected characteristics and never perceive our true reality in objective terms. It is out-with the bounds of this study to discuss in depth the sociological and psychological complexities of this topic, yet it is worth-while bringing to light some key concepts to further the understanding of the interactions between this study's participants.

An identity is a tool of recognition. It "allows us to recognise individuals, categories, groups and types of individuals," Wiley (1994:130). More than this, it is also a tool of categorisation and emotional cues. It implies a "conscious awareness by members of a group, some positive or negative emotional feelings towards the characteristics which members of a group perceive themselves as sharing and in which they perceive themselves as differing from others," Mennell (1994:177). Goffman (1968) further expounds these aspects of recognition by dividing identity into three sections: the 'personal identity', the 'ego identity' and the 'social identity.' The personal identity is the unique identification that each individual possesses to differentiate themselves technically, legally and realistically from all others. The ego identity is a purely subjective observation that is built from a multitude of social experiences and is a sense of one's own particular state and nature. The social identity provides a way of categorising people and connects each person with a set of attributes and characteristics thought to be in keeping with the members of their respective categories. Individuals that possess commonalities in the form of thought, action, nature, experience or lifestyle can all be grouped into various social identities. Examples of social identities are nationality, gender, music-taste, age, profession and political views. It is important to stress that while individuals may only hold one personal and ego identity, they are able to juggle multiple social identities which have varying degrees of focus from situation to situation.

Akerlof and Kranton (2000) put forward the theorem that our perception of our ego identity can have a significant influence on our decisions and actions. Their theorem sheds light on a number of seemingly irrational choices. Actions that are of apparent detriment to an individual can be viewed as a form of behaviour that it used to create a more unique self identity. Similarly, steps may be taken to symbolise the assumption of a particular identity or the membership of a certain group, be they conscious or otherwise. Men do not generally wear dresses, and as such this behavioural code is unconsciously subscribed to by the majority of men. Any behaviour to the contrary poses a challenge not merely to the social norm, but to the identity of manhood itself. Attempts to manipulate an individual's decisions can be based on the notion of identity. In order to influence people to buy their products, companies create advertisements that often show a stylised form of a particular identity that people may aspire to. Finally, as identity can play such a large role in determining our economic decisions and behaviour, and assuming that individuals can choose their own identity, Akerlof and Kranton (2000) conclude that identity choices can be a major factor in a person's overall economic well-being, a conclusion strengthened by the theory of role-identities.

It is difficult to determine to what extent our identity is prescriptive or descriptive in relation to our own actions, but nevertheless the dual concepts of 'identity' and 'role' are deeply interconnected. Lyman and Scott (1970:136) clarify this link by stating that "roles are identities mobilised in a specific situation; whereas role is always situationally specific, identities are trans-situational." By assuming an identity, we also assume a role. Despite that the fact that this role varies from situation to situation, it is at all times consistent with the assumed identity. It is from this notion that expectations and metaperspectives are formed. Laing et al. (1966) pioneered the belief that it is not 'I' but 'you' that is important. More specifically they claimed that we are all deeply influenced by considering our view of others' view of ourselves and in such a way develop a 'role-performance' that conforms to the expectation others have of our behaviour so as to receive positive feedback or avoid negative feedback. In order to assess these expectations and act accordingly, we must first judge what these expectations are. With strangers, this is problematic, and as such we "orient ourselves toward them in terms only of the ill-specified contours of their social roles," (McCall and Simmons, 1978:70). In this respect, we are only able to form rough estimations of a person's true identity and thereby rely on our perception of how they fit into vague boundaries of social identities. When we perceive people this way, our perception of the attributes they possess as ascribed by their social identity is often completely arbitrary when viewed in the context of an objective character analysis. The perception and reality can at times be poles apart, decreasing in accuracy with increasing 'social distance'.

Identities and Social Preferences

Akerlof (1997) defined social distance as a measure of social proximity between individuals. The model he created summarises that people gain benefits from interacting to those socially closer to themselves, with these benefits decreasing with isolation. This model is supported by empirical evidence that suggests that trust and reciprocity are linked with social connection and that members of the same nationality and race exhibit greater degrees of these attributes towards one another (Glaeser et al. 1999). A common method of analysing social distance is through the construction of groups in experiments and many studies of this kind have provided quite significant results.

Studying the effects of group membership on cooperation, Orbell et al. (1988) find that subjects are far more likely to cooperate with in-group members than out-group members, with 79% of participants showing cooperation with the former and only 30% showing cooperation with the latter. Through using a variation on the dictator game, Frey and Bohnet (1997) also showed how group membership affects social preferences. The experiment observed that in-group members were allocated far more of the total endowment than out-group members suggesting some correlation with membership and altruism and inequity aversion. An important finding of the literature on the topic of group membership is that subjects react to membership in a very subjective manner, disregarding objective considerations. Billing and Tajfel (1973) observe that the even most minimal connections within a group still give rise to in-group positive discrimination. Although subjects realised that the basis of group composition was entirely random, they still discriminated toward their fellow members in a very significant way. The fact that the weakest bonds are able to create positive in-group interaction is an important consideration when examining the relation between perceptions of social identity and expressions of social preferences.

While group membership is a powerful force, transnational studies have shown that the cooperation inducing group mentality is not a universally consistent attribute. Buchan, Croson and Johnson (1999) find that subjects from the U.S. are more trusting when paired with in-group members but that this is not the case for subjects from China and Japan, who are more trusting in general, regardless of whom they were paired with. Buchan and Croson (1999) also find variations across genders observing that although participants trust men and women equally, women are seen to reciprocate more than men in Trust Games and are more generous in Dictator Games, findings that are consistent internationally. Another consistency that was found across nationalities in this study was the effect that communication between players had on trust and reciprocation levels, a conclusion also mirrored in other experiments. Roth (1995) found that even simple, seemingly irrelevant conversations significantly increased the levels of these social preferences.

Regardless of variations across nationality, gender and communication levels, it is apparent that there is a clear connection between identity and social preferences. As we categorise individuals into social categories, we not only presume they possess certain qualities and attributes but we also predict how they react. In the same way we use metaperspectives to shape our own actions based on vague notions of the social identities of others, we also use these imperfect images to form inherently imperfect expectations of future interactions. The perception and reality can at times be irreconcilable and yet any initial interaction uses this as its basis. McCall and Simmons (1978) put forward the idea that any interaction that takes place is solely based on images that are constructed in the minds of those interacting. Taking into account the inaccuracy of these constructs when strangers interact, we can see how this translates into the layman's term of 'prejudice', a concept closely linked with trust.

The concept of trust, as mentioned earlier is based on confidence and at the heart of confidence is a deep reliance on predictions and expectations which are in turn based on the rough identities that we perceive others to possess. This results in great variance in trust levels which, although proven in studies referred to above, is readily seen in everyday life. Trust can be unquestioned with interactions with family members and friends but displayed with lesser and lesser extents to strangers and those who we perceive as untrustworthy. Just as signalling is used in the employment markets, so it is in our trust-dependant interactions. One may ask a well-dressed, polite and friendly stranger to watch over some personal belongings in a library but may be loath to leave anything unattended when in the presence of hooded youth. The hood can be seen as a signal that the wearer is dangerous and cannot be trusted. It is perceived as the expression of an identity, the perception of which can influence our attitudes and behaviour.

The studies above also show that identity can greatly affect reciprocity, inequity aversion and altruism. Experiments based around group membership, however minimal, show the great influence groups have on these social preferences. One explanation of this is the concept of metaperspectives, in that individuals are more generous in experimental games because they believe that their counterpart expects them to be. Akerlof's (1997:1008) model of social distance also sheds some light on this by theorising that individuals benefit from lesser amounts of social distance between them and thus have the incentive to conform to expectations, what he labels The Conformist Model. A reduction of social distance between players can also be achieved by perceived acts of kindness and so experimental game players may be willing to sacrifice monetary gains so as to achieve social gains with another player. This incentive however, is again based on social distance and those players who feel socially far apart may feel no need to become socially closer, a feeling that is ultimately merely based on their perception of the current social distance and social identities.

Two significant ways in which individuals identify themselves and others is by their nationality and gender. At the outset of mankind's evolution, gender has been a universal divider of the human race, preceding all other identities. Rooted in our biology, gender is the simplest form of classification, but its implications are far more wide-reaching than simple physiology. To the opposite sex, gender implies certain generalised roles, attitudes, commitments, experiences and lifestyles. The source of such clear social stereotypes is only in part biological and many academics are of the belief that behavioural and psychological differences are created and perpetuated by unbalanced power and privilege structures in society (Flax 1990). The amplification of social distance is caused by the notion that qualities are gender specific, with masculinity and femininity being attributes in themselves, and the fact that men and women are commonly associated with their relative positions in both family life and work life. Lockheed (1985) supposes that women are conceived as compliant followers and men dominant leaders only because of the common minority and majority balance that is common in social and work situations. The large disparity between the social identity and actual realities of members of the opposite sex provides a good opportunity to explore to what extent interaction is based on unqualified perceptions and to map the effect of variations in this perception.

Unlike gender identities, nationalism is a relatively new force in the world (Smith 1995). It can be seen as a group identity that has transcended some cultures, as seen in the ethnically diverse nations such as India and Russia, but divided others as seen in the cases of North and South Korea and the Taiwanese and Chinese separation and is manifested in positive discrimination towards fellow nationals and negative discrimination towards foreigners (Macesich, 1985). Breton (1964:378) notes that governments utilise "nationalistic instruments… for the purpose of increasing the share of assets in a given assets in a given territory owned by the nationals of that territory". Breton (1964) also observes that nationalistic redistribution of investment and capital results in a lower rate of return than would be realised if resources were allocated efficiently, an observation that draws parallels with the nature of social preferences on a much larger scale. The practice of promoting these nationalistic policies that are not beneficial to certain population segments is centred on the formation of a nation-wide group identity that promotes solidarity in the same way that smaller scale groups do.

The membership of these nation-groups is defined according to several commonalities. Members share an economy, a historic territory, myths and memories, a public culture, and a set of legal rights (Smith 1991). What is clear from this definition is the lack of consistent personal characteristics, illustrating that members of a nation-group vary considerably in their social and personal identities. The minimal nature of the nation-group is accepted by many academics, some seeing nationalism as an ironic tool that encourages members to appreciate things that are national for the mere fact that it is national (Breton 1964). Karl Deutsch (1969:3) aptly described a nation as a "group of people united by a common error about their ancestry and a common dislike of their neighbours," evoking the notion that national identity is a predominately social construct inaccurately perceived to be connected to common characteristics, descent and preferences. (Smith 1996) stresses that the perception of one's own nationality and that of others is inherently only emotional, implying a subjective disregard for objective considerations that results in large social distances between foreigners and nationals, and smaller social distances between nationals. However erroneous, the very substantial influence nationality exerts can be seen through the stereotypical actions of distrusting of foreigners and supporting fellow country-men, making nationality another excellent candidate identity to examine how variations in perceived identity cause variations in the social preferences displayed.

Experimental Design and Implementation

It is social identity and its inherently variable quality that is at the heart of this study. It is this study's aim to discover in what way the perception of this identity can affect the extent that social preferences are displayed and whether or not a shift of focus from one form of social identity to another will cause a change in degree of social preferences manifested. Given its ability to expose these preferences, an extension of the Trust Game is used in this study. Building on the theorem that individuals possess multiple identities that are brought to the foreground in particular situations, this study equates multiple identities to multiple social distances that exist between any given two individuals. An example of this variance in perceived identity can be seen in the interactions of football supports. A particular football club supporter may exhibit ferocious hatred to a supporter of a rival team while the two teams are playing a match but the same supporters may feel a bond of fellowship while together supporting the national team. This is a simple but profound example of the effect of perceived identity and implies that a shift in perspective can result in a shift in preferences. According to the theory above, players orient themselves to the social identity of other players and so intuition follows that the results from interactions with a focus on nationality will be significantly different from interactions with a focus on gender. In essence this study aims to clarify, using economic game theory and statistical analysis, the intuition that social identities and distances are vague social constructs that although effect our everyday interactions, are ultimately mere fallacies of our perception.

As the identities of nationalities and gender were being used in this experiment, an equal number of male and female participants of various nationalities were sought. A total number of eighteen individuals participated, six of them Scottish nationals, six of them English nationals and six of them Chinese nationals. Gender was distributed equally among nationalities with each group comprising of three males and three females. Recruitment was from contacts within the University of Edinburgh and several blanket email advertisements. All participants were undergraduate students within the age range of 18 to 23 years old. The description of the experiment was relatively vague, inviting participants to take part in a study to help a fellow student, with no information as to the purpose or context of the study intentionally omitting any mention of social preferences and social identities. The chance of winning a cash prize of £20 was offered.

The experiment took place in two stages: the nationality stage and the gender stage. On arrival, after reading the instructions (see Appendix A) participants were invited to enter one of three rooms according to their nationality. In each room video clips were played which had some bearing on the nationality in question. Among the clips shown in each room were scenes from national parades, famous sporting wins and iconic scenery from the respective countries. Upon viewing, participants were asked to write down, without submitting for inspection, their thoughts on what it means to be of their particular nationality as opposed to a member of any other nationality.

Replicating real world constructs of the social identities of gender and nationalities was key in this experiment. Identity group salience is linked with the exaggeration of behavioural differences and stereotypical perception (Turner 1982) but care was taken so as to only attempt the creation of social identities and not arbitrary groups that would artificially affect preferences as a result of group identity influences over and above typical group mentalities associated with collective social identities. The video clips and request to write down thoughts were used to stimulate personal contemplation on the definitions society provides for nationality. Any group discussion on this matter which would also have encouraged a focus on the national identity would have had the side effect of not only creating a group dynamic but also the interactions that participants would experience when discussing would have significantly skewed results. Participants were asked to not communicate with any of the other participants at any point in the study, a request enforced by an invigilator. As referred to earlier, Roth (1995) discovered that any form of communication can have profound effects on social preferences and thus the rule of no communication was strictly enforced.

Participants were then invited in pairs to another room to play a variation on the Trust Game, explained in further detail below. Players were informed that their end payoffs would equate to virtual lottery tickets for the cash prize, the larger the payoff, the greater the chance of winning the cash prize. After each participant had played one instance of the Trust Game, they were then invited to enter one of two rooms according to their gender. Again, in each room video clips were played which had some bearing on the gender in question. Among the clips shown were scenes from James Bond and suffragette protests. Repeating the previous format, participants wrote down their thoughts on what it means to be of their particular gender. A second game was played by each of the participants with care taken to ensure that no player was paired with the same player from the previous stage. After all games were finished, payoffs were totalled and converted to virtual lottery tickets before a draw was made and the winner given £20 thereby formally ending the experiment.

Figure 1 shows the exact composition of pairings in both stages of the experiment. As can be seen in each stage there were three pairings of players of the Same Nationality but Different Gender (SNDG), two pairings of players of a Different Nationality and Different Gender (DNDG) and four pairings of players of a Different Nationality but the Same Gender (DNSG). Although some of the subject pairings are the same in both stages in terms of their identity classifications the individual subjects themselves are never paired with the individual from the previous stage and the majority of subjects are paired with a subject of a different identity to that of the previous stage. With the exception of two players, subjects through-out both stages play once as the Investor and once as the Trustee.

The Trust Game used in this experiment differs slightly from the traditional format in that both the Investor and the Trustee are given initial endowments of £10 as opposed to only the Investor receiving an endowment. Excepting this variation, the Trust game used is otherwise the same as in previous studies. The Investor is first given the option to give any amount £X between £0 and £10 which will be transferred to the Trustee as £3X. Upon receiving this, the Trustee is given the option to give any amount £Y between £0 and £10 + £3X to the Investor. The purpose of this format is to rule out inequity aversion on the part of the Investor and focus on the amount of trust they place in the Trustee, making it easier to map effects across variations in identity. Given the fact that initial endowments are equal, investors should feel no sympathy towards the Trustee and no compulsion to create an equitable distribution. Any amount sent, therefore, should be viewed as an investment with an expected return, the value of the investment varying with this expectation and thus varying with the degree of trust.

In each stage participants would enter the room in which the Game would be played and take a seat on either side of a table with a stack of ten casino tokens to represent the money involved. Before the Game began the experimenter clearly explained the format of the Trust Game to ensure no uncertainties. He also stated the identity of the two players in accordance with the current stage. In the first stage, the experimenter stated the nationality of the two players and in the second stage the experimenter stated the gender of the two players. After this, the Investor was asked to begin the game by moving his or her chosen number of tokens across the table to the Trustee. If the Investor passed none, the game was declared over and the players taken back to their respective rooms. If the Investor did choose to pass an amount, the experimenter multiplied the number of tokens sent by three and passed these on to the Trustee. The Trustee was then invited to finish the game by passing back a certain number of tokens or passing back none at all. Players were not made aware of the decisions of any other player, excepting the one he or she was paired with and participants were prohibited from discussing results during or after the experiment.

There are several limitations in the experiment that should be mentioned before continuing. The small numbers of participants suggests that the results presented here can only be used as the premise for future experiments. Although there are definite trends in how players interact, eighteen players, nine pairs and two games played puts into question any conclusion, however significant, when compared to experiments utilising of hundreds of subjects. Players also played the Trust Game face-to-face. Hoffman et al. (1994) find that experiments using a 'double-blind' design in which players' identities were hidden both from each other and the experimenter result in more selfish actions. Hoffman et al. (1994) suggest that subjects will be more selfless in the presence of observation because they presume they are expected to be, reminiscent of metaperspective influences.

In this study, this concern is theoretically only a matter of levels. If participants are influenced by the presence of the experimenter this would be a constant for each pair as the experimenter was present at each game instance. If the participants were influenced by knowing that the other player had knowledge of their own identity, this influence should vary between pairs. The theory of social distance implies that there are less incentives to conform at greater social distances (Akerlof 1997) and so, theoretically, there should be the same amount of variation across pairs in both double-blind and face-to-face designs. Berg et al (1995) implement a double-blind design to prevent results being affected by reputation, collusion or the threat of punishment, but these again should vary according to perceived social distance. Problems may arise if the fact that face-to-face interaction breaks down social identities and allows certain players to perceive their counterpart's role-identity as defined by their particular idiosyncratic interpretations (McCall and Simmons, 1978) thereby making it more difficult to control for variations in social identity. Owing to the fact that participants played the Trust Game twice, albeit with different players in different roles, expectations may have been built that influenced predictions as to the actions of Trustees. Although with higher numbers it would be possible for participants to play just once, the design in which the same participants play twice emphasises the variability and malleability of social identity. This format in fact draws parallels with real world interactions in which individuals repeat similar interactions, learning and changing their actions, thus forming expectations and stereotypes. The inconsistency of subjects' social preferences in the face of a change in identity focus demonstrates the fickle nature of these stereotypes.

Predictions

Putting aside design considerations and before analysing the experiment's results, it is important to clearly state how these results should be viewed. The main hypothesis of this study is that perceived identity is related to the expression of social preferences namely those of reciprocity, inequity aversion, altruism and trust. If there is a common identity between the Investor and the Trustee, social preferences will be displayed to a greater extent. This gives rise to five hypotheses. [1] The significant influence of nationality in the first stage, but not in the second, and that of gender in the second stage but not in the first will verify the hypothesis that social distance is a perceived and inherently variable concept. This influence will be seen in the degrees of social preferences manifest. [2] Greater levels of trust will be displayed by Investors interacting with those of a similar social identity in accordance with the particular identity focus of the stage. This will be seen as Investors give more to the Trustees they 'trust'. [3] Trustees faced with an Investor of a similar focused social identity will show reciprocity by giving back more than the amount they received. [4] These same Trustees will also display greater signs of inequity aversion by more frequently returning double the amount that they received thereby equalising the final payoffs. [5] Altruism will be displayed by these Trustees giving amounts over and above double that of what they received. The implication of each of these hypotheses is that these indicators of social preferences will be displayed less frequently and in lesser degrees when the pairings have no focused social identity in common.

Results

The focus on different identities affects the outcome of the game in the amount of trust the Investor places in the Trustee but does not appear to significantly affect the value returned by the Trustee. The value returned nonetheless exhibits signs that the players do possess social preferences. Figures 2, 3 and 4 show the differences between the values Investors gave to Trustees, the values Trustee returned to Investors, the profit made by the Investors and the total payoffs of both players when mapped across the two stages of first nationality and then gender and across the variations of pair makeup. At first glance, it is evident that those pairings involving subjects of the same focused social identity boast higher averages in each of amount sent, returned, Investor profit and total payoffs.

In the first stage, Investors from SNSG pairings sent an average of £8.667, substantially higher than the average values sent by Investors from DNDG and DNSG, investments of £4.5 and £5.25 respectively. Trustees from SNDG pairings returned an average of £14.667, much higher than the averages of £6 and £8.5 of DNDG and DNSG pairings although, as will be discussed below, this indicator is not entirely relevant. Investors' profits were much higher in SNDG pairings, an average of £6 compared to averages of £1.5 and £3.25 in DNDG and DNSG pairings. The result of these transactions saw Investors and Trustees in SNDG pairings achieve average total payoffs of £18.667 compared to the lower values in DNDG and DNSG pairings of £14.5 and £15.25 respectively. In the second stage, these values are higher for DNSG pairings. Averages of Sent, Returned, Profit and Total from these pairings are similar to those of SNDG in the first stage with values of £8, £14.25, £6.25 and £18 respectively. In this stage pairings of SNDG are comparable to those of DNSG in the first stage with average values of £4.333, £8.5, £4.5 and £14 respectively. As before, DNDG pairings posses the lowest average values of £3.5, £7, £3.5 and £13.5 respectively.

Although the averages of these categories is higher for nationality in the first stage, they are higher for gender in the second stage thus appearing to verify hypothesis [1] that social distance is inherently variable yet this must be confirmed through further analysis.

It should be noted that Stage 2 data has been adjusted to correct for an outlying data point. This outlier was the result of a Trustee player, a Chinese Female, 'giving up' and deciding to leave the game, returning everything in her possession to the Investor, a Chinese Male. To the best of the experimenter's knowledge this had no connection with culture or anything other than the peculiarities of that individual. The omission of this data point means that only two points were used to calculate Average Returned and Average Profit and four points to calculate Average Total in the SNDG pairing category. The data point used to calculate Average Sent is still valid as the Investor was not aware of the Trustees intentions. Relevant regression analyses and graphs of stage 2 also exclude this outlier.

Figures 5 and 6 show the distribution of amount sent and proportion returned across pairings in both stages of the game with the acronyms on the X-axis representing the pair type. In the first stage, all nine Investors transferred some amount of money, and in the second stage only one Investor did not transfer any amount, paralleling findings from Berg et al. (1995). In the first stage, one Investor transferred their whole wealth of £10, and this was repeated by two Investors in the second stage. In both stages, all Trustees returned some amount, excluding one instance in the second stage in which the Trustee received no money. Any amount that Trustees did return was consistently at least the amount given to them. Also shown in figures 5 and 6 is the 'equitable level' at which, should the Trustee return this value, both players will have equal payoffs. As can be seen, in the both stages four Trustees returned at least this value, with one pairing in each round seeing the Trustee give slightly more than this value.

Investor Giving

In the first stage, after participants had watched videos designed to reinforce their national identity, Investors tended to give more to a Trustee of the same nationality. In the second stage, after participants had watched videos designed to reinforce their gender identity, Investors tended to give more to a Trustee of the same gender than those of the same nationality. In both stages Investors do still give slightly more to Trustees of the same, but unreinforced identity - gender in the first stage and nationality in the second - when compared to Trustees of neither the same gender nor the same nationality. Figure 7 shows the results of running an Ordinary Least Squares Regression investigating the effect of both gender and nationality on the amount that the Investor gives in the first stage of the experiment. The regression shows that Investors do indeed give more to those of the same nationality, a result that is statistically significant at the ten percent level. In contrast, gender not only has a much smaller effect on the amount Investors give, increasing the value by a mere £0.75 compared to the £3.25 increase nationality sees, but this effect is also statistically insignificant. Figure 8 shows the same regression on the amount the Investor gives but in the second stage of the experiment. In this stage, it can be seen that Investors give more to those of the same gender but that they still give more to those of the same nationality than to those with whom they have no commonality. Although the figures show that gender and nationality increase the value of the amount sent by £9 and £5 sent, given the insignificant intercept one should only examine the difference between the two values which is largely evident. These results are both statistically significant at the one percent level. It is possible that in the second stage the prevalence of relatively high giving to those of the same nationality, while not as high as to those of the same gender, is a result of repeating the game and identity constructs not being fully replaced. What should be taken from this analysis is that the amount sent to those of the same gender in the second stage is both high and significant, an outcome that is vastly different to that of the first stage and gives much weight to the core hypothesis of this study.

Amount Returned

To measure the extent and variation of social preferences in Trustees, merely looking at the amount of money that they return to Investors is a deceptive indicator of the relationship between perceived identity and social preferences. It is logical to assume the presence of reciprocal preferences in participants and hence assume that the amount returned by Trustees is dependent on the amount sent by Investors, an assumption given credence by Berg et al. (1995). A more useful indicator of the effect of perceived identity is to look at what percentage of the amount Trustee receives is given back to Investors. If perceived identity had a reasonably strong effect, it would follow that Trustees would display greater degrees of altruism and inequity aversion and therefore return a higher percent of the amount given to them.

A simple glance shows that the amount returned by Trustees to Investors varies with identities and stages. As with the amount sent by Investors, this amount increases when dealing with those with a similar identity according to which stage of the experiment the game is played in. Figures 2, 3 and 4 show that the value of the amount returned is greater to those of the same nationality in stage one, and greater to those of the same gender in stage two. As stated above, the actual amount returned to the Investor is of little value in determining a connection between perceived identity and social preferences and so percentage returned is used to look into the variation between games. Each Trustee returned at least the same amount, regardless of the identity of their Investor, a clear sign of the presence of reciprocation but nevertheless shows the redundancy of using the amount returned as an indicator of the effect of identity. Figure 9 shows the amount Trustees returned as a percentage of the amount sent, showing a comparison between pairings of Same Nationality Different Gender, Different Nationality and Same Gender, and Different Nationality and Different Gender. Had perceived identity affected the percentage that Trustees returned, the points from the pairings of Same Nationality and Different Gender participants would have been noticeably higher than the points of other pairings. As one can see, this is not the case, a finding that will be discussed below.

Figure 10 shows a scatter plot chart in the same format for the second stage of the experiment. In this stage, had perceived identity affected the percentage Trustees returned, the points of from pairings of Different Nationality and Same Gender would have been noticabley higher than other pairings. Again, this is not the case. In addition to the outlying data point, one additional data point has been excluded due to the fact that the Investor did not send any amount.

To be certain that perceived identity had no effect on the percentage Trustees returned, an Ordinary Least Squares Regression was run, investigating the effect of gender, nationality and the amount that Investors sent. Figure 11 shows the results for the first stage. Although common nationality increases the percentage returned by 75%, this value, and the value attributed to common gender is not statistically significant. There is also no indication that the amount sent increases the percentage returned, contrasting with findings from Berg et al. (1995).

Figure 12 shows the same regression in the second stage. Again, it is impossible to suggest that identity had any real affect on the percentage Trustees returned although there is some indication that Trustees reciprocated Investor trust by returning higher values after receiving higher values, which unlike findings in the first stage, is in line with results from Berg et al. (1995).

Comparisons with Current Research

The results presented here also conform to those of other studies. Figure 13 shows the relationship between the amount Investor's sent and their final payoffs. Given that money sent can be equated to trust levels, it follows that higher levels of trust result in higher levels of wealth. These results are statistically significant at the 5% level, as shown in figure 14, demonstrating that subjects received on average £0.80 in total payoff more for every £1 invested in the Trustee. This is paralleled not only by Buchan et al. (2006) in their Trust Game experiments but also by Knack and Keefer (1997) in their empirical study relating higher levels of trust within countries with higher economic performance. The large influence that social identity has on wealth levels gives weight to Akerlof's (2000) claim that one's identity choices are a major factor in one's overall economic well-being. Investors paired with those of a different focused identity were less likely to trust their Trustees, therefore receiving lower average payoffs adding experimental evidence to Breton's (1964) theory that nationalism does not give rise to the highest income levels.

Although in the second stage of the experiment, nationality did not appear to be a major influence on interaction, this does not invalidate findings from Glaeser et al. (1999) but rather strengthens the conviction that social distance is variable. Similarly, the findings that gender was a significant influence on trust levels in the second stage do not invalidate findings from Buchan et al. (1999) again merely strengthening the hypothesis. These results can be seen to strengthen the conclusions of Billing and Tajfel (1973) in that inaccurate social identity mirrors minimal group membership. This study adds to Billing and Tajfel (1973) in showing that minimal group membership, although relevant, is easily altered. Due to the small sample size this study is unable to draw any conclusions as to the country or gender specific sensitivity to social identity variations. It may still hold that certain subjects consistently display greater degrees of social preferences but in the experiment's current format, a great deal more participants would be needed before any analysis was attempted.

Interpretations and Conclusions

Due to the fact that an initial equal distribution of endowments rules out inequity aversion as a motivation for Investor giving, it is fair to assume that the variations in the amount sent by Investors relates to the amount of trust placed in the Trustee in the anticaption that there will be benefit from the 'investment'. From the results we can see that investors place more trust in those of a common identity, confirming hypothesis [2] and conforming with Akerlof's (1997) theory of social distance. This trust, however, varies depending on the stage of the experiment. In the first stage, Investors place more trust in Trustees of the same nationality but in the second stage more trust is placed in Trustees of the same gender. This is a clear indicator that trust is merely dependent on how the identity of an individual is perceived, rather than on the actual identity of the individual, thereby verifying hypothesis [1]. This stands as emperical evidence of the notion developed by McCall and Simmons (1978) that individuals are only able to interact with strangers according to vague perceptions of their social identity.

There is strong evidence of reciprocity across games, given that every Trustee returned at least the amount they were given. Although this does not disprove the theory of self-interest given the fact that Trustees may be maximising other preferences, this does prove that maximising monetary gain is not a dominating motivation. It is evident that hypothesis [3] does not hold as there is too much variation in the levels of reciprocity to determine any clear link with social identity. Over and above motivations of reciprocity, displays of inequity aversion were very much apparent when looking at the percentages of investments returned but due to the fact that no clear trend could be found between these figures and variations in identity hypotheis [4] does not hold. The same is true for hypothesis [5] although the mere fact that some altruism was displayed adds weight to the revelance of social preferecnces. The precence of social preferences adheres to Amartya Sen's (1977) belief that traditional self-interest theory does not explain all forms interaction. Very few subjects displayed signs of being 'social morons'.

One possible reason for the lack of connection between amounts returned and social identity is because the basis of people's actions towards strangers when observing their identity is prejudice. People make assumptions and predictions on the actions a person may take based on a judgement that takes into consideration very little information. In the experiment examined in this study, Investors decided how much trust to place in Trustees with very little information, the vast majority of which was centred on their identity of gender or nationality and very little else. Trustees, however, were party to additional information, namely how much trust the investor was willing to place in them. It is possible that they were no longer able to make a prejudiced decision on how much to give to Investor as they were aware of how much the Investor would give to them. This personal interaction of sitting with the Investor face to face may also have had an effect. Trustees may not have wished to appear selfish to the person sitting across from them after receiving payment from them. It could be seen that the Trustees viewed the Investor no longer as an individual with a certain gender or nationality, but as an individual who has entrusted them with money with the expectation of reward, thereby over-riding social identities and going someway to explaining the lack of correlation with the two focal identities.

Although social identity and social preferences have been studied extensively, this study has attempted to highlight just how variable these concepts are, not across demographics but within individuals themselves. The relative ease at which subjects assumed and changed their social identity focus reinforces evidence that minimal groups are still effective but emphasizes that these groups are in no way lasting. The finding that manipulating an individual's identity focus can change their preferences has wide-reaching connotations. In advertising, attempts should be made to not only sell products on the basis of identity but also to persuade consumers to bring particular identities to the fore. At present, some foodstuffs are advertised in a nationalistic fashion with large stamps saying that it is from the consumer's own country. With this particular example, marketers should be able to find more success if they are able to evoke nationalism within the consumer via say, images of national pride thereby encouraging consumers to buy based on a strong social identity connection. There are also implications for government initiatives for cultural integration. Many initiatives focus on diversity and tolerance and yet this study would suggest that initiatives that build on the commonalities through-out cultures would be more effective.

Given such substantial implications and this study's significance even with a small sample size, it is this researcher's hope that this will act as a pilot study for future larger and more expansive experiment's to be based.

Bibliography

• Agell, Jonas and Lundborg, Per (1995): "Theories of Pay and Unemployment: Survey Evidence from Swedish Manufacturing Firms", Scandinavian Journal of Economics Vol. 97, pp. 295-308
• Akerlof, George (1997): "Social Distance and Social Decisions" Econometrica Vol. 65, No. 5 pp. 1005-1027
• Akerlof, George A. and Kranton, Rachel E. (2002): "Economics and Identity", The Quarterly Journal of Economics Vol. CXV, No. 3, pp. 715-753
• Andreoni, James and Miller, John (2002): "Giving according to GARP: An Experimental Test of the Consistency of Preferences for Altruism", Econometrica, Vol. 70, No.2, pp. 737-753
• Baughn, C. Christopher and Yaprak Attila (1996): " Economic Nationalism: Conceptual and Empirical Development" Political Psychology Vol. 17, No.4, pp. 759-778
• Bearden, William O. and Etzel, Michael J. (1982): "Reference Group Influence on Product and Brand Purchase Decisions", The Journal of Consumer Research, Vol. 9, No. 2, pp. 183-194
• Berg, Joyce; Dickhaut, John and McCabe, Kevin (1995): "Trust, Reciprocity and Social History" Games and Economic Behavior, Vol. 10, No.1, pp.122-142
• Bewley, Truman (1999) Why Wages don't fall during a Recession, Harvard University Press,
• Billig, Michael and Tajfel, Henri (1973): "Social Categorization and Similarity in Intergroup Behavior,'' European Journal of Social Psychology, Vol. 3, No. 1, pp. 27-52.
• Bowles, Samuel and Gintis, Herbert (2000): "The Evolution of Strong Reciproc