Master thesis Psychology, specialization: Social and Organizational Psychology
Institute of Psychology
Faculty of Social and Behavioral Sciences – Leiden University Date: 01-06-2017
Student number: S1237233
First examiner of the university: Jorg Gross
Second examiner of the university: Carsten De Dreu
Governing the commons
The role of leadership on cooperation and trust in a
public goods game
Robbin Dercksen
Abstract
Cooperation is a behavior that often occurs in social situations. Trust plays an important role in cooperation. Former research in naturalistic settings shows that leadership has an influence on trust and on cooperation between people. It shows that more democratic settings are better for cooperation and trust between people. Former research shows that personality differences are also a variable of interest in investigating cooperation. This research manipulated leadership arrangements, one group was a dictatorship, the other a democracy. Our expectation based on former research was that a democracy is better for cooperation and trust. We found that form of leadership had no effect on trust, and that a dictatorship was better for cooperation than a democracy. Our research further shows that personality differences only play a marginal role in cooperation and trust.
Introduction
Collective cooperation is a behavior that often occurs in human interaction, even though cooperation is not always the most viable strategy for individuals (Rand & Nowak, 2013). Cooperation is a very difficult subject with some contradictions. Is it a smart idea to cooperate when others clearly don’t cooperate? Why help a potential competitor? In theory cooperation might be adaptive and maladaptive at the same time (Rand & Nowak,
2013). It is this discrepancy between the fact that cooperation might be maladaptive for individuals and cooperative behavior being widespread has become the topic of much research (Nowak, 2006). Some researchers state that cooperation is not adaptive from an evolutionary perspective and try to find conditions under which cooperation might still be apparent in today’s human interactions (Nowak & Sigmund, 1998; Panchanathan & Boyd, 2004).
Social Dilemmas as a way of measuring cooperation
A method to experimentally test cooperation is the use of social dilemmas. The core of a social dilemma is the choice for maximizing one’s own outcomes, or maximizing the outcomes of the group: the choice between cooperation and defection (Baron et al., 2007; Forsyth, 2014; Rand & Nowak, 2013). Motives for cooperation of defection are thus in conflict with each other in this kind of games. There are different versions of social dilemmas, but two very basic ones are the so called ‘Commons Dilemma’ and the ‘Public Goods Dilemma’.
The Commons Dilemma can be illustrated very vividly on the basis of a story, which depicts the essence of the dilemma. Consider a very large patch of grass in the desert. There are several shepherds who share this patch of grass for their sheep to graze. The patch of grass is able to feed all the sheep effectively, but some shepherds decide to add some extra sheep to their herd, to make an extra profit of their sheep. As more
shepherds decide to do this the amount of grass starts to deteriorate the sheep have less to eat, become malnourished and soon after die of starvation (Hardin, 1968). Commons Dilemmas are categorized by a resource a group of people share, of which they can decide to take something for them self. If everyone takes more than their fair share, the resource will soon deplete. A thin line between short-term personal gain and long-term losses arises (Forsyth, 2014; Pruitt, 1998).
Another form of social Dilemmas is the so called ‘Public Goods Games’ (PGG). A Public Goods Game is an experimental measure to test cooperation which is often played for
multiple rounds (r). As opposed to a Commons Dilemma, which is a ‘take some’ scenario, a Public Goods Game is a ‘give some’ scenario. In each round of the Public Goods Game participants receive an amount of tokens called the ‘endowment’ (Pe).
Together with n other participants the participant has to decide how much of their endowment to contribute to a common pool (ci) keeping only a part of the tokens for
themselves (πi = 20 – ci). All participants make this contribution decision simultaneously.
Tokens that are contributed are multiplied with a certain factor (x). The personal outcomes (πi ) of a participant in a round of a PPG would come down to:
πi = Σ (ci * x) / n (Pe – ci)
If three participants would all contribute 10 tokens in a given round, which are multiplied by factor 1.5, the personal outcomes for the participants would be: (10 + 10 + 10) * 1,5 / 3 + (20 -10) = 25. For every point contributed every members receives 1 * 1,5 / 3 = .50 extra on top of the endowment they did not contribute.
Out of this formula the conflict of interest of the PGG arises: should I contribute so the entire group benefits or should I defect so I can maximally benefit from the contributions of other participants? Participants who decide to contribute their entire endowment add to maximizing the groups outcomes. When sticking to the example above, for every token contributed, 1.5 token is generated, which is more than 1 token. Because the generated tokens are divided equally amongst all participants, one receives only .5 of a token, which is lower than 1 token. Cooperating may cause to decrease their personal outcome at the risk that others will not reciprocate. When other participants decide to free ride and to contribute nothing to the common pool, participants who do cooperate receive a lower personal outcome than their initial endowment (Haigner & Wakolbinger, 2010; Wall & Nolan, 1987). When cooperation is a risky choice, choosing to do so might be based on some level of trust in other people to cooperate (Arrow, 1972).
Trust seems to play an important role in contribution in public goods dilemma’s, in which higher levels of trust lead to higher contribution (Kocher, Martinsson, Matzat & Wollbrant, 2015). Trust in other people seems to be an aspect of daily interaction
between human beings. Higher levels of trust lead to a higher intention for cooperation, which in turn lead to better (economic) outcomes (Arrow, 1972; Johnson & Mislin, 2011). Trust is important in economic and political exchanges between people (Buchan et al., 2002; Kosfeld et al., 2005). Trust in terms of the Public Goods Games is based on the decision how much people are willing to contribute, which ranges from full cooperation to defecting and free riding (Fischbacher, Gachter, & Fehr, 2001; McNamara, Stephens, Dall, & Houston, 2009). In the Public Goods Game participants don’t know how much the other participants will contribute, so full cooperation seems to be a risky choice (Wall & Nolan, 1987). The expectation (or the trust in other participants) to contribute seems to foster higher levels of cooperation. Higher levels of trust lead to higher contributions to the common pool (Shikishima et al., 2015).
A common way of measuring interpersonal trust is with the so called ‘Trust Game’, a game in which person A has to transfer money to person B (Johnson & Mislin, 2011). One of the first Trust Games was created by Camerer and Weigelt (1988). This version of the game became the most prominently used in the field and was revised an adapted in many other experiments (Johnson & Mislin, 2011). This game is basically a transfer of one’s endowment to another player. The amount of money which is
transferred is multiplied by a certain factor and there after given to the other participant. The receiver can choose how much of the received endowment to return to the sender. The measure of trust in this game is sending money to another participant in the hope he or she reciprocates; which is a risky move. Higher trust should lead to higher transferred sums of money (Espin et al., 2016). Most if not all studies researching cooperation in Social Dilemmas fail to take interpersonal trust into account. We expect that people who are higher in trust, show higher levels of cooperation in a Public Goods Game.
If people don’t trust each other there are ways to boost cooperation in Social Dilemmas. Many Public Goods Games work through a voluntary contribution
decide for themselves how much they would like to contribute to the common pool of resources. Most Public Goods Games use VCM as the rule for contribution (Drazen & Nowak, 2014). Most studies show, when the VCM-rule is applied in a Public Goods Game which is played for multiple rounds, cooperation eventually reaches zero. There are ways to adapt the VCM- rule, so people lose the (direct) control over how much to contribute, for instance introducing a leader (Rivas & Sutter, 2011). The leader decides how much every participant has to contribute to the common pool Participants in the Public Goods Game lose the opportunity to decide how much points they are willing to contribute to the common pool. This decision how much every participant should
contribute is transferred to a leader, who decides for everyone. When adapting the Public Goods Game different forms of leadership can be applied (Chiang & Fsu, 2016; Drazen & Ozbay, 2015). An often drawn distinction is between a endogenous (democratic) and an exogenous (dictatorial) condition. In real life this can be seen as the difference that can be drawn in how governments run their country: the difference between a democracy and a dictatorship (Deacon, 2009; Haigner & Wakolbinger, 2009; Rivas & Sutter, 2011).
As said a broad distinction can be made in ways countries are governed. Most if not all countries in the Western World can be seen as democracies, while many third world countries are seen as dictatorships (Pollack & Mateur, 2006; Chipkin, 2009; Camprubi, 2016; Acemoglu & Robinson, 2005; Ziblatt, 2006; Corduwener. 2016). Democracy originates in ancient Greece. ‘Demos’ being the Greek word for ‘people’ and ‘cratos’ being the Greek word for ‘power; power to the people. Democratic countries are characterized by a more equal division of power under the population (Arrow &
Robinson, 2005). In a Public Goods Game with an democratic leader participants have the opportunity to decide who they want as a leader or to vote on basic rules for cooperation.
As opposed to power for the people in a democratic political system, a dictatorial system is characterized by a very small group of persons who have the entire basis of power. This political system has as a consequence that it mostly fulfills the basic needs and interests of a small group of elite which often goes hand in hand with high levels of corruption (Deaco, 2009; Acemoglu & Robinson, 2005). Corruption usually takes the
form of government officials putting public funds in their own pockets. In a Public Goods Game a dictator might decide that he or she defects, but decides to allocate the entire endowments of the subordinates to the common pool. The dictator exploits the subordinates for their own profit, while not contributing anything from their own endowment. Large differences in those factors can be noticed between two political systems (Deaco, 2009).
The democratic systems has the consequence that leaders should gather the support of the people to implement policies and therefore they should be aimed at the general population. The consequence of policies implemented by leaders chosen in a democratic way, as opposed to dictators who decide alone, are generally seen as more fair and therefore accepted faster (Magalhaes, 2016). This procedural justice plays an
important role in the acceptance of decisions the leadership makes for the rest of the group (Foster & Putterman, 2010; Tyler & Blader, 2003; Bó).
The form of authority might play a role on people’s willingness to cooperate and to contribute to the pool of common resources in Social Dilemmas (Chiang & Fsu, 2016). The effects which leadership might have on cooperation are illustrated by a study carried out by Bó et al., (2003). The researchers let participants play a prisoners dilemma, in which a participant has to choose between maximizing one’s own outcomes or the outcomes of both themselves and the other participant. This game lets the participant explicitly chose to cooperate or to defect, as opposed to the PGG, which only lets participants decide how much they’d like to cooperate. This game was played in groups of four and before each of the ten rounds the participant was randomly matched with another participant to play the upcoming round. Before this Prisoner’s Dilemma started, the group of four participants had to decide whether to impose a fine or tax on non-cooperation (the endogenous condition). For another group of participants the computer randomly decided whether a fine or tax was incorporated in the game (the exogenous condition). The outcome was that participants who were in the endogenous condition showed more cooperation and support for the fine than groups who were in the
exogenous condition; procedural justice plays an important role. The question remains what the role of trust is in this process, which this research does not investigate.
Besides some form of fair procedures in decisions made by leaders, reciprocity also plays a role: leaders in a democratic system should gather the support from subordinates to carry out their policies, and followers need to be sure that a leader has their interests at heart (Charness & Rabin, 2002). This reciprocity between leaders and followers is displayed in another experiment by Drazen and Ozbay (2015). Groups of three participants took part in this experiment. There were two conditions in this experiment: in the ‘Election condition’ three participants took part of which one is
appointed as a citizen, the other two were appointed as candidates. The citizen has to vote which of the two candidates is chosen as the leader. In the ‘Appointed condition’ one of the three participants was randomly chosen, without an election. Leaders were randomly given a policy of their preference and soon after their election/appointment they learned the preference of the other two participants. Leaders who were elected showed a higher preference for choosing a policy that was not their own, in comparison to leaders who were appointed without an election. This seems to indicate that democratically chosen leaders act in a more non-selfish fashion and take the wishes of the citizens into account. Leaders who were randomly chosen showed the tendency to choose a policy that was not their own to a far lesser extent. Only 25 percent of appointed leader chose a policy other than their own. This study shows that democratically chosen leaders are less likely to exploit their subordinates in favor of their own goals, than leaders that are appointed as leaders. In a study conducted by Brandts, Cooper and Weber (2015) in the same setting, also showed that when people had a voice in the process of choosing their leader in an effort investing game, participants showed higher levels of cooperation compared to people whom could not chose their leader.
Most of the studies regarding cooperation in social dilemmas use games which put cooperation in an ‘all or none’ fashion, which means people either cooperate completely or defect completely. A Public Goods Game offers participants a continuum to decide how much they would like to cooperate. Another thing former studies fail to do is reversing the process of leadership; what could be the consequences of removing leadership after participants have to play the Public Goods Game on cooperation? We expect that people who have had a voice in choosing their leader show higher levels of
cooperation when the leadership is removed in a Public Goods Game than people who did not get to choose their leader.
So far we looked at the effects of trust and leadership on cooperation separately. Trust and form of authority form an interesting interaction, when comparing people who live in democracies versus people who live in dictatorships on (interpersonal) trust. People who live under dictatorships are less trusting of the leadership to begin with, but are also less trusting of other people in general too (Kim, 2014). This effect is illustrated by comparing different countries who differ in political systems and looking at different levels of trust. People who live in dictatorships indicate they trust the regime less than people who live in democracies and are more skeptical about engaging in political changes (Kim, 2014; Nicholas, 2012; Yakinthou & Croeser, 2016). In Public Goods Games this effect can be noted by comparing different populations to each other. In a set of studies measuring cooperation in Public Goods Games in different populations
worldwide and comparing the levels of cooperation find interesting differences. Comparing for instance Kamchatka collectives (a population in Eastern Siberia mostly left unaffected by the strict Soviet Regime) in Post-Soviet Russia to people living in urban areas shows there are large differences. Cooperation seems to be much higher to populations left unaffected by a dictatorship than populations effected by it. Same goes for comparing the Kamchatka collectives to people from other dictatorships than Soviet Russia (e.g. Zimbabwe) which supports this theory even further (Gerkey, 2013; Henrich, 2000; Barr, 2001).
Trying to integrate all this research shows that people who lived in dictatorships are less willing to cooperate in the Public Goods Game, as compared to people who lived in more democratic societies of are left unaffected by the dictatorship (Gerkey, 2013; Henrich, 2000; Barr, 2001). Furthermore people who live under dictatorships show less (interpersonal) trust and less trust in the regime (Kim, 2014; Nicholas, 2012; Yakinthou & Croeser, 2016). Like the studies of Drazen and Ozbay (2015) Brandts et al. (2015) and Bó et al., (2003) showed there are differences in how leaders behave towards their subordinates, based on how they gained their power. All this research leaves us with the notice of differences between groups, but failed to either manipulate leadership when
measuring trust or cooperation or failed to measure trust after an experimental
manipulation of leadership. This present study manipulates leadership in a Public Goods Game and thereafter measures trust to gather a comprehensive overview of the effects of different forms of leadership on (interpersonal) trust and cooperation. This study will is also aimed at finding out what the consequences are of removing leadership on
cooperation in a Public Goods Game.
As mentioned are people who are democratically chosen in a Social Dilemma seem to be less exploitive of their subordinates than dictators who are not democratically chosen (Acemoglu & Robinson, 2005; Deaco, 2009; Drazen and Ozbay, 2015). People in democratic conditions seem to be more cooperative than people who live under
dictatorial conditions (Bó, Foster & Putterman, 2010; Tyler & Blader, 2003). People who live in dictatorships experience less trust in general compared to people who live in democracies, so we assume this would also play an important role in the interpersonal context (Kim, 2014; Nicholas, 2012; Yakinthou & Croeser, 2016). Trust and reciprocity are also very important in cooperation. People low in trust are therefore more likely to cooperate less (Arrow, 1972; Baron et al., 2007; Johnson & Mislin, 2011).
Hypothesis 1: We expect leaders in the dictator condition to be more exploitive in terms of forcing others to contribute more to a common pool of resources than leaders who are elected.
Hypotheses 2: We expect participants in the democratic condition to be more trustful to the other participants in the trust game than participants in the dictatorial condition. Hypothesis 3: We expect subordinates in the democratic condition to contribute more to the common pool of resources when the political systems are removed compared to people in the dictatorial condition
As we have noted are dictatorships often aimed at enriching just a small portion of the population or the leadership itself and exploit subordinates when possible (Acemoglu & Robinson, 2005; Deaco, 2009). A sense of hierarchy gives the leader the power to
administer outcomes to subordinates (Greer & van Kleef, 2010; Nikiforakis, Oechssler & Shah, 2014). Nikiforakis et al. (2014) illustrated this by a piece of research in which a leader has to make a suggestion to how much effort a subordinate has to locate to a certain task. The leader has the chance to exploit the subordinate without consequences. Having a sense of power might have people act differently in which personality may also play a role (Chemers, 2000). Might there be an interaction between personality and leadership?
When looking at personality research in combination with exploitive tendencies a cluster of personality characteristics called ‘The Dark Triad’ arises. This Dark Triad consists of Machiavellianism, psychopathy, and narcissism. These three personality characteristics are often aimed at using and abusing other people in reaching your own goals, without taking the interests of other people in regard. Especially Machiavellianism correlates high with dishonest behaviors such as exploiting others for your own short term gain (Béla Birkás, Csathó, Gács, & Bereczkei, 2015; Brankley & Rule, 2014; McCain et al., 2016; Wisse & Sleebos, 2016).
Besides the Dark Triad another personality measure can help in explaining differences in cooperation and exploitation. People differ in their preferences for distributing outcomes between themselves and others. A way of measuring this is with the Social Value Orientation (SVO). (Loewenstein, Thompson & Bazerman, 1989; Forsyth, 2014). SVO states that a conflict between concern for one’s own outcomes or the outcome of the others determines willingness to cooperate (pro-self or pro-social). Four subtypes can be determined: Individualists are pro-self. Competitors are pro-self but also aimed at minimizing the outcome for the other(s). Cooperators are both pro-social and pro-self. Altruists are pro-social but not pro-self. One’s SVO might determine how leaders distribute outcomes in a Public Goods Dilemma. People who score high on personality measures in the Dark Triad, such as Machiavellianism seem to be more exploitive towards other people and aimed at dishonest behaviors than people who score
lower on these concepts. (Béla Birkás, Csathó, Gács, & Bereczkei, 2015; Brankley & Rule, 2014; McCain et al., 2016; Popper, 2016; Wisse & Sleebos, 2016). Based on this reasoning we expect:
Hypothesis 4: We expect leaders who are more exploitive towards their subordinates to score higher on measures of the Dark Triad than leaders who are less exploitive.
Leaders who exploit other people are mostly aimed at enriching themselves, a trait which can be conceptualized by measures of Social Value Orientation (Forsyth, 2014;
Loewenstein, Thompson & Bazerman, 1989). Based on the subtypes of the SVO, two subtypes seem suitable to conceptualize leaders who exploit and leaders who don’t exploit their subordinates (van Lange et al., 2007):
Hypothesis 5a: We expect leaders who are more exploitive towards their subordinates to be more pro-self than leaders who don’t exploit their subordinates.
Hypothesis 5b: We expect leaders who are less exploitive towards their subordinates to be more pro-social than leaders who exploit their subordinates.
Method Participants
210 participants participated in this study of which 162 were females (Age: M = 21,51,
enrolled themselves via the research website of Leiden University (SONA). First year psychology students participated for partial course credit, students from other studies participated for money or because they wanted to participate in an experiment.
Procedure
Participants enter the lab where a research assistant is waiting for their arrival.
Participants are given an informed consent (see Appendix A.1.) after signing they were taken to a computer. The research consists of four parts, which are described below. Before every part participants receive basic instructions on the computer screen.
The first part of the experiment is a Public Goods Game (PGG), participants play with two other participants. There are two conditions in this first part: a democratic condition and a dictator condition.
Participants read in the instructions that in the PGG that they are about to play the leader decides how much everyone contributes to the common pool of resources. In the dictator condition people are either appointed leader or follower. In this condition participants read that the leader cannot be changed during the game. In the democratic condition in this PGG participants read that the leader is chosen every three rounds. Participants vote on who the leader is for the coming three rounds. The PGG is played for a total number of 18 rounds. After each rounds participants are presented an overview of how much points every player earned. The outcomes of every round are added up and displayed on the end of round 18.
After the PGG participants play a trust game. Participants have to decide how much of a maximum of 10 tokens they transfer to either one of the other participants. After this transaction the tokens are tripled and the receivers get an overview of how much tokens they have received. After this the receivers have to decide how much tokens they transfer back to the sender.
The third part is a new PGG. In the PGG the VCM-rule is applied and is again played for a total of 18 rounds. The last part consists of three questionnaires. The first questionnaire regards participant’s Social Value Orientation (SVO). The second questionnaire regards the Dark Triad. The third questionnaire consists of demographic questions. After the last questionnaire the participants are debriefed and payed for their participation.
Independent variables:
There are several variables we measure in our research to effectively test our hypothesis. Exploitation
Exploitation can be measured by calculating the difference between what leader
contribute and what leader makes the subordinates contribute. This difference averaged over rounds in the Public Goods game gives an indication of how much the leader exploited the subordinates on average.
Cooperation
To measure the extent participants cooperate a Public Goods Game is used. The original Public Goods Game used in Haigner and Wakolbinger (2009) is adapted to test our hypothesis. Participants receive and endowment of 20 Monetary Units (MU’s) to contribute to the common pool. The endowment corresponds with real money: 1 MU is worth 4 eurocents. MU’s which are endowed to the common pool are multiplied with a factor of 1.5. Cooperation can be measured by the extent people choose to endow their 20 MU’s to the common pool; higher amounts added to the common pool correspond with a higher level of cooperation. When all participants cooperate, a total number of 60 MU’s are allocated to the common pool. After division every participant earns the maximum amount of MU’s; 30 MU’s.
Whether to see whether there are personality differences between people who get in a position of leadership in terms of who abuse their power and who don’t the Dirty Dozen Scale (DDS) is used to assess whether these differences exist. The DDS measures three constructs, namely Machiavellianism, psychopathy, and narcissism; together called the ‘Dark Triad’. The Dirty Dozen Scale is designed by Jonason and Webster (2010; Webster and Jonason, 2013). This scale derived from a few other scales and established via a factor analyses three subscales were developed.
The first one is Machiavellianism, which contains five items. An example of a question on this scale is “I tend to exploit others towards my own end”. People have to report on a 7-point likert scale, ranging from 1 (I disagree very much) to 7 (I agree very much). The Cronbach’s Alpha for consistency between the 5 items on this subscale is .92.
The second subscale is psychopathy, which contains five items. An example of a question on this scale is “I tend to not be too concerned with morality or the morality of
my actions”. People have to report on a 7-point likert scale, ranging from 1 (I disagree
very much) to 7 (I agree very much). The Cronbach’s Alpha for consistency between the 5 items on this subscale is .84.
The first one is Machiavellianism, which contains five items. An example of a question on this scale is “I tend to want others to pay attention to me”. People have to report on a 7-point likert scale, ranging from 1 (I disagree very much) to 7 (I agree very much). The Cronbach’s Alpha for consistency between the 5 items on this subscale is .92. Social Value Orientation (SVO)
To assess the differences between subjects in terms of their Social Value Orientation (SVO) the scale used in van Lange, Bekkers, Schuyt & van Vugt (2007) is used. This scale contains 9 items. An example of a question is) You get 480 other gets 80, you get 540 other gets 280 or you get 480, other gets 480. People then have to indicate which situation they prefer. On the basis of a code schema of answers, the SVO of the
participant can be determined. People get a range of 9 possibilities in the 9 questions to choose from. Each choice corresponds with a score; selfish choices earn less points and
pro-social choices earn more points. On the basis of their answers people who are prosocial score higher on the Social Value Angle (SVA) than people who are more selfish.
Results Exploitation
We expected leaders in the dictator condition to be more exploitive towards their
subordinates than leaders in the democratic condition. To test hypothesis 1 we calculated the average exploitation for leaders in the dictator condition and leaders in the democratic condition. The averages are displayed in Table 1. Leaders in the dictator condition exploit their subordinates on average (M = 3.47, SD = 4.40) more than leaders in the democratic condition (M = 2.53 SD = 3.05). This difference is however not significant (t (68) = -1.011, p =.315). This means that leaders in the dictatorship do not exploit their subordinates significantly more than leaders in the democratic condition.
To test if leaders show average differences in exploitation over rounds we did a Repeated measures ANOVA with exploitation as the dependent variable and rounds as within subject variable and condition as between subjects variable. The averages for exploitation over rounds are plotted in Figure 1. Testing this difference showed no significant main effect (F (1,68) = 1.023, p = .315). An analyses of the linear contrast for the interaction between the conditions and exploitation also gives a non-significant result (F (1,68) = 0.00 p = .994). Figure 1 also shows that there is no clear trend over rounds across conditions for exploitation. Hypothesis 1 is therefore rejected.
We expected participants in the democratic condition to be more trustful towards the other participants than participants in the dictator condition. We therefor expected the participants in the democratic condition to transfer more points to the other participants than participants in the dictator condition. To test hypothesis 2 we calculated the average amount of points transferred in the trust game for both conditions. The averages are displayed in Table 1. Participants in the democratic condition allocated on average somewhat more points to the other participants (M = 7.20, SD = 2.86) than the participants in the dictator condition (M = 7.05, SD = 2.60). An ANOVA with points allocated as dependent variable and condition as independent variable indicates that this main effect is not significant (F (208,1) = .613, p = .687). These results show that participants in the democratic condition are on average not more trustful towards others than participants in the dictatorship condition. Hypothesis 2 is therefore
Figure 1. Exploitation over rounds across conditions (blue = democratic condition, green
= dictator condition).
Cooperation
We expected participants in the democratic condition to contribute more than the
participants in the dictatorship once the institutions are removed. To test hypothesis 3 we calculated the amount of points given to the common pool. The averages are displayed in
Table 1. Contrary to our expectation, participants in the democratic condition contributed
on average less to the common pool (M = 10.1, SD = 6.95) than participants in the dictatorship (M = 11.9, SD = 6.38). To test this difference we did an ANOVA with average cooperation as the dependent variable and condition as the independent variable. The analysis indicates that this difference is marginally significant (F (208,1) = 3.506, p = .063). This results indicate that there is a trend in which participants in the democratic condition contribute less to the common pool than participants in the dictator condition. To test if leaders show average differences in cooperation over rounds we did a Repeated measures ANOVA with cooperation as the dependent variable, rounds as within subject variable and condition as between subjects variable. The averages for cooperation over rounds are plotted in Figure 2. Testing this difference showed a marginally significant main effect of condition on cooperation (F (208,1) = 3.506, p = .063). The linear trend for the interaction between condition and cooperation is not significant (F (208,1) = .008, p = .931). Figure 2 also shows that it is hard to see a clear difference across conditions. Hypothesis 3 is therefore rejected.
Table 1. Mean exploitation, trust and cooperation across conditions.
Dictatorship Democratic
Exploitation 3.47 4.40 2.53 2.05
Trust 7.20 2.86 7.05 2.60
Cooperation 10.1 6.95 11.9 6.38
Figure 2. Average voluntary contributions over rounds across conditions (blue = democratic condition, green = dictatorship).
The classification for the Social Value Orientation is based on the Social Value Angle (SVA). Participants that score higher on this scale are more prosocial, participants that score lower on this scale are more pro-self.
In our sample we found that 2 participants that classified as altruists which means they cared more about the outcomes of the other, than about their own outcome. 1
participant classified as competitive which means this participant was very keen on minimizing the outcome of the other person. 63 participants classified as individualists, which means they focused on maximizing just their own outcome. 144 participants classified as prosocial, which means they focused on maximizing both their own
outcomes as maximizing the outcomes of the other. To capture the amount of selfish and pro-social participants, we collapsed altruist and the prosocials into one category called ‘prosocials’. We did the same with the individualists and the participant who classified as competitive, which were collapsed into a category called ‘selfish. This leaves us with 146 prosocial participants and 64 selfish participants.
We expected leaders that are more exploitive towards their subordinates to be more proself. This means leaders that are more exploitive score lower on the SVA. To test hypothesis 5 we looked at the correlation between the SVA and exploitation. There is a medium significant correlation between Social Value Angle and Exploitation in the public goods game (r = -.245, p = .041). This means that leaders that score lower on the Social Value Angle exploit their subordinates more than leaders that score higher on the Social Value Angle; more selfish leaders exploit their subordinated more. The correlation between the Social Value Angle and exploitation for dictators bears a marginally
significant result (r = -.332, p = .059). This means that dictators that are more pro-self also tend to exploit their subordinates more than dictators that are more prosocial. Hypothesis 5 is therefore confirmed.
Further we expected that leaders that score higher on Machiavellianism to be more exploitive towards their subordinates. To test hypothesis 4 we looked at the correlation between Machiavellianism and exploitation. There is no significant
subordinates (r = -.025, p = .836). Machiavellianism does not predict exploitative behaviors in leaders in a PGG.
What we do see is a trend between Social Value Angle and Machiavellianism scores. Selfish individuals (who score lower on the SVA) score higher on the
Machiavellianism scale (M = 2.391, SD = 1.214) than prosocial participants (who score higher on the SVA; M = 1.781, SD = 1.169). The difference is significant (t (208) = -3.438, p = .001) and so is the correlation between the Social Value Angle and
Machiavellianism (r = -.235, p = .001). The results indicate that participants that score lower on the SVA score higher on Machiavellianism.
Discussion
The aim of this experiment was to investigate the links between leadership arrangements and exploitation in a social dilemma. Further we investigated the effect of leadership on trust and cooperation. A body of research has been done on either leadership and
cooperation (Acemoglu & Robinson, 2005; Bó, Foster & Putterman, 2010; Deaco, 2009; Drazen and Ozbay, 2015, Kim, 2014; Nicholas, 2012; Tyler & Blader, 2003; Yakinthou & Croeser, 2016) or trust and cooperation (Arrow, 1972; Baron et al., 2007; Johnson & Mislin, 2011) in social dilemmas which found that trust and cooperation are both
influenced by leadership arrangements. The downside of former research is that it either failed to manipulate the leadership context by only looking at naturalistic settings (e.g. comparing real life dictatorships with democratic regimes (Barr, 2001; Gerkey, 2013; Henrich, 2000;) hence makes it impossible to capture causal relations between leadership and trust for instance. Other research did manipulate leadership, but did not measure the effects on trust and cooperation (Barr, 2001; Gerkey, 2013; Henrich, 2000; Kim, 2014; Nicholas, 2012; Yakinthou & Croeser, 2016).
In this study we directly manipulated leadership by exposing people to either a democratic or a dictator condition, and thereby could investigate the effects different leadership arrangements on cooperation and trust. Part one of the experiment contained a public goods game in which participants either got to democratically elect their leader or a dictator was installed. The leader decided how much every participant had to contribute to the common pool. Former research seems to suggest that dictators tend to be more exploitative (Acemoglu & Robinson, 2005; Deaco, 2009) compared to leaders that are democratically chosen (Bó et al., 2003; Brandts, Cooper and Weber, 2015; Charness & Rabin, 2002; Drazen and Ozbay 2015). We therefore expected dictators in a public goods game to exploit their subordinates (by making them contribute, but not contributing themselves) more than democratically chosen leaders. The results of this study indicate that dictators on average exploit their subordinates somewhat more than democratically chosen leaders do. This hypothesis was not supported by our data. Our data shows low rates of exploitation and very marginal differences in exploitation across leadership arrangements.
In the second Public Goods game the leadership was removed. Former research provides evidence that a democracy is better for cooperation than a dictatorship
(Acemoglu & Robinson, 2005) and that regardless leadership arrangements, cooperation eventually declines (Rivas & Sutter, 2011). We expected participants who were in the dictatorship condition to be less cooperative than participants in the democratic condition once the leadership was removed. The results however were contrary to our expectations. The analyses indicated that participants in the dictator condition actually contributed more of their MUs than did participants in the democratic condition, although the
differences did not reach statistical significance. There was no clear trend in cooperation in the expected direction: cooperation did not decline significantly. When leadership was removed participants gave across all rounds approximately half of their endowment to the common pool with no clear decline in contribution.
Lastly we expected the participants that experienced the basic disadvantages of a dictatorship to be less trustful towards other participants in a separate Trust Game compared to participants in the democratic condition (Kim, 2014; Yakinthou & Croeser,
2016; Nicholas, 2012; Gerkey, 2013; Henrich, 2000 & Barr, 2001). This means concretely that we expected those participants to transfer less MUs to the other participants in the Trust Game. Participants in the dictatorship condition indeed
transferred on average less MUs to the other participants. However this difference in trust was rather small and not statistically significant.
Based on former research we expected a cluster of personality differences called the Dark Triad plays a role in taking advantage of other people for your own profit (Popper, 2016). Especially Machiavellianism plays an important role in exploitative behavior (Popper, 2016). We expected a significant correlation between
Machiavellianism and exploitative behavior, especially for dictators. The results showed that Machiavellianism and exploitative tendencies were not significantly correlated in our dataset. The correlations between exploitation and this trait are rather and do not predict the extent of exploitative behavior.
Secondly, former research shows the importance of SVO in choices in social contexts (Lange et al, 2007). We expected a significant correlation between exploitation and pro-self-scores on the SVA. Correlations for the Social Value Angle and exploitation reached statistical significance. These results indicate that people who score higher on exploitation tend to be more pro-self. Lastly the correlation between Machiavellianism and the SVA was significant. Hence people who score higher on Machiavellianism tend to be more pro-self.
The main expectation of this research was that leaders in the dictatorship exploit their subordinates. The analyses indicated that the manipulation in leadership failed to reach a statistical difference. Dictators were rather pro-social towards their subordinates and, even though they had the chance to, hesitated to exploit their subordinates
systemically more than democratic leaders did. Why the experiment failed to manipulate the differences in leadership effectively could have different reasons. We noticed that a lot of the participants had difficulty with the instructions. Leaders might have inferred that it was best to contribute their MUs too, because then the common pool would be larger, which would lead to more points. People had to calculate that this is not the case
in the comprehension questions for starting the experiment, but participants still failed to see that exploitation was better for their personal gain. Parallel to this it might be that the manipulation that was not strong enough. We intentionally avoided labels like ‘exploit’ or ‘dictator’ in the instructions, but this might have led to a manipulation that was no strong enough to spark the idea in the minds of leaders that they could exploit their subordinates for their own profit.
Another reason is the large number of people with a prosocial Social Value Orientation. Around two thirds of the participants were prosocial as compared to one third of the participants being pro-self. Because pro-socials prefer equal outcomes when allocating points (Lange et al., 2007) this might have prevented them from exploiting others for their own profit when they had the chance to in a leadership position.
It is hard to draw practical implications from this study. Strictly this study states that dictatorships and democracies do not affect cooperation and trust between
individuals. The results even indicate that a dictatorship influences voluntary cooperation positively. It is hard to generalize the results in this study to real life leadership
arrangements and their effects on cooperation and trust in people. It might be true that a leader that is not democratically elected does not by design be detrimental for
cooperation and trust among individuals. However exploitation seems to be important in this situations. The fact that most dictators in a naturalistic setting exploit their
subordinates when they get the chance may play a role in the detrimental effects of dictatorship on cooperation and trust. It might be that it is not the type of leadership per se that influences trust and cooperation, but the implementation of the leadership style. A leader that is not democratically chosen and does not exploit their subordinates may be equal to a democratically chosen leader when it comes to promoting cooperation and trust.
Another limitation of the study is that participants who may have known each other were in the same condition. We tried to avoid this by recruiting participants from different studies, but we could not prevent this entirely from happening hence it might have been harder for participants to exploit someone they know or might even be friends
with. These limitations may have led to results in which leaders did not exploit other participants whom they know. A third limitation is the fact that we had to work with screens to create our own cubicles. To prevent reciprocal interactions or reputation formation that would go beyond the experiment, it was important that people could not see each other and could not interact. Because the screens could not entirely limit interaction, this might have led to situations in which participants saw each other when they were busy with the experiment. Participants even could have interacted with each other which may have interfered with the results. A large strength of the study is the large sample size and the fact that we did not only use psychology students for our research, but that we recruited participants from all kinds of studies on Leiden University, from example history, language and physics students.
Future research can try to replicate this study, with a stronger manipulation of the exploitation in the first Public Goods Game. In new research leaders could get more clear instructions that exploitative behavior might be beneficial for their own profit. This may cause a shift to more exploitative behaviors and might even influence future cooperation and trust. Once larger differences in exploitation between different leadership
arrangements exist, good causal relationships can be investigated. The existing gap in the literature to capture causal relations of the effect leadership arrangements on cooperation and trust still remains till now and new research could try to fill in the blanks.
References
Acemoglu, D., & Robinson, J., A. (2005). Economic Origins of Dictatorship and
Democracy. Cambridge university press: New York City, NY.
Arrow, K. (1972). Gifts and exchanges. Philosophy and Public Affairs, 1, 343–362.
Baron, R. A., Byrne, D., & Branscombe, N. R. (2007) Mastering Social Psychology. Pearson Education Inc.: Bosten, MA.
Barr, A. 2001. Social dilemmas and shame-based sanctions: experimental results from Zimbabwe. Centre for the Study of African Economies Working Paper Series,
149, 1–22.
Birkás, B., Csathó A., Gács B., & Bereczkei T. (2015). Nothing ventured nothing gained: Strong associations between reward sensitivity and two measures of Machiavellianism. Personality and Individual Differences, Volume 74, 112-11
Bó P. D., Foster, L., & Putterman, L. (2010). Institutions and Behavior: Experimental Evidence on the Effects of Democracy. Am Econ Rev, 100, 2205-2229.
Brankley, A. E., & Andrew, E. (2014). Threat perception: How psychopathy and Machiavellianism relate to social perceptions of competition. Personality and
Individual Differences, 71, 103-107
Buchan, N. R., Croson, R. T. A., & Dawes R. M. (2002). Swift Neighbors and Persistent Strangers: A Cross-Cultural Investigation of Trust and Reciprocity in Social Exchange. American Journal of Sociology, 108, 168-206.
Burks, S. V., Carpenter, J. P., & Verhoogen, E. (2003). Playing both roles in the trust game. Journal of Economic Behavior & Organization., 51, 195-216.
Camprubi, L (2016). Science Policies and Twentieth-Century Dictatorships: Spain, Italy, and Argentina. Volume: 107, 681-682.
Camerer, C., & Weigelt, K. (1988). Experimental tests of a sequential reputation model. Econometrica, 56, 1–36.
Charness, G. and M. Rabin (2002), “Understanding Social Preferences with Simple Tests,” Quarterly Journal of Economics, 117, 817-869.
Chemers, M. M. (2000). Leadership Research and Theory: A Functional Integration.
Group Dynamics: Theory, Research and Practice, 4, 27-43.
Chipkin, I. (2009). Democracy and dictatorship. Social dynamics-a journal of the
centre for african studies university of cape town, 35, 375-393
Corduwener, P. (2016) ‘democracy as a contested concept in post-war western europe: a comparative study of political debates in France, West-Germany, and Italy’, The historical journal, 59, 197–220.
Deacon, R. T. (2009). Public good provision under dictatorship and democracy.
Public Choice, 139, 241-262.
Drazen A., & Ozbay, E. Y. (2015). Does “Being Chosen to Lead” Induce Non-Selfish Behavior? Experimental Evidence on Reciprocity.
Espin, A. M., Hernan-Gonzalez, R., Kujal, P., Rassenti, S., Corgnet, B., Espín, A. M., & Hernán-González, R. (2016). To trust, or not to trust: Cognitive reflection in trust games. Journal of Behavioral and Experimental Economics., 64, 20-27.
Fehr, E., & Fischbacher, U. (2003). The nature of human altruism. Nature, 425, 785-791
Fischbacher, U., Gachter, S. & Fehr, E. 2001 Are people conditionally cooperative? Evidence from a public goods experiment. Econ. Lett. 71, 397–404.
Forsyth, D. R. (2014). Group Dynamics. Wadsworth Cengage Learning: Boston, MA.
Greer, L. L., & van Kleef, G. A. (2010). Equality Versus Differentiation: The Effects of Power Dispersion on Group Interaction. Journal of Applied Psychology, 95, 1032-1044.
Hage, S. R., Ziegler, W., Ackermann, H., & Hage, S. R. (2014). Abstract. The
Behavioral and Brain Sciences an International Journal of Current Research and Theory with Open Peer Commentary, 37, 529-546.
Hardin, G. (1968). The tragedy of the Commons. Science, 162, 1243-1248.
Henrich, J. 2000. Does culture matter in economic behavior? Ultimatum game bargaining among the Machiguenga. American Economic Review 90, 973– 979.
Kelly, K. J. (2012). Political quid pro quo and the impact of perceptions of corruption on democratic behavior. Election Law Journal, 15, 160-174
Kim, H. (2014). Generalised Trust, Institutional Trust and Political Participation. Asian Journal of Social Science., 42, 695-721.
Kocher, M., Martinsson, P., Matzat, D., & Wollbrant, C. (2015). The role of beliefs, trust, and risk in contributions to a public good. Journal of Economic Psychology:
Research in Economic Psychology and Behavioral Economics, 51, 236-244.
Kosfeld M., Heinrichs, M., Zak, P. J., Fischbacher, U., & Fehr, E. (2005). Oxytocin increases trust in humans. Nature, 435, 673-676.
Johnson, N. D., & Mislin, A. A. (2011). Trust Games: A Meta-analyses. Journal of
Economic Psychology 32, 865–889.
Jonason, P. K., Li, N. P., & Teicher, E. A. (2010). Who is James Bond? The Dark Triad as an agentic social style. Individual Differences Research, 8, 111–120
Lange van, P. A. M., Bekkers R., Schuyt, T. N. M., & Van Vugt, M. (2007) From Games to Giving: Social Value Orientation Predicts Donations to Noble Causes.
Loewenstein, G. F., Thompson, L., & Bazerman, M. H. (1989). Social Utility and Decision Making in Interpersonal Contexts. Interpersonal Relations and Group
Processes, 57, 426-441.
Magalhaes, P. (2016). Economic Evaluations, Procedural Fairness, and Satisfaction with Democracy. Political Research Quarterly : Official Journal of Western Political
Science Association, Pacific Northwest Political Science Association, Southern California Political Science Association, Northern California Political Science Association., 69(3), 522-534.
McCain, J. L., Borg, Z. G., Rothenberg A. H., Churillo, K. M., Weiler, P., & Campbell W. K. (2016). Personality and selfies: Narcissism and the Dark Triad.
Computers in Human Behavior, 64, 126-133
McNamara, J., Stephens, J., Dall, P., & Houston, S. (2009). Evolution of trust and trustworthiness: Social awareness favours personality differences. Proceedings,276, 605-613.
Nicholas, A. (2012). Fairness as a constraint on reciprocity: Playing simultaneously as dictator and trustee. The Journal of Socio-economics, 41, 211-221.
Nikiforakis, N., Oechssler, J., & Shad, A. (2014). Hierarchy, coercion, and exploitation: an Experimental analysis. Journal of Economic Behavior &
Organization, 97, 155-168.
Popper, M. (2016). Followership, deity and leadership. The Theory of Social
Behavior, 46, 211-228.
Pollack, B., & Matear A. (1996). Dictatorship, democracy and corruption in Chile.
Crime, Law and Social Change 25, 371–382.
Pruitt, D. G. (1998). Social Conflict. In D. T. Gilbert, S. T. Fiske, & G. Lindzey, The
Rand, D. G., & Nowak, M. A. (2013) Human Cooperation. Trends in Cognitive
Sciences, 17, 413-425.
Robinson, E. W. (2014). The Greek Polis and the Invention of Democracy: A
Politico-cultural Transformation and Its Interpretation. Polis: The Journal for Ancient
Greek Political Thought, 31, 435-442.
Tyler, T. R., & Blader, S. L. (2003). The Group Engagement Model: Procedural Justice, Social Identity and Cooperative Behavior. Personality and Social Psychology
Review, 7, 349-361.
Wall, V. D., Jr., & Nolan, L. L. (1987). Small group conflict: A look at equity satisfaction and styles of conflict management. Small Group Behavior, 18, 188-211.
Webster, G. D., & Jonason, P. K. (2013). Putting the “IRT back in “dirty”: item response theory analyses of the Dark Triad Dirty Dozen –an efficient measure of narcissism, psychopathy and Machiavellianism. Personality and Individual
Differences, 54, 302-306
Yakinthou, C., & Croeser, S. (2016). Transforming Tunisia: Transitional Justice and Internet Governance in a Post-Revolutionary Society. International Journal of
Transitional Justice, 10, 230-249.
