Journal of Conflict Resolution http://jcr.sagepub.com/
Inducing Cooperative Behavior among Proselfs versus Prosocials: The Moderating Role of Incentives and Trust Christophe Boone, Carolyn Declerck and Toko Kiyonari Journal of Conflict Resolution 2010 54: 799 originally published online 15 June 2010 DOI: 10.1177/0022002710372329 The online version of this article can be found at: http://jcr.sagepub.com/content/54/5/799
Published by: http://www.sagepublications.com
On behalf of: Peace Science Society (International)
Additional services and information for Journal of Conflict Resolution can be found at: Email Alerts: http://jcr.sagepub.com/cgi/alerts Subscriptions: http://jcr.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations: http://jcr.sagepub.com/content/54/5/799.refs.html
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
Inducing Cooperative Behavior among Proselfs versus Prosocials: The Moderating Role of Incentives and Trust
Journal of Conflict Resolution 54(5) 799-824 ª The Author(s) 2010 Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/0022002710372329 http://jcr.sagepub.com
Christophe Boone1, Carolyn Declerck1, and Toko Kiyonari2
Abstract This study investigates how an individual’s social value orientation (SVO) interacts with explicit cooperative incentives on one hand, and intrinsic and extraneously induced trust on the other hand, to affect cooperative behavior. In three experiments, subjects (n ¼ 322) played a one-shot prisoner’s dilemma (PD; with weak cooperative incentives) and an assurance game (AG; with strong cooperative incentives) in conditions with or without trust signals. The authors found, as expected, that cooperative behavior is strongly spurred by explicit incentives, but not by trust, among people with a proself value orientation. Conversely, trust is very important to enhance cooperative behavior of participants with a prosocial value orientation, whereas explicit incentives are less important compared to proselfs. The authors conclude that this study reveals two fundamentally different logics of cooperative behavior: one based on extrinsic incentives and the other on trust.
1 Department of Management, Faculty of Applied Economics, University of Antwerp, Antwerpen, Belgium 2 School of Social Informatics, Aoyama Gakuin University, Sagamihara-city, Kanagawa, Japan
Corresponding Author: Carolyn Declerck, Faculty of Applied Economics, University of Antwerp, Prinsstraat 13, 2000 Antwerpen, Belgium Email:
[email protected]
799 Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
800
Journal of Conflict Resolution 54(5)
Keywords cooperation, social dilemma, social value orientation, trust, game theory Understanding how people can be induced to act cooperatively in social dilemmas, in which individual and collective interests conflict, is an important objective in the social sciences. Mapping the conditions that make people refrain from acting selfish for the benefit of all is a prerequisite for solving real-life social dilemma problems (Dawes and Messick 2000). Past experimental research in gametheoretic decision contexts has identified two, fundamentally different, classes of antecedents of cooperative behavior (see reviews in Kollock 1998; Colman 2003). The economics literature has stressed the importance of explicit incentives embedded in the payoff structure of the game (Bornstein 2003). Such ecological information enters the overt strategic calculations of interacting subjects by revealing the strategic opportunities afforded by the situation. In psychology, the focus has generally been on the impact of social information that is ‘‘external’’ to the game but plays a role in tacit reasoning about the trustworthiness of the interacting person (Haley and Fessler 2005; Boone, Declerck, and Suetens 2008). Interestingly, incentives and trust are the major ingredients of the thirty-year-old goal-expectation theory of Pruitt and Kimmel explicating that two conditions have to be met for an individual to cooperate in social dilemmas: (1) she or he has to be motivated to cooperate (‘‘willingness to cooperate’’) and (2) she or he has to expect that partners are also willing to do so (Pruitt and Kimmel 1977; Yamagishi and Sato 1986). However, this literature ignores that people differ systematically in the extent to which they hold self-regarding versus other regarding preferences. These individual differences in the so-called social value orientation (SVO) reflect how people evaluate outcomes for self and others (Messick and McClintock 1968; Kollock 1998; Van Lange 2000). People with a prosocial value orientation, who tend to maximize joint outcomes, are by default willing to cooperate, whereas proselfs tend to pursue their self-interest (Van Lange 1999, 2000). The implication is that the underlying drivers of cooperative behavior are likely to be fundamentally different for prosocials compared to proselfs. Indeed, based on a comprehensive review of the SVO literature, Bogaert, Boone, and Declerck (2008) present a conceptual framework, embedded in the general theory of cooperation of Pruitt and Kimmel, which unravels why and when people with different SVOs come to cooperate. Their major proposition is that two fundamentally different logics of cooperative behavior—one based on extrinsic incentives and the other based on trust—can be discerned depending on a person’s SVO. Because prosocials care for other people’s outcomes, they adopt a more socially oriented type of rationality (Van Lange 2000). Their natural inclination to cooperate makes them vulnerable for being exploited by noncooperative alters. However, prosocials are not ‘‘irrational’’ and will only cooperate when they expect that their cooperative move will be reciprocated (e.g., Van Lange and SeminGoossens 1998; De Cremer and Van Lange 2001). Bogaert, Boone, and Declerck
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
Boone
801
(2008) therefore hypothesize that prosocials will be very sensitive to information and cues that signal trustworthiness and reduce the fear of being exploited. Prosocials will hesitate to cooperate when they lack trust. As a result, (generalized or extraneously induced) trust will be an important moderator of prosocials’ cooperative behavior. Proselfs, in contrast, have pure self-regarding preferences and try to behave according to rational economic principles. Their default is to exploit other parties whenever it pays off to be greedy (Smeesters et al. 2003). However, rational proselfs will cooperate when it is in their best interest to do so. Therefore, Bogaert, Boone, and Declerck specifically hypothesize that explicit incentives embedded in the social interaction are the main moderators of proselfs’ cooperative behavior. The purpose of the current article is to empirically test the major implications of the model of Bogaert, Boone, and Declerck (2008) by studying the coexistence of dual routes toward cooperation, that is, incentive versus trust based. The hypothesis that prosocials need trust whereas proselfs need incentives to cooperate forms the major conclusion of their review, but the literature from which this conclusion is derived provides little direct empirical evidence. Indeed, the interplay between trust, explicit incentives, and SVO has, to our knowledge, never been explicitly studied together. For this purpose, we gathered data from three experiments conducted to investigate the impact of this interplay. Explicit incentives were manipulated by having all participants (n ¼ 322) play a one-shot prisoner’s dilemma (PD) and an assurance game (AG), also known as the coordination game, trust dilemma, or stag hunt. As will be explained in more detail below, the PD game has weak extrinsic incentives to cooperate. In an AG, in contrast, the extrinsic incentive to cooperate is much stronger. We also measured the subjects’ SVO and generalized trust. In addition, in one experiment, we induced extraneous trust by allowing subjects to have prior contact before playing the games—a minimal extraneous cue that is expected to generate a climate of trust (as in Boone, Declerck, and Suetens 2008). Before elaborating the hypotheses, we first present the basic structure of the experimental games used in the current study.
Social Values and the Role of Incentives and Trust Both the PD and AG have been used extensively to study cooperative behavior in the social sciences (Camerer and Knez 2000; Bornstein and Gilula 2003; Skyrms 2004). Figure 1 shows the payoff matrices of the PD and AG we used in the current experiments. The PD is a mixed motive game in which motives of fear and greed tend to pull people toward noncooperative behavior. While mutual cooperation is collectively the most efficient choice (outcome [L,K] in figure 1A), a greedy person can always obtain a better outcome by defecting (choosing S). Anticipating this, a cooperative but fearful person will defect too (choosing P). As a result, defect is the dominant strategy in the PD game and mutual defection (outcome [S,P] in figure 1A) is the only Nash equilibrium. However, when all parties defect, everybody is worse off.
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
802
Journal of Conflict Resolution 54(5)
1A
Partner K
P
Partner
You
Partner
€8
L You €8
€ 12 You
€1 Partner
S
Partner
€1 You
€ 4 You
€ 12
€ 4
1B
K
Partner
Partner
You
L You €7
P Partner
€7
€5 You
€1 Partner
S
Partner
€1 You
€5
€5 You
€5
Figure 1. Payoff matrices of prisoner’s dilemma (1A) and assurance game (1B).
Many game theorists have therefore investigated ways to transform the incentive structure of the game to make cooperation more appealing (reviewed in Bornstein 2003; Skyrms 2004). For example, by decreasing the payoff from defection in a PD to or below the level of the payoff from mutual cooperation, one transforms the PD to an AG. By doing so, the temptation to free ride is eliminated and incentives to cooperate are generated.1 While decision making still yields an uncertain outcome in the AG, the motive of greed is removed and the game becomes one of coordinating on each other’s decision. If one expects the other party to cooperate (defect), the optimal strategy is to cooperate (defect) too. More technically, unlike the PD, the AG has two Nash equilibria. Mutual cooperation (outcome [L,K] in figure 1B) is the so-called payoff dominant equilibrium, while mutual defection the risk dominant (outcome [S,P] in figure 1B). It is only when one does not trust the other party to cooperate that it makes sense
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
Boone
803
to do the same.2 Although cooperation is not at all obvious in the AG with its dual equilibria (Kelley et al. 2003; Skyrms 2004), people generally cooperate more in AG compared to PD games due to the absence of the temptation to free ride in the former (Kollock 1998). Even in PD games, people are much more cooperative than predicted by the self-interest–based Nash equilibrium (Komorita and Parks 1996; Colman 2003). This appears also to be the case in simultaneous and sequential one-shot games where the probability of future exchange is zero, which puzzles many scholars (Gintis et al. 2003). An often cited explanation is that actors subjectively transform the ‘‘given’’ or ‘‘objective’’ payoffs and then select choices based on the transformed dilemma (Kelley and Thibaut 1978). In doing so, people are often biased by heuristics that have been shaped by natural selection to fine tune decision making in the face of uncertainty (Cosmides and Tooby 1992). Many experiments have revealed that heuristic decision making prompts people to cooperate even in oneshot interactions, when there are no apparent benefits associated with it and when explicit information regarding the outcome is lacking (Kiyonari, Tanida, and Yamagishi 2000; Yamagishi et al. 2007; Declerck, Kiyonari, and Boone 2009). The mere perception that one is in a situation involving social exchange is apparently sufficient to trigger heuristics that subjectively transform mixed motive into coordination incentives, facilitating coordination on the cooperative outcome: the PD is then played as if it were an AG.
SVO and Explicit Incentives Simpson (2004, 387) noted that heuristics provide ‘‘a powerful explanation of the process through which PD is transformed into AD but does not explain who transforms PD into Assurance.’’3 He argues that only prosocials have a natural tendency to play the PD as an AG because they lack the urge to free ride. Greed is not a motive for prosocials because they prefer to maximize joint outcomes in the first place. Therefore, prosocials are not expected to perceive any incentive differences between the PD and AG. Proselfs, in contrast, do not subjectively transform the PD into AG. Given their inclination to maximize self-interest, they will therefore be tempted to defect in a PD. To substantiate this claim, Simpson provides the following pieces of evidence with regard to the one-shot PD: (1) prosocials prefer the mutual cooperation outcome, (2) proselfs prefer the outcome in which they defect and alter cooperates, (3) prosocials behave more cooperatively in a simultaneous one-shot game, and (4) second-player prosocials in a sequential PD game reciprocate first-player cooperative behavior much more often than proselfs. In a sequential PD, fear (or lack of trust in alter’s behavior) is eliminated because the decision of the first mover has been revealed to the second mover. Thus, prosocial’s higher level of cooperation in the sequential PD has to be due to their lower level of greed. Simpson (2004) implies from this that a trust-based explanation likewise cannot explain why prosocials cooperate more in a simultaneous one-shot PD, only a greed-based
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
804
Journal of Conflict Resolution 54(5)
explanation can. Put differently, the difference in cooperative behavior between prosocials and proselfs in the simultaneous PD game must be the result of prosocials transforming PD into AG. The evidence Simpson presents, however, is in our view too circumstantial to derive such strong conclusions. First, ruling out trust-based explanations of behavior in simultaneous games necessitates the manipulation of conditions that stimulate trust and/or the measurement of trust. Second, an unequivocal test of the subjective transformation hypothesis requires comparing the cooperative behavior in simultaneous one-shot PD and AG among prosocials and proselfs. If indeed prosocials play the PD as an AG, the difference in their level of cooperation between both games should be small. The objective transformation of incentives does not matter for prosocials because they already subjectively transform the PD game to an AG. In contrast, proselfs, who play PD as PD, will show more cooperation in an AG than a PD. Because cooperation in the AG is a payoff dominant equilibrium, the cooperative outcome is aligned with the proself’s pursuit of self-interest. The literature on SVO provides many examples indicating that alignment of the cooperative goal with self-interest increases cooperation especially among proselfs (De Cremer and Van Vugt 1999; De Cremer et al. 2008). In these studies, goal alignment was achieved by experimentally inducing group identity, which transforms the personal goal to a collective one (Brewer and Kramer 1986). As prosocials already assign significant weight to others’ outcomes by default, externally induced group identity is less likely to affect them (Bogaert, Boone, and Declerck 2008). Furthermore, in two experiments, Simpson (2006) showed that identity-based enhancement of cooperation is more effective in dilemma situations where the incentives to defect stem from greed rather than fear. Because greed is a primary motive for proselfs (but not prosocials), increased group identity is more likely to transform the goal of proselfs compared to prosocials. To sum up, we expect that SVO will moderate the effect of game type on cooperative behavior such that Hypothesis 1: Among proselfs, the likelihood of cooperation will be higher in a simultaneous one-shot AG compared to a PD. Among prosocials, the likelihood of cooperation will not differ between an AG and a PD.
SVO and Trust Although prosocials’ cooperative behavior is expected to be less sensitive to explicit incentives, it is not unconditional. They may prefer equality and joint outcomes but also want to avoid exploitation. Indeed, when confronted with a noncooperative partner, prosocials will reduce their own level of cooperation (Van Lange and SeminGoossens 1998; De Cremer and Van Lange 2001). Similarly, prosocials are not likely to cooperate when they do not expect others to cooperate. Given that they are very
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
Boone
805
vulnerable to being exploited, Bogaert, Boone, and Declerck (2008) argue that prosocials’ cooperative decisions will strongly depend on their judgment of alters’ trustworthiness. The latter is affected by a person’s generalized trust and/or by extraneous signals that generate a climate of trust. Rotter’s (1967, 1971) research clearly revealed stable individual differences in generalized trust with important implications for behavior in a variety of situations. Salient for the current article is that high compared to low trusters regard others to be more trustworthy by default (Kosugi and Yamagishi 1998). Similarly, we expect high trusters to be more confident that the other party will also cooperate in social dilemmas. Consequently, we predict that prosocials will behave more cooperatively when they have a high level of generalized trust, as this reduces their fear of ending in the defect–defect Nash equilibrium in both the PD and AG. This is consistent with the proposition of Acevedo and Krueger (2005, 449) that prosocials may end up defecting, even in an AG, to reflect a ‘‘prudent goal not to be suckered.’’ As proselfs adhere to rational, calculative economic principles and focus on maximizing their own payoffs based on task related information, we do not expect generalized trust to affect their cooperative behavior. Thus, we predict that cooperative behavior will be affected by the interaction between SVO and generalized trust such that Hypothesis 2: Among prosocials, the likelihood of cooperation will be higher when their generalized trust is high. Among proselfs, the likelihood of cooperation will not be affected by their generalized trust. For similar reasons, Bogaert, Boone, and Declerck (2008) hypothesized that extraneous cues signaling trustworthiness of others would be important determinants of the level of cooperation of prosocials. Such cues help to build a climate of trust, which facilitates acts ‘‘that voluntary expose oneself to the greater positive and negative externalities by the action of others’’ (Yamagishi et al. 2005, 277). This is substantiated by much experimental research that indicates more cooperation toward people of flesh and blood than toward computers (Kiesler, Sproull, and Waters 1996), toward friends than toward strangers (Yamagishi and Sato 1986), and to someone named a ‘‘partner’’ compared to an ‘‘opponent’’ (Burnham, McCabe, and Smith 2000). Bohnet and Frey (1999) show that plain identification, without verbal communication, also increases solidarity, trust, and cooperative behavior. Finally, Boone, Declerck, and Suetens (2008) found that minimal prior contact between partners induced trust and increased cooperative behavior in a repeated mixed-motive game that offered no natural synergies. Following Bogaert, Boone, and Declerck (2008), we expect that especially prosocials, with their natural inclination to cooperate, would benefit from trust signals in the form of prior contact. Some research already points in this direction. For example, although information signaling moral and honest partners elicits more cooperation from everybody compared to information about dishonest partners
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
806
Journal of Conflict Resolution 54(5)
(Liebrand et al. 1986; Smeesters et al. 2003), its effect is more pronounced for prosocials (Van Lange and Kuhlman 1994) than for proselfs. The sensitivity of prosocials to trustworthiness in general is also underscored in a study by Stouten, De Cremer, and Van Dijk (2005) who show that prosocials experience a strong emotional reaction when trust is violated in social interactions. Proselfs, however, seem less affected by a breach of trust. To summarize, we expect that minimal prior contact will affect cooperative behavior more among prosocials than proselfs. Trust signals that increase the likelihood of reaching a mutually cooperative outcome should be more important for prosocials who engage in heuristic decision making and less important for proselfs who are calculative with respect to the outcome of the dilemma. Therefore, we suggest Hypothesis 3: Among prosocials, the likelihood of cooperation will be higher in the presence of minimal prior contact compared to no prior contact. Among proselfs, the likelihood of cooperation will not be affected by minimal prior contact.
Method Participants and Settings A total of 376 students (206 females, 170 males, mean age ¼ 20.7, SD ¼ 3.16) participated in one of three experiments conducted at the University of Antwerp. Experiment 1 (n ¼ 131) took place in March 2008, experiment 2 (n ¼ 128) took place in April 2008, and experiment 3 (n ¼ 117) took place in June 2008. All participants were recruited via university Web mail and registered in advance for participation. Participants were explained that the experiment consisted of making decisions in economic games for which they would be remunerated. Monetary incentives were emphasized and full anonymity was guaranteed. They signed an informed consent form before participating.
Design The three experiments had a dual purpose. First, the experiments tested the impact of a number of factors on cooperative decision making. In experiment 1, we tested the effect of the hormone oxytocin (compared to a placebo) on cooperation between anonymous partners. In experiment 2, we tested the effect of oxytocin (versus a placebo) on cooperation when partners had engaged in prior contact with each other. In experiment 3, we tested the effect of the experimentally induced emotion of shame on cooperation between anonymous partners. The results on the effects of oxytocin and shame are to be published elsewhere. We stress that, other than these additional experimental factors, the three experiments were identical. We also control for these additional factors in the statistical analysis. By doing so, the hypotheses presented in this article are tested on data obtained in a wide set of conditions, which adds to the robustness and generalizability of the findings. Second and equally important, the experiments were
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
Boone
807
designed to collect systematic data on SVO and generalized trust to test the hypotheses set forth in the previous section. In all three experiments, participants play three consecutive games, each time with a different, anonymous partner sitting in a different room. The games (a within factor in the design) comprise a simultaneous PD, an AG, and a sequential PD. In each game, a single decision is to be made based on the payoff matrix shown in Figure 1A and B: to choose option L (the cooperative decision) or option S (the defect decision). The equivalent options for the partners are denoted by K and P. In the simultaneous PD and the AG, decisions are made without any knowledge of the partner’s decision. The payoff for these games is calculated based on actual partner matching. For the sequential PD, the partner’s decision is shown beforehand. In contrast to the PD and the AG, no actual matching took place and all participants were informed that their partner had chosen option K (the cooperative option). The purpose of including a sequential PD was to check the validity of SVO and to replicate Simpson’s (2004) findings (see results section). Second-mover cooperative behavior in a sequential PD is a strong indicator of intrinsic cooperative values. Showing that greed is not a motivation for prosocials is of vital importance to our hypothesis that their cooperative decisions are trust based. The sequential PD was always played last, while the order of the PD and the AG was systematically varied.
Procedures The three experiments together comprised a total of twelve experimental sessions, with a minimum of twenty-eight and a maximum of thirty-six participants per session. For each session, participants gathered in a large foyer for check in. Next, they filled out a consent form and a personality questionnaire. SVO was assessed by means of the decomposed measure (Van Lange 2000) in which either a prosocial or a proself choice has to be indicated for nine consecutive decision outcomes. Generalized trust was assessed with a translated version of the Trust questionnaire by Yamagishi (1998). Participants were split into four groups (approximately eight per group) and guided to four different classrooms for the actual interactive games. The games were preceded by a half hour of activity and/or waiting time. This was the only part that differed between the three experiments.4 In none of the experiments were participants in a room allowed to converse with one another. An experimenter who was blind to the purpose of the experiment remained in the room at all times. Experiment 2 included a manipulation of extraneously induced trust, following the procedures of Boone, Declerck, and Suetens (2008). During the waiting time prior to playing the interactive games, participants in these sessions (n ¼ 128) were called one by one out of the experimental room to meet in another group of eight inside a conference room. Thus, each participant met seven other people that were participating in the experiment but were sitting in different rooms. They were asked to introduce themselves by name, state their favorite hobby, and shake hands with everybody. The
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
808
Journal of Conflict Resolution 54(5)
interaction lasted about five minutes. We call this the ‘‘prior contact’’ (PC) group. In all other sessions (comprising experiments 1 and 3), partners remained anonymous and spoke to nobody other than the room supervisor. The instructions of the subsequent games explained that participants were to make a number of decisions for which real money could be earned. In the simultaneous PD and AG games, subjects were pre-matched with a different anonymous partner in another room. The outcome of the decision depended on their decision as well as on the decision of their partner. In the PC condition, the instructions also explained that they were matched with one of the people they had just met but was now sitting in a different room. Note that this PC manipulation does not allow exact identification of the partner, which prevents possible retaliation or reputation effects. To ensure that all participants adequately understood the monetary payoffs of the games, the game instructions also included eight questions that subjects had to answer correctly before continuing with the actual experiment. The answers were checked by the experimenter. The actual decisions were made with paper and pencil in answer booklets. At the end of the experiment, participants filled in a short post-experimental questionnaire while they waited for payoffs to be computed. All participants were paid in truth the actual amount they earned.5 Debriefing occurred by sending each participant an e-mail referring them to a Web site containing a full description of the intent, procedures, and results of the experiment.
Variables The dependent variable is the cooperative decision, which has the value of 1 (if the participant cooperates) or 0 (if the participant defects). The focal independent variables are the within factor game type (AG_PD) which equals 1 for AG and 0 for PD, and the between factors SVO, generalized trust, and prior contact (1 for the subjects that participated in experiment 2 featuring a face-to-face group meeting before interaction, zero for all participants of experiments 1 and 3). SVO was assessed using the nine-item, triple dominance decomposed measure, which distinguishes between a cooperative, individualistic, and competitive orientation. Respondents are classified into one of these three orientations, if they make six of the nine choices consistent with one orientation (e.g., Van Lange et al. 1997; Van Lange 2000). Fifty-four subjects did not make consistent choices and could therefore not be classified. Following standard practice, we dropped these subjects from subsequent analyses (leaving a total of 322 subjects; 119 in experiment 1, 101 in experiment 2, and 102 in experiment 3). The cooperative choices correspond to the prosocial orientation, taking on the value of 1. As is usually done, the individualistic and competitive orientations are combined to form the proself orientation, which takes on the value of 0. Many researchers have confirmed that the reliability and validity of the nine-item decomposed measure are adequate (Au and Kwong 2004; Bogaert, Boone, and Declerck 2008).
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
Boone
809
Generalized trust was assessed with a six-item questionnaire, scored on a seven-point Likert-type scale (Yamagishi 1998). Sample items include ‘‘most people are basically honest’’ and ‘‘most people trust a person if the other person trusts them.’’ Cronbach a for this population equals .722. A principal component analysis confirmed that only one factor with eigenvalue above 1 (explaining 43.3 percent of the variance) could be extracted from the data. We include several control variables: the order in which AG and PD are played (0 if AG is played first, 1 if PD is played first), gender (0 ¼ female; 1 ¼ male), and age. The latter two variables are included because previous studies have reported age (Van Lange et al. 1997) and gender effects (Van Vugt, De Cremer, and Janssen 2007) on cooperative behavior. Finally, the different experiments are accounted for by adding one additional dummy variable for experiment 3. Given that our prior contact variable is in fact an indicator variable for experiment 2 (see above), experiment 1 serves as the comparison.
Analyses To increase the statistical efficiency of our estimates, participants’ decisions in the simultaneous PD and AG games were pooled and analyzed in panel data form treating each individual decision as the unit of analysis. The total number of observations, therefore, equals 644 (322 subjects 2 decisions).6 Pooled data often exhibit autocorrelation as the same subjects are observed several times, in this case twice. In estimating the coefficients of our binary logistic regressions, we therefore used the method of generalized estimating equations (GEE) of Liang and Zeger (1986), which generalizes quasi-likelihood estimation to the panel data context. This method is a very flexible way to deal with autocorrelation resulting from clustered data and is widely used in the social sciences (Zorn 2001). In the current study, GEE corrects the estimates for potential within-subject correlation of errors between the two games. Because the two decisions of each subject cannot be assumed to be independent, we also report robust standard errors of the coefficients using the sandwich estimators developed by Huber (1967) and White (1982). The development of summary goodness-of-fit statistics for such GEE models is problematic as the residuals are correlated (Zorn 2001; Ballinger 2004). Instead, we report Wald chi-square statistics that test the null hypothesis that all regression coefficients are equal to zero. This statistic is, however, not a goodness-of-fit measure, and is, therefore, not suited to compare the goodness of fit of alternative models (Ballinger 2004, 146).
Results Descriptives and Correlations Descriptive statistics and zero-order correlations among the variables under study are reported in table 1 at the subject level (n ¼ 322). The zero-order correlations reveal
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
810
Journal of Conflict Resolution 54(5)
Table 1. Descriptives and Zero-Order Correlationsa
1. 2. 3. 4. 5. 6. 7. 8.
Cooperation in AG Cooperation in PD Cooperation in seq. PD Order Gender Age SVO Generalized trust
Mean
SD
1
2
3
4
5
6
7
.59 .34 .39 .54 .48 20.71 .48 34.16
.49 .47 .49 .50 .50 3.16 .50 6.30
.36* .22 .00 .02 .11* .14* .10
.43* .01 .09 .15* .30* .18*
.05 .03 .21* .39* .20*
.01 .03 .01 .09
.08 .05 .08
.10 .02
.10
a
n ¼ 322, *p < .05 (two-tailed). AG ¼ assurance game; PD ¼ prisoner’s dilemma; SD ¼ standard deviation; SVO ¼ social value orientation.
that SVO, trust, and age are significantly and positively related to cooperative behavior. The order in which the games were played apparently did not affect the participants’ behavior. Finally, the correlations between the independent variables are low, indicating that potential problems of multicollinearity are not present in this study. With respect to the distribution of participants, slightly more than half are female (52 percent). Approximately 48 percent of the participants were classified as prosocials, while 52 percent were proselfs. Among the proselfs, 81 percent were of the individualistic orientation and 19 percent of the competitive orientation. This SVO distribution is close to the ones reported in previous studies (Bogaert, Boone, and Declerck 2008). Across both games, prosocials cooperated significantly more than proselfs (57 percent vs. 36 percent of decisions; w2 ¼ 30.0, df ¼ 1, p < .001). To check the validity of the SVO scale, we analyzed the relationship between SVO and cooperative behavior in the sequential PD. Only 21 percent of the proselfs decided to reciprocate the first-mover cooperative choice, whereas this percentage amounts to 59 percent among the prosocials (w2 ¼ 48.3, df ¼ 1, p < .001). These proportions are very similar to those reported by Simpson (2004) and underscore the intrinsic willingness of prosocials to cooperate. As expected, the proportion of participants that cooperate is much higher in the AG (59 percent) than in the simultaneous PD (34 percent; w2 ¼ 54.5, df ¼ 1, p < .001). In the sequential PD, 39 percent of participants cooperated. This percentage does not differ statistically from cooperation in the simultaneous PD (w2 ¼ 2.16, df ¼ 1, p < .14) but is significantly less than the cooperation rate in the AG (w2 ¼ 23.9, df ¼ 1, p < .001). Manipulating extraneous trust only slightly increased cooperation. In the prior contact condition, 49 percent of decisions were cooperative, compared to 45 percent in the anonymous experiments 1 and 3. This difference is not significant (w2 ¼ .59, df ¼ 1, p < .441).
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
Boone
811
Test of Hypotheses The baseline model 1 of table 2 does, however, reveal that the level of cooperation, controlling for experiment 3, is significantly higher in experiment 2 (where prior contact was introduced) compared to the same experiment without prior contact (experiment 1). Older subjects are also more likely to cooperate. Furthermore, there is a strong effect of extrinsic incentives: subjects cooperate much more in the AG compared to the PD. As expected, prosocials cooperate significantly more than proselfs. In model 2, generalized trust is added to the equation, which appears to have a significant and positive effect on the likelihood of cooperation. Interestingly, adding trust does not change the estimate of SVO, suggesting that the effect of the latter is not the result of prosocials being more trusting (see also Simpson 2004). The large, significantly negative coefficient of the interaction term AG_PD SVO in model 3 is consistent with the moderation predicted in hypothesis 1. To visualize this interaction, we plot the percentage of cooperative decisions in AG and PD for prosocials and proselfs, respectively, in figure 2. This reveals, as expected, that proselfs are extremely sensitive to extrinsic incentives compared to prosocials. In the AG, 52 percent of the proselfs cooperated, while only 20 percent cooperated in the PD (w2 ¼ 36.8, df ¼ 1, p < .0001). Unlike predicted, however, prosocials apparently also increase cooperation in the AG (66 percent) compared to the PD (49 percent), but this effect is much less pronounced (w2 ¼ 9.55, df ¼ 1, p < .002). We speculate that the latter might be due to prosocials’ anticipation of being exploited by greedy alters in the PD, an incentive that is absent in the AG. In model 4, we tested for the interaction between SVO and generalized trust. The coefficient of this product term is positive and significant as expected, indicating that generalized trust spurs cooperative behavior, especially among prosocials. This is in support of hypothesis 2. To visualize this interaction, we plot the percentage of cooperative decisions comparing low-, medium-, and high-trusting proselfs, with the same categories of prosocials. To do this, we split the continuous generalized trust scores into terciles. Cooperation frequencies (averaged over both games) in these terciles appear to differ significantly for prosocials (w2 ¼ 8.1, df ¼ 2, p < .017) but not for proselfs (w2 ¼ 4.6, df ¼ 2, p < .10). Figure 3 shows that the percentage of cooperative decisions (averaged over both games) for prosocials was 46 percent for the low trusters, 61 percent for the medium trusters, and 65 percent for the high trusters. For proselfs, these percentages (again averaged over both games) were 35 percent, 33 percent, and 40 percent for the low, medium, and high trusters, respectively. In appendix A, we decompose the histogram of figure 3 to show the cooperation levels of prosocials and proselfs for PD and AG separately. This reveals the same pattern. It becomes clear that generalized trust increases cooperation among prosocials but not among proselfs. Model 5 supports the interaction proposed in hypothesis 3 showing that the estimate of SVO Prior contact is positive and significant. Figure 4 shows that prior contact indeed positively affects the level of cooperation of prosocials but not of proselfs.
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
a
.26 (.25) .03 (.20) .20 (.20) .07* (.03) 1.12*** (.14) .87*** (.20) .36 (.23) .04** (.02)
94.62*** 644
.29 (.25) .01 (.19) .23 (.19) .08* (.04) 1.10*** (.14) .91*** (.20) .39* (.23)
86.52*** 644
Model 2
93.18*** 644
.25 (.25) .03 (.20) .20 (.20) .07* (.04) 1.50*** (.22) 1.27*** (.25) .36 (.24) .04** (.02) .75** (.29)
Model 3
*p < .05, **p < .01, ***p < .001 (one-tailed), robust standard errors in parenthesis.
Experiment 3 Order Gender Age AG_PD SVO Prior contact Trust AG_PD SVO SVO Trust SVO Prior contact Trust Prior contact SVO Trust Prior contact Wald chi-square N
Model 1
92.77*** 644
.06* (.03)
.26 (.25) .03 (.20) .20 (.20) .07* (.03) 1.12*** (.14) 1.13 (1.03) .40* (.24) .01 (.02)
Model 4
99.03*** 644
.87* (.43)
.29 (.25) .03 (.20) .22 (.20) .07* (.03) 1.13*** (.14) .60** (.24) .02 (.30) .04** (.02)
Model 5
100.33*** 644
.29 (.25) .03 (.20) .22 (.20) .07* (.03) 1.49*** (.22) .87 (1.05) .02 (.31) .01 (.02) .71** (.29) .05* (.03) .84* (.43)
Model 6
103.18*** 644
.30 (.25) .002 (.20) .22 (.20) .07* (.03) 1.49*** (.22) 1.79 (1.28) 1.31 (1.55) .004 (.03) .71** (.29) .08* (.04) 3.93* (2.24) .04 (.04) .09 (.07)
Model 7
Table 2. Logistic Regression Generalized Estimating Equations Estimates of the Impact of Incentives, Social Value Orientation, and Trust on Cooperationa
812 Journal of Conflict Resolution 54(5)
Boone
813
Figure 2. The impact of incentives on cooperation for prosocials and proselfs. AG ¼ assurance game; PD ¼ prisoner’s dilemma. Y axis shows percentage cooperative decisions. Standard errors (SE) of percentages are denoted inside the columns.
For prosocials, 54 percent of decisions are cooperative (averaged over both games) in the anonymous condition, while 67 percent are cooperative in the prior contact condition. For proselfs, the percentage of cooperative decisions (averaged over both games) is 37 percent and 34 percent in the anonymous and prior contact conditions, respectively. The effect of prior contact is significant for prosocials (w2 ¼ 4.92, df ¼ 1, p < .027) but not for proselfs (w2 ¼ .14, df ¼ 1, p < .705).7 Again, the same pattern holds if figure 4 is decomposed to show cooperation levels separately for PD and AG. This is illustrated in appendix B. In model 6, the three two-way interactions are entered simultaneously. This does not affect the focal estimates, rendering confidence in the findings. To check for the possibility that high-trusting prosocials are more susceptible to prior contact than lowtrusting prosocials, we estimated a model including the three-way interaction between SVO, trust, and prior contact (see model 7). This interaction, however, does not appear to be significant.
Discussion and Implications The combined results of three experiments corroborate the conceptual model of Bogaert, Boone, and Declerck (2008), which suggests that there are two routes toward cooperation corresponding to people’s SVO. The first route is consistent with rational economic principles and is mainly applicable to those holding a proself orientation. It requires adding (context-specific) signals that accentuate the win-win nature of
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
814
Journal of Conflict Resolution 54(5)
Figure 3. The impact of generalized trust on cooperation for prosocials and proselfs. Low, medium, and high trusters are determined on the basis of terciles of generalized trust scores. Y axis shows percent cooperative decisions. Standard errors (SE) of percentages are denoted inside the columns.
cooperation, so that the self-interested inclination of proselfs can be aligned with the interest of the larger collective. Indeed, ‘‘goal alignment’’ through extrinsic incentives is an important moderator of proself’s cooperative behavior (e.g., De Cremer et al. 2008). A second way to establish cooperation involves a more socially oriented rationality and appears to be especially effective for prosocials. Their principles and expectations are guided by fairness and equality (Van Lange 2000). However, as they also have a heterogeneous view of the social world, prosocials are fully aware that other people vary in their natural inclinations to cooperate and are alert to the possibility of exploitation (Van Lange 1992). To overcome the fear of exploitation, trust is an important moderator of cooperation for prosocials.8 Our study shows that both intrinsic generalized trust as well as extraneously induced trust signals9 significantly enhance cooperative behavior of prosocials. We believe this study contributes to the extant literature on cooperation in two important ways. The finding that neither economic incentives nor trust is a necessary or sufficient condition for cooperation to occur, because their relative importance depends on a person’s value orientation—and hence on one’s intrinsic motivation to cooperate—is important as it reveals the existence of fundamentally different logics underlying cooperative behavior. It underscores the importance of simultaneously studying the effect of several cooperative antecedents rooted in different disciplines of the social sciences, including explicit incentives, subtle
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
Boone
815
Figure 4. The impact of prior contact on cooperation for prosocials and proselfs. Y axis shows percentage cooperative decisions. Standard errors (SE) of percentages are denoted inside the columns.
contextual cues, and personality. More generally, our study points to the need to develop a more comprehensive picture of human rationality (Barrett and Kurzban 2006), one that incorporates how deep-level, individual differences affect the way in which people process the complex bundle of task and non–task-related cues typically associated with social exchanges. A second major contribution of this study is that it substantiates the existence of stable differences in cooperative types. A number of studies have linked cooperation with existing personality traits (Lu and Argyle 1991; Wilson, Near, and Miller 1998; Kurzban and Houser 2001), while other studies have suggested distinct typologies to account for different cooperative strategies (Kurzban and Houser 2005; Weber and Murnighan 2008). The result of the current studies suggests a novel and promising cooperative typology based on an individual’s ‘‘adjustment behavior.’’ In this respect, there are apparently three types of players: consistent defectors (who defect in PD and AG), consistent cooperators (who cooperate in PD and AG), and adjusters (who defect in PD and cooperate in AG).10 If we link SVO to this typology, it becomes clear that adjusters and consistent defectors are dominated by proselfs (see table 3 showing a cross-tabulation between SVO and behavioral type). In addition, adjustment behavior proves to be based on self-interest. Nearly three quarters of the adjusters defect when they have the chance to reciprocate a cooperative decision as a second player in a sequential PD game (see table 4 showing a cross-tabulation of behavioral type with second-player decision in a sequential PD game). An advantage to using ‘‘adjusting behavior’’ as a typology of cooperative behavior is that it simultaneously takes into
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
816
Journal of Conflict Resolution 54(5)
Table 3. Social Value Orientation by Incentive Adjustment Behaviora,b
Proselfs Prosocials Total
Consistent Cooperators
Consistent Defectors
Adjusters (Defect in PD; Cooperate in AG)
26 (29%) 65 (71%) 91 (100%)
73 (63%) 42 (37%) 115 (100%)
60 (61%) 38 (38%) 98 (100%)
a Cell frequencies are reported (with percentage of column totals in parenthesis). Chi-square (2 df) ¼ 29.42, p < .001. b n ¼ 304 (18 of the 322 subjects who defected in AG and cooperated in PD are excluded from this analysis).
Table 4. Second-Mover Cooperative Choice in Sequential PD Game by Incentive Adjustment Behaviora,b
Second-mover defector Second-mover cooperator Total
Consistent Cooperators
Consistent Defectors
Adjusters (Defect in PD; Cooperate in AG)
24 (27%) 67 (73%) 91 (100%)
88 (77%) 27 (23%) 115 (100%)
73 (74%) 25 (26%) 98 (100%)
a Cell frequencies are reported (with percentage of column totals in parenthesis). Chi-square (2 df) ¼ 64.92, p < .001. b n ¼ 304 (18 of the 322 subjects who defected in AG and cooperated in PD are excluded from this analysis).
account the strategic opportunities of the decision-making environment, as well as the personal cooperative inclination of an individual. Future research could shed more light on this typology. For example, several questions remain with respect to the ‘‘consistent defectors.’’ This type is dominated by proselfs (63 percent, see table 3), raising the question as to why not all proselfs opt to adjust their behavior when the environment offers cooperative advantages. However, there are also a number of prosocials (37 percent) that consistently defect. Probably, these are prosocials with low generalized trust, who do not respond to induced trust signals. They remain fearful of betrayal, even in an AG, so that they have internalized defect as a stable strategy. The consistently defect strategy is more problematic for proselfs, because their motive of greed should not lead them to a defect decision in an AG. To conclude, we believe that the results of this study have important implications for solving real-life social dilemmas and developing communal or corporate policies. The results we have presented here show that trust alone does not suffice for selfinterested people who approach social dilemmas with economically rational principles. Incentives, in contrast, may align a self-interest and a cooperative goal. However, this appears to have considerably less effect on socially motivated people who naturally transform a dilemma to a coordination task. This means that groups of people
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
Boone
817
living or working together must take into account individual differences in SVO to succeed and continue to exist. If group existence depends on cooperative efforts, different tactics will need to be combined to boost performance and avoid free riding. Trust building, through personal contact, is essential for low trusting, cooperative people. However, care needs to be taken not to generalize this principle to the population as a whole, as some individualists may exploit trust to favor themselves. For these self-serving people, trust signals need to be kept minimal while cooperative incentives are made prominent.
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
818
Journal of Conflict Resolution 54(5)
Appendix A
Figure A1. The impact of generalized trust on cooperation for prosocials (above) and proselfs (below). Low, medium, and high trusters are determined on the basis of terciles of generalized trust scores. AG ¼ assurance game; PD ¼ prisoner’s dilemma. Y axis shows percentage cooperative decisions. Standard errors (SE) of percentages are denoted inside the columns
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
Boone
819
Appendix B
Figure B1. The impact of prior contact on cooperation for prosocials (above) and proselfs (below). AG ¼ assurance game; PD ¼ prisoner’s dilemma. Y axis shows percentage cooperative decisions. Standard errors (SE) of percentages are denoted inside the columns.
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
820
Journal of Conflict Resolution 54(5)
Notes 1. We do not mean to imply a one-to-one relationship between trust and incentives, on one hand, and game type, on the other. Specifically, trust is needed in both the AG and the PD to coordinate on the mutually beneficial outcome. The point we try to make is that trust is relatively less important in the AG compared to the PG because the removal of the temptation to free ride in the former strengthens incentives to cooperate. 2. Kelley et al. (2003) describe a second reason for not cooperating in an AG. An individual with a competitive motivation seeking to maximize relative advantage over the other (rather than maximizing the outcome for self) will also defect but for reasons that are unrelated to trust. 3. AD refers to assurance dilemma that we refer to as AG. 4. For experiments 1 and 2, this included the oxytocin or placebo administration and waiting time. In experiment 3, the waiting time was spent writing an essay. 5. Participants earned between 18 and 30 euros for a one hour experiment. 6. As a robustness check, we redid the analyses with respect to hypothesis 3 only using the data of experiments 1 and 2, as both experiments were completely identical except for the prior contact factor (n ¼ 119 and n ¼ 101 in the anonymous and prior contact conditions, respectively). This leaves us with 440 observations (220 2 decisions) instead of 644. These findings are more pronounced but qualitatively identical to the ones we report in the article. 7. The findings with respect to prior contact suggest that our PC manipulation did not trigger self-regarding reputational concerns. If this would have been the case, prior contact would probably enhance cooperation among proselfs and less so among prosocials. As mentioned in the Method section, we took great care in our PC manipulation that exact identification of the partner was not possible precisely to prevent reputational concerns (see also Piazza and Bering 2008). Note also that this specific PC manipulation has been shown to increase trust behavior in a trust game (Boone, Declerck, and Suetens 2008). 8. As mentioned in the results section, we speculate that this same fear of exploitation among prosocials might also explain why they cooperated less in the PD compared to the AG. 9. Although the positive effect of subtle trust cues on prosocial behavior is well documented, the precise underlying mechanism of our prior contact manipulation is not known and remains open to future research. 10. Only eighteen individuals (5 percent) cooperated in the PD and defected in the AG. These were dropped from further analyses.
Authors’ Note Replication files are available at http://jcr.sagepub.com/ Declaration of Conflicting Interests The authors declared no potential conflicts of interests with respect to the authorship and/or publication of this article.
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
Boone
821
Funding The authors disclosed receipt of the following financial support for the research of this article: This research was supported by an NOI grant #1044 from the University of Antwerp awarded to Carolyn Declerck and Christophe Boone. References Acevedo, Melissa, and Joachim I. Krueger. 2005. Evidential reasoning in the Prisoner’s Dilemma. American Journal of Psychology 118:431-57. Au, Wing-Tung, and Jessica Y. Y. Kwong. 2004. Measurement and effects of social value orientation in social dilemmas: A review. In Contemporary psychological research on social dilemmas, ed. R. Suleiman, D. Budescu, I. Fischer, and D. Messick, 71-98. Cambridge: Cambridge University Press. Ballinger, Gary A. 2004. Using generalized estimating equations for longitudinal data analysis. Organizational Research Methods 7:127-50. Barrett, H. Clark, and Robert Kurzban. 2006. Modularity in cognition: Framing the debate. Psychological Review 113:628-47. Boone, Christophe, Carolyn H. Declerck, and Sigrid Suetens. 2008. Subtle social cues, explicit incentives, and cooperation in social dilemmas. Evolution and Human Behavior 29:179-88. Bogaert, Sandy, Christophe Boone, and Carolyn H. Declerck. 2008. Social value orientation and cooperation in social dilemmas: A review and conceptual model. British Journal of Social Psychology 47:452-80. Bohnet, Iris, and Bruno S. Frey. 1999. The sound of silence in Prisoner’s Dilemma and dictator games. Journal of Economic Behavior and Organization 38:43-57. Bornstein, Gary. 2003. Intergroup conflict: Individual, group, and collective interests. Personality and Social Psychology Review 7:129-45. Bornstein, Gary, and Zohar Gilula. 2003. Between-group communication and conflict resolution in assurance and chicken games. The Journal of Conflict Resolution 47:326-39. Brewer, Marilynn B., and Roderick M. Kramer. 1986. Choice behavior in social dilemmas. Effects of social identity, group size, and decision framing. Journal of Personality and Social Psychology 50:543-49. Burnham, Terence, Kevin McCabe, and Vernon L. Smith. 2000. Friend-or-foe intentionality priming in an extensive form trust game. Journal of Economic Behavior and Organization 43:57-73. Camerer, Colin F., and Mark Knez. 2000. Increasing cooperation in Prisoner’s Dilemma by establishing a precedent of efficiency coordination in games. Social Science Working Paper 1080. Division of the Humanities and Social Sciences. California Institute of Technology, Pasadena, California. Colman, Andrew M. 2003. Cooperation, psychological game theory, and limitations of rationality in social interaction. Behavioral and Brain Sciences 26:139-98. Cosmides, Leda, and John Tooby. 1992. Cognitive adaptations for social exchange. In The adapted mind: Evolutionary psychology and the generation of culture, ed. J. H. Barkow, L. Cosmides, and J. Tooby, 163-227. Oxford: Oxford University Press.
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
822
Journal of Conflict Resolution 54(5)
Dawes, Robyn M., and David M. Messick. 2000. Social dilemmas. International Journal of Psychology 35:111-6. Declerck, Carolyn H., Toko Kyonari, and Christophe Boone. 2009. Why do responders reject unequal offers in the Ultimatum Game? An experimental study on the role of perceiving interdependence. Journal of Economic Psychology 30:335-43. De Cremer, David, Daan van Knippenberg, Eric van Dijk, and Esther van Leeuwen. 2008. Cooperating if one’s goals are collective-based. Social identification effects in social dilemmas as a function of goal transformation. Journal of Applied Social Psychology 38:1562-79. De Cremer, David, and Paul A. M. Van Lange. 2001. Why prosocials exhibit greater cooperation than proselfs: The roles of social responsibility and reciprocity. European Journal of Personality 15:S5-18. De Cremer, David, and Mark Van Vugt. 1999. Social identification effects in social dilemmas: A transformation of motives. European Journal of Social Psychology 29:871-93. Gintis, Herbert, Samuel Bowles, Robert Boyd, and Ernst Fehr. 2003. Explaining altruistic behavior in humans. Evolution and Human Behavior 24:153-72. Haley, Kevin J., and Daniel M. T. Fessler. 2005. Nobody’s watching? Subtle cues affect generosity in anonymous economic games. Evolution and Human Behavior 26:245-56. Huber, Peter J. 1967. The behavior of maximum likelihood estimates under non-standard conditions. In Proceedings of the Fifth Berkeley Symposium in Mathematical Statistics, and Probability, ed. L. M. Le Cam and J. Neyman, 221-33. Berkeley: University of California Press. Kelley, Harold H., John G. Holmes, Norbert L. Kerr, Harry T. Reis, Caryl E. Rusbult, and Paul A. M. Van Lange. 2003. An atlas of interpersonal situations, 177-91. Cambridge: Cambridge University Press. Kelley, Harold H., and John W. Thibaut. 1978. Interpersonal relations: A theory of interdependence. New York: Wiley. Kiesler, Sara, Lee Sproull, and Keith Waters. 1996. A prisoner’s dilemma experiment on cooperation with people and human-like computers. Journal of Personality and Social Psychology 29:1123-34. Kiyonari, Toko, Shigehito Tanida, and Toshio Yamagishi. 2000. Social exchange and reciprocity: Confusion or a heuristic? Evolution and Human Behavior 21:411-27. Kollock, Peter. 1998. Social dilemmas: The anatomy of cooperation. Annual Review of Sociology 24:183-214. Komorita, Samuel S., Craig D. and Parks. 1996. Social dilemmas. Boulder, CO: Westview. Kosugi, Motoko, and Toshio Yamagishi. 1998. Generalized trust and judgments of trustworthiness. The Japanese Journal of Psychology 69:349-57. Kurzban, Robert, and Daniel Houser. 2001. Individual differences in cooperation in a circular public goods game. European Journal of Personality 15:S37-52. ———. 2005. Experiments investigating cooperative types in humans: A compliment to evolutionary theory and simulations. Proceedings of the National Academy of Sciences 102:1803-7. Liang, Kung-Yee, and Scott L. Zeger. 1986. Longitudinal data analysis using generalized linear models. Biometrika 73:13-22.
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
Boone
823
Liebrand, Wim B. G., Ronald W. T. L. Jansen, Victor M. Rijken, and Cor J. M. Suhre. 1986. Might over morality: Social values and the perception of other players in experimental games. Journal of Experimental Social Psychology 22:203-15. Lu, Luo, and Michael Argyle 1991. Happiness and cooperation. Personality and Individual Differences 12:1019-30. Messick, David, and Charles G. McClintock. 1968. Motivational bases of choice in experimental games. Journal of Experimental Social Psychology 4:1-25. Piazza, Jared, and Jesse M. Bering. 2008. Concerns about reputation via gossip promote generous allocation in an economic game. Evolution and Human Behavior 29:172-8. Pruitt, Dean G., and Melvin J. Kimmel. 1977. 20 years of experimental gaming—Critique, synthesis, and suggestions for the future. Annual Review of Psychology 28:363-92. Rotter, Julian B. 1967. A new scale for measurement of interpersonal trust. Journal of Personality 35:651-5. ———. 1971. Generalized expectancies for interpersonal trust. American Psychologist 26:44352. Simpson, Brent. 2004. Social values, subjective transformations, and cooperation in social dilemmas. Social Psychology Quarterly 67:385-95. ———. 2006. Social identity and cooperation in social dilemmas. Rationality and Society 18:443-70. Skyrms, Brian 2004. The stag hunt and the evolution of social structure. Cambridge: Cambridge University Press. Smeesters, Dirk, Luk Warlop, Eddy Van Avermaet, Olivier Corneille, and Vincent Yzerbyt. 2003. Do not prime hawks with doves: The interplay of construct activation and consistency of social value orientation on cooperative behavior. Journal of Personality and Social Psychology 84:972-87. Stouten, Jeroen, David De Cremer, and Eric Van Dijk. 2005. All is well that ends well, at least for proselfs: Emotional reactions to equality violation as a function of social value orientation. European Journal of Social Psychology 35:767-83. Van Lange, Paul A. M. 1992. Confidence in expectations. A test of the triangle hypothesis. European Journal of Personality 6:371-9. Van Lange, Paul A. M. 1999. The pursuit of joint outcomes and equality in outcomes: An integrative model of social value orientation. Journal of Personality and Social Psychology 77:337-49. Van Lange, Paul A. M. 2000. Beyond self-interest: A set of propositions relevant to interpersonal orientations. In European review of social psychology, vol. 11, ed. W. Stroebe and M. Hewstone, 297-331. New York: Wiley. Van Lange, Paul A. M., and D. Michael Kuhlman. 1994. Social value orientations and impressions of partner honesty and intelligence: A test of the might versus morality effect. Journal of Personality and Social Psychology 67:126-41. Van Lange, Paul A. M., Wilma Otten, Ellen M. N. De Bruin, and Jeffrey A. Joireman. 1997. Development of prosocial, individualistic, and competitive orientations: Theory and preliminary evidence. Journal of Personality and Social Psychology 73:733-46. Van Lange, Paul A. M., and Astrid Semin-Goossens. 1998. The boundaries of reciprocal cooperation. European Journal of Social Psychology 28:847-54.
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
824
Journal of Conflict Resolution 54(5)
Van Vugt, Mark, David De Cremer, and Dirk P. Janssen. 2007. Gender differences in cooperation and competition. The male warrior hypothesis. Psychological Science 18:19-23. Weber Mark J., and J. Keith Murnighan. 2008. Suckers or saviors? Consistent contributors in social dilemmas. Journal of Personality and Social Psychology 95:1340-53. White, Halbert. 1982. Maximum likelihood estimation of misspecified models. Econometrica 50:1-25. Wilson, David S., David C. Near, and Ralph R. Miller. 1998. Individual differences in Machiavellianism as a mix of cooperative and exploitative strategies. Evolution and Human Behavior 19:203-12. Yamagishi, Toshio. 1998. The structure of trust: The evolutionary game of mind and society. Tokyo: University of Tokyo Press. Yamagishi, Toshio, Satoshi Kanazawa, Rie Mashima, and Shigeru Terai. 2005. Separating trust from cooperation in a dynamic relationship. Rationality and Society 17:275-308. Yamagishi, Toshio, and Kaori Sato. 1986. Motivational bases of the public goods problem. Journal of Personality and Social Psychology 50:67-73. Yamagishi, Toshio, Shigeru Terai, Toko Kiyonari, Nobuhiro Mifune, and Satoshi Kanazawa. 2007. The social exchange heuristic: Managing errors in social exchange. Rationality and Society 19:259-91. Zorn, Christopher J. W. 2001. Generalized estimating equation models for correlated data: A review with applications. American Journal of Political Science 45:470-90.
Downloaded from jcr.sagepub.com at Aoyama Gakuin University on October 10, 2010
