Reward, Punishment, and Cooperation: A Meta-Analysis

Abstract

How effective are rewards (for cooperation) and punishment (for noncooperation) as tools to promote cooperation in social dilemmas or situations when immediate self-interest and longer term collective interest conflict? What variables can promote the impact of these incentives? Although such questions have been examined, social and behavioral scientists provide different answers. To date, there is no theoretical and/or quantitative review of rewards and punishments as incentives for cooperation in social dilemmas. Using a novel interdependence-theoretic framework, we propose that rewards and punishments should both promote cooperation, and we identify 2 variables—cost of incentives and source of incentives—that are predicted to magnify the effectiveness of these incentives in promoting cooperation.A meta-analysis involving 187 effect sizes revealed that rewards and punishments exhibited a statistically equivalent positive effect on cooperation (d =0.51 and 0.70, respectively). The effectiveness of incentives was stronger when the incentives were costly to administer, compared to free. Centralization of incentives did not moderate the effect size. Punishments were also more effective during iterated dilemmas when participants continued to interact in the same group, compared to both (a) iterated dilemmas with reassignment to a new group after each trial and (b) one-shot dilemmas. We also examine several other potential moderators, such as iterations, partner matching, group size, country, and participant payment. We discuss broad conclusions, consider implications for theory, and suggest directions for future research on rewards and punishment in social dilemmas.

Full text

Reward, Punishment, and Cooperation: A Meta-Analysis
Daniel Balliet
Singapore Management University and
VU University Amsterdam
Laetitia B. Mulder
University of Groningen
Paul A. M. Van Lange
VU University Amsterdam
How effective are rewards (for cooperation) and punishment (for noncooperation) as tools to promote
cooperation in social dilemmas or situations when immediate self-interest and longer term collective
interest conflict? What variables can promote the impact of these incentives? Although such questions
have been examined, social and behavioral scientists provide different answers. To date, there is no
theoretical and/or quantitative review of rewards and punishments as incentives for cooperation in social
dilemmas. Using a novel interdependence-theoretic framework, we propose that rewards and punish-
ments should both promote cooperation, and we identify 2 variables—cost of incentives and source of
incentives—that are predicted to magnify the effectiveness of these incentives in promoting cooperation.
A meta-analysis involving 187 effect sizes revealed that rewards and punishments exhibited a statistically
equivalent positive effect on cooperation (d⫽0.51 and 0.70, respectively). The effectiveness of
incentives was stronger when the incentives were costly to administer, compared to free. Centralization
of incentives did not moderate the effect size. Punishments were also more effective during iterated
dilemmas when participants continued to interact in the same group, compared to both (a) iterated
dilemmas with reassignment to a new group after each trial and (b) one-shot dilemmas. We also examine
several other potential moderators, such as iterations, partner matching, group size, country, and
participant payment. We discuss broad conclusions, consider implications for theory, and suggest
directions for future research on rewards and punishment in social dilemmas.
Keywords: punishment, reward, cooperation, social dilemma, meta-analysis
Good and evil, reward and punishment, are the only motives to a
rational creature: these are the spur and reins whereby all mankind are
set on work, and guided.
—John Locke, Some Thoughts Concerning Education
If people are good only because they fear punishment, and hope for
reward, then we are a sorry lot indeed.
—Albert Einstein, quoted in All the Questions You Ever
Wanted to Ask American Atheists
Reward and punishment are incentives that tend to capture strong
views of human nature, as well as beliefs regarding public policy,
political structures, and organizational systems. Some, like John
Locke, believe that incentives are effective tools that help regulate
individuals in their pursuit of self-interest. Others, like Albert Ein-
stein, believe that incentives may undermine autonomy, authenticity,
and, most importantly, the true motive to be good. One of the
strongest views was expressed by Thomas Hobbes (1651), who ar-
gued that people who want collective interest to triumph over self-
interest should support Leviathan, an authority or government that
enforces social order. More recently, Hardin (1968) noted that coer-
cion may be the most effective tool to encourage people to sacrifice
self-interest for collective benefit. These ideas ignited interdisciplin-
ary research across the biological and social sciences on the effect of
incentives as a solution to cooperation (Edney & Harper, 1978; Fehr
&Ga¨chter, 2000; Lynn & Oldenquist, 1986; Ostrom, Walker, &
Gardner, 1992; Sigmund, 2007; Yamagishi, 1986). This research has
generally supported the position of Hobbes and Hardin: Incentives for
cooperation do encourage people to sacrifice their self-interest for
collective benefit (e.g., Ga¨chter, Renner, & Sefton, 2008; Rand,
Dreber, Ellingsen, Fudenberg, & Nowak, 2009).
Aligned with Albert Einstein’s view, however, other perspec-
tives suggest that incentives may undermine cooperation.
1
For
1
In this article, we use the term incentives to refer to both rewards and
punishments of behavior in social dilemmas. Moreover, we use the term
effectiveness of incentives to refer to the positive effect of incentives on
cooperation in social dilemmas. We are not referring to the overall effi-
ciency of incentives as a solution to social dilemmas. Although incentives
may be effective at enhancing cooperation in social dilemmas, incentives
could still remain a very inefficient solution to a social dilemma due to the
cost of monitoring behavior and providing the incentive.
This article was published Online First May 16, 2011.
Daniel Balliet, School of Social Sciences, Singapore Management Uni-
versity, Singapore, and Department of Social and Organizational Psychol-
ogy, VU University Amsterdam, Amsterdam, the Netherlands; Laetitia B.
Mulder, Department of Human Resource Management and Organizational
Behavior, University of Groningen, Groningen, the Netherlands; Paul
A. M. Van Lange, Department of Social and Organizational Psychology,
VU University Amsterdam, Amsterdam, the Netherlands.
This research was supported by Singapore Management University
Faculty Research Grant 09-C242-SMU-009.
Correspondence concerning this article should be addressed to Daniel
Balliet, Department of Social and Organizational Psychology, VU Univer-
sity, Van der Boechorststraat 1, 1081 BT, Amsterdam, the Netherlands.
E-mail: dp.balliet@psy.vu.nl
Psychological Bulletin © 2011 American Psychological Association
2011, Vol. 137, No. 4, 594– 615 0033-2909/11/$12.00 DOI: 10.1037/a0023489
594

example, incentives may undermine autonomy and the intrinsic mo-
tivation to cooperate (Deci & Ryan, 2000; Ryan & Deci, 2000), which
can reduce persistence in cooperation (Cardenas, Stranlund, & Willis,
2002; Lepper & Greene, 1978), result in resistance to comply with the
external forces (Brehm, 1966), and/or influence a decline in cooper-
ation if the incentive is no longer present (Deci, Koestner, & Ryan,
1999). Social incentives can also evoke costly retaliation (Denant-
Boemont, Masclet, & Noussair, 2007; Hopfensitz & Rueben, 2009;
Oliver, 1980), transform an ethical decision into a business decision
(Gneezy & Rustichini, 2000; Tenbrunsel & Messick, 1999), and even
undermine trust in others (Chen, Pillutla, & Yao, 2009; Mulder, van
Dijk, De Cremer, & Wilke, 2006a), and thus create their own neces-
sity (Mulder, van Dijk, Wilke, & De Cremer, 2005). Although these
disadvantages of incentives may not be a threat to their effectiveness
as long as they are strong and inescapable (Tenbrunsel & Messick,
1999), it is often unfeasible and/or undesirable to install strong incen-
tives for cooperation and a watertight monitoring system. Whereas
incentives can exert negative psychological effects (e.g., reactance, a
lack of intrinsic obligation and distrust), one cannot expect an incen-
tive system to be an all-encompassing and long-term solution to social
dilemmas. Instead, one should understand under what conditions
incentives are most effective.
The major purpose of the present meta-analytic review is to
provide a comprehensive analysis of three broad questions. First,
do incentives to cooperate promote and sustain cooperation in
small group social dilemmas? Second, what variables might influ-
ence the effectiveness of incentives? Finally, do reward and pun-
ishment differ in their ability to promote and sustain cooperation?
As we discuss shortly, we adopt an interdependence-theoretical
analysis for understanding whether incentives might promote co-
operation and when these incentives might be especially effective.
Social Dilemmas and the Free-Rider Problem
Social dilemmas are situations of interdependence characterized by
a conflict between immediate self-interest and longer term collective
interest (Dawes, 1980; Van Lange & Joireman, 2008). Many social
dilemmas take the form of either a public goods dilemma or a
resource dilemma. In a public goods dilemma, people decide how
much to contribute to the installment, maintenance, or improvement
of a public good, such as paying for public transportation, contributing
to a group project, or engaging in teamwork (Ga¨chter & Herrmann,
2009). Similarly, in a resource dilemma, people decide how much to
take from a shared resource. For example, fishermen may decide how
many fish to catch in an area where certain fish are endangered,
knowing that, over time, overharvesting will lead to depletion of that
particular fish (Komorita & Parks, 1995).
One major challenge is the free-rider problem. In social dilem-
mas, it is individually tempting to contribute as little as possible to
public goods or take as much as possible in resource dilemmas,
while enjoying the benefits of others’ cooperation, such as the
provision of a public good or access to a sustainable resource. To
reduce free riding and encourage cooperation, authorities often
provide rewards (e.g., public recognition) and/or punishments
(e.g., fines). Additionally, groups and collectives often develop
social norms that are informally enforced to regulate individual
behavior and promote cooperation. How can one best understand
the effect of these incentives on cooperation in social dilemmas?
An Interdependence Analysis of Rewards and
Punishment in Social Dilemmas
Since Skinner (1953), research has shown that reinforcements and
punishments are important forms of situational feedback that promote
learning and performance (for an overview, see Kazdin, 2001). Per-
formance and learning are outcomes that are relevant to the individ-
ual, and they typically serve one’s own self-interest in the shorter
term, in the case of performance (e.g., receiving a high grade), or
longer term, in the case of learning (e.g., mastering a particular skill).
In contrast, incentives in the context of social dilemmas are concep-
tually distinct, in that they serve as motivators of cooperation and
include both (a) a cost to oneself and (b) a benefit to all other members
of the group. Moreover, the benefit for others, in absolute terms, is
greater than the cost to oneself. Thus, in social dilemmas, incentives
seek to motivate people to act against their immediate self-interest so
as to serve the collective interest.
Classic and contemporary formulations of interdependence theory
provide a conceptual framework that acknowledges the interdepen-
dence between the self and others, as well as a domain of situations in
which self-interest and collective interest conflict (Kelley et al., 2003;
Kelley & Thibaut, 1978; Rusbult & Van Lange, 2003). More gener-
ally, interdependence theory is a broad theory of the structure of
interpersonal situations and how these situations afford the expression
of certain motives among individuals. From this perspective, we
propose two broad principles that are useful for understanding the
effectiveness of incentives in social dilemmas.
A first principle focuses on the interpersonal motives of the self.
People may be strongly focused on enhancing immediate outcomes
for self (direct self-interest) but may also take into account broader
considerations, such as the goal to minimize (absolute) differences in
outcomes for self and other (egalitarianism or fairness), the goal to
enhance outcomes for the collective, or the goal to enhance outcomes
in the future. The idea that these broader considerations (referred to as
transformations by Kelley et al., 2003) can shape behavior and
interactions has received strong support in past research (for an
overview, see Van Lange, De Cremer, van Dijk, & Van Vugt, 2007).
Specifically, this first principle, which we term the given matrix
thesis, states that an individual’s initial preferences are shaped by the
direct concrete outcomes of an interaction for the self (e.g., the
monetary payoffs for the self during social dilemmas). Although
motives other than direct self-interest might come into play, interde-
pendence theory suggests that self-interest represents a powerful mo-
tive in interdependence situations. This thesis is shared by other
theories and frameworks and has the important implication that peo-
ple should be responsive to structural changes in interdependence
patterns underlying social dilemmas (Kelley et al., 2003; Olson, 1965;
Yamagishi, 1986, 1988).
2
Importantly, incentives can affect the struc-
ture of outcomes in the given matrix by enhancing the correspondence
of outcomes and reducing the conflict of interests between partici-
pants in the dilemma. That is, when incentives are present, there is less
discrepancy between self-interest and collective interests compared to
2
The notion that self-interest represents a powerful motive is shared by
classic theorizing in the social and behavioral sciences. Also, the assump-
tion that other-regarding motives might also be relevant to explain behavior
in interdependence situations is increasingly acknowledged in several
contemporary models and theories of social preferences and human coop-
eration (e.g., Fehr & Schmidt, 1999; Van Lange, 1999).
595
REWARD, PUNISHMENT, AND COOPERATION

when incentives are absent. Specifically, both rewards and punish-
ments of equal magnitude will result in a similar effect, rendering
cooperation in the self-interest of each individual in the dilemma
(Rapoport & Chammah, 1965; see also Komorita & Parks, 1995).
Indeed, prior research has suggested that reward of cooperation or
punishment of noncooperation results in greater levels of cooperation
in social dilemmas relative to when incentives are absent (Fehr &
Ga¨chter, 2000; Mulder, 2008; Ostrom et al., 1992; Parks, 2000; Rand
et al., 2009; Yamagishi, 1986). Thus, following the given matrix
thesis of interdependence theory, incentives are predicted to increase
cooperation (Hypothesis 1).
Interdependence theory extends many other theories by assum-
ing that self-interest is not the only interpersonal motive that is
relevant to interdependence situations, such as social dilemmas. It
also assumes that people might adopt broader motives, including
other-regarding motives, and that people are oriented toward un-
derstanding others’ behavior in terms of such motives. Thus, a
second principle focuses on the perceived interpersonal motives of
interdependent others. This principle states not only that people
might adopt broader motives, including other-regarding motives in
situations of interdependence, but also that people are oriented
toward understanding others’ behavior in terms of such motives.
For example, people are oriented toward understanding whether
self-interest or other-regarding motives provide the most plausible
account of behavior. To illustrate, strong deviations from self-
interested preferences by Mary that benefit John promote John’s
trust. This trust, in turn, may enhance John’s pro-partner behavior,
which may affect Mary’s trust and pro-partner behavior as well.
Such dynamics explain why trust and commitment in relationships
can be maintained and grow over time (Wieselquist, Rusbult,
Foster, & Agnew, 1999; see also Murray & Holmes, 2009). As we
suggest below, this principle can help identify features of the
situation that can make incentives more or less effective at en-
hancing cooperation.
What Factors Promote the Effectiveness of Reward
and Punishment?
According to an interdependence-theoretical analysis, the per-
ceived motives by others are likely to influence the effectiveness
of incentives. Generally speaking, incentives should be more ef-
fective if the goals of those who administer incentives are per-
ceived as aimed at enhancing collective interest, rather than their
own interests. That is, incentives should be more effective to the
degree that they are perceived as guided by cooperative motives
(e.g., Kelley et al., 2003; Mulder & Nelissen, 2011; Van Lange &
Rusbult, in press). This premise has implications for two variables
and their impact on the incentives–cooperation relationship: (a)
the cost of incentives and (b) the source of incentives.
Cost of Incentives
According to an interdependence perspective, incentives should
be more effective to the extent that incentives are seen as genuine
tools that others use to promote collective interests. When others
are expending greater costs to provide reward (for cooperation)
and deliver punishments (for noncooperation), people are likely to
believe that these others seek to promote cooperation, rather than
their self-interests. In fact, research suggests that people who are
genuinely concerned about collective outcomes tend to sacrifice
self-interest to provide punishment, with the punishment largely
directed toward noncooperators (Egas & Riedl, 2008; Fehr &
Ga¨chter, 2000, 2002; Herrmann, Thöni, & Ga¨chter, 2008; Masclet,
Noussair, Tucker, & Villeval, 2003). That is, such punishment is
costly to the punisher, who is not simply acting in self-interest.
Thus, cooperation should increase as a function of group members’
perceptions of a linkage between the costliness of incentives
provided by others and those others’ cooperative motives. We aim
to provide more conclusive evidence relevant to the prediction that
costly incentives will increase cooperation in social dilemmas,
compared to when incentives are free to administer (Hypothesis 2).
The Source of Incentives
Past research has examined two distinct sources: decentralized
incentives, which are provided by individuals who participate in
the dilemma themselves and have equal roles, versus centralized
incentives, which are provided by some kind of external force,
often an authority figure (e.g., a government, manager, teacher, or
the experimenter). Any difference in perceived cooperative mo-
tives between similar-status ingroup members and a central au-
thority may influence the effectiveness of the incentive system.
In fact, power is often associated with self-interest motives. As
power increases, people have been shown to be more concerned
with their self-interest (Keltner, Langner, & Allison, 2006), feel a
sense of entitlement (De Cremer & van Dijk, 2005, 2008), and tend
to become less cooperative in social dilemmas (De Cremer & van
Dijk, 2005; Samuelson & Allison, 1994; van Dijk & De Cremer,
2006). Moreover, people in a position of (unilateral) power tend to
be trusted less than those who hold equal power and have similar
roles (Kramer, 1999; Muthusamy & White, 2006; Van Vugt,
Hogan, & Kaiser, 2008). People are also more inclined to coop-
erate with ingroup members with similar status (e.g., Kalkhoff &
Barnum, 2000; Oldmeadow, Platow, Foddy, & Anderson, 2003;
Turner, 1991) and those who share similar characteristics (Parks,
Sanna, & Berel, 2001). Additionally, people tend to react more
positively to power when this influence is mutual and not the result
of a single power holder (Carpenter & Matthews, 2009; Muth-
usamy & White, 2006). Therefore, dispensers of incentives who
are not part of the social dilemma may be seen as more strongly
guided by self-interest motives (e.g., Komorita, Sheposh, &
Braver, 1968; Kramer, 1999; Mulder, Verboon, & De Cremer,
2009), fueling the perception that they are administering an incen-
tive for their own personal gains rather than concern for the
collective outcome. Thus, we predict centralized incentive sys-
tems—provided by authority figures and not direct participants—
will be less effective than decentralized incentive systems (Hy-
pothesis 3).
Overview of the Meta-Analysis
We report a meta-analytic review of the effect of both rewards
and punishment on cooperation in experimental social dilemmas.
In so doing, we examine the above-noted predictions derived from
interdependence theory. Specifically, we estimate the overall av-
erage effect of both rewards and punishments on cooperation and
then test whether these relationships are moderated by the cost and
source of incentives. Besides testing the interdependence perspec-
596 BALLIET, MULDER, AND VAN LANGE

tive, we also consider several other perspectives about the differ-
ences between reward and punishment. Specifically, we examine if
punishments are more effective compared to rewards in the short
term, while also considering the possibility that rewards are more
effective than punishments in the long term. Last, we examine the
moderating role of the type of dilemma, number of iterations,
group size, cost-to-fine ratio, participant payment, partner match-
ing, and country of participants.
Method
Search for Studies
We searched several databases for published studies, including
PsycINFO, PsycARTICLES, Econlit, Google Scholar, Social Sci-
ences Citation Index, Web of Science, Worldwide Political Sci-
ence Abstracts, and Dissertations Online. We searched the entire
text of English written journal articles by using the terms punish-
ment, rewards, incentive, reinforcer, and sanction along with
social dilemmas, public goods dilemmas, resource dilemmas, and
voluntary contributions mechanism. We searched the references of
all research and review articles for relevant studies. We also
contacted over 150 researchers who attended the 2007 and 2009
International Conference for Social Dilemmas for published and
unpublished data. We also posted a call for data to the Economic
Science Association methods discussion group (http://
groups.google.com/group/esa-discuss). Finally, we contacted sev-
eral authors who had published papers in the past 5 years with
requests for additional data.
Study Criteria
There were several criteria for the selection of studies. First,
studies had to be conducted on adult participants (ages 18 years
and above). Second, all studies had to examine the effect of either
a reward or punishment on cooperation in a social dilemma. We
included only studies that compared a reward or punishment
condition to a control condition and excluded conditions when
both reward and punishments were used simultaneously. The in-
centives could have been delivered by a fine imposed by the
experimenter, by providing participants an opportunity to sanction
each other during the dilemma, or by a leader in the dilemma. The
social dilemma could be either a public goods or give-some
dilemma, prisoner’s dilemma, resource or take-some dilemma, or
other possible matrix games that included some degree of a con-
flict between individual and collective interests. Specifically, mu-
tual cooperation had to yield higher outcomes than mutual defec-
tion, and according to the individual, defection always yielded a
higher personal outcome than cooperation (Dawes, 1980). We
excluded games that did not fit the strict criteria above, including
ultimatum bargaining games, negotiations, trust games, and dicta-
tor games.
3
We applied these criteria to evaluate potentially rele-
vant studies, which uncovered a total of 103 papers. However,
some papers were excluded because they either did not have a
control condition or failed to report the statistics necessary to
calculate the effect size. This resulted in a total of 76 papers that
contained 187 effect sizes (150 published and 37 unpublished).
Coding Procedure
Cost of incentive: Costly versus free. Many experiments
have participants make a payment to either punish or reward other
group members. However, some experiments do not include a cost
for providing incentives to fellow participants (e.g., voting if a
player should be punished, experimenter provides a sanctioning
system free of charge to the participants). Therefore, we coded
whether the incentive was costly (k⫽116), free (k⫽70), or not
determined (k⫽1).
Source of incentives: Centralized versus decentralized. In
social dilemma experiments with incentives, researchers often
allow participants in the dilemma an option to either reward or
punish their partners. During the first stage of these experiments,
participants make their choices in the dilemma. In the second
stage, participants are provided feedback about the choices of their
partners and given the opportunity to pay or vote toward the end of
rewarding or punishing other members. For example, a participant
might pay $1 to reduce the earnings of another member by $3. In
these experiments, punishment is confidential, and participants are
not informed which of the other partners paid to reduce their own
earnings. In other experiments, however, an incentive system is
implemented by the experimenter that either rewards cooperators
or punishes noncooperators (e.g., the earnings of the least coop-
erative member are reduced by a specific amount). This second
scenario is much like a fine for noncooperation. These studies
often vary in their probability of defection being detected. We
coded if the incentive was decentralized, that is, delivered by the
participants in the social dilemma (k⫽102); centralized, meaning
that an incentive mechanism (e.g., a fine) was imposed by the
experimenter (k⫽83); or not applicable (k⫽2).
Cost-to-fine ratio in punishment studies. In many studies,
punishment varied not only in cost to the punisher but also in the
impact it had on the punished. In some studies, participants paid
one monetary unit to decrease their partner’s earnings by one
monetary unit (k⫽6). However, in other studies, one monetary
unit reduced the partner’s earnings by either two (k⫽11), three
(k⫽59), or four (k⫽5) monetary units. In several other studies,
each monetary unit purchased a 10% reduction in a group mem-
ber’s earnings (k⫽10).
Group size. We coded group size as a continuous variable.
The mode of group size was a four-person group (k⫽88).
Excluding one outlier (a 1,000-person group), the mean of group
3
We did not include these paradigms because they (a) do not share the
same game theoretical or interdependence features (e.g., a dictator game
represents unilateral dependence rather than interdependence), (b) add
qualities that make the game different (e.g., most negotiation games in-
clude an element of coordination, represent incomplete information about
their partner’s preferences, may be time bound, and can be asymmetrical,
in that the roles of the players differ), or (c) make salient psychological
states or orientations that are somewhat different from social dilemmas
(e.g., the trust game calls for trust, less so for cooperation; the ultimatum
bargaining appears to challenge violations of fairness). Furthermore, they
have been primarily used in a two-person format rather than an n-person
format.
597
REWARD, PUNISHMENT, AND COOPERATION

size is a four-person group, and the range is between two- and
16-person groups.
Type of experimental protocol: Partner design versus
stranger design. A common method of punishment protocol
included in this meta-analysis is the paradigm developed by Fehr
and Ga¨chter (2000). Fehr and Ga¨chter had participants play an
iterated four-person public goods dilemma allowing participants
an opportunity to punish their partners in the dilemma after each
trial. Originally, Fehr and Ga¨chter composed two conditions—the
stranger design and the partner design. In the stranger design,
participants play the same dilemma for several trials but are
randomly assigned to a new group after each trial. In the partner
design, however, participants remain anonymous but are not reas-
signed to a new group after each trial and remain in the same group
for the entire experiment. Of the studies that employed the Fehr
and Ga¨chter protocol, we coded if the study used the partner design
(k⫽62) or the stranger design (k⫽32).
One-shot versus iterated dilemma. In social dilemma stud-
ies, participants are allowed to interact only once, or the dilemma
may occur repeatedly for several iterations. The sample of studies
includes both one-shot (k⫽54) and iterated (k⫽133) dilemmas.
We also coded the number of iterations as a continuous variable
(M⫽12, Mdn ⫽10, mode ⫽10).
Type of dilemma. In this analysis, public goods dilemmas
(k⫽139) were the most common dilemma, followed by the
prisoner’s dilemma (k⫽19) and the resource dilemma (k⫽11).
Also, a few effect sizes were from studies that used several of these
dilemmas coded above (k⫽6) or used a different type of social
dilemma (k⫽12). Additionally, the social dilemma paradigms in
this sample of studies involved either decisions about real money
(k⫽168) or decisions in hypothetical scenarios (k⫽19).
Country of participants. The studies included in our analy-
sis were conducted in 27 different countries. Most studies were
conducted in the United States (k⫽63), the Netherlands (k⫽48),
Australia (k⫽13), Switzerland (k⫽13), England (9), Japan (k⫽
7), and Russia (k⫽6). Other countries represented include Aus-
tria, China, Greece, Israel, Saudi Arabia, Turkey, Ukraine, and
several more. The country of each effect size is labeled in Table 1.
Overview of Analysis
We use the dstatistic as the measure of effect size. This value
is the difference between the mean levels of cooperation in the
treatment and control conditions divided by the pooled standard
deviation. This is a common measure of effect size for examining
the effect of a manipulated dichotomous variable (incentives vs.
control) on a continuous dependent variable (degree of coopera-
tion). A positive dvalue indicates greater cooperation in the
reward/punishment condition, relative to a control condition.
When the means and standard deviations were not directly re-
ported, the dvalues were calculated using the sample size along
with the Fscore, tvalue, or chi-square value (see Lipsey &
Wilson, 2001).
Several studies allowed us to code multiple effect sizes. For
example, one study could include a control condition and then
three separate punishment treatments, affording the calculation of
three separate effect sizes. However, the effect sizes are noninde-
pendent because they share several methodological features as
well as the same control condition. Therefore, we applied Cooper’s
(1998) shifting-units-of-analysis approach to handling noninde-
pendent effect sizes when conducting analyses. Using this ap-
proach, we averaged over all the effects abstracted from a single
study. This creates a single effect size for each study with multiple
nonindependent effect sizes. These combined effect sizes are then
used in each analysis.
To begin our analysis, we first estimate the average effect sizes
for both punishment and reward with cooperation using a random-
effects model. A fixed-effects model is inappropriate because we
assume that we do not have the entire population of studies and
that there will be systematic between-study variation. Specifically,
we are assuming that the effect of incentives on cooperation in
social dilemmas will have systematic variation that will be ex-
plained, in part, by the study characteristics we mentioned above
(see also Lipsey & Wilson, 2001). We then assess the amount of
estimated variation in the effect size distribution using several
indicators of heterogeneity (T,T
2
, and I
2
) and homogeneity (Q)of
variance. We then apply a mixed-effects model when examining
moderators because a random-effects model assumes only random
variation in the effect size distribution. However, one limitation of
a mixed-effects model is that it may be too conservative and result
in Type II errors, compared to a fixed-effects model (Lipsey &
Wilson, 2001). We report any discrepancies between fixed- and
mixed-effects analyses. Finally, because several of the moderators
are correlated, we conclude our analyses with a multiple regression
model with several study characteristics predicting the effect size.
Analyses were conducted using the Hedges and Olkin (1985)
method with comprehensive meta-analysis.
Results
The Effects of Punishment Versus Reward on
Cooperation
We begin our analysis by reporting the main effects of both
punishment and rewards on cooperation in social dilemmas and
then comparing the effectiveness between these incentives.
Throughout the Results section, we report the effects of punish-
ment first, and then rewards, because punishments have received
the most attention in prior research and we have obtained a much
larger sample of punishment effect sizes (k⫽154) compared to
reward effect sizes (k⫽33).
Punishment and cooperation. Studies examining the
punishment–cooperation relationship used in the meta-analysis,
including their effect sizes and study characteristics, are listed in
Table 1. Prior to analyses, we created a single averaged effect size
for studies that reported multiple nonindependent effect sizes,
which reduced the sample of effect sizes (k⫽126). As expected,
punishment had a medium-sized, positive effect on cooperation in
social dilemmas (d⫽0.70, 95% confidence interval [CI] lower
limit [LL]⫽0.60, upper limit [UL]⫽0.80, 90% prediction
interval LL ⫽0.18, UL ⫽1.58). The indicators of heterogeneity of
the effect size distribution suggest that there is variation in the true
effect size distribution (T
2
⫽.28, T⫽.53) and that a substantial
amount of this variation can be explained by between-study dif-
ferences (I
2
⫽90.55%). Moreover, the overall punishment–
cooperation effect size distribution contained more variation than
would be expected by chance, Q(125) ⫽1,322, p⬍.001.
598 BALLIET, MULDER, AND VAN LANGE

Table 1
Studies on the Punishment–Cooperation Relationship
Study NCO dLL/UL DV GS IT(#) SD/PM DC/C $/Fr P/NP
Bell, Petersen, & Hautaluoma (1989) 60 US 0.22 ⫺0.30/0.74 RD 3 IT(7) — C Fr P
Bochet, Page, & Putterman (2006) 116 US 0.96 0.57/1.70 PGD 4 IT(10) PM DC $ P
Bornstein & Weisel (2010a) 72 IS 1.42 0.76/2.07 PGD 4 IT(18) PM DC $ P
Sample b 72 IS 0.52 0.03/1.01 PGD 4 IT(18) PM DC $ P
Bornstein & Weisel (2010b) 72 IS 1.65 0.94/2.36 PGD 4 IT(18) PM DC $ P
Sample b 72 IS 0.71 0.20/1.23 PGD 4 IT(18) PM DC $ P
Caldwell (1976) 130 US 0.48 0.13/0.83 PD 5 IT(40) — DC Fr NP
Camera & Casari (2009) 80 US 0.62 ⫺0.24/1.07 PD 2 IT(100) — DC $ P
Carpenter (2007b) 46 US 0.55 0.14/0.97 PGD 7 IT(10) SD C $ P
Sample b 46 US 0.44 0.02/0.85 PGD 7 IT(10) SD C $ P
Sample c 46 US 0.18 ⫺0.23/0.59 PGD 7 IT(10) SD C $ P
Carpenter & Matthews (2009) 100 US 0.59 0.19/0.99 PGD 4 IT(10) PM DC $ P
Carpenter, Matthews, & Ong’ong’a
(2004) 72 US 0.94 0.45/1.44 PGD 4 IT(10) PM DC $ P
Casari & Luini (2009) 60 IT 1.60 0.30/1.27 PGD 5 IT(10) PM DC $ P
Chen, Pillutla, & Yao (2009) Study 2 50 US 0.52 ⫺0.05/1.08 PGD 4 OS — C Fr NP
Cinyabuguma, Page, & Putterman
(2005) 128 US 0.96 0.12/1.80 PGD 16 IT(15) — DC $ P
Dickinson (2001) 80 US 1.26 0.78/1.74 PGD 4 IT(8) — C Fr P
Sample b 80 US 1.29 0.81/1.77 PGD 4 IT(8) — C Fr P
Dreber, Rand, Fudenberg, & Nowak
(2008) 52 US 0.78 0.71/1.45 PD 2 IT(24) — DC $ P
Sample b 52 US 0.37 ⫺0.78/0.98 PD 2 IT(24) — DC $ P
Eek, Loukopoulos, Satoshi, & Ga¨rling
(2002) 208 SE 2.16 1.56/2.75 PD 1,000 OS — C Fr NP
Egas & Riedl (2008) 324 NL 0.08 ⫺0.14/0.31 PGD 3 IT(6) SD DC $ P
Sample b 324 NL 0.50 0.27/0.72 PGD 3 IT(6) SD DC $ P
Sample c 306 NL 0.21 ⫺0.01/0.44 PGD 3 IT(6) SD DC $ P
Sample d 324 NL ⫺0.09 ⫺0.31/0.13 PGD 3 IT(6) SD DC $ P
Etran, Page, & Putterman (2009) 80 US 0.37 0.14/0.59 PGD 4 IT(3) PM DC $ P
Fehr & Ga¨chter (2000) 24 CH 1.35 0.46/2.23 PGD 4 IT(10) SD DC $ P
Sample b 80 CH 1.65 1.14/2.15 PGD 4 IT(10) PM DC $ P
Fehr & Ga¨chter (2002) 236 CH 1.35 1.17/1.53 PGD 4 IT(6) SD DC $ P
Fuster & Meier (2010) 15 US 1.10 0.50/1.74 PGD 4 IT(6) PM DC $ P
Sample b 19 US 0.94 0.40/1.48 PGD 4 IT(6) PM DC $ P
Ga¨chter & Herrmann (2009) 141 CH 0.31 0.14/0.48 PGD 3 OS SD DC $ P
Sample b 102 CH 0.30 0.11/0.50 PGD 3 OS SD DC $ P
Sample c 180 RU ⫺0.05 ⫺0.20/0.09 PGD 3 OS SD DC $ P
Sample d 180 RU ⫺0.08 ⫺0.23/0.06 PGD 3 OS SD DC $ P
Ga¨chter & Herrmann (2011) 205 RU ⫺0.07 ⫺0.21/0.07 PGD 3 OS SD DC $ P
Sample b 105 RU 0.07 ⫺0.13/0.26 PGD 3 OS SD DC $ P
Sample c 143 RU 0.18 0.01/0.36 PGD 3 OS SD DC $ P
Sample d 153 RU ⫺0.17 ⫺0.33/⫺0.01 PGD 3 OS SD DC $ P
Ga¨chter, Renner, & Sefton (2008) 105 CH 0.71 0.31/1.11 PGD 3 IT(10) PM DC $ P
Sample b 102 CH 1.95 1.48/2.42 PGD 3 IT(50) PM DC $ P
Ga¨chter & Thöni (2005) 51 CH 1.85 1.19/2.50 PGD 3 IT(10) SD DC $ P
Sample a 126 CH 0.20 ⫺0.15/0.56 PGD 3 IT(10) SD DC $ P
Sample b 54 CH 0.62 0.07/1.16 PGD 3 IT(10) SD DC $ P
Herrmann, Thöni, & Ga¨chter (2008) 56 US 1.36 1.00/1.72 PGD 4 IT(10) PM DC $ P
Study 2 56 UK 1.38 1.02/1.75 PGD 4 IT(10) PM DC $ P
Study 3 68 DK 1.12 0.81/1.42 PGD 4 IT(10) PM DC $ P
Study 4 60 DE 1.42 0.82/1.47 PGD 4 IT(10) PM DC $ P
Study 5 48 CH 1.87 1.40/2.34 PGD 4 IT(10) PM DC $ P
Study 6 96 CH 1.21 0.95/1.48 PGD 4 IT(10) PM DC $ P
Study 7 68 BY 0.55 0.29/0.80 PGD 4 IT(10) PM DC $ P
Study 8 44 UA 0.08 ⫺0.22/0.38 PGD 4 IT(10) PM DC $ P
Study 9 152 RU 0.39 0.23/0.56 PGD 4 IT(10) PM DC $ P
Study 10 44 GR ⫺0.16 ⫺0.46/0.14 PGD 4 IT(10) PM DC $ P
Study 11 64 TR 0.33 0.08/0.58 PGD 4 IT(10) PM DC $ P
Study 12 48 SA ⫺0.15 ⫺0.43/0.14 PGD 4 IT(10) PM DC $ P
Study 13 52 OM ⫺0.01 ⫺0.28/0.27 PGD 4 IT(10) PM DC $ P
Study 14 84 KR 1.64 1.31/1.97 PGD 4 IT(10) PM DC $ P
Study 15 96 CN 1.23 0.96/1.49 PGD 4 IT(10) PM DC $ P
Study 16 40 AU 2.14 1.58/2.71 PGD 4 IT(10) PM DC $ P
(table continues)
599
REWARD, PUNISHMENT, AND COOPERATION

Table 1 (continued)
Study NCO dLL/UL DV GS IT(#) SD/PM DC/C $/Fr P/NP
Hopfensitz & Rueben (2009) 81 NL 1.08 0.47/1.70 PD 2 OS — DC $ P
Kieruj, Mulder, & Nelissen (2008) 56 NL 2.80 2.07/3.54 PGD 4 IT(6) — C Fr P
Kocher, Martinsoon, & Visser (2008) 120 ZA 0.22 0.03/0.40 PGD 3 OS SD DC $ P
Komorita & Barth (1985) 60 US ⫺0.21 ⫺0.78/0.38 PD 3 IT(24) — C Fr P
Kroll, Cherry, & Shogren (2007) 35 US 0.88 0.49/1.27 PGD 5 IT(10) PM DC $ P
Martichuski & Bell (1991) 48 US 0.66 0.08/1.24 RD 3 IT(15) — C Fr P
McCusker & Carnevale (1995) 124 US 0.82 0.43/1.20 RD 8 IT(12) — C $ P
Study 2 48 US 0.66 0.04/1.27 RD 8 IT(12) — C $ P
Mulder (2005) 154 NL 0.58 0.26/0.90 PGD 4 OS — C Fr P
Mulder (2008) 76 NL 0.47 0.02/0.93 PGD 4 OS — C Fr P
Mulder, van Dijk, De Cremer, & Wilke
(2001) 100 NL 0.45 0.05/0.84 PGD 4 OS — C Fr P
Sample b 38 NL 0.73 0.07/1.38 PGD 4 OS — C Fr P
Mulder, van Dijk, De Cremer, & Wilke
(2002) 106 NL 0.53 0.15/0.92 PGD 4 OS — C Fr P
Mulder, van Dijk, De Cremer, & Wilke
(2003) 126 NL 0.35 ⫺0.01/0.71 PGD 4 OS — C Fr P
Mulder, van Dijk, De Cremer, & Wilke
(2006a) 113 NL 0.90 0.51/1.28 PGD 4 OS — C Fr P
Study 2 159 NL 0.59 0.27/0.91 PGD 4 OS — C Fr P
Mulder, van Dijk, De Cremer, & Wilke
(2006b) 50 NL 0.62 0.05/1.18 PGD 4 OS — C Fr P
Study 2 123 NL 0.69 0.33/1.06 PGD 4 OS — C Fr P
Sample b 80 NL 0.39 ⫺0.05/0.83 PGD 4 OS — C Fr P
Study 3 100 NL 0.53 0.13/0.93 PGD 4 OS — C Fr P
Mulder, van Dijk, Wilke, & De Cremer
(2005) 124 NL 0.05 ⫺0.30/0.41 PGD 4 OS — C Fr P
Mulder & van Prooijen (2006) 154 NL 0.48 0.16/0.80 PGD 4 OS — C Fr P
Myers (2009) 72 US 1.12 0.63/1.62 PGD 4 IT(20) SD DC $ P
Nelissen & Mulder (2010) 56 NL 2.80 2.07/3.54 PGD 4 IT(6) PM DC Fr P
Nikiforakis (2008) 144 AU 1.03 0.67/1.40 PGD 4 IT(10) SD DC $ P
Sample b 144 AU 1.63 1.23/2.04 PGD 4 IT(10) PM DC $ P
Sample c 144 AU 0.21 ⫺0.14/0.56 PGD 4 IT(10) SD DC $ P
Sample d 144 AU 0.51 0.16/0.86 PGD 4 IT(10) PM DC $ P
Nikiforakis & Normann (2008) 48 AU 0.50 ⫺0.07/1.08 PGD 4 IT(10) PM DC $ P
Sample b 48 AU 1.09 0.49/1.70 PGD 4 IT(10) PM DC $ P
Sample c 48 AU 2.07 1.37/2.77 PGD 4 IT(10) PM DC $ P
Sample d 48 AU 2.24 1.52/2.97 PGD 4 IT(10) PM DC $ P
Nikiforakis, Normann, & Wallace
(2010) 44 AU 1.09 0.45/1.72 PGD 4 IT(10) PM DC $ P
Sample b 48 AU 1.20 0.58/1.81 PGD 4 IT(10) PM DC $ P
Sample c 44 AU 1.84 1.14/2.55 PGD 4 IT(10) PM DC $ P
Sample d 48 AU 1.85 1.18/2.53 PGD 4 IT(10) PM DC $ P
O’Gorman, Henrich, & Van Vugt
(2008) 44 UK 0.35 0.05/0.66 PGD 4 IT(6) SD DC $ P
Sample b 44 UK ⫺0.06 ⫺0.36/0.24 PGD 4 IT(6) SD C $ P
O’Gorman & Van Vugt (2010) 48 UK 0.80 0.45/1.13 PGD 4 IT(6) SD DC $ P
Sample b 44 UK 0.38 0.08/0.69 PGD 4 IT(6) SD C $ P
Page, Putterman, & Unel (2005) 128 US 1.29 0.91/1.67 PGD 4 IT(20) PM DC $ P
Patel, Cartwright, & Van Vugt (2010) 66 UK 0.20 ⫺0.39/0.80 PGD 4 IT(10) PM DC $ P
Sample b 36 UK 0.38 ⫺0.45/1.21 PGD 4 IT(10) PM DC $ P
Sample c 72 UK 0.21 ⫺0.36/0.78 PGD 4 IT(10) PM DC $ P
Sample d 36 UK ⫺0.79 ⫺1.71/0.13 PGD 4 IT(10) PM DC $ P
Rand, Dreber, Ellingsen, Fudenberg, &
Nowak (2009) 26 US 1.61 0.71/2.52 PGD 4 IT(50) PM DC $ P
Rapoport & Au (2001) 30 HK 0.58 ⫺0.20/1.37 RD 5 IT(90) — C Fr P
Reuben & Riedl (2009) 39 NL 4.51 3.33/5.70 PGD 3 IT(10) PM DC $ P
Sample b 39 NL 0.90 0.24/1.56 PGD 3 IT(10) PM DC $ P
Sample c 42 NL 2.38 1.59/3.17 PGD 3 IT(10) PM DC $ P
Sample d 39 NL 1.77 1.00/2.54 PGD 3 IT(10) PM DC $ P
Riedl & Ule (2009) 12 NL 0.88 ⫺0.49/2.24 PD 6 IT(60) — DC $ P
Sample b 12 NL 0.81 ⫺0.54/2.15 PD 2 IT(60) — DC $ P
Sample c 13 NL 1.87 ⫺0.17/3.90 PD 2 IT(60) — DC $ P
Sample d 11 NL 0.68 ⫺0.70/2.05 PD 2 IT(60) — DC $ P
Sato (1987) 36 JP 1.97 1.18/2.77 RD 4 IT(12) — C Fr P
Sefton, Shupp, & Walker (2007) 18 US 0.32 ⫺0.15/0.79 PGD 4 IT(10) PM C Fr P
600 BALLIET, MULDER, AND VAN LANGE

This effect size estimate could have been influenced by publi-
cation bias, as the majority of studies included in the analysis were
published studies. To examine the possibility of a publication bias,
we first considered the funnel plot where all studies were plotted
according to their sample size and standard error. In this plot, we
observed a reasonable amount of symmetrical dispersion of the
effect sizes, but to examine this more formally, we used Duval and
Tweedie’s (2000) trim and fill approach. This method examines
the symmetry of the effect sizes in the funnel plot and removes the
most extreme small studies from either side of the plot while
recalculating the average effect size at each iteration until symme-
try is achieved. If there is publication bias, then we can expect that
this approach will remove studies above the estimated overall
effect size and provide an estimate of a lower estimated effect size
without publication bias. Through this approach, we found that
there were no studies that were trimmed above the estimated effect
size. Asymmetry did exist, however, characterized by missing
studies below the estimated effect size. The trim and fill approach
added 11 studies above the estimated effect size and recalculated
an estimated average effect size larger than the original estimated
effect size (d⫽0.81, 95% CI LL ⫽0.69, UL ⫽0.93). Moreover,
we found a statistically significant Egger’s regression intercept,
intercept ⫽3.58, t(124) ⫽6.29, p⬍.001, which indicates bias in
the data. Taken together, these results suggest that the sample did
not contain a publication bias that is usually characterized by
missing null findings. Instead, there seems to be a slight bias of
missing studies with larger effect sizes.
As an additional measure of publication bias, we also estimated
Orwin’s (1983) fail-safe N, a statistic designed to estimate the
number of unpublished findings with null results necessary to
Table 1 (continued)
Study NCO dLL/UL DV GS IT(#) SD/PM DC/C $/Fr P/NP
Sell & Wilson (1999) 64 US 0.70 0.36/1.02 PGD 4 IT(22) — NA NA P
Shaw (1976) 160 US 0.42 0.11/0.73 PD 2 IT(20) — NA Fr P
Shinada & Yamagishi (2007) 106 JP 0.91 0.51/1.31 PD 3 OS — C Fr P
Sample b 103 JP 0.50 0.11/0.89 PD 3 OS — DC $ P
Study 2 96 JP 0.68 0.27/1.09 PD 3 OS — DC $ P
Sample b 96 JP 0.34 ⫺0.06/0.74 PD 3 OS — DC $ P
Sutter, Haigner, & Kocher (2010) 160 AT 1.08 0.74/1.43 PGD 4 IT(10) — DC $ P
Sample b 165 AT 0.75 0.43/1.43 PGD 4 IT(10) — DC $ P
Sutter, Linder, & Platsch (2009) 120 AT 0.43 0.04/0.81 PGD 4 IT(10) — DC $ P
Tan (2008) 48 NL 1.33 0.71/1.96 PGD 4 IT(15) PM DC $ P
Tenbrunsel & Messick (1999) 84 US ⫺0.43 ⫺0.87/0.00 PD 2 OS — C Fr NP
Study 2 56 US ⫺0.67 ⫺1.21/⫺0.13 PGD NA OS — C Fr NP
Study 3 90 US 0.42 ⫺0.10/0.94 PGD NA OS — C Fr NP
Sample b 90 US ⫺0.23 ⫺0.69/0.24 PGD NA OS — C Fr NP
Tyran & Feld (2004) 84 CH 0.00 ⫺0.43/0.43 O 3 OS — DC Fr P
van Prooı´jen, Gallucci, & Toeset
(2008) 80 NL ⫺0.20 ⫺0.64/0.24 O 4 IT(5) — C Fr NP
Sample b 80 NL ⫺0.91 ⫺1.37/⫺0.44 O 4 IT(5) — C Fr NP
Study 2 36 NL 0.20 ⫺0.45/0.86 O 4 IT(5) — C Fr P
Sample b 36 NL ⫺0.78 ⫺1.46/⫺0.10 O 4 IT(5) — C Fr NP
Van Vugt & De Cremer (1999) 90 UK 0.76 0.33/1.19 PGD 6 IT(4) — C Fr P
Walker & Halloran (2004) 48 US ⫺0.04 ⫺0.32/0.25 PGD 4 OS SD DC Fr P
Sample b 48 US 0.11 ⫺0.18/0.39 PGD 4 OS SD DC Fr P
Wit & Wilke (1990) 61 NL 0.52 ⫺0.01/1.06 PGD 10 OS — C Fr NP
Study 2 120 NL 0.08 ⫺0.34/0.51 PGD 10 OS — C Fr NP
Study 3 120 NL 0.00 ⫺0.36/0.36 PGD 10 OS — C Fr NP
Study 4 63 NL 0.61 0.08/1.14 PGD 10 OS — C Fr NP
Xiao & Kunreuther (2010) 63 US 0.31 ⫺0.19/0.81 O 2 IT(10) PM C $ P
Sample b 55 US 0.86 0.30/1.42 O 2 IT(10) PM C $ P
Sample c 47 US 0.47 ⫺0.11/1.05 O 2 IT(10) PM C $ P
Sample d 46 US 1.72 1.04/2.40 O 2 IT(10) PM C $ P
Sample e 63 US 0.47 ⫺0.03/0.97 O 2 IT(10) PM C $ P
Sample f 56 US 1.56 0.95/2.18 O 2 IT(10) PM C $ P
Sample g 47 US 0.34 ⫺0.24/0.91 O 2 IT(10) PM C $ P
Sample h 46 US 1.68 1.01/2.36 O 2 IT(10) PM C $ P
Yamagishi (1986) 128 JP 0.61 0.25/0.98 PGD 4 IT(12) — C $ P
Yamagishi (1988) 196 US 1.16 0.86/1.47 PGD 4 IT(24) — C $ P
Yamagishi (1992) 240 JP 2.14 1.36/2.92 PGD 6 IT(12) — C $ P
Note. Dashes indicate that the particular characteristic is not applicable to that specific study. Study ⫽an independent coded effect size; Sample ⫽a
nonindependent coded effect size; N⫽number of participants included in the effect size estimate; CO ⫽country; AT ⫽Austria; AU ⫽Australia; BY⫽
Belarus; CH ⫽Switzerland; CN ⫽China; DE ⫽Germany; DK ⫽Denmark; GR ⫽Greece; HK ⫽Hong Kong; IS ⫽Israel; IT ⫽Italy; JP ⫽Japan; KR ⫽
Korea; NL ⫽the Netherlands; OM ⫽Oman; RU ⫽Russia; SA ⫽Saudi Arabia; SE ⫽Sweden; TR ⫽Turkey; UA ⫽Ukraine; UK ⫽United Kingdom;
US ⫽United States; ZA ⫽South Africa; d⫽standardized mean difference; LL/UL ⫽95% confidence interval with lower limit/upper limit; DV ⫽
dependent variable; PGD ⫽public goods dilemma; PD ⫽prisoner’s dilemma; RD ⫽resource dilemma; GS ⫽group size; IT(#) ⫽iterated dilemma
(number of iterations); OS ⫽one-shot dilemma; SD ⫽stranger design; PM ⫽partner matching design; DC ⫽decentralized punishment; C ⫽centralized
punishment; $ ⫽punishment is costly; Fr ⫽punishment is free; P ⫽paid for decisions in the dilemma; NP ⫽not paid for decisions in the dilemma.
601
REWARD, PUNISHMENT, AND COOPERATION

reduce the overall effect size to nonsignificance (d⫽0.05). We
found that Orwin’s fail-safe Nis 1,058. According to Hedges and
Olkin (1985), to assure confidence in the results, Orwin’s fail-safe
Nshould be greater than 5 times the number of studies (here, 5 ⫻
126 ⫽630) plus 10 (630 ⫹10 ⫽640). Thus, we can conclude that
the estimated average effect size for the punishment–cooperation
relationship is robust to the presence of unpublished null results.
Rewards and cooperation. Table 2 reports the sample of
studies included in this analysis that examined the rewards–
cooperation association (k⫽33). Rewards had a medium-sized,
positive effect on cooperation in social dilemmas (d⫽0.51, 95%
CI LL ⫽0.31, UL ⫽0.70, 90% prediction interval LL ⫽0.27,
UL ⫽1.29). Examining the heterogeneity of the effect size distri-
bution indicates that there is variation in the true effect size (T
2
⫽
.20, T⫽.45) and that much of this variation can be explained by
between-study differences (I
2
⫽80.81%). The overall effect size
distribution for these studies also contained more variation than
would be expected by chance alone, Q(24) ⫽130.24, p⬍.001.
Considering the possibility of a publication bias, we examined
the funnel plot using Duval and Tweedie’s (2000) trim and fill
approach. This approach did indicate some asymmetry in the effect
size, but this asymmetry was again the result of missing effect
sizes larger than the overall estimated effect size. This approach
did not remove any studies above the estimated average effect size
but did remove four studies below the estimated effect size and
estimated an unbiased average effect size (d⫽0.64, 95% CI LL ⫽
0.43, UL ⫽0.85), which was larger than the original effect size.
Egger’s regression intercept, however, was nonsignificant, inter-
Table 2
Studies on the Rewards–Cooperation Relationship
Study NCO dLL/UL DV GS IT(#) DC/C $/Fr P/NP
Cha (1999) 98 US 1.37 0.93/1.81 RD 8 IT C Fr P
Chen, Pillutla, & Yao (2009) 64 US 0.46 ⫺0.04/0.96 PGD 4 OS C Fr NP
Clark (2002) 60 NZ 0.07 ⫺0.44/0.58 PGD 5 IT(10) C $ P
Dickinson (2001) 80 US 1.04 0.57/1.51 PGD 4 IT(8) C Fr P
Sample b 80 US 0.47 0.02/0.91 PGD 4 IT(8) C Fr P
Komorita & Barth (1985) 60 US 0.11 ⫺0.45/0.67 PD 3 IT(20) C Fr P
Martichuski & Bell (1991) 48 US 0.66 0.08/1.24 RD 3 IT(15) C Fr P
McCusker & Carnevale (1995) 124 US 2.17 1.64/2.69 RD 8 IT(12) C $ P
Study 2 48 US 1.88 1.07/2.67 RD 8 IT(12) C $ P
Mulder, van Dijk, De Cremer, &
Wilke (2001) 100 NL 0.08 ⫺0.31/0.47 PGD 4 OS C Fr P
Sample b 38 NL 0.13 ⫺0.51/0.77 PGD 4 OS C Fr P
Mulder (2008) 74 NL ⫺0.21 ⫺0.66/0.24 PGD 4 OS C Fr P
Mulder, van Dijk, De Cremer, &
Wilke (2001) 100 NL 0.08 ⫺0.31/0.47 PGD 4 OS C Fr P
Sample b 38 NL 0.13 ⫺0.51/0.77 PGD 4 OS C Fr P
Mulder, van Dijk, De Cremer, &
Wilke (2006a) 206 NL 0.61 0.17/1.07 PGD 4 OS C Fr P
Sample b 80 NL 0.02 ⫺0.42/0.45 PGD 4 OS C Fr P
Mulder, van Dijk, Wilke, & De
Cremer (2005) 120 NL 0.15 ⫺0.21/0.51 PGD 4 OS C Fr P
Parks (2000) 124 US 0.64 0.23/1.05 PGD 4 IT(20) C Fr P
Sample b 124 US 1.17 0.72/1.62 PGD 4 IT(20) C Fr P
Sample c 124 US 0.71 0.30/1.21 PGD 4 IT(20) C Fr P
Sample d 124 US 0.40 ⫺0.01/0.80 PGD 4 IT(20) C Fr P
Sample e 124 US 0.29 ⫺0.12/0.69 PGD 4 IT(20) C Fr P
Sample f 124 US 0.00 ⫺0.42/0.42 PGD 4 IT(20) C Fr P
Rand, Dreber, Ellingsen, Fudenberg, &
Nowak (2009) 54 US 1.41 0.56/2.27 PGD 4 IT(50) C $ P
Rapoport & Au (2001) 30 HK 0.20 ⫺0.56/0.96 RD 5 IT(90) C Fr P
Sefton, Shupp, & Walker (2007) 72 US 0.87 ⫺0.10/1.83 PGD 4 IT(10) DC $ P
Sutter, Haigner, & Kocher (2010) 180 AT 0.43 0.14/0.73 PGD 4 IT(10) DC $ P
Sample b 284 AT 0.82 0.56/1.08 PGD 4 IT(10) DC $ P
Sutter, Linder, & Platsch (2009) 92 DE 0.63 0.40/0.85 PD 2 IT(15) DC $ P
Walker & Halloran (2004) 48 US ⫺0.17 ⫺0.45/0.12 PGD 4 OS DC Fr P
Sample b 48 US ⫺0.11 ⫺0.39/0.18 PGD 4 OS DC Fr P
Wit & Wilke (1990) 61 NL 0.52 ⫺0.01/1.06 PGD 10 OS C Fr NP
Study 2 120 NL ⫺0.43 ⫺0.87/0.00 PGD 10 OS C Fr NP
Study 3 120 NL 0.34 ⫺0.03/0.71 PGD 10 OS C Fr NP
Study 4 63 NL 0.47 ⫺0.05/0.99 PGD 10 OS C Fr NP
Note. Study ⫽an independent coded effect size; Sample ⫽a nonindependent coded effect size; N⫽number of participants included in the effect size
estimate; CO ⫽country; AT ⫽Austria; DE ⫽Germany; HK ⫽Hong Kong; NL ⫽the Netherlands; NZ ⫽New Zealand; US ⫽United States; d⫽
standardized mean difference; LL/UL ⫽95% confidence interval with lower limit/upper limit; DV ⫽dependent variable; PGD ⫽public goods dilemma;
PD ⫽prisoner’s dilemma; RD ⫽resource dilemma; GS ⫽group size; IT(#) ⫽iterated dilemma (number of iterations); OS ⫽one-shot dilemma; DC ⫽
decentralized punishment; C ⫽centralized punishment; $ ⫽punishment is costly; Fr ⫽punishment is free; P ⫽paid for decisions in the dilemma; NP ⫽
not paid for decisions in the dilemma.
602 BALLIET, MULDER, AND VAN LANGE

cept ⫽0.03, t(24) ⫽0.02, p⫽.78. However, this later result may
be due to low statistical power. The estimated fail-safe Nwas 258,
indicating that the rewards–cooperation effect size may be robust.
Overall, we conclude that our data did not contain a publication
bias characterized by systematically missing studies below the
average effect size. Instead, there was slight bias in the data to
underestimate the effects of both rewards and punishments on
cooperation.
Punishments versus rewards and cooperation. Although
the punishment effect was slightly stronger than the rewards effect,
comparing these two estimates using a mixed-effects analysis
indicated that the effectiveness of punishments and rewards did not
statistically differ in their impact on cooperation, Q(1) ⫽2.95, p⫽
.09. Previous research has suggested that punishments may be
more effective during one-shot dilemmas, whereas rewards may be
more effective during iterated dilemmas. To examine these per-
spectives, we considered the relative effectiveness of rewards and
punishments during either one-shot or iterated dilemmas. During
one-shot interactions, punishments (d⫽0.32, 95% CI LL ⫽0.20,
UL ⫽0.43) did not significantly differ from rewards (d⫽0.16,
95% CI LL ⫽⫺0.03, UL ⫽0.35), Q(1) ⫽1.92, p⫽.16, even
though the punishment effect was statistically significant and the
reward effect was not. During iterated dilemmas, there was no
difference between punishment (d⫽0.87, 95% CI LL ⫽0.74,
UL ⫽0.99) and rewards (d⫽0.81, 95% CI LL ⫽0.55, UL ⫽
1.07), Q(1) ⫽0.15, p⫽.70.
Moderators
We now consider the possible moderators of the punishment–
cooperation and rewards–cooperation effect size distributions. In
the following analyses, we consider each possible moderator in
turn while examining its impact on each relationship, respectively.
The results of these moderator analyses on the punishment–
cooperation and rewards–cooperation effect sizes are reported in
Tables 3 and 4, respectively.
Cost of incentive: Costly versus free. The punishment–
cooperation association was stronger when punishment was costly
(d⫽0.78, 95% CI LL ⫽0.66, UL ⫽0.91) compared to when it
was free (d⫽0.52, 95% CI LL ⫽0.34, UL ⫽0.70), Q(1) ⫽5.41,
p⫽.02. Similarly, the rewards–cooperation relationship was
stronger when rewarding was costly (d⫽0.97, 95% CI LL ⫽0.57,
UL ⫽1.37) compared to when it was free (d⫽0.32, 95% CI LL ⫽
0.12, UL ⫽0.52), Q(1) ⫽8.07, p⫽.005.
Table 3
Results of the Categorical Univariate Moderator Analyses on the Punishment–Cooperation Relationship
Variable and class
Between-class
effect (Qb)kN dT
95% confidence interval
for d
(lower limit/upper limit)
Costly vs. free 5.41
ⴱ
Costly 86 8,264 0.78 .54 0.66/0.91
Free 39 3,862 0.52 .52 0.34/0.71
Centralized vs. decentralized 4.31
ⴱ
Centralized 40 4,252 0.55 .51 0.38/0.72
Decentralized 82 7,647 0.78 .55 0.65/0.91
Experiment protocol 24.50
ⴱ
Partner matching 48 3,324 1.02 .60 0.83/1.21
Stranger matching 25 3,320 0.37 .40 0.20/0.54
Iterations vs. one-shot 41.79
ⴱ
Iterations with partner matching 74 5,875 0.92 .57 0.78/1.07
Iterations with stranger design 15 1,919 0.62 .50 0.35/0.90
One-shot dilemmas 37 4,396 0.32 .30 0.20/0.43
Type of dilemma 1.54
Public goods dilemma 104 10,147 0.74 .54 0.62/0.85
Resource dilemma 13 1,373 0.54 .48 0.25/0.84
Prisoner’s dilemma 6 346 0.77 .38 0.38/1.17
Participant payment 4.65
ⴱ
Not paid 12 1,219 0.31 .63 ⫺0.07/0.69
Paid 114 10,971 0.74 .52 0.64/0.85
Cost-to-fine ratio 0.61
1 to 1 5 458 0.87 .31 0.53/1.22
1 to 2 7 644 0.76 .17 0.55/0.96
1 to 3 49 3,697 0.84 .60 0.66/1.02
1 to 4 4 405 0.90 .45 0.41/1.39
Country of participants 124.59
ⴱ
Australia 5 568 1.27 .50 0.79/1.75
Israel 4 288 1.03 .43 0.51/1.55
Japan 5 705 1.07 .51 0.56/1.57
The Netherlands 30 2,761 0.77 .58 0.54/1.00
Russia 7 1,023 ⫺0.03 .09 ⫺0.12/0.07
Switzerland 13 1,324 1.00 .62 0.64/1.36
United States 35 3,446 0.62 .41 0.46/0.77
ⴱ
p⬍.05.
603
REWARD, PUNISHMENT, AND COOPERATION

Incentive system: Decentralized versus centralized. The
punishment–cooperation association was significantly stronger in
a decentralized system (d⫽0.78, 95% CI LL ⫽0.65, UL ⫽0.91)
compared to a centralized system (d⫽0.55, 95% CI LL ⫽0.38,
UL ⫽0.72), Q(1) ⫽4.31, p⫽.03. However, regarding the
rewards–cooperation effect size, there was no statistical difference
between rewards being administered by a centralized (d⫽0.50,
95% CI LL ⫽0.23, UL ⫽0.77) or a decentralized system (d⫽
0.52, 95% CI LL ⫽0.24, UL ⫽0.80), Q(1) ⫽0.02, p⫽.91.
Group size. We coded the size of the group facing the
dilemma and use this as continuous measure predicting the effect
size in meta-regression. Group size did not moderate either the
punishment–cooperation association, b⫽⫺.01, Q(1) ⫽0.15, p⫽
.70 or the reward–cooperation association, b⫽.04, Q(1) ⫽0.28,
p⫽.35.
Type of experimental design: Partner versus stranger de-
sign. For this analysis, we included only studies that employed
the Fehr and Ga¨chter (2000) protocol. The punishment–
cooperation association was stronger in the partner design (d⫽
1.02, 95% CI LL ⫽0.83, UL ⫽1.21) than in the stranger design
(d⫽0.37, 95% CI LL ⫽0.20, UL ⫽0.54), Q(1) ⫽25.50, p⬍
.001.
One-shot versus iterated dilemmas. In some experiments on
punishment, researchers have participants play a social dilemma
repeatedly, but during each round, they are randomly assigned
with a new group of participants in the dilemma (the stranger
design). When testing the effect of iterations on the punishment–
cooperation effect size, we considered the difference among three
conditions: (a) iterations with a partner matching design, (b) iter-
ations with a stranger design, and (c) one-shot interactions. How-
ever, for the rewards–cooperation relationship, there were no
studies that combined iterations with a stranger design, and there-
fore, we can compare iterations only of a partner matching design
and one-shot interactions.
The punishment–cooperation association was stronger in the
iterated dilemmas with partner matching (d⫽0.92, 95% CI LL ⫽
0.78, UL ⫽1.06) compared to the iterated dilemmas with a
stranger design (d⫽0.62, 95% CI LL ⫽0.35, UL ⫽0.90) and the
one-shot dilemmas (d⫽0.32, 95% CI LL ⫽0.21, UL ⫽0.42),
Q(2) ⫽44.42, p⬍.02. Similarly, the rewards–cooperation asso-
ciation was stronger during iterated dilemmas (d⫽0.76, 95% CI
LL ⫽0.50, UL ⫽1.03) than in the one-shot dilemmas (d⫽0.16,
95% CI LL ⫽⫺0.03, UL ⫽0.35), Q(1) ⫽13.03, p⬍.001.
We examined whether the incentives–cooperation association
would become stronger with an increase in the number of itera-
tions of the dilemma. In a subsequent analysis, we use the number
of iterations as a continuous variable predicting the effect size in
meta-regression. Iterations were nonsignificantly related to the
effect size distribution for both punishment, b⫽.005, Q(1) ⫽
3.38, p⫽.06, and rewards, b⫽.003, Q(1) ⫽1.10, p⫽.29.
However, this random-effects model may be conservative, espe-
cially with a small sample of studies. Using a fixed-effects model,
there was a significant positive effect of iterations on the rewards–
cooperation effect size, b⫽.009, Q(1) ⫽6.60, p⫽.01.
Type of dilemma. There are three main types of dilemmas
represented in our analysis: public goods dilemmas, prisoner’s
dilemmas, and resource dilemmas. The punishment–cooperation
association did not differ across the type of dilemma, Q(2) ⫽1.52,
p⫽.46. Specifically, this effect size did not statistically differ
between the prisoner’s dilemma (d⫽0.54, 95% CI LL ⫽0.25,
UL ⫽0.84), the public goods dilemma (d⫽0.74, 95% CI LL ⫽
0.62, UL ⫽0.85), and the resource dilemma (d⫽0.77, 95% CI
LL ⫽0.38, UL ⫽1.17).
However, for the rewards–cooperation association, there was a
significant difference between the types of dilemmas, Q(2) ⫽7.80,
p⫽.02. The rewards–cooperation linkage was the strongest in the
resource dilemma (d⫽1.27, 95% CI LL ⫽0.60, UL ⫽1.94)
compared to the public goods dilemma (d⫽0.34, 95% CI LL ⫽
Table 4
Results of the Categorical Univariate Moderator Analyses on the Reward–Cooperation Relationship
Variable and class
Between-class
effect (Qb)kn dT
95% confidence interval
for d
(lower limit/upper limit)
Costly vs. free 8.07
ⴱ
Costly 8 992 0.97 .51 0.57/1.38
Free 18 1,954 0.32 .35 0.12/0.52
Centralized vs. decentralized 0.01
Centralized 19 2,338 0.50 .53 0.23/0.77
Decentralized 7 608 0.52 .30 0.24/0.80
Iterations 15.47
ⴱ
One-shot 12 1,250 0.16 .23 ⫺0.03/0.35
Iterations 14 1,656 0.81 .43 0.55/1.07
Type of dilemma 7.80
ⴱ
Public goods dilemma 19 1,878 0.34 .32 0.17/0.52
Prisoner’s dilemma 1 120 0.11 .00 ⫺0.45/0.67
Resource dilemma 5 828 1.27 .70 0.60/1.93
Participant payment 3.30
†
Not paid 6 898 0.23 .30 ⫺0.08/0.54
Paid 20 1,928 0.59 .56 0.36/0.82
Country 9.78
ⴱ
The Netherlands 10 1,805 0.16 .24 ⫺0.05/0.37
United States 11 958 0.88 .61 0.48/1.28
†
p⫽.07.
ⴱ
p⬍.05.
604 BALLIET, MULDER, AND VAN LANGE

0.17, UL ⫽0.52) and the prisoner’s dilemma (d⫽0.11, 95% CI
LL ⫽⫺0.45, UL ⫽0.67).
Participant payment: Paid versus not paid. Punishment
was more effective when participants were paid for the outcomes
of the dilemma (d⫽0.75, 95% CI LL ⫽0.64, UL ⫽0.85) relative
to when they were making decisions about hypothetical amounts
of money (d⫽0.31, 95% CI LL ⫽⫺0.07, UL ⫽0.69), Q(1) ⫽
4.65, p⫽.03. The rewards–cooperation association was margin-
ally significantly larger when participants were paid for their
decisions (d⫽0.59, 95% CI LL ⫽0.36, UL ⫽0.82) than not paid
for their decisions in the dilemma (d⫽0.23, 95% CI LL ⫽⫺0.08,
UL ⫽0.54), Q(1) ⫽3.30, p⫽.07.
Cost-to-fine ratio in punishment studies. We found that the
cost-to-fine ratio did not moderate the punishment–cooperation
effect size, Q(3) ⫽0.61, p⫽.89. Specifically, we did not observe
a statistical difference in the estimated punishment–cooperation
effect size between conditions when one monetary unit purchased
a reduction in one (d⫽0.87), two (d⫽0.76), three (d⫽0.84), or
four (d⫽0.90) monetary units of the person punished.
Country of participants. When analyzing the effect of pun-
ishment on cooperation across countries we considered those
countries represented by four or more effect sizes. We found that
the impact of punishment on cooperation varied significantly
across countries, Q(7) ⫽124.61, p⬍.001. The punishment–
cooperation effect size was strongest in studies from Australia
(d⫽1.27), Japan (d⫽1.07), Israel (d⫽1.03), Switzerland (d⫽
1.00), the Netherlands (d⫽0.77), and the United States (d⫽0
.62). However, punishment had no effect on cooperation in studies
conducted in Russia (d⫽⫺0.03).
The rewards–cooperation effect size sample contained much
less diversity in countries. We found that rewards were more
effective at enhancing cooperation in the United States (d⫽0.88)
than in the Netherlands (d⫽0.16), Q(1) ⫽9.78, p⫽.002.
Multiple Regression Model
Some moderators are intercorrelated—for example, there was an
exceptionally strong correlation between cost of incentives and
whether the incentive was centralized or decentralized (r⫽.83,
p⬍.001). To examine such overlapping moderators, we con-
ducted a random-effects multiple regression analysis using method
of moments estimations with SPSS macros provided by Lipsey and
Wilson (2001). In this analysis, we combined both punishment and
reward effect sizes as the dependent variable and included seven
predictors of the effect size: (a) type of incentive (punishment vs.
reward), (b) cost of incentive (free vs. costly), (c) source of
incentive (centralized vs. decentralized), (d) number of iterations,
(e) participant payment (paid vs. not paid), (f) group size, and (g)
whether the study was conducted by either an economist or psy-
chologist. The model predicted a significant amount of variance in
the effect size distribution, R
2
⫽.11, Q(7) ⫽22.90, p⫽001.
Supporting earlier analyses, we found that three variables moder-
ated the effect size: cost of incentive (␤⫽.26, p⫽.04), partici-
pant payment (␤⫽.21, p⫽.02), and iterations (␤⫽.12, p⫽.08).
After controlling for the variance explained by these other vari-
ables, source of incentives did not moderate the effect size.
4
Similarly, there was no effect of type of incentive, group size, or
whether studies were conducted by economists versus psycholo-
gists.
Discussion
Using an interdependence-theoretic framework, we proposed
that rewards and punishment should promote cooperation and
identified two variables—cost of incentives and source of incen-
tives—that might influence the effectiveness of these incentives in
promoting cooperative behavior. A meta-analysis involving 187
effect sizes revealed that reward and punishment exhibit a sub-
stantial and statistically equivalent positive effect on cooperation.
5
Thus, the main finding is that incentives reliably enhance cooper-
ative behavior. Furthermore, we found that the effectiveness of
these incentives is stronger when the incentives are costly to
administer. We suggested earlier that the administration of costly
incentives might signal a strong commitment to the goal of pro-
moting cooperation, rather than simply pursuing self-interest. It is
the perceptions of these benevolent interpersonal motives that may
explain why incentives are more effective to the degree that they
are more costly to administer.
We also believed that incentives that were decentralized (i.e.,
provided by actual participants in the cooperative venture) would
induce more cooperation than centralized incentives (i.e., those
provided by external sources). Whereas the meta-analysis provided
support for this hypothesis, after statistically controlling for cost,
centralized and decentralized incentive systems did not signifi-
cantly differ in their impact on cooperation. We also found that
punishments were more effective during iterated dilemmas when
participants continued to interact in the same group, compared to
both iterated dilemmas with reassignment to a new group after
each trial and one-shot dilemmas. The incentive–cooperation ef-
fect size differed across countries but was found to be unrelated to
other study characteristics (e.g., group size or the cost-to-fine
ratio).
In what follows, we evaluate the evidence relevant to the hy-
potheses of interdependence theory, discuss some intriguing issues
raised by the meta-analysis, and suggest directions for future
research. We start with the first question raised in this article.
4
We also conducted a random-effects multiple regression model includ-
ing the cost of incentives and source of incentives as predictors of the effect
size. Again, cost of incentives was the only significant predictor of the
effect size in this model.
5
We are the first, to our knowledge, to conduct a comprehensive
meta-analytic review of the effect of incentives on cooperation. At the
same time, we should acknowledge an earlier meta-analysis, conducted by
Zelmer (2003), which reviewed factors affecting cooperation in public
good settings. This meta-analysis provided more preliminary evidence, in
that the analysis was smaller in scope (27 studies) and included only a few
studies in which punishment was manipulated. Moreover, nearly all studies
included were conducted by behavioral economists, and no moderators of
the effect sizes were examined. The present meta-analysis complements
this past work especially by including a much broader sample of studies
from both economics and psychology that enabled us to test several
theoretically interesting moderators. Moreover, the present meta-analysis
complements recent narrative reviews of incentives and cooperation in
both economics (Chaudhuri, 2011) and psychology (Shinada & Yamagishi,
2008).
605
REWARD, PUNISHMENT, AND COOPERATION

Are Rewards and Punishment Effective
Promoters of Cooperation?
A great deal of research has focused on the role of incentives as
tools for promoting cooperation in social dilemmas. Incentives can
be seen as structural solutions to resolving social dilemmas in that
they affect the interdependence structure of the social dilemma by
reducing the conflicts of interest (e.g., Kollock, 1998; Van Lange
& Joireman, 2008; Yamagishi, 1986, 1988). We note immediately
that some investigators have proposed that incentives may ener-
gize a set of interrelated psychological processes that might reduce
cooperation in social dilemmas, including (a) decreasing intrinsic
motivation for cooperation; (b) undermining feelings of autonomy,
control, and freedom; and (c) fueling psychological reactance.
Thus, a key question drove our investigation: Given that research
is generally supportive of these psychological mechanisms related
to hidden costs of incentives (Lepper & Greene, 1978), can incen-
tives effectively increase cooperation?
In support of Hypothesis 1, the present findings provide un-
equivocal evidence that rewards and punishment are effective
promoters of cooperation in social dilemmas, explaining approx-
imately 3%–12% of the variance in cooperation. We posit that
incentives influence cooperation in a structural manner, such that
the conflicts among group members are reduced. In line with
interdependence theory (and game theory), people are more likely
to cooperate to the degree that the incentives of collective interests
are enhanced and the incentives for self-interest are reduced (An-
denaes, 1971; Bankston & Cramer, 1974; Komorita & Parks, 1995;
Rapoport & Chammah, 1965; Zimring, 1971). As such, the first
contribution of the present meta-analysis is that, although incen-
tives might undermine intrinsic motivation and even cooperation
in several important ways (e.g., Lepper & Greene, 1978), the costs
of these undermining mechanisms are outweighed by incentives
reducing the conflicts of interest. In other words, reward and
punishments effectively increase cooperation.
What Factors Promote the Effectiveness of Reward
and Punishment?
Interdependence theory is highly useful in understanding how
social circumstances might magnify the effectiveness of incen-
tives. In particular, interdependence theory emphasizes the inter-
personal motives of a person as well as the interpersonal motives
that a person perceives in others. Indeed, we have suggested that
rewards and punishment are effective when they are associated
with the pursuit of collective interest, rather than self-interest. In
fact, the decision to provide incentives can be viewed as a second-
order social dilemma: People in the dilemma can either free ride on
others’ provision of incentives for cooperation or sacrifice self-
interest to provide incentives for collective benefit. Indeed, mod-
ern economic theories and research have provided evidence that
inequality aversion can predict sacrifice for collective interest in
terms of both cooperation (Fehr & Schmidt, 1999) and the pro-
pensity to punish noncooperators (e.g., Fowler, Johnson, & Smir-
nov, 2005). Although delivering incentives typically involves the
sacrifice of self-interest for collective benefit, this does not nec-
essarily mean that others will always perceive that incentives are
directed toward collective benefit and the question becomes, What
situations enable perceptions that incentives are being adminis-
tered in the pursuit of collective interest?
Cost of incentive. The meta-analysis provides good support
for Hypothesis 2, the prediction that incentives would be more
effective when the incentives were costly to administer compared
to when incentives were free. We reasoned from interdependence
theory that when individuals receive incentives involving cost to
the provider, they are more likely to perceive that the provider is
relatively more concerned about the collective, relative to when
incentives are free. In fact, Fehr and Ga¨chter (2002) labeled costly
punishment altruistic because they demonstrated that people use
costly punishment in the absence of immediate, material benefits
for the self or their current group. If individuals perceive that
others provide costly incentives to cooperate as a result of coop-
erative intentions, this may affect other downstream psychological
processes, such as the formation of expectations of subsequent
behavior, development of social norms, and formation of positive
reputations, all of which may increase cooperation.
Recent research has suggested complementary (and alternative)
explanations of why cost of incentives moderates their impact on
cooperation. Several studies have shown a so-called price effect of
punishment, such that people punish more when it costs less (C. M.
Anderson & Putterman, 2006; Andreoni, Harbaugh, & Verster-
lund, 2003; Carpenter, 2007a; Egas & Riedl, 2008; Ostrom et al.,
1992). A possible implication of more frequent costless incentives
is that they heighten the chances of receiving incentives and thus
may increase cooperation. However, this perspective would pre-
dict a stronger incentive–cooperation linkage under conditions of
costless sanctions—which is at odds with the results of this meta-
analysis.
Another implication of the price effect, however, is that low
incentive costs may evoke more antisocial punishment (e.g., pun-
ishment of cooperators by defectors; Cinyabuguma, Page, & Put-
terman, 2005; Falk, Fehr, & Fischbacher, 2005; Herrmann et al.,
2008), whereas high incentive costs may make people more
thoughtful of when and to whom to provide incentives, leading to
more well-considered incentive choices. More specifically, costly
rewards (punishments) will more selectively be delivered to coop-
erators (defectors). Thus, under conditions of costly incentives,
perceived fairness of the incentives will be enhanced, and a norm
of cooperation may be communicated more clearly. Alternatively,
a costly incentive from a fellow participant in a dilemma may carry
additional social reinforcement of the desired behavior, relative to
an equivalent incentive that is free (Masclet et al., 2003; Noussair
& Tucker, 2005). Future research should further examine if per-
ceived cooperative intent motives mediate the effect of costly
incentives on cooperation, while controlling for alternative expla-
nations, such as the frequency of incentives, fairness of incentives,
communication of norms, and social reinforcement.
Source of incentive. In our sample of studies, source of
incentives was strongly correlated with the cost of incentives. Past
research has mostly used paradigms that involve either decentral-
ized and costly incentives or centralized and free incentives. When
we controlled for the variance in the effect size accounted for by
cost of incentives, source of incentives did not account for a
significant amount of variance in the effect size. Hence, Hypoth-
esis 3 was not supported.
One explanation is that centralized sanctioning systems may
actually increase cooperation more than decentralized sanctioning
systems. Earlier, we reasoned that centralized systems may be less
effective than decentralized systems because (a) an authority who
606 BALLIET, MULDER, AND VAN LANGE

administers incentives can be regarded as powerful, (b) people
often associate power with self-interest motives (compared to
similar status others), and (c) people tend to have negative reac-
tions to power asymmetries. At the same time, people are often
more likely to be influenced by more, compared to less, powerful
persons. Traditionally, power is defined in terms of the ability to
move others through a wide range of outcomes (Thibaut & Kelley,
1959), and the abilities to reward others and to punish others form
two important bases of power and influence (French & Raven,
1959). Complementing these classic conceptualizations, more re-
cent research has revealed that feelings of power tend to reinforce
the influence that power affords (C. Anderson, Keltner, & John,
2003). For example, in negotiations, low-power individuals con-
cede more to high-power individuals than the other way around
(Butt & Choi, 2010; De Dreu & van Kleef, 2004; Overbeck, Neale,
& Govan, 2010; Sinaceur & Tiedens, 2006; van Kleef, De Dreu, &
Manstead, 2004). Also, when groups are unsuccessful in managing
a public good (Messick et al., 1983; Rutte & Wilke, 1984), then
group members see the installation of a leader as a fruitful way to
solve the free-rider problem (Samuelson, 1991; Yamagishi, 1986).
In addition, the coding of source of incentives may reflect not
only a difference in power or centralization but also a difference of
personal involvement in the social dilemma. After all, an authority
(e.g., the experimenter) may be less involved in the social dilemma
itself than a typical group member (e.g., a participant). Thus, our
reasoning that the source of incentives reflects power differentials
in the dilemma may be confounded with personal involvement.
This consideration is important because, in decentralized systems,
a fellow group member may promote various feelings, such as
common fate (“we are in this together”) or identity (i.e., he or she
is “one of us”) that might promote trust (Brewer & Kramer, 1986).
In contrast, feelings of common fate or shared identity with au-
thorities may be considerably weaker because the personal in-
volvement of authorities is weaker than that of the prototypical
group member (Hogg & Terry, 2000). Thus, it is plausible that,
along with differences in power, feelings of common fate and
shared identity help people understand psychological conse-
quences of centralized and decentralized incentive systems. We
encourage investigators to manipulate power or centralization
while keeping personal involvement constant. Such research not
only could systematically test how power and centralization affect
the extent to which incentive systems influence cooperation but
also could inform the underlying processes, such as the undermin-
ing of trust in others, perceived cooperative motives, or perceived
fairness of incentives.
Insofar as there was an effect of source of incentive, this effect
disappeared, while the effect of cost of incentive remained signif-
icant. One might speculate that the perceived cost of incentives
might to some degree override the potential effect of source of
incentives. What matters most is not who administers the incen-
tives but the intentions that are perceived to underlie the admin-
istration of incentives. Genuine intentions to safeguard and pro-
mote collective interests, rather than self-interest, are seemingly
more persuasive when such incentives are costly. This finding
informs us about the importance of designing studies where cost of
incentives and source of incentives are orthogonally manipulated.
The results of such studies will inform theory and may provide
relatively concrete guidelines as to how to effectively administer
rewards and punishments in social dilemmas.
Are Rewards and Punishment Equally Effective?
Earlier research is inconclusive on this topic. Although some
studies have reported that rewards are more effective than punish-
ment (Komorita & Barth, 1985; McCusker & Carnevale, 1995),
other studies have suggested that punishment is more effective
than reward (Andreoni et al., 2003; Rapoport & Au, 2001; Stern,
1976) or that there is no difference between the two incentives in
terms of effectiveness (Rand et al., 2009; Sefton, Shupp, &
Walker, 2007). The results of the meta-analysis suggest that re-
wards and punishment have equivalent main effects on coopera-
tion. Although there is a slightly larger effect size estimate of
punishment (d⫽0.70) than rewards (d⫽0.51), this is only
marginally significant, and the analysis of bias in our sample
suggests the rewards–cooperation effect size may be underesti-
mated. Therefore, there is no strong empirical reason to conclude
that reward and punishments differ in their overall effectiveness as
motivators of cooperation in social dilemmas.
Previous theory and research have suggested that punishments
should exert a quicker effect on behavior compared to rewards
(Azrin & Holz, 1966; Driscoll, 2005; Gershoff, 2002; Skinner,
1953). Conversely, rewards have been considered to result in
better long-term learning than punishments (Driscoll, 2005; Ger-
shoff, 2002; Holz & Azrin, 1962; Skinner, 1953). We analyzed the
differences between reward and punishment in both one-shot and
iterated dilemmas. In one-shot dilemmas, rewards had a small
marginally significant positive effect on cooperation (d⫽0.16),
whereas punishments had a slightly larger significant positive
effect on cooperation (d⫽0.32), although these differences were
not statistically significant. Similarly, there was no significant
difference between rewards and punishment during iterated dilem-
mas. Therefore, in the context of social dilemmas, there is no
difference between rewards and punishment as either short-term or
long-term promoters of cooperation.
Although the present findings provide strong evidence that
rewards and punishments do not substantially differ in their impact
on cooperation, it may still be that these incentives differentially
affect either expectations of cooperation, intrinsic motivation, or
ethical concerns during social dilemmas. More insight into these
questions may lead to the conclusion that, despite the fact that
rewards and punishments increase cooperation to an equal extent,
it nevertheless matters which incentive system is installed. This is
especially true in situations in which cooperation levels depend, at
least in part, on individuals’ willingness to cooperate or situations
in which there might be some uncertainty regarding the actual
administration of the incentive (e.g., when it is not easy to catch
the noncooperators). Under these conditions, rewards or punish-
ments may be more or less effective depending on their indirect
effect on individuals’ thoughts, feelings, and motivations in social
dilemmas.
Emerging Issues for Future Study: The Psychology
and Economics of Incentives
One benefit of meta-analytic reviews is that they identify areas
of empirical investigation that are especially promising. Moreover,
the current meta-analysis includes studies from different disci-
plines, which gives rise to some intriguing issues relevant to all
disciplines that study human cooperation. We now discuss the
607
REWARD, PUNISHMENT, AND COOPERATION

issues of (a) incentives over time, (b) effectiveness versus effi-
ciency of incentives, (c) participant payment, (d) and the impor-
tance of culture.
Incentives over time: Beyond reinforcement and punish-
ment. Interestingly, we found that incentives, both reward and
punishments, more strongly increased cooperation during iterated
dilemmas than in one-shot dilemmas and that this association
became even stronger with increased iterations. So, incentives
became more effective over time. At first glance, this result high-
lights the role of incentives in learning cooperation. After all,
learning theory suggests that, over time, rewards (punishments)
will increase (decrease) the probability of subsequent sanctioned
behavior (Skinner, 1953). In such situations, learning and perfor-
mance are enhanced over time through patterns of positive task
feedback on performance.
However, the finding that incentives become more effective
over time is also in line with an interdependence-theoretical per-
spective. In social dilemmas, incentives generally encourage not
only one’s motivation to behave cooperatively (i.e., one’s own
motives) but also one’s expectations about others’ cooperation
(i.e., believed or perceived others’ motives). For example, proso-
cial motives to cooperate may not affect cooperation unless the
perceivers expect others to cooperate (see Pruitt & Kimmel, 1977),
and expectations are strongly influenced by past behavior (e.g.,
Van Lange, 1999; Van Lange, Ouwerkerk, & Tazelaar, 2002).
Therefore, incentives may be more effective after several trials
because people have developed perceived cooperative motives via
group members’ sanctioning behavior, generating expectations of
future cooperation. Similarly, the activation of reputational con-
cerns, the development of social norms prescribing cooperation,
and the weakening of potentially undermining processes (such as
feelings of injustice, feelings of being gypped; see Kerr et al.,
2009; Mulder et al., 2006a; Nowak & Sigmund, 2005; Tenbrunsel
& Messick, 1999; Yamagishi, 1986) may each unfold over time,
thereby accounting for longer term rather than immediate effec-
tiveness of incentives. Also, the emotions that can be evoked by
rewards (e.g., joy, gratitude, and pride) and punishment (e.g.,
anger, guilt, and shame) might affect the duration of their impact
by promoting a set of positive versus negative cycles of social
interaction (e.g., Keltner, van Kleef, Chen, & Kraus, 2008). Al-
though such issues are speculative at the moment, they are impor-
tant as future research may help people to understand the psycho-
logical processes that underlie the impact of incentives on
cooperation over time.
As both individual learning and group processes may explain
the differential effects in one-shot and iterated dilemmas, we also
compared the effectiveness of incentives in a stranger design to
both one-shot and iterated dilemmas using the partner matching
design. Recall that, in the stranger design, individuals play a
dilemma repeatedly but, for each trial, are randomly assigned to a
new group of partners. If learning is the primary mechanism
responsible for incentives’ greater effectiveness during iterated
dilemmas, then one could expect that (a) the stranger design would
have an equally effective impact on the punishment–cooperation
association, compared to iterated dilemmas with a partner match-
ing design, and (b) both the stranger and the partner matching
designs would show stronger punishment–cooperation effects than
one-shot dilemmas. Yet this is not what we found. In fact, the
partner matching studies with iterations had a larger punishment–
cooperation effect size compared to the stranger design studies
with iterations, which had a larger effect than one-shot dilemma
studies. These findings suggest that the effect of punishment
during iterated dilemmas is not only due to learning principles but
also due to processes related to group norms, generating expecta-
tions of group cooperation, and strategically developing reputa-
tions (e.g., Fehr & Fischbacher, 2004; Fehr, Fischbacher, & Ga¨ch-
ter, 2002). Whether this is also the case for rewards remains
unclear as we did not find studies in which the stranger protocol
was employed while studying the impact of rewards on coopera-
tion.
Paid versus hypothetical dilemmas. We were particularly
intrigued by the finding that both rewards and punishments were
more effective when participants were actually paid for their
decisions rather than making hypothetical decisions without mon-
etary consequences. Similarly, Yamagishi (1988) found that when
cooperation was more costly for individuals, then punishment was
more likely to result in higher levels of cooperation. Two simple
messages emerge from the current findings. First, the effect of
costliness of cooperation on the punishment–cooperation associ-
ation generalizes to the rewards–cooperation association. Second,
researchers interested in testing theory of the incentives–-
cooperation linkage should test these hypotheses under conditions
of costly cooperation.
Moreover, it is interesting that disciplines differ in the norms for
paying participants (or not). For example, in psychology, it is not
uncommon to examine social dilemmas involving hypothetical
outcomes, whereas, in (behavioral) economics, there are stronger
norms to examine social dilemmas involving actual cash. The
present findings suggest the possibility that incentives may matter
more when the stakes are greater (as for real money). This possi-
bility, could open a new line of research because some social
dilemma outcomes are indeed material, but many social dilemmas
also involve immaterial outcomes (e.g., contributing to a positive
group spirit) or intermediate ones, which sometimes are quite easy
to translate into actual cash (e.g., time and effort). Thus, the current
results suggest that incentives are more effective in conditions with
concrete material outcomes or when outcomes are framed accord-
ing to these material benefits of cooperation. At the same time,
future research may consider whether incentives are indeed rela-
tively less effective during dilemmas involving immaterial out-
comes and, if so, how the effectiveness of incentives under these
conditions can be enhanced.
Incentives and group size. Group size did not moderate the
effect of incentives on cooperation.
6
Yamagishi (1992) suggested
that group size would increase the magnitude of the incentive–
6
Casari (2005) argued that any difference between punishment behav-
iors in larger, compared to smaller, groups may result from different
punishment protocol used in experiments. Specifically, during decentral-
ized punishment protocol, researchers either allow a fixed or variable
fine-to-fee ratio. In short, Casari noticed that the variable fine-to-fee ratio
used in these studies creates a condition where it is cheaper to punish
defectors in larger, compared to smaller, groups, and so, people are more
likely to punish in larger groups. We examined the effect of group size on
the punishment–cooperation relationship in 14 studies that used a fixed
fine-to-fee ratio. We did not find that group size moderated the effect size.
We did not have enough studies that used a variable fine-to-fee ratio to test
for the association under this condition.
608 BALLIET, MULDER, AND VAN LANGE

cooperation relationship because people are more likely to expect
others to pursue immediate self-interest and support the incentive
system. Although there is evidence supporting this line of reason-
ing, such studies were not included in the meta-analysis because
they did not examine effects of rewards and punishment on coop-
eration and typically involved relatively large group sizes (e.g.,
Kerr, 1989). In contrast, our meta-analysis included only studies
on rewards and punishment, and that body of research typically
includes small group sizes (where the modal of group size was four
persons). Thus, it is still possible that group size moderates the
effectiveness of punishment when group size is observed across a
more extensive range. To date, relevant studies have compared
relatively small groups (e.g., five- vs. 10-person groups: Carpenter,
2007b; Carpenter, Bowles, Gintis, & Hwang, 2009; four- vs.
eight-person groups: Yamagishi, 1992), and the results are incon-
clusive.
One might speculate that the psychological effects of group size
become especially pronounced if one moves beyond at least five
members. For example, according to the core configurations of
group size, one should move beyond the five-person group to
understand the unique adaptive challenges posed by larger groups
(Caporael, 1997). Compared to relatively small groups (roughly
seven members or fewer), which are often challenged by concrete
forms of cooperation (often face-to-face interactions, such as for-
aging, or working on a common task), larger groups (e.g., 30
members or more) feature forms of exchanges (e.g., migration,
environmental behavior, or information exchange) that are more
global, abstract, and less strongly rooted in social interactions
yielding concrete team products (Brewer & Caporael, 2006; Ca-
porael, 1997). We suggest that one important function of a meta-
analytic review is that it identifies gaps in the literature or domains
that have not been fully examined. As such, the present research
suggests that the study of larger groups may provide important
answers to the impact of group size on the use and effectiveness of
rewards and punishment in social dilemmas.
Effectiveness versus efficiency. As noted earlier, the current
meta-analysis leaves little doubt about the effectiveness of rewards
and punishment. A next, more challenging question is whether
these incentives are efficient. Do the material gains for the collec-
tive outweigh the sum of the material costs of the incentives that
are administered? Indeed, this question relates to the issue of group
efficiency, which has recently been addressed in prior research on
punishment in particular. It appears that group efficiency is under-
mined shortly after the administration of punishments (e.g., Dre-
ber, Rand, Fudenberg, & Nowak, 2008; Egas & Riedl, 2008). In
particular, there is a reduction in group efficiency with only a few
iterations of the dilemma (e.g., 10 iterations or fewer). In these
dilemmas, when incentives are added, there is an initial decrease in
group payoffs, but over iterations, an increase in such payoffs
occurs (e.g., Fehr & Ga¨chter, 2000; Ga¨chter et al., 2008; Nikifora-
kis & Normann, 2008).
Although various mechanisms are possible, one is that early
(and frequent) punishment of noncooperators encourages cooper-
ation, which reduces the need for punishment on subsequent trials.
There is also evidence that the potential for reputation monitoring
(Rockenbach & Milinski, 2006), for coordination in the provision
of incentives (Boyd, Gintis, & Bowles, 2010), and for self-chosen
punishment systems (vs. enforced punishment systems; Gu¨rerk,
Irlenbusch, & Rockenbach, 2006) can magnify the efficiency of
incentives. Thus, there is tentative evidence suggesting that incen-
tives are efficient, at least over time, and that several contextual
variables need to be examined to understand when and why they
are especially efficient. Perhaps the most intriguing dilemma is the
conflict between two pressures: (a) Incentives that are more costly
logically undermine efficiency, and (b) incentives that are more
costly are more effective at increasing cooperation and so may
bring about more benefits at the collective level.
The importance of culture. Building on earlier advances
(Henrich et al., 2001), recent research has started to examine how
features of the situation vary in their impact on cooperation across
different cultures (e.g., Herrmann et al., 2008). The findings of the
present meta-analysis, like those of Herrmann et al. (2008), reveal
that the impact of incentives on cooperation varies significantly
between countries. To illustrate, in Russian samples (Ga¨chter &
Herrmann, 2011; Herrmann et al., 2008), decentralized incentives
tend to have no impact on cooperation (d⫽⫺0.03) and reduce
group profits. In many other countries, however, punishment has a
positive impact on cooperation (e.g., United States d⫽0.62) and
enhances group efficiency. Herrmann and colleagues, in an excep-
tionally extensive cross-cultural study, found that countries varied
substantially in their tendency to engage in antisocial (or perverse)
punishment, that is, the punishment of cooperators by defectors.
Among the more plausible accounts of these cultural differences is
the notion that cultures differ in social norms about how to interact
with strangers—people outside of informal networks that are stud-
ied in these experiments (Gintis, 2008). For example, cultures may
differ in terms of how much trust they have in strangers (especially
their generosity) and the desire to communicate the norm to not be
excessively generous with strangers.
Setting aside specific explanations, we regard these findings
as important because if we did not have such an extensive
cross-cultural database, we might have been tempted to gener-
alize our results beyond our largely Western sample (Herrmann
et al., 2008). For example, the fact that costly incentives en-
hance cooperation, compared to costless incentives, may not
extend to the Russian sample. Given the limited number of
studies in several other countries, we were unable to examine
the effect of moderators in a sample of each country. For
example, certain countries may more or less benefit from a
centralized, opposed to decentralized, incentive system. Exam-
ining such questions will undoubtedly make an important con-
tribution. Additionally, future research on culture and cooper-
ation should move beyond simply comparing samples from
different countries (see Ga¨chter, Herrmann, & Thöni, 2010).
Instead, this work may profit by comparing samples from
different groups of countries that have different cultural back-
grounds (e.g., ex-communist vs. advanced industrial societies;
Inglehart & Baker, 2000). This approach may more effectively
identify dimensions of culture that affect the impact of incen-
tives on cooperation in social dilemmas.
Strengths and Limitations
We should acknowledge at least three limitations of the
present research. First, psychologists and economists have em-
ployed different paradigms in manipulating incentives. For
example, whereas economists have traditionally used a decen-
tralized system of punishment, psychologists have used a cen-
609
REWARD, PUNISHMENT, AND COOPERATION

tralized system of punishment. It may be that some subtle
differences in experimental protocol between economists and
psychologists may explain the differences observed between
certain moderators coded along this divide. Yet we are not
overly concerned about this limitation. In fact, we coded
whether the studies were conducted by either an economist or
psychologist and controlled for this variable in a multiple
regression model predicting the effect size. This factor was not
a significant moderator. Intriguingly, there is recent evidence
suggesting that participants in experiments conducted by psy-
chologists versus economists may differ in their interpersonal
orientations: Prosocial orientation (i.e., the tendency to enhance
equality in outcomes and joint outcomes) is the dominant ori-
entation among psychology students, whereas individualistic
orientation (i.e., the tendency to enhance outcomes for self with
no or very little regard for others’ outcomes) is the dominant
orientation among economics students (see Van Lange, Schip-
pers, & Balliet, in press). Such evidence is potentially important
to understanding how participants might respond to external
forces that seek to promote cooperation.
Second, there are fewer studies on rewards than punishments,
reducing statistical power to detect moderators of the rewards–
cooperation association. This is a key area for further investi-
gation.
Third, some studies were not included in this analysis be-
cause they lacked the required statistical information to calcu-
late the effect size. Accordingly, we employed a mixed-effects
analysis, assuming we did not have all studies in the population
of studies represented in our sample. A mixed-effects analysis
is a more conservative test of moderator effects and should
provide more confidence that the results will generalize across
the population of studies.
Concluding Remarks
Reward and punishment, as two distinct incentives, are clas-
sic concepts with a long history in psychology. They are central
in the basics of psychology, providing the conceptual founda-
tion for behaviorism and the study of performance and learning.
Even though incentives may provide hidden costs, such as
reduced intrinsic motivation to cooperate, the results of this
meta-analysis support the position of Hobbes (1651) and Hardin
(1968): Incentives do promote cooperation in social dilemmas.
More interesting, however, is that a generalized learning mech-
anism thought to underlie the effect of incentives on perfor-
mance and learning may not be an all-encompassing explana-
tion of the effect of incentives on cooperation with
interdependent others. Punishment was most effective when
individuals interacted with the same group over several trials,
compared to when individuals were reassigned to a new group
after each trial. Therefore, several other more explicitly social
mechanisms may underlie the effectiveness of incentives, such
as reputation, the communication of social norms, and the
perceived motives of others in the dilemma.
Aligned with this perspective, interdependence theory pro-
vides an insightful analysis that helps one understand how the
effectiveness of incentives can be further promoted. This anal-
ysis links an individual’s psychological processes to the broader
interpersonal circumstances in which behavioral patterns evolve
and are reinforced. Incentives are more effective when incen-
tives are costly. Not only is this finding of great theoretical
importance, it may be essential to the successful functioning of
teams, schools, and organizations. After all, these groups are all
faced with the basic question, How can one promote coopera-
tive behavior among group members? Whether it makes human
beings a sorry lot or not, rewards and punishments do matter
(just as do hopes and fears). Knowing that certain individuals
provide incentives with the noble intent to advance collective
interests, even at a cost to themselves, seems a source of
inspiration, rather than a source of frustration.
References
References marked with an asterisk indicate studies included in the
meta-analysis.
Andenaes, J. (1971). Deterrence and specific offenses. University of Chi-
cago Law Review, 38, 37–53. doi:10.2307/1599029
Anderson, C., Keltner, D., & John, O. P. (2003). Emotional convergence
between people over time. Journal of Personality and Social Psychol-
ogy, 84, 1054–1068. doi:10.1037/0022-3514.84.5.1054
Anderson, C. M., & Putterman, L. (2006). Do non-strategic sanctions obey
the law of demand? The demand for punishment in the voluntary
contribution mechanism. Games and Economic Behavior, 54, 1–24.
doi:10.1016/j.geb.2004.08.007
Andreoni, J., Harbaugh, W., & Versterlund, L. (2003). The carrot or the
stick: Rewards, punishments, and cooperation. American Economic Re-
view, 93, 893–902. doi:10.1257/000282803322157142
Azrin, N. H., & Holz, W. C. (1966). Punishment. In W. A. Honig (Ed.),
Operant behavior: Areas of research and application (pp. 380–447).
New York, NY: Appleton-Century-Crofts.
Bankston, W. B., & Cramer, J. A. (1974). Toward a macro-sociological
interpretation of general deterrence. Criminology: An Interdisciplinary
Journal, 12, 251–280. doi:10.1111/j.1745-9125.1974.tb00635.x
*Bell, P. A., Petersen, T. R., & Hautaluoma, J. E. (1989). The effect of
punishment probability on overconsumption and stealing in a simulated
commons. Journal of Applied Social Psychology, 19, 1483–1495. doi:
10.1111/j.1559-1816.1989.tb01460.x
*Bochet, O., Page, T., & Putterman, L. (2006). Communication and
punishment in voluntary contribution experiments. Journal of Economic
Behavior & Organization, 60, 11–26. doi:10.1016/j.jebo.2003.06.006
*Bornstein, G., & Weisel, O. (2010a). Punishment, cooperation, and
cheater detection in “noisy” social exchange. Games, 1, 18–33. doi:
10.3390/g1010018
*Bornstein, G., & Weisel, O. (2010b). [A replication of the study reported
in “Punishment, Cooperation, and Cheater Detection in ‘Noisy’ Social
Exchange”]. Unpublished raw data.
Boyd, R., Gintis, H., & Bowles, S. (2010, April 30). Coordinated punish-
ment of defectors sustains cooperation and can proliferate when rare.
Science, 328, 617–620. doi:10.1126/science.1183665
Brehm, J. W. (1966). A Theory of psychological reactance. Oxford, Eng-
land: Academic Press.
Brewer, M. B., & Caporael, L. R. (2006). An evolutionary perspective on
social identity: Revisiting groups. In M. Schaller, J. Simpson, & D.
Kenrick (Eds.), Evolution and social psychology (pp. 143–161). New
York, NY: Psychology Press.
Brewer, M. B., & Kramer, R. M. (1986). Choice behavior in social
dilemmas: Effects of social identity, group size, and decision framing.
Journal of Personality and Social Psychology, 50, 543–549. doi:
10.1037/0022-3514.50.3.543
Butt, A. N., & Choi, J. N. (2010). Does power matter? Negotiator status as
a moderator of the relationship between negotiator emotion and behav-
610 BALLIET, MULDER, AND VAN LANGE

ior. International Journal of Conflict Management, 21, 124–146. doi:
10.1108/10444061011037378
*Caldwell, M. D. (1976). Communication and sex effects in a five-person
prisoner’s dilemma game. Journal of Personality and Social Psychol-
ogy, 33, 273–280. doi:10.1037/0022-3514.33.3.273
*Camera, G., & Casari, M. (2009). Cooperation among strangers under the
shadow of the future. American Economic Review, 99, 979–1005. doi:
10.1257/aer.99.3.979
Caporael, L. R. (1997). The evolution of truly social cognition: The core
configurations model. Personality and Social Psychology Review, 1,
276–298. doi:10.1207/s15327957pspr0104_1
Cardenas, J. C., Stranlund, J., & Willis, C. (2002). Economic inequality
and burden-sharing in the provision local environmental quality. Eco-
logical Economics, 40, 379–395. doi:10.1016/S0921-8009(01)00285-3
Carpenter, J. P. (2007a). The demand for punishment. Journal of Economic
Behavior & Organization, 62, 522–542. doi:10.1016/j.jebo.2005.05.004
*Carpenter, J. P. (2007b). Punishing free-riders: How group size affects
mutual monitoring and the provision of public goods. Games and
Economic Behavior, 60, 31–51. doi:10.1016/j.geb.2006.08.011
Carpenter, J., Bowles, S., Gintis, H., & Hwang, S. (2009). Strong reci-
procity and team production. Journal of Economic Behavior & Organi-
zation, 71, 221–232. doi:10.1016/j.jebo.2009.03.011
*Carpenter, J., & Matthews, P. H. (2009). What norms trigger punishment?
Experimental Economics, 12, 272–288. doi:10.1007/s10683-009-9214-z
*Carpenter, J. P., Matthews, P., & Ong’ong’a, O. (2004). Why punish?
Social reciprocity and the enforcement of prosocial norms. Journal of
Evolutionary Economics, 14, 407–429. doi:10.1007/s00191-004-0212-1
Casari, M. (2005). On the design of peer punishment experiments. Exper-
imental Economics, 8, 107–115. doi:10.1007/s10683-005-0869-9
*Casari, M., & Luini, L. (2009). Cooperation under alternative punishment
institutions: An experiment. Journal of Economic Behavior & Organi-
zation, 71, 273–282. doi:10.1016/j.jebo.2009.03.022
*Cha, Y. S. (2000). Using the concepts of framing and loss aversion to
answer pertinent questions on cooperative behaviors in social dilemmas.
Dissertation Abstracts International: Section B. Sciences and Engineer-
ing, 60(5), 5831.
Chaudhuri, A. (2011). Sustaining cooperation in laboratory public goods
experiments: A selective survey of the literature. Experimental Econom-
ics, 14, 47–83. doi:10.1007/s10683-010-9257-1
*Chen, X., Pillutla, M. M., & Yao, X. (2009). Unintended consequences of
cooperation inducing and maintaining mechanisms in public goods
dilemmas: Sanctions and moral appeals. Group Processes & Intergroup
Relations, 12, 241–255. doi:10.1177/1368430208098783
*Cinyabuguma, M., Page, T., & Putterman, L. (2005). Cooperation under
the threat of expulsion in a public goods experiment. Journal of Public
Economics, 89, 1421–1435. doi:10.1016/j.jpubeco.2004.05.011
*Clark, J. (2002). Recognizing large donations to public goods: An exper-
imental test. Managerial and Decision Economics, 23, 33–44. doi:
10.1002/mde.1044
Cooper, H. M. (1998). Synthesizing research: A guide for literature re-
views (3rd ed.). Thousand Oaks, CA: Sage.
Dawes, R. M. (1980). Social dilemmas. Annual Review of Psychology, 31,
169–193. doi:10.1146/annurev.ps.31.020180.001125
Deci, E. L., Koestner, R., & Ryan, R. M. (1999). A meta-analytic review
of experiments examining the effects of extrinsic rewards on intrinsic
motivation. Psychological Bulletin, 125, 627–668. doi:10.1037/0033-
2909.125.6.627
Deci, E. L., & Ryan, R. M. (2000). The “what” and “why” of goal pursuits:
Human needs and the self-determination of behavior. Psychological
Inquiry, 11, 227–268. doi:10.1207/S15327965PLI1104_01
De Cremer, D., & van Dijk, E. (2005). When and why leaders put
themselves first: Leader behaviour in resource allocations as a function
of feeling entitled. European Journal of Social Psychology, 35, 553–
563. doi:10.1002/ejsp.260
De Cremer, D., & van Dijk, E. (2008). Leader–follower effects in resource
dilemmas: The roles of leadership selection and social responsibility.
Group Processes & Intergroup Relations, 11, 355–369. doi:10.1177/
1368430208090647
De Dreu, C. K. W., & van Kleef, G. A. (2004). The influence of power on
the information search, impression formation, and demands in negotia-
tion. Journal of Experimental Social Psychology, 40, 303–319. doi:
10.1016/j.jesp.2003.07.004
Denant-Boemont, L., Masclet, D., & Noussair, C. N. (2007). Punishment,
counterpunishment, and sanction enforcement in a social dilemma ex-
periment. Economic Theory, 33, 145–167. doi:10.1007/s00199-007-
0212-0
*Dickinson, D. L. (2001). The carrot vs. the stick in work team motivation.
Experimental Economics, 4, 107–124. doi:10.1007/BF01669275
*Dreber, A., Rand, D. G., Fudenberg, D., & Nowak, M. A. (2008, March
20). Winners don’t punish. Nature, 452, 348–351. doi:10.1038/
nature06723
Driscoll, M. P. (2005). Psychology of learning for instruction. Boston,
MA: Peterson.
Duval, S., & Tweedie, R. (2000). Trim and fill: A simple funnel-plot-based
method of testing and adjusting for publication bias in meta-analysis.
Biometrics, 56, 455–463. doi:10.1111/j.0006-341X.2000.00455.x
Edney, J. J., & Harper, C. S. (1978). The commons dilemma: A review of
contributions in psychology. Environmental Management, 2, 491–507.
doi:10.1007/BF01866708
*Eek, D., Loukopoulos, P., Satoshi, F., & Ga¨rling, T. (2002). Spill-over
effect of intermittent costs for defection in social dilemmas. European
Journal of Social Psychology, 32, 801–813. doi:10.1002/ejsp.122
*Egas, M., & Riedl, A. (2008). The economics of altruistic punishment and
the maintenance of cooperation. Proceedings of the Royal Society:
Series B. Biological Sciences, 275, 871–878. doi:10.1098/
rspb.2007.1558
*Ertan, A., Page, T., & Putterman, L. (2009). Who to punish? Individual
decisions and majority rule in mitigating the free rider problem. Euro-
pean Economic Review, 53, 495–511. doi:10.1016/j.euroecorev
.2008.09.007
Falk, A., Fehr, E., & Fischbacher, U. (2005). Driving forces of informal
sanctions. Econometrica, 73, 2017–2030. doi:10.1111/j.1468-
0262.2005.00644.x
Fehr, E., & Fischbacher, U. (2004). Social norms and human cooperation.
Trends in Cognitive Sciences, 8, 185–190. doi:10.1016/j.tics
.2004.02.007
Fehr, E., Fischbacher, U., & Ga¨chter, S. (2002). Strong reciprocity, human
cooperation, and the enforcement of social norms. Human Nature, 13,
1–25. doi:10.1007/s12110-002-1012-7
*Fehr, E., & Ga¨chter, S. (2000). Cooperation and punishment in public
goods experiments. American Economic Review, 90, 980–994. doi:
10.1257/aer.90.4.980
*Fehr, E., & Ga¨chter, S. (2002, January 10). Altruistic punishment in
humans. Nature, 415, 137–140. doi:10.1038/415137a
Fehr, E., & Schmidt, K. M. (1999). A theory of fairness, competition, and
cooperation. Quarterly Journal of Economics, 114, 817–868. doi:
10.1162/003355399556151
Fowler, J. H., Johnson, T., & Smirnov, O. (2005, January 6). Egalitarian
motives and altruistic punishment. Nature, 433, E1. doi:10.1038/
nature03256
French, J. R. P., Jr., & Raven, B. (1959). The bases of social power. In D.
Cartwright (Ed.), Studies in social power (pp. 150–167). Ann Arbor:
University of Michigan, Institute for Social Research.
*Fuster, A., & Meier, S. (2010). Another hidden cost of incentives: The
detrimental effect on norm enforcement. Management Science, 56, 57–
70. doi:10.1287/mnsc.1090.1081
*Ga¨chter, S., & Herrmann, B. (2009). Reciprocity, culture and human
cooperation: Previous insights and a new cross-cultural experiment.
611
REWARD, PUNISHMENT, AND COOPERATION

Philosophical Transactions of the Royal Society: Series B. Biological
Sciences, 364, 791–806. doi:10.1098/rstb.2008.0275
*Ga¨chter, S., & Herrmann, B. (2011). The limits of self-governance when
cooperators get punished: Experimental evidence from urban and rural
Russia. European Economic Review, 55, 193–210. doi:10.1016/
j.euroecorev.2010.04.003
Ga¨chter, S., Herrmann, B., & Thöni, C. (2010). Culture and cooperation.
Philosophical Transactions of the Royal Society: Series B. Biological
Sciences, 365, 2651–2661. doi:10.1098/rstb.2010.0135
*Ga¨chter, S., Renner, E., & Sefton, M. (2008, December 5). The long-run
benefits of punishment. Science, 322, 1510. doi:10.1126/science
.1164744
*Ga¨chter, S., & Thöni, C. (2005). Social learning and voluntary coopera-
tion among like minded people. Journal of European Economic Asso-
ciation, 3, 303–314. doi:10.1162/jeea.2005.3.2-3.303
Gershoff, E. T. (2002). Corporate punishment by parents and associated
child behaviors and experiences: A meta-analytic and theoretical review.
Psychological Bulletin, 128, 539–579. doi:10.1037/0033-2909.128
.4.539
Gintis, H. (2008, March 7). Punishment and cooperation. Science, 319,
1345–1346. doi:10.1126/science.1155333
Gneezy, U., & Rustichini, A. (2000). A fine is a price. Journal of Legal
Studies, 29, 1–17. doi:10.1086/468061
Gu¨rerk, Ö., Irlenbusch, B., & Rockenbach, B. (2006, April 7). The com-
petitive advantage of sanctioning institutions. Science, 312, 108–111.
doi:10.1126/science.1123633
Hardin, G. (1968, December 13). The tragedy of the commons. Science,
162, 1243–1248. doi:10.1126/science.162.3859.1243
Hedges, L. V., & Olkin, I. (1985). Statistical methods for meta-analysis.
Orlando, FL: Academic Press.
Henrich, J., Boyd, R., Bowles, S., Camerer, C., Fehr, E., Gintis, H., &
McElreath, R. (2001). In search of homo economicus: Behavioral ex-
periments in 15 small-scale societies. American Economic Review, 91,
73–78. doi:10.1257/aer.91.2.73
*Herrmann, B., Thöni, C., & Ga¨chter, S. (2008, March 7). Antisocial
punishment across societies. Science, 319, 1362–1367. doi:10.1126/
science.1153808
Hobbes, T. (1651). Leviathan. Cambridge, England: Cambridge University
Press.
Hogg, M. A., & Terry, D. J. (2000). Social identity and self-categorization
processes in organizational contexts. Academy of Management Review,
25, 121–140. doi:10.2307/259266
Holz, W. C., & Azrin, N. H. (1962). Interactions between the discrimina-
tive and aversive properties of punishment. Journal of the Experimental
Analysis of Behavior, 5, 229–234. doi:10.1901/jeab.1962.5-229
*Hopfensitz, A., & Rueben, E. (2009). The importance of emotions for the
effectiveness of social punishment. Economic Journal, 119, 1534–1559.
doi:10.1111/j.1468-0297.2009.02288.x
Inglehart, R., & Baker, W. E. (2000). Modernization, cultural change, and
the persistence of traditional values. American Sociological Review, 65,
19–51. doi:10.2307/2657288
Kalkhoff, W., & Barnum, C. (2000). The effects of status-organizing and
social identity processes on patterns of social influence. Social Psychol-
ogy Quarterly, 63, 95–115. doi:10.2307/2695886
Kazdin, A. E. (2001). Behavior modification in applied settings. Belmont,
CA: Wadsworth.
Kelley, H. H., Holmes, J. G., Kerr, N. L., Reis, H. T., Rusbult, C. E., & Van
Lange, P. A. M. (2003). An atlas of interpersonal situations. Cambridge,
England: Cambridge University Press.
Kelley, H. H., & Thibaut, J. W. (1978). Interpersonal relations: A theory
of interdependence. New York, NY: Wiley.
Keltner, D., Langner, C. A., & Allison, M. L. (2006). Power and moral
leadership. In D. Rhode (Ed.). Moral leadership: The theory and prac-
tice of power, judgment, and policy (pp. 177–194). San Francisco, CA:
Jossey-Bass.
Keltner, D., van Kleef, G. A., Chen, S., & Kraus, M. W. (2008). A
reciprocal influence model of social power: Emerging principles and
lines of inquiry. Advances in Experimental Social Psychology, 40, 151–
192. doi:10.1016/S0065-2601(07)00003-2
Kerr, N. L. (1989). Illusions of efficacy: The effects of group size on
perceived efficacy in social dilemmas. Journal of Experimental Social
Psychology, 25, 287–313. doi:10.1016/0022-1031(89)90024-3
Kerr, N. L., Rumble, A. C., Park, E., Ouwerkerk, J. W., Parks, C. D.,
Gallucci, M., & Van Lange, P. A. M. (2009). How many bad apples does
it take to spoil the whole barrel? Social exclusion and toleration for bad
apples. Journal of Experimental Social Psychology, 45, 603–613. doi:
10.1016/j.jesp.2009.02.017
*Kieruj, N., Mulder, L. B., & Nelissen, R. M. A. (2008). Opdrijven of
beklijven? Het verschil tussen boetes en sociale afkeuring in sociale
dilemma’s [Forcing up or sinking in? The difference between fines and
social disapproval in social dilemmas]. Jaarboek Sociale Psychologie,
2007, 193–202. doi:urn:nbn:nl:ui:12-3132860
*Kocher, M. G., Martinsoon, P., & Visser, M. (2008). Does stake size
matter for cooperation and punishment? Economic Letters, 99, 508–511.
doi:10.1016/j.econlet.2007.09.048
Kollock, P. (1998). Social dilemmas: The anatomy of cooperation. Annual
Review of Sociology, 24, 183–214. doi:10.1146/annurev.soc.24.1.183
*Komorita, S. S., & Barth, J. M. (1985). Components of reward in social
dilemmas. Journal of Personality and Social Psychology, 48, 364–373.
doi:10.1037/0022-3514.48.2.364
Komorita, S. S., & Parks, C. D. (1995). Social dilemmas. Boulder, CO:
Westview Press.
Komorita, S. S., Sheposh, J. P., & Braver, S. L. (1968). Power, the use of
power, and cooperative choice in a two-person game. Journal of Per-
sonality and Social Psychology, 8, 134–142. doi:10.1037/h0025300
Kramer, R. M. (1999). Trust and distrust in organizations: Emerging
perspectives, enduring questions. Annual Review of Psychology, 50,
569–598. doi:10.1146/annurev.psych.50.1.569
*Kroll, S., Cherry, T. L., & Shogren, J. F. (2007). Voting, punishment, and
public goods. Economic Inquiry, 45, 557–570. doi:10.1111/j.1465-
7295.2007.00028.x
Lepper, M. R., & Greene, M. D. (1978). The hidden costs of reward.
Hillsdale, NJ: Erlbaum.
Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-analysis. London,
England: Sage.
Lynn, M., & Oldenquist, A. (1986). Egoistic and nonegoistic motives in
social dilemmas. American Psychologist, 41, 529–534. doi:10.1037/
0003-066X.41.5.529
*Martichuski, D. K., & Bell, P. A. (1991). Reward, punishment, privati-
zation, and moral suasion in a commons dilemma. Journal of Applied
Social Psychology, 21, 1356–1369. doi:10.1111/j.1559-1816.1991
.tb00475.x
Masclet, D., Noussair, C., Tucker, S., & Villeval, M. (2003). Monetary and
nonmonetary punishment in the voluntary contributions mechanism.
American Economic Review, 93, 366–380. doi:10.1257/
000282803321455359
*McCusker, C., & Carnevale, P. J. (1995). Framing in resource dilemmas:
Loss aversion and the moderating effects of sanctions. Organizational
Behavior and Human Decision Processes, 61, 190–201. doi:10.1006/
obhd.1995.1015
Messick, D. M., Wilke, H. A. M., Brewer, M. B., Kramer, R. M., Zemke,
P. E., & Lui, L. (1983). Individual adaptations and structural change as
solutions to social dilemmas. Journal of Personality and Social Psychol-
ogy, 44, 294–309. doi:10.1037/0022-3514.44.2.294
*Mulder, L. B. (2005). [Sanctions and social values in a public good
dilemma]. Unpublished raw data, Department of Human Resource Man-
612 BALLIET, MULDER, AND VAN LANGE

agement and Organizational Behavior, University of Groningen, Gro-
ningen, the Netherlands.
*Mulder, L. B. (2008). The difference between punishments and rewards in
fostering moral concerns in social decision making. Journal of Experi-
mental Social Psychology, 44, 1436–1443. doi:10.1016/j.jesp
.2008.06.004
Mulder, L. B., & Nelissen, R. M. A. (2011). When rules really make a
difference: The effect of cooperation rules on moral norms in social
dilemmas. Journal of Business Ethics. Advance online publication.
doi:10.1007/s10551-011-0795-z
*Mulder, L. B., van Dijk, E., De Cremer, D., & Wilke, H. A. M. (2001).
[Sanctions, rewards and trust in a public good dilemma]. Unpublished
raw data, Department of Human Resource Management and Organiza-
tional Behavior, University of Groningen, Groningen, the Netherlands.
*Mulder, L. B., van Dijk, E., De Cremer, D., & Wilke, H. A. M. (2002).
[Sanctions and trust in a public good dilemma]. Unpublished raw data,
Department of Human Resource Management and Organizational Be-
havior, University of Groningen, Groningen, the Netherlands.
*Mulder, L. B., van Dijk, E., De Cremer, D., & Wilke, H. A. M. (2003).
[Sanctions and trust in a public good dilemma]. Unpublished raw data,
Department of Human Resource Management and Organizational Be-
havior, University of Groningen, Groningen, the Netherlands.
*Mulder, L. B., van Dijk, E., De Cremer, D., & Wilke, H. A. M. (2006a).
Undermining trust and cooperation: The paradox of sanctioning systems
in social dilemmas. Journal of Experimental Social Psychology, 42,
147–162. doi:10.1016/j.jesp.2005.03.002
*Mulder, L. B., van Dijk, E., De Cremer, D., & Wilke, H. A. M. (2006b).
When sanctions fail to increase cooperation in social dilemmas: Con-
sidering the presence of an alternative defection option. Personality and
Social Psychology Bulletin, 32, 1312–1324. doi:10.1177/
0146167206289978
Mulder, L. B., van Dijk, E., Wilke, H. A. M., & De Cremer, D. (2005). The
effect of feedback on support for a sanctioning system in a social
dilemma: The difference between installing and maintaining the sanc-
tion. Journal of Economic Psychology, 26, 443–458. doi:10.1016/
j.joep.2004.12.007
*Mulder, L. B., & van Prooijen, J. W. (2006). [Sanctions and framing in
social dilemmas]. Unpublished raw data, Department of Human Re-
source Management and Organizational Behavior, University of Gro-
ningen, Groningen, the Netherlands.
Mulder, L. B., Verboon, P., & De Cremer, D. (2009). Sanctions and moral
judgments: The moderating effect of sanction severity and trust in
authorities. European Journal of Social Psychology, 39, 255–269. doi:
10.1002/ejsp.506
Murray, S. L., & Holmes, J. G. (2009). The architecture of interdependent
minds: A motivation-management theory of mutual responsiveness.
Psychological Review, 116, 908–928. doi:10.1037/a0017015
Muthusamy, S. K., & White, M. A. (2006). Does power sharing matter?
The role of power and influence in alliance performance. Journal of
Business Research, 59, 811–819. doi:10.1016/j.jbusres.2006.01.018
*Myers, C. D. (2009). Participation and punishment. Unpublished manu-
script, Department of Politics, Princeton University.
*Nelissen, R. M. A., & Mulder, L. B. (2010). What makes a sanction
“stick”? The effects of financial and social sanctions on norm compli-
ance. Unpublished manuscript, Department of Social Psychology, Til-
burg University, Tilburg, the Netherlands.
*Nikiforakis, N. (2008). Punishment and counter-punishment in public
good games: Can we really govern ourselves? Journal of Public Eco-
nomics, 92, 91–112. doi:10.1016/j.jpubeco.2007.04.008
*Nikiforakis, N., & Normann, H. (2008). A comparative statistical analysis
of punishment in public-good experiments. Experimental Economics,
11, 358–369. doi:10.1007/s10683-007-9171-3
*Nikiforakis, N., Normann, H., & Wallace, B. (2010). Asymmetric en-
forcement of cooperation in a social dilemma. Southern Economic
Journal, 76, 638–659. doi:10.4284/sej.2010.76.3.638
Noussair, C., & Tucker, S. (2005). Combining monetary and social sanc-
tions to promote cooperation. Economic Inquiry, 43, 649–660. doi:
10.1093/ei/cbi045
Nowak, M. A., & Sigmund, K. (2005, October 27). Evolution of indirect
reciprocity. Nature, 437, 1291–1298. doi:10.1038/nature04131
*O’Gorman, R., Henrich, J., & Van Vugt, M. (2008). Constraining free
riding in public goods games: Designated solitary punishers can sustain
human cooperation. Proceedings of the Royal Society: Series B. Biology,
276, 323–329. doi:10.1098/rspb.2008.1082
*O’Gorman, R., & Van Vugt, M. (2010). [Does allowing participants the
option to choose their punisher increase contributions in a public goods
game.] Unpublished raw data, Department of Psychology, University of
Essex, Colchester, England.
Oldmeadow, J. A., Platow, M. J., Foddy, M., & Anderson, D. (2003).
Self-categorization, status, and social influence. Social Psychology
Quarterly, 66, 138–152. doi:10.2307/1519844
Oliver, P. (1980). Rewards and punishments as selective incentives for
collective action: Theoretical investigations. American Journal of Soci-
ology, 85, 1356–1375. doi:10.1086/227168
Olson, M. (1965). The logic of collective action: Public goods and the
theory of groups. Cambridge, MA: Harvard University Press.
Orwin, R. G. (1983). A fail-safe Nfor effect size in meta-analysis. Journal
of Educational Statistics, 8, 157–159. doi:10.2307/1164923
Ostrom, E., Walker, J., & Gardner, R. (1992). Covenants with and without
the sword: Self-governance is possible. American Political Science
Review, 86, 404–417. doi:10.2307/1964229
Overbeck, J. R., Neale, M. A., & Govan, C. L. (2010). I feel, therefore you
act: Intrapersonal and interpersonal effects of emotion on negotiation as
a function of social power. Organizational Behavior and Human Deci-
sion Processes, 112, 126–139. doi:10.1016/j.obhdp.2010.02.004
*Page, T., Putterman, L., & Unel, B. (2005). Voluntary association in
public goods experiments: Reciprocity, mimicry, and efficiency. Eco-
nomic Journal, 115, 1032–1053. doi:10.1111/j.1468-0297.2005.01031.x
*Parks, C. D. (2000). Testing various types of cooperation rewards in
social dilemma. Group Processes & Intergroup Relations, 3, 339–350.
doi:10.1177/1368430200003004001
Parks, C. D., Sanna, L. J., & Berel, S. R. (2001). Actions of similar others
as inducements to cooperate in social dilemmas. Personality and Social
Psychology Bulletin, 27, 345–354. doi:10.1177/0146167201273008
*Patel, A., Cartwright, E., & Van Vugt, M. (2010). Punishment cannot
sustain cooperation in a public good game with free-rider anonymity.
Unpublished manuscript, Department of Economics, University of
Gothenburg, Gothenburg, Sweden.
Pruitt, D. G., & Kimmel, M. J. (1977). Twenty years of experimental
gaming: Critique, synthesis, and suggestions for the future. Annual
Review of Psychology, 28, 363–392. doi:10.1146/annurev.ps.28
.020177.002051
*Rand, D. G., Dreber, A., Ellingsen, T., Fudenberg, D., & Nowak, M. A.
(2009, September 4). Positive interactions promote public cooperation.
Science, 325, 1272–1275. doi:10.1126/science.1177418
*Rapoport, A., & Au, W. T. (2001). Bonus and penalty in common pool
resource dilemmas under uncertainty. Organizational Behavior and Hu-
man Decision Processes, 85, 135–165. doi:10.1006/obhd.2000.2935
Rapoport, A., & Chammah, A. M. (1965). Prisoner’s dilemma: A study in
conflict and cooperation. Ann Arbor: University of Michigan Press.
*Reuben, E., & Riedl, A. (2009). Enforcement of contribution norms in
public goods games with heterogeneous populations. Unpublished man-
uscript, Management Division, Columbia University, New York, NY.
*Riedl, A., & Ule, A. (2009). Exclusion and cooperation in social network
experiments. Unpublished manuscript, Faculty of Economics and
Econometrics, Center for Research in Experimental Economics and
political Decision-making (CREED), Amsterdam, the Netherlands.
613
REWARD, PUNISHMENT, AND COOPERATION

Rockenbach, B., & Milinski, M. (2006, December 7). The efficient inter-
action of indirect reciprocity and costly punishment. Nature, 444, 718–
723. doi:10.1038/nature05229
Rusbult, C. E., & Van Lange, P. A. M. (2003). Interdependence, interac-
tion, and relationships. Annual Review of Psychology, 54, 351–375.
doi:10.1146/annurev.psych.54.101601.145059
Rutte, C. G., & Wilke, H. A. M. (1984). Social dilemmas and leadership.
European Journal of Social Psychology, 14, 105–121. doi:10.1002/
ejsp.2420140109
Ryan, R. M., & Deci, E. L. (2000). Self-determination theory and the
facilitation of intrinsic motivation, social development, and well-being.
American Psychologist, 55, 68–78. doi:10.1037110003-066X.55.1.68
Samuelson, C. D. (1991). Perceived task difficulty, causal attributions,
and preferences for structural change in resource dilemmas. Person-
ality and Social Psychology Bulletin, 17, 181–187. doi:10.1177/
014616729101700210
Samuelson, C. D., & Allison, S. T. (1994). Cognitive factors affecting the
use of social decision heuristics in resource-sharing tasks. Organiza-
tional Behavior and Human Decision Processes, 58, 1–27. doi:10.1006/
obhd.1994.1027
*Sato, K. (1987). Distribution of the cost of maintaining common re-
sources. Journal of Experimental Social Psychology, 23, 19–31. doi:
10.1016/0022-1031(87)90023-0
*Sefton, M., Shupp, R., & Walker, J. M. (2007). The effect of rewards and
sanctions in provision of public goods. Economic Inquiry, 45, 671–690.
doi:10.1111/j.1465-7295.2007.00051.x
*Sell, J., & Wilson, R. K. (1999). The maintenance of cooperation:
Expectations of future interaction and the trigger of group punishment.
Social Forces, 77, 1551–1571. doi:10.2307/3005886
*Shaw, J. I. (1976). Response-contingent payoffs and cooperative behavior
in the prisoner’s dilemma game. Journal of Personality and Social
Psychology, 34, 1024–1033. doi:10.1037/0022-3514.34.5.1024
*Shinada, M., & Yamagishi, T. (2007). Punishing free riders: Direct and
indirect promotion of cooperation. Evolution and Human Behavior, 28,
330–339. doi:10.1016/j.evolhumbehav.2007.04.001
Shinada, M., & Yamagishi, T. (2008). Bringing back the leviathan into
social dilemmas. In A. Biel, D. Eek, T. Garling, & M. Gukstafsson
(Eds.), New issues and paradigms in research on social dilemmas (pp.
93–123). New York, NY: Springer. doi:10.1007/978-0-387-72596-3_7
Sigmund, K. (2007). Punish or perish? Retaliation and collaboration among
humans. Trends in Ecology & Evolution, 22, 593–600. doi:10.1016/
j.tree.2007.06.012
Sinaceur, M., & Tiedens, L. Z. (2006). Get mad and get more than even:
The benefits of anger expressions in negotiations. Journal of Experi-
mental Social Psychology, 42, 314–322. doi:10.1016/j.jesp.2005.05.002
Skinner, B. F. (1953). Science of human behavior. New York, NY: Mac-
millan.
Stern, P. C. (1976). Effect of incentives and education on resource con-
servation decisions in a simulated common dilemma. Journal of Per-
sonality and Social Psychology, 34, 1285–1292. doi:10.1037/0022-
3514.34.6.1285
*Sutter, M., Haigner, S., & Kocher, M. G. (2010). Choosing the carrot or
the stick? Endogenous institutional choice in social dilemma situations.
Review of Economic Studies, 77, 1540–1566. doi:10.1111/j.1467-
937X.2010.00608.x
*Sutter, M., Linder, P., & Platsch, D. (2009). Social norms, third-party
observation, and third-party rewards. Unpublished manuscript, Depart-
ment of Public Finance, University of Innsbruck, Innsbruck, Austria.
*Tan, F. (2008). Punishment in a linear public good game with productivity
heterogeneity. De Economist, 156, 269–293. doi:10.1007/s10645-008-
9094-1
*Tenbrunsel, A. E., & Messick, D. M. (1999). Sanctioning systems, deci-
sion frames, and cooperation. Administrative Science Quarterly, 44,
684–707. doi:10.2307/2667052
Thibaut, J. W., & Kelley, H. H. (1959). The social psychology of groups.
New York, NY: Wiley.
Turner, J. C. (1991). Social influence. Milton Keynes, England: Open
University Press.
*Tyran, J., & Feld, L. P. (2004). Achieving compliance when Legal
sanctions are non-deterrent. Unpublished manuscript, Department of
Economics, University of Copenhagen, Copenhagen, Denmark.
van Dijk, E., & De Cremer, D. (2006). Self-benefiting in the allocation of
scarce resources: Leader-follower effects and the moderating effect of
social value orientations. Personality and Social Psychology Bulletin,
32, 1352–1361. doi:10.1177/0146167206290338
van Kleef, G. A., De Dreu, C. K. W., & Manstead, A. S. R. (2004). The
interpersonal effects of anger and happiness in negotiations. Journal of
Personality and Social Psychology, 86, 57–76. doi:10.1037/0022-
3514.86.1.57
Van Lange, P. 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, 337–349. doi:10.1037/0022-
3514.77.2.337
Van Lange, P. A. M., De Cremer, D., van Dijk, E., & Van Vugt, M. (2007).
Self-interest and beyond: Basic principles of social interaction. In A. W.
Kruglanski & E. T. Higgins (Eds.), Social psychology: Handbook of
basic principles (pp. 540–561). New York, NY: Guilford Press.
Van Lange, P. A. M., & Joireman, J. A. (2008). How we can promote
behavior that serves all of us in the future. Social Issues and Policy
Review, 2, 127–157. doi:10.1111/j.1751-2409.2008.00013.x
Van Lange, P. A. M., Ouwerkerk, J. W., & Tazelaar, M. J. A. (2002). How
to overcome the detrimental effects of noise in social interaction: The
benefits of generosity. Journal of Personality and Social Psychology, 82,
768–780. doi:10.1037/0022-3514.82.5.768
Van Lange, P. A. M., & Rusbult, C. E. (in press). Interdependence theory.
In P. A. M. Van Lange, A. W. Kruglanski, & E. T. Higgins. (Eds.),
Handbook of theories of social psychology. Thousand Oaks, CA: Sage.
Van Lange, P. A. M., Schippers, M. C., & Balliet, D. (in press). Who
volunteers in psychology experiments? An empirical review of prosocial
motivation in volunteering. Personality and Individual Differences.
*van Prooijen, J., Gallucci, M., & Toeset, G. (2008). Procedural justice in
punishment systems: Inconsistent punishment procedures have detri-
mental effects on cooperation. British Journal of Social Psychology, 47,
311–324. doi:10.1348/014466607X218212
*Van Vugt, M., & De Cremer, D. (1999). Leadership in social dilemmas:
The effects of group identification on collective actions to provide public
goods. Journal of Personality and Social Psychology, 76, 587–599.
doi:10.1037/0022-3514.76.4.587
Van Vugt, M., Hogan, R., & Kaiser, R. B. (2008). Leadership, follower-
ship, and evolution: Some lessons from the past. American Psychologist,
63, 182–196. doi:10.1037/0003-066X.63.3.182
*Walker, J. M., & Halloran, M. A. (2004). Rewards and sanctions and the
provision of public goods in one-shot settings. Experimental Economics,
7, 235–247. doi:10.1023/B:EXEC.0000040559.08652.51
Wieselquist, J., Rusbult, C. E., Foster, C. A., & Agnew, C. R. (1999).
Commitment, pro-relationship behavior, and trust in close relationships.
Journal of Personality and Social Psychology, 77, 942–966. doi:
10.1037/0022-3514.77.5.942
*Wit, A., & Wilke, H. A. (1990). The presentation of rewards and pun-
ishments in a simulated social dilemma. Social Behaviour, 5, 231–245.
*Xiao, E., & Kunreuther, H. (2010). Punishment and cooperation in
stochastic social dilemmas. Unpublished manuscript, The Wharton
School, University of Pennsylvania.
*Yamagishi, T. (1986). The provisioning of a sanctioning system as a
public good. Journal of Personality and Social Psychology, 51, 110–
116. doi:10.1037/0022-3514.51.1.110
*Yamagishi, T. (1988). Seriousness of social dilemmas and the provision
614 BALLIET, MULDER, AND VAN LANGE

of a sanctioning system. Social Psychology Quarterly, 51, 32–42. doi:
10.2307/2786982
*Yamagishi, T. (1992). Group size and the provision of a sanctioning
system in a social dilemma. In W. B. G. Liebrand, D. M. Messick, &
H. A. M. Wilke (Eds.), Social dilemma: Theoretical issues and research
findings (pp. 267–287). Oxford, England: Pergamon Press.
Zelmer, J. (2003). Linear public goods experiments: A meta-analysis.
Experimental Economics, 6, 299–310. doi:10.1023/A:1026277420119
Zimring, F. (1971). Perspectives on deterrence. Washington, DC: National
Institute of Mental Health.
Received May 23, 2010
Revision received February 16, 2011
Accepted February 22, 2011 䡲
Online First Publication
APA-published journal articles are now available Online First in the PsycARTICLES database.
Electronic versions of journal articles will be accessible prior to the print publication, expediting
access to the latest peer-reviewed research.
All PsycARTICLES institutional customers, individual APA PsycNET威database package sub-
scribers, and individual journal subscribers may now search these records as an added benefit.
Online First Publication (OFP) records can be released within as little as 30 days of acceptance and
transfer into production, and are marked to indicate the posting status, allowing researchers to
quickly and easily discover the latest literature. OFP articles will be the version of record; the
articles have gone through the full production cycle except for assignment to an issue and
pagination. After a journal issue’s print publication, OFP records will be replaced with the final
published article to reflect the final status and bibliographic information.
615
REWARD, PUNISHMENT, AND COOPERATION