2010 07online
2010 07online
Dictator Games:
A Meta Study
Christoph Engel
Christoph Engel
March 2010
Max Planck Institute for Research on Collective Goods, Kurt-Schumacher-Str. 10, D-53113 Bonn
http://www.coll.mpg.de
Dictator Games: A Meta Study
Christoph Engel
Abstract
Over the last 25 years, more than a hundred dictator game experiments have been published.
This meta study summarizes the evidence. Exploiting the fact that most experiments had to fix
parameters they did not intend to test, in multiple regression the meta study is able to assess the
effect of single manipulations, controlling for a host of alternative explanatory factors. The re-
sulting rich dataset also provides a testbed for comparing alternative specifications of the statisti-
cal model for analysing dictator game data. It shows how Tobit models (assuming that dictators
would even want to take money) and hurdle models (assuming that the decision to give a posi-
tive amount is separate from the choice of amount, conditional on giving) provide additional in-
sights.
1
1. Introduction
Some 25 years ago, Daniel Kahneman first had students in his classroom play an ultimatum
game (Güth, Schmittberger et al. 1982), which was then followed by the question:
“You will be matched at random with two other students, and you will get to share
some money with one or both of them. If the two people made different decisions
in the first stage (e.g. one of them took $10 and one took $18), then you must
make a decision about how to allocate the money. Call the person who took $10
and gave the other one $10 student E (for even). Call the person who took $18 and
gave the other one $2 student U (for uneven). Your choices are as follows: you
may allocate $5 to yourself, $5 to student E, and nothing to student U; or you may
allocate $6 to yourself, nothing to student E, and $6 to student U”.
74 % of their participants chose the first option although this cost them $1 (Kahneman,
Knetsch et al. 1986:S290 f.). This is how a literature started to which in the meantime 129
contributions have been published, testing a total of 616 different treatments.1 The experi-
mental paradigm has proven so powerful precisely because it is so simple. Actually the suc-
cessors of Kahneman have further simplified the game (starting with Forsythe, Horowitz et al.
1994). The game now is typically stripped of the third party punishment component and
played in isolation. The dictator's action space is complete, so that she may distribute the pie
at her will between the recipient and herself.
Kahneman invented the game as part of his programme that turned textbook assumptions into
behavioural hypotheses. While normally a sizeable fraction of participants does indeed give
nothing, as predicted by the payoff maximisation hypothesis, only very rarely this has been
the majority choice. It by now is undisputed that human populations are systematically more
benevolent than homo oeconomicus. Later experiments have explored this predisposition in
two dimensions: situational and demographic. The former implicitly sticks to the claim that, at
least at the population level, behavioural dispositions are human universals. It engages in re-
fining the conditions under which benevolence is to be expected. The latter puts the research
question upside down and uses the extremely simple design as a tool for quantifying system-
atic behavioural differences between populations.
The dictator game has become popular among experimentalists. In the one year of 2008, 30
new papers with this game have been published. It therefore is time to take stock, and to make
the existing body of evidence accessible. Yet this paper aims beyond merely providing orien-
tation. Implicitly, through their design choices, experimenters have generated data on inde-
pendent variables they have not explicitly set out to test. They for instance have played a one-
shot game with students, asking dictators to divide a pie of $10 given to them between them-
selves and an anonymous recipient from the same subject pool. What looks like a perfectly
standard dictator game implicitly provides data on one shot versus repeated games; on games
with students versus other populations; on manna from heaven versus earned money; on
stakes; on a specified degree of social distance; on dictator-recipient anonymity versus dicta-
1 For detail, see the list of papers used for the meta study in Appendix 1.
2
tor identification. This evidence is untapped as yet. It is useful for two purposes. The basis for
testing the effect of isolated manipulations becomes much broader. More interestingly even, it
becomes possible to control for alternative explanations to a degree that by far transcends
what is feasible in individual experimental studies. That way one learns which effects are ro-
bust, and how big effects are once one controls for other factors that have been shown to mat-
ter for the willingness of dictators to give.
The remainder of this paper is organised as follows. Section 2 explains how the sample has
been collected and addresses analytic methodology. Section 3 treats all experiments as contri-
butions to one question: how much are dictators willing to give? Section 4 focuses on individ-
ual experimental manipulations. Section 5 simultaneously uses all these independent variables
to explain generosity with multiple regressions. In this section I also discuss different options
for specifying the statistical model. Section 6 concludes.
Two papers have engaged in a similar exercise. In 2003, Camerer has done a meta-study of 11
experiments (Camerer 2003:57 f.). In 2008 Cardenas and Carpenter have done the same for
2 The database PsychInfo reports 4,685 entries with the word „meta analysis” in the title of the paper.
3 The database EconLit reports 491 entries with the keyword „meta analysis”, most of which refer to work
in econometrics.
3
10 experiments playing the dictator game in developing countries (Cárdenas and Carpenter
2008:317). The present meta-study covers all 129 papers published between 1992 and the end
of 2009, including 4 papers to come out in 2010 but already available through advance access.
4 papers do not report sample size. The remaining papers cover a total of 41,433 observations.
The search for papers has started with the two predecessor studies. The keyword “dictator
game” produced 89 hits in the EconLit database, and 240 hits in RePEc. Finally I have
checked the references of the papers thus found. Unsurprisingly, these sources frequently
pointed to the same publications. Moreover, in particular but not only in RePEc, not so rarely
one and the same publication is listed repeatedly within the same database, usually since ear-
lier and later versions are kept. Subtracting duplicates, the resulting gross sample includes 255
papers. I have excluded 76 papers since they test a different game. I in particular have left
aside papers that give the recipient any kind of power (and thereby bring the experiment close
to an ultimatum game); papers (like (Kahneman, Knetsch et al. 1986) that started the litera-
ture) that mix the motive of benevolence with one of sanctioning the recipient; twelve papers
that limit the dictator's action space such that it does not include the option to give nothing
(again like (Kahneman, Knetsch et al. 1986)). Six experiments play the right game, but ask a
different research question. They for instance want to know how much unendowed observers
expect dictators to give (Aguiar, Branas-Garza et al. 2009). Another 15 papers do not report
the data such that it can be used for meta-analysis. Seven papers have no original data. Anoth-
er five are surveys or (the two) meta-studies. 20 are pure theory papers. One is a pure econo-
metrics paper. The papers included in the final sample comprise a total of 616 treatments. I
deliberately include working papers, even if they have never been published, to guard against
publication bias (as recommended by Thompson and Pocock 1991).
In dictator games, the dependent variable is (quasi)continuous. Dictators may give any frac-
tion between 0 and 100 % of their endowments, albeit frequently only in integers of $10.
Classic meta-analysis works with one datapoint per experiment and condition. In my case, this
is the mean fraction of the pie that dictators give recipients. For 498 treatments, this measure
has been reported. For the remaining treatments, it can be calculated from the pie size and
information on individual contributions.
While all experiments reported in this paper play a dictator game, they use this game to test a
rich array of research questions. To do so, they manipulate a host of context factors. Given
manipulations differ, one should expect differences in outcome. More importantly, heteroge-
neity across studies can be traced back to explicit differences in experimental design, includ-
ing factors held constant in the individual experiment, but different across experiments. The
appropriate technique for capturing this observed heterogeneity is meta regression (Harbord
and Higgins 2008).
4
Meta-analysis exploits the fact that individual studies have different size and different vari-
ance.4 So called fixed effects meta-regression weights each data point by the respective stand-
ard error.5 The weight thus reflects the precision of the information from each study. Yet in
my sample standard deviations or standard errors are only reported for 191 treatments. Fortu-
nately, for another 254 treatments, the standard error can be reconstructed from information
on the distribution of give rates. Frequently, this information is presented graphically, as a
histogram or as a cumulative distribution, with no exact numbers reported. I then have
checked whether my measurement of the height of bars is sufficiently precise. I have accepted
the outcome if numbers add up to a result within [.95 1.05]. I thus have accepted a translation
mistake of at most 5%. This procedure seemed more reliable than ex post readjusting the vis-
ual impression such that the numbers add up to exactly 1.
Fixed effects meta-analysis assumes that all studies are taken from a sample with the same
residual error. So called random effects meta-analysis relaxes this assumption. It takes the
possibility into account that different experiments not only differ by observed design features,
but also by unobserved factors. Following the procedure proposed by (DerSimonian and Laird
1986), random effects meta-regression allows the standard error of each study to itself result
from a random draw. It thus works with model
yi x i ' u ei
where yi is the fraction of the pie one participant gives the recipient, x i ' is a vector of ex-
planatory variables with its coefficient vector, u ~ N (0, 2 ) is the between studies error, with
variance 2 , and ei ~ N (0, i 2 ) is the within study (residual) error. Consequently, each study is
weighted with 1 / i 2 2 (Harbord and Higgins 2008). The weight thus also reflects how
much the respective treatment contributes to explaining the between studies variance.6 This
procedure mitigates the loss in information resulting from unobserved heterogeneity across
studies.
Normally, meta-analysis cannot go any further. Fortunately, the literature on dictator games
frequently not only reports means and standard errors, but also distributions. The measure is
slightly imprecise though. Even if dictators’ action space was not limited to integers from 0 to
10, usually only deciles are reported. Moreover, I mostly had to translate graphs into numbers
for the purpose, with a small risk of translation error. With these limitations, based on distri-
bution information and on the N of the respective study, for those 83 papers with complete
distribution information, I am able to reconstruct the original data, resulting in 20,813 obser-
vations.7 The procedure to do so is straightforward. If a publication reports 100 observations
4 While this is standard in the neighbouring disciplines, weighting by a measure for precision is still unusu-
al in experimental economics, but see (Croson and Marks 2000; Zelmer 2003) who weight by the number
of observations.
5 Terminology slightly differs from the terminology that is standard in econometrics. The „fixed effects”
model actually does not have an individual specific effect in the model, but exclusively captures differ-
ences between studies by the weight.
6 For the restricted maximum likelihood estimator of 2 , see (Harbord and Higgins 2008).
7 (Prante, Thacher et al. 2007; Hopfensitz 2009; Weizsäcker 2010) also work with meta datasets.
5
per treatment, and if in the first treatment 30% have kept their entire endowment, this results
in 30 reconstructed datapoints with the independent variables from this treatment and 0% giv-
ing as the normalized dependent variable. Note that I have full information of independent
variables in that they are fix per treatment.
I use this second data set for complementary analysis. It in particular makes it possible to go
beyond means per treatment as the exclusive dependent variable, and to also investigate dis-
tributions. If I work with this data set, I have to account for the fact that each study has gener-
ated multiple data points. I use two complementary approaches. Clustering at the level of
treatments corrects standard errors for the potential lack of independence. Treatment dummies
control for unobserved heterogeneity.8 Along with the analysis based on individual data, clas-
sical meta-analysis is always reported since it is more conservative, and since it can cover
more papers.
The reconstructed meta-dataset provides a further opportunity. As is patent from the graphs
usually reported in dictator game manuscripts, giving is not distributed normally. A substan-
tial fraction tends to give nothing. Frequently there is a second peak at the equal split. The
rich dataset that reflects characteristic distributions makes it possible to discuss alternative
statistical approaches for analysing this data.
3. Overall generosity
If one calculates the grand mean from all reported or constructed means per 616 treatments,
dictators on average give 28.35 % of the pie. As Figure 1 demonstrates, the distribution of
means is left skewed. Dictators are more likely to give little. In 6 of 616 treatments, they on
average give zero. Only in one of these six cases, this results from the fact that dictators had a
take option.9
8 This approach has also been used by (Croson and Marks 2000; Hopfensitz 2009).
9 In (Brosig, Riechmann et al. 2007) giving nothing established the equal split between both players’ en-
dowments; in the pertinent treatment of (Leider, Möbius et al. 2009) social distance was high and effi-
ciency of giving low; in (Oxoby and Spraggon 2008) the dictator had earned the pie.
6
.15
.1
Fraction
.05
0
0 .2 .4 .6 .8 1
mean
unweighted, N = 616
Figure 1
Distribution of Mean Giving per Treatment
Finally, using the supplementary dataset with reconstructed individual observations, we learn
that contributions are very unevenly distributed over the unit interval, see Figure 2. 36.11 %
of all participants give nothing to the recipient. 16.74 % choose the equal split. As many as
5.44 % give the recipient everything.
10 The result from fixed-effects meta-analysis differs though. If weights are not corrected for the contribu-
tion of a treatment to explaining between studies variance, the estimated give rate goes down to 20.4 %.
11 In technical terms, parameter 2 .
7
.4
.3611
.3
Fraction
.2
.1674
0 .2 .4 .6 .8 1
give
data only from 328 treatments with full range information, N = 20813
Figure 2
Distribution of Individual Give Rates
4. Explanatory Factors
Experiments are designed to test the effect of one, maybe two or three, manipulations. In the
motivation of their papers, experimentalists link their tests to the related literature. Yet if ex-
periments are not meant to test formal theory, which is normally not the case in dictator
games, it is not always clear how one effect is related to another. In a fully satisfactory way,
these links could only be built if the mediating factors, from the environment or from partici-
pants’ attitudes, were isolated; this would require new experiments. Here, I must content my-
self with ex post generating order in this literature. In so doing, I start with those manipula-
tions that are closest to standard economic theory. A number of experiments have slightly
altered the incentive structure (a). Others have given the dictator all the power, but have ex-
posed her to social control. This tests the power of social expectations and of the risk of social
sanctions (b). Money maximising agents only care about their own payoff. Many experi-
mental games point to the fact that participants are also sensitive to relative, not only to abso-
lute payoffs. Many dictator games have explored the motivating force of payoff comparisons
by making the dictator and the recipient differently deserving (c). Some experiments have
introduced context by framing the situation. Since context defies standardisation, for this me-
ta-study only one manipulation is analysed in detail: the effect of having participants handle
real coins or notes (d). I may be willing to give to the victims of an earthquake at the opposite
side of the globe, and my neighbours may have annoyed me so much in the past that they
would be the least to whom I would want to be generous. Yet if such contextual factors are
excluded by design, one would expect that people give the less, the higher the social distance
between themselves and the recipient (e). Finally, demographics like age, gender, race, or the
country of origin matter (f). Many of these explanatory variables are dummies, some are ordi-
nally scaled, some are continuous. For detail, the reader is referred to the descriptive statistics
in Appendix 2.
8
As is typical for meta-studies, these independent variables are not distributed equally across
the (meta) sample. Moreover some independent variables are not reported in all studies. For
detail, the interested reader is again referred to Appendix 2. Since the dataset is rich, despite
the imbalance many effects turn out significant. In this section, I introduce the independent
variables one by one and present tests. Even if effects are significant, once one controls for
study heterogeneity, the additional explanatory power due to a single explanatory variable is
usually fairly low. I nonetheless report regressions throughout. Knowing the marginal effect
and its significance level is more informative than just descriptive statistics, which would be
the alternative. Also, low explanatory power is what one should expect. All models aim at
explaining variance in giving in the entire dataset. Yet the variance of individual explanatory
factors across studies is low for most individual explanatory factors. As laid out in Appendix
2, most individual explanatory factors are distributed very unevenly in this dataset. Therefore
frequently a small number of datapoints have to provide all the explanation. In Section 5,
models are presented that simultaneously control for multiple explanatory factors. They ex-
plain a much higher fraction of the variance.
a) Incentives
In most dictator games, the action space is quasi continuous. For instance the stake is $10, and
dictators can give any amount of integer dollars they deem fit, but they usually cannot split
the amount by cents. Sometimes, experimenters have (further) constrained the action space.
They for instance have excluded the equal split, or they have only given dictators a choice
between keeping everything and contributing half of the pie (both is, e.g. used in Bolton, Ka-
tok et al. 1998). Meta-regression with this explanatory variable is insignificant. However if
one uses individual data, limiting the action space has a significant positive effect: OLS, with
fixed effects for each treatment (not reported), cons .342***, limited action space .121***, N
= 20,813, additional adj.R2 <.001.12
In the standard dictator game, the dictator may be sure that her decision determines her pay-
off. Some papers have introduced uncertainty about the dictator's own payoff. A lottery de-
termines whether the dictator's choice is implemented, or whether it is replaced by a random
draw (a key contribution is Andreoni and Bernheim 2009). Again meta-regression is insignifi-
cant. However using individual data, one establishes a significant negative effect. The less the
dictator is sure that intended benevolence becomes effective, the less she gives in the first
place: OLS, cons .342***, degree of uncertainty -.265***, N = 20,813, additional adj.R2
<.001.
12 All OLS models have heteroskedasticity robust standard errors. Throughout the paper, significance levels
are reported by *** p < .001, ** p < .01, * p < .05, + p < .1. Since the model contains a regressor for each
treatment of an experiment, the overall adjusted R2 (which is .230) is not informative. Instead the increase
in the adjusted R2 resulting from adding the regressor(s) in question is reported. In the concrete instance it
is close to 0, despite the fact that the regressor is highly significant, t = 4.59, p < .001.
9
Normally, there is just one dictator. What if more than one dictator competes? The key contri-
bution to this question is a study with school children. After the whole class had played, the
payoff ranking was posted on the blackboard, and the dictator with the highest payoff got an
additional treat (Houser and Schunk 2009). Once more, meta-regression is insignificant, as is
a regression with individual data.13
It is standard in dictator games to pay out each individual choice. Yet if experimenters have
repeated the game, or if they wanted to use high stakes, they sometimes have only paid some
choices, or some dictators for that matter, at random. Occasionally, experimenters have not
used pecuniary incentives at all, and have just asked the hypothetical question how much par-
ticipants would be happy to give, were they to have the money. Again, the effect of this ma-
nipulation is insignificant in meta-regression and when using individual data.14
The standard stakes in a dictator game are $10. Especially when repeating the game, or when
using the strategy method (Selten 1967), stakes can be fairly low. An additional challenge
stems from the fact that many experiments have been run in countries with a national currency
other than the dollar. If papers have indicated the exchange rate at the time of the experiment,
I have used the dollar equivalent. Nonetheless, for this independent variable, I only have 440
(rather than 616) treatments. Stakes range from $0 to $130, with mean $21.77. This time, both
meta-regression and a regression with the original data are insignificant. Meta-regression is
significant, though, if one reduces the sample to those 158 treatments that have manipulated
stakes. Higher stakes reduce the willingness to give. If there is more to gain, dictators keep
more, not only in absolute, but also in relative terms (although the effect is very small): meta-
regression, cons .326***, stake -.002**, N = 158, adj.R2 .036.
The standard dictator game is one shot. Some experimenters have repeated the game, but
changed recipients every round. Others have used the strategy method. I have coded both ma-
nipulations as multiple tests. Multiple tests have a clear negative effect on giving: meta-
regression, cons .297***, multiple tests -.071**, N = 445, adj.R2 .058.
What if the dictator is not an individual, but a group? Again, both meta-regression and re-
gression with individual data are insignificant.15
13 If one does not add study dummies (but clusters standard errors for treatments), dictators give less if they
are under competitive pressure, OLS, cons .273***, dictator competition -.074*, N = 20,813, adj.R2
.0008.
14 If one does not add study dummies (but clusters standard errors for treatments), one finds a significant
negative effect for random pay, while hypothetical payment is still not significantly different from paying
out every choice, OLS, cons .377***, random pay -.117***, hypothetical pay -.045, N = 20,813, adj.R2
.010.
15 There is a highly significant effect when using individual data and not adding treatment dummies (but
clustering standard errors at the treatment level). Groups then are more selfish than individuals: OLS,
cons .274***, group decision -.083***, N = 20,813, adj.R2 .002.
10
b) Social Control
Many experimenters have wondered whether dictator greed would disappear with increased
social control. The most obvious way to increase social control is making recipients identify
the person in whose hands they are. One for instance makes dictators stand up, so that recipi-
ents observe who decides (Frey and Bohnet 1995). Here, meta-regression yields a surprising
result. The effect of identification is insignificant. But it becomes significant if one controls
for stakes. Yet then the effect is negative: meta-regression, cons .283***, identification -
.060*, stake .0002, N = 346, adj.R2 .045. At the level of individual data, the effect reverses
and thus is strongly positive, as one would expect: OLS, cons .342***, identification .658*, N
= 20,813, additional adj.R2 <.001. Comparing distributions, one sees the likely reason for the
surprising result of meta-regression, Figure 3. If dictators are identified, this has three effects:
they are less likely to give nothing; the mode shifts from 0 to .5; dictators are less likely to
give more than half of the pie. Seemingly, even in the dictator game benevolence may crowd
out.
no identification identification
.4
Fraction
.2
0
0 .5 1 0 .5 1
give
Figure 3
Effect of Dictator Identification
based on individual data
A less intrusive form of increasing social control is giving the dictator a social cue. For in-
stance one study has shown part of the dictators three large dots that were arranged in a way
that evokes a face (Rigdon, Ishii et al. 2009). Since standard dictator games are devoid of any
social cues, it is meaningful to compare them. While the effect is not significant in meta-
regression, it is when using individual data (and controlling for study heterogeneity), OLS,
cons .342***, social cue .121***, N = 20,813, additional adj. R2 <.001.
Other experimenters have not increased, but reduced the degree of social control, compared to
the standard design of dictator games. To that end, they have given the dictator a chance to
hide her decision, such that the recipient does not learn (for sure) that her payoff is due to a
11
dictator's decision (see e.g. Dana, Cain et al. 2006). This reduces dictators’ generosity: meta-
regression, cons .288***, concealment option -.077*, N = 445, adj.R2 .007.
Early on, experimenters were concerned that dictator-recipient anonymity would be insuffi-
cient to measure true benevolence. They were afraid that participants would give because they
were aware the experimenter would learn their choices. Fairly elaborate procedures have been
invented to also guarantee dictator-experimenter anonymity (starting with Hoffman, McCabe
et al. 1994). Both in meta-regression and when using individual data, such doubleblind proto-
cols do not have a significant effect. This changes if one controls for one shot versus repeated
interaction. One then finds a weakly significant, small reduction of generosity: meta-
regression, cons .309***, repeated interaction -.082***, doubleblind -.036+, N = 445, adj.R2
.066. This suggests that the choice between single and double blind designs matters only in
more complicated designs.
c) Distributive Concerns
Ultimately, a dictator game is about distribution. One should therefore expect that dictators
give more if the recipient’s claim to her fair share is made more legitimate. One way of ma-
nipulating legitimacy is property rights. In the standard dictator game, the dictator receives
money from the experimenter and is free to give a fraction to the recipient. What if, instead,
the money is allocated to the recipient, but the dictator is free to take a fraction (as first inves-
tigated by Eichenberger and Oberholzer-Gee 1998)? While this does not have a significant
effect in meta-regression, it does when using individual data, and controlling for study hetero-
geneity, cons .342***, take option -.342***, additional adj.R2 <.001. The constant and the
regressor almost perfectly neutralize each other, indicating that, when they have the option to
also take money, on average dictators leave the recipient with no gain.
In the standard dictator game, the recipient is not needy. Many experiments have checked
how generosity increases if it is made clear that the recipient is deserving. They for instance
have asked dictators whether they were happy to make a donation to a charity. This has a
clear and strong effect: meta-regression cons .261***, deserving recipient .115***, N = 445,
adj.R2 .075. Figure 4 demonstrates that this can be decomposed into two effects. If the recipi-
ent is deserving, only half as many dictators keep everything. More than 20% give everything.
The difference in distributions is significant: Epps Singleton, p < .0001.16
16 On methodology see (Epps and Singleton 1986; Goerg and Kaiser 2009). The test is preferred over Kol-
mogorov Smirnov since it has higher statistical power.
12
ordinary recipient deserving recipient
.4
Fraction
.2
0
0 .5 1 0 .5 1
give
Figure 4
Giving to a Deserving Recipient
Dictators do take even less if the recipient had earned the money into which they are now free
to tap: meta-regression, cons .286***, recipient earned money -.059, takeoption -.084,
earned*takeoption .616+, N = 445, adj.R2 .008.
The standard dictator game is a zero sum game. One dollar won for the recipient is one dollar
lost for the dictator. Quite a few experiments have turned the game into a positive sum game.
They for instance have stipulated that one dollar lost for the dictator increases the recipient's
payoff by three dollars. Other experiments have reduced the efficiency of giving, so that one
dollar lost for the dictator creates less than a dollar for the recipient. These manipulations
have a significant effect in the expected direction. The larger the multiplier, the more dictators
give, and vice versa: meta-regression, cons .248***, multiplier (fraction or multiple of 1)
.030*, N = 445, adj.R2 .016.
What if there is more than one recipient? One might have thought that the presence of a se-
cond recipient serves as an excuse for giving little to both. Actually, the opposite is true. If
there is another recipient, both receive substantially more: meta-regression, cons .280***,
multiple recipients .112*, N = 445, adj.R2 .022.
In the standard dictator game, the recipient is poor while the dictator is rich. If the recipient
also receives an endowment upfront (as, for instance, in Eckel, Grossman et al. 2005), this
strongly reduces giving: meta-regression cons .291***, recipient endowment -.213***, N =
445, adj.R2 .052. As Figure 5 shows, if the recipient has received a positive endowment at the
start of the interaction, the reduction is almost perfectly proportional to the size of the en-
dowment; datapoints almost perfectly lie on the regression line.
13
1
.8
.6
mean
.4
.2
0
0 .2 .4 .6 .8 1
recipient endowment
Figure 5
Effect of Upfront Endowment to Recipient
Perceived distributive fairness is also altered if the pie is not manna from heaven, but if the
dictator had to earn it. This strongly reduces her generosity: meta-regression, cons .295***,
earned pie -.169***, N = 445, adj.R2 .084. The mere fact that the dictator is ostentatiously
entitled with the pie does not have a significant effect in meta-regression. With original data,
and controlling for study heterogeneity, there is a small, weakly significant effect in the ex-
pected direction: cons .342***, entitlement -.098+, additional adj.R2 <.001.
d) Framing
The standard dictator game is presented as a context free opportunity structure, giving the dic-
tator all the decision power, and the recipient none. Experimenters have added a variety of
frames to the dictator game, like buying and selling a commodity (Hoffman, McCabe et al.
1994). Individual frames can be quite influential (see e.g. Branas-Garza 2007). Yet the ab-
stract fact that a frame has been added does not have a significant effect; each frame is differ-
ent. Another manipulation that is related to framing has a remarkably strong effect though. If
dictators handle coins or notes, they give substantially more: meta-regression, cons .268***,
real money .075***, N = 445, adj.R2 .023. As Figure 6 demonstrates, the effect has two
sources. If they handle real money, participants are less likely to give nothing. More remarka-
bly even, they become much more likely to give everything. The difference in distributions is
significant: Epps Singleton, p < .0001.
14
no coins or notes coins or notes
.4
Fraction
.2
0
0 .5 1 0 .5 1
give
Figure 6
Effect of Handling Real Money
e) Social Distance
In the standard dictator game, it is common knowledge that the dictator and the recipient are
randomly selected members of the same student subject pool. Experimenters have manipulat-
ed the degree of social distance, ranging from a total stranger over another member of the sub-
ject pool to a member of a more closely defined group, to the friend of a friend of a friend, to
the friend of a friend, to an immediate friend (this has been an active area of research recently,
see Leider, Möbius et al. 2009; Brañas-Garza, Cobo-Reyes et al. 2010; Goeree, McConnell et
al. 2010).17 In the meta-regression, the degree of social distance has a surprising negative ef-
fect. The model predicts that dictators give the less the closer the recipient: meta-regression,
cons .355***, degree of social distance -.063***, N = 445, adj.R2 .098. Inspecting Figure 7,
one should become suspicious though. If dictators and recipients are any closer than being
members of the same subject pool, their generosity is practically stable. The significant effect
is a statistical artefact, driven by other explanatory factors inducing ordinary members of the
subject pools to contribute substantially more. This of course strongly indicates that multiple
regression is preferable over single regression.18
17 Note that social distance measures proximity, not distributive concerns. If, for instance, dictators are in-
vited to give to a charity, social distance is high, but there is a higher than ordinary social expectation of
giving.
18 If one controls for unobserved heterogeneity, in multiple regression one recovers the positive effect of
social proximity found in those studies that explicitly manipulated this factor, see below section 5.
15
1
.8
.6
mean
.4
.2
0
0 1 2 3 4 5
degree of social distance
Figure 7
Effect of Social Distance
f) Demographics
The normal participant of a dictator game is a student. On average, non-students give much
more: meta-regression, cons .398***, student -.151***, N = 445, adj.R2 .153. As Figure 8
shows, students are much more likely to give nothing, and they are much less likely to choose
the equal split, or to even give everything. The difference in distributions is statistically sig-
nificant: Epps Singleton, p < .0001. These are remarkable findings. Experimental economists
have often been critized for generating artificial findings with a subject pool that has little to
do with the much harsher reality of economic relations (see e.g. Levitt and List 2007). At least
in dictator games, the opposite turns out to be true. Student experiments underestimate the
deviation from the textbook prediction.19
non-student student
.4
Fraction
.2
0
0 .5 1 0 .5 1
give
Figure 8
Comparing Students to Non-Students
19 Note, however, that the effect reverses in model 3 of Table 1, i.e. when controlling for other explanatory
factors and for unobserved heterogeneity by way of treatment dummies.
16
Starting with (Eckel and Grossman 1998), many have tested gender effects. Since in ordinary
papers on dictator games gender is not reported, meta-regression with all data would not be
meaningful. If one confines the sample to those papers that have explicitly tested gender, it
turns out that women give significantly more: meta-regression, cons .212***, female .058*, N
= 12, adj.R2 .106.
If one does the same with dictator race, there is no significant effect. For this independent var-
iable, meta-regression is the only option since no paper that has manipulated dictator race has
also reported distributions.
Women do not only give more in dictator games, they also get more as recipients. In a meta-
regression confined to those experiments that have explicitly tested recipient gender, this fac-
tor alone explains 73.2 % of the observed variance, cons .052, female recipient .150***, N =
39. If one controls for recipient gender, dictator gender is insignificant, cons .041, female dic-
tator .016, female recipient .143*, N = 33, adj. R2 .7203. Recipient race does not have a sig-
nificant effect.
Anthropologists have frequently used the dictator game as a technology for quantifying the
sociability of developing or primal cultures (the classic study is Henrich and Boyd 2005). If
one codes Western subject pools as 0, pools from developing countries as 1, and pools from
primal societies as 2, one establishes a remarkable significant positive effect: meta-regression,
cons .275***, society characteristic .053**, N = 445, adj.R2 .024. The more a society is pri-
mal, the more dictators are willing to share. If one treats the two non-Western origins as cate-
gorical variables, seemingly the difference between Western and developing countries disap-
pears, while the difference between developed and primal societies is all the stronger: meta-
regression, cons .275***, developing country .050, primal society .108**, N = 445, adj.R2
.020.
As
Figure 9 demonstrates, this impression is misleading. In Western societies, the typical picture
from student populations is replicated. Close to 40 % of participants give nothing, less than 20
% choose the equal split, some 5 % give everything. Both in developing countries and in pri-
mal societies, giving more than 50% is rare. However, in primal societies the equal split is the
mode, and giving little is rare. Developing countries are in the middle. Giving nothing is
much less frequent than in Western societies, but much more frequent than in primal societies.
Contribution patterns basically distribute over the range [0, .5]. All bilateral comparisons of
distributions are statistically significant: Epps Singleton, p < .0001.
17
Western developing
.4
.2
0
Fraction
0 .5 1
primitive
.4
.2
0
0 .5 1
give
Figure 9
Society of Origin
Age also has a strong effect. If one codes children with 0, students with 1, middle-aged adults
with 2, and the elderly with 3, there is a highly significant, substantial effect: meta-regression,
cons .187***, age .098***, N = 445, adj.R2 .038. If one treats each age class as a categorical
variable, in meta-regression, the behaviour of children is not significantly different from the
behaviour of students, while there is a significant difference with respect to the remaining age
classes: meta-regression, cons .269***, child .036, middle age .138**, elderly .443***, N =
445, adj.R2 .122. Again distributions are more informative than means. Children are unlikely
to give more than half of the pie, and many give less. This explains why there is no difference
in means, compared to students. Yet children are much less likely to give nothing. Giving
nothing is even rarer in participants of middle age, and it never happens in the elderly. For
people of middle age, the equal split is the mode, while for the elderly this is giving every-
thing. All bilateral comparisons of distributions are statistically significant: Epps Singleton,
p < .0001.
18
child student age
.6
.4
.2
Fraction
0
middle age old age
.6
.4
.2
0
0 .5 1 0 .5 1
give
Figure 10
Effect of Age
5. Multiple Regression
In single regression most effects turns out significant. Yet single regression has little explana-
tory power. This becomes patent through the measure for the adjusted R2. In most regressions,
it is below .1, implying that more than 90 % of the variance remains unexplained. Compared
to the meta-analysis of means, which left 97.1 % of the variance unexplained, this is only a
marginal improvement. Multiple regression has a considerably better fit. Table 1 uses comple-
mentary analytic strategies. The complete meta-regression model, i.e. the regression with
weighted study means as the dependent variable, explains almost half of the variance, even if
one adjusts the R2 for the fact that one uses 24 explanatory variables.20
20 In these regressions three explanatory variables are left out since including them would severely reduce
the number of observations: gender, race, and stakes.
19
ols ols
no treat treat truncated
meta-regression dummies dummies tobit logit 0 ols logit 50 logit 100
limited action space -0.062+ 0.038 -0.029 0.027 0.281 0.131** -1.093** 0.05
degree of uncertainty -0.036 -0.07 -0.654*** -0.068 0.303 -0.2 0.584 0
incentive -0.01 -0.040** -0.052** -0.067*** 0.418** -0.004 -0.305* -0.142
repeated game -0.064** -0.018 -0.130* -0.024 0.037 -0.025 -0.566*** 0.409
group decision -0.054+ -0.108* -0.015 -0.103+ -0.07 -0.210*** -0.097 0
identification 0.042 0.049* 0.243*** 0.077* -0.522* 0.035 0.575** 0.016
social cue 0.005 -0.031 0.225*** -0.033 -0.026 -0.06 0.062 -0.343
concealment -0.065* -0.028 -0.123*** -0.035 0.083 -0.032 -0.147 -0.14
double blind -0.024 -0.021 -0.262*** -0.023 0 -0.028 0.185 -0.447
take option 0.067 -0.037 -0.038 -0.083 0.443 0.043 0.371 0.125
deserving recipient 0.086*** 0.168*** 0.534*** 0.226*** -0.913*** 0.117* -0.520* 1.830***
recipient earned 0.128* 0.169*** -0.132* 0.275*** -0.922** 0.220*** 1.006*** 0.866
efficiency recipient 0.026+ 0.007 -0.180*** 0.022 -0.294** -0.026 -0.394* 0.875**
multiple recipients 0.148*** 0.038* -0.065+ 0.028 0.187* 0.125*** -0.611*** 0.539+
recipient endowment -0.173*** -0.058 0.204** -0.147 0.886* 0.062 -0.795+ 0.248
dictator earned -0.174*** -0.191*** -0.126** -0.374*** 1.489*** -0.213*** -1.556*** 0
real money 0.025 0.062+ 0.216*** 0.076+ -0.141 0.092+ -0.443* 2.058***
degree of social proximity -0.053*** 0.002 0.191*** 0.005 0.036 0.017 -0.832** 2.675***
student -0.104** -0.220*** 0.216*** -0.233*** 0.456 -0.301*** -0.076 -2.185***
child -0.117** -0.195** -0.007 -0.172* -0.311 -0.385*** 0.932+ -6.404***
middle age 0.001 -0.044 0.435*** 0.031 -1.475+ -0.258*** 1.888** -2.434***
old age 0.336*** 0.189** 0.181+ 0.247** 0 0 1.293* -1.384*
developing country 0.015 0.01 0.231*** 0.042 -0.617* -0.052+ 0.211 -1.329*
primal society -0.009 -0.098 -0.047 -0.027 -1.847* -0.329*** 0.492 -4.147***
constant 0.416*** 0.518*** 0.210* 0.460*** -1.377+ 0.651*** 0.556 -5.241***
N 603 20813 20813 20813 20663 13298 20813 19402
adj.R2 / pseudo R2 .485 0.149 (0.242) 0.105 0.074 0.077 0.268
Table 1
Multiple Regression
codes are explained in section 4
20
In meta-regression, not all regressors turn out significant. The degree of uncertainty is insignifi-
cant, as is the kind of incentives, dictator identification, the presence of a social cue, dictator-
experimenter anonymity, a take option, the use of real money, the fact that the dictator has mid-
dle age, and that she comes from a developing country or from a primal society. Figure 11 col-
lects those regressors that remain at least marginally significant if one controls for all other ex-
planatory variables. Old age has by far the strongest positive effect. A fairly strong effect also
stems from having more than one recipient. The remaining three positive variables concern dif-
ferent dimensions of recipient need and legitimacy. By contrast, if the recipient’s desire is only
weakly legitimate, this strongly reduces giving. This may result from the fact that the recipient
had an endowment in the first place, or that the dictator had to earn the pie. In demographic
terms, children and students give less. Dictators also exploit concealment options. Finally they
give less if the game is repeated, if the action space is limited, if groups decide and, surprisingly,
the smaller the social distance to the recipient.
old age
multiple recipients
recipient earned
deserving recipient
efficiency recipient
group decision
repeated game
concealment
student
child
recipient endowment
dictator earned
Figure 11
Significant Regressors in the Complete Model
meta-regression
If one replaces study means by the original data (column 2 of Table 1), one works with 20,813
instead of 445 datapoints. One should expect that this makes it easier to establish significance;
standard errors are the standard deviation, divided by the square root of N, after all. This indeed
holds for dictator identification and for the use of real coins and notes. Both now also have their
positive effect if one controls for all the other explanatory variables. Interestingly, when control-
ling for alternative explanations the effect of incentivising individual choices turns out negative;
dictators give less. The equally negative effect of group decision is now significant at conven-
tional levels. Two more regressors become significant at a more demanding level. Yet there are
also three regressors that were significant in meta-regression, and that become insignificant when
21
using individual choices. Two more regressors are only significant at a lower level. Yet if one
reduces the sample of the meta-regression to those studies that reported distribution information,
one finds the same or even lower significance levels. The differences are thus due to sample at-
trition. As usual, working with individual rather than aggregate data leads to considerably more
unexplained variance; means tend to exhibit greater regularity.
Model 3 also works with the original data, but adds treatment dummies to the model. This has an
advantage and a disadvantage. Any unexplained heterogeneity across studies is “dummied out”.
This yields many more significant regressors. Participants give less the higher the uncertainty.
They give more if there is a social cue. They give less when double blind procedures are used.
They give more if they handle real money, when they have middle age or when they come from
a developing country. All of this is as expected. It is interesting that, in this specification, stu-
dents give substantially more. It is comforting that, in this model, a higher degree of social prox-
imity has the expected positive sign (not a negative one, as in meta-regression). Yet other regres-
sors are surprising. The negative effect of being a child is no longer significant, nor is the
positive effect of old age. Higher efficiency for the recipient has a counterintuitive negative ef-
fect, and the fact that the recipient already holds an endowment has an equally counterintuitive
positive effect. All these surprising effects result from the fact that almost all the variance with
respect to these explanatory variables results from variance between, not within treatments.
Therefore in this specification the effects picked up by both meta-regression and OLS without
treatment fixed effects go into the treatment dummies. For the explanatory variables no variance
is left, or it comes from a few outliers. Consequently eventually both specifications of the OLS
model should be seen as complementary, not as mutually exclusive.
An equally plausible analytic approach assumes that the decision to make a positive contribution,
and the decision how much to give, conditional on the willingness to give at all, are two separate
processes. Then, a hurdle model is appropriate. It first analyses the binary decision to give or not
to give with a logit model. For the size of positive contributions, conditional on giving anything,
21 To allow for direct comparisons with the OLS model, the Tobit model reports marginal effects, not the coef-
ficients of the latent variable.
22
it uses OLS, adjusting the distribution assumption to the fact that observations are taken from a
truncated normal distribution (for background see McDowell 2003).22
From this exercise, one learns that the willingness to give anything is indeed not driven by the
same forces as the choice of the size of the contribution to the recipient. Let us first compare the
logit component with the least squares model of all data. One now finds a strongly significant
effect of the factor by which dictator generosity translates into recipient benefit. The larger this
factor, i.e. the higher the efficiency of a dollar given, the less likely a dictator is to give nothing.
Conversely, the larger the upfront endowment of the recipient, the more a dictator is inclined to
keep everything. In this model, one also reestablishes the difference between Western and devel-
oping countries or primal societies. In developing countries, dictators are less likely to give noth-
ing. In primal societies this is extremely unlikely. Finally there is a fairly pronounced, although
only weakly significant negative effect of middle age.
For the truncated least squares model, two comparisons are of interest: with unconditional least
squares, and with the logit model. The first comparison reveals that dictator-experimenter ano-
nymity significantly reduces the willingness to give, as does middle age and being a member of a
primal society, while a limited action space increases conditional giving; none of these four ef-
fects was significant in the unconditional model. For another four regressors, the significance
level goes up, compared to the unconditional model. This holds true for the negative effect of
group decision and of being a child, and for the positive effect of the recipient having earned her
endowment, and of the dictator being faced with multiple recipients.
Interestingly, in the very same respects truncated OLS also differs from the hurdle equation. The
logit model is either insignificant in these respects, or it has a lower level of significance. This
suggests that the hurdle model indeed outperforms OLS. Apparently, the decision to give nothing
is indeed driven by forces that differ from the conditional choice of a contribution level. Three
regressors only matter for the conditional choice, not for the willingness to give at all. Students
and groups make smaller contributions. If dictators handle real coins and notes, this slightly in-
creases their contributions, conditional on their willingness to give. By contrast, a number of ex-
planatory factors only matter for the willingness to give anything, not for the size of the contribu-
tion. This holds for the question whether each and every choice is paid out, for the factor by
which a dollar given translates into a dollar received, for the question whether the recipient has
her own endowment, and whether the dictator comes from a developing country. Actually, no
more than three regressors significantly explain both the willingness to make a positive contribu-
tion and its size. If the recipient has earned the pie, she is more likely to get something, and the
size of the contribution increases. Interestingly, the remaining two regressors that are significant
in both models predict conflicting effects. Being of middle age and being a member of a primal
society reduces the likelihood of giving nothing, but it also reduces the size of contributions.
Such dictators think they should give something, but they give rather little.
22 The number of observations is, of course, confined to those participants who have made a positive contribu-
tion.
23
While giving nothing is the mode, the distribution of dictator giving may also be described as tri-
modal, with the equal split and giving everything as the two other peaks. This invites two more
logit models. The first tests the hypothesis that a dictator goes for the equal split, against the nul
that she makes another decision. This model too is revealing. The equal split is most popular in
middle age and in primal societies. If the recipient identifies the dictator, this also pushes the
dictator towards the equal split. By contrast, if the game is repeated, or if there are multiple re-
cipients, the equal split becomes considerably less likely.
The final logit model tests the hypothesis that the dictator gives everything, against the nul that
she makes a different choice.23 The strongest force pulling dictators into this direction is social
proximity. Handling real coins and notes and knowing that the recipient is deserving also make
total generosity much more likely. If a dollar given is worth more to the recipient, this also
makes it more likely that the dictator gives all. By contrast, the elderly, dictators from develop-
ing countries and students are quite unlikely to donate the entire pie. This is extremely unlikely
in middle aged dictators, in members of primal societies, and in children. Note that this final
model explains a larger fraction of the variance than any other model working with the original
data.
Table 2 summarises the evidence. In this table, if a cell is empty, the respective regressor does
not have a significant effect. + signs indicate a significant positive effect, – signs indicate a sig-
nificant negative effect. If either meta-regression or least squares at the level of individual data
demonstrates a positive effect, one should expect less dictators to give nothing. This expectation
is borne out for deserving recipients and for recipients who have earned their endowment. By the
same logic, dictators who have earned the pie give less and are more likely to give nothing. One
has a similar picture if dictators handle real coins and notes and if each individual choice is paid,
although here the overall effect is insignificant in meta-regression. The logic is violated by the
effect of deciding for multiple recipients. The overall effect is positive. Nonetheless, dictators are
more likely to give nothing. Yet they are also more likely to give everything. This manipulation
thus polarises choices.
Being a student, being a child or deciding as a group have a significant overall effect, but do not
significantly influence the decision to make a positive contribution. If the action space is limited,
the game is repeated, the dictator can conceal her action, social proximity is different from being
a member of the same subject pool or if the recipient has a positive endowment, this explains
differences in study means, but it does not explain individual choices (in the reduced sample).
Having middle age, being a member of a primal society or of a developing country or providing
dictator-experimenter anonymity does not explain overall effects. Yet these regressors become
significant if one separately analyses the decision to make a positive contribution, and its size.
Manipulating the degree of uncertainty, providing the dictator with a social cue, or giving her a
take option does not have a significant effect once one controls for the other explanatory factors.
23 The number of observations is smaller since with 3 independent variables participants never gave everything.
24
ols ols
no treat treat
meta-reg dummies dummies tobit logit 0 truncols logit 50 logit 100
deserving recipient + + + + – + – +
recipient earned + + – + – + +
old age + + + + + –
multiple recipients + + – + + – +
real money + + + + +
identification + + + – +
efficiency recipient + – – – +
dictator earned – – – – + – –
student – – + – –
child – – – – + –
group decision – – – –
incentive – – – +
limited action space – – –
repeated game – – –
concealment – –
degree of social proximity – + – +
recipient endowment – + +
middle age + – – + –
primal society – – –
developing country + – –
double blind –
degree of uncertainty –
social cue +
take option
constant + + + + – + –
Table 2
Significant Effects in Multiple Regression
6. Conclusion
The main substantive results of this study can be read off Table 2 and Figure 11. The more con-
sistent effects are across specifications of the statistical model (Table 2), the more firmly an ef-
fect is established. The following effects are very robust: If the recipient is deserving, she gets
more. If the dictator has old age, she gives more. If the dictator is identified, she is more gener-
ous. Children give less, as do groups. Other effects only show up when means are replaced by
distributions. If one does, one finds that dictators give more when handling real money, and
when they are identified. If choices are incentivized, this reduces generosity. Yet other effects
only become visible if one further controls for unobserved heterogeneity. One then finds that
dictators give more when they get a social cue, when they are middle aged (rather than being a
student), or when they come from a developing country. Further effects were already visible in
meta-regression, but become apparent in the original data only when controlling for unobserved
heterogeneity. One then finds that dictators give less if the game is repeated, and if they have a
concealment option.
25
For some effects, it is important to separately analyse the decision to give anything, and the
amount given, conditional on passing the hurdle. One then for instance better understands the
effect of identification. If dictators cannot hide behind anonymity, they feel urged to give some-
thing; they do not feel obliged to give more. The same holds if a dollar given is multiplied before
it reaches the recipient. If dictators had to earn the pie, they are significantly more likely to keep
everything. The same holds, once their choices are incentivized and if the recipient also has an
endowment. Middle aged dictators, and those from developing countries or primal societies are
significantly less likely to keep their entire endowments.
Effect sizes are visualized in Figure 11. By far the biggest effect is demographic. Pensioneers
give more than 30 % more than average, children give more than 10 % less. Next in line are dis-
tributional concerns. If the dictator had to earn the pie, or if the recipient has her own endow-
ment, generosity drops by almost 20 %. By contrast, the recipient keeps more than 10 % more if
she had to earn the pie, and she receives some 8 % more if she is visibly deserving. Situational
features have smaller effects. If there are multiple recipients, generosity goes up by more than
10 %. If the game is repeated, if the action space is limited, or if groups decide, generosity goes
down by less than 10 %.
The picture is thus complicated. There is, of course, also a simple message. Originally, Daniel
Kahneman and his co-authors had invented the dictator game to refute the income maximisation
assumption of economics textbooks. If one translates the assumption into the categorical hypoth-
esis that all humans maximise income, their first experiment was sufficient to falsify the hypoth-
esis. After 25 years of experimentation, the original result has not been refuted. Actually, in the
sample of those experiments that provide distribution information, the probability of giving a
positive amount, and hence violating the income maximisation hypothesis, comes close to the
original result. In Kahneman’s experiment, 74 % did not seize the option to gain an additional
dollar. In later dictator game experiments, 63.89 % of all participants, i.e. 13,298 of 20,813 dicta-
tors, made a positive contribution.
Yet if this was everything one wanted to test, 129 experiments would certainly have been an
overkill. The research question becomes richer, and more meaningful for understanding social
interaction, if one turns the finding upside down. While 63.89 % violate the income maximisa-
tion hypothesis, 36.11 % do not. Hence more than a third of a typical population does indeed
consist of subjects who have no reticence to leave a recipient with nothing, although this recipi-
ent is at their mercy. Clearly, generosity is not a human universal. Moreover, those who are will-
ing to make a donation do neither give everything, nor do they split the pie equally. On average,
they give 42.64 %. Hence even those who in principle are generous to a degree exploit the op-
portunity to their advantage. Even generous subjects thus tend to have a selfish side.
Consequently, the ample evidence on dictator games is better read as an exercise in exploring
human heterogeneity. The radically simple design of the game makes it a powerful tool for the
systematic variation of conditions that moderate sociality. As differential psychology, using dif-
ferent experimental approaches, has established long ago (Ross and Nisbett 1991), (perceived)
26
situation and personality interact. In and of themselves, neither situational nor personality varia-
bles suffice to predict behaviour. One must understand how a person with a certain personality
trait reacts to the situation, as she perceives it. Dictator games are helpful for exploring human
sociality, because the situation can be tightly controlled, because the simple game can be played
with all classes of participants and, of course, because decisions are incentivised.
Ironically, the radically simple dictator game provides ample evidence that human sociality is far
from simple. The existing evidence is best compared with a map. After 25 years of expeditions
into human sociality, some major roads have been safely documented. Yet aside those roads,
there are still many white spots. It is therefore safe to predict that dictator game experimentation
will continue to thrive. To date, it is an open question though whether experimenters will ever be
able to draw a near complete map of sociality. It is even less clear whether those relying on this
knowledge in the interest of improving their individual well-being or of making better policy
will be able to correctly infer where an interaction partner is located in this landscape. The better
the driving forces of human behaviour are understood, the more it becomes obvious that institu-
tions are not only needed for overcoming antisocial motives. They are at least as important for
making behaviour predictable (Engel 2005).
27
AGUIAR, FERNANDO, PABLO BRANAS-GARZA, RAMÓN COBO-REYES, NATALIA JIMENEZ and LUIS
M. MILLER (2009). "Are Women Expected to be More Generous?" Experimental Econom-
ics 12: 93-98.
ANDREONI, JAMES and B. DOUGLAS BERNHEIM (2009). "Social Image and the 50-50 Norm. A
Theoretical and Experimental Analysis of Audience Effects." Econometrica 77: 1607-
1636.
BOLTON, GARY E., ELENA KATOK and RAMI ZWICK (1998). "Dictator Game Giving. Rules of
Fairness versus Acts of Kindness." International Journal of Game Theory 27: 269-299.
BROSIG, JEANNETTE, THOMAS RIECHMANN and JOACHIM WEIMANN (2007). Selfish in the
End?:An Investigation of Consistency and Stability of individual Behavior
http://mpra.ub.uni-muenchen.de/2035/1/MPRA_paper_2035.pdf.
CAMERER, COLIN F. (2003). Behavioral Game Theory. Experiments in Strategic Interaction. New
York, NY, Sage u.a.
CÁRDENAS, JUAN CAMILO and JEFFREY CARPENTER (2008). "Behavioural Development Econom-
ics: Lessons from Field Labs in the Developing World." Journal of Development Studies
44: 311-338.
CROSON, RACHEL T.A. and MELANIE MARKS (2000). "Step Returns in Threshold Public Goods.
A Meta- and Experimental Analysis." Experimental Economics 2(3): 239-259.
DANA, JASON, DAYLIAN M. CAIN and ROBYN M. DAWES (2006). "What You Don't Know Won't
Hurt Me. Costly (But Quiet) Exit in Dictator Games." Organizational Behavior and Human
Decision Processes 100: 193-201.
DERSIMONIAN, REBECCA and NAN LAIRD (1986). "Meta-Analysis in Clinical Trials." Controlled
Clinical Trials 7: 177-188.
28
DRUCKMAN, DANIEL (1994). "Determinants of Compromising Behavior in Negotiation. A Meta-
analysis." Journal of Conflict Resolution 38(3): 507-556.
ECKEL, CATHERINE C. and PHILIP J. GROSSMAN (1998). "Are Women Less Selfish Than Men?:
Evidence from Dictator Experiments." Economic Journal 108: 726-735.
ECKEL, CATHERINE C., PHILIP J. GROSSMAN and RACHEL M. JOHNSTON (2005). "An Experi-
mental Test of the Crowding Out Hypothesis." Journal of Public Economics 89: 1543-
1560.
EICHENBERGER, REINER and FELIX OBERHOLZER-GEE (1998). "Rational Moralists: The Role of
Fairness in Democratic Economic Politics." Public Choice 94: 191-210.
ENGEL, CHRISTOPH (2005). Generating Predictability. Institutional Analysis and Institutional De-
sign. Cambrige, Cambridge University Press.
EPPS, T.W. and KENNETH J. SINGLETON (1986). "An Omnibus Test for the Two-sample Problem
Using the Empirical Characteristic Function." Journal of Statistical Computation and
Simulation 26: 177-203.
FORSYTHE, ROBERT, JOEL L. HOROWITZ, N.E. SAVIN and MARTIN SEFTON (1994). "Fairness in
Simple Bargaining Experiments." Games and Economic Behavior 6: 347-369.
FREY, BRUNO and IRIS BOHNET (1995). "Institutions affect Fairness." Journal of Institutional and
Theoretical Economics 151: 286-303.
GOEREE, JACOB K., MARGARET A. MCCONNELL, TIFFANY MITCHELL, TRACEY TROMP and
LEEAT YARIV (2010). "The 1/d law of Giving." American Economic Journal: Microeco-
nomics 2(1): 183-203.
GÜTH, WERNER, ROLF SCHMITTBERGER and BERND SCHWARZE (1982). "An Experimental Anal-
ysis of Ultimatum Bargaining." Journal of Economic Behavior and Organization 3: 367-
388.
HARBORD, ROGER M. and JULIAN P.T. HIGGINS (2008). "Meta-regression in Stata." Stata Journal
8: 493-519.
HARLESS, DAVID W. and COLIN F. CAMERER (1994). "The Predictive Utility of Generalized Ex-
pected Utility Theories." Econometrica 62: 1251-1289.
29
HENRICH, JOSEPH and ROBERT BOYD (2005). "'Economic Man' in Cross-Cultural Perspective.
Behavioral Experiments in 15 Small-Scale Societies." Behavioral and Brain Sciences 28:
795-815.
HOFFMAN, ELIZABETH, KEVIN MCCABE, KEITH SHACHAT and VERNON L. SMITH (1994). "Prefer-
ences, Property Rights, and Anonymity in Bargaining Games." Games and Economic Be-
havior 7: 346-380.
HOPFENSITZ, ASTRID (2009). Previous Outcomes and Reference Dependence. A Meta Study of
Repeated Investment Tasks With and Without Restricted Feedback http://mpra.ub.uni-
muenchen.de/16096/1/Hopfensitz_2009.pdf.
HOUSER, DANIEL and DANIEL SCHUNK (2009). "Social Environments With Competitive Pres-
sure: Gender Effects in the Decisions of German Schoolchildren." Journal of Economic
Psychology 30: 634-641.
HUCK, STEFFEN, HANS-THEO NORMANN and JÖRG OECHSSLER (2004). "Two are Few and Four
are Many. Number Effects in Experimental Oligopolies." Journal of Economic Behavior &
Organization 53: 435-446.
JONES, GARETT (2008). "Are Smarter Groups More Cooperative? Evidence from Prisoner's Di-
lemma Experiments, 1959-2003." Journal of Economic Behavior & Organization 68(3-4):
489-497.
KAHNEMAN, DANIEL, JACK L. KNETSCH and RICHARD THALER (1986). "Fairness and the As-
sumptions of Economics." Journal of Business 59: S285-S300.
LEIDER, STEPHEN, MARKUS M. MÖBIUS, TANYA ROSENBLAT and QUOC-ANH DO (2009). What
Do We Expect from Our Friends?
https://mercury.smu.edu.sg/rsrchpubupload/15345/ExpectFriends.pdf.
LEVITT, STEVEN D. and JOHN A. LIST (2007). "What Do Laboratory Experiments Measuring So-
cial Preferences Reveal About the Real World?" Journal of Economic Perspectives 21:
153-174.
MCDOWELL, ALLEN (2003). "From the Help Desk. Hurdle Models." Stata Journal 3: 178-184.
OOSTERBEEK, HESSEL, RANDOLPH SLOOF and GIJS VAN DE KUILEN (2004). "Cultural Differences
in Ultimatum Game Experiments. Evidence from a Meta-Analysis." Experimental Eco-
nomics 7: 171-188.
OXOBY, ROBERT J. and JOHN SPRAGGON (2008). "Mine and Yours. Property Rights in Dictator
Games." Journal of Economic Behavior & Organization 65: 703-713.
PERCOCO, MARCO and PETER NIJKAMP (2009). "Estimating Individual Rates of Discount. A Me-
ta-analysis." Applied Economics Letters 16(12): 1235-1239.
30
PRANTE, TYLER, JENNIFER A. THACHER and ROBERT P. BERRENS (2007). "Evaluating Coasean
Bargaining Experiments with Meta-analysis." Economics Bulletin 3(68): 1-7.
RIGDON, MARY, KEIKO ISHII, MOTOKI WATABE and SHINOBU KITAYAMA (2009). "Minimal So-
cial Cues in the Dictator Game." Journal of Economic Psychology 30: 358-367.
ROSS, LEE and RICHARD E. NISBETT (1991). The Person and the Situation. Perspectives of Social
Psychology. New York, McGraw-Hill.
SELTEN, REINHARD (1967). Die Strategiemethode zur Erforschung des eingeschränkt rationalen
Verhaltens im Rahmen eines Oligopolexperiments. Beiträge zur experimentellen
Wirtschaftsforschung. Ernst Sauermann. Tübingen, Mohr: 136-168.
STANLEY, T.D. (2001). "Wheat from Chaff. Meta-analysis as Quantitative Literature Review."
Journal of Economic Perspectives 15(3): 131-150.
THOMPSON, SIMON G. and STUART J. POCOCK (1991). "Can Meta-analyses be Trusted?" Lancet
338(8775): 1127-1130.
WEIZSÄCKER, GEORG (2010). "Do We Follow Others when We Should? A Simple Test of Ra-
tional Expectations." American Economic Review 100: 2340-2360.
31
Appendix 1: List of Papers Used for the Meta Study
(Sefton 1992; Forsythe, Horowitz et al. 1994; Hoffman, McCabe et al. 1994; Bohnet and Frey
1995; Frey and Bohnet 1995; Eckel and Grossman 1996; Hoffman, McCabe et al. 1996; Schot-
ter, Weiss et al. 1996; Cason and Mui 1997; Frey and Bohnet 1997; Bolton and Katok 1998; Bol-
ton, Katok et al. 1998; Eckel and Grossman 1998; Eichenberger and Oberholzer-Gee 1998; Ruf-
fle 1998; Selten and Ockenfels 1998; Bohnet and Frey 1999; Anderson, Rodgeres et al. 2000;
Eckel and Grossman 2000; Harbaugh and Krause 2000; Johanneson and Persson 2000; Andreoni
and Vesterlund 2001; Cherry 2001; Fershtman and Gneezy 2001; Frohlich, Oppenheimer et al.
2001; Saad and Gill 2001; Brandstätter and Güth 2002; Cherry, Frykblom et al. 2002; Burnham
2003; Gowdy, Iorgulescu et al. 2003; Harbaugh, Krause et al. 2003; Small and Loewenstein
2003; Ben-Ner, Kong et al. 2004; Ben-Ner, Putterman et al. 2004; Carpenter, Burks et al. 2004;
Cox 2004; Diekmann 2004; Ensminger 2004; Gurven 2004; Marlowe 2004; Song, Cadsby et al.
2004; Carpenter, Verhoogen et al. 2005; Carter and Castillo 2005; Greiner, Güth et al. 2005; Ha-
ley and Fessler 2005; Holm and Danielson 2005; Holm and Engseld 2005; Kamas, Baum et al.
2005; Ashraf, Bohnet et al. 2006; Branas-Garza 2006; Capra and Li 2006; Carpenter, Liati et al.
2006; Cox and Deck 2006; Dana, Cain et al. 2006; Dufwenberg and Muren 2006a; Dufwenberg
and Muren 2006b; Mittone and Ploner 2006; Rankin 2006; Takezawa, Gummerum et al. 2006;
Tan and Bolle 2006; Bekkers 2007; Benenson, Pascoe et al. 2007; Branas-Garza 2007; Broberg,
Ellingsen et al. 2007; Brosig, Riechmann et al. 2007; Cappelen, Hole et al. 2007; Chaudhuri and
Gangadharan 2007; Fisman, Kariv et al. 2007; Fong 2007; Knafo and Israel 2007; List 2007;
Stanton 2007; Vanberg 2007; Whitt and Wilson 2007; Ahmed 2008; Ahmed and Salas 2008;
Asheim, Helland et al. 2008; Bardsley 2008; Bellamare, Kröger et al. 2008; Ben-Ner, Kramer et
al. 2008; Boschini, Muren et al. 2008; Bosco 2008; Cárdenas, Candelo et al. 2008; Cárdenas and
Carpenter 2008; Carlsson, He et al. 2008; Carpenter, Connolly et al. 2008; Castillo and Cross
2008; Charness and Gneezy 2008; Cox, Sadiraj et al. 2008; Farina, O'Higgins et al. 2008;
Gurven, Zanolini et al. 2008; Koch and Normann 2008; Korenok, Millner et al. 2008; List and
Cherry 2008; Mohlin and Johannesson 2008; Oberholzer-Gee and Eichenberger 2008; Oxoby
and Spraggon 2008; Slonim and Garbarino 2008; Stephen and Pham 2008; Swope, Cadigan et al.
2008; van der Merwe and Burns 2008; Yamagishi and Mifune 2008; Yamamori, Kato et al.
2008; Ackert, Gillette et al. 2009; Andrade and Ariely 2009; Andreoni and Bernheim 2009; Barr,
Wallace et al. 2009; Branas-Garza 2009; Branas-Garza, Durán et al. 2009; Branas-Garza and
Ottone 2009; Cadsby, Servátka et al. 2009; Carter and Castillo 2009; Dalbert and Umlauft 2009;
Dickson 2009; Duffy and Kornienko 2009; Fong and Luttmer 2009; Heinrich, Riechmann et al.
2009; Houser and Schunk 2009; Klempt and Pull 2009; Lazear, Malmendier et al. 2009; Leider,
Möbius et al. 2009; Luhan, Kocher et al. 2009; Rigdon, Ishii et al. 2009; Schurter and Wilson
2009; Xiao and Houser 2009; Anderson and Dickinson 2010; Bosch-Domènech, Nagel et al.
2010; Branas-Garza, Cobo-Reyes et al. 2010; Burns 2010; Ellingsen, Johannesson et al. 2010)
32
ACKERT, LUCY F., ANN B. GILLETTE, JORGE MARTINEZ-VAZQUEZ and MARK RIDER (2009). Risk
Tolerance, Self-Interest, and Social Preferences
http://excen.gsu.edu/workingpapers/GSU_EXCEN_WP_2009-04.pdf.
AHMED, ALI M. (2008). Are Religious People More Prosocial? A Quasi-Experimental Study
with Madrasah Pupils in a Rural Community in India
http://ideas.repec.org/p/hhs/gunwpe/0330.html.
AHMED, ALI M. and OSVALDO SALAS (2008). In the Back of Your Mind. Subliminal Influences
of Religious Concepts on Prosocial Behavior
https://guoa.ub.gu.se/dspace/bitstream/2077/18838/4/gupea_2077_18838_4.pdf.
ANDERSON, CLARE and DAVID L. DICKINSON (2010). "Bargaining and Trust. The Effects of 36-h
Total Sleep Deprivation on Socially Interactive Decisions." Journal of Sleep Research 19:
54-63.
ANDERSON, LISA R., YANA V. RODGERES and ROGER R. RODRIGUEZ (2000). "Cultural Differ-
ences in Attitudes Toward Bargaining." Economics Letters 69: 45-54.
ANDRADE, EDUARDO B. and DAN ARIELY (2009). "The Enduring Impact of Transient Emotions
on Decision Making." Organizational Behavior and Human Decision Processes 109: 1-8.
ANDREONI, JAMES and B. DOUGLAS BERNHEIM (2009). "Social Image and the 50-50 Norm. A
Theoretical and Experimental Analysis of Audience Effects." Econometrica 77: 1607-
1636.
ANDREONI, JAMES and LISE VESTERLUND (2001). "Which is the Fair Sex? Gender Differences in
Altruism." Quarterly Journal of Economics 116: 293-312.
ASHEIM, GEIR B., LEIF HELLAND, JON HOVI and BJORN HOYLAND (2008). "Self-serving Dicta-
tors."
ASHRAF, NAVA, IRIS BOHNET and NIKITA PIANKOV (2006). "Decomposing Trust and Trustwor-
thiness." Experimental Economics 9: 193-208.
BARR, ABIGAIL, CHRIS WALLACE, JEAN ENSMINGER, JOSEPH HENRICH, CLARK BARRETT, ALEX-
ANDER BOLYANATZ, JUAN CAMILO CARDENAS, MICHAEL GURVEN, EDWINS GWAKO, CAR-
OLYN LESOROGOL, FRANK MARLOWE, RICHARD MCELREATH, DAVID TRACER and JOHN
ZIKER (2009). Homo Æqualis: A Cross-Society Experimental Analysis of Three Bargain-
ing Games http://ssrn.com/abstract=1485862.
BEKKERS, RENÉ (2007). "Measuring Altruistic Behavior in Surveys. The All-or-Nothing Dictator
Game." Survey Research Methods 1: 139-144.
33
BELLAMARE, CHARLES, SABINE KRÖGER and ARTHUR VAN SOEST (2008). "Measuring Inequity
Aversion in a Heterogeneous Population Using Experimental Decisions and Subjective
Probabilities." Econometrica 76: 815-839.
BEN-NER, AVNER, FANMIN KONG and LOUIS PUTTERMAN (2004). "Share and Share Alike? Gen-
der-pairing, Personality, and Cognitive Ability as Determinants of Giving." Journal of
Economic Psychology 25: 581-589.
BEN-NER, AVNER, AMIT KRAMER and ORI LEVY (2008). "Economic and Hypothetical Dictator
Game Experiments. Incentive Effects at the Individual Level." Journal of Socio-Economics
37: 1775-1784.
BEN-NER, AVNER, LOUIS PUTTERMAN, FANMIN KONG and DAN MAGAN (2004). "Reciprocity in
a Two-part Dictator Game." Journal of Economic Behavior & Organization 53: 333-352.
BENENSON, JOYCE F., JOANNA PASCOE and NICOLA RADMORE (2007). "Children’s Altruistic Be-
havior in the Dictator Game." Evolution and Human Behavior 28: 168-175.
BOHNET, IRIS and BRUNO FREY (1995). "Ist Reden Silber und Schweigen Gold?" Zeitschrift für
Wirtschafts- und Sozialwissenschaften 115: 169-209.
BOHNET, IRIS and BRUNO FREY (1999). "Social Distance and Other-Regarding Behavior in Dic-
tator Games. Comment." American Economic Review 89: 335-339.
BOLTON, GARY E. and ELENA KATOK (1998). "An Experimental Test of the Crowding Out Hy-
pothesis. The Nature of Beneficent Behavior." Journal of Economic Behavior & Organiza-
tion 37: 315-333.
BOLTON, GARY E., ELENA KATOK and RAMI ZWICK (1998). "Dictator Game Giving. Rules of
Fairness versus Acts of Kindness." International Journal of Game Theory 27: 269-299.
BOSCHINI, ANNE, ASTRI MUREN and MATS PERSSON (2008). Constructing Gender in the Eco-
nomics Lab http://www.ne.su.se/paper/wp09_15.pdf.
34
BRANAS-GARZA, PABLO (2009). "Personal Identity. A Theoretical and Experimental Analysis."
Journal of Economic Methodology ***: ***.
BRANAS-GARZA, PABLO, MIGUEL A. DURÁN and MARÍA PAZ ESPINOSA (2009). "The Role of
Personal Involvement and Responsibility in Unfair Outcomes." Rationality and Society 21:
225-248.
BRANAS-GARZA, PABLO and STEFANIA OTTONE (2009). Third-party Punishment is More Effec-
tive on Women. Experimental Evidence
http://www.ugr.es/~teoriahe/RePEc/gra/wpaper/thepapers09_08.pdf.
BRANDSTÄTTER, HERMANN and WERNER GÜTH (2002). "Personality in Dictator and Ultimatum
Games." Central European Journal for Operations Research & Economics 10: 191-215.
BROBERG, TOMAS, TORE ELLINGSEN and MAGNUS JOHANNESSON (2007). "Is Generosity Invol-
untary?" Economics Letters 94: 32-37.
BROSIG, JEANNETTE, THOMAS RIECHMANN and JOACHIM WEIMANN (2007). Selfish in the
End?:An Investigation of Consistency and Stability of individual Behavior
http://mpra.ub.uni-muenchen.de/2035/1/MPRA_paper_2035.pdf.
BURNS, JUSTINE (2010). Race and Social Interactions in Post-Apartheid South Africa. Discrimi-
nation in an Unequal World. Miguel Angel Centeno und Katherine S. Newman. Oxford,
Oxford University Press: 88-108.
CADSBY, C. BRAM, MAROS SERVÁTKA and FEI SONG (2009). Gender and Generosity. Does De-
gree of Anonymity or Group Gender Composition Matter?
http://www.economics.uoguelph.ca/econ/Research/DisPapers/2009_13.pdf.
CAPPELEN, ALEXANDER W., ASTRI DRANGE HOLE, ERIC O. SORENSEN and BERTIL TUNGODDEN
(2007). "The Pluralism of Fairness Ideals: An Experimental Approach." American Eco-
nomic Review 97: 818-827.
CAPRA, MONICA C. and LEI LI (2006). Conformity in Contribution Games. Gender and Group
Effects http://www.economics.emory.edu/Working_Papers/wp/capra_06_01_paper.pdf.
35
CÁRDENAS, JUAN-CAMILO, NATALIA CANDELO, ALEJANDRO GAVIRIA, SANDRA POLONÍA and
RAJIV SETHI (2008). Discrimination in the Provision of Social Services to the Poor. A Field
Experimental Study http://www.iadb.org/res/publications/pubfiles/pubR-544.pdf.
CÁRDENAS, JUAN CAMILO and JEFFREY CARPENTER (2008). "Behavioural Development Econom-
ics: Lessons from Field Labs in the Developing World." Journal of Development Studies
44: 311-338.
CARLSSON, FREDERIK, HAORAN HE and PETER MARTINSSON (2008). Easy Come, Easy Go. The
Role of Windfall Money in Lab and Field Experiments http://nordicconference.org/wp-
content/uploads/2009/10/Carlsson.pdf.
CARPENTER, JEFFREY, STEPHEN BURKS and ERIC VERHOOGEN (2004). "Comparing Students to
Workers. The Effects of Social Framing on Behavior in Distribution Games." Research in
Experimental Economics: 261-289.
CARPENTER, JEFFREY, CRISTINA CONNOLLY and CAITLIN KNOWLES MYERS (2008). "Altruistic
Behavior in a Representative Dictator Experiment." Experimental Economics 11: 282-298.
CARPENTER, JEFFREY, ALLISON LIATI and BRIAN VICKERY (2006). They Come to Play: Supply
Effects in an Economic Experiment
http://sandcat.middlebury.edu/econ/repec/mdl/ancoec/0602.pdf.
CARPENTER, JEFFREY, ERIC VERHOOGEN and STEPHEN BURKS (2005). "The Effect of Stakes in
Distribution Experiments." Economics Letters 86: 393-398.
CARTER, MICHAEL R. and MARCO CASTILLO (2005). Morals, Markets and Mutual Insurance. Us-
ing Economic Experiments to Study Recovery from Hurricane Mitch. The Social Econom-
ics of Poverty. Christopher B. Barrett. New York, Routledge: 268-287.
CARTER, MICHAEL R. and MARCO CASTILLO (2009). "Trustworthiness and Social Capital in
South Africa. Analysis of Actual Living Standards Data and Artefactual Field Experi-
ments." Economic Development and Cultural Change ***: ***.
CASON, TIMOTHY N. and VAI-LAM MUI (1997). "A Laboratory Study of Group Polarisation in
the Team Dictator Game." Economic Journal 107: 1465-1483.
CASTILLO, MARCO and PHILIP J. CROSS (2008). "Of Mice and Men. Within Gender Variation in
Strategic Behavior." Games and Economic Behavior 64: 421-432.
CHARNESS, GARY and URI GNEEZY (2008). "What’s in a Name? Anonymity and Social Distance
in Dictator and Ultimatum Games." Journal of Economic Behavior & Organization 68: 29-
35.
CHAUDHURI, ANANISH and LATA GANGADHARAN (2007). "An Experimental Analysis of Trust
and Trustworthiness." Southern Economic Journal 73: 959-985.
36
CHERRY, TODD L. (2001). "Mental Accounting and Other-regarding Behavior. Evidence from the
Lab." Journal of Economic Psychology 22: 605-615.
CHERRY, TODD L., PETER FRYKBLOM and JASON F. SHOGREN (2002). "Hardnose the Dictator."
American Economic Review 92: 1218-1221.
COX, JAMES C. (2004). "How to Identify Trust and Reciprocity." Games and Economic Behavior
46: 260-281.
COX, JAMES C. and CARY A. DECK (2006). "When are Women More Generous Than Men?"
Economic Inquiry 44: 587-598.
COX, JAMES C., KLARITA SADIRAJ and VJOLLCA SADIRAJ (2008). "Implications of Trust, Fear,
and Reciprocity for Modeling Economic Behavior." Experimental Economics 11: 1-24.
DALBERT, CLAUDIA and SÖREN UMLAUFT (2009). "The Role of the Justice Motive in Economic
Decision Making." Journal of Economic Psychology 30: 172-180.
DANA, JASON, DAYLIAN M. CAIN and ROBYN M. DAWES (2006). "What You Don't Know Won't
Hurt Me. Costly (But Quiet) Exit in Dictator Games." Organizational Behavior and Human
Decision Processes 100: 193-201.
DICKSON, ERIC S. (2009). "Do Participants and Observers Assess Intentions Differently During
Bargaining and Conflict?" American Journal of Political Science 53: 910-930.
DIEKMANN, ANDREAS (2004). "The Power of Reciprocity. Fairness, Reciprocity, and Stakes in
Variants of the Dictator Game." Journal of Conflict Resolution 48: 487-505.
DUFFY, JOHN and TATIANA KORNIENKO (2009). Does Competition Affect Giving?
http://www.econ.pitt.edu/papers/John_dictatorcontest.pdf.
DUFWENBERG, MARTIN and ASTRI MUREN (2006a). "Gender Composition in Teams." Journal of
Economic Behavior & Organization 61: 50-54.
DUFWENBERG, MARTIN and ASTRI MUREN (2006b). "Generosity, Anonymity, Gender." Journal
of Economic Behavior & Organization 61: 42-49.
ECKEL, CATHERINE C. and PHILIP J. GROSSMAN (1998). "Are Women Less Selfish Than Men?:
Evidence from Dictator Experiments." Economic Journal 108: 726-735.
ECKEL, CATHERINE C. and PHILIP J. GROSSMAN (2000). "Volunteers and Pseudo-Volunteers. The
Effect of Recruitment Method in Dictator Experiments." Experimental Economics 3: 107-
120.
37
EICHENBERGER, REINER and FELIX OBERHOLZER-GEE (1998). "Rational Moralists: The Role of
Fairness in Democratic Economic Politics." Public Choice 94: 191-210.
ELLINGSEN, THORE, MAGNUS JOHANNESSON, SIGVE TJOTTA and GAUTE TORSVIK (2010). "Test-
ing Guilt Aversion." Games and Economic Behavior 68: 95-107.
ENSMINGER, JEAN (2004). Market Integration and Fairness. Evidence from Ultimatum, Dictator,
and Public Goods Experiments in East Africa. Foundations of Human Sociality. Economic
Experiments and Ethnographic Evidence from Fifteen Small-Scale Societies. Joseph Hen-
rich und Robert Boyd. Oxford, Oxford University Press: 356-381.
FARINA, FRANCESCO, NIALL O'HIGGINS and PATRIZIA SBRIGLIA (2008). Eliciting Motives for
Trust and Reciprocity by Attitudinal and Behavioural Measures
http://ftp.iza.org/dp3584.pdf.
FERSHTMAN, CHAIM and URI GNEEZY (2001). "Discrimination in a Segmented Society. An Ex-
perimental Approach." Quarterly Journal of Economics 116: 351-377.
FISMAN, RAYMOND, SHACHAR KARIV and DANIEL MARKOVITS (2007). "Individual Preferences
for Giving." American Economic Review 97: 1858-1876.
FONG, CHRISTINA M. and ERZO F.P. LUTTMER (2009). "What Determines Giving to Hurricane
Katrina Victims? Experimental Evidence on Racial Group Loyalty." American Economic
Journal: Applied Economics 1: 64-87.
FORSYTHE, ROBERT, JOEL L. HOROWITZ, N.E. SAVIN and MARTIN SEFTON (1994). "Fairness in
Simple Bargaining Experiments." Games and Economic Behavior 6: 347-369.
FREY, BRUNO and IRIS BOHNET (1995). "Institutions affect Fairness." Journal of Institutional and
Theoretical Economics 151: 286-303.
FREY, BRUNO and IRIS BOHNET (1997). "Identification in Democratic Society." Journal of Socio-
Economics 26: 25-38.
FROHLICH, NORMAN, JOE OPPENHEIMER and J. BERNARD MOORE (2001). "Some Doubts About
Measuring Self-interest Using Dictator Experiments. The Costs of Anonymity." Journal of
Economic Behavior & Organization 46: 271-290.
GOWDY, JOHN, RALUCA IORGULESCU and STEPHEN ONYEIWU (2003). "Fairness and Retaliation
in a Rural Nigerian Village." Journal of Economic Behavior & Organization 52: 469-479.
38
GREINER, BEN, WERNER GÜTH and RO'I ZULTAN (2005). Let the Dummy Talk! Unilateral Com-
munication and Discrimination in Three-Person Dictator Experiments
ftp://papers.econ.mpg.de/esi/discussionpapers/2005-18.pdf.
GURVEN, MICHAEL (2004). "Economic Games Among the Amazonian Tsimane. Exploring the
Roles of Market Access, Costs of Giving, and Cooperation on Pro-Social Game Behavior."
Experimental Economics 7: 5-24.
GURVEN, MICHAEL, ARIANNA ZANOLINI and ERIC SCHNITER (2008). "Culture Sometimes Mat-
ters. Intra-cultural Variation in Pro-social Behavior Among Tsimane Amerindians." Jour-
nal of Economic Behavior & Organization 67: 587-607.
HALEY, KEVIN J. and DANIEL M.T. FESSLER (2005). "Nobody’s Watching? Subtle Cues Affect
Generosity in an Anonymous Economic Game." Evolution and Human Behavior 26: 245-
256.
HARBAUGH, WILLIAM T. and KATE KRAUSE (2000). "Children's Altruism in Public Good and
Dictator Experiments." Economic Inquiry 38: 95-109.
HARBAUGH, WILLIAM T., KATE KRAUSE and STEVEN G. LIDAY (2003). Bargaining by Children
http://economics.uoregon.edu/papers/UO-2002-4_Harbaugh_Kid_Bargaining.pdf.
HEINRICH, TIMO, THOMAS RIECHMANN and JOACHIM WEIMANN (2009). Game or Frame? Incen-
tives in Modified Dictator Games http://ideas.repec.org/p/mag/wpaper/09008.html.
HOFFMAN, ELIZABETH, KEVIN MCCABE, KEITH SHACHAT and VERNON L. SMITH (1994). "Prefer-
ences, Property Rights, and Anonymity in Bargaining Games." Games and Economic Be-
havior 7: 346-380.
HOFFMAN, ELIZABETH, KEVIN MCCABE and VERNON L. SMITH (1996). "Social Distance and
Other-Regarding Behavior in Dictator Games." American Economic Review 86: 653-660.
HOLM, HAKAN J. and ANDERS DANIELSON (2005). "Tropic Trust versus Nordic Trust. Experi-
mental Evidence from Tanzania and Sweden." Economic Journal 115: 505-532.
HOLM, HAKAN J. and PETER ENGSELD (2005). "Choosing Bargaining Partners. An Experimental
Study on the Impact of Information About Income, Status and Gender." Experimental
Economics 8: 183-216.
HOUSER, DANIEL and DANIEL SCHUNK (2009). "Social Environments With Competitive Pres-
sure: Gender Effects in the Decisions of German Schoolchildren." Journal of Economic
Psychology 30: 634-641.
39
KAMAS, LINDA, SANDY BAUM and ANNE PRESTON (2005). "Altruistic Responses to the Septem-
ber 11 Terrorist Attacks. Some Evidence from Dictator Games." Eastern Economic Journal
31: 551-562.
KLEMPT, CHARLOTTE and KERSTIN PULL (2009). Generosity, Greed and Gambling. What Differ-
ence Does Asymmetric Information in Bargaining Make?
http://www.econbiz.de/archiv1/2009/94186_generosity_greed_gambling.pdf.
KNAFO, ARIEL and S. ISRAEL (2007). "Individual Differences in Allocation of Funds in the Dicta-
tor Game Associated with Length of the Arginine Vasopressin 1a Receptor RS3 Promoter
Region and Correlation between RS3 Length and Hippocampal mRNA." Genes, Brain and
Behavior 7: 266-275.
KOCH, ALEXANDER and HANS-THEO NORMANN (2008). "Giving in Dictator Games. Regard for
Others or Regard by Others?" Southern Economic Journal 75: 223-231.
KORENOK, OLEG, EDWARD L. MILLNER and LAURA RAZZOLINI (2008). Experimental Evidence
on Inequality Aversion. Dictators Give to Help the Less Fortunate
http://www.people.vcu.edu/~okorenok/DictatorEquity_080828.pdf.
LAZEAR, EDWARD P., ULRIKE MALMENDIER and ROBERTO A. WEBER (2009). Sorting and Social
Preferences http://www.econ.berkeley.edu/~ulrike/Papers/Sorting_Paper_2009-06-19.pdf.
LEIDER, STEPHEN, MARKUS M. MÖBIUS, TANYA ROSENBLAT and QUOC-ANH DO (2009). What
Do We Expect from Our Friends?
https://mercury.smu.edu.sg/rsrchpubupload/15345/ExpectFriends.pdf.
LIST, JOHN A. (2007). "On the Interpretation of Giving in Dictator Games." Journal of Political
Economy 115: 482-493.
LIST, JOHN A. and TODD L. CHERRY (2008). "Examining the Role of Fairness in High Stakes Al-
location Decisions." Journal of Economic Behavior & Organization 65: 1-8.
LUHAN, WOLFGANG J., MARTIN KOCHER and MATTHIAS SUTTER (2009). "Group Polarization in
the Team Dictator Game Reconsidered." Experimental Economics 12: 26-41.
MARLOWE, FRANK (2004). "What Explains Hazda Food Sharing?" Research in Economic An-
thropology 23: 69-88.
MITTONE, LUIGI and MATTEO PLONER (2006). Is It Just Legitimacy of Endowments? An Exper-
imental Analysis of Unilateral Giving
http://eprints.biblio.unitn.it/archive/00001115/01/papero06_02.pdf.
40
OBERHOLZER-GEE, FELIX and REINER EICHENBERGER (2008). "Fairness in Extended Dictator
Game Experiments." B.E. Journal of Economic Analysis and Policy 8(16): 1-19.
OXOBY, ROBERT J. and JOHN SPRAGGON (2008). "Mine and Yours. Property Rights in Dictator
Games." Journal of Economic Behavior & Organization 65: 703-713.
RANKIN, FREDERICK W. (2006). "Requests and Social Distance in Dictator Games." Journal of
Economic Behavior & Organization 60: 27-36.
RIGDON, MARY, KEIKO ISHII, MOTOKI WATABE and SHINOBU KITAYAMA (2009). "Minimal So-
cial Cues in the Dictator Game." Journal of Economic Psychology 30: 358-367.
RUFFLE, BRADLEY J. (1998). "More Is Better, But Fair Is Fair. Tipping in Dictator and Ultima-
tum Games." Games and Economic Behavior 23: 247-265.
SAAD, GAD and TRIPAT GILL (2001). "The Effects of a Recipient's Gender in a Modified Dictator
Game." Applied Economics Letters 8: 463-466.
SCHOTTER, ANDREW, AVI WEISS and INIGO ZAPATER (1996). "Fairness and Survival in Ultima-
tum and Dictatorship Games." Journal of Economic Behavior & Organization 31: 37-56.
SCHURTER, KARL and BART WILSON (2009). "Justice and Fairness in the Dictator Game." South-
ern Economic Journal 76: 130-145.
SEFTON, MARTIN (1992). "Incentives in Simple Bargaining Games." Journal of Economic Psy-
chology 13: 263-276.
SELTEN, REINHARD and AXEL OCKENFELS (1998). "An Experimental Solidarity Game." Journal
of Economic Behavior & Organization 34: 517-539.
SLONIM, ROBERT and ELLEN GARBARINO (2008). "Increases in Trust and Altruism from Partner
Selection. Experimental Evidence." Experimental Economics 11: 134-153.
SMALL, DEBORAH and GEORGE LOEWENSTEIN (2003). "Helping a Victim or Helping the Victim.
Altruism and Identifiability." Journal of Risk and Uncertainty 26(5-16).
SONG, FEI, C. BRAM CADSBY and TRISTAN MORRIS (2004). "Other-Regarding Behavior and Be-
havioral Forecasts. Females versus Males as Individuals and as Group Representatives."
International Journal of Conflict Management 15: 340-363.
STEPHEN, ANDREW T. and MICHEL TUAN PHAM (2008). On Feelings as a Heuristic for Making
Offers in Ultimatum Negotiations http://mpra.ub.uni-
muenchen.de/8779/1/MPRA_paper_8779.pdf.
41
SWOPE, KURTIS J., JOHN CADIGAN, PAMELA M. SCHMITT and ROBERT SHUPP (2008). "Personali-
ty Preferences in Laboratory Economics Experiments." Journal of Socio-Economics 37:
998-1009.
TAKEZAWA, MASANORI, MICHAELA GUMMERUM and MONIKA KELLER (2006). "A Stage for the
Rational Tail of the Emotional Dog. Roles of Moral Reasoning in Group Decision Mak-
ing." Journal of Economic Psychology 27: 117-139.
TAN, JONATHAN H.W. and FRIEDEL BOLLE (2006). "On the Relative Strengths of Altruism and
Fairness." Theory and Decision 60: 35-67.
VAN DER MERWE, WILHELM GERHARD and JUSTINE BURNS (2008). "What's in a Name? Racial
Identity and Altruism in Post-Apartheid South-Africa." South African Journal of Econom-
ics 76: 266-275.
WHITT, SAM and RICK K. WILSON (2007). "Public Goods in the Field. Katrina Evacuees in Hou-
ston." Southern Economic Journal 74: 377-387.
XIAO, ERTE and DANIEL HOUSER (2009). "Avoiding the Sharp Tongue. Anticipated Written
Messages Promote Fair Economic Exchange." Journal of Economic Psychology 30: 393-
404.
YAMAGISHI, TOSHIO and NOBUHIRO MIFUNE (2008). "Does Shared Group Membership Promote
Altruism? Fear, Greed, and Reputation." Rationality and Society 20: 5-30.
YAMAMORI, TETSUO, KAZUHIKO KATO, TOSHIJI KAWAGOE and AKIHIKO MATSUI (2008). "Voice
Matters in a Dictator Game." Experimental Economics 11: 336-343.
42
Appendix 2: Descriptives
left hand column: frequency in meta-study; right hand column: frequency in reconstructed original data
43