Upload
avner-ben-ner
View
216
Download
4
Embed Size (px)
Citation preview
Journal of Economic Psychology 25 (2004) 581–589
www.elsevier.com/locate/joep
Share and share alike?Gender-pairing, personality, and cognitive
ability as determinants of giving
Avner Ben-Ner a,*, Fanmin Kong b, Louis Putterman c
a Industrial Relations Center, Carlson School of Management, University of Minnesota,
Minneapolis, MN 55455, USAb Guanghua School of Management, Peking University, Beijing, 100871, Chinac Department of Economics, Brown University, Providence, RI 02912, USA
Received 26 July 2002; received in revised form 28 November 2002; accepted 10 April 2003
Available online 22 July 2003
Abstract
We conduct dictator game experiments in which women and men are allowed to split $10
with a completely unknown person or a person of known gender. Subjects also complete per-
sonality and cognitive tests. We find that (a) gender information significantly affects giving
only in the case of women, who give systematically less to women than to men and persons
of unknown gender; (b) largely on account of this difference, women give less than men on
average, although the difference is not statistically significant; and (c) giving is significantly ex-
plained, especially for women, by personality measures and the cognition score.
� 2003 Elsevier B.V. All rights reserved.
PsycINFO classification: 2223
JEL classification: C91; D64
Keywords: Dictator game; Gender; Altruism; Giving
* Corresponding author.
E-mail addresses: [email protected] (A. Ben-Ner), [email protected] (F. Kong),
[email protected] (L. Putterman).
0167-4870/$ - see front matter � 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S0167-4870(03)00065-5
582 A. Ben-Ner et al. / Journal of Economic Psychology 25 (2004) 581–589
1. Introduction
In a dictator game, an individual is given some amount of money and asked to
decide how much, if any, to share with another individual. Potential determinants
of sharing in this game – especially gender – have been isolated for study (Andreoni& Vesterlund, 2001; Eckel & Grossman, 1998). The effect on sharing of information
about the identity of the recipient has also begun to be studied (Eckel & Grossman,
1996).
The present paper studies how, if at all, the decisions of men and women are af-
fected by their gender, by knowledge of the gender of the person with whom they are
paired, and by measurable dimensions of personality and cognitive ability.
The paper proceeds as follows. In Section 2, we introduce the themes of our anal-
ysis: gender, gender-pairing, personality, and cognitive ability. Section 3 provides adescription of our experiments. Section 4 analyzes subjects� behaviors. Section 5 con-cludes the paper.
2. Theoretical considerations
The dictator game was introduced to test if sharing of endowments, which is com-
monly observed in the related ultimatum game (Camerer & Thaler, 1995), is still ob-served if recipients have no veto power. Early experiments found that dictators still
shared an average of about three dollars in a 10-dollar endowment, but that the
amount shared declined when neither recipient nor experimenter could associate a
decision with a specific dictator (i.e., in a double-blind treatment).
This paper extends the literature on the determinants of individual differences in
sharing in a dictator game. We focus on four possible determinants of sharing. The
first is gender. Men and women may differ in their propensities to share a windfall for
a variety of reasons. Differences in reproductive and family roles may lead to thevaluing of nurturing and conciliating qualities in women and of competitive qualities
in men (Eagly & Crowley, 1986; Eckel & Grossman, forthcoming). While such dif-
ferences tend to be inculcated through socialization, some differences could have bi-
ological roots.
The second factor we study is gender-pairing. Persons of either gender may
vary their behaviors depending upon the gender of the person with whom they are
interacting. For example, women may feel greater solidarity with other women than
with men, or, on the contrary, women may view other women as competitors. Theformer could be considered an instance of in-group/out-group behavior, as studied
in many psychological experiments (Tajfel, 1981). The latter would conform to the
predictions of evolutionary psychology (Buss, 1998; Campbell, 1999; Kanazawa,
2002).
The third factor we look at is personality type. Psychologists note that people dif-
fer in various attributes of personality, probably due to a combination of genetic and
environmental differences. One leading psychological scoring system is the NEO five-
factor inventory (Briggs, 1992), which generates measures of personality dimensions
A. Ben-Ner et al. / Journal of Economic Psychology 25 (2004) 581–589 583
called ‘‘neuroticism,’’ ‘‘extraversion,’’ ‘‘openness to experience,’’ ‘‘agreeableness,’’
and ‘‘conscientiousness.’’ How the propensities towards selfishness versus altruism,
generosity, or rule-based behavior play out in a particular individual may be signif-
icantly linked to such personality variables, as shown recently by Boone, De Bra-
bander, and van Witteloostuijn (1999), Koole, Jager, van den Berg, Vlek, andHofstee (2001), and LePine and Van Dyne (2001). It is of interest to know whether
not-otherwise-explained differences in sharing behavior are purely random, or
whether they in fact correlate with measurable personality indicators.
The fourth and last factor that we consider is cognitive ability. Decision-making
errors and misunderstanding of instructions are often suggested as explanations of
experimental outcomes seemingly at odds with those of neoclassical economics.
Since the neoclassical prediction in the dictator game, assuming strictly self-inter-
ested agents, is sending zero, while only positive amounts or zero can be sent, errorscannot center on the predicted value but are of necessity positive. Thus to the extent
that errors play an important role in explaining sending and that errors are less likely
to be made by subjects with stronger cognitive abilities, there should be a significant
negative relationship between amounts sent and a reliable measure of cognitive abil-
ity. A related possibility is that cognitive ability is systematically related to generosity
for reasons other than proclivity to error: for instance, perhaps Machiavellians are
right that ‘‘only fools are kind when there is nothing in it for them.’’ Although
our experimental design does not permit us to distinguish between these two inter-pretations, both make it interesting to look for correlations between cognitive ability
and sharing in a dictator game. Accordingly, we administered a test of cognitive abil-
ity, the Wonderlic. 1
3. The experiment
The experiment was conducted by a team of researchers from the University ofMinnesota, first at that university, and subsequently at Brown University. It in-
volved 224 first semester undergraduate students at Minnesota, of whom 112 were
assigned the role of dictator, and 84 first semester undergraduates at Brown, of
whom 42 were assigned that role. The Minnesota subjects responded to an e-mail in-
vitation to the university�s entire freshman class (which numbered about 5000), whilethe Brown subjects responded by e-mail to a flyer distributed to all freshmen (about
1600). The e-mail invitations were issued individually (the identities of other recipi-
ents were not shown on the invitation), and both these and the flyers at Brown askedfor participation in an economics/psychology experiment that would last up to 2
1 The Wonderlic personnel test is a timed, 50-item cognitive ability measure commonly used in the pre-
employment selection context. Wonderlic scores are highly consistent with other well-recognized measures
such as the Wechsler Adult Intelligence Scale, the General Aptitude Test Battery, and the Stanford
Achievement Test (see, e.g., Hawkins, Faraone, Pepple, & Seidman, 1990; McKelvie, 1989). The
Wonderlic is also used as a cognitive ability measure in LePine and Van Dyne, cited above.
584 A. Ben-Ner et al. / Journal of Economic Psychology 25 (2004) 581–589
hours, that would require no physical effort, that would assure subjects� anonymity,and that would earn them a $15 participation fee and possibly additional earnings. 2
Subjects who subsequently learned of their roles as dictators, and those as-
signed the recipient role, entered different rooms in separate buildings and were
assured explicitly of their anonymity. At both campuses, all rooms in which dicta-tors sat contained similar number of male and female subjects seated without regard
to gender, so there were no general cues about the salience of gender in the experi-
ment.
In addition to the $15 fee for participation, each participant in the sender rooms
was given $10 in 1 dollar bills that he or she was asked to keep or divide with an
anonymous recipient. The experiments were fashioned in a manner that closely fol-
lows the double-blind design of Hoffman, McCabe, Shachat, and Smith (1994) and
Eckel and Grossman (1996) so as to assure anonymity both among subjects and be-tween subjects and experimenters. However, the instructions follow more closely the
‘‘divide an endowment’’ language used by Forsythe, Horowitz, Savin, and Sefton
(1994) rather than the ‘‘set a price’’ language used by Hoffman et al. (1994) (see Bol-
ton, Katok, & Zwick, 1998 for a discussion of a possible difference in framing
effects).
In a departure from more basic dictator games, the written instructions given to
roughly 70% of dictators included information about their potential recipient. At
Minnesota, 23 subjects were given no information about the person with whom theywere dividing their $10, 20 subjects were told that this person was a female, 26 were
told that this person was a male, and 43 subjects, whose decisions are not analyzed
here, were given information about their recipient�s place of origin. At Brown,21 subjects were given no information about the person with whom they were divid-
ing their $10 and 21 subjects were told that this person was a female. Limitations on
the number and genders of subjects who showed up for the experiment at Brown
meant that both the sending-to-male condition and conditions with information
about the place of origin of the recipient could not be included there. Our focus ison the sending-to-person (no information), sending-to-female, and sending-to-male
conditions, to which a total of 111 subjects, 69 at Minnesota and 42 at Brown, were
assigned. Because subjects were told nothing about whether information was being
given to other subjects, or if so, what kind of information, the absence of certain con-
ditions at one of the two schools cannot have affected the understanding or framing
of the task facing the subjects whose decisions we analyze. 3 Recipients of the gender
2 The relatively high participation fee, which could have some bearing upon the amounts sent, was
deemed necessary due to the time required by participants to complete the NEO and Wonderlic
instruments and an additional survey and tasks not pertinent to the present paper (but discussed in Ben-
Ner, Putterman, Kong, & Magan, forthcoming).3 The complete instructions given to participants are available upon request.
A. Ben-Ner et al. / Journal of Economic Psychology 25 (2004) 581–589 585
indicated (or of either gender, in the ‘‘sending-to-person’’ condition), who were
seated in other rooms, received the bills sent to them, as promised by the experi-
menter. 4
4. Experimental results and analysis
4.1. Description and analysis of variance
Table 1 displays definitions of the variables, and means of the continuous vari-
ables, included in our analysis. Table 2 shows the average amount sent by dictators
overall, and a breakdown of sending by males and females to recipients of unknown
gender and recipients known to be male or female. The most important finding isthat female subjects sent less to known female recipients than they did to known
male recipients and to persons about whom no information was provided. That in-
formation on recipient�s gender mattered to female subjects is partially supported byan analysis of variance. The results of a one-way ANOVA indicate that knowing the
receiver�s gender significantly influenced female dictators� giving, F ð2; 67Þ ¼ 3:13,p < 0:05, g2 ¼ 0:085. A similar effect did not obtain for male subjects, F ð2; 38Þ ¼0:05, p < 0:95, g2 ¼ 0:003. We also conducted a two-way ANOVA, with the dollaramount sent as the dependent variable. Results do not suggest a significant main ef-fect of dictator�s gender, F ð1; 105Þ ¼ 0:85, p < 0:36, g2 ¼ 0:007, on sharing. Nor dothe results suggest a main effect of receiver�s gender, F ð2; 105Þ ¼ 0:68, p < 0:51,g2 ¼ 0:011, on sharing. The interaction effect, F ð2; 105Þ ¼ 1:34, p < 0:27, g2 ¼0:024, on sharing is not significant either. A Mann–Whitney test of the combined
Minnesota and Brown sample shows that women sent less to women than to men
and to persons (both differences are significant at the 5% level). 5 Mann–Whitney
tests confirm these differences at the 10% level in the Minnesota subsample, al-
though not in the Brown subsample. 6 A Mann–Whitney test on the combined sam-ple also shows that women sent less to women than men did, significant at the 10%
level.
4 Recipients also received show-up fees, completed tests and a survey, and carried out other tasks not
pertinent to this paper (see again Ben-Ner et al., forthcoming).5 The OLS regression on which the two-way ANOVA is based also provides evidence of the gender
difference. In this regression, amount sent is the dependent variable, and the gender-pairing dummy
variables female-to-female, female-to-male, female-to-person, male-to-female and male-to-male are the
explanatory variables. The coefficient on the female-to-female dummy variable is negative and significant
at the 10% level.6 This may be partly a function of sample size. Average behaviors by gender are remarkably similar at
the two schools. Both a Mann–Whitney test and a Kolmogorov–Smirnov test show no significant
differences between overall sending at Minnesota and at Brown, no significant difference between sending
by women at Minnesota and by women at Brown, and no significant difference between sending by men at
Minnesota and by men at Brown.
Table 2
Amounts sent, by gender-pairing
Receiver�s gender Total
Male Female Person
Dictator’s GenderMale 3.50a 3.71a 3.41a 3.54a
(2.99)b (2.89)b (2.24)b (2.60)b
10c 14c 17c 41c
Female 3.81a 2.19a 3.30a 2.99a
(2.26)b (2.34)b (2.07)b (2.29)b
16c 27c 27c 70c
Total 3.69a 3.71a 3.34a 3.19a
(2.51)b (2.61)b (2.11)b (2.41)b
26c 41c 44c 111c
aMean.b Standard deviation.cNumber of observations.
Table 1
Variable definitions, and means of continuous variables
Variable Definition Mean
Dependent variable
Send Dollar amount sent ð$0; $1; $2; . . . ; $10Þ 3.19
Independent variables
Agreeable Raw score dictator earned on NEO agreeableness scale 32.51
Conscientious Raw score dictator earned on NEO conscientiousness scale 30.76
Extraversion Raw score dictator earned on NEO extraversion scale 30.68
Neurotic Raw score dictator earned on NEO neuroticism scale 22.95
Open Raw score dictator earned on NEO openness scale 32.29
Cognition Raw score dictator earned on Wonderlic personnel test on problem-
solving ability
29.18
Female-to-female Gender pairing dummy, 1 if a female dictator was informed that her
paired recipient is also a female; 0 otherwise
Female-to-male Gender pairing dummy, 1 if a female dictator was informed that her
paired recipient is a male; 0 otherwise
Female-to-person Gender pairing dummy, 1 if a female dictator was NOT informed
any gender or origin information of her paired recipient; 0 otherwise
Male-to-male Gender pairing dummy, 1 if a male dictator was informed that his
paired recipient is also a male; 0 otherwise
Male-to-female Gender pairing dummy, 1 if a male dictator was informed that his
paired recipient is a female; 0 otherwise
Male-to-person Gender pairing dummy, 1 if a male dictator was NOT informed of
any gender or origin information of his paired recipient; 0 otherwise
586 A. Ben-Ner et al. / Journal of Economic Psychology 25 (2004) 581–589
4.2. Regression analysis
We study the effects of cognitive ability and personality jointly with those of gen-
der and gender-pairing by estimating multivariate regressions. Table 3 presents OLS
Table 3
Determinants of sending: Gender and no information conditions
Variable OLS Ordered logit
All Males Females All Males Females
Agreeable 0.12�� 0.16� 0.12� 0.10�� 0.17�� 0.11��
(0.05) (0.08) (0.06) (0.04) (0.07) (0.05)
Conscientious 0.02 0.09 )0.08� 0.01 0.07� )0.08��
(0.04) (0.06) (0.04) (0.03) (0.04) (0.04)
Extraversion )0.07�� )0.16�� 0.005 ) 0.06� )0.14�� 0.01
(0.04) (0.07) (0.04) (0.03) (0.06) (0.04)
Neurotic )0.03 0.03 )0.08�� )0.02 0.02 )0.07��
(0.03) (0.06) (0.03) (0.02) (0.04) (0.03)
Open 0.06 0.06 0.03 0.05 0.03 0.04
(0.04) (0.07) (0.04) (0.03) (0.05) (0.04)
Cognition )0.09� )0.03 )0.11�� )0.06� )0.01 )0.10�
(0.05) (0.08) (0.05) (0.04) (0.06) (0.05)
Female-to-female )1.19 )1.16�� )1.34�� )1.30��
(0.76) (0.56) (0.61) (0.55)
Female-to-male 0.65 0.55 0.05 0.21
(0.86) (0.67) (0.65) (0.59)
Female-to-person )0.10 )0.24(0.74) (0.58)
Male-to-male 1.27 1.73 0.47 1.32
(1.03) (1.25) (0.82) (0.95)
Male-to-female 0.67 0.62 0.39 0.56
(0.85) (0.96) (0.69) (0.71)
Constant 2.40 1.60 5.70�
(2.69) (5.01) (2.99)
N 111 41 70 111 41 70
Prob: > F =v2 0.05 0.26 0.002 0.03 0.14 0.001
Adjusted R2 0.08 0.06 0.23
Note: Standard errors of parameter estimates are reported in parentheses. Significance levels are marked
with � for p6 0:10, and �� for p6 0:05. For ordered logit, the estimates for ancillary parameters are notreported here.
A. Ben-Ner et al. / Journal of Economic Psychology 25 (2004) 581–589 587
and ordered logit estimates for the combined Minnesota and Brown samples, with
both genders combined, and for male and female dictators separately. The regression
equations include dummy variables for each sender-recipient gender pair, sending-
to-person by males being the excluded gender-pairing condition in the combined re-
gressions. Chow tests fail to reject at the 10% level the null hypothesis that male and
female behaviors conform to the same parameters; the tests fail to reject at the 10%
588 A. Ben-Ner et al. / Journal of Economic Psychology 25 (2004) 581–589
level the null hypotheses that behaviors at Brown and those at Minnesota are de-
scribed by the same parameters. 7
Among females, the amount sent to another female is significantly less, consistent
with the Mann–Whitney tests above, and three of the personality factors and cogni-
tion are also significant predictors of sending. In particular, more ‘‘agreeable’’ femaledictators shared more with their recipients, while more ‘‘conscientious,’’ ‘‘neurotic,’’
and cognitively high-scoring female dictators shared less. 8
The estimates for males predicts behavior less well, with no significant effects ap-
parent for gender-pairing condition and cognitive score. However, two of the five
personality factors do have predictive power for male sending. Of these one, extra-
version, was not a significant predictor for female dictators. The sign and magnitude
of the coefficient on agreeableness are quite similar for the two genders, by contrast.
5. Conclusions
Our dictator game experiment suggests that personality, cognitive ability, and the
gender of sender and recipient affect how much is sent to an anonymous recipient in
a one-shot dictator game. Interestingly, all of these factors have a more pronounced
effect for women than for men. In separate sessions at two universities, female sub-
jects on average shared less of their endowments than did males, mainly because wo-men paired with other women sent less than those paired with men or with persons
about whom no information was provided. Women with lower cognitive scores sent
more, and four of five personality factors were also significant predictors of sending,
although only one of these was predictive in both the male and female subsamples.
The model is more successful at explaining female than male sending.
Acknowledgements
We would like to thank John Dickhaut for his help in designing the experiment,
and Dan Magan and Shu-yi Oei for their help in carrying out the experiment, which
was funded by the Carlson School of Management at the University of Minnesota.
7 For OLS, the test statistic for pooling men of both schools with women of both schools is F ¼ 0:205,
which falls far short of the critical value at 5%, F �0:05;6;99 ¼ 2:192. A Chow test of whether Brown and
Minnesota observations can be pooled, disregarding gender differences, also fails to reject the null
hypothesis even at the 10% level, with F ¼ 0:001 < F �0:1;8;95 ¼ 1:736. The test statistic for pooling males of
the two schools is F ¼ 0:514 < F �0:1;6;29 ¼ 1:99. The test statistic for pooling females of the two schools is
F ¼ 0:10 < F �0:1;6;58 ¼ 1:88. A Chow-type test, calculated for the ordered logit regressions, gives similar
results.8 In terms of magnitudes, the OLS estimate for female dictators implies that a one standard deviation
increase in the cognitive score would decrease the amount shared by a female sender by about 56 cents, or
by about 19%. A one standard deviation increase in ‘‘agreeableness’’ would increase female sending by 58
and a one standard increase in ‘‘neuroticism’’ or ‘‘conscientiousness’’would decrease female sending by 70
and 50 cents, respectively. (These calculations use the standard deviations in the 70 female observations.)
A. Ben-Ner et al. / Journal of Economic Psychology 25 (2004) 581–589 589
An earlier version of this paper was presented at the American Economics Associa-
tion panel on Economics, Values and Organization, Allied Social Sciences Associa-
tion meetings, Chicago, January 1998. We thank many readers, seminar and
conference participants for useful comments, and the editor and referees of this jour-
nal for help in focusing the paper.
References
Andreoni, J., & Vesterlund, L. (2001). Which is the fair sex? Quarterly Journal of Economics, 116(1), 293–
312.
Ben-Ner, A., Putterman, L., Kong, F., & Magan, D. (forthcoming). Reciprocity in a two part dictator
game. Journal of Economic Behavior and Organization.
Bolton, G., Katok, E., & Zwick, R. (1998). Dictator game giving: Rules of fairness versus acts of kindness.
International Journal of Game Theory, 27, 269–299.
Boone, C., De Brabander, B., & van Witteloostuijn, A. (1999). The impact of personality on behavior in
five prisoner�s dilemma games. Journal of Economic Psychology, 20, 343–377.Briggs, S. R. (1992). Assessing the five-factor model of personality description. Journal of Personality, 60,
253–293.
Buss, D. (1998). Evolutionary psychology: The new science of the mind. Boston: Allyn and Bacon.
Camerer, C., & Thaler, R. (1995). Ultimatums, dictators and manners. Journal of Economic Perspectives,
9(2), 209–219.
Campbell, A. (1999). Staying alive: Evolution, culture, and women�s intrasexual aggression. Behavioral andBrain Sciences, 22, 203–252.
Eagly, A. H., & Crowley, M. (1986). Gender and helping behavior: A meta-analytic review of the social
psychological literature. Psychological Bulletin, 100, 283–308.
Eckel, C., & Grossman, P. (1996). Altruism in anonymous dictator games. Games and Economic Behavior,
16, 181–191.
Eckel, C., & Grossman, P. (1998). Are women less selfish than men?: Evidence from dictator games. The
Economic Journal, 108, 726–735.
Eckel, C., & Grossman, P. (forthcoming). Differences in the economic decisions of men and women:
Experimental evidence. In C. Plott, V. Smith (Eds.), Handbook of results in experimental economics.
Forsythe, R., Horowitz, J. L., Savin, N. E., & Sefton, M. (1994). Fairness in simple bargaining
experiments. Games and Economic Behavior, 6, 347–369.
Hawkins, K., Faraone, S., Pepple, J., & Seidman, L. (1990). WAIS-R validation of the Wonderlic
personnel test as a brief intelligence measure in a psychiatric sample. Psychological Assessment, 2, 198–
201.
Hoffman, E., McCabe, K., Shachat, K., & Smith, V. (1994). Preferences, property rights, and anonymity
in bargaining games. Games and Economic Behavior, 7, 346–380.
Kanazawa, S. (2002). Is discrimination necessary to explain the sex gap in earnings? Unpublished,
Department of Psychology, University of Canterbury.
Koole, S., Jager, W., van den Berg, A., Vlek, C., & Hofstee, W. (2001). On the social nature of personality:
Effects of extraversion, agreeableness, and feedback about collective resource use on cooperation in a
resource dilemma. Personality and Social Psychology Bulletin, 27(3), 289–301.
LePine, J., & Van Dyne, L. (2001). Voice and cooperative behavior as contrasting forms of contextual
performance: Evidence of differential relationships with big five personality characteristics and
cognitive ability. Journal of Applied Psychology, 86(2), 326–336.
McKelvie, S. (1989). The Wonderlic personnel test: Reliability and validity in an academic setting.
Psychological Reports, 65, 161–162.
Tajfel, H. (1981). Human groups and social categories. Cambridge: Cambridge University Press.