Altruism in individual and joint-giving decisions: What's gender got to do with it?

Abstract

This paper uses dictator experiments to examine gender differences in altruistic behavior in the United States when decisions are made individually and jointly. In anonymous individual giving to charity, women give substantially more than men, and in paired settings, mixed-sex groups give the most while all male pairs give the least. Evidence supports social information and negotiation effects as participants change giving toward that of their partners. Social image effects are found only in mixed-sex groups, indicating a gender-based component to the value of the social signal sent. Although men and women appear to have similar influence, the positive social image effect pushes giving in mixed-sex pairs above the sum of the members' individual gifts because the less altruistic partners (usually men) adjust their giving upward more than the more altruistic partners (usually women) reduce giving. Therefore, increasing women's participation in traditionally male spheres of decision making may result in more altruistic economic behavior.

Keywords:

• Altruism
• gender
• and experimental economics
• JEL Codes: D64, J16, C92

Acknowledgments

We gratefully acknowledge funding for this project from research grants provided by Santa Clara University and the Leavey School of Business. We also would like to thank the anonymous referees whose helpful suggestions greatly improved the paper.

Notes

  ¹ Studies finding women make larger offers than men include Clifford Norwell and Sara H. Tinkler (1994), Catherine Eckel and Philip Grossman (1998), Rachel Croson and Nancy Buchan (1999), and Sara J. Solnick and Maurice E. Schweitzer (1999). No significant difference in offers made by men and women was found by Gary E. Bolton and Elena Katok (1995); Timothy N. Cason and Vai-Lam Mui (1997); C. Bram Cadsby and Elizabeth Maynes (1998); Catherine Eckel and Philip Grossman (2001); Sara J. Solnick (2001); James C. Cox (2002); Avner Ben-Ner, Fanmin Kong, and Louis Putterman (2004); and Martin Dufwenberg and Astri Muren (2006b). Jamie Brown-Kruse and David Hummels (1993) reported that men made larger offers. James Andreoni and Lise Vesterlund (2001) and James C. Cox and Cary A. Deck (2006) find that the more generous sex depends on the price of giving.

  ² Catherine Eckel and Philip Grossman (2008; forthcoming) and Rachel Croson and Uri Gneezy (2004) survey the literature on gender differences in economic decision making. Eckel and Grossman find that in experiments where women are exposed to risk, such as monetary loss or feeling cheated in ultimatum games or public goods games, they behave similarly to men, while in games without risk, such as dictator games, women act more altruistically than men. Croson and Gneezy conclude that differences in experimental findings may result from the fact that women are more sensitive to the social conditions of an experiment.

  ³ While Eckel and Grossman (1998) find women to be more generous in dictator games, Bolton and Katok (1995); Cason and Mui (1997); Cox (2002); Ben-Ner, Kong, and Putterman (2004); and Dufwenberg and Muren (2006b) report no significant gender differences.

  ⁴ Some studies conclude that the more generous sex depends on the context: Andreoni and Vesterlund (2001) find that women are more generous than men when the price of giving is high, while men give more when the price is low; and Cox and Deck (2006) conclude that the more generous sex depends on the size of the stakes, degree of anonymity, and possibility of reciprocity.

  ⁵ A number of studies compare giving when the dictator has information about the recipient, including gender. See Iris Bohnet and Bruno S. Frey (1999); Gary Charness and Uri Gneezy (2008, forthcoming); Gad Saad and Tripat Gill (2001a, 2001b); Jorn P. W. Scharlemann, Catherine C. Eckel, Alex Kacelnik, and Rick K. Wilson (2001); Terence C. Burnham (2003); Ben-Ner, Kong, and Putterman (2004); and Dufwenberg and Muren (2006b). As researchers increasingly conduct dictator experiments beyond the confines of the developed Western countries, it may become clear that cultural and sociological forces also impact gender differences in behavior.

  ⁶ Norman Frohlich, Joe Oppenheimer, and J. Bernard Moore (2001) report that dictators who stated that they viewed the experiment as a game they were trying to win gave less to recipients than dictators who did not view the experiment as a game.

  ⁷ See Gerald Marwell and Ruth Ames (1981), John Carter and Michael Irons (1991), and Robert H. Frank, Thomas Gilovich, and Dennis T. Regan (1993). In addition, Ernst Fehr, Micahel Naf, and Klaus M. Schmidt (2006) find that economics/business majors more often choose allocations of payoffs based on efficiency motives while other majors tend to prefer allocations that reduce inequality among the recipients.

  ⁸ Eckel and Grossman (1996) also used The American Red Cross in their dictator experiments.

  ⁹ The Nash equilibrium is a cooperative agreement between the two team members that occurs when the alternative, or failure to reach an agreement, is associated with some threat point payoff for each member. We assume a simple utility function such that an individual's utility is highest when his or her donation is what was chosen in the individual round and drops off in a linear, one-to-one fashion as the donation deviates from that initial level in a positive or negative direction. For person one who gave D₁ in the individual round and gives D_t in the team round, utility in the team setting would be [D₁−|D₁−D_t|]. We also assume that each individual's threat point, the donation they will give if they fail at bargaining, is their partner's donation. In the situation where two individuals gave different amounts in the individual setting and must agree on a common gift, D_t, the Nash equilibrium is the D_t that maximizes the product of the excess of person one's utility of D_t above his utility of the threat point and the excess of person two's utility of D_t above his utility of the threat point. The D_t that maximizes this product is (D₁+ D₂)/2, or the average of the partners' first round gifts, and the Nash equilibrium team gift would be D₁+ D₂.

¹⁰ Social information may influence giving because it acts as a reference point or because the person wants to conform to social norms (Jen Shang and Rachel Croson 2006), so that obtaining such information may affect giving even when others are not aware of how much a person gives, that is, in situations that preserve anonymity. The influence of social information has been examined by: Timothy N. Cason and Vai-Lam Mui (1998), Iris Bohnet and Richard Zeckhauser (2004), Bruno S. Frey and Stephan Meier (2004), Shang and Croson (2006), and Stephan Meier (2006, 2007).

¹¹ The possibility that social image or non-anonymity may affect decisions has been studied by Gary E. Bolton and Rami Zwick (1995); Alvin E. Roth (1995); Elizabeth Hoffman, Kevin McCabe, and Vernon L. Smith (1996); Gary E. Bolton, Elena Katok, and Rami Zwick (1998); and Dufwenberg and Muren (2006b).

¹² Though this paper focuses on measuring gender differences in altruism, an in-depth explanation of the causes of differences in such behavior is beyond the scope of the present work. Biological, social, psychological, or spiritual considerations may influence behavior. For example, as pointed out by a reviewer, evolutionists might argue that an increased generosity of males when paired with females is a type of sexual signaling behavior. On the other hand, women may be socialized to be more giving than men, and women's identification as mothers or caregivers may lead to altruistic acts. Finally, spiritual or religious values usually foster care or compassion for others.

¹³ Linda Babcock and Sara Laschever (2003) show that women are less likely to negotiate and less effective at negotiating starting salaries.

¹⁴ Equation (2) imposes the constraint that the coefficient on g ₁ in equation (1) is one and allows the constant to represent the sole social image effect. If in estimating equation (1), the coefficient on g ₁ were greater than one, for example, the coefficient on g ₁ as well as the constant would incorporate the social image effect. The initial estimate of giving before any influence by one's partner would be greater than one's initial gift by some percentage, and that percentage would be a part of the social image effect. In the empirical section we test whether this constraint is justified.

¹⁵ All monetary amounts reported in this paper are in US dollars.

¹⁶ Because we wanted anonymous individual giving to serve as a benchmark with which to compare the joint-giving decisions, we did not reverse the order of the allocation decisions because then the previous discussion with the partner would affect the individual decisions. Cason and Mui (1997) ran their experiments with both orders, individual first then team and team first then individual. They did not find changes in giving between the two orders to be significantly different.

¹⁷ Eckel and Grossman (1996) report that subjects gave 30.1 percent of the money to the American Red Cross compared to the 10.6 percent given to anonymous recipients. They recruited participants from economics, accounting, finance, and psychology courses, which may partly explain the smaller giving than found here, where students come from a more diverse set of courses and were not recruited. Since we are primarily interested in differences in giving among participants and changes in giving from individual to joint giving, the absolute amount of average giving in the population of participants should not affect our findings.

¹⁸ Eckel and Grossman (1998) find in their US experiments that women donate twice as much as men ($1.60 compared to $0.82 out of $10), and men are more likely to keep all of the money (60.0 percent of men and 46.7 percent of women).

¹⁹ While US business majors give significantly less than other majors, we do not find this to be true for economics majors, thus we include the latter with other majors. This result suggests that students with a more self-interested outlook may choose to major in business rather than economics when the choice is available, while such students may major in economics in institutions that do not offer an undergraduate business degree. Similarly, Bruno S. Frey and Stephan Meier (2003) report that business economics majors in Switzerland (which is similar to US business administration majors) contribute less to a social fund than other majors while political economics majors (similar to the US economics major) do not contribute less than other majors.

²⁰ These results are consistent with those of Dufwenberg and Muren (2006a) who also find that mixed-sex groups are the most generous.

²¹ Cason and Mui (1997) find that more pairs increase giving above the mean of individual giving than reduce giving below the mean and conclude this is evidence supporting the Social Comparison Theory. However, they do not separate groups by gender, so their result may have been driven by mixed-sex groups.

²² This result, which conflicts with that of Cason and Mui (1997) who found women to have greater influence, might seem strange since Tables 6 and 7 show that mixed-sex group giving does tend to move towards the more altruistic partner, which is often the woman. However, the social image effect in these groups may push team giving above the Nash equilibrium.

²³ The sample is not really the full sample since in groups where members give the same team gift, the data from the two individuals are mirror images and cannot both be used. The full sample includes one member from each group where the group members give the same amount (the more altruistic member when initial giving is different) and both members from each group where members give different team gifts.

²⁴ Econometric tests (adding a dummy variable for sex to the equation [3] specification) reject the hypothesis that the social image effect differs for men and women. We also estimate the equation for the mixed-sex groups who agree on team giving but use only the males as our observations. In this specification the coefficient on the difference between partner and own gift is 0.52 implying a larger impact of women than men but not significantly so. In this specification, the constant is positive although not significantly different from zero.

²⁵ In the teams that agree on a common gift, by construction the influence of the two partners sums to one. In same-sex teams, each partner's influence is roughly equal while in mixed-sex teams the less altruistic partner has about three times the influence (0.74) of the more altruistic partner (0.26) In teams that do not agree on a common gift there is a big gap between team members' giving as each member only moves about 17 percent of the way toward his or her partner's gift.