Giving, taking, and gender in dictator games

Transcription

1 CBESS Discussion Paper Giving, taking, and gender in dictator games by Subhasish M. Chowdhury*, Joo Young Jeon*, and Bibhas Saha** * School of Economics, University of East Anglia ** Economics in the Business School, Durham University Abstract We investigate whether a pure framing has any effect on the decisions made in a dictator game. We run a between subject dictator game with a giving and a taking frame whilst keeping the strategy space the same. Complying with the literature we find no overall difference in the amount allocated to the recipient across treatments. Contributing to the literature we further find that females are not only more altruistic than males; they also allocate significantly more to the recipient in the taking game compared to the giving game. Males do not show such behavior. A taking frame makes males significantly more selfish, but females significantly more egalitarian compared to a giving frame. JEL classification codes C91, D64, D84, J16 Keywords altruism, dictator game, taking game, framing, gender. Centre for Behavioural and Experimental Social Science University of East Anglia Norwich Research Park Norwich NR4 7TJ United Kingdom

2 Giving, Taking and Gender in Dictator Games * Subhasish M. Chowdhury, a Joo Young Jeon, a and Bibhas Saha b a School of Economics, Centre for Behavioural and Experimental Social Science, and ESRC Centre for Competition Policy, University of East Anglia, Norwich NR4 7TJ, UK b Economics in the Business School, Durham University, Durham DH1 3LE, UK October 07, 2014 Abstract We investigate whether a pure framing has any effect on the decisions made in a dictator game. We run a between subject dictator game with a giving and a taking frame whilst keeping the strategy space the same. Complying with the literature we find no overall difference in the amount allocated to the recipient across treatments. Contributing to the literature we further find that females are not only more altruistic than males; they also allocate significantly more to the recipient in the taking game compared to the giving game. Males do not show such behavior. A taking frame makes males significantly more selfish, but females significantly more egalitarian compared to a giving frame. JEL Classifications: C91 ; D64 ; D84 ; J16 Keywords: Altruism; Dictator-game; Taking-game, Framing, Gender * Corresponding author: Subhasish M. Chowdhury (s.modak-chowdhury@uea.ac.uk). We thank Jim Andreoni, Mike Brock, Anna Dreber, Bob Sugden, and the seminar participants at the University of East Anglia for useful comments and Zoë Bett for excellent research assistance. Any remaining errors are our own. 1

3 1. Introduction Researchers in economics and in other disciplines of social science have shown a considerable interest in understanding social preferences. Exploring the issues of why one individual behaves altruistically with others, or how to measure such altruism have fascinated researchers. In this context, the Dictator game has been employed as one of the most popular workhorse methods to understand and estimate altruism. In its most standard form (Kahneman et al., 1986; Forsythe et al., 1994) a subject, called the dictator, is given a certain amount of money and is asked to decide upon how much of that money to allocate between himself and a passive subject, called the recipient. Since the dictator does not otherwise have any incentive to share the money with the recipient, amount transferred is often used as a measure of altruism. Indeed, several experiments find consistently that dictators, on average, allocate non-trivial amounts of money (Camerer, 2003; Engel, 2011). Various social preference theories such as pure altruism (Becker, 1974), inequality aversion (Fehr and Schmidt, 1999; Bolton and Ockenfels, 2000), impure altruism (Andreoni, 1989), conditional altruism (Konow, 2010) etc. are proposed to explain this behavior. Due to its easy and interpretable structure, this game is employed in various forms and frames to investigate specific questions related to altruism. A simple Google scholar search to date returns almost 800 published and unpublished papers in economics, management, psychology, and other research areas implementing a dictator game. The interpretation that one gives in the game, however, is not free from criticism. It is observed that this game can be prone to the house money effect (Ruffle, 1998), be affected by experimenter demand effect (Zizzo, 2010), become sensitive to contexts (Eckel and Grossman, 1998) and cues (Bateson et al., 2006). Finally, it is also argued that the outcomes can be sensitive to the framing of the experiment. Understandably, if the results do depend on framing, they cannot be applied in a general way. In investigating the effects of framing for the first time, Suvoy (2003) introduces a taking game. Whereas in a standard dictator game (the giving game, or GG) the dictator owns some money and decides to give some of this to the recipient, in the taking game (TG) the recipient owns some money and the dictator decides upon how 2

4 much to take from him. Suvoy (2003) implements a within subject design in which the dictators play the two games simultaneously and could not find a difference in allocation (i.e., the money given in a GG or the money not taken in a TG) to the recipient. Bardsley (2008) and List (2007) extend this to the case when the dictator can either give to the recipient or take from the recipient. Hence, along with changing frames these designs also alter the strategy space. Results from both the experiments show that the availability of a taking option in a giving game decreases the amount given to the recipient. This confirms the sensitivity to framing in dictator allocation. 1 Since then, there have been a flurry of studies which employs this framing in investigating diverse questions. When both GG and TG framing are employed the result is, at most, mixed. Grossman and Eckel (2012) use the two frames where the recipient is a real charity and find no difference in allocation. The same result appears in Dreber et al. (2013) who run lab and online GG / TG games with money; and Rubinstein (2014), who runs them online with no money. Alevy et al. (2014) employ a 2 2 design in which GG and TG are played anonymously versus when the dictator decisions are observed. They find no difference in allocation under anonymity. However, males take less whilst females remain unaffected in allocation decisions when being observed. In a study along the same lines, Kettner and Ceccato (2014) investigate the effects of framing while interacting with revelation of dictator and recipient gender. They find no overall framing effect on allocation when gender is not revealed. But dictators significantly take less when the recipient is of the opposite gender. Korenok et al. (2013) employ a frame similar to List (2007), and find that not taking is not equivalent to giving. Cappelen et al. (2013) also employ a 2 2 design where GG and TG are interacted with whether the amount is earned versus when it is a windfall. They find that introducing a take option always decreases the allocation to recipient, but the effect of the origin of the amount is not significant. 2 Finally, Engel (2011) runs a meta-analysis and finds no effect of framing in dictator allocation. Zhang and Ortmann (2014) use the same data as Engel (2011) but employ a different technique and find result similar to Bardsley (2008) and List (2007). 1 Applying similar framing Brosig et al. (2007) investigate within subject behavior in repeated GG and TG games. They find that dictators take almost all the amount with experience. Keysar et al. (2008) apply deception and find that the act of giving is objectively considered more generous than the act of not taking. 2 Two further studies employ only a TG frame. Heinrich and Weimann (2013) find that dictators are sensitive to taking price. Chowdhury et al. (2014) find gender difference in the effect of social cue in TG. 3

5 Investigating whether giving in a GG frame is equivalent to not taking in a TG frame is important, since this allows one to understand the robustness of the game, as well as robustness of the behavioral theories that are constructed and tested using this game. The effect of strategy space is another important question to consider. But studies cannot compare framing whenever the strategy space is altered simultaneously with frames. Hence, the studies of Bardsley (2008), List (2007) and others who use similar designs are not appropriate to test pure framing effect on altruism. A specific test of the framing would be to compare GG and TG, while keeping the strategy space the same. Furthermore, to avoid any behavioral spillover, the design should be between-subject. In this study we do exactly this. We employ between subject GG and TG frames in which the dictator has the option to allocate the amount of 10. Our study is very close to that of Dreber et al. (2013) who run laboratory and online dictator games with GG and TG frames. They do not in find any difference in the amount allocated by framing. We further observe that the results in a dictator game, even under anonymity, are known to be affected by dictator gender (Eckel and Grossman, 1998). Female dictators are on average more generous than males (Engel, 2011). The context of an experiment also has asymmetric effect by gender (Croson and Gneezy, 2009). Moreover, the change in frame from GG to TG has implications regarding property rights of the pie (Oxby and Spraggon, 2008), related entitlement issues (Bylsma and Major, 1992), and possible cognitive biases such as endowment effects (Thaler, 1980) and status-quo bias (Samuelson and Zeckhauser, 1988). Since these issues are not gender independent (Dommer and Swaminathan, 2013), it is important to understand the effect of gender while investigating the framing effects. As a result, we replicate the study by Dreber et al. (2013), but both in design and in analysis maintain a specific focus on dictator gender. We find support for the overall result by Grossman and Eckel (2012) and Dreber et al. (2013) that giving in GG is indeed equivalent to not taking in TG. But investigating further, the framing has opposing effects on males and females. The latter are significantly more generous than males in TG but not in GG. They are also significantly more generous in TG than in GG, but the males are only insignificantly less generous in TG than in GG. Moreover, there is no difference in the allocation distribution in GG by gender. But the allocation distributions are polar opposites for males and females in TG. 4

6 Whereas most male dictators show pure selfishness under TG, females tend to behave as egalitarians. We hence conclude that framing does not affect overall allocation, but its effect is asymmetric and opposing across gender. The approach is as follows. Section 2 describes the experimental design, Section 3 reports the results, and Section 4 concludes. 2. Experimental Design We employed a between-subject Dictator game with 2 treatments and 40 dictators in each treatment. To keep it gender balanced, in each treatment we recruited 20 male and 20 female dictators. Only one treatment was run in a particular session. In each session, subjects were randomly and anonymously placed into pairs and were asked to sit in cubicles. They were then assigned the role as either a dictator or a recipient. Each subject played only one role and the roles remained the same until the end of the session. Subjects were told they would receive a 3 show-up fee, which was the same for all subjects in this experiment. In the Giving treatment the dictator was given access to an additional 10 and could transfer any amount between 0 and 10 to the recipient. In the Taking treatment the recipient was given access to an additional 10 and the dictator could transfer any amount between 0 and 10 to himself. We implemented neutral words transfer instead of give and take, and access instead of belong to minimize any instruction effect. The roles of the recipients were passive, meaning they had to accept the dictators decision (please see the instructions in the Appendix). Each session consisted of two parts. In the first part, the dictators made decisions. In the second part, the recipients had to guess the amount the dictator had given or taken. If the absolute difference between the actual amount and the guess was within 50 pence, then the recipient received an extra 1. 3 As we did not find any treatment or gender effect in the guesses, this element is not discussed further. Subjects were students at the University of East Anglia. They were recruited through the online recruitment system ORSEE (Greiner, 2004) and did not have any prior experience with GG or TG. The sessions were computerized with z-tree (Fischbacher, 3 This incentive mechanism for guess is similar to the ones in Brañas-Garza and Rodriguez-Lara (2014) or Chowdhury and Jeon (2013). The instructions for the second part was given only after the first part was done. It was mentioned in the instruction of the first part that the recipient s decision was payoff irrelevant to the dictator, restricting any strategic interaction between dictator decision and recipient anticipation. 5

7 2007). A subject could participate in only one session. Each session took around 30 minutes and the average payment was Results Table 1 shows the average amount allocated (amount given in the GG, or 10 amount taken in the TG) for both aggregated data and also for male and female by treatment. In the GG, an average of 1.93 is given to the recipients. In the TG, the dictators take on average 7.77 and as a result, the amount left to the recipient is A Mann-Whitney test shows no significant difference in final amount allocated to recipients between the two treatments (p-value = 0.78). This result is consistent with the results of Dreber et al. (2013) who also find no framing effects in all three experiments they run. Table 1. Average (Standard Dev) allocation to recipient Data Giving game Taking game Mann-Whitney test (Giving vs. Taking) All (40 obs. / treatment) (1.739) (2.156) No difference (p=0.778) Male (20 obs. / treatment) (1.611) (1.702) No difference (p = 0.112) Female (20 obs. / treatment) (1.895) (2.097) Difference at 10% ( p = 0.079) Mann-Whitney test (Male vs. Female) No difference (p=0.879) Difference at 1% (p= 0.005) We now investigate these actions across gender. Male dictators on average allocate 1.83 in GG and 1.20 in TG to the recipient, but a Mann-Whitney test shows marginal indifference at conventional significance level (p-value = 0.11). However, the average allocation by female dictators are 2.03 in GG and 3.26 in TG, and the difference is significant (p-value = 0.08). The results also confirm within treatment gender differences. In the GG, there is no difference in the amount given by gender. The 6

8 TG, however, shows gender differences in the amount left for the recipient. The average amount left by male dictators is only 1.20 whereas for female dictators it is A Mann-Whitney test confirms that the difference is significant at a 1% level. To test for robustness we run a set of OLS regressions as reported in Table 2. The dependent variable is the amount allocated to the recipient and the independent variables are treatment dummy, gender dummy, their interactions and an age dummy (Age 21 =1 if age 21). We run the analysis for the whole data and separately for males and females. Table 2. Regression of amount allocated to the recipient OLS Total Total Male Female Intercept *** ** *** *** (0.416) (0.449) (0.438) (0.492) Giving game ** (0.442) (0.565) (0.650) Female *** (0.431) TG Female *** (0.586) GG Male (0.604) GG Female (0.586) Age (0.449) (0.439) (0.571) (0.664) # of Obs Adjusted R Standard errors in parentheses. ***, ** and * indicates significance at the 1%, 5%, and 10% level. In the first regression given by Table 2, we use the dummy variable of GG to test for the framing effect while controlling for gender and age. Complying with the result in Table 1, the coefficient for GG is insignificant but the coefficient of Female is positive and significant at 1% level. This implies that there is no treatment effect, but that female dictators are more generous than males. This mirrors the findings of Dreber et al. (2013, Table 1), who report the same result from their lab experiment. 7

9 This result, however, does not reveal any effect of framing by gender. Dreber et al. (2013) also do not focus on this issue. Therefore, to examine the gender effect further, we introduce an interaction of gender and treatment. This is given in Colum 3 of Table 2 and shows that females allocate significantly higher amounts in the TG frame compared to their male counterparts. Then we run the same regression, without the interaction terms, for males and females separately. Discovering a gender-wise treatment effect, we find that males do not change their behavior due to framing (p-value for GG = 0.188), but females allocate more amount in the TG compared to the GG (p-value for GG = 0.049). Until now we have analyzed the average amount allocated, and found asymmetric effects of framing on gender. But this analysis cannot reveal the mechanism through which this asymmetry arises, i.e., any distribution differences. Understanding this spread, however, is important since it enables us to understand also the distribution of the social type of dictators. To visualize the allocation distribution in these two frames, we plot the proportion of dictators by amount allocated to the recipients in Figure 1. We note now that whereas the allocation in TG follows a bimodal distribution between selfish (who takes the whole pie) and egalitarian (who splits the pie to half) dictators. However, it does not seem to follow a particular pattern for the GG. This suggests that although in terms of the average allocation giving is indeed equivalent to not taking, in terms of the social type of the dictators the two frames seem not to be equivalent. Figure 1. Allocation in GG and TG. % of Dictators 50% 40% 30% 20% 10% 0% Amount aloocated GG TG Figure 1 prompts us to investigate the framing effect on dictators social type by gender and we plot the allocation distribution in these two frames by gender in Figure 2. 8

10 As can be noted, the allocation distribution for GG does not show a specific pattern for either gender. But it is right skewed for males and left skewed for females in the TG, i.e., the overall bimodal distribution in the TG comes from a peak in the egalitarian allocation by the female dictators and a peak in the selfish allocation by the male dictators. Figure 2. Allocation in GG and TG by gender Male Female % of Dictators 60% 50% 40% 30% 20% 10% 0% Amount allocated GG TG % of Dictators 60% 50% 40% 30% 20% 10% 0% Amount allocated GG TG To test the significance of these observations, we run two sets of Probit models. The dependent variable in the first set is a dummy for whether a subject is selfish (allocates 0), and in the second set a dummy for whether a subject is egalitarian (allocates half), with independent variables of frames and age. We run the regressions for the whole data, and then separately for males and females. The marginal effects for the gender specific regressions are reported below. Table 3. Probit Regressions of dictators type Male Female Y=Selfish Y=Egalitarian Y=Selfish Y=Egalitarian Giving game ** * (0.156) (0.106) (0.146) (0.146) Age (0.166) (0.111) (0.147) (0.161) # of Obs Pseudo R Standard errors in parentheses. ***, ** and * indicates significance at the 1%, 5%, and 10% level. 9

11 Further supporting the observations in Figure 2, the probit models reveal that there indeed exists a framing effect across gender. The first two regressions show that the male dictators tend to be more selfish in the TG compared to the GG, but the frame does not affect their likelihood of being egalitarian. In contrast, the female dictators tend to be more egalitarian in the TG compared to the GG, but the frame does not affect their likelihood of being selfish. Finally, the frames do not show a significant effect in the whole data (and hence we do not report the regressions), further supporting the overall results of the existing studies (Grossman and Eckel, 2012; Dreber et al., 2013) Discussion We investigate whether a pure framing effect exists on a dictator game when the strategy space remains the same across treatments. We employ a giving and a taking frame and compare the amount given in the giving frame with the amount left for the recipient in the taking frame. Both non-parametric tests and regressions reaffirm the observations of Grossman and Eckel (2012) and Dreber et al. (2013) that overall there is no framing effect within the dictator game and giving is indeed equivalent to not taking. However, we further investigate the results by gender and find that framing has opposing effects for female and male dictators. Females take less in a taking game compared to a giving game yet males do the opposite, although the effect is significant only for females. Furthermore, we find that this occurs due to a spike in egalitarian-type female dictators and selfish type male dictators in the taking game. The male dictators are significantly more likely to be selfish in the taking frame compared to the giving frame, but the female dictators are significantly more likely to be egalitarian in the taking frame compared to the giving frame. We find these results despite implementing a computerized experiment with strictly neutral instructions, using words such as access to instead of belongs to and transfer instead of give or take. We believe that the results would have been stronger if a more frame specific instruction was implemented and the experiment was run 4 Although it cannot capture the difference in subject social-type, to further investigate any difference in allocation distribution we also ran Kolmogorov-Smirnov tests corresponding to Figure1 and Figure 2. These tests, conforming to the OLS results, show no framing effect in allocation in the whole data and for males (p-values of and 0.238, respectively); but a significant effect for females (p-value = 0.048). 10

12 manually (giving a stronger context effect). Nevertheless, our results show that a framing effect exists via gender even in the weakest possible design. We infer that the gender specific framing effect is observed due to two possible rationales. First, the framing of GG and TG introduces a salience of property right of the pie to be divided. Whereas in the GG frame the property right belongs to the dictator, in the TG frame it belongs to the recipient. This affects the notion of personal entitlement among the dictators. Existing studies (Bylsma and Major, 1992) have already pointed out gender differences in personal entitlement, and have shown that males perceive more entitlement that females. As a result, in TG frame females take less and males take more. Second, the change in frame also brings in issues of cognitive biases such as endowment effects (Thaler, 1980) or a status-quo bias (Samuelson and Zeckhauser, 1988). It is known from the study of Croson and Gneezy (2009) that the context of an experiment has asymmetric effect on males and females. It is very much true also for the cognitive biases discussed above (Dommer and Swaminathan, 2013). Hence, once we introduce these biases through the framing, they inject gender specific effects. Females respond to the endowment effects or the status-quo bias by taking less (specifically being egalitarian, as seen in Figure 2) in TG, while males do the opposite. This result is of interest for various reasons. First, this result essentially implies that one can alternatively use a giving or a taking frame without changing the strategy space, so long as the investigation is not focused on gender. If this is true, the stream of literature that investigates the dictator game will be able to interpret the results in a more succinct manner. Second, this result sheds light on a very active literature regarding the gender effect on behavioral decision making. Since the giving frame does not invite social-type differences in the dictator by gender but a taking frame does, this warrants further investigation in this area of research. Whilst possible explanations for the difference are given above, these cannot be certified within the current experiment. A further and concrete investigation of the reason of our discovered gender difference is seen as vital and we leave it as a key topic of future research. 11

13 References Alevy, J. E., Jeffries, F. L., & Lu, Y. (2014). Gender-and frame-specific audience effects in dictator games. Economics Letters, 122(1), Andreoni, J. (1989). Giving with impure altruism: applications to charity and Ricardian equivalence. Journal of Political Economy 97 (6): Bardsley, N. (2008). Dictator game giving: Altruism or artefact? Experimental Economics, 11(2), Bateson, M., Nettle, D., & Roberts, G. (2006). Cues of being watched enhance cooperation in a real-world setting. Biology Letters, 2(3), 412. Becker, G.S. (1974). A theory of social interactions. Journal of Political Economy 82: Bolton, G., & Ockenfels, A. (2000). ERC A theory of equity, reciprocity and competition. American Economic Review 90: Brañas-Garza, P., & Rodriguez-Lara, I. (2014). What Do We Expect of Others? MPRA Working Paper No Brosig, J., Riechmann, T., & Weimann, J. (2007). Selfish in the end?: An investigation of consistency and stability of individual behavior. FEMM Working Paper No. 05/2007. Bylsma, W. H., & Major, B. (1992). Two routes to eliminating gender differences in personal entitlement: Social comparisons and performance evaluations. Psychology of Women Quarterly, 16(2), Camerer, C. (2003). Behavioral game theory: Experiments in strategic interaction. Princeton University Press. Cappelen, A. W., Nielsen, U. H., Sørensen, E. Ø., Tungodden, B., & Tyran, J. R. (2013). Give and take in dictator games. Economics Letters, 118(2), Chowdhury, S.M., & Jeon, J. (2013). Altruism, Anticipation, and Gender, UEA Centre for Behavioural and Experimental Social Science Working Paper No Chowdhury, S.M., Jeon, J. & Saha, B. (2014). Eye-images in altruism experiments: Social cue or experimenter demand effect? UEA Economics Working Paper No. 67. Croson, R., & Gneezy, U. (2009). Gender differences in preferences. Journal of Economic Literature 47: Dommer, S. L., & Swaminathan, V. (2013). Explaining the endowment effect through ownership: the role of identity, gender, and self-threat. Journal of Consumer Research, 39(5), Dreber, A., Ellingsen, T., Johannesson, M., & Rand, D. G. (2013). Do people care about social context? Framing effects in dictator games. Experimental Economics, 16(3), Eckel, C.C., & Grossman, P.J. (1998). Are Women Less Selfish than Men? Evidence from Dictator Experiments, Economic Journal 108: Engel, C. (2011). Dictator games: a meta study. Experimental Economics, 14,

14 Fehr, E., & Schmidt, K. (1999). A theory of fairness, competition, and cooperation. Quarterly Journal of Economics 114: Fischbacher, U. (2007). z-tree: Zurich toolbox for ready-made economic experiments, Experimental Economics, 10 (2), Forsythe, R., Horowitz, J., Savin, N., & Sefton, M. (1994). Fairness in Simple Bargaining Experiments, Games and Economic Behavior 6: Grossman, P., & Eckel, C. (2012). Giving versus taking: a real donation comparison of warm glow and cold prickle in a context-rich environment. Monash University Department of Economics Working paper No Heinrich, T., & Weimann, J. (2013). A note on reciprocity and modified dictator games. Economics Letters, 121(2), Kahneman, D., Knetsch, J., & Thaler, R. (1986). Fairness as a constraint on profit seeking: entitlements in the market. American Economic Review 76: Kettner, S. E., & Ceccato, S. (2014). Framing Matters in Gender-Paired Dictator Games. University of Heidelberg Working paper No Keysar, B., Converse, B. A., Wang, J., & Epley, N. (2008). Reciprocity Is Not Give and Take Asymmetric Reciprocity to Positive and Negative Acts. Psychological Science, 19(12), Konow, J. (2010). Mixed feelings: Theories of and evidence on giving Journal of Public Economics 94: Korenok, O., Millner, E. L., & Razzolini, L. (2013). Taking, giving, and impure altruism in dictator games. Experimental Economics, List, J. A. (2007). On the interpretation of giving in dictator games. Journal of Political Economy, 115(3), Oxoby, R. J., & Spraggon, J. (2008). Mine and yours: Property rights in dictator games. Journal of Economic Behavior & Organization, 65(3), Rubinstein, A. (2014). A Typology of Players: between Instinctive and Contemplative, Mimeo. Available at Ruffle, B. J. (1998). More Is Better, But Fair Is Fair: Tipping in Dictator and Ultimatum Games. Games and Economic Behavior 23: Samuelson, W., & Zeckhauser, R. (1988). Status Quo Bias in Decision Making. Journal of Risk and uncertainty, 1, Suvoy, R. (2003). The effects of give and take framing in a dictator game. Unpublished Honors Thesis, University of Oregon. Thaler, R. (1980). Towards a Positive Theory of Consumer Choice. Journal of Economic Behavior and Organization, 1, Zhang, L., & Ortmann, A. (2013). The effects of the take-option in dictator-game experiments: a comment on Engel s (2011) meta-study. Experimental Economics, 1-7. Zizzo, D. J. (2010). Experimenter demand effects in economic experiments. Experimental Economics, 13(1),

15 Appendix: Instructions 1. Instruction for Dictator in Taking game Welcome to this experiment! In this experiment each of you will be paired with a different person. You will not be told who you are matched with during or after the experiment, and he or she will not be told who you are either during or after the experiment. Your decisions will be strictly anonymous and cannot be linked to you in any way. The experiment has two parts and is conducted as follows: Everyone in this room has already been allocated a show up fee of 3. You have been paired with someone else in the room. The other person you are paired with has access to an additional 10. In the first part of the experiment, you will have to make a simple decision. You have to decide what portion, if any, of the 10 to transfer to yourself. Your choice can be anywhere from 0 to 10, in 1p increments. Your take-home earnings from this experiment will be your initial 3 show up fee plus the money you transfer from the person you are paired with. The earnings of the person you are paired with will be his/her 3 show up fee plus the money left over from the 10 after you transfer to yourself. In the second part of the experiment, the person you are paired with will make a decision, but that decision will NOT affect your earnings. You will have 1 minute to come to a decision about your choice. Please do not talk to the other people in this room until your session is completed. Do not be concerned if other people make their decisions before you. 14

16 2. Instruction for Dictator in Giving game Welcome to this experiment! In this experiment each of you will be paired with a different person. You will not be told who you are matched with during or after the experiment, and he or she will not be told who you are either during or after the experiment. Your decisions will be strictly anonymous and cannot be linked to you in any way. The experiment has two parts and is conducted as follows: Everyone in this room has already been allocated a show up fee of 3. You have been paired with someone else in the room. You have access to an additional 10. The other person you are paired with does not have access to that extra 10. In the first part of the experiment, you will have to make a simple decision. You have to decide what portion, if any, of the 10 to transfer to the person you are paired with. Your choice can be anywhere from 0 to 10, in 1p increments. Your take-home earnings from this experiment will be your initial 3 show up fee plus the money left over from the 10 after you transfer to the person you are paired with. The earnings of the person you are paired with will be the amount you transfer to him/her plus his/her 3 show up fee. In the second part of the experiment, the person you are paired with will make a decision, but that decision will NOT affect your earnings. You will have 1 minute to come to a decision about your choice. Please do not talk to the other people in this room until your session is completed. Do not be concerned if other people make their decisions before you. 15