Journal of Experimental 59 (2015) 67–76

Contents lists available at ScienceDirect

Journal of Experimental Social Psychology

journal homepage: www.elsevier.com/locate/jesp

The Scrooge effect revisited: Mortality salience increases the satisfaction derived from prosocial behavior

Tomasz Zaleskiewicz a,⁎, Agata Gasiorowska a, Pelin Kesebir b a University of Social Sciences and Humanities, Faculty in Wroclaw, Ostrowskiego 30b, 53-238 Wroclaw, Poland b University of Wisconsin-Madison, Madison, WI 53706, USA

HIGHLIGHTS

• People reminded of their mortality act more generously in simple economic games • Under mortality salience, satisfaction increases with more generous decisions • Satisfaction derived from generosity is linked to lower thought accessibility

article info abstract

Article history: According to , people deal with the potential for that results from the Received 15 May 2014 knowledge of the inevitability of death by holding on to sources of value that exist within their cultural Revised 26 March 2015 worldview. Acting prosocially is one such source of value, and previous research suggests that reminders of Available online 8 April 2015 mortality increase the desire for prosociality. In three studies, we tested the hypothesis that mortality reminders would lead to more generous allocation of financial resources and to more satisfaction derived from acting Keywords: Terror management theory generously. Using the dictator game (Study 1), the ultimatum game (Study 2), and a quasi-naturalistic giving Mortality salience situation (Study 3) we showed that participants reminded of their mortality were not only more generous but Prosocial behavior also more satisfied the more money they donated. Moreover, Study 3 demonstrated that people reminded of Satisfaction their mortality derived higher satisfaction from prosocial behavior and such behavior was associated with better suppression of death-related thoughts. We conclude that acting prosocially in the face of mortality thoughts effectively soothes death anxiety and in turn produces psychological satisfaction. © 2015 Elsevier Inc. All rights reserved.

Classical economic theory assumes that in situations involving the current paper is to inquire into how awareness of personal mortality decisions about the allocation of financial goods between oneself and dif- affects prosociality in economic decision-making and in particular, how it ferent targets, people are motivated by material self-interest, and strive to affects the satisfaction one derives from prosocial behaviors. maximize personal benefits (Mullainathan & Thaler, 2001). However, as Terror management theory (Greenberg, Solomon, & Pyszczynski, empirical studies are increasingly making clear, people frequently reveal 1997) posits that in order to manage the anxiety born from the aware- social (other-regarding) preferences in economic decision-making and ness of their mortality, people strive for value and meaning in the show a willingness to share their material assets with even anonymous context of their worldviews. Based on this insight and previous research interaction partners (Camerer, 2003; Cartwright, 2011). A concern with on the topic demonstrating the “Scrooge effect” (Jonas, Schimel, preventing unfairness in allocation decisions and punishing free riders Greenberg, & Pyszczynski, 2002), we expected that reminders of also emerge as prevalent behavior patterns in these studies (Fehr & mortality would increase the desire for prosociality, leading to more Schmidt, 1999). That said, prosociality concerns in economic decision- generous and less selfish allocation of financial resources. More impor- making are affected by various situational factors (Camerer, 2003). One tantly, we hypothesized that mortality salience would increase people's such factor is the salience of thoughts related to one's mortality. Previous satisfaction with their prosocial behaviors. If mortality reminders research has established that under certain conditions, thinking about induce a shift in the more prosocial direction and make people genuine- one's mortality increases prosocial attitudes and behavior. The aim of ly interested in giving and sharing, then we would find that in the face of mortality thoughts, participant satisfaction would increase with more generous allocation of resources, as such a behavior would help to ⁎ Corresponding author. Fax: +48 71 7507208. E-mail addresses: [email protected] (T. Zaleskiewicz), suppress death-related thoughts. That satisfaction, in turn, should be [email protected] (A. Gasiorowska), [email protected] (P. Kesebir). related to lower accessibility of death-related thoughts. Such a pattern

http://dx.doi.org/10.1016/j.jesp.2015.03.005 0022-1031/© 2015 Elsevier Inc. All rights reserved. 68 T. Zaleskiewicz et al. / Journal of Experimental Social Psychology 59 (2015) 67–76 of results would support our argument that unselfish economic be- arouse personal fear or remind participants of their own vulnerability, haviors in the face of mortality thoughts soothe existential anxiety such as signing an card or helping to a wheelchair- and produce genuine satisfaction. bound person, they in fact found that mortality thoughts significantly decreased support instead of increasing it. Terror management theory More recently, Jonas, Sullivan, and Greenberg (2013) applied focus theory of normative conduct (Cialdini, Kallgren, & Reno, 1991)tofurther Humans, unlike other animals, are sophisticated enough in their refine the conditions under which mortality reminders lead to a pro- mental abilities to be aware of the fragility of life and the inevitability social shift, particularly towards an outgroup. They found that although of ultimate death. Terror management theory (TMT; for overviews, mortality reminders reduced donations to a foreign charity, in line with see Kesebir & Pyszczynski, 2012; Pyszczynski, Greenberg, Solomon, & research showing that MS induces negativity toward the outgroup Koole, 2010) proposes that the awareness of mortality has the potential (Greenberg et al., 1997), this effect of MS on decreased outgroup to generate paralyzing anxiety and that the management of this poten- charitability was eliminated when the norm of generosity was primed. tial anxiety is essential for effective functioning. According to the theory, Similarly, the researchers demonstrated that MS led to higher generos- people develop an anxiety buffering system that, as long as it is func- ity toward a stranger at the expense of the self in a dictator game para- tional, protects against existential anxiety and provides psychological digm, but only when combined with a fairness prime. As demonstrated equanimity. The key ingredients of this anxiety buffer are a sense of by this study, a concern for fairness may also contribute to people mak- meaning, security, value, relatedness, and transcendence. These ingredi- ing more generous monetary choices when they are reminded of their ents are typically found in cultural worldviews, self-esteem, and close mortality. Given that a fair world is a more benevolent, safe and control- personal relationships. Because these psychosocial entities buffer lable place and the importance of fairness to societies (Fehr & Schmidt, against death anxiety, people are highly motivated to seek and maintain 1999), upholding the fairness value could confer protection against anx- them and defend them against threats (Greenberg et al., 1997; Solomon, iety in the face of death thoughts. Indeed, Van den Bos and Miedema Greenberg, & Pyszczynski, 1991). Since the inception of TMT, a large (2000) showed that thinking about death made fairness a more impor- body of research has supported hypotheses generated by the theory tant issue to participants. (Burke, Martens, & Faucher, 2010). Accordingly, when thoughts of mor- In summary, considerable evidence suggests that mortality thoughts tality are activated, people become more invested in their cultural lead to more prosocial attitudes and behaviors, especially toward the worldview, self-esteem, and close relationships. Conversely, when ingroup. In the present work, we aimed to conceptually replicate this one's cultural worldview, self-esteem or close personal relationships finding and more importantly, to refine our understanding of what drives are threatened, anxiety increases and death-related thoughts become this effect. Specifically, we hypothesized that in the face of mortality more salient in the consciousness (Hayes, Schimel, Arndt, & Faucher, thoughts, people would derive higher satisfaction from acting prosocially, 2010). as prosocial behaviors are a potent source of existentially protective ingredients such as meaning, value, safety, and self-transcendence. Mortality salience and prosociality Generosity toward another person or a prosocial cause can fulfill cru- cial psychological needs including self-esteem, belongingness, and According to TMT, acts that provide a person with a sense of value, competence, which in turn would soothe concerns about one's fi- meaning, relatedness, and self-transcendence help to buffer death anx- niteness and fragility. We thus were interested in the hypothesis iety. In virtually any culture, prosocial behaviors are widely endorsed, that in the presence of mortality thoughts, people would derive and are capable of producing the aforementioned key ingredients. more joy from giving than in the absence of such thoughts. In addi- TMT also predicts that reminders of mortality increase the perceived tion, we expected that the joy of acting prosocially would be associ- importance of social norms and conformity to them. People adhere to ated with more effective suppression of the death-related thoughts. social standards, because this offers a sense of security and self- esteem. It thus stands to reason that making death thoughts salient would cause people to become more generous and less selfish in their Overview of the studies choices. Acting in this way would affirm people's faith in their world- view and enhance their sense of security and value, thereby creating a To test these hypotheses, we conducted three experiments, in which buffer from death fears. participants received a financial endowment and were asked to allocate Supporting this line of reasoning, research has revealed that thinking it between themselves and another, anonymous person (Studies 1 and about one's mortality leads to increased prosocial attitudes and charita- 2) or a prosocial cause (Study 3). The first experiment employed the dic- ble behaviors. In a first demonstration of this, Jonas et al. (2002) found tator game, and the second experiment the ultimatum game. Both games that people interviewed in front of a home, compared to those have been extensively used to study people's economic decisions interviewed several blocks away, reported more positive attitudes to- (Camerer, 2003; Chaudhuri, 2009). Study 3 adopted a less structured ward charities. They furthermore showed that participants reminded and more real-world-like scenario than economic games, in which partic- of their mortality donated more money to a national (but not foreign) ipants (university students) were given some money and then asked to charity than participants in the control condition. The researchers divide it between themselves and their Student Government. To examine dubbed this the “Scrooge effect” in reference to Ebenezer Scrooge in the effects of mortality thoughts on monetary choices, in all studies partic- Charles Dickens's novella A Christmas Carol, who was transformed into ipants were randomly divided into two groups: the mortality salience a kind and generous person from a miserly and bitter one after an inti- group and the control group. We also examined the effects of mortality mate encounter with his own mortality. Further research on the Scrooge salience on participants' satisfaction with their decisions and on their effect has revealed some moderators and boundary conditions of this death thought availability. Participants were asked to report their own phenomenon. For example, mortality reminders have been demonstrat- satisfaction as well as their estimates of the satisfaction experienced by ed to lead to a prosocial shift mainly among individuals who endorse the other player (in Studies 1 and 2, which employed economic games). more proselfish attitudes (Joireman & Duell, 2005) and lower self- We expected, in line with the Scrooge effect, that participants in the transcendent values (Joireman & Duell, 2007, Study 1) to begin with, mortality salience condition would be more generous in their allocation and not necessarily among people already high in prosocial orientation. decisions than participants in the control condition. Moreover, we hy- Hirschberger, Ein-Dor, and Almakias (2008) furthermore showed that pothesized that higher prosociality would be associated with stronger whereas mortality reminders increase prosocial behaviors, the nature personal satisfaction in the mortality salience condition and reduced of the prosocial cause plays a decisive role. For prosocial causes that accessibility of death-related thoughts. T. Zaleskiewicz et al. / Journal of Experimental Social Psychology 59 (2015) 67–76 69

Amount of money of which was to push the recently evoked thoughts about death out of sent to the other consciousness. Previous research has established that such delay and party distraction tasks following mortality reminders lead to more robust ter- ror management effects (Pyszczynski, Greenberg, & Solomon, 1999). Mortality After participants completed the crossword puzzle, they were told salience Satisfaction that they have randomly been assigned to the role of a person who de- cides about allocating the sum of 30 PLN (i.e., Proposer). They were also informed that the participants in the study who have been assigned the Fig. 1. Hypothesized model of moderated mediation in Studies 1 and 2. Receiver role (supposedly students from the same university) were waiting in another room, but that they would remain mutually anony- Study 1. Mortality salience and decision-making in a dictator game mous. In reality, no second group of Receivers existed, and all parti- cipants in the study had been assigned to the Proposer role. All In Study 1, we employed a basic version of the dictator game (DG; participants thus received an envelope containing 30 PLN, in the form Forsythe, Horowitz, Savin, & Sefton, 1994) to test the effect of mortality of two 10 PLN banknotes and two 5 PLN coins. There also were two salience on prosocial behavior. In this basic version of the game, Player 1 empty envelopes on the desk, one labeled “Me”, and the other labeled (Proposer) receives a specific sum of money and has to divide this sum “Player 2”. Participants' task was to divide the money between the between herself and the anonymous Player 2 (Receiver). Receiver does two envelopes in any way they desired. To assure full confidentiality, not have the option to reject the offer. Hence a strictly rational, self- they were instructed to seal both envelopes, even if they decided not interested Proposer would be expected to keep all the money to herself. to give any money to the Receiver. Participants were also informed be- Offering any sum of money to Receiver signals prosociality on the part of fore they played the game that 1/3 of the players would be randomly Proposer. chosen to receive real payoffs after the study. We tested our hypotheses using a moderated mediation model (see After playing the DG, participants were asked to indicate how satis- Fig. 1 for a depiction of the model). Specifically, we predicted that mor- fied they were with their decision and how satisfied, in their opinion, tality salience (MS) would be both an independent variable affecting the second player would be. Both ratings were made on a scale ranging the amount of money sent to the other player, and a moderator of the from 1 (not satisfied at all)to10(fully satisfied). After that, the experi- relationship between the amount of money sent and the level of person- menter approached each participant, to decide by a toss of the dice if al satisfaction. The amount of money sent was hypothesized to mediate the participant and Player 2 would receive real payoffs. the relationship between MS manipulation and satisfaction with one's decision. Participants in the control condition were expected to keep more money for themselves than in the mortality salience condition. Results Moreover, participants in the MS condition were hypothesized to report higher satisfaction the more money they allocated to the other person No differences were found between the MS and control conditions in η2 b or the prosocial cause. either positive affect, F(1, 60) = 0.06, p = .80, .01 or negative affect, F(1, 60) = 2.21, p = .14, η2 = .03. The two groups, however, differed in Method how much money they allocated to the Receiver, F(1, 60) = 4.90, p = .01, η2 = .11. In line with our hypothesis, Proposers allocated more Participants money to Receivers in the MS condition (M = 15.32, SD = 4.99) than Sixty-two students (43 women) at a Polish university with a mean in the control condition (M =11.94,SD = 4.77). However, because age of 26.89 (SD = 8.51) participated in the study for credit. During the number of participants in both experimental conditions was rela- recruitment, participants were informed that they could get up to 30 tively small, we carried out a post hoc analysis of test power using Polish zlotys (app. $10) in the experiment. Sample size was not G*Power 3.1 software (Faul, Erdfelder, Lang, & Buchner, 2007). The predetermined, but was decided by how many participants signed up achieved power for ANOVA was 0.78, slightly below the conventionally for the time period during which the study was run (predetermined used statistical power level of .8 (Cohen, 1988). to be 3 days). No participants were excluded from the study. We also We also conducted an analysis of covariance to control for potential- fi confirm that we reported all the measures and conditions in all experi- ly related variables. No signi cant effects on money allocated to the η2 ments in this paper. Receiver were observed for gender, F(1, 56) = 3.06, p =.09, = 0.05, positive affect, F(1, 56) = 1.07, p = .31, η2 = 0.02, or negative af- η2 Procedure fect, F(1, 56) = 0.85, p = .36, = 0.01. The effect of age was marginally fi η2 1 After arriving in the lab, participants were seated in individual cubi- signi cant, F(1, 56) = 3.63, p =.06, = 0.06 . The main effect of the fi cles and had neither visual nor verbal contact with each other. They experimental manipulation remained signi cant when controlling for η2 were told that the study would examine the relationship between per- these four variables, F(1, 56) = 5.85, p =.02, = .10, achieved sonality and decision-making, and asked to complete a pen-and-paper power = 0.73. questionnaire. Participants in the MS and control conditions did not differ in their At the beginning of the experiment, participants were randomly personal satisfaction with their decision, F(1, 60) = 0.95, p =.76, η2 b assigned to either the mortality salience (n = 31) or control condition .01, or the predicted satisfaction for the Receiver, F(1, 60) = 1.78, η2 (n = 31). In the MS condition, participants completed a “fear of death p = .19, = .03. However, as predicted, MS moderated the relation- questionnaire” consisting of 12 short statements, to which they could ship between the amount sent to the Receiver and the level of personal respond with a ‘yes’ or ‘no’ (e.g., “I am very much afraid to die”). The satisfaction. To test this moderated mediation model (see Fig. 1), we aim of this questionnaire was not to measure fear of death, but to acti- conducted a multiple regression analysis. The dependent variables vate mortality thoughts. Participants in the control condition, in con- were personal satisfaction with one's decision and predicted satisfaction trast, filled out a “dental anxiety questionnaire” consisting of similarly for the Receiver. The model included three predictors: experimental worded items (e.g., “I am very much afraid of dental work”). 1 Following the manipulation, all participants completed the brief ver- The correlation between age and amount of money donated was positive, r(62) = 0.24, p = 0.06. We also tested for a potential interaction between mortality sa- sion of the Positive-Negative Affect Scale (PANAS; Watson, Clark, & lience and age on the amount of money donated, but the interaction was not significant, α Tellegen, 1988; Cronbach's 's = .79 for positive affect and .91 for neg- b = 0.16, se = 0.15, t =1.05,p = 0.30, 95% CI [−0.14, 0.45]. Older participants donated ative affect). Next, participants solved a crossword puzzle, the purpose more money than younger ones, irrespectively of the experimental condition. 70 T. Zaleskiewicz et al. / Journal of Experimental Social Psychology 59 (2015) 67–76

either, β = −.17, se = .11, t = −1.59, p = .12, 95% CI [−0.38, 0.04]. Amount of money sent to Receiver In other words, the amount of money sent to the Receiver was positively 0.33** 0.06 related to the predicted satisfaction of the Receiver in both the control and mortality salience conditions. Mortality 0.02 Proposer salience Satisfaction Discussion

Amount of money 0.44** Study 1 demonstrated, in the context of a simple economic game, sent to Receiver that people become more prosocial and offer more money to an anony- X mous person when primed with thoughts of their mortality. This result Mortality Salience is consistent with the assumptions of TMT and previous work on the Scrooge effect: In the face of death thoughts, people cope with existen- Fig. 2. Moderated mediation effect of mortality salience on amount of money sent by tial anxiety by upholding highly valued cultural norms that provide a Proposer and Proposer's satisfaction in the Dictator game. Standardized regression sense of meaning, value, security, and transcendence. Prosocial acts coefficients, **p b .01. are culturally endorsed resources for these critical ingredients, and as a result, prosociality is increased by mortality concerns. manipulation, amount of money sent to Receiver and the interaction of Importantly and in line with our predictions, when death thoughts these two variables (Fig. 2). Following recent advances in mediation were salient, participants derived higher joy from giving more. This analysis and recommendations concerning the interpretation of indirect was unlike the control condition, in which being generous came at a effects (Hayes, 2013), we conducted our moderated mediation analysis personal psychological cost. On the other hand, MS did not change using sampling with replacement, with a bias-corrected bootstrapping participants' predictions of how happy the Receivers would be with procedure (in this case, 10,000 samples were taken). All variables the amount of money they got. This suggests that when people act were z-scored prior to the analysis, to allow for standardized regression prosocially in the face of mortality thoughts, this has more to do with fi coef cients in the output. finding such behavior to be personally rewarding, than with imagining Investigating the predictors for satisfaction with one's decision, we that the other person will benefit more from the generous act. It seems found that the regression model with experimental manipulation, that prosociality does a good job of satisfying people's psychological amount of money sent to Receiver and their interaction as predictors needs when mortality thoughts are activated. was significant, F(3, 58) = 4.07, p = .01, R2 = .17, achieved power = 0.84. We found that no significant effect was exerted by either experi- Study 2. Mortality salience and decision-making in an mental manipulation, β =.06,se =.13,t = 0.51, p =.61,95%CI ultimatum game [−0.19, 0.32], or the amount of money sent to Receiver, β =.02, se =.13,t =0.14,p = .89, 95% CI [−0.23, 0.27]. However, as expected, To examine the same hypotheses about the effect of mortality re- the interaction between experimental manipulation and the amount of minders on prosocial orientation and the satisfaction derived from it, money was significant, β =.44,se =.13,t = 3.41, p = .001, 95% CI in Study 2 we used a different economic game, namely the ultimatum [0.18, 0.69]. In other words, the impact of the amount of money sent game (UG). Unlike the dictator game, in the UG, Player 2 is actively in- to the Receiver on Proposer satisfaction depended on whether the Pro- volved and has the power to reject unfair offers (Güth, Schmittberger, poser had been exposed to MS or not. & Schwarze, 1982). However, as recent studies show, Proposers in this We formally tested for moderated mediation using the boots- game make decisions not only to minimize the risk of their offer being trapping procedure. In our data, the 95% bootstrapped confidence inter- rejected, but also because they care about the norms of generosity and val for the indirect effect of money allocated to Receiver was between fairness (Camerer, 2003; Chaudhuri, 2009). If activation of death- −0.29 and −0.02, indicating that amount of money sent to the Receiver related thoughts increases prosociality, then we would expect the Pro- was a significant moderated mediator of the relationship between mor- posers in the MS condition to make more generous offers than the Pro- tality salience and Proposer satisfaction (indirect effect size = −.12, posers in the control condition. As in Study 1, we also examined effects boot se = .07). Accordingly, mortality thoughts increased the amount of mortality salience on participants' satisfaction following their choices. of money sent to the Receiver, β =.33,se =.12,t =2.73,p =.01, We tested the same moderated mediation model, in which mortality sa- 95% CI [0.09, 0.58]. Furthermore, after being reminded of their mortality, lience appears as both the independent variable influencing the amount the more money participants sent to the Receiver, the higher was their of money sent to the second player and the moderator of the relation- reported satisfaction, β = .50, se = .17, t = 2.84, p = .01, 95% CI [0.15, ship between the amount sent and personal satisfaction. 0.85]. In contrast, in the control condition, the more money participants sent to the Receiver, the less satisfaction they reported, β = −.37, se = .18, t = −2.01, p = .05, 95% CI [−0.74, −0.001]. To sum up, partici- Method pants primed with mortality thoughts sent more money to the Receiver and experienced higher satisfaction, the more money they sent. Partic- Participants ipants in the control condition sent less money and experienced lower Forty-one students (25 women) at a Polish university with a mean satisfaction, the more money they offered. age of 27.46 (SD = 9.10) participated in a two-condition experiment Investigating participants' predictions of the Receiver's satisfaction, in exchange for cafeteria coupons worth of 5 PLN (app. $1.70). During we found that the regression model with experimental manipulation, the recruitment, participants were informed that they could get up to amount of money sent to Receiver and their interaction as predictors 30 PLN (app. $10) in the experiment. Sample size was not decided in ad- was significant, F(3, 58) = 14.92, p b .001, R2 = .44, achieved vance, but was shaped by how many participants signed up for the time power = 0.99. The amount of money sent to the Receiver significantly period during which the study was run (predetermined to be 3 days). affected Proposers' predictions, β =.66,se =.10,t = 6.33, p b .001, No participants were excluded from the study. 95% CI [0.45, 0.87]. The more money they sent, the more they predicted that the Receiver would be satisfied. Experimental manipulation did not Procedure significantly affect predicted satisfaction, β = −.05, se =.10, The procedure was similar to the one used in the previous study, t = −0.49, p =.63,95%CI[−0.26, 0.16]. The interaction between the with the exception that participants played the ultimatum game instead manipulation and amount of money sent did not reach significance of the dictator game and the current study was computerized. T. Zaleskiewicz et al. / Journal of Experimental Social Psychology 59 (2015) 67–76 71

At the beginning of the study, as in Study 1, participants assigned to the experimental condition (n = 21) completed the fear of death ques- Amount of money sent to Responder tionnaire, whereas participants in the control condition (n = 20) com- 0.35* -0.14 pleted the dental anxiety questionnaire. Next, they reported their mood on PANAS (Cronbach's α's = .85 for positive affect and .92 for negative Mortality 0.09 Proposer affect), and solved a crossword puzzle. In the final part of the study, par- salience Satisfaction ticipants received detailed instructions about the ultimatum game. As noted above, the ultimatum game differs from the dictator game in Amount of money 0.50** that Player 2 (Responder) has the power to accept or reject the sent to Responder Proposer's offer. In the case of acceptance, the money is divided accord- X ing to the proposal made by Proposer. If Responder rejects the offer, Mortality Salience however, none of the players get anything. After learning about the UG, participants were informed that they would be randomly assigned to the role of either Person A Fig. 3. Moderated mediation effect of mortality salience on amount of money sent by Proposer and Proposer's satisfaction in the Ultimatum game. Standardized regression (i.e., Proposer) or Person B (i.e., Responder). They were told that they coefficients, *p b .05, **p b .01. would be matched with students from the same university, who were seated in another room and would remain anonymous during the study and after it. In reality, as in Study 1, all participants were assigned the amount of money they sent depended on whether their mortality to the Proposer role. Participants received an endowment of 30 PLN and, was salient or not (see Fig. 3). using 1 PLN increments, had to divide the money between themselves We again tested for moderated mediation using the bootstrapping and the Responder. Unlike Study 1, no physical money or envelopes procedure. The 95% confidence interval for the indirect effect of the were present, and all interaction took place on a computer interface. amount of money sent to Responder was between −0.55 and −0.05, After participants made their allocation decision, they were asked to indicating that it was a significant moderated mediator of the relation rate how satisfied they were with their decision and how satisfied, in between MS and Proposer satisfaction (indirect effect size = −.22, their opinion, the second player would be. Both ratings were made on boot se = .12). As noted earlier, mortality thoughts increased the a scale ranging from 1 (not satisfied at all)to10(fully satisfied). amount of money sent to the Responder, β = .35, se = .15, t = 2.34, Finally, participants were informed about the decision made by Re- p = .02. Furthermore, in the MS condition participants reported higher sponder (acceptance or rejection). In reality, the computer made that satisfaction the more money they offered, β = .36, se = .19, t = 1.97, decision, and offers were randomly accepted with a probability of 1/3. p b .06, 95% CI [−0.01, 0.75]. In contrast, sending more money was as- Participants whose offers were accepted collected their payoffs from sociated with decreased satisfaction in the control condition, the game in addition to cafeteria coupons. β = −.62, se =.30,t = −2.04, p b .05, 95% CI [−1.24, −0.003]. To sum up, participants primed with mortality sent more money to the other player and experienced higher satisfaction after sending larger Results sums, whereas participants in the control condition sent less money and experienced higher satisfaction after sending smaller sums. No differences were observed between the MS and control condi- Analyzing participant predictions of the Responder satisfaction, we tions in either positive affect, F(1, 39) = 0.75, p = .39, η2 =.02,orneg- found that the regression model with experimental manipulation, ative affect, F(1, 39) = 0.31, p = .58, η2 = .01. Consistent with our amount of money sent to Receiver and their interaction as predictors predictions, however, the two conditions differed in the amount of was only marginally significant, F(3, 37) = 2.21, p =.1,R2 =.15, money the Proposer sent to the Responder, F(1, 39) = 5.50, p = .02, most probably due to the low achieved power of the test that was η2 = .12, achieved power = 0.63. Participants in the MS condition of- 0.56. The amount of money sent to the Responder was a significant pre- fered more money (M =16.60,SD = 4.02) than participants in the con- dictor, β =.47,se =.18,t =2.57,p = .01, 95% CI [0.10, 0.84]. The effect trol condition (M = 14.19, SD =2.40). of experimental manipulation was not significant, β = −16, se =.16, An analysis of covariance was conducted to control for other poten- t = −1.00, p = .33, 95% CI [−0.50, 0.17]; neither was the interaction tially related variables. No significant effects were found on the amount between manipulation and amount of money sent, β = −.22, se = of money allocated for gender, F(1, 35) = 0.01, p =.92,η2 b .001, age, .18, t = −1.21, p = .23, 95% CI [−0.59, 0.15]. F(1, 35) = 0.33, p =.86,η2 = .001, positive affect, F(1, 35) = 0.94, p =.34,η2 = .03, or negative affect, F(1, 35) = 0.05, p =.82,η2 = Discussion .002. The main effect of the experimental manipulation was still intact after controlling for these four variables, F(1, 35) = 5.36, p =.03, Study 2 completely replicated the results of Study 1 in the context of η2 =.13,achievedpower=0.67. a different economic game. We found that mortality reminders motivat- Echoing results from Study 1, participants in the mortality salience ed participants to allocate more money to an anonymous partner in an and control conditions did not differ in their personal satisfaction with ultimatum game. Furthermore, as in Study 1, mortality thoughts caused their decision, F(1, 39) = 0.11, p = .75, η2 b .01, or in their predicted a switch in participants from associating increased giving with lower satisfaction for the second player, F(1, 60) b 0.001, p =.99,η2 b .01. satisfaction to associating it with higher satisfaction. Also, when partic- The regression model with experimental manipulation, amount of ipants became more prosocial in the mortality salience condition, this money sent to the Responder and their interaction as predictors was appeared to be driven more by an intrinsic desire to give, rather than marginally significant, F(3, 37) = 2.72, p = .058, R2 =.18,achieved by the perceived fear that the Responder would find the amount unsat- power = 0.66. Investigating the predictors for personal satisfaction, isfactory and reject it. we found no significant effects for either experimental manipulation, β = .09, se = .16, t = 0.57, p = .57, 95% CI [−0.23, 0.42], or the amount Study 3. Mortality salience and prosociality revisited of money sent to the Responder, β = −.14, se =0.18,t = −0.77, p = .45, 95% CI [−0.50, 0.23]. However, as hypothesized, there was a signif- Studies 1 and 2 demonstrated that mortality salience increases gen- icant interaction between experimental manipulation and the amount erous, prosocial behavior and also increases the satisfaction derived of money allocated to the Responder, β =.50,se =.18,t =2.77, from such behavior. In Study 3, we again tested the hypothesis that act- p b .01, 95% CI [0.13, 0.86]. The satisfaction participants derived from ing prosocially would be associated with higher personal satisfaction 72 T. Zaleskiewicz et al. / Journal of Experimental Social Psychology 59 (2015) 67–76 after reminders of mortality. This time, we observed participants' be- raising campaign organized by the Student Government. The goal of havior in a setting less structured, and potentially less artificial, than the campaign was explained to be buying additional appliances for the economic games. Specifically, we examined how participants chose to “student zone”—a lounge area where the students work, rest and social- naturally divide the money they were offered between themselves ize together. Next, we manipulated the incentive structure to guide par- and a prosocial cause, after either being reminded of their mortality or ticipant decisions in either a prosocial or selfish direction. In the not. Differently from Studies 1 and 2, in this study we also attempted “prosocial” condition (n = 69), participants were told that they would to actively guide participants' allocation decisions. It is clearly difficult receive 0.5 PLN for every token they decided to keep for themselves. to manipulate people's decisions into a prosocial vs. selfish direction in In contrast, for every token they chose to donate, the Student Govern- a subtle manner, and successful manipulations of this, to our good ment would receive 2 PLN. In the “selfish” condition (n = 67), the pay- knowledge, are rare. Inspired by Aknin et al. (2013, Study 3), in our at- ment structure was reversed. Participants were informed that they tempt to influence participant decisions we manipulated the incentive would receive 2 PLN for every token they kept for themselves, whereas structure. This meant that for half of the participants acting selfishly the Student Government would receive 0.5 PLN for every token they maximized overall profits, whereas for the second half acting pro- chose to donate. socially did. Study 3 was thus designed as a 2 (mortality salience vs. After putting their chosen amount of tokens into the envelopes and control condition) × 2 (incentive structure: prosocial vs. selfish) sealing both of these envelopes, participants indicated how satisfied between-subject experiment. The incentive structure manipulation they felt with their decision by responding to four items: “I am satisfied served an instrumental role in influencing participants' behavior. Our with the decision I made”, “If I had to divide the tokens once again, I main dependent variable was satisfaction with one's decision. As the would do the same”, “I believe that the decision I have made is the satisfaction arises from the actual decision, and not from the incentive most appropriate one”,and“If I had to divide the tokens again, I structure itself, our independent variable was participants' subsequent would do it differently” (reversely coded). All ratings were made on a behavior—that is, the extent to which they engaged in prosocial giving. scale ranging from 1 (totally disagree)to7(totally agree). Because the In addition to satisfaction derived from the decision, we also measured four items were highly correlated, we averaged them to form a general the accessibility of death thoughts as a second dependent variable. If measure of participants' satisfaction (Cronbach's α = .88, M = 6.21, prosocial acts effectively mollify the death anxiety evoked by MS, then SD = 1.10). this should be manifested in the lower accessibility of death-related Participants' final task was a word-stem completion task that mea- cognition following one's prosocial decision. In turn, lower accessibility sured the accessibility of death thoughts. Word-stem completion tasks of death thoughts should predict higher satisfaction with one's decision. are commonly used in terror management research (Hayes et al., We thus hypothesized that prosocial behavior would suppress death 2010). Participants were asked to complete 20 word fragments in Pol- thoughts triggered by mortality salience and thereby make our partici- ish. Ten of these words could have been completed in a death-related pants more satisfied. manner. For example, T R _ _ _ A could be completed as tratwa (raft) or trumna (coffin). Death thought accessibility was operationalized as Method the number of fragments that participants completed in a death- related manner (M = 1.45, SD =1.23). Participants One hundred and thirty-eight students at a Polish university partic- Results ipated in the study in exchange for credit. To improve the statistical power of our analyses relative to the previous studies, we aimed to sig- Preliminary analyses nificantly increase the number of participants. Sample size was decided No differences were found between the MS and control conditions in by how many participants signed up for the time period during which either positive affect, F(1, 134) = 1.63, p = .20, η2 = .01 or negative af- the study was run (predetermined to be 8 days). We excluded two par- fect, F(1, 134) = 1.34, p =.25,η2 = .01. ticipants (one from the mortality salience/prosocial incentive structure Analyzing the number of tokens donated, we confirmed that our at- and one from the dental salience/prosocial incentive structure), because tempts to influence participant decisions were effective: When the in- they admitted after the study that they were not fluent in Polish and had centive structure encouraged selfish decisions, participants donated experienced problems with the crossword puzzle and the death fewer tokens (M =6.13,SD = 3.44) than when it encouraged prosocial thought accessibility measure. The final sample consisted of 136 partic- decisions (M =9.03,SD =2.09),F(1, 132) = 35.01, p b .001, η2 =.21, ipants (118 women) with a mean age of 25.65 (SD =8.41). achieved power = 0.99. Neither the mortality salience manipulation, nor the interaction between the two manipulations affected the number Procedure of tokens kept by the participants, F(1, 132) = 0.56, p = .46, η2 b .01 and After arriving in the lab, participants were seated in individual F(1, 132) = .01, p =.93,η2 b .001, respectively. In other words, it was cubicles and had neither visual nor verbal contact with each other. only the structure of incentives that motivated our participants to be- They were told that the study examined the relationship between per- have more selfishly or more prosocially. Preliminary analyses suggested sonality and decision-making, and asked to complete a pen-and-paper that the interaction between the MS and the incentive structure manip- questionnaire. ulation was not significant for either the satisfaction or the death At the beginning of the experiment, as in the previous studies, par- thought accessibility variables.2 However, our predictions concerned ticipants assigned to the experimental condition (n = 69) completed not what we wanted our participants to do, but what they in fact did. the fear of death questionnaire, whereas participants in the control con- Hence, we carried out an analysis, in which the number of tokens donat- dition (n = 67) completed the dental anxiety questionnaire. Next, they ed was used as the independent variable. Mortality salience manipula- reported their mood on PANAS (Cronbach's α = .74 for positive affect tion, number of tokens donated, and the interaction of these two were and .88 for negative affect), and solved a crossword puzzle meant for distraction and delay. 2 We ran a 2 (MS vs. control) × 2 (prosocial vs. selfish decision context) analysis of var- In the next part of the study, participants were told that they would iance. For the satisfaction variable, there was no effect of MS, F(1, 132) = 0.80, p = .34, be rewarded for their participation in the study and each of them was η2 = .01, no effect of the incentive structure, F(1, 132) = 1.89, p = .17, η2 = .01, and 2 handed an envelope containing 10 tokens to be exchanged for money. no MS × incentive structure interaction, F(1, 132) = 1.60, p =.21,η = .01. For the death thought accessibility measure, there was marginally significant effect of the MS manipula- There were also two empty envelopes on the desk, one labeled “Me”, tion, F(1, 132) = 2.87, p =.09,η2 = .02, no effect of the incentive structure, F(1, “ ” 2 and the other labeled Student zone . Participants were told that they 132) = 0.46, p = .50, η b .01, and no interaction of the two, F(1, 132) = 0.85, p =.36, had the opportunity to donate some of their tokens to the money- η2 =.01. T. Zaleskiewicz et al. / Journal of Experimental Social Psychology 59 (2015) 67–76 73 entered into a multiple regression. The dependent variables were par- satisfaction compared to the control condition. The less they donated, ticipants' satisfaction with their decision and death thought accessibili- the stronger was this conditional effect of the mortality salience manip- ty. All variables were standardized to z-scores prior to the analysis. ulation on their satisfaction.

Satisfaction with decision Death thought accessibility The regression model with mortality salience, number of tokens do- For the second dependent variable, the regression model with mor- nated to the Student Government and their interaction as predictors of tality salience, number of tokens donated to the Student Government satisfaction with decision was significant, F(3, 132) = 9.79, p b .001, and their interaction as predictors was significant, F(3, 132) = 8.59, R2 = .18, achieved power = 0.99. The mortality salience manipulation p b .001, R2 = .16, achieved power = 0.99. We found a marginally sig- did not significantly affect satisfaction with one's decision, β = −.09, nificant effect of the MS manipulation, β =.17,se =0.08,t =1.97, se =.08,t =1.21,p =.23,95%CI[−0.25, 0.06], however, the number p = .05, 95% CI [0, 0.31] and a significant effect of the number of tokens of tokens donated did, β =.36,se =0.08,t = 4.56, p b .001, 95% CI [0.20, donated, β = −.23, se =0.08,t = −2.89, p b .01, 95% CI [−0.39, 0.52]. Higher levels of donation predicted higher satisfaction. As expect- −0.07]. As expected, participants tended to generate more words relat- ed, the interaction between mortality salience and the number of tokens ed to death in the MS condition, and across the board, the more they do- donated was also significant, β =.22,se = 0.08, t = 2.77, p = .01, 95% CI nated, the lower was the accessibility of death-related thoughts. [0.06, 0.38]. This interaction indicates that the impact of the amount of Furthermore, the interaction between mortality salience and donation tokens donated on participants' satisfaction depended on whether level was also significant, β = −.30, se =.08,t = −3.77, p b .001, they had been exposed to MS or not. In particular, in the control condi- 95% CI [−0.46, −0.14]. The impact of mortality salience on the accessi- tion there was no association between the number of tokens donated bility of death-related thoughts thus depended on the amount of tokens and satisfaction with one's decision, β = .14, se = .11, t = 1.23, p = the participants decided to donate. Decomposing the interaction, we .22, 95% [−0.08, 0.36]. In contrast, in the mortality salience condition found no relationship between the number of tokens donated and the more the participants donated, the higher was their reported satis- death thought accessibility in the control condition, β =.07,se =.11, faction, β = .57, se = .11, t = 5.19, p b 0.001, 95% CI [0.36, 0.80] (see t =0.66,p = .51, 95% CI [−0.15, 0.30]. However, when mortality Fig. 4A). The observed positive association between prosocial behavior thoughts were made salient, the amount of tokens the participants and personal satisfaction in the face of mortality thoughts completely donated predicted death thought accessibility, β = −.53, se = .11, replicated our findings from Studies 1 and 2. t = −4.70, p b .001, 95% CI [−0.75, −0.30]. As can be seen in Fig. 5A, Next, we investigated the nature of this interaction using the John- the higher the extent of prosocial behavior in the face of mortality son–Neyman technique (Hayes, 2013; Preacher, Rucker, & Hayes, salience, the lower was the accessibility of death-relevant thoughts. 2007). Our analysis revealed that the effect of mortality salience on per- Using the Johnson–Neyman technique to explore the interaction ceived satisfaction was significant when participants donated less than shown in Fig. 5 revealed that the effect of mortality salience on death 6.6 tokens (Fig. 4B). In other words, if participants donated more than thought accessibility was significant only when participants donated 6.6 tokens on average, the experimental condition did not have an effect less than 7.6 tokens on average, and the less they donated, the stronger on their satisfaction. Yet, if they kept a substantial number of tokens for was the conditional effect of the mortality salience manipulation on themselves after being reminded of death, they reported lower (A) (A)

(B) (B)

Fig. 4. (A) Participants' satisfaction as a function of mortality salience and the number of Fig. 5. (A) Number of death-related words as a function of mortality salience and the tokens donated to the Student Government. (B) Johnson–Neyman regions of significance number of tokens donated to the Student Government. (B) Johnson–Neyman regions of for the conditional effect of mortality salience on participants' satisfaction at raw values of significance for the conditional effect of mortality salience on number of death-related number of tokens donated to the Student Government. words at raw values of number of tokens donated to the Student Government. 74 T. Zaleskiewicz et al. / Journal of Experimental Social Psychology 59 (2015) 67–76 death thought accessibility. It seems that for participants in the mortal- (respectively: c path: β =.57,se =.12,t =4.64,p b .001; c' path: ity salience condition, donating a substantial part of their tokens was β =.40,se = .13, t =3.05,p b .01; indirect effect = .17, boot se = helpful in suppressing the death-related cognition evoked earlier by .11, 95% boot CI [0.01, 0.46]). To sum up, when mortality thoughts the experimental manipulation. In contrast, when participants kept a were made salient, behaving prosocially reduced the availability of higher number of tokens (three or more) for themselves in the MS death-related thoughts. This effect seems to account for why behaving condition, mortality thoughts were more readily available to their prosocially was more satisfying than behaving selfishly in the mortality consciousness. salience condition.

Death thought accessibility as a mediator of the relationship between Discussion mortality salience, prosociality and satisfaction To test the idea that participants' satisfaction with their decision Study 3 conceptually replicated the results of Studies 1 and 2 in a set- might be explained by the fact that donating to the Student Government ting different from economic games. Unlike the previous studies, this reduced their death thought accessibility, we conducted a moderated time we also used a manipulation to guide participant decisions in ei- fi mediation analysis using the bootstrapping procedure. First, we ther a prosocial or sel sh direction. In keeping with the two previous introduced the number of death-related words as a mediator to the re- studies, we observed that making mortality thoughts salient changed gression predicting participants' satisfaction from the mortality salience the way our participants perceived donating money. However, in manipulation, number of tokens donated, and their interaction (see Study 3 we have not observed a negative relation between donating Fig. 6). The regression model explained a substantial part of the variance money and satisfaction in the control condition as in Studies 1 and fi of dependent variable, F(4, 131) = 12.13, p b 0.001; R2 = 0.27, achieved 2. There simply was no signi cant correlation between amount of power = 0.99. Death thought accessibility was a significant predictor of money donated and reported satisfaction in the control condition of participants' satisfaction, β = −.32, se = .08, t = −3.98, p b .001, 95% CI Study 3. The lack of a negative correlation as observed in the previous [−0.49, −0.16]. The effect of the interaction between mortality salience studies might be due to the fact that participants perceived donation ‘ ’ and number of tokens donated on participant satisfaction was no longer for a student zone as sending money indirectly to themselves. They significant, β = −.12, se =.08,t = 1.54, p = .13, 95% CI [−0.03, 0.28], could potentially use in the future the appliances that would be bought suggesting a full mediation. A bootstrapping mediation test with 10,000 with the money they donated, so perhaps they did not feel like the allo- samples confirmed the indirect effect (indirect effect size = .10, boot cation decision involved a zero-sum game as in the previous studies and se = .05, 95% boot CI [0.02, 0.23]). were thus not as bothered by giving away money that could otherwise To decompose this moderated mediation, we performed two sepa- be theirs. Nevertheless, replicating Studies 1 and 2, we found that dona- rate mediation analyses: first in the control condition and second in tion was a source of personal satisfaction in the mortality salience the mortality salience condition. As noted earlier, in the control condi- condition. Moreover, behaving prosocially in the mortality salience con- tion the number of tokens donated was not related to the accessibility dition decreased the accessibility of death thoughts, which predicted of death thoughts, (a path: β = .07, se =.10,t =0.77,p =.45,95%CI increased satisfaction with one's decision. This provided additional sup- [−0.12, 0.27]), while the accessibility of death thoughts predicted port for our argument that prosocial acts are an effective buffer in the lower satisfaction (b path: β = −.32, se =.12,t = −2.78, p b .01, face of death thoughts, and this buffering effect possibly explains the 95% CI [−0.56, −0.09]). Total, direct and indirect effects of donation higher satisfaction people derive from acting prosocially when death on satisfaction were insignificant (respectively: c path: β = .14, se = thoughts are active. .10, t =1.43,p = .16, c'path:β = .16, se =.09,t = 1.76, p = .08; indi- rect effect = −.02, boot se =0.03,95%bootCI[−0.12, 0.03]). In the MS General discussion condition, on the other hand, the more tokens the participants donated, the less death-related words they generated in a subsequent task (a In three behavioral studies, we demonstrated that mortality re- path: β = −.53, se =.12,t = −4.20, p b .001, 95% CI [−0.78, minders increase prosocial acts and the satisfaction associated with act- −0.28]), which in turn enhanced their satisfaction (b path: β = −.32, ing prosocially. This effect occurs presumably because the acts of se = .11, t = −2.82, p b .01, 95% CI [−0.14, −0.67]). All total, direct upholding cherished social values and the sense of meaning, value, con- and indirect effects of donation on satisfaction were significant nectedness, transcendence and security that can be derived from giving

Death thought Mortality 0.17 accessibility salience -0.29*** -0.09 (-0.04) -0.23** Amount of tokens donated Participant 0.36*** (0.28***) Satisfaction -0.30*** Amount of tokens donated X 0.22** (0.12) Mortality salience

Fig. 6. The effect of mortality salience on participants' satisfaction depends on the amount of tokens they donated and is mediated by death thought accessibility. Standardized regression coefficients, †p b .10, *p b .05, **p b .01, ***p b .001. Values in parentheses represent regression coefficient while controlling for number of death-related words. T. Zaleskiewicz et al. / Journal of Experimental Social Psychology 59 (2015) 67–76 75 and sharing soothe existential anxiety. Mortality thoughts thus lead on others (e.g., gift-giving, charitable donations) emerged as associated people not only to behave more generously, but also to perceive giving with higher happiness than spending money on oneself (Dunn,Aknin,& as more pleasant than painful. As we have seen, in the mortality salience Norton, 2014). Interestingly, however, people were unaware of the condition participants reported higher personal satisfaction the more happiness-inducing powers of prosocial spending and when asked, generous their decision was, whereas no such relationship was ob- they consistently predicted that they would be happier if they spent served in the control condition. In fact, in Studies 1 and 2, participants money on themselves. Our findings show that mortality reminders reported higher satisfaction in the control condition, the less money make people bypass this “virtue blind spot” (Sandstrom & Dunn, they gave. The estimated satisfaction of the second player in these stud- 2013) and act in ways that are more conducive to happiness. We can ies, on the other hand, did not differ as a function of being exposed to thus think of mortality salience as providing a valuable perspective mortality thoughts. This suggests that when people choose to behave and prompting people to behave in ways that will make them happier more generously after mortality thoughts, this has more to do with eventually. one's personal satisfaction than with increased concern about the satis- To sum up, our research shed light on the question of how thoughts faction the other player will experience. about one's mortality affect prosocial behavior in economic giving The finding that mortality salience increases prosocial orientation contexts. In a situation where anonymity was fully assured and no con- replicated the Scrooge effect reported earlier by Jonas et al. (2002). ceivable external pressure to behave prosocially existed, just thinking Our research went beyond the existing studies in showing that the en- about their mortality for a brief period made participants not only hanced prosociality brought about by mortality salience is associated more generous, but also more likely to associate generosity with in- with decreased accessibility of death thoughts and in turn, with in- creased personal satisfaction. These findings stand at the intersection creased personal satisfaction. To our knowledge, our work is the first of terror management theory, economic psychology, and prosocial to demonstrate this finer mechanism underlying prosocial decision- behavior literatures, and contribute to each. making in the face of mortality thoughts. Our studies, admittedly, had some limitations. For one, we had low power of tests, especially in Study 2, and to a smaller degree in Study Acknowledgments 1. The conventionally accepted level of statistical power is 0.8, and we reported .78, .73, .84 and .99 for the tests in Study 1, and .63, .67, .66, The research was financed by Polish Ministry of Science and Higher and .56 for those in Study 2. These suboptimal levels of achieved Education grant NN106133237. power were mainly due to small sample size. We tried to address this concern in Study 3 by enlarging our sample size and power was above .99 for all tests conducted as part of Study 3. Notwithstanding the rela- References tively low power observed in Study 2, we believe that the consistent replication of similar results across three studies attests to the reliability Aknin, L.B., Barrington-Leigh, C.P., Dunn, E.W., Helliwell, J.F., Burns, J., Biswas-Diener, R., and robustness of our effects. et al. (2013). Prosocial spending and well-being: Cross-cultural evidence for a psy- chological universal. Journal of Personality and Social Psychology, 104(4), 635–652. Another important limitation of our work revolves around the issue Burke, B.L., Martens, A., & Faucher, E.H. (2010). Two decades of terror management theo- of causation. We have argued throughout our paper that following mor- ry: A meta-analysis of mortality salience research. Personality and Social Psychology tality reminders, prosocial decision-making leads to higher satisfaction. Review, 14,155–195. Camerer, C.F. (2003). Behavioral game theory: Experiments in strategic interaction. However, this argument is supported more by correlational than exper- Princeton, NJ: Princeton Univ. Press. imental evidence. In Studies 1 and 2 we did not manipulate prosocial Cartwright, E. (2011). Behavioral economics. Routledge advanced texts in economics and decision-making; hence, our ability to make causal inferences about finance. London: Routledge. Chaudhuri, A. (2009). Experiments in economics: Playing fair with money. New York: the effect of such decisions per se remains limited. In Study 3, on the Routledge. other hand, we tried to guide participant responses in a more selfish Cialdini, R.B., Kallgren, C.A., & Reno, R.R. (1991). A focus theory of normative conduct: A or prosocial direction with a manipulation, but our analyses relied on theoretical refinement and reevaluation of the role of norms in human behavior. In M.P. Zanna (Ed.), Advances in experimental social psychology, Vol. 24.(pp.201–234). participant decisions and not the manipulation itself. Given the correla- San Diego: Academic Press. tional nature of our results, we would advocate caution when inter- Cohen, J. (1988). Statistical power analysis for the behavioural sciences. New York: Academ- preting these effects. As difficult as they might be to come up with, we ic Press. believe that future work in this area would benefit a lot from successful Dunn, E.W., Aknin, L.B., & Norton, M.I. (2014). Prosocial spending and happiness: Using money to benefit others pays off. Current Directions in Psychological Science, 23(1), manipulations of prosocial vs. selfish decision making. 41–47. Our research contributes to the burgeoning literature on how death Faul, F., Erdfelder, E., Lang, A. -G., & Buchner, A. (2007). G*Power 3: A flexible statistical thoughts affect people's relationship to money. We know that money power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39,175–191. in general serves an existentially soothing function: MS makes people Fehr, E., & Schmidt, K.M. (1999). A Theory of fairness, competition, and cooperation. The value money more, and money primes decrease self-reported death anx- Quarterly Journal of Economics, 114(3), 817–868. iety (Zaleskiewicz, Gasiorowska, Kesebir, Luszczynska, & Pyszczynski, Forsythe, R., Horowitz, J.L., Savin, N.E., & Sefton, M. (1994). Fairness in simple bargaining experiments. Games and Economic Behavior, 6(3), 347–369. 2013). Both spending and saving have also been shown to serve existen- Greenberg, J., Solomon, S., & Pyszczynski, T. (1997). Terror management theory of self- tial functions. Even though mortality reminders make people more inter- esteem and cultural worldviews: Empirical assessments and conceptual refinements. ested in conspicuous consumption goals (Kasser & Sheldon, 2000), In M.P. Zanna (Ed.), Advances in experimental social psychology, Vol. 29. (pp. 61–139). New York: Academic Press. research suggests that saving behavior is more effective in mollifying Güth, W., Schmittberger, R., & Schwarze, B. (1982). An experimental analysis of ultima- death anxiety than spending behavior (Zaleskiewicz, Gasiorowska, & tum bargaining. Journal of Economic Behaviour and Organizations, 3,367–388. Kesebir, 2013). The current finding that mortality reminders make peo- Hayes, A.F. (2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. New York: The Guilford Press. ple less interested in keeping money for themselves and more interested Hayes, J., Schimel, J., Arndt, J., & Faucher, E.H. (2010). A theoretical and empirical review of in benefiting others suggests that socially and culturally based terror the death-thought accessibility concept in terror management research. Psychological management defenses, such as sharing with an ingroup member, might Bulletin, 136,699–739. trump individual defenses (e.g., spending and saving money for oneself). Hirschberger, G., Ein-Dor, T., & Almakias, S. (2008). The self-protective altruist: Terror management and the ambivalent nature of prosocial behavior. Personality and In future work, we want to more directly test the hypothesis that gener- Social Psychology Bulletin, 34,666–678. osity/prosociality is a more effective soother of death anxiety than greed/ Joireman, J., & Duell, B. (2005). Mother Teresa versus Ebenezer Scrooge: Mortality sa- selfishness. lience leads proselfs to endorse self-transcendent values (unless proselfs are reassured). Personality and Social Psychology Bulletin, 31,307–320. Research yields strong empirical support for the idea that prosocial Joireman, J., & Duell, B. (2007). Self-transcendent values moderate the impact of mortality spending increases happiness. In a series of studies, spending money salience on support for charities. Personality and Individual Differences, 43,779–789. 76 T. Zaleskiewicz et al. / Journal of Experimental Social Psychology 59 (2015) 67–76

Jonas, E., Schimel, J., Greenberg, J., & Pyszczynski, T. (2002). The Scrooge effect: Evidence Sandstrom, G.M., & Dunn, E.W. (2013). Is efficiency overrated? Minimal social interac- that mortality salience increases prosocial attitudes and behavior. Personality and tions lead to belonging and positive affect. Social Psychological and Personality Social Psychology Bulletin, 28,1342–1353. Science, 5(4), 436–441. Jonas, E., Sullivan, D., & Greenberg, J. (2013). Generosity, greed, norms, and death: Differ- Solomon, S., Greenberg, J., & Pyszczynski, T. (1991). A terror management theory of social ential effects of mortality salience on charitable behavior. Journal of Economic behavior: The psychological functions of self-esteem and cultural worldviews. In M.P. Psychology, 35,47–57. Zanna (Ed.), Advances in experimental social psychology, Vol. 24. (pp. 91–159). San Kasser, T., & Sheldon, K.M. (2000). Of wealth and death: Materialism, mortality salience, Diego: Academic Press. and consumption behavior. Psychological Science, 11,348–351. Van den Bos, K., & Miedema, J. (2000). Toward understanding why fairness matters: The Kesebir, P., & Pyszczynski, T. (2012). The role of death in life: Existential aspects of human influence of mortality salience on reactions to procedural fairness. Journal of motivation. In R.M. Ryan (Ed.), The Oxford handbook of human motivation (pp. 43–64). Personality and Social Psychology, 79,355–366. New York, NY: Oxford University Press. Watson, D., Clark, L.A., & Tellegen, A. (1988). Development and validation of brief Mullainathan, S., & Thaler, R. (2001). Behavioral economics. In N.J. Smelser, & P.B. Baltes measures of positive and negative affect: The PANAS scales. Journal of Personality (Eds.), International encyclopedia of social and behavioral sciences (pp. 1094–1100). and Social Psychology, 54, 1063–1070. Oxford: Pergamon Press. Zaleskiewicz, T., Gasiorowska, A., & Kesebir, P. (2013). Saving can save from death Preacher, K.J., Rucker, D.D., & Hayes, A.F. (2007). Addressing moderated mediation hy- anxiety: Mortality salience and financial decision-making. PLoS ONE, 8(11), e79407. potheses: Theory, methods, and prescriptions. Multivariate Behavioral Research, 42, http://dx.doi.org/10.1371/journal.pone.0079407. 185–227. Zaleskiewicz, T., Gasiorowska, A., Kesebir, P., Luszczynska, A., & Pyszczynski, T. (2013). Pyszczynski, T., Greenberg, J., & Solomon, S. (1999). A dual-process model of defense Money and the fear of death: The symbolic power of money as an existential anxiety against conscious and unconscious death-related thoughts: An extension of terror buffer. Journal of Economic Psychology, 36,55–67. management theory. Psychological Review, 106,835–845. Pyszczynski, T., Greenberg, J., Solomon, S., & Koole, S.L. (2010). Experimental existential psychology: Coping with the facts of life. In S.T. Fiske, D.T. Gilbert, & G. Lindzey (Eds.), Handbook of social psychology (pp. 724–757). Hoboken, NJ: Wiley.