Competitive Helping in Online Giving

Transcription

1 Report Competitive Helping in Online Giving Graphical Abstract Authors Nichola J. Raihani, Sarah Smith Corresponence In Brief Raihani an Smith show competitive helping in onations mae to online funraising pages. Males respon competitively to onations mae by other males, but only when giving to an attractive female funraiser. Female onors o not compete in this way. These finings suggest a role for sexual selection in explaining conspicuous generosity. Highlights We present evience of competitive helping in a real-worl setting We measure onor responses to large onations on online funraising pages Males compete with other males when giving to attractive female funraisers This suggests a role for sexual selection in explaining unconitional generosity Raihani & Smith, 2015, Current Biology 25, May 4, 2015 ª2015 Elsevier Lt All rights reserve

2 Current Biology Report Competitive Helping in Online Giving Nichola J. Raihani 1, * an Sarah Smith 2 1 Department of Genetics, Evolution an Environment, University College Lonon, Lonon WC1E 6BT, UK 2 Department of Economics, University of Bristol, Bristol BS8 1TN, UK *Corresponence: nicholaraihani@gmail.com SUMMARY Unconitional generosity in humans is a puzzle. One possibility is that iniviuals benefit from being seen as generous if there is competition for access to partners an if generosity is a costly an therefore reliable signal of partner quality [1 3]. The competitive helping hypothesis preicts that people will compete to be the most generous, particularly in the presence of attractive potential partners [1]. However, this key preiction has not been irectly teste. Using ata from online funraising pages, we emonstrate competitive helping in the real worl. Donations to funraising pages are public an mae sequentially. Donors can therefore respon to the behavior of previous onors, creating a potential generosity tournament. Our test of the competitive helping hypothesis focuses on the response to large, visible onations. We show that male onors show significantly stronger responses (by onating more) when they are onating to an attractive female funraiser an responing to a large onation mae by another male onor. The responses for this conition are aroun four times greater than when males give to less-attractive female (or male) funraisers or when they respon to a large onation mae by a female onor. Unlike males, females o not compete in onations when giving to attractive male funraisers. These ata suggest that males use competitive helping isplays in the presence of attractive females an suggest a role for sexual selection in explaining unconitional generosity. RESULTS AND DISCUSSION Despite iniviual incentives to free-rie, humans often cooperate in social ilemmas. In repeate, two-player games, iniviuals can benefit if the partner reciprocates [4], whereas in larger groups, the possibility that cheating will be punishe [5] or that generous actions will be reware [6] are possible explanations for cooperation. Most of the evience for punishment an rewaring, however, comes from laboratory stuies, while the prevalence of these strategies in real-worl settings has been questione [5]. A plausible alternative yet selom acknowlege mechanism is so-calle competitive helping [2] (also known as competitive altruism [1]). This theory assumes that there is a biological market [2], where iniviuals compete for access to partners with the highest market value by signaling their value through costly helping isplays [1]. Signal reliability is maintaine by the cost associate with sening it [3]. Extravagant helping isplays might therefore serve as an honest signal of an iniviual s unerlying quality [7], incluing access to resources or cooperative intent. Although the term competitive altruism is commonly use to escribe extravagant generosity (e.g., [1, 8 13]), competitive helping isplays are not truly altruistic since the signaler is expecte to erive personal benefits from their actions. Several laboratory stuies offer evience suggestive of competitive helping. People are more generous when they are observe [8 10], an the most helpful iniviuals are preferentially chosen for subsequent interactions requiring cooperation [11 13]. Costly helping isplays coul be use to attract sexual partners. Although males ten to prioritize physical cues of fertility when choosing mates [14], females place a higher premium on resource acquisition [14] an have also been shown to be more sensitive to cooperative tenency in sexual partners [15] both of which may be avertise via helpful actions. Thus, competitive helping might be particularly pronounce among males (although this oes not preclue female-female competition in this or other contexts). Inee, males are more sensitive than females to the presence of an opposite sex auience when performing helping behavior [9, 16]. While these various stuies inicate that people are motivate to acquire a goo reputation an that iniviuals might be preferre as partners on this basis, the aci test of competitive helping requires evience that people compete irectly, by increasing generosity in response to isplays from competitors [1, 13]. To our knowlege, no stuy has provie evience of responsive competitive helping. We o this here. We use ata from a large, UK-base, online funraising platform to test a key preiction of the competitive helping hypothesis: males respon competitively to the generosity bis of other males in the presence of attractive females. Online funraising platforms provie a unique forum to test this iea in a real-worl setting. Funraisers host funraising pages where they provie personal information (name, photo, charity, an event they are being sponsore for) an collect onations, with nearly all onations coming from onors personally known to them. Donations to funraising pages are mae sequentially, an the names an contributions of all onors to a funraiser s page are visible in chronological orer (unless onors opt for anonymity) so that new onors can see who has onate before them an how much [17]. This creates a potential tournament in which onors may compete by responing to how much others have given. Current Biology 25, , May 4, 2015 ª2015 Elsevier Lt All rights reserve 1183

3 Change in amount given ( ) Male onors Competitive helping scenario All other scenarios Female onors Figure 1. Male an Female Donor Responses to Large Donations Mae by Other Donors Change in amount given (, effect size an confience intervals from a single LMM) among male onors (n = 1,800) an female onors (n = 1,295) in response to a large onation relative to the PRE-mean. Competitive helping scenario refers to the case where competitive helping woul be expecte. For males, this case is an attractive female funraiser an a large onation by another male onor. For females, this case is an attractive male funraiser an a large onation by another female onor. All other scenarios refers to all other cases. Previous work has shown that existing onations on a page act as an anchor for current onors, inicating that onors o pay attention to other onors actions in this setting [17]. Here, we ask whether onors behavior was affecte by the gener an attractiveness of the funraiser an whether they compete with other onors of the same gener. For our analysis sample of 2,561 funraising pages, the mean (±SEM) number of onations per page was 42.8 (±0.71), averaging 30.3 (±0.22) per onor. The average total amount raise per page was 1, (±28.4) (see Table S1 for all escriptive statistics). Regarless of their gener, more-attractive funraisers raise more money than less-attractive funraisers: a 1-SD increase in attractiveness was associate with an average (±54.0) increase in total amount raise by funraisers, controlling for their age (linear regression: F 1, 2544 = 11.4; p = 0.001; Table S2). Attractive funraisers of both sexes receive more onations (linear regression: F 1, 2544 = 8.41; p = 0.004; Table S2) an, to a lesser extent, also receive larger onations (linear regression: F 1, 2544 = 2.78; p = 0.10; Table S2). For female funraisers, attractiveness was significantly correlate with maximum onation size (linear regression: F 1, 2543 = 9.08; p = 0.003; Table S2), which increase by 41.1 (±13.6) with a 1-SD increase in attractiveness. All onation ata are available online (Data S1). Our test for competitive helping focuse on onors responses to large, visible onations on a funraising page (large was efine as at least twice the previous page mean an more than 50, sensu [17]). Responses to large onations were measure by calculating the ifference in amount given after the large onation ( ) relative to the mean onation size for that page prior to the large onation (hereafter referre to as the PRE-mean ). The PRE-mean was calculate using up to ten (where available, sensu [17]) onations mae prior to the large onation. Note that we only use onations mae by onors of the same gener, so the male response is calculate relative to the PRE-mean efine for male onors. For each funraising page, we consiere the responses of up to 15 onors (where available) following the large onation, yieling a sample size of 1,800 for male onor responses an 1,295 for female onor responses. Funraising page ID was inclue as a ranom term in each moel to control for the effects of repeate observations for the same funraiser an funraising page on the istribution of the ata. Arriving on a funraising page after large onations has been shown to have a positive effect on the size of onations, with no effect on the quantity of onations [17]. In our sample, 1,829 pages ha large onations (mean large onation size: ± 2.69). Of these, 420 were mae by males an 248 by females, with the remainer not gener assignable. We focus our analysis on the 668 large onations for which we coul assign a gener. Summary statistics are given in Table S1. The size of a large onation i not significantly iffer by either onor or funraiser gener (tests for equality of means, all p > 0.05; see Supplemental Information). Results from a linear mixe moel (LMM) with maximum likelihoo shows that, in line with previous results [17], a large onation ha a positive effect on subsequent amounts, increasing onations by 9.55 (±1.27). Uner the competitive helping hypothesis, we expecte a significantly stronger response by onors when (1) the funraiser was an attractive member of the opposite sex (where attractive was efine as being within the top quartile), an (2) the large onation was mae by someone of the same sex. We contraste the responses by onors in this competitive helping conition to their responses in all other cases. Results are summarize in Figure 1. For males, we foun that a large onation was associate with an aitional response of (±7.75) in the competitive helping case, over an above the average response in all other cases, which was 9.61 (±1.61). This aitional response by male onors in the competitive helping conition was significantly ifferent to that in all other cases (c 2 1, 1800 = 13.38; p < 0.001). However, when we ran a similar moel for female onors, asking whether females woul show a greater response when giving to an attractive male funraiser an when the large onation was mae by another female, we foun that female onors i not show greater responses in the competitive helping case (c 2 1, 1295 = 0.54; p = 0.461). We then explore variation in male onor responses in more etail. Using the same response term (change in onation amount,, relative to the PRE-mean), we ran a LMM with funraising page ID as the ranom term an the three-way interaction between three categorical explanatory variables: funraiser gener (male/female), funraiser attractiveness (plain/average/ attractive), an large onor gener (male/female). Thus, the moel ha a esign (Table S3) allowing us to check whether male responses woul be strongest in the scenario preicte by competitive helping theory (i.e., when giving to an 1184 Current Biology 25, , May 4, 2015 ª2015 Elsevier Lt All rights reserve

4 Change in amount given ( ) A Plain Average Attractive males o not conform to the majority when making onations but instea compete irectly with other males when onating to attractive females. Excessively cooperative iniviuals can be shunne or punishe [19] an sometimes opt for anonymity when making large onations [20]. Our finings provie a possible explanation for this, inicating that overt generosity can be a competitive rather than a cooperative act. EXPERIMENTAL PROCEDURES Change in amount given ( ) B Female funraisers Male funraisers Figure 2. Male Donor Responses to Large Donations Mae by Other Donors Accoring to Large Donor Gener, Funraiser Gener, an Funraiser Attractiveness (A an B) Change in amount given (, effect size an confience intervals from a single LMM) among male onors (n = 1,800) in response to a large onation relative to the PRE-mean. Responses to a large onation varie with the gener an attractiveness of the funraiser an varie epening on whether the large onation was mae by a male (A) or female (B) onor. Male onors increase their giving by more when giving to an attractive female funraiser an when responing to a large onation mae by a male competitor than in any other case. attractive female funraiser an when the large onation was mae by another male) compare to any of the other 11 possible scenarios. The results are shown in Figure 2 an confirm the expecte pattern: male onors respone to a large onation to the greatest extent when giving to a female funraiser who was attractive an when the large onation came from another male onor. Their response in this case was significantly greater (see Table S4) an aroun four times larger than their response in any of the other 11 cases. These results support a key preiction of competitive helping theory [1, 2, 13] by showing that male onors compete irectly with other males in the presence of an attractive, opposite-sex auience, although we fin no evience for this in females. Whether competitive helping isplays prouce fitness benefits remains an open avenue for further exploration, although previous work has shown that more-cooperative iniviuals are preferre as sexual partners [15]. We also note that competitive helping responses are not necessarily conscious responses either to the onations of others or to the perceive attractiveness of the funraiser but may instea reflect responses of an evolve psychology to maximize the benefits associate with helping in ifferent contexts. Previous work has shown that onors are sensitive to the onations of others an, in particular, conform to escriptive social norms by giving what they believe is the normative amount [18]. In contrast, here we show that Our initial sample consiste of 4,581 pages for the 2014 Virgin Lonon Marathon for which funraisers ha uploae one profile photo, allowing us to obtain an attractiveness rating. Of this initial sample, 91 ha a URL that i not link to a page, leaving 4,490 pages. For these pages, we sought four inepenent beauty ratings (on a scale of 0 to 10) of each funraiser base on their photo (following [21]) by recruiting 1,189 raters (651 males; 520 females; 16 i not specify) from the Amazon Mechanical Turk (MTurk) crowsourcing platform ( The mean age of the raters was 31 ± 0.3 (range: 18 72). To minimize any confouning effects of intercultural ifferences in attractiveness ratings [22], we aske only US-base workers to provie ratings since the majority of workers on MTurk are base in the US, an the US an UK are categorize as belonging to the same worl culture [23]. Each rater was expecte to look at 20 pages. As well as rating attractiveness, raters were also aske the gener of the funraiser, what sort of clothes they were wearing (e.g., sporting/fancy ress), whether they were smiling, the color of their hair, an approximate age. The questions the raters were expecte to answer are available in the Supplemental Information. Data on factors affecting funraiser attractiveness are available online (Data S2). Donor gener was not observe irectly but was inferre on the basis of their names, acquire from the funraising pages. We were able to assign a gener to 46.1% of the onors in our sample. Cases that we coul not inclue ha ambiguous names (e.g., Sam, Chris), multiple names (e.g., Sue an Davi), an anonymous onations. We juge 2,561 funraisers to have a vali set of attractiveness ratings where (1) they ha at least three non-missing ratings, an (2) all raters agree on the funraiser s gener. The main reasons for missing ratings were that the photo was not of a person or that it was of more than one person, such that the funraiser coul not be ientifie. We foun a high level of agreement among raters over the attractiveness of funraisers: for our sample of 2,561 funraisers, Cronbach s alpha was 0.88 (0.63 for male funraisers an 0.94 for female funraisers). For our analysis, we create stanarize aggregate measures of attractiveness (a i ) for each funraiser in the following way (following [24]). Each rater rate up to 20 pictures, an each funraiser ha three or four ratings. We first transforme the rating by rater j of funraiser i into a Z score an then took the average of the three or four Z scores for each funraiser. In line with previous stuies (e.g., [14]), gener, hair color, age, an whether the person was smiling were all significantly correlate with attractiveness (see Supplemental Information). For further statistical etails an tables, please refer to the Supplemental Information. SUPPLEMENTAL INFORMATION Supplemental Information inclues Supplemental Experimental Proceures, four tables, an two ata files an can be foun with this article online at AUTHOR CONTRIBUTIONS Both authors contribute equally to this stuy. ACKNOWLEDGMENTS We thank P. Barclay, R. Bshary, A. Garner, D. Hamermesh, I. Penton-Voak, an anonymous reviewers for useful comments an I. Cuthill an S. English for statistical avice. This work was fune by a Royal Society University Research Fellowship to N.J.R. an an ESRC grant to S.S. N.J.R. has one Current Biology 25, , May 4, 2015 ª2015 Elsevier Lt All rights reserve 1185

5 consulting work for a UK-base online funraising company. All ata on onations use in this paper were freely available in the public omain. Receive: November 26, 2014 Revise: February 9, 2015 Accepte: February 17, 2015 Publishe: April 16, 2015 REFERENCES 1. Roberts, G. (1998). Competitive altruism: from reciprocity to the hanicap principle. Proc. Biol. Sci. 265, Barclay, P. (2011). Competitive helping increases with the size of biological markets an invaes efection. J. Theor. Biol. 281, Grafen, A. (1990). Biological signals as hanicaps. J. Theor. Biol. 144, Trivers, R.L. (1971). The evolution of reciprocal altruism. Q. Rev. Biol. 46, Raihani, N.J., Thornton, A., an Bshary, R. (2012). Punishment an cooperation in nature. Trens Ecol. Evol. 27, Milinski, M., Semmann, D., an Krambeck, H.J. (2002). Reputation helps solve the tragey of the commons. Nature 415, Zahavi, A., an Zahavi, A. (1997). The Hanicap Principle: A Missing Piece of Darwin s Puzzle. (New York: Oxfor University Press). 8. Hary, C.L., an Van Vugt, M. (2006). Nice guys finish first: the competitive altruism hypothesis. Pers. Soc. Psychol. Bull. 32, Böhm, R., an Regner, T. (2013). Charitable giving among females an males: an empirical test of the competitive altruism hypothesis. J. Bioecon. 15, Van Vugt, M., an Hary, C.L. (2009). Cooperation for reputation: wasteful contributions as costly signals in public goos. Group Process. Intergroup Relat. 13, Sylwester, K., an Roberts, G. (2013). Reputation-base partner choice is an effective alternative to inirect reciprocity in solving social ilemmas. Evol. Hum. Behav. 34, Barclay, P. (2004). Trustworthiness an competitive altruism can also solve the tragey of the commons. Evol. Hum. Behav. 25, Barclay, P., an Willer, R. (2007). Partner choice creates competitive altruism in humans. Proc. Biol. Sci. 274, Buss, D. (1989). Sex ifferences in human mate preferences: evolutionary hypotheses teste in 37 cultures. Behav. Brain Sci. 12, Phillips, T., Barnar, C., Ferguson, E., an Reaer, T. (2008). Do humans prefer altruistic mates? Testing a link between sexual selection an altruism towars non-relatives. Br. J. Psychol. 99, Van Vugt, M., an Ireale, W. (2013). Men behaving nicely: public goos as peacock tails. Br. J. Psychol. 104, Smith, S., Winmeijer, F., an Wright, E. (2014). Peer effects in charitable giving: evience from the (running) fiel. Econ. J. Publishe online June 12, Croson, R., an Shang, J. (2013). Limits of the effect of social information on the voluntary provision of public goos: evience from fiel experiments. Econ. Inq. 51, Herrmann, B., Thöni, C., an Gächter, S. (2008). Antisocial punishment across societies. Science 319, Raihani, N.J. (2014). Hien altruism in a real-worl setting. Biol. Lett. 10, Hamermesh, D.S. (2006). Changing looks an changing iscrimination : the beauty of economists. Econ. Lett. 93, Penton-Voak, I.S., Jacobson, A., an Trivers, R. (2004). Populational ifferences in attractiveness jugements of male an female faces: comparing British an Jamaican samples. Evol. Hum. Behav. 25, Inglehart, R., an Baker, W.E. (2000). Moernization, cultural change, an the persistence of traitional values. Am. Sociol. Rev. 65, Hamermesh, D.S., an Bile, J. (1994). Beauty an the labour market. Am. Econ. Rev. 84, Current Biology 25, , May 4, 2015 ª2015 Elsevier Lt All rights reserve