The effect of social norms estimation in non-strategic giving: Discarding the role of anchoring and relative incentives

CARLOS MAXIMILIANO SENCI; BRENDA ELIZABETH RYAN; NATALIA GREGORIETTI; FREIDIN ESTEBAN

Wageningen

Conferencia; Behavioural Insights in Research and Policy Making; 2016

Wageningen University

Background: Social norms seemingly play an ubiquitous role in economic decisions. Here we focus on the fact that participants that estimate others? opinions about the socially appropriate behavior in the Dictator Game (i.e., the prescriptive norm), then make more generous offers (Krupka & Weber, 2009). Our study attempts to explore whether this results survives after controlling for the effect of factors unrelated to social norms. To start with, we dealt with the problem that the incentivized estimation of norms changes the earning prospects in the session. Our concern was that participants could become more generous, not because they recalled the prescriptive norm, but because they could gain more money in the session. To control for this, the standard social norms treatment was compared against a treatment in which participants could gain extra money by accurately guessing others? opinion of a matter unrelated to social norms (Einstein?s death age). Participants in Einstein treatment were similarly generous than those in the control treatment and less so than participants in the social norms treatment. These results did not support the hypothesis that the effect of social norms on DG decisions depended on increased earnings prospects. Second, we were concerned with the possibility of an anchoring effect given its pervasiveness across contexts, including normatively charged settings such as legal decision making (Englich et al., 2006). More specifically, we tested whether the estimation of the prescriptive norm could play its role by anchoring participants? responses to the number associated with the fair share in the estimation (see also Raihani & McAuliffe, 2014). With this goal in mind, we ran a treatment in which the anchoring prediction was pitted against the social norms prediction. We implemented a treatment in which participants? guess prior to the DG still involved thinking about others? normative opinions, but where the numeric anchor, if effective, would drive participants away from the fair share. Results did not support the anchoring hypothesis either.Methods: We recruited 115 university students from a variety of disciplines (e.g., economics, biology, education, and engineering, among others) in Bahía Blanca, Argentina. We ran paper-and-pencil sessions during the second semester of 2015 at the Universidad Nacional del Sur. In experiment 1a, participants in the prescriptive norms treatment (n=34) had to estimate previous participants? opinions about the socially appropriate behavior in the DG before actually playing the game for real money. We compared this condition against a control treatment in which participants made their decisions in the DG without any prior estimation (n=20). In experiment 1b, we ran a treatment (n=25) in which, prior to playing the DG, participants could earn extra money if they correctly guessed previous participants? estimation of Einstein?s death age. The goal of the Einstein treatment was to control whether participants were more generous simply because they had the expectation of earning more money in the session. Last, in experiment 1c, we ran a treatment (n=36) in which participants had to guess the social norm in a DG with a pool of AR$100 before making their decisions in a DG with a pool of AR$50. The goal of this condition was to assess whether the numbers involved in the estimation (AR$50 was the fair split in the AR$100 game) anchored the subsequent DG decision away from the fair share (which was AR$25 in the AR$50 DG).Results: In experiment 1a, participants were more generous in the treatment in which they had to estimate previous participants? normative opinions than in the control treatment (Mann-Whitney U test, Z=2.36, P= 0.018). Furthermore, participants who had the chance of earning extra money by guessing others? estimation of Einstein death age offered less money in the DG than participants that could earn the extra money by guessing others? normative opinions (Mann-Whitney U test, Z=2.55, P= 0.01). In fact, participants? DG offers in the Einstein treatment did not differ from those in the control treatment (P>0.10). Last, in experiment 1c, participants estimating the prescriptive norm in the AR$100-pool DG were similarly generous than those estimating the norm in the AR$50-pool DG (Mann-Whitney U test, Z=0.21, P= 0.84), and more so than in the control treatment without estimation (Mann-Whitney U test, Z=2.55, P= 0.01).219Discussion: In the present study, we could successfully replicate the effect of prescriptive social norms on DG decisions (see Krupka & Weber, 2009; Bicchieri & Xiao, 2009). Alike US samples, Argentine participants made more generous decisions after thinking of the prescriptive social norm than in the control treatment without any estimation prior to the DG (experiment 1a). Not only that, but we showed that the effect of the prescriptive norm on DG decisions did not depend on the extra incentives present in the incentivized estimation exercise. When participants could gain extra money by guessing others? estimation of Einstein?s death age prior to playing the DG, they were no more generous than the control treatment without estimation (experiment 1b). Last but not least, the mean offer after estimating the social norm in a DG with an AR$100-pool was no different than in the original treatment in which the estimation was done in an AR$50-pool DG. This result implies that even when participants thought of AR$50 as the fair share in the AR$100 game, their decisions in the AR$50 game showed no sign of being anchored near the number 50, but instead were closer to the AR$25 offer which was the equitable share in the AR$50-pool game. In summary, we found evidence that the effect of estimating prescriptive norms on socio-economic decisions is robust, and depends on neither the extra earning prospects of the estimation exercise nor a numeric anchor after the estimation.