The paper “Cooperation and Punishment in Public Goods Experiments” by Fehr and Gächter from 1999 was a milestone for the change of RCT from its orthodox versions to the adoption of elements from non-economic fields. Main result of the experiment was that in a public good situation subjects not only started with a contribution rate of about 50 %, but that with announcing the opportunity to punish free riders alone already, contributions made a strong jump upwards and converged to nearly 100 % in the following rounds – although punishment was expensive for punishers and should not occur for selfish rational actors.

The contribution investigates the scope of possible explanations for the observed patterns in the F&G-experiments by Rational Choice Theory (RCT) extended by motives of reciprocity, with the model of frame selection (MFS). MFS refers to processes of cognitive activation of specific types of “definition of the situation” (Esser and Kronberg 2015). Main result is that most findings can be reconstructed rather easily by means of both approaches – with one exception: After starting with punishment and after withdrawal of this option after 10 rounds subjects should following RCT react immediately with at least some defection, following MFS, however, with keeping a high level of cooperation, independently of motives of subjects.

An independent empirical test with data also from other experiments (Hermann et al. 2008) showed exactly this: no change in cooperation, not even by egoists. Alternative RCT-explanations aiming to find cooperative equilibria for keeping cooperation unchanged by egoists could be the assumption of reputation-effects in finite iterated games. This interpretation, however, seems to be not plausible: F&G tried to control explicitly for reputation effects for all versions, and at least for the stranger-version this attempts should have been successful. The effect, however, appeared in both versions, and for the stranger-version of the data set by Hermann et al. even stronger than in the original experiment.

Fehr, E., and S. Gächter. 1999. Cooperation in Public Goods Experiments. Working Paper No. 10. Institute for Empirical Research in Economics. University of Zürich.

Esser, H., and C. Kroneberg. 2015: Towards an Integrative Theory of Action: The Model of Frame Selection. S. 63–85 in: E. L., Sh. Thye and J. Yoon (Eds.), Order on the Edge of Chaos: Social Psychology and the Problem of Social Order. Cambridge, Mass.: CUP.

Hermann, B., Ch. Thöni and S. Gächter. 2008. Antisocial Punishment Across Societies. Science 319: 1362–1367.


In this paper, we test how constrained choices affect the provision of public goods. Specifically, we implement two public good games with different levels of discretization in contribution choices and find that limiting options leads to higher payoffs. Using a novel experimental design, we are able to distinguish two separate channels through which payoffs improve. As might be expected, higher payoffs can be explained, in part, by more cooperation and higher rates of giving in the treatment with constrained choices. However, we also find that subjects provide public goods in a more cost-effective manner when facing fewer options, suggesting that constrained choice can lead to better decisions, even conditional on the same level of giving. Together our results indicate that the simple availability of choices, or lack thereof, can influence behavior and outcomes, with important implications for a wide range of settings like charitable giving and environmental conservation efforts.

I develop a theoretical framework for studying choice bracketing based on general preference axioms. A decision maker is characterized by a system of brackets with coarser systems of brackets corresponding to a less narrow view of the decision environment. For each bracket the decision maker evaluates the dimensions of her choice inside the bracket jointly, while keeping all dimensions outside the bracket fixed at the reference level. As a consequence, the narrow decision maker underestimates the complementarities and substitutabilities between dimensions that are in distinct brackets and behaves as if she would apply a budgeting heuristic.

Network structure has often proven to be important in understanding the decision behavior of individuals or agents for different interdependent situations. Computational studies predict that network structure has a crucial influence on behavior in iterated 2 by 2 asymmetric `battle of the sexes' games. We test such behavioral predictions in an experiment with 240 human subjects. We found that as expected the less `random' the network structure, the better the experimental results are predictable by the of the computational models. In particular, there is an effect of network clustering on the heterogeneity of convergence behavior in the network. We also found that degree centrality and having an even degree are important predictors for the decision behavior of the subjects in the experiment. We thus find empirical validation of previous computational models using human subjects.

Both economists and psychologists are interested in understanding decision making under uncertainty. Yet, they rely on different concepts to analyze human behaviour: Economists use economic preference parameters rooted in utility theory, while psychologists use personality traits to describe responses to uncertain situations. Using a sample of university students, this study examines and contrasts 5 economic preference parameters and 6 psychological personality traits that are commonly used in the literature for studying individuals' attitudes towards uncertainty. A novelty of this paper is to consider both the economic concept of ambiguity aversion as well as the personality trait of ambiguity intolerance. We analyze the relation between these measures, along with their influence on a variety of real life outcomes. Our main result shows that the psychological personality trait of ambiguity intolerance is considerably more related to selected important life outcomes than standard economic preference parameters.

We aim to understand how two sources of social influences modulate participants’ risk preferences. We used the Balloon Analogue Risk Task (BART), which is one of the few risk tasks that has reliably predicted participants’ risk behavior in a wide variety of decision domains, to decompose social modulation of risk preferences into contexts shaped by beliefs about the likely behaviors of others, and from direct interaction. Participants in the BART pump a balloon to accumulate tokens and aim to cash in these tokens before the balloon pops. In social contexts, two participants, each with their own balloon, play simultaneously and besides not popping their own balloons, should cash in higher than their opponent to secure their cashed tokens. In the belief only context, the identity of the opponent was known, but actions during play were hidden. In the interaction only context, opponent identity was hidden, but balloon pump actions were visible in real-time. To measure participants’ beliefs about balloons’ pop point and their opponents’ behavior, we introduced a novel belief elicitation task. In groups of 5-12 we collected data from a total of 159 participants stemming from 29 experimental sessions. In each experimental session, one participant underwent fMRI while playing the BART alone and in social contexts, in real-time, against other session participants who were seated outside the scanner. Participants’ behavior was substantially influenced by social contexts. In the belief only context participants increase their willingness to take on risk (compared to playing alone) by pumping the balloon an average an additional 6 times. In the interaction only context, participants decrease the average number of pumps (2.5 less). However, the average interaction-only behavior does not tell the complete story. Participants show a preference for pumping to a similar level as their opponent, pumping less when the opponent ends at a lower number and more when the opponent pumps higher. FMRI analyses provide support for separate neural mechanisms underlying social belief and direct interaction. Although both contexts activate theory of mind regions, the belief only context shows a more pronounced activity pattern in frontal medial regions, whereas the interaction only context activates lateral components of brain areas associated with social processing. In sum, participants’ risk preferences are influenced by social contexts, demonstrating social modulation of risk preferences that can both increase and decrease tolerance for risk as compared to behavior alone. These behavioral differences are supported by separable neural mechanisms which together provide clues for how social factors influence risk preferences in social situations. Furthermore, we replicate findings that behavior in the BART correlates with substance use, indicating the relevance of our task on understanding behavior outside the laboratory.

The representativeness heuristic (RH) has been proposed to be at the root of several types of biases in judgment. In this project, we study if RH is helpful in organizing data describing two kinds of choices in the context of gambling. Specifically, in a field experiment with naturalistic stimuli and a potentially extremely high monetary pay-out, we give each of our subjects a choice between a lottery ticket with a random-looking number sequence and a ticket with a patterned sequence; we subsequently offer them a small cash bonus if they switch to the other ticket. In the second task, we investigate the gambler’s fallacy, asking subjects what they believe the outcome of a fourth coin toss after a sequence of three identical outcomes will be. We find that most subjects prefer “random” sequences, and that approximately half believe in dependence between subsequent coin tosses. There is no correlation, though, between the initial choice of the lottery ticket and the prediction of the coin toss. Nonetheless, subjects who have a strong preference for certain number combinations (i.e., subjects who are willing to forgo the cash bonus and remain with their initial choice) also tend to predict a specific outcome (in particular a reversal, corresponding to the gambler’s fallacy) in the coin task.

Uncertainty is an inherent feature of decision-making in finance. In real world market settings, dividend processes as well as fundamental values are guided by vagueness; neither objective probabilities of realization returns are known for certain, nor is the actual amount of returns. Whereas classical risk and vague probabilities (“classical” ambiguity) have been widely analyzed in different contexts, vague outcome realizations have gotten surprisingly little attention so far. We add to the existing literature in two ways. First, we are able to relate individual attitudes towards risk and vagueness to actual trading behavior in markets and market outcomes. Second, we are able to identify the effects of different types of vagueness on market outcomes, in particular by including the novel influence of vagueness in outcome realizations. We conduct a large-scale laboratory experiment with 320 subjects. On average, we neither find a significant risk premium, nor a significant premium for vagueness in probabilities, outcomes, as well as the combination of both types of vagueness in neither the individual task nor the market experiment. Testing whether a theoretical prediction based on general equilibrium theory holds in our market experiment, we find that the difference is not significantly different to zero, which leads to markets actually behaving close to and indistinguishable from general equilibrium predictions.

Theory and empirical research have established that repeated interactions foster cooperation in social dilemmas. Therefore, in social dilemmas, actors have incentives for strategic tie formation in the sense of establishing long-term relations involving repeated interactions. We introduce and analyze a simple game-theoretic model that captures the effects of repeated interactions and simultaneously endogenizes the formation of long-term relations. We assume strict game-theoretic rationality as well as self-regarding preferences. We highlight the commitment-feature of tie formation: through establishing a long-term relation, at cost, actors ensure that they would suffer themselves from future sanctions of own opportunism. This allows for mutually beneficial cooperation in the first place.

While the paper does not yet include experimental work, it offers testable implications as well as further suggestions for experimental work and macro-implications.

This paper aims to test the effect of fairness considerations on the decision to follow the law using experimental economics methodology. In a repeated public good game, we implement an incentivized obligation (i.e., minimum contribution level) that is enforced probabilistically. We implement the asymmetry of the enforcement by assigning heterogeneous monitoring probabilities to otherwise identical subjects in a public goods game. To understand how unfair implementation of the law affects individuals’ behavior, we compare average contributions in treatments with low, high, and asymmetric enforcement. Initial analysis of the results indicate no significant difference under symmetric or asymmetric enforcement of the law. The subjects reacted only to their own obligations and how those obligations were enforced upon others had no effect on their behavior.

When there is an opportunity to gain a positive reputation, individuals are more willing to sacrifice their immediate self-interest, and to behave cooperatively. Evolutionary models of cooperation pose that reputation-based partner choice can be an alternative to indirect reciprocity (Nowak and Sigmund, 2005), or a useful complement to it (Roberts, 2015). According to Sommerfeld et al. (2007), gossip is an effective alternative to direct observation in games of indirect reciprocity, and exchanging information can effectively support cooperation in a Public Goods Game (Wu et al, 2016). However, these studies overlook the fact that individuals can have strategic motives to lie, and therefore gossip can be completely unreliable. Tooby, Cosmides, & Barkow (1995) suggest that selection would have favored our disseminating information in the interests, not of objective truth but of our own success in social competition. Gossipers would have incentives to deceive receivers in ways that benefit the signaler (Hess, Hagen 2006), thus derogating rivals and masking their faults. Also, noise and unintentional errors are unavoidable features of information transmission, thus raising further doubts about the efficacy of gossip in sustaining cooperation over time and across different groups. In this experimental study, we use a combination of Public Good Games and gossip rounds in order to test to what extent gossip can remain truthful and sustain cooperation when participants can lie. 160 individuals played a combination of repeated PGG in groups of 4 individuals, and a one-shot final round in groups of 8. For the final game 2 randomly selected leaders formed the groups, and in the partner selection treatment, only the group with the highest score was rewarded. In both conditions, participants could send messages by filling a form in which they had to indicate the target, the receiver, and the source of the gossip, choosing between themselves as identifiable sources and an unknown other. Our preliminary analyses show that partner choice made individuals more cooperative, but this did not reduce the manipulative potential of gossip. Both source and content manipulation were used, but deception rates were similar between the two conditions.

Corruption marks a major societal challenge. Although many corrupt practices such as bribery are outlawed in national codes of laws, stark differences in (perceived) corruption levels persist around the world. To explain this gap between legal norms and actual behavior, the current corruption literature emphasizes the importance of social norms, the unwritten rules that guide behavior. Recent lab research suggests that short social norms messages can reduce people’s perceptions of descriptive norms about bribery and lower their own inclination to bribe. In pursuit of first field evidence for this link, we conducted a pre-registered lab-in-the-field study in South Africa. Throughout town we distributed posters that contained a descriptive norms messages about bribery. Inside a mobile, we used an incentivized assessment of social norms (descriptive and injunctive) and a behavioral bribery task. Outside the lab, we assessed the missing stock in a local pharmacy as a real-life measure of corruption. Our results reveal that only during the period in which the poster was put up, participants’ perceived descriptive norms of bribery, and their own willingness to engage in bribery decreased. We discuss the findings in light of their relevance for the interdisciplinary literature dealing with social norms and (anti-) corruption.

Previous literature shows that individuals care about the ability of their competitor and the social recognition of their achievements when choosing effort in competitions. We are the first who study how these factors influence behavior in competitions in which cheating is possible. Our laboratory experiment consists of two rounds of competition. First, we study if the matching of competitors based on the first round outcome (High/Low) has an effect on behavior. When two High compete against each other, men significantly increase cheating while women significantly increase effort. Due to the opposing behavior of men and women, the overall increase of effort and cheating is insignificant. There is no significant difference in behavior if two Low compete against each other. Second, we study how social status affects behavior. Private as well as public feedback increases cheating but has no effect on effort. We find that the effect of competing against peers varies with the rewards. In addition, we investigate the behavior of first round winners and losers. While winners reduce cheating in the second round, losers increase cheating.

Lying costs and social preferences are perhaps the two most well-documented deviations from selfish maximization behavior in the economic literature. We hypothesize that in situations where both motives - a motive not to lie and a motive to be fair - are present, but in conflict, individuals may self-servingly place more weight on the motive that helps them to increase their earnings while maintaining a positive self-image. We test this hypothesis using a laboratory experiment that allows us to document whether such motivated motive selection is present. We find evidence that subjects behave as if they care more about equality when caring about fairness implies higher earnings, while they behave as if they care more about truth-telling when telling the truth implies higher earnings.

We introduce a modified version of the die-in-the-cup paradigm to study dis(honest) behavior under time pressure. Replacing the regular die with one that has a distinct color on its either side enables us to manipulate the amount of familiarity with the randomization device. This both removes the limitations of the original paradigm and allows for a test of theories that suggest that dis(honest) behavior is affected by the relative difficulty of generating false reports.

We also replace the cup with a simple mechanical device for better control over the very process of rolling the die, and collect mouse movement data from the participants to investigate the present behavioral archetypes.

Our main finding is that time pressure leads to more dishonest behavior but only if the regular die is used. We also find that when given the time to deliberate, the participants generally report lower values if the regular rather than color die is used.

Wealthy individuals have a large influence on the income distribution in society through politics and the corporate world. How do they think about redistribution? We find that wealthy individuals from the top 5% in the U.S. prefer less redistribution than a representative sample of the bottom 95%. The affluent want to further decrease taxes for the rich and were more likely to support Donald Trump in the last presidential election. This difference in tax attitudes and political decisions can be largely attributed to differences in distributional preferences, which we measure with an incentivized experiment. The wealthy (N = 467) and the general population (N = 415) could redistribute real earnings between real workers, who received unequal compensation for their work. The top 5% accepted more inequality than the general population by redistributing less of the earnings between workers. Individuals who climbed the income ladder, such as successful entrepreneurs and investors appear to drive the gap in distributional preferences. By contrast, individuals who are born rich have distributional preferences much closer to the preferences of the general population. Our findings raise the possibility that wealthy individuals contribute to the persistent income inequality in the U.S.

We use a ‘multi-player dictator game’ (MDG) to examine how individual redistribution behaviour is influenced by the observable actions of peers, and whether average contributions as well as the selected behavioural strategy is subject to an ‘anchoring effect’. We find that in the aggregate, individuals positively condition their redistribution choices on the contributions of first-movers in their group, suggesting conformity to peers. However, we observe that the average contributions of second-movers are affected by the first observable peer contribution (the anchor) that is presented to them using a sequential strategy. Additionally, the anchor is found to influence the behavioural strategy that individuals engage in; specifically, low anchors increase the likelihood of selecting self-interested strategies, whilst high anchors increase the likelihood of giving strategies. The distribution of ‘types’ is therefore dependent on the initial conditions of play – specifically, the initial amount that players observe - in the strategy game.

We investigate how the dependence of payoffs affects preferences for redistribution. For a fixed outcome, we experimentally implement and compare a zero-sum world and a setting in which everyone can be simultaneously successful. First, two subjects’ performances in a real effort task translate into chances of gaining a prize. Across treatments we then vary the interdependence of payoffs: either, there is only a single prize, or both subjects can potentially gain a prize at the same time. Afterwards, a third subject can redistribute the prize money. Removing the direct dependency of subjects’ payoffs decreases the average amount of redistribution by 14-22%. If the outcome of the allocation process solely hinges on relative performance and not partially on chance, the role of payoff dependence remains unchanged. However, we find that the mere presence of randomness increases redistribution – even though there is no uncertainty about the (relative) performance of the two subjects.

One of the questions that guides this study is if, when a basic income is introduced, people are going to invest less time in paid and unpaid work. Further, we discuss peoples’ preferences concerning different redistribution schemes in the context of the German welfare state. Based on Fröhlich and Oppenheimer (1990), we designed an experiment to find out if these preferences vary when peoples’ positions in society are unknown. One scheme is a stylized representation of the actual German welfare state, i.e. with a means-tested minimal income and relatively progressive income tax system. The next scheme represents a situation where there is no redistribution and functions as a control. The last depicts a scenario where there is an unconditional basic income financed by a flat income tax. The experiment was designed aiming the discussion of two issues. The first concerns the hypothetical effects of the introduction of a basic income on people’s paid and unpaid work supply decisions. And the second is about the participant’s preferences concerning redistribution schemes and the influences of a constructed veil of ignorance on these preferences. This veil of ignorance enables the simulation of a situation under which the participants are not aware of their social positions, which are represented in the experiment by hourly wages and distributional schemes. Among the objectives of this experiment is the contribution to the further development of lab experiments on basic income and to the discussion on possible reforms for the German welfare state.

Using data from a household survey and an experiment in rural Tanzania, we study a potential determinant of women's bargaining power that has not previously been studied: intra-couple preference differences. We find substantial differences in time and risk preferences between spouses. These differences in preferences are strongly and negatively associated with women's involvement in household decision-making. Our findings suggest that preference conflicts between spouses may contribute to marginalizing women in households in gender-unequal societies.

We investigate the causal impact of unfair chances and gender discrimination on labor supply decisions. We conduct a large-scale experiment in which workers individually engage in the same task for a fixed piece-rate wage. We employ two payment schemes with equal wages, and three payment schemes with unequal wages generated through different procedures: (1) fair chances, (2) unfair chances based on an unspecified source, and (3) unfair chances based on gender discrimination. We find that, at a given wage, negative gender discrimination reduces the labor supply of workers substantially compared to equal wages (-22%). This effect is twice as large as the decrease induced by unequal wages from non-discriminatory fair and unfair chances (-11%). Moreover, the unfairness of non-discriminatory chances does not influence labor supply. Advantaged workers do not react to the unequal wages generated by our different procedures. Overall, our results support a novel supply-side consequence of discrimination in labor markets.

We study the relationship between gender, willingness to compete and career choices in a setting that includes individuals from the whole ability distribution and career options covering the whole range of skill requirements. We collect incentivized experimental choices from more than 1500 Swiss lower-secondary school students and relate them to their choice of post-compulsory education. This enables us to investigate two novel questions: 1. How does the gender gap in willingness to compete vary with ability? 2. Can willingness to compete predict choices between career options that cover the whole range of skill requirements, including choices between different types of vocational education and the choice between vocational and academic education? Our main results are: 1. The gender gap in willingness to compete is small among the lowest-ability students, but increases steadily with ability and is largest for the highest-ability students. 2. Willingness to compete predicts choices both of academic specializations and of vocational careers. Finally, we combine these two results to analyze their implications for our understanding of gender differences in career choices at different skill requirement levels.

This paper examines experimentally whether the menstrual cycle of women influences their bargaining behavior and bargaining outcomes. We propose that the menstrual cycle - and possibly the hormones progesterone and oestrogen - influences bargaining behavior. Particularly, we hypothesize that Women bargain more aggressively during the ovulation phase, compared to the other phases, and especially, compared to the premenstrual phase. In our study women are asked to track their menstrual cycle for three months before the experiment. Thereafter, they come to the laboratory experiment to participate in an unstructured bargaining game followed by risk and social preference elicitation tasks. Our results strongly confirm a more aggressive bargaining behavior during the ovulation phase compared to the premenstrual phase. Aggressiveness is measured by the magnitude of the offers and demands made by the players in a bargaining context. Furthermore, we also observed that variations in bargaining behavior over the menstrual cycle don't seem to be explained by variations in risk and social preferences.

In this study, I analyze the results of a principal-agent experiment, where (i) agents can cause positive as well as negative externalities on behalf of their principals, and where (ii) effort is a partial substitute for negative externalities. Externalities are modeled as an increase or decrease in a charitable payment. The results suggest a fundamentally asymmetric response with agents being reluctant to donate on behalf of the principal, yet on average willing to increase the payoff of the principal at the expense of the charity. Agents with higher social value orientation provide more effort and take less from the charity, such that principals earn higher payoffs. However, agents who fully renounce damages produce significantly lower payoffs for principals than agents who are willing to cause maximum damages, irrespective of whether or not the principals’ profit targets required damages. Finally, agents who refuse to take from the charity generate the largest economic surplus. This points to a fundamental incentive problem and contradicts the neoclassical claim that the economic surplus is maximized when agents maximize shareholder wealth.

This paper experimentally investigates how using tracking and tracing technologies impacts the work performance, division of tasks and stress levels of individuals in team work. In one treatment, contributions to team performance are tracked. In an additional treatment individual productivity is estimated and subjects are paid based on the individual productivity. The results do not indicate that the tracking per se affects the performance of teams. Payment contingent on individual performance indicators, however, significantly affects the distribution of work contributions within the team but not the group output level. Neither applying tracking technology nor using it to determine individual pay increase subjects' stress levels in this setting. Our findings imply that the effects of tracking and of individual pay on stress levels and on team output are modest.

Pretend-But-Perform mechanism (also called a Game of Pretension) is a revelation mechanism that allows players to declare their types freely but mandates players to act according to their declared types. Using a between subject design in a controlled laboratory environment, we are hoping to measure the effect of this mechanism on a Cournot duopoly. Theoretical predictions dictate that regulating a Cournot duoply by Pretend-but-Perform mechanism will result in an increase in social welfare. In this specific setting, subjects will make announcements about their production costs and in return the mechanism will force them to produce the amount predicted by Nash equilibrium given their cost declarations.

One of the core threats of climate change are increasing flood events in severity and frequency (Hirabayashi et al. 2013; IPCC, 2012). The main private adaptation measure to cope with possible damages from such events is insurance. However, private insurance protection is low (Kunreuther 1984; Schwarze et al. 2011). One factor underinsurance can be attributed to is the individuals’ expectation of governmental relief or charitable aid in case of a damage. This reliance on others’ support and not sufficiently undertake own measures is called ‘charity hazard’ (see e.g. Andor et al. 2017; Browne & Hoyt 2000).

We analyse how participants’ insurance decision is influenced by different compensation schemes and damage experience experimentally. Specifically, we test the ‘charity hazard’ of the two compensation schemes: (i) certain partial compensation which is in place in Austria and (ii) uncertain total compensation which reflects the German strategy. Furthermore, the experience of a damage might influence the subsequent insurance decision depending on the number of events and the discounting of the experienced effects. We conducted an incentivised experiment and have three main contributions to the literature: Firstly, we compare compensation schemes mirroring actual policies in Germany and Austria. Secondly, we provide to the best of our knowledge the first experimental analysis of charity hazard in insurance markets with monetary incentives. Thirdly, we run the experiment with student subjects as well as subjects facing real flood risks.

In our experiment subjects were randomly assigned to either one of the compensation schemes (i) certain partial compensation or (ii) uncertain total compensation. Each participant need to make a personal insurance decision (yes/no) in ten consecutive rounds. We run the experiment in the mLab at the University of Mannheim in October 2018 with 127 subjects. Additionally, we run the same experiment online with 47 participants from the city Dornbirn (Austria) living in flood prone districts in December 2018.

First results show that individuals respond to the different compensation schemes depending on their risk attitude. Controlling for previous experience of a damage within the experiment reveals that only slight differences in participants insurance decisions. Furthermore, on average the student subject pool and the population of Dornbirn are not significantly different reacting to neither the compensation scheme nor the experience of events in their insurance decision.

I study the effect of a debate and argumentation program on students' ability to reason, and to successfully identify fake news. In a field setting, Czech high schools, I find no effect on either reasoning ability or news literacy. Using a post-intervention survey and interviews with teachers, I identify three possible reasons why teaching "correct" argumentation may backfire: 1) student overconfidence in their baseline abilities, 2) lack of motivation of students to engage with difficult material, and 3) student inability to apply the material in practice.

With these three issue in mind, I design a follow-up lab experiment to replicate my field findings. I provide a feedback mechanism with monetary incentives to counter both overconfidence and lack of motivation, and restructure the learning materials to include more practical examples. I deepen the analysis by exploring the (lack of) argumentative skills of students with unusually extreme opinions on topics covered in the articles assigned for analysis, as field experiment results suggest these students could systematically differ from their peers both in their abilities and in their motivation to apply their skills.

[Lab experiment data pending; will be available for the conference.]

We study the influence of argument source on the evaluation of argument validity. We ask, whether partial, impartial and reluctant sources generate different evaluations of argument validity. Previous research has revealed that impartial and reluctant sources promote persuasion. We, however, look at evaluations of argument validity, rather than attitude change, because the assessment of argument quality is also an essential part of democratic discussion. We explore the source effect via a survey experiment where participants are asked to evaluate politically relevant arguments. To avoid the influence of participants’ personal opinions on the issues, participants are asked to take a general perspective when they evaluate arguments. Respondents are randomly allocated into four treatments: a Control treatment with no argument source; a Partial Source treatment, a Reluctant Source treatment and an Impartial Source treatment. Participants (n = 1600) are recruited from a pool of volunteers, and they form a representative sample of the Finnish population in terms of age, sex and residential area. The survey was conducted online in June 2018. Data is analysed mainly through OLS-regressions. Based on our results we can reject the null hypothesis of no impact of the argument source. Further, the impartial source gives rise to higher evaluations of argument quality compared to a reluctant or a partial source. We tentatively suggest that this influence goes through trustworthiness. We conclude that the argument source has an influence on the evaluations of argument validity, and that an impartial source generates highest evaluations of argument validity.

Using a public good experiment with pre-play face-to-face communication (FFC) this paper investigates two channels over which FFC influences contributions by the subjects. Firstly, the contents of the FFC are investigated by categorizing specific strategic information and using simple meta-data. Secondly, a machine-learning approach to analyze facial expressions of the subjects during their communications is implemented. These approaches constitute the first of their kind, analyzing content and facial expressions in FFC aiming to predict the behavior of the subjects in a public good game. Although both approaches are conducted independently the results are consistent: verbally agreeing to fully contribute to the public good until the very end and communicating through facial clues reduces the commonly observed end-game behavior. The length of the FFC quantified in number of words is further a good measure to predict cooperation behavior towards the end of the game. The accuracy of predictions based on the best performing models lies around 7% above the trivial classifier. The obtained findings provide first insights how a priori available information can be utilized to predict free-riding behavior in public good games.

In a one-shot two-person lab experiment we study how behavior is affected by whether player 1 reads the message that player 2 has sent him/her. The issue of one’s communication being considered is relevant in organizations, bargaining and in public decision-making. Our work is motivated by our general interest in how free-form communication affects economic behavior. We have data from three treatments: Optional Communication (N=220), Fixed Communication (N=110) and No Communication (N=110), with the two latter treatments acting as controls. The experiment has two stages. In stage 1 of Optional Communication, player 1 has to choose between two payoff distributions affecting both players 1 and 2. Before player 1 makes the decision, player 2 can send a free-form message to player 1. Before player 1 makes the decision he can choose to read the message or not. In stage 2, player 2 can reward or punish player 1. We study behavior in two different tasks, which differ in the two payoff distributions that are possible. In task 1 (task 2) player 1 has to choose between the following payoff distributions: A ($24, $6) and B ($18, $12) (A ($24, $6) B ($4, $26)). In task 1 player’s 1 decision can be easily influenced by player 2’s message, whereas in task 2 it seems less likely that player 1 can be influenced. The experiments were run at the MONLEE lab at Monash using z-tree. The results show that player 2’s behavior is consistent with reacting differently to kind and unkind treatment and not by whether he/she is listened to or not. Player 1’s behavior is consistent with curiosity or respect and not with the exploitation of moral wiggle room or self-image concerns. We relate our results to several models of social preferences.

We study how the powerful perceive power abuse, and how negative experience related to it influences the appropriateness judgments of the powerless. We create an environment conducive to unfair exploitation in a repeated Public Goods game where one player (punisher) is given a further ability to costlessly subtract money from others (victims). We find that punishers who choose to abuse their power rationalize their behavior by believing that free-riding, while forcing others to contribute to the public good, is not inappropriate. Victims of such abuse also start to believe that punishers' free-riding and punishment are justifiable. Both punishers and victims are not aware that their beliefs are adjusting in this way. In addition, subjects assigned to the role of power, regardless of how they use it, think that outside observers share their beliefs about the appropriateness of their actions. All these observations are explained by the Belief in a Just World hypothesis, which states that people rationalize any wrongful acts in order to maintain a coherent picture of the world that is orderly and lawful. Our findings demonstrate the fearsome capacity of humans to exculpate abusive behavior by themselves and others.

Climatic changes, population growth, and economic scarcity alone and together create shared problems that can be tackled effectively through cooperation and coordination. Perhaps because cooperation and coordination is fragile and easily breaks down, human societies also provide individual solutions for shared problems, like private means of transportation and protection (e.g. gun ownership), or privatized healthcare and retirement planning. What remains unknown is whether and how the availability of individual solutions, and the concomitantly reduced co-dependence on others, affects free-riding on others’ cooperation on the one hand, and the efficient creation of public goods on the other. This we examined using a new experimental paradigm in which groups of individuals faced a shared problem and could use personal resources to solve it either individually or collectively. Across different cost-benefit ratios of solving the shared problem individually vs. collectively, individuals display a remarkable tendency towards group-independent, individual solutions. Such ‘individualism’ leads to inefficient resource allocation and coordination failure. Only when cooperation can save more than 70% of the resources, groups start to coordinate on collective action. Peer punishment helps groups to coordinate on the more efficient collective solution but also leads to wasteful punishment feuds between ‘individualists’ and ‘collectivists’. Our results indicate that societal and economic innovations that reduce co-dependence and reciprocity concerns can create a “modern tragedy of the commons.” In the presence of individual solutions to shared problems, such as those triggered by climate change and population growth, societies not only need to find solutions to the free-rider problem but also need to find a balance between self-reliance and collective efficiency.

Human societies function by virtue of cooperation. Peer punishment seems an important regulatory factor for the functioning of human societies, as various studies showed that the opportunity to punish peers can deter free-riding and sustain high levels of cooperation. However, here we show experimentally that peer punishment fails to deter free-riding in diversified groups with pluriform populations. This insight is important because – although a common assumption in the peer punishment literature is that free-riding takes place in demarcated groups with rather uniform populations – diversity is an inherent feature of human societies. Humans have various social ties and belong to a wide range of collectives within their local community and society. In such diversified groups with pluriform populations, peer punishment opportunities allow and may invite individuals to apply double standards of cooperation and be psychologically reactant to punishments by members with whom they are less affiliated. We argue and demonstrate that this undermines the effectiveness of peer punishment in promoting cooperation.

In our experiment, a total of 144 students participated in an iterated public goods game with real monetary stakes. In groups of four members, participants played 20 rounds of the public goods game with and 20 rounds without peer punishment opportunities. There were two treatment conditions: A uniform condition and a pluriform condition. In the uniform condition, the group consisted of four players all sharing the same real social affiliation. In the pluriform condition, the group consisted of two players sharing a real social affiliation and two players sharing another real social affiliation. In both conditions, participants were not informed about the identity of the others in their group, only about each other’s social affiliation.

Our results show that more antisocial punishment takes place in diversified groups than in demarcated groups, as well as that pluriform populations invite individuals to apply double cooperation standards and be psychological reactant to ‘outgroup’ punishment. More importantly, we show that, although peer punishment opportunities are effective in deterring free-riding in demarcated groups, the opportunity to punish peers does not deter free-riding in diversified groups and is ineffective in stabilizing cooperation. Our results question the notion that peer punishment is an important regulatory factor for the functioning of human societies, as peer punishment opportunities only seem effective under very specific and rather artificial conditions.

To encourage long-term cooperation in social dilemmas such as common pool resources, the importance of sanctioning is often stressed. Economist Elinor Ostrom advocates graduated sanctioning: the severity of a defector’s punishment is related to the extent of criminal acts in the past. Graduated sanctioning is especially suggested because a society might contain ‘vengeful’ persons: people that do not adhere to the rules after receiving a punishment they feel to be disproportional compared to their misconduct. This study compares the effect of graduated and strict sanctioning on cooperation in commons on the micro- and macro-level. Theoretically, hypotheses are derived using game theoretic predictions of behaviour based on extensions of the standard model with vengeful actors. In addition, we distinguish whether the type of sanctions is determined exogenously or voted for by the actors in the game. A Common Pool Resource game is used in a laboratory experiment, integrating crucial elements of social structure and rule-making mechanisms within a common. The results confirm the hypothesis that more vengeful subjects behave more cooperatively under graduated sanctioning than under strict sanctioning. However, in a population without the vengeful type, strict sanctioning yields the highest level of cooperation. When looking at macro-level outcomes of cooperation, having a possibility to vote for a sanctioning mechanism influences the effect of graduated sanctioning positively, and strict sanctioning negatively. For micro-level outcomes, this has no effect.

This research examined whether the better-than-average (BTA) effect is a possible explanation for norm misperceptions. Data from an online study demonstrates that people perceive their risk and proenvironmental behavior as more socially approved than the same behavior in others, suggesting that norm misperceptions are driven by the BTA effect. This suggestion is further substantiated by the observation that when a proenvironmental lifestyle is important to a person, that person’s perception that their proenvironmental behavior is more socially approved of than the same behavior in others becomes more pronounced. This moderator effect of importance is typically reported in the BTA literature. Norm misperceptions were also more pronounced when behaviors were framed in a socially disapproving manner when compared to those framed in a socially approving manner. This framing effect underpins that the BTA effect drives norm misperceptions. We expect that these findings advance the conceptualization of interventions based on social norms and contribute to BTA literature.

In a large field experiment I study how repeated public praise impacts the short and long run performance of 900 teachers in 39 schools. For both recipients and non-recipients of recognition, I analyze the effect of repeated public praise on performance as measured by student grades, attendance, and performance on anonymously graded high-stake exams. In a random half of the schools, the best teachers (according to teacher value added) are praised in an online message posted on the school messaging board. When teachers are praised (not praised) in the first round, their students perform significantly better (worse) in subsequent months. Using the fact that praise is repeatedly given, I test different mechanisms that could explain teacher behavior. Results are in line with teachers having status concerns, and learning about their relative performance through praise. Recognition has large and persistent effects, explained by real increases in effort, as opposed to teachers “cheating” on student assessments.

Scholars from diverse disciplines try to identify the supporting mechanisms that explain the large degree of cooperation in human societies. Recent evidence suggests that social networks play an important role in the evolution of cooperation, however the role of social influence on other-regarding preferences has often been overlooked. Unfortunately, neither meaningful social contacts nor preferences can be exogenously manipulated, which makes causal inference less attainable than through controlled and randomized experiments. Nevertheless, deep social contacts such as friendships, dynamic preferences and social influence are important features of societies and play an important role for human large-scale cooperation.

Our approach allows to study the mechanisms of cooperation in real-world environments, which have before only been studied using computer simulations or artificial laboratory experiments. We thus combine incentivized, fine-grained and repeated measurements of other-regarding preferences with the dynamic friendship networks of 57 school classes (N=1258). We apply matching methods using substantial information about individuals, their friends, peers and teachers to construct suitable comparison groups that allow to study the causal effect social environments have on individual preferences.

Our results suggest that social environments substantially influence individual preferences and hereby contribute to the development of homogenous clusters. Furthermore, we find that cooperative individuals systematically try to avoid social contacts with uncooperative ones, whereas uncooperative individuals seek for new relations towards cooperators. We conclude that social influence and conformity play an import role for the development of other-regarding preferences and paired with a weak partner-selection process contribute substantially to the evolution of cooperation by creating homogenous clusters.

We examine (i) the impact of terrorist attacks on the level of hate speech against refugees in online discussions, and (ii) how the effect of terrorist attacks depends on the ambiguity of social norms of prejudice expression. To this end, we report on the results of a unique combination of a natural and a lab-in-the-field experiment. We exploit the occurrence of two consecutive Islamist terrorist attacks in Germany, the W{\"u}rzburg and Ansbach attacks, in July 2016. Hateful comments towards refugees, but not in other topics, increased as a result of the attacks. The experiment compares the effect of the terrorist attacks in contexts where a descriptive norm against the use of hate speech is emphasized to contexts in which the norm is ambiguous. We find that participants were especially reliant on normative cues in their context after the terrorist attacks. We elaborate a mechanism that explains the increase in online hate as contingent on terrorist attacks creating uncertainty about the social acceptability of the public expression of hate. The implications of the findings for the literature on social norms are discussed.

The Rentier State Hypothesis states that taxation promotes government accountability. The argument is that citizens demand more accountability for spending of tax revenue than for spending of windfall revenue (e.g., natural resource revenue and aid). This paper presents causal evidence from a novel between-subject experiment that tests the effect of taxation on demand for accountability. To investigate the underlying mechanisms of the effect, the design focuses on two main features that distinguish tax from windfall revenue: Tax revenue is produced by citizens' work and has been in their possession before being collected as tax. Consistent with the Rentier State Hypothesis, the main finding is that taxation causes a higher demand for accountability when both features of taxation are present. The paper thus sheds light on the political economy of government revenues, and more generally contributes to our understanding of how features of the tax system shape behavior.

In a large natural field experiment (N>80,000), we evaluate new strategies on how to increase compliance of potential TV-fee evaders by sending them letters and randomly varying the text and envelope. We use two new text strategies aimed at (i) the elicitation of preference for fee designation, and (ii) the explanation of fee purpose. We also employ three well–known strategies that have so far given conflicting results in the literature: (iii) highlighting the formal consequences of evasion, (iv) stating the value of services obtained in exchange for the fee, and (v) invoking social norms. We also test two modifications of the envelope and aim at recipients' reciprocity and attention stimulation by (vi) placing a picture of a fairy-tale cartoon character on the envelope and an identical sticker inside, or by (vii) placing there a red inscription "Important" instead. Our results show that the only treatment more efficient than the baseline is using the deterrence principle spelling out the formal consequences of not paying. Both envelope-modifying treatments marginally decrease the response and registration rates. In a preceding incentivized laboratory pre-test, students had predicted the ranking of the text treatments accurately.

Can immorality judgments be predicted by the three core components of the Theory of Dyadic Morality (Schein and Gray, 2015, 2018) negative affect, harm perception and norm violation? And furthermore, are these immorality judgments and their predictive elements sensitive to framing effects and political orientation? We implemented a novel experimental design to empirically test the theoretical predictions of the Theory of Dyadic Morality and further investigated sensitivity towards framing effects as well as the impact of political orientation. Using an online experiment, we investigated how voluntary payments for an online news website are judged regarding their perceived immorality, harm, anger and social norms. To test for framing effects we varied four wordings of a voluntary payment mechanism in a between-subjects four-factorial design: Pay-What-You-Want, Pay-What-You-Can, Pay-What-You-Believe-Is-Fair and Pay-What-It-Is-Worth-To-You. We contribute to the previous literature by providing empirical evidence for Schein and Gray's Theory of Dyadic Morality in the applied setting of voluntary payments and thus link the disciplines of moral psychology and behavioral economics. 614 U.S. Americans were recruited via Amazon Mechanical Turk and participated in the online experiment. Our results indicate that harm predicts immorality judgments, providing empirical evidence for the theoretical assumptions of the Theory of Dyadic Morality. We further find that the impact of framing effects on own payment behavior, harm, anger, social norm perceptions as well as on immorality judgments in a voluntary payment context is rather low.

Autonomous driving fuels an important debate on the morality of weighing negative externalities in the context of machine learning. Beyond the programming of autonomous cars, negative externalities in the analogue world are also present in other domains of algorithm-based decision making (e.g., HR recruiting). Algorithm programming is rule based. It has to be decided a priori (i.e., before an ethical dilemma occurs) which outcome variable an algorithm should maximize. This demands for an explicit commitment to an ethically preferred position. However, for some ethical dilemmas, it will be very difficult to reach consensus in societal discussions. In these cases, outcome randomization could be a way forward in programming machine-learning systems. We analyze experimentally, if decision randomization by algorithms is generally accepted as a means to address ethical dilemmas. Moral dilemmas are created by inducing decisions with real negative consequences for others as well as for the decision-makers themselves. We find that many decision makers are willing to delegate their ethical dilemmas to an algorithm for outcome randomization. This is especially true, if they are not at risk to incur the negative externality themselves. Our work contributes to the debate on algorithm aversion and its potential causes. It emphasizes that in the need to program algorithms a priori the option of randomization should be included in the societal and economic discourse.

An established research result is that people's environmental attitudes only loosely translate into actions effectively reducing their environmental impact, something known as the attitude-behaviour gap. On the other hand, correct information and environmental education are often considered a key to promote sustainability, which raises the question of when attitudes can actually work as a lever to promote environmental objectives and, conversely, when other factors have a better chance to succeed. To answer these questions, we tested the effect of environmental attitudes in an online experiment with real money at stake and real-world climate mitigation consequences. We found that environmental attitudes mainly affected behaviour in a low cost situation, while their effect was reduced when the stakes were higher. This finding is consistent with the low cost hypothesis of environmental behaviour and has important consequences for the shaping of more effective climate policies in a democratic context.

How individual decisions influence global outcomes is a fundamental question in the social sciences. Societies across the globe face important challenges. Among these are climate change. For instance, the extent of people’s civic engagement determines the efficacy of neighbourhoods, communities, and political entities in developing and applying environmental policies.

How can we solve these challenges? Social norms, which can be defined as informal behavioural rules that are supported by empirical and normative expectations and potentially backed by enforcement, seem to be an important and promising way of helping solving these issues. Social norms and social interactions influence the spread and resilience of those policies. In fact, when we try to understand the uptake of a climate change policy we can no longer focus on how people would behave individually. Rather, we have to consider how single agents interact within the society. Civic engagement as expression of cooperative behaviour is one of the most important measures for the stability of the society. In light of this interpretation the main goal of our work is to study the socio-economic issues associated with the Climate Change Challenge that is, to study how to develop and sustain environmental-friendly social norms in society. To do so we try to investigate, by the mean of a long run and online experiment, two main goals: 1) the role of social learning and evolution of social norms in feeding dysfunctional macro-behaviours; 2) the specific role of individuals’ expectations in learning the social norms governing each group and to understand, once a norm exists, under which conditions behavioural spill-overs actually matter in the endorsement of a cooperative action.

Among nowadays collective action problems, the climate change, i.e. the responsible of natural disasters such as flooding, desertification, ecosystem and habitat destruction, represents one of the most important challenges of our century. Considering the climate as a non-excludable good we inevitably incur in the well-known outcome of the tragedy of the commons. In fact, when attempting to avoid global warming, individuals face a social dilemma in which, besides securing future benefits, it is also necessary to reduce the chances of future losses. Unfortunately, individuals, regions or nations may opt to be free riders, hoping to benefit from the efforts of others while choosing not to make any effort themselves. Pioneering work by Barrett [1–3] proposes a theory of full international cooperation that can be applied to the climate change problem. His findings provide a formal proof that only a “small” number of countries can sustain full cooperation by means of self-reinforcing environmental agreements. More importantly, his work shows that the constraint on international cooperation is free-rider deterrence, not treaty compliance enforcement. This last evidence is also supported in [4, 5], showing the importance of sanctions against free-riders for accomplishing cooperative climate agreements. A recent theoretical model has been proposed by the 2018 Nobel laureate Nordhaus on the formation of coalitions, or clubs, suggesting efficient solutions to tackle the dilemma and achieving high cooperation level outcomes [6]. At the individual level, social norms, defined as informal behavioural rules that are supported by empirical and normative expectations and potentially backed by enforcement [7], seem to be a promising way of helping solving climate change games situations in local interactions, but no experiments have been performed to measure them in this framework. The experimental setup that I will present is strictly related to the recent work by Nordhaus [6]. The basic idea is that international cooperative treaties can be achieved through the formation of climate clubs, i.e. coalitions of countries committed to abate their carbon fossil emissions, that sanction, by the implementation of harsher taxation regimes, countries outside the coalition. We investigate whether this simple theoretical mechanism can be applied to collective action problems when human participants, and not countries, are involved. 2 The experimental design is a repeated climate change game where each round is composed by three stages. In the first one participants pledge their participation to the climate club, in the second one they ratify it according to the number of other committed participants while in the last one they decide their contribution amount (considering that club members cannot free-ride). As in previous work, the disaster impact depends on the total contribution value but, now, club members also benefit from sanctions at expenses of non-member participants. Theoretical models predict that two Nash equilibria are present and can be reached more easily, or not, depending game parameters. In fact, the introduction of a sanctioning/rewarding system modifies the setup from a cooperation scenario to a coordination one. Accordingly, individuals can thus coordinate their actions to either full contribution or none.

In light of the recent popularity of online social networking websites and the increasing number of users consuming political content on these websites, the study of opinion formation in online settings is more relevant than ever before. Crucial to this study is social influence, a fundamental sociological concept that serves as the micro-mechanism to theories about cohesion, social distinction, and political polarization. On the one hand, there is a very rich theoretical literature studying the collective outcomes of social influence in networks. On the other hand, existing models are based on many competing assumptions about influence and empirical research testing these competing assumptions remains scarce. In particular, attempts to find negative influence, the tendency to increase opinion differences to disliked sources of influence, have so far remained fruitless, often criticized for being based on problematic sampling approached and inadequate experimental design.

We introduce a general theoretical model of social influence that is able to capture the most prominent assumptions about social influence. Next, we conduct a power analysis in order to inform the design a laboratory experiment able to test these competing assumptions against each other. In particular, we derive conclusions about a study’s necessary sample size, and prior opinion distribution.

Various mechanisms have been proposed that could explain observed opinion shifts, both at the individual and group level. These mechanisms largely fall into three categories: assimilation, distancing and alignment of opinions. There is, however, not yet an integrative framework that could explain when what type of influence is experienced. This is important since neither mechanism alone can explain the patterns of opinion polarization and convergence that we observe in society. When bridging the micro-macro gap in the opinion dynamics literature, a strong empirical basis for the behavioral rules that govern opinion change is required.

In order to test the three influence mechanisms we present a general theoretical framework that researchers can apply in experimental settings where fine grained data on pre and post stimulus opinion measurement are available. First, the broad landscape of social-influence theories is integrated into two hybrid formal models: a linear social influence and a so-called bounded confidence model. With these two models, we are able to distinguish between the experienced influence direction and strength, and the general persuasiveness of an argument, as a function of opinion discrepancy. Estimating the core parameters of the two models proves particularly challenging since survey measurement as well as behavioral measurement of opinions is usually truncated, resulting in odd-shaped multimodal distributions even for latent Gaussian distributed traits. Detailed power analyses with simulated datasets provide confidence in the estimation method and identify the most volatile model parameters.

Second, a general experimental setting is presented that could be applied in different contexts wherein -or subjects on which- opinion change is expected. Various conditions under which the influence function might differ are discussed. Lastly, an example of such a setting is presented in which we maximize opinion discrepancy in order to find out whether there is empirical evidence of negative influence.

Experimental methods are used in a rapidly increasing number of studies in economics, political science, and sociology. These disciplines share many topics – e.g., public goods, large groups, social preferences – and many theoretical puzzles. But findings do cross community borders only to a very limited degree. One consequence is that different conceptual and methodological approaches and criteria are applied in these separated communities, without much awareness and transparency about that circumstance. Therefore, the x-hub project aims to establish an infrastructure that fosters transparency, replicability, and reuse of experimental data across disciplinary borders. Diverging methods and paradigms applicable to primary data sets shall become visible at first sight, so that researchers from any sub-discipline can find and evaluate the data and research from their respective discipline-specific perspective (http://gepris.dfg.de/gepris/projekt/251955964?language=en).

Modeling the micro-macro link: Understanding macro-level outcomes using randomization tests on micro-level data

Analytical sociology explains macro-level outcomes by referring to micro-level behaviors, and its hypotheses thus take macro-level entities (e.g. groups) as their units of analysis. The statistical analysis of these macro-level units is problematic, since macro units are often few in number, leading to low statistical power. Additionally, micro-level processes take place within macro units, but tests on macro-level units cannot adequately deal with these processes. Consequently, much analytical sociology focuses on testing micro-level predictions. We propose a better alternative; a method to test macro hypotheses on micro data, using randomization tests. The advantages of our method are (i) increased statistical power, (ii) possibilities to control for micro covariates, and (iii) the possibility to test macro hypotheses without macro units. We provide a heuristic description of our method and illustrate it with data from a published study. Data and R-scripts for this paper are available in the Open Science Framework (https://osf.io/scfx3/).

Some fields of economics, studying health and well-being for instance, have so far relied on non-incentivized survey data, mostly because the survey answers are unverifiable. It has long been thought that it is impossible to incentivize people to truthfully report something as subjective as how they feel, or how happy they are. Over the past fifteen years, several methods have been proposed to incentivize truth-telling for such unverifiable questions, the best-known being the Bayesian truth-serum of Prelec (2004). We test the impact of truth-telling incentives in a large-scale, online experiment about health and well-being. We exogenously introduce a quantifiable bias in responses using defaults. We find that, in the presence of a default, incentives induce more effort, which reduces the default bias. We also find that, in the absence of default, data quality is good and incentives have no impact.

The “transportability” of laboratory findings to other instances than the original implementation entails the robustness of rates of observed behaviors and estimated treatment effects to changes in the specific research setting and in the sample under study. In four studies based on incentivized games of fairness, trust, and reciprocity, we evaluate (1) the sensitivity of laboratory results to locally recruited student-subject pools, (2) the comparability of behavioral data collected online and, under varying anonymity conditions, in the laboratory, (3) the generalizability of student-based results to the broader population, and (4), with a replication at Amazon Mechanical Turk, the stability of laboratory results across research contexts. For the class of laboratory designs using interactive games as measurement instruments of prosocial behavior we find that rates of behavior and the exact behavioral differences between decision situations do not transport beyond specific implementations. Most clearly, data obtained from standard participant pools differ significantly from those from the broader population. This undermines the use of empirically motivated laboratory studies to establish descriptive parameters of human behavior. Directions of the behavioral differences between games, in contrast, are remarkably robust to changes in samples and settings. Moreover, we find no evidence for either anonymity effects nor mode effects potentially biasing laboratory measurement. These results underscore the capacity of laboratory experiments to establish generalizable causal effects in theory-driven designs.

There are several advantages to eliciting individual preferences and behavioural tendencies using surveys. Surveys are cost-saving in terms of time and require relatively simple subject engagement strategies. However, social scientists who conduct incentivized experiments have repeatedly argued that survey measures of preferences could be nothing more than cheap talk and that, outside the lab, there could be considerable loss of control. Here, we present a set of studies the objective of which was to inform and validate the design of an unincentivized survey approach to estimate experimental treatment effects. We present three sets of results. The first is derived from a between-subject analysis of two independent, but comparable samples of non-student adults. One sample participated in a standard, incentivized laboratory experiment and the other participated in an unincentivized survey experiment. The two methods returned remarkably similar treatment effects. The second set of results relates to a sample of students drawn from a behavioural laboratory’s pool of registered subjects. They participated in both the incentivized lab and unincentivized survey experiments. We perform a between-subject comparison of the two treatment-elicitation methods but, this time, focusing on the same sample of subjects. Again, the treatment effects are very similar. Finally, we establish that within-subjects there is some consistency between decisions made under the two methods.

While increasing the number of donors, standard matching has been shown to cause considerable crowding out (Eckel and Grossman, 2003, Karlan and List, 2007, or Huck and Rasul, 2011). Can we design an alternative matching scheme that would lead to an increase in the values of gifts given? We propose a form of a threshold matching that matches any donation above a certain threshold with a fixed amount. First, we study how such thresholds should be chosen depending on past donations. We find that asking for less than the amount given in the past results in lower donations and lower revenue. Asking for the amount given in the past or a higher amount increases donations with a threshold of 60-75% higher than the past donation being optimal. However, while asking for less result in high compliance, asking for more also triggers some contrarian behavior. Second, we study how to use such personalized matching for prospective donors based on observable characteristics. Finally, we compare personalized threshold matching to a non-personalized version with the same threshold for all participants. In the sample of previous non-donors, higher threshold results in lower response rate, higher average donation, and lower return per mail-out.

How do voting institutions affect incentives of committees to vote for ethical alternatives? We model the problem of a committee, whose members have different depths of reasoning, that decides whether to approve an ethical proposal. Members who vote for the proposal receive expressive utility, but all members pay a cost if the proposal is passed. The model suggests that institutional features that reduce the probability of a member being pivotal – such as larger committee size, or a more restrictive majority rule – increase the expected share of votes in favour of the ethical alternative. A laboratory experiment with a charitable donation framing demonstrates comparative statics that are in line with these results. We also structurally estimate the distribution of expressive preferences across individuals, and find that a high proportion of them are strategically naive.

The competition for donations between charities is tough. Yet, little is known about how giving behavior is affected by competition between charities. Do people have a need to satisfy their demand for giving by contributing to a particular charity? Or can the demand for doing good be satisfied by giving to any organization? In a donation dictator game, I vary competition between charities by i.) altering the set of similar real charities to which subjects can donate and ii.) changing the relative price of giving to a randomly selected charity in the choice set by introducing a matching grant. I find weak substitution between charities when giving to more than one charity is possible, as the donated amounts to individual charities decrease with the size of the choice set. At the same time, aggregate giving to all charities increases when charities are in competition. Intensified competition through an increase in the charitable giving market seems to attract new giving. A reduced price of giving to one charity (matching grant), however, only increases giving to the matched charity. Price competition does not attract new donations as subjects substitute part of their own contributions with the matching grant and leave their total contributions unchanged.

Are charitable gifts are complements or substitutes; are charitable fundraisers rivals? These questions are important to fundraisers, policymakers, and economic models of generosity. However, observational data lacks shocks that are clearly specific to giving to one charity; motivating experimental variation. We report on recent experimental work (our own and others’) from a variety of laboratory and field settings, under a range of design choices providing convergent validation. Most work presents simultaneous or proximately repeated donation opportunities to subjects, with noticeably varying “prices”, promotional information, and choice sets. Here, specific shocks strongly increase giving to the targeted charities, and lead to decreased giving to the remaining charities, particularly where charities serve similar goals. However, this may be driven by experimental demand and logical responses to obvious contrasts, which may not reflect real-world behavior. Our web-based experiments offer robustness to these critiques. We pay participants for completing two surveys unrelated to charitable giving, separated by multiple weeks. Some (all) participants have the opportunity to donate from their earnings to a single charity at the end of the first (second) survey. Between subjects, we vary the number of asks and the similarity of the first and second charities. The results offer further evidence of "expenditure substitution", but less support for the idea that similar charities crowd out each other more.

Please note: We are uploading a previous version of the paper; we aim to have a version integrating data from a range of recent experiments within the next few months.

We conducted an incentivized experiment to measure competitiveness from a sample of the members of a representative survey panel ( N=1674 ). Individuals who compete in the experiment are more highly educated, choose more ambitious college majors, earn more and are more likely to be in a high-level professional position. One year later, we elicited competitiveness from all members of the panel through unincentivized survey measures ( N=5268 ). We show that the unincentivized and incentivized measures are strongly correlated at the individual level and predict the same outcomes. The predictive power of competitiveness for education and labor market outcomes is robust to controlling for other traits, including risk seeking, confidence and the Big Five personality traits. For most outcomes, the predictive power of competitiveness exceeds that of other traits.

We jointly investigate last place and rank reversal aversion, two mechanisms potentially weakening support for redistributive policies, to disentangle their effects. While (rank-dependent) rank reversal aversion nests last place aversion, that ranks are inverted is not a necessary condition to find oneself in the last place. Moreover, while last place aversion affects individuals close to the bottom of the distribution, rank reversal aversion might bite in all income brackets. We shed light on a potential reason for the failed replication of previous results on this topic, and discuss it in light of the literature on replicability of experimental results: After introducing a small change in the design, our data indicates that both mechanisms drive subjects’ behaviours. We find evidence for rank reversal aversion (independent of rank) having the strongest impact on subjects’ behaviour, in addition to pure last place aversion. We discuss implications for policy design from both the public finance and management science perspectives.

In a laboratory experiment, we compare the stability of bidding rings in the English auction and the first-price sealed-bid auction in a heterogeneous-value setting. In both a re-matching condition and a fixed-matching condition, we observe that biddings rings are more stable in the English auction than in the first-price sealed-bid auction. In both conditions, the first-price sealed-bid auction dominates the English auction in terms of average revenue. The English auction outperforms the first-price sealed-bid auction in terms of efficiency.

This paper analyses a multiple prize contest model with expectation-based loss averse contestants a la Koszegi and Rabin (2006, 2007). Contestants simultaneously exert costly effort and prizes are allocated to the contestants with highest effort. A contestant's cost of effort depends on his effort level as well as his ability, which is private information. The contest designer maximizes the total expected effort by allocating a limited amount of resources into prizes. The model is able to align the empirical evidence on effort provision, which is hard to reconcile with the standard economic assumptions. More specically, high-ability contestants overexert effort while low-ability contestants withhold effort in comparison to the predictions of standard preferences. Moreover, the optimal prize allocation differs markedly in the presence of lossaverse contestants: multiple prizes becomes optimal in the cases where the standard preferences predict the optimality of a single grand prize.

Can attitudes towards minorities, an important cultural trait, be changed? We show that the presence of African American soldiers in the UK during World War II reduced anti-minority prejudice, a result of the positive interactions which took place between soldiers and the local population. The change has been persistent: in locations in which more African American soldiers were posted there are fewer members of the UK’s leading far-right party, less implicit bias against blacks and fewer individuals professing racial prejudice, all measured around 2010. We show that persistence has been higher in rural areas and areas with less subsequent in-migration.

When asked to choose between candidates in elections, voters take a range of candidate characteristics into account, including their gender, ethnicity, and social class. Whether a candidate has a disability potentially plays a role too, given the stigma and stereotypes that continue to exist about people with disabilities in our societies. This study is the first to examines whether and how candidate disability affects voter perceptions of their traits, beliefs, and competences and, consequently, the vote choice, using a conjoint experiment in the UK. The findings from a pilot study suggest that candidates with depression but not those using a wheelchair are considered less competent, strong, and hard-working. Both types of disability, but more so the physical disability, are associated with perceptions of concern about and competence in handling disability-related issues. Disabled candidates are also seen as more left-wing. Meanwhile, the effects on the vote choice are limited. The findings provide important insights about the role of stereotypes in voting behaviour and have potential to contribute to addressing the underrepresentation of disabled people in politics.

Contact theory (Allport, 1954) proposes that prejudices towards an outgroup may be reduced when individuals have personal contact with members from the outgroup. Ample empirical evidence suggests that individuals who frequently interact with an outgroup have less biased attitudes. Whether this relationship is causal and in which direction the causality runs is hard to establish, however. We run a field experiment in which high school pupils engage in a group assignment during an introductory lecture at a Dutch university. Each group is accompanied by a university undergraduate which, by random assignment, is either a native Dutch student or a student with Arabic - North African roots. A month later the high school pupils participate in two experimental games in which they are randomly assigned to a partner with either a native Dutch or an Arabic first name. We find that contact with a non-native undergraduate student during the introductory lecture significantly reduces the behavioral bias against a partner with an Arabic first name. This only holds, however, for pupils who have no prior exposure to non-native pupils in their school class.

In the recent development of the sharing economy and other peer-to-peer digital markets, (ethnic) discrimination has become an important issue. Reputation systems are often put forward as the most promising solution to (ethnic) discrimination in the platform economy. This claim is based on the finding that the ethnic gap is smaller for users with reviews than for users without reviews. However, as reviews are only written after completed interactions, the chance to get a review may not be equal for all users. Hence, initial differences between users in the probability to be selected for a transaction may accumulate over time, thereby diminishing the potential of reputation systems to decrease discrimination. Previous studies have overlooked this potentially negative effect of reputation systems. In the current study we use data use data from a peer-to-peer motorcycle sharing platforms to study the relation between reputation systems and discrimination. In the current paper we do not find evidence for the potential of reputation systems to reduce the ethnic gap. Rather, we find that the ethnic gap persists even for renters with a positive rating. Regardless of the reputation of the renter, requests from renters with more non-Dutch sounding names are less likely to be accepted. This decreases their probability to get a (positive) review, which in turn negatively affects their chances to participate in future interactions.

Daily life provides individuals with many occasions to compare themselves to others. For example, neighbors compare their houses, athletes compare their performances, coworkers compare their incomes, and academics compare their publications. Such comparisons can have important effects: they might alter utility and ultimately affect behavior. Previous studies have shown that social comparisons matter in settings as diverse as consumption decision, effort choices (Bandiera et al., 2010), happiness (Clark & Oswald, 1996), risk taking (Schwerter, WP), financial decisions (Bursztyn et al., 2014), and social preferences (Fehr & Schmidt, 1999), amongst others. A key ingredient of such social comparisons is the determination of the relevant comparison group (i.e. the social reference point). In most domains, individuals have the chance to compare themselves to a large variety of individuals. But are all comparisons equally relevant? Most likely not. A casual marathon runner, for example, will probably care more about the time achieved by his best friend than by the time achieved by professional runners. Similarly, workers are probably more affected by learning the salary of their coworkers than by learning the salary of the CEO of another company. To this point, much of the evidence on peer effects and social reference points has focused on environments in which peers are exogenously assigned to individuals. However, in many settings individuals have the opportunity to actively choose whom to compare to. For example, a casual marathon runner can choose to compare to his friends and not look at the time achieved by professional runners. Even in settings in which it might appear that individuals do not explicitly single out a peer, the individual can still choose who to attend to and ignore how other individuals behave. At the workplace for example, young professionals often pick one coworker as a role model, and focus much less on how other coworkers perform.

This lab experiment tries to identify potential mechanisms behind peer effects on debt taking. Various studies have shown that social comparison influences consumption and spending decisions but the mechanisms behind these effects are still not clear. In our experiment, participants can earn money in an IQ-style test. They are paid according to their performance in relation to others in the session. Afterwards, they can buy different quality pens using the money that they have earned during the IQ-task. There is one quality pen that corresponds to each level of earning. However, all participants can take a loan that enables them to purchase a higher quality pen. There are three different treatments that vary the way participants decide and communicate which pen they would like to buy. In the private treatment participants make the decision simultaneously and the decision is kept private. In the public treatment, participants again make the decision simultaneously but know that they have to announce their choice of pen publicly later on. In the information treatment, participants make the decision sequentially in random order. Therefore, we can show the participants which pen previous participants bought. However, the identity of these are kept private. Following the consumption decision, all participants perform a slider task for four minutes. The way the slider task is performed is independent of the previous treatment. The money that participants earn here is used to repay any loan that they may have taken in the shopping round. Results so far suggest that participants in the information treatment take the most debt. Surprisingly, there is no difference between the private and the public treatment in the amount of debt taken. However, individuals in the public treatment spend significantly less on pens, meaning they buy a pen that is of lower quality than the one that they could afford. The same holds for participants in the information treatment. Thus, spending behavior in this treatment is the most heterogenous. Looking at the slider task, we find that individuals in the information treatment exert significantly more effort than individuals in the private treatment. The difference between private and public treatment is statistically not significant. Contrary to our expectations, persons who take out more debt in general also perform worse in the slider task. First results indicate that this may be driven by poor focus or lower ability, rather than the experimental treatment.

Social influence and conformity have been thoroughly studied by social psychologists, leading political scientist to the postulation of the bandwagon effect. However, despite its relevance, the bandwagon effect, claiming that seeing pre-election polls makes votes for majority options more likely, has not been properly tested for political voting and evidence regarding this effect is mixed. Experiments either were ran in very abstract contexts only vaguely representing political voting, or only testing the effect of polls on opinions or hypothetical votes . We present an unique experimental design with more realism than previous designs, keeping the experimental control needed to make causal claims.

We tested in an online experiment with 1115 participants from the US how votes change when pre-election polls are shown. Our experimental design is unique in that the votes of the participants have real-world consequences within their electoral area (the US), empower actual political organizations falling on a left-right spectrum, and the votes are on issues currently debated in US politics (firearms, abortion, immigration, and environment). Per issue participants chose between three different charities, representing different positions within the political spectrum. As a result of the experiment 1200$ were distributed across these charities as suggested by the vote.

In line with the bandwagon-effect, we find clear evidence that seeing poll results makes votes for majority opinion more likely. After seeing the surveys, majority opinions received an extra ~7% of votes. In our experiment this effect did not depend on the electoral system and was robust against controlling for the gender and age of voters and the self-assessment on a left-right spectrum. However, we find evidence that under extreme-polarization (where moderate position are the least popular option) the bandwagon-effect is much weaker.

Improvements in the upward social mobility of a group or society can paradoxically lead to more frustration (relative deprivation) among its members. At present, there is little understanding of the conditions under which this paradox occurs. Boudon created a game-theoretic model of relative deprivation that can be used to derive predictions for these conditions. We extend his model, and through computer simulations provide predictions for the relationship between societal mobility and relative deprivation, of which a subset is tested within a laboratory experiment. Among other things, we predict and find that if social mobility moves from a low to medium level, and the costs of investing in a better position are low, relative deprivation increases.

How do investors assess the risk of investments? Traditional finance does not leave any room for discussion: variance of returns (volatility) is the one and only risk measure. However, concepts like loss aversion already indicate that investors care more about downside than upside potential. So, do investors primarily care, for example, about skewness, semi-variance or loss likelihood, to name just a few possible risk measures? This study explores the important questions of whether there is any objective risk measure that can best explain what investors implicitly have in mind when they assess the risk of investments, and how this affects their willingness to invest. Finding an answer to this research question is of high importance not only for researchers but for business practice and the financial regulator alike. For example, the understanding of what is risk for investors will considerably facilitate how risks can or should be communicated.

We conducted a series of five experiments and 1000+ participants in which we present participants with various different return distributions and ask them how risky they perceive these investments and how likely they would invest. By a careful experimental design we keep the expected return constant for all different return distributions, but we achieve pronounced differences in various risk measures. We thereby analyze a large variety of different measures: standard deviation (volatility), semi-variance, loss probability, skewness, kurtosis, maximum return, minimum return, value-at-risk.

We find that volatility, most frequently used by financial advisors and even the regulator, fails to explain how investors perceive risk. In contrast, our results highlight the crucial role that the probability of losing plays in determining the perceived risk by an investor. This measure alone can explain a very surprising 98% of the variations in the average perceived risk in our baseline experiment, and it is the only factor that delivers a non-zero explanatory power for the individual perceived risk. We further find that perceived risk is the main driver of investment propensity, affecting it negatively. We obtain very similar results in our four control experiments in which we control for the effect of color coding of gains and losses (Exp. 2), investor sophistication (Exp. 3), incentives for the investment task (Exp. 4) and exclusive focus on investment decision without asking for the riskiness of the investment (Exp. 5).

One of our conclusions, having shown that loss probability is the most important risk factor, which could lead to suboptimal decisions, is that (robo) advisors should out more effort in explaining to their clients the consequences of a loss, for example with simulation tools, which have been shown already to be effective in the case of mis-estimation of loss probability (see Kaufmann et al. 2013, Bradbury et al., 2015). The regulator should consider using alternative approaches to communicate the risks of investments. Our paper is intended to set the stage for a new research agenda in the field of investor risk perception and effective risk communication, to enable investors to make better and more informed decisions.

People are susceptible to various risks that can bring damage to their homes, the content of their homes, and even to themselves. Two frequently applied theories in academia have analyzed how people deal with risk and risk prevention: Protection Motivation Theory (PMT; Rogers, 1975) and the Health Belief Model (HBM; Rosenstock, 1966). The basic idea behind PMT and HBM is that the perceived likelihood of a risk and its perceived consequences create a motivation for self-protection, and a (perhaps implicit) cost-benefit analysis that results in taking action or not. The suggested determinants (vulnerability, severity, costs, effort, and effectiveness), have been extensively tested in various areas of prevention behavior research. We now test them in the area of domestic risk prevention.

Perhaps more importantly, in this paper we highlight an issue that has been neglected throughout the literature, as far as we can see. Suppose that one finds that, say, the perceived effectiveness of a behavior correlates with the probability that someone indeed performs this behavior as many researchers have found (Floyd et al., 2000; Janz & Becker, 1984; Milne, Sheeran, & Orbell, 2000). What does this mean? It could mean that persons who find this particular behavior more effective than other persons, perform the behavior more often (an effect “between-persons”). Or, it could mean that for a given person, prevention behaviors that are perceived as more effective than other behaviors are more likely to be performed (an effect “between-behaviors-within-persons”). Although these two implications are not the same, the literature on PMT and HBM does not make this distinction. Most studies do not clearly state which of these two interpretations is the one they imply, although the (mostly implicit) general argument that is being used is on the within-person level: if, say, the perceived effectiveness of a prevention behavior would become higher for a given person (and everything else remained equal) that person would be more likely to perform that behavior. However, although the general argument is on the within-person level, correlational studies perform analyses solely between-persons. The difference may sound subtle, but whether the effects are primarily within- or between-persons has important implications for initiatives directed at influencing prevention behavior. When effects are mainly between persons and hence depend on personal characteristics (a person’s general perception of the effectiveness of prevention behaviors, socio-demographics, etc.), it would make sense to direct general prevention initiatives at specific target groups. However, when effects are mainly between behaviors within persons, it would make more sense to try and influence specific behaviors for a general audience.

We use survey research (n=263) in which we ask our participants to evaluate several prevention behaviors and their characteristics. Our results show that all determinants influence domestic prevention behavior (as hypothesized). Given our design, we can separate the effects, and find that the bulk of the effects run within persons (between behaviors), not between persons. We discuss our findings and conclude with implications for policy makers and others interested in influencing people to increase their (domestic) risk prevention behavior.

A decoy is an irrelevant option that, when added to a binary choice, is not selected but nonetheless alters the subjects’ preferences between the other two options, systematically biasing towards one of them (the target of the decoy). Since their first experimental observation (Huber, Payne, & Puto, 1982), the decoy effect has been considered an important anomaly of rational decision-making, albeit recently its applicability to real-life choice scenarios has been challenged (Shane, Lee, & Baskin, 2014; Yang & Lynn, 2014) and is now a matter of some debate (Huber, Payne, & Puto, 2014). In particular, decoys have been often studied in choices between outcomes occurring at different points in time, i.e. intertemporal choices (Kowal & Faulkner, 2016; Gluth, Hotaling, & Rieskamp, 2017), with mixed results: sometimes decoys are impactful, sometimes they are not, and in general they seem to be more effective in biasing towards larger- and-later (LL) outcomes, rather than towards sooner-and-smaller (SS) rewards.

We suggest that this puzzling set of results is due to an underappreciation of two important influencing factors: time pressure (Pettibone, 2012) and immediacy / certainty (Weber & Chapman, 2005). Moreover, we argue that decoy effects constitute an excellent testbed to assess similarities and differences between intertemporal choice and risky decision-making, which constitutes another open issue of debate (for discussion, see Green & Myerson, 2004).

Two studies are presented to support these claims. In Study 1 (N=92), we demonstrate that asymmetrically dominated decoys influence both economical intertemporal choice and risky decision-making only in the absence of time pressure, since otherwise the comparative process required for the decoy to have an impact cannot occur, consistently with predictions made by connectionist models of decision (e.g., multialternative decision field theory, Roe, Busemeyer, & Townsend, 2001). In Study 2 (N=53), we show that, when the SS option is no longer presented as immediate / certain, the impact of decoys on intertemporal and risky choice becomes symmetrical – that is, decoys can prompt subject to become either more economically patient / daring or more impulsive / prudent, since the anomalous element of immediacy and certainty has been removed from the equation.

We conclude by discussing the implications of these findings for our understanding of the multifaceted role of time and chance in decision making.

We run an incentivized choice process experiment along the lines of Caplin et al (2011) to assess the robustness and nature of the Asymmetric Dominance Effect (ADE, Huber et al., 1982), the most prominent example of context effects. According to the ADE, adding a dominated option to a choice set increases the choice share of the newly dominant option at the expense of other options. While widely replicated, the ADE is usually found in hypothetical or payoff-irrelevant situations, and without considering the choice process. We systematically vary the utility difference of the options and track the choice process in real time.

We use an expenditure minimization task. Subjects are given a budget to buy three (fictitious) liters of gasoline and keep what they do not spend. We split price per liter into two different dimensions (quantity and price per such quantity), so that subjects have to estimate unit prices.

We vary across decision tasks the relative price of the target (the dominant option) to the competitor (the non-comparable other option) keeping fixed the premium of the target with respect to the decoy (the dominated option). This allows us to assess the importance of the ADE away from indifference and to measure the cost sustained by each subject in following the ADE.

Subjects face each choice for a given time and are free to change their mind during this time. We keep track of all the provisional choices. Once the time has ran out, we draw a random uniform moment and the item chosen at that moment is binding. If no choice had been made, then an item is chosen at random. Subjects are incentivized to give a fast reply and then reconsider.

We find that the ADE is a transitory phenomenon, that disappears when subject are given enough time and incentives to ponder their choices. The ADE emerges for the most part only in the early stages of the choice process. Consumers provisionally choose the asymmetrically dominant option to avoid the dominated decoy and then progressively switch until choice shares come to correspond to price differences only.

We also allow subjects to differ by types. We define four types of decision-makers: pure price (P), who disregard dominance and decide based on price estimates only; pure heuristic (H), who exploit dominance irrespective of prices; heuristic then price (HP), who first use dominance then adjust based on prices; and random (R).

We estimate the proportion of each type with a mixture model. 50% of our participants are HP, 18% are P, 26% are H, and less than 2% are random.

Our results show that the ADE consistently affects only a subset of subjects, while for most others it disappears given time and incentives. Its importance in the applied literature and popular press seems in light of our results to have been exaggerated.

How should an algorithm deal with the danger of making a wrong decision? The current work addresses this question in an experiment. In it, subjects set a parameter which controls a data-driven algorithm. This parameter, a threshold, determines whether a person’s preference for risk is high enough for a financial risk to be taken. A lower threshold means that the algorithm implements risky choices more often and that the rate of false positives is higher. Conversely, a higher threshold implies less risky choices but also more false negatives. The findings are as follows: i) When setting the threshold for themselves, subjects leave the algorithm little influence in the eventual decisions by either setting the threshold very high or very low. When setting it for others, they leave the algorithm more influence. On average, however, chosen thresholds are the same when subjects set it for decisions which affect themselves as when they set it for decisions which the algorithm makes for others. ii) Subjects respond to conflicts of interests: When setting the threshold for decisions which affect others, subjects respond to a bonus for inducing risky decisions. They do so by setting the threshold lower than when the bonus is absent. iii) Threshold choices in all treatments are unaffected by whether the choice procedure is framed as being based on a “computer algorithm” or a human-developed “decision rule”.

We introduce DOSE - Dynamically Optimized Sequential Experimentation - and use it to estimate individual-level loss aversion in a representative sample of the U.S. population (N=2,000). DOSE elicitations are more accurate, more stable across time, and faster to administer than standard methods. We find that around 50% of the U.S. population is loss tolerant. This is counter to earlier findings, which mostly come from lab/student samples, that a strong majority of participants are loss averse. Loss attitudes are correlated with cognitive ability: loss aversion is more prevalent in people with high cognitive ability, and loss tolerance is more common in those with low cognitive ability. We also use DOSE to document facts about risk and time preferences, indicating a high potential for DOSE in future research.

Prosociality is defined as voluntary behavior intended to benefit another. When prosocial behavior comes at a cost and without immediate advantage for the actors, the classical rational choice theory (RCT) using the assumption of materialistic egoism predicts that prosocial behavior will not occur. However, a plethora of experimental investigations shows a completely different picture. Although the use of social instead of egoistic preferences can resolve some of the differences between prediction and observation, still a number of systematic deviations, such as framing effects, cannot be adequately explained.

One promising approach to close this gap comes from the dual process theories (DPT), according to which the RCT represents only one of two ideal-typical variants of human decision-making behavior. In contrast to rational considerations, actors might choose actions via an intuitive decision-making system. Whether actors make their decision via the rational or the intuitive system depends on individual as well as on situational characteristics. Hence, RCT is only a special case of DPT and in order to be an improvement over the simpler RCT any application of the DPT must meet two crucial criteria: First, it must reliably predict whether actors will choose their actions via the rational or the intuitive system and, second, it must tell us what the intuitive action will be.

For this reason, we present experimental evidence on the situational conditions for prosocial behavior. In our work, we systematically vary the situational factors that, according to the most common variants of the DPT, should have an impact on the decision-making process as well as on the decision itself, such as time constraints or framing of the situation. With this research we contribute to the empirical assessment of the predictive value of DPT in the context of prosocial behavior.

Subconsciously, web browsing opens up a world of styles and impressions. To some of those styles we are used to, but often it is just unfamiliar language, new images and experiences. Does it matter that we read text, information, entertainment written in style (and language) that is not our own ? The current debate both in academia and the real world about the negative effects of the internet on political polarization are well known. The arguments circle around overt features of communication, social networks and outright misinformation. What is often overlooked though is the impressionistic richness of our online lives. This online field experiment / project tests predictions on the effect of style on 1) biased reasoning and 2) political fragmentation. Our conjecture is the idea that unfamiliar style makes us actually more realistic about power and the need for compromise. Using online field experiments, we provide evidence that diverse style, as opposed to diverse content, can enhance contrasting and political realism.

This paper tests with an experiment on a minimum effort game Keynes’s conjectures that ‘animal spirits’ both affect decision making in the macroeconomy and that their influence is often more significant than that of rational calculation based on the facts. We do this by using a psychological theory of explanatory styles to spin news stories with the same facts either optimistically or pessimistically. In this way, the experiment also tests the power of ‘narratives’ to influence the macroeconomy. We find evidence to support both of Keynes’s conjectures and we draw policy insights that might flow from this test.

Moral reminders can be a powerful drivers of behavior change. Existing literature also suggests that the most compelling moral appeals have a number of good ingredients, including targeted framing, timing and saliency of the message, and social norm interventions. Our study builds on this line of research to inquire: is dissemination of normative information powerful in itself? Or does the medium through which one disseminates information have a distinct upshot on eliciting desired behavior? To address these questions, we designed an experimental study to evaluate the impact of social norms and communication media on dishonesty. Our research examined whether who communicates a social norm (male vs female), and how the social norm is communicated (written vs audio vs visual instructions) have any impact on cheating behavior. We recruited 1725 participants on Amazon’s Mturk and they were presented with a roll-a-die task. In our two-by-three between subjects design, the treatment messages varied along two dimensions: 1) Whether the social norm conveyed was normative or empirical (respectively: Most people do not cheat/ Most people believe one should not cheat) and 2) whether the message was conveyed in written (message on their screen), audio (audio instructions via either male or female voice) or visual form (an image).

Our study also points at the realisation that moral gender disparities might be accentuated by specific communication media, notably visuals. Secondly, people who believed that the majority of other participants engaged in cheating behavior, were more likely to cheat, thus conforming to their perceived social norm.Thirdly, when we further analysed the data to observe whether the gender of the audio condition influences cheating, we find that the most effective audio treatment is one that combines empirical norms and female voice. Counterintuitively, treatments with a female voice had a lower cheating rate than a male voice, serving as a more powerful motivation for ethical behavior (difference of about 3 percentage points). We observe that this finding has support in the real-world when we think about navigation systems that default to female voice such as Siri or Google's GPS satellite systems. Currently, we are working on isolating this gender effect in audio treatments. We also find other interesting results on participant’s dispositions to risk-taking, honesty and norm perceptions.

Overall, this research proposes that the success or failure of a moral appeal - sometimes - is not directly proportional to the way in which the message is disseminated. Indeed, social norm interventions need to be activated, made salient and delivered in the right context, but we also speculate the “who” communicating the norm may equally matter. In conclusion, we hope these results offer a unique perspective on the vividness of different media on norms, especially as we think about designing large-scale behavior change initiatives.

Voice has been suggested to be a cue of cooperativeness since several acoustic features influence cooperativeness ratings. However, no experimental study has investigated whether these acoustic features are associated with actual and not only perceived cooperativeness. Still, this question is crucial to disentangle whether inferences of traits from voices only reveal stereotypes or enable hearers to extract useful information with respect to cooperativeness. To fill this gap, we quantified cooperativeness of 64 native French men through a one-shot public good game, and measured mean fundamental frequency, pitch variations, roughness and breathiness from spontaneous speech. We found that men with lower-pitched voices and with greater pitch variations were more cooperative. This is the first evidence of acoustic correlates of cooperativeness. Combined with the literature on face-based cooperation detection, we suggest that more than one sensory modality advertise cooperativeness and could therefore be simultaneously used to assess cooperativeness more accurately.

The emotion that someone expresses has consequences for how that person is treated. We study whether people strategically adjust their expressions in games with different incentive structures. In two laboratory experiments, participants play task delegation games in which managers assign a task to one of two workers. When assigning the task, managers see pictures of the workers and we vary whether getting the task is desirable or not. We find that workers strategically choose emotional expressions and that it indeed pays off to do so.

Sequentiality of moves in an infinitely repeated prisoner's dilemma does not change the conditions under which mutual cooperation can be supported in equilibrium as compared to simultaneous decision-making. The nature of the interaction is different, however, given that the second mover in a sequential-move game does not face strategic uncertainty in the stage game. We study in an experiment whether sequentiality has an effect on cooperation rates. We find that with intermediate incentives to cooperate, sequentiality increases cooperation rates by around 40 percentage points after learning, whereas with very low or high incentives to cooperate, cooperation rates are respectively very low or high in both settings. The findings are generally in line with the notion that strategic uncertainty is a key determinant of behavior in repeated games.

We report on an experiment conducted to study the effect of strategic substitutability and strategic complementarity on cooperation in innitely repeated two-player games. We find that choices in the first rounds of the repeated games are significantly more cooperative under strategic substitutes than under strategic complements and that players are more likely to choose joint-payoff maximizing choices in the former than in the latter case. We argue that this effect is driven by the fact that it is less risky to cooperate under substitutes than under complements. We also find that choices under strategic substitutes do not remain more cooperative than under complements over the course of the rounds within the repeated games. We show that this is because best-reply dynamics come into the picture: players are more inclined to follow cooperative moves of the partner under complements, offsetting the treatment effect observed in the first rounds.

We investigate the effectiveness of status as a means of enabling players to coordinate on socially superior outcomes despite having conflicting individual interests. Theoretically, we illustrate how an exogenous and arbitrary status ordering, ranking all members of a social group from high to low, permits coordination on an efficient -- yet fragile -- equilibrium if and only if the social group reaches a consensus on the appropriate compensation for conceding the privilege of high status to another player. We further analyse the conditions under which non-coercive transfer institutions help reaching such consensus. Experimentally, we find the availability of status to lift coordination clearly above the mixed equilibrium. And yet, coordination falls significantly short of the efficient status equilibrium due to (i) substantial under-investment into compensation, and (ii) a surprising reluctance of middle ranks (rather than low ranks) to concede.

We experimentally examine the effects of varying time pressure in a coordination game with a label salient focal equilibrium. We consider both a pure coordination game (payoff symmetry) and a battle of the sexes game with conflict of interest (payoff asymmetry). In symmetric games, there are no effects of time pressure, since the label-salient outcome is highly focal regardless of how much time subjects have to decide. In asymmetric games, less time results in greater focality of the the label-salient action, and it becomes significantly more likely that any coordination is on the focal outcome.

The interaction of culture and institutions is of great relevance for social and economic interactions. Specific cultural trajectories in different societies (that base on different histories) can influence the adoption and success of (newly installed) institutions. Identifying causality between culture and institutions, however, is difficult with field data. To investigate the role of culture on the (acceptance and) prevalence of sanctioning institutions, we conduct a series of controlled lab experiments in two countries with different cultural trajectories: Germany and Turkey. If imposed exogenously, an institution with individual sanctioning opportunities performs well in establishing cooperation in Germany, but less so in Turkey. If the same sanctioning institution is one of two alternatives that people can freely choose in an endogenous choice setting, then the sanctioning institution is the clear winner against a non-sanctioning institution, in Germany, as well as in Turkey. Though we find differences in people's initial institutional preferences and contribution behaviors, the dynamics of institution choice and both the evolution of contributions and sanctioning behaviors are remarkably similar in both countries.

Immigration has rapidly changed the demographic profile of most European societies, increasing ethnic, linguistic and religious diversity. A large literature based mainly on observational evidence claims that high levels of ethnic diversity undermine cooperation and the provision of public goods in modern societies. Yet, experimental tests of the proposition that ethnic diversity leads to a lower capacity to solve collective action problems have produced inconclusive results. Moreover, it is unclear whether the mechanisms that are known to be effective in stabilizing cooperation in homogeneous groups work similarly in heterogeneous groups. We conducted a public goods experiment with Italian and immigrant residents of the metropolitan city of Milan to examine the effects of diversity on cooperation. Results show that cooperation is initially lower in mixed than in homogeneous Italian groups because of both compositional as well as contextual effects. We also find that one of the most frequently citied mechanisms sustaining high-levels of cooperation in homogeneous groups - peer sanctioning - is less effective in bringing about cooperation in some heterogeneous groups but not others, calling for a more careful examination of the composition of diverse groups and more generally, the mechanisms that can restore cooperation in the presence of diversity.

Producing goods often creates negative externalities, such as bad working conditions. Usually, people do not want to harm a third party when they can decide individually but competition might promote unethical behavior. In a social dilemma game with an affected but inactive third party we let players form a costly institution that allows informal punishment to reduce externalities and competitive disadvantages. An exogenous implemented punishment institution reduces the harm imposed on a third party but social welfare decreases compared to the standard game due to losses by severe punishment. When players can choose to join groups that implemented the punishment institution endogenously, such groups act in favor of the third party and do not punish each other. In this case social welfare increases significantly. However, only a third of all participants choose the punishment institution, so that social welfare in total is not affected.

This paper reports on an experiment designed to test the effectiveness of coordinated punishment in teams. In our setting, participants decide how much to contribute to a team account in the first stage of the game and punish other team members in the second. While team production technology is linear, punishment requires coordination, as punishers pay in full the points they send but only the minimum number of points is received by punished participants. We study how the presence of an enforcing institution in one domain (sanctions) shapes behavior in another (contributions). We compare two punishment institutions (of High and Low effectiveness) with three benchmarks (with and without punishment). Coordinated punishment significantly, but moderately, increases contributions in both the High and Low conditions, and significantly increases earnings only in the former. We find strong evidence of behavioral spillovers between the punishment and contribution stages in the High condition. Successful teams learn to contribute most of their endowment without learning to punish, as they fail to coordinate sanctions.

We report experimental results and test cooperative models of unstructured bargaining by checking the empirical relevance of the underlying axioms. Our data support strong efficiency, symmetry, independence of irrelevant alternatives and monotonicity, and reject scale invariance. Individual rationality and midpoint domination are violated by a significant fraction of agreements that implement equal division in highly unequal circumstances. Two well-known bargaining solutions that satisfy the confirmed properties explain the observed agreements reasonably well. The most frequent agreements in our sample are the ones suggested by the equal-division solution. In terms of the average Euclidean distance, the theoretical solution that best explains the data is the deal-me-out solution (Sutton, 1986; Binmore et al., 1989, 1991). Popular solutions that satisfy scale invariance, individual rationality, and midpoint domination, as the well-known Nash or Kalai-Smorodinsky bargaining solutions, perform poorly in the laboratory.

Many-player divide-the-dollar games have been a workhorse for theoretical and experimental analysis on multilateral bargaining. If we deal with a loss, that is, if we consider many-player ``divide-the-penalty" games for location choices of obnoxious facilities, allocation of burdensome chores, and climate change summit to reduce carbon dioxide emissions, the theoretical predictions do not merely have flipped signs of those in the divide-the-dollar games. We show that the stationary subgame perfect equilibrium (SSPE) is no longer unique in payoffs. The most ``egalitarian" equilibrium among the stationary equilibria is mirror-imaged to the unique SSPE in the Baron-Ferejohn model. That equilibrium is fragile in the sense that allocations are sensitively responding to the changes in parameters while the most ``unequal" equilibrium is not affected by the changes in parameters. Experimental evidence clearly support the most unequal equilibrium: Most of approved proposals under a majority rule involve an extreme allocation of the loss to a few members.

Second-price auctions with public information, such as those on eBay, provide an opportunity for sellers to use the information from finished and ongoing auctions when acting strategically in future auctions. Sellers have frequently been observed to bidding on their own item with the intent to artificially increase its price. This is known as shill bidding. Using lab experiments with two sequential auctions, we study the effect of shill bidding when the seller can choose to shill bid in the second auction. We also study the impact of different information revelation policies regarding the provision of the first auction bidding history to the seller. The experimental data confirm that shill bidding in the second auction affects outcomes in both auctions. Our findings are consistent with the predictions that the threat of shill bidding in the second auction does increase the bidders' final bid in the first auction. However, providing the seller with the bidding history from the first auction does not affect any important outcome variables.

Asking-prices in negotiations serve as anchors because the final agreement often settled near the mid-point between the asking-price and the counteroffer. Precise, rather than round asking-prices serve as stronger anchors. Consequently, a popular advice for sellers is to set precise asking-prices, drawing on research focusing on a buyer’s market – where supply exceeds demand. Here, four pre-registered experiments as well as an analysis of real estate data reveal that in a seller’s market, where demand exceeds supply and buyer bid above the asking-price, setting a precise asking price is suboptimal. In a seller’s market, precise asking-prices lead to lower counteroffers, an effect driven by people’s adjustment on a finer-tuned scale, but not by their evaluations of the seller’s competence or the competitiveness of the market. Analyzing real-estate data from a fierce seller’s market shows that by increasing the roundness of the asking price by one level (for example from precise to the thousands to precise to the tens of thousands), sellers can gain thousands of euros per deal. Precision should be viewed and used in context.

Narratives pervade almost any aspect of our life and play a particularly important role in moral and prosocial decision-making. We study how positive (stories in favor of a prosocial action) and negative (stories in favor of a selfish action) narratives influence prosocial behavior. Our main findings are that positive narratives increase giving substantially, especially for selfish types, compared to a baseline with no narratives. Negative narratives, on the other hand, have a differential effect. Prosocial types decrease their giving, while selfish types give more than in the baseline. We also find that positive narratives lead to a binary response (comply or not comply), while negative narratives induce a more gradual trade-off.

A growing body of evidence suggests that individuals are on average overconfident about their ability, affecting career and financial decisions, among others. We investigate how overconfidence may persist in the face of objective feedback. Self-attribution biases are said to exist when we take credit for good outcomes, but blame poor outcomes on external factors. While heavily studied in social psychology, and often referenced in economics, rigorous evidence is scarce. We present a modified Bayesian model of self-attribution bias, which gives testable predictions for two types of this bias, (1) noisy: which generates positive asymmetric updating about oneself, and as such has been studied previously, and (2) fundamental: individuals mis-attribute positive feedback to themselves, and negative feedback to an external fundamental. We test the theory in an experiment where subjects are matched into two person teams. Individuals are overconfident, and take too much credit for positive feedback. However, both types of self-attribution bias are rejected, as subjects significantly under-weight negative feedback, without attribution to either teammate.

False beliefs about natural, health, economic and social issues are pervasive in society. Many persist even when contradicted by scientific evidence. Some examples are the beliefs that vaccines cause autism, that GM foods are not safe, or negation of climate change. Research in cognitive psychology shows that this behavior is related to several cognitive biases that affect the human mind. Consequently, dispelling misconceptions may be hard, even after exposure to sound information. Indeed, evidence shows that college level students stick to false previous beliefs on economic issues after a semester-long exposure to an economic principles course, independently of grade performance.

In this paper we report on the results of field and lab experiments designed to investigate the ability of a particular communication strategy, the refutation text, to reduce a widespread economic misconception: the belief that rent controls make housing available to more people. The refutation text explicitly states that a given belief is a misconceptions, provides data and arguments contradicting it, and shows the negative consequences of holding the false belief. This type of text has been used in psychology and other fields but not in economics, to the best of our knowledge. The field experiment is conducted using two cohorts of college students enrolled in an introductory economics course. Two surveys are conducted, one at the beginning and the second at the end of the semester, to obtain students’ beliefs. The control cohort is exposed to a standard lecture on price controls, while the treatment cohort is exposed to the refutation text. Using Angrist and Pischke’s (2009) approach, we find that the intervention has a statistically significant impact on the previously held misconception, inducing a belief change aligned with expert reasoning and evidence.

We then take the field to the lab in order to obtain additional insights about the workings of the refutation text. We ask four questions: i) do field results replicate in a different environment? ii) do some particular cognitive factors affect the change of beliefs? iii) does individual or group reading of the refutation text affect results? and iv) are short-run and long-run effects different? In the lab experiment we use a control group and two treatment groups, one where the refutation text is read individually and a second one where it is discussed in small groups. Questionnaires include a cognitive reflection test and a Wason task. Preliminary results show that the refutation text reduces the misconception also in the lab.

Willful ignorance is a strategy aimed at avoiding accountability and at reducing moral costs of dishonest, inappropriate, or immoral actions. Most of the literature focuses on the demand side of willful ignorance and does not consider that, in many cases, ignorance comes from the decision of a different agent, e.g., advisor, newspapers’ editor, etc., to share or communicate information. With this project we experimentally study whether and how informed agents, i.e., advisors, decide to suppress information about potentially negative consequences of decision makers’ actions for non-strategic reasons—when advisors are not in competition—and for strategic reasons—when they advisors compete to advise. The results of the experiment show that: (i) advisors are willing to suppress information in the absence of competition and suppression depends on their personal preferences to be informed; (ii) Suppression of news with and without competition are not statistically different.

I examine whether gender quotas have improved political participation and reduced the gender gap in politics for women in the context of Delhi. I disaggregate and measure political participation in general and context specific ways. I exploit the natural experiment of random quota assignment, where in 50% wards were randomly reserved for women in 2012 and 2017. I embed a two wave panel survey of 1500+ HHs across 18 wards in Delhi within this natural experiment to examine the difference in levels of participation and gender gap between reserved and non-reserved wards. Apart from women's participation and gender gap outcomes, I examine whether quotas were able to mobilise both men and women in the context of anti-corruption protests, anti-rape/ sexual violence and environmental protests that have gripped Delhi since 2012 till date. Lastly, I embed a survey experiment to examine whether being informed about the quota policy as well as gender and quality of the councillor has any effect on perceptions of inclusion and democratic accountability in the political system. While previous studies have examined village level quotas, this is the first examination of quotas on broader participation in an urban context. ​

Prior research has suggested gender differences in negotiation outcomes, particularly when parties are advocating for themselves in a fixed pie (distributive negotiations) context (Walters, Stuhlmacher, & Meyer, 1998; Stuhlmacher & Walters, 1999; Mazei, Hüffmeier, Freund, Stuhlmacher, Bilke, & Hertel, 2015). On average, male negotiators walk away with significantly more value than female negotiators. What is left unanswered is how much of any gender difference is due to party’s own behavior (men and women making different opening offers and conceding at different rates) and how much is due to the other party’s reaction towards different genders (men and women making different opening offers and conceding at different rates depending of the gender of their counterpart). We have designed a 2 (true gender) X 2 (assigned gender) virtual reality study whereby we can disentangle processes and outcomes related to “true” gender from those related to “assigned” gender. We will have essentially 4 conditions: 1) true and assigned gender male, 2) true and assigned gender female, 3) true gender male and assigned gender female, and 4) true gender female and assigned gender male. This allows us to disentangle supply side effects (own behavior) from demand side effects (expectations about and reactions to the other party because of gender), and to see how much each might contribute to negotiation processes and outcomes.

This paper examines how social acceptability of xenophobic views affects willingness to publicly express these views, and how normative influence varies conditional on gender. As a proxy for open support of anti-immigrant views we use donations to either an anti-immigrant or pro-immigrant organization. Scholars have repeatedly established the existence of a social norm against the public expression of hate, which makes the expression of prejudice more likely in a private than in a public context. This anti-hate norm prevents people from openly expressing xenophobic attitudes. Based on literature on social norms of prejudice expression, we hypothesize that increasing the perceived social acceptability of prejudice should increase its public expression. Moreover, we test whether the extent of normative influence depends on the gender of the individual.

To this end, we designed an online experiment (N=2283) in which participants were invited to participate in an online forum discussing refugees and immigration issues. We manipulate the social acceptability of expressing prejudice by increasing the proportion of comments considered hateful - violations of the anti-hate norm. In the treated conditions the number of comments considered hateful increased with each consecutive forum page, whereas in the not-treated condition there are no hateful messages and the tone of the comments remains stable across forum pages. Participants are given the possibility to make a donation of 1 euro to a randomly drawn organization that could either be anti- or pro-immigration. Both organizations were selected by means of a pre-experimental online survey (N=200) in which we selected the best-known organizations in Germany: AfD, and ProAsyl. The donation decision was randomized in a manner so different participants were asked to make the decision at different stages of the forum. The treatment conditions thus vary along three dimensions: i) the type of organization, ii) the number of comments the participant sees in the online forum before the decision, and iii) the fraction of those comments that are hateful. Across systematic variations thereof, we measure how the proportion of norm violations of the anti-hate norm influences the decision to donate.

The empirical results show that overall people are more willing to donate to the pro-immigration organization than to the anti-immigration, and that women are particularly reluctant to donate to anti-immigrant organization. Results also show that women reduced even more their donations to this organization when the anti-immigrant comments raised normative concerns. We explain this result as women displaying greater social desirability bias and more willingness to follow the social norm against prejudice. This paper is part of a growing literature on social norms, prejudice expression, peer effects, and support for xenophobic attitudes. Results in this paper can help explain how changes in the normativity of openly expressing xenophobic views can impact the success of right wing populist parties, and how the anti-hate norm could potentially prevent them to gain further support.

When responding to survey questions, participants use cognitive processes that help them find the answers: 1, comprehension: they focus on a question and interpret it, 2, retrieval: they search for relevant beliefs in their memory, 3, judgement: they integrate them to final opinion, 4: response: they adjust the opinion according to potential response categories (Tourangeau et al. 2000). There is an ongoing discussion, which type of scale helps the participants more in their effort when answering survey questions. On the one hand, full labeling of response scales provides more information how to interpret the scales, which means lesser cognitive load and more precise answers (Arce-Ferrer 2006; Johnson et al. 2005; Weng 2004). On the other hand, end labeling of response scales proves to be easier to remember and does not suffer from language ambiguity (Krosnick – Fabrigar 1997).

In our experiment, we focused on a question whether different types of scales provide different type of answers and results. We had designed split-ballot experiment that we integrated as a part of International Social Survey Programme 2018 module „Religion“ in Slovakia. Participants (N=1470) were randomly divided into two groups and as a part of a survey they both answered the question „In the last five years, how often have you or a member of your immediate family come across a public official who hinted they wanted, or asked for, a bribe or favour in return for a service?“.

One group of participants chose from five-point full labeled scale (1 = never, 2 = seldom, 3 = occasionally, 4 = quite often, 5 = very often), the other one was asked to choose from five-point scale that was labeled only at the ends (1 = never, 5 = very often). Our results showed, that there were no significant differences between these two groups.

This conclusion it is in contrast previous research, which suggests that end labeling invokes extreme response style, meaning that participants are prone to choose extreme endpoints on rating scale (Moors et al. 2014). However, our experiment results indicate that when asked about their experience, in this case with corruption, the type of scale does not influence the answers of participants.

The decision to cheat involves intertemporal decision making as well as an assessment of the risk about being discovered. We run a lab experiment to study the connection between time preferences and cheating at the individual level, controlling for individuals' risk attitudes; our hypothesis is that the willingness to cheat is higher among individuals who attribute more importance to the present. Our experiment, designed to preserve anonymity, also allows us to record socio-demographic details and information on logic ability, overconfidence and a number of psychological traits such as altruism and self-control. We observe widespread cheating, and robust evidence of a negative correlation between cheating and time discounting. Cheating also turns out to be positively correlated with over-confidence.

Political rent-seeking is the act of diverting resources away from a productive activity towards cultivating connections with corrupt political agents. Yet it is difficult to observe. We create and conduct a novel rent-seeking laboratory game which captures real political dynamics of the behavior with actual political leaders in Mozambique and their citizens. We find that contrary to the one-shot subgame-perfect Nash equilibrium prediction, both citizens and leaders are willing to engage in and reward rent-seeking activity respectively. The result is a transfer of resources from poorer citizens to wealthier leaders. Critically these deviations are strongly correlated with real world behavior. Citizens engage in significantly more rent-seeking with leaders who are actually observed appropriating community money. Beyond this we find additional relationships with rent-seeking and leaders' willingness to engage in corrupt activities.

The spread of false information (e.g. fake news, gossip, secrets) in social media has caught a great deal of attention by both academic research and popular news. Importantly, it has been shown that most people do not share misinformation to hurt others but to gain personal benefits that reinforce their well-being, status or political beliefs. In other words, misinformation has a self-perpetuating nature, which sparked interest in identifying measures that reduce its dissemination. The dominant measures proposed are linked to the use of verification. The underlying assumption is that if information is easy to identify as false, most individuals will not disseminate it. We test this assumption through a controlled online experiment. We vary how individuals verify the truthfulness of the information they receive in three conditions: no verification, exogenous verification (individuals are forced to verify) and endogenous (individuals can choose to verify or not). Our results show the conditions under which verification is e↵ective in preventing the spread of lies in social networks. In particular, we find a positive but moderate e↵ect of verification. It reduces lying but only to a certain extent. Moreover, the strongest e↵ect comes from endogenous verification, where individuals choose whether to observe information that may go against their material interest. Finally, the role of verification influences a shift in who is responsible for lying. While in networks without verification the liar can be in any position, in those with verification lying is shifted to the final players.

The recent literature of social preferences for inequality and redistribution has highlighted a concrete challenge: shifting attitudes towards redistributive policies is inherently hard. Survey experiments with informational and emotional treatments proved effective to improve knowledge and remove biases about specific issues, and, moreover, to increase concern for inequality in general. However, they largely failed to increase preferences for specific socio-economic policies aiming at reducing inequality, especially those involving governmental interventions. Often, the effect is much reduced, or even reversed, for republican respondents, who might even become more polarized on the issue. Two further limitations on the literature on social preferences for redistribution are found: (i) little is known about the effect of informational and emotional treatments on behavior (rather than just preference), and (ii) there is an almost complete neglect for network effects. In fact, social networks are the foci of human political decision making and the preferential tool to achieve behavioral change—or opinion change with a stake—via complex contagion processes. Therefore, networks appear as the natural settings for an experiment aiming at shifting preferences for socio-economic outcomes. In this study, we address the question “Can we study preferences for redistributive policies in network settings and make a targeted network intervention to shift these for left-wing and right-wing individuals alike?” We designed a network-based online experiments using the nodeGame platform; the experiment is currently being implemented and we expect to perform the whole data collection in March 2019 and to present the first results at the IMEBESS conference 2019.

Previous literature demonstrates that people’s beliefs about the determinants of income inequality play a major role in the decision whether to support income redistribution. However, not much is known about how people form these beliefs and the degree to which self-serving biases influence their formation. This study investigates whether people form self-serving beliefs regarding the extent to which work versus luck determined their income, and particularly whether such beliefs are driven by the desire to support personally advantageous redistributive policies. To address this question, I conduct a laboratory experiment that exogenously varies participants’ incentive to distort their beliefs. I find that participants attribute income inequality to work when they are rich, and to luck when poor, but the financial incentive of benefitting from an advantageous redistributive policy does not make this effect any stronger. I also measure preferences over redistributive policies, and confirm that these are influenced by the elicited beliefs.

Questions of distributive justice are not only important for the understanding of various political outcomes (e.g. tax laws) and the development of generally accepted social policies but are also core for our understanding of human nature. The main aim of the contribution is to find out who relies on which principles of justice (equality, merit and need) when and why? In order to address problems of social desirability bias and to measure people’s justice attitudes in a direct manner, a modified version of a survey experiment was developed. In this modified design, respondents are asked to distribute a specified amount of money among three people described in vignettes (including indicators for need and merit, as well as information such as sex and ethnic background). This methodological approach combines the possibilities of distribution tasks in laboratory settings with the interdependency and visual presentation of a choice experiment and the convenient metric outcomes of factorial surveys. It also allows us to consider the complex interplay of individual-level, contextual and situational factors in the formation of justice attitudes. Since the experiment is embedded in a PAPI or online questionnaire, the results of the survey experiment can be analysed in combination with other questions and socio-demographic variables. Another advantage of the method is that the situation can easily be adapted, in this case to include distributions among family members, friends, students applying for scholarships and at the workplace. Furthermore, the amount to be distributed was varied in order to test if it matters for the respondents; the assumption being, that a lower amount will force participants to focus on one principle, while a higher amount will allow them to use a more fine-grained distributional rule. For external validity and as a means of testing the influences of context, the survey was distributed among a random sample of the Swiss general population, as well as to two student populations at the University of Bern, Switzerland and the University of Princeton, USA. Preliminary results show: (1) The situation for which respondents are asked to distribute resources substantially shapes the choice of the primary justice principle applied to the task. For example, respondents were approximately three times as likely to apply the equality principle in the family setting compared to any other situation. (2) Context matters! The geographical region in which the respondents were socialised affects the way they perceive distributional justice. (3) Individual level factors such as sex, age and class background are also important predictors of attitudes towards distributional justice. The results are mainly in line with self-interest motives and reduction of cognitive dissonance mechanisms (e.g. upper class people and men are more in favour of the merit principle than lower class people and women). (4) What is more, vignette people were treated unequally based on their ethnicity. (5) Further analyses are expected to provide additional insight and help uncover the underlying mechanisms. We can then fit the existing pieces of the puzzle, add to it and achieve a broader picture and generalizability.

One of the main questions of distributive justice concerns how we judge inequality in returns from labor. How unjust an inequality is perceived inter alia depends on the factors that generated the inequality. This project investigates a situation where people can choose both, how much effort they are willing to provide and whether they want a safe rate of return to their effort or a risky one. In such a setting, people can be held accountable for their choices and following the accountability principle, differences in earnings created by different decisions should not be offset. Using a third party spectator design, I investigate whether people do indeed follow the accountability principle or whether they are influenced in their allocation decision by the level of effort exerted by the risk-taker such that if the risk-taker put high effort and won, the resulting inequality might be perceived as more just than if the risk taker put low effort and won. In the same vein, if the risk-taker put high effort and was unlucky the resulting inequality might be perceived as less just than if the risk-taker put low effort and was unlucky.

Increasing active transportation is both a health and environmental policy priority in many European countries. Transport choices have a large impact on our life and on the environment, hence behaviour change in this direction can provide large societal benefits. Understanding which incentive works best is extremely important since transport behaviour is heavily habit dependent. A vast body of research can be found on the psychological determinants of mode choice and on the effectiveness of economic instruments. However, there is a lack of controlled studies combining the two literatures and that can disentangle how different measures work in the field.

This study tests people’s responsiveness to different incentive schemes for cycling using an innovative app tracking technology. In particular, we run a Randomized Controlled Trial (RCT) in the field to test whether monetary incentives can increase people’s cycling frequency and distance. Tax exemption or payment schemes are already used to stimulate cycling in some countries such as the Nederland, France and Belgium. In Norway, the socialist party suggested in 2015 to implement measures where cyclist and pedestrian can deduct part of their taxes based on how many Km they cycle or walk. The suggestion is appealing, but it presents challenges on how to document activity and for the national budget.

Insights from behavioural economics paired up with new mobile app sensing technologies such as Sense.Dat can give us the right tools to approach the problem. From Kahneman and Tversky (1979) seminal paper on prospect theory, we learn that people tend to overweight small probabilities and that they are attracted to lotteries with low probabilities and high rewards. Hence, we hypothesize that a well-designed lottery could work as well as, or better than a small incentive for all. Another advantage of the lottery is that it would be a much cheaper measure than tax deduction.

When implemented correctly, RCTs are the most rigorous way of determining whether a cause-effect relation exists between treatment and outcome. This experiment provides a causal relationship between the type of incentive and amount of cycling. In particular, we compare the effect of a riskless and flat rate incentive for all, with the possibility for few to win a large sum of money through a lottery. Moreover, we draw from recent literature and include Regret Aversion in one treatment.

Results are in line with the hypotheses. Compared to a control group, payment of a flat rate to everyone and the possibility for few of winning a lottery resulted in a significantly increase in cycling activity (measured in both Km cycled and amount of days with non-zero cycling activity). Our results show that the strongest effect is found in the “pay per km” treatment, where we register 45% km increase, and in the Treatment with regret aversion with 36% increase.

We present a longitudinal online experiment (n=1,240) testing the effect of differently framed Implementation Intention (II) interventions on vegetarian food choices. Participants are randomly allocated to either a control group, or one of three different II interventions where the benefits of a vegetarian diet are presented in terms of either personal wellbeing, or planetary wellbeing, or both personal and planetary wellbeing. Participants make choices between vegetarian and non-vegetarian food items in an incentive-compatible task. After the food choice task, they have also the opportunity to donate money to a charity. Finally, participants answer a follow-up survey three days after the food task. We find that, compared to the control group, participants are twice as likely to choose a vegetarian option when exposed to any of the three II interventions. There are no statistically significant differences in the proportions of vegetarian options across the three II conditions. However, in the follow-up survey participants in the II intervention where the benefits of a vegetarian diet are framed in terms of both personal and planetary wellbeing report consuming vegetarian food more frequently than in the control group. Moreover, we do not find any evidence of negative spillover effects on subsequent charitable donations. If anything, participants in the joint personal and planetary condition are more likely to donate to a charity following the food choice task, indicating possible positive spillover effects. Taken together, these results point to the promise of combining the information about the personal and planetary benefits of a vegetarian diet when designing behavioural interventions to nudge diet change.

Obesity and overweight have become a global epidemic. As a result, a wide variety of preventive measures have been implemented around the world. However, effecting changes toward healthier eating and lifestyle habits is known to be extremely difficult, especially in adulthood (Hill, 2009). For this reason, and because health status acquired during childhood has a major influence on the patterns of adult life, preventing obesity and overweight among children is a critical task.

Most interventions directed to children examine the impact of educational programs and obtain mixed results. Recent evidence indicates that results may depend on how the educational message is framed. Most of the prevention programs in health have a fear-based approach (focusing on health risks if the subject does not adopt the recommended behavior). But it is not clear whether this is the most effective approach, especially when dealing with children. Therefore, it is important to examine the influence of positive versus negative messages on children's snack choices. In this paper, we design a randomized control trial experiment that allows us to identify the impact of visual information (positive and negative) about snack quality on children's immediate snack-choices.

The experiment was carried out in October 2018 at seven elementary schools in Pamplona (Spain), involving 258 students in 4th grade (8-9 years old). Selected schools captured a wide range of the socioeconomic diversity of elementary-age children in Pamplona. Randomization was performed at the individual level. We included two treatment groups (in the positive treatment, kids were exposed to photos of a smiling emoticon surrounded by fruits; in the negative, kids were exposed to photos of an angry emoticon surrounded by unhealthy snacks), and one control group (without any visual message). Along with the visual information, kids had to choose, in private, between two snack boxes: one healthy (with fresh fruit) and another unhealthy (with cookies and similar), both of them providing the same amount of calories intake.

Data analysis shows that the average treatment effect of visual messages (either negative or positive) is not significantly different from the lack of messages (control group). However, when gender is considered, the average treatment effect of positive visual messages is significantly different from the control group, encouraging the choice of the healthy snack for girls (p<0.05), but not for boys. Negative visual information about unhealthy snacks has no effect neither on boys nor girls. Our results are robust to school characteristics, nature of snacks brought the previous day of the experiment and estimation method.

Our results suggest that fear-based messages do not work to promote healthy choices among children. We also show that girls are more receptive to positive educational messages than boys. Public health interventions directed to children in the area of nutrition should take these results into account.

Customers of Swiss electricity providers had a choice between „green“ and „grey“ energy. In contrast to grey energy, green energy is neither generated by fossil fuel nor by nuclear power. However, customers had to pay a slightly higher price for ecofriendly green power than they had to pay for grey power. The price difference is about 4 % for households and 5 % for businesses (0.01 of 0.25 Swiss Francs per kWh for households and 0.01 of 0.18 for businesses). There is no difference in the private utility of green or grey electricity. Hence, choosing green energy is a contribution to the collective good of a better environment with less risky nuclear energy. Households and businesses alike face a social dilemma with the alternatives of cooperation (green energy) or defection (grey energy) whereby the choice of grey energy is a dominating strategy in monetary terms. By this logic it is clearly understandable that the overwhelming majority opted for grey energy.

But there is another psychological factor to be taken into account. In former years grey electricity was the reference category. In previous years several utilities switched the reference category for ordering electricity from grey energy to green energy. Did the change in the default value have an impact on the proportion of customers buying green energy? Many studies report relatively large effects of the default category on behavior. But does the effect persist? And, even more interesting, is the default effect observable for businesses as well as for private households? Economists would expect that businesses act more in accordance with the Homo oeconomicus model than private households. Also, one would expect that customers‘ willingness to order green energy will decline with the amount of electrical energy consumption.

We explored default effects of energy consumption using data from two electrical suppliers. Both companies recorded the energy consumption before and after the intervention, i.e. the switch of the default category from grey to green energy. The novel aspects of our investigation are that we were able to analyze a large data set of more than 250‘000 customers and that the sample included businesses as well as private households. Moreover, we observed the demand for green energy for several years before and after the intervention. These data made it possible to explore the persistency or fading out of default effects. Simple comparison of differences as well as more refined econometric analysis clearly showed that there is even a surprisingly large and persistent effect of a change in the reference category for businesses.

The volunteer’s dilemma (VOD) is a step-level collective goods game in which the contribution of one individual is necessary and sufficient to provide a benefit for the entire group. Doing the dishes after a flat-share dinner, calling the police if someone disturbs everybody’s sleep, vetoing an unpopular motion in a committee meeting, are all situations that can be conceived as VODs.

We investigate how small groups tacitly coordinate in the VOD and thereby develop latent norms. Latent norms are behavioral regularities that emerge over time when members of a group react to each other’s actions. There are two types of latent norms that are often observed in the repeated VOD: turn-taking, by which group members take turns in providing the public good, and single volunteering, by which the same group member provides the public good repeatedly while others free-ride. Both latent norms can lead to collective efficiency, but turn-taking leads to equal payoffs between group members while single volunteering leads to unequal payoffs

In two experiments, we address the question whether the latent norms we observe are determined by group members’ other-regarding preferences, or whether they are a result of the structural properties of the VOD. To answer this question, we measured subjects’ other-regarding preferences and varied the asymmetry of the VOD experimentally. We manipulated two types of asymmetry: payoff asymmetry and focality. In a VOD with payoff asymmetry, one of the group members has lower (or higher) costs of producing the public good. Focality is established by singling out one group member with an arbitrary visual cue. Both types of asymmetries might help groups to coordinate on the latent norm of single volunteering, by letting the odd one out volunteer repeatedly.

In our first experiment, we test the hypothesis that other-regarding preferences promote tacit coordination on turn-taking in the symmetric VOD (as this leads to equal earnings over time), but hamper coordination on single volunteering by the member with the lowest costs in the asymmetric VOD. In our second experiment we test the hypothesis that singling out one group member by an arbitrary visual cue is sufficient to bring about single volunteering by the focal group member, even if the VOD payoffs are symmetric.

We find that the structural properties of the VOD have a larger bearing on what type of latent norm emerges than group members’ other-regarding preferences. In the symmetric VOD, the large majority of groups develop a latent norm of turn-taking. However, the relation between asymmetry and the latent norms that emerge is not clear-cut. If there is one group member with lower costs, single volunteering by this member is coordinated on by most groups. However, single volunteering rarely develops if there is one member with higher costs. Likewise, when payoffs are symmetric, but there is one focal player, the variation in types of latent norms that emerge is larger, and few groups coordinate on single volunteering.

Conventions are customary and arbitrary rules of behaviour that coordinate our interactions with others if all parties share an interest to conform (Lewis 1969). When conformity runs counter to individual self-interest, interactions may be regulated by social norms, which are endowed with prescriptive pressure and supported by sanctions (Cialdini et al 1990, Horne, 2001, Bicchieri 2006). Conventions and social norms have been kept distinct along these lines. However, both can emerge tacitly, as unintended consequences of individual actions (Sugden 1986, Young 2015, Centola and Baronchelli 2015), and what originates as a mere convention may, over time, acquire a prescriptive force and turn into a social norm (Opp 2004, Tummolini et al 2013). Although there have been first attempts (Guala 2013, Diekmann and Przepiorka 2016), no study so far has systematically measured the normativity of emerging conventions.

Here we induce and measure the emergence of social norms from conventions in the repeated volunteer’s dilemma (Diekmann 1985, 1993). The volunteer’s dilemma (VOD) is a binary choice, n-person game in which a single actor’s cooperation is necessary and sufficient to provide the collective good for the entire group. Previous research has shown that an ‘egalitarian’ convention, in which each person sequentially incurs the cost of volunteering (turn taking), can emerge in the symmetric VOD, in which the cost of volunteering is the same for everyone. In an asymmetric VOD, in which one person has a lower cost of volunteering, an ‘exploitative’ regularity often emerges; the person with the lower cost takes a disproportionately large share of volunteering (strongest-always-volunteer) (Diekmann & Przepiorka, 2016).

Building on these results, we conduct a two-part experiment in which subjects start in either the symmetric or asymmetric VOD and then move to the other VOD in the second part. Drawing on the work of Bicchieri and colleagues (Bicchieri, Lindemans, & Jiang, 2014), we employ a set of incentivised measures to capture when conventions turn into social norms. Our design allows us to explore (1) whether conventions such as turn-taking or the strongest-always-volunteer turn into social norms and (2) whether these social norms make the prescribed behaviour ‘sticky’ in face of structural changes (i.e. when subjects move from the symmetric to the asymmetric VOD or vice versa). We also measure social approval providing us with additional information on the normativity of the conventions that emerge.

Our results show that groups with a higher consensus on which convention should be followed, turn-taking or strongest-always-volunteers, take longer to change their behaviour when they move to the VOD with an alternative payoff structure. However, the exploitative norm is less stable than the egalitarian one. Although the stability of the egalitarian norm is also affected by changes in incentives, it is more likely to persist in an environment that would be otherwise conducive to unequal division and exploitation. Our analysis of approval scores reveals that once established even exploitative norms are perceived as legitimate, suggesting that normative judgments can swiftly shift in support of the social order that emerges due to changes in incentives.

Loss aversion is widely studied in the monetary domain as well as with respect to material goods. However, little is known about its relevance in the non-material domain. We conduct a laboratory experiment to explore whether the concept of loss aversion applies to social image concerns, a non-material good. Social image concerns are important in an individual decision-making since people typically care about their reputation. We analyze whether an exogenous improvement or harm to one's reputation follows the same pattern as gaining and losing money or material goods. Our experimental design attempts to quantify the effect of loss aversion in social image concerns via the scope of lying. We aim at fixing a within-subject reference point in reputation, then inducing an exogenous reputational gain or loss and observing how much subject lie in a potentially image-improving task.

We employ a laboratory experiment to study the relative relevance of social value orientations and group norms for the recognition of others’ needs. According to the Model of Frame Selection the definition of the situation is as important as the mode of information processing and variation in social preferences in the explanation of human behavior. Goal Frame Theory, in turn, systematizes possible frames into three main categories of goals, that is, the hedonic, the gain, and the normative frame. In this paper we ask whether goal frames affect the recognition of others’ needs in a distribution game. In the first stage of the experiment, we measure and elicit distributive preferences as a result of induced goal frames, aggregated into a signal of shared group preference. Then, subjects play a distribution game in three-line networks in which they negotiate in dyads about the allocation of points to group members. If a dyad agrees on a distribution of the points, this distribution is implemented. In order to reach the final stage of the period, in which further points can be earned by individually accomplishing tasks, subjects need to satisfy a threshold of some minimum number of points, which varies across individuals. We expect the satisfaction of needs thresholds of “third players” who are not part of the agreeing dyads to vary with the preferences that are induced by the goal frames. Without any signal of a prevailing group norm, we expect the satisfaction of own needs to be the dominant motive of action, which also depends on social value orientations. After a signal suggesting the prevalence of egoistic preferences in the group, we expect the satisfaction of the third players’ needs to decline, whereas we expect it to rise after a signal of the prevalence of a norm of need-based justice. Inter alia, we find that the display of a signal indicating the prevalence of a norm promoting a need-based distribution enhances the probability of need satisfaction, compared to situations where no signal or an egoistic signal were displayed.

Previous experiments address the intergroup bias in decision making, measuring differences between decisions towards in-group and out-group agents. However, we can find contexts in which the group identity of some agents is unknown. This makes it important to study decision making towards this third set of existent agents whose group identity is not visible. Through studying behavior in a coordination game that includes interaction decision making, this experiment addresses the intergroup bias on interaction preferences, including agents whose group identity group remains unknown, and its dependence on the group identity sense saliency. Results show that in the weak group identity sense environment, in-group favouritism and out-group derogation discrimination patterns emerge. Agents exhibit preferences for interacting with other agents that belong to the same group identity group, and provide middle and low effort levels in the coordination context. Furthermore, also in the weak group identity sense environment, a negative discrimination for not interacting with agents whose group identity is not known emerge, and affects not only agents that provide middle and low effort levels in the coordination context, but it even affects agents that provide high effort levels in the coordination context. These positive and negative discrimination patterns in the weak group identity senses environment almost cancel out each other and make the total welfare reached in the coordination context to be only slightly lower than in the environment in which economic agents had not been instilled any group identity sense. We also found that when the instilled group identity sense is enhanced using a group-solving task, agents’ initial in-group favouritism on interaction preferences is similar than in the environment of weak group identity sense. However, initial willingness to coordinate at high effort levels in the coordination context is much higher in the strong group identity sense environment. We show that the higher initial willingness to coordinate at high effort levels in the strong group identity sense environment makes the initial in-group favouritism on interaction preferences to disappear, and leads to higher frequencies of interactions between own, other and unknown group identity agents. As a consequence of these higher interaction frequencies, and the higher willingness to coordinate at high effort levels in the coordination game, total welfare levels reached in the strong group identity sense environment more than doubling total welfare levels reached in both, weak group identity sense environment and the environment without group identities.

The voluminous literature on polarization and sorting among members of the mass public in the U.S. raises questions about why individuals choose to identify with the Republican or Democratic parties. What do individuals who identify with the Democratic (Republican) party say they like about that party, and what they dislike about the Republican (Democratic) party? How, if at all, have these explanations evolved over recent decades? This paper uses large-scale text analysis of open-ended survey responses from 1984 to 2016 to answer this question.

Humans often favor members of their own group over members of outgroups, a preference that drives prejudices, discrimination and intergroup conflicts. Across cultures and using various behavioral and attitude measures, such ingroup bias has been documented. However, whether the level of ingroup bias is comparable across different natural groups and whether it interacts with the type of outgroup one faces are largely unknown. Here, we present field evidence of ingroup bias in antisocial group decision-making from an economic experiment with 192 male members of three natural groups in Ethiopia. Natural variance in intergroup relations (enmity versus neutral) and the strength of group identity (strong versus weak) allows us to test for group-level drivers of antisocial behavior. In our sample, lack of outgroup concerns is universal: Almost all subjects chose the antisocial option when targeting an outgroup member, irrespective of friendly or hostile intergroup relations. Members of the two groups with strong ethnic identities exhibited less antisocial behavior towards their own group members. There is substantial individual heterogeneity in antisocial choices; about half of our subjects never chose the antisocial option when the target was an ingroup member, and about one fourth of subjects were antisocial regardless of the group composition. A simple model of ingroup bias organizes the choices of more than 90% of subjects. Our results imply that previous work using Western samples grossly underestimates the extent of antisocial behavior targeting outgroup members.

Recent waves of migration to Europe and North America have fueled intense debates and public anxieties over the potentially adverse economic, social and cultural impacts of immigration. In this context, migrants' socioeconomic status plays a critical role in shaping reactions to new arrivals, with poorly-educated, low-skilled immigrants consistently eliciting greater opposition on measures of explicit attitudes. Little is known however about the effect of immigrants' socioeconomic status on exclusionary reactions rooted in implicit outgroup biases. We report findings from a field experiment examining implicit bias against high- and low-status immigrants in Milan, Italy. We measure natives' implicit biases as manifested in aversive reactions to sharing a personal space with immigrants in everyday encounters. Results from 831 trials demonstrate that immigrants are physically avoided regardless of their socioeconomic status. Further exploratory analysis reveals this effect to be driven by native women avoiding contact with immigrant men. We interpret these results within a dual process framework which separates implicit negative reactions towards immigrants from explicit attitudes based upon instrumental comparisons of the potential impact of high- vs. low-skilled immigration.

We assess the motive of indirect reciprocity in a gift exchange experiment. In particular, we test if individuals interpret someone’s behavior towards a third actor as a signal about how they would be treated and consequently reward or punish it. Alternatively, indirect reciprocity can be interpreted as a notion of altruism. In our experiment one employer is matched with two employees to whom she can pay different wages. In the first treatment the employees solely know their own wage. Here, a positive correlation between wages and effort indicates direct reciprocity. In a second and third treatment, employees are informed about both wages or solely about their coworker’s wage. We find evidence for indirect reciprocity in terms of a significant positive correlation between the coworker’s wage and effort in the third treatment. This correlation becomes insignificant and negative when introducing one’s own wage. We therefore conclude, indirect reciprocity rather being a self-referred reaction than an act of altruism.

Stylized facts of financial markets, such as excess volatility and bubbles and crashes, have drawn a lot of attention in the last decade. The question as to what explains these bubbles has spurred considerable debate between economists. In a series of so-called Learning to Forecast laboratory experiments, it has been shown that -- even in a very stylized setting -- the self-fulfilling nature of financial market expectations (that is, if traders expect stock prices to increase, and increase their demand of the stock based upon that belief, stock prices will indeed increase) may cause bubbles and crashes in asset prices in the laboratory (see e.g. Hommes, Sonnemans, Tuinstra and van de Velden, 2005, 2008, and Heemeijer, Hommes, Sonnemans and Tuinstra, 2009). However, these laboratory markets are inhabited by traders that have short run horizons: they are only interested in next period's asset price. On actual financial markets some investors may have a longer investment horizon, and this may decrease the incidence of bubbles and crashes in asset prices. We present results from a Learning to Forecast laboratory experiment designed to study the effects of increasing the forecasting horizon on asset market price volatility. Two competing effects may emerge as the forecasting horizon increases: (i) prices stabilize as non-fundamental predictions affect dynamics to a lesser extent when the forecasting horizon increases and (ii) the more cognitively demanding task to predict the price for more distant future periods destabilizes price dynamics. Participants to the experiment have to repeatedly predict the price of the asset, where their forecast for future periods determines the current market price, and we pay them for the accuracy of their predictions. We vary the initial history of prices from stable to unstable to investigate how the increasing forecasting horizon affects dynamics under different conditions. We find that increasing the forecasting horizon in markets with stable histories stabilizes dynamics. On the contrary, in markets with an unstable price history, an increase in forecasting horizon increases instability, although, the effect is diminishing. This can partly be explained by the (partial) breakdown of coordination of individual expectations.

The questions of whom to trust and whether to cooperate with another person are fundamental questions of social life. Experimental studies of these questions are legion and often use the one-shot Prisoner’s Dilemma as the paradigm choice situation. In the social network literature, homophily and reciprocity are fundamental forces that shape tie formation, as ties often occur disproportionately between individuals who are similar to each other on significant dimensions of social differentiation and also occur with greater than chance regularity when the tie for one person to another matches an incoming tie from the second to the first. Studies that demonstrate the importance of homophily and reciprocity are also legion.

Our research question explores the integration of the basic tie formation forces with the questions of trust and cooperation: that is, to what extent and under what circumstances are homophily and reciprocity conducive to trust and cooperation with others. The research is motivated in part by a desire to better understand two mechanisms, attraction and repulsion, that could produce homophily. In the first mechanism, individuals form social ties with others based on “attraction” to those who are similar to them along important social dimensions. In the second mechanism, the driving force in tie formation is “repulsion” from those who differ on these important social dimensions. In the first case, the overrepresentation of ties between persons of similar background is a direct consequence of an “inbreeding” bias while in the second case, it is an indirect consequence of a “rejection” bias directed at dissimilar others. Specifically, we hypothesize that the visibility of the distinguishing marker of group identity impacts an individual’s decision to form ties of trust and cooperation with others. Our experiments seek to evaluate this claim. At the same time our experiments enable us to evaluate the relative strength of homophily versus reciprocity in sustaining trust and cooperation when the two forces are at odds and the question is whether to reciprocate trust with an individual from the other identity group.

Participants in our experiment were given a series of exchange opportunities. The design of our experiment allows us to treat each subject-alter dyad in terms of a Prisoner’s Dilemma. However, the decision to cooperate or defect in the game is driven not only by the subject’s expectations about how likely the alter is to reciprocate, but also by the subject’s expectations about how likely other participants in the experiment are to share their resources with the subject. That is, the decision to cooperate or defect is shaped by expectations of both direct and generalized reciprocity, and the latter are a function of whether the other players are members of the subject’s in-group.

Understanding a complex social system requires an accurate model of the behavior of the micro-entities, as even seemingly innocent differences between micro models can generate markedly different macro dynamics. While the theoretical understanding of complex social systems has increased tremendously in the past decade, the field lacks empirical research testing competing micro-models and their macro implications. As a consequence, theoretical models are based on various competing micro-models - such as myopic best-response, reinforcement learning, and imitation - that have not been validated.

Here, we report results of a pilot study informing a bigger laboratory experiment on a simple coordination problem in a network of 15 to 20 nodes. We experimentally manipulated which information was provided to subjects in order to allow them to or prevent them from using alternative decision rules. We tested which micro-model best described observed decisions depending on the provided information, using both the micro-data and the macro-patterns that we observe during the experiment.

In many social situations, economic agents are concerned about their relative standing compared with others. One typical example is conspicuous consumption, whereby individuals compete for relative position in a society by spending on visible, status-signaling goods (e.g., expensive cars, gifts, branded haute couture). However, positional concerns extend beyond conspicuous consumption to any social situation in which rewards are allocated based on relative positions. These include all-pay auctions, an arms race between countries, rank-order tournaments and investment in rent-seeking.

One common aspect of the literature on positional concerns is that both the theoretical and the empirical works assume that individuals make comparisons relative to everybody else in their society or in a larger group to which they belong (e.g., village, neighborhood, age, or income group). However, a growing body of economic literature on social networks has demonstrated that individuals interact with a smaller set of social contacts, and the structure of social connections has a significant impact on the individual outcomes. In this regard, a theoretical contribution by Ghiglino and Goyal (2010) found that individuals’ consumption of positional goods is determined by the Katz-Bonacich centrality of their position in the social network. Moreover, the overall network structure, especially the number of links, also affects individual choices.

We designed an experimental game based on the theoretical contribution of Ghiglino and Goyal (2010) in which individuals are embedded in a network of four individuals, and allocated a fixed endowment between a private and a positional good for 30 consecutive rounds. Our study is centered around four main research questions.

The first question studied whether individuals exhibit positional concerns in social networks or choose more efficient allocations. The second research question aimes to uncover the relationship between the centrality of an individual’s network position and the individual’s consumption choices and utility. The third question studies the impact of the overall network structure, in particular, the number of links, on the consumption choices and welfare. The aim of the fourth question is to understand the learning process in the game. In particular, if the game play converges to the Nash equilibrium, how do the individuals learn to play the Nash equilibrium given that the game is relatively complex?

Tax evasion is an illegal concealment of a taxable activity and a worrisome phenomenon. Understanding how to motivate individuals to pay their taxes has become one of the main targets for policy makers. From the standard economic literature we know that a taxpayer makes a decision under risk, maximizing his expected utility. In particular, the payment of taxes is a highly structured decision process. Indeed, by choosing whether to comply with the law or to engage in the tax evasion, individuals take into account a set of social norms along with their role in the society. The behavior of other individuals within the same community matters since it increases the moral cost of non-compliance. But how does it affect individual decisions? Given the magnitude of this phenomenon and its economic consequences, a better understanding of its determinants represents an essential starting point for the definition of effective policy interventions. By using an experimental approach, in this paper we study how the type of information provided to our individuals within the same network affects their level of tax compliance. Social information has a particularly strong impact on tax compliance, especially when information flows within an interconnected network of individuals. The type of information subjects are provided with reveals to be pivotal in their decision process, especially when it refers to subjects belonging to their own reference group. This is because individuals attribute a greater value to the behavior of people they know better. To this end, by introducing the same monetary incentives to cheat, in our experiment design we test four different treatments: No Info, Full Info, PositiveInfo and NegativeInfo. In our baseline treatment, NoInfo, individuals are provided with full information on the event to be audited, the relative outcome, and their final payoff. Then, by manipulating the type of information given to individuals on their neighbors’ behavior, we explore the other three types of treatment. Specifically, in the PositiveInfo treatment, subjects receive information on the neighbors they are connected with, by obtaining details on the audit event, and its relative output, i.e. if their neighbors have been found complaint with the tax payment of the requested taxes. In the NegativeInfo treatment, instead, subjects are again informed about their neighbors about the audit event, but they will be told only if their neighbors have been caught non-compliant. Finally, in our last treatment FullInfo, participants receive complete details on the others two 2 neighbors there are connected with, which means subjects will be fully aware of the audit event, and its relative output either when their own neighbors have been complainant, or not.

How do separate cultures arrive at similar category systems across a wide range of semantic domains? Foundational work in cognitive science argues that cross-cultural coherence in category systems emerges because people independently produce the same categories as a result of universal psychological processes. Category coherence across cultures is thus viewed as incompatible with social theories of categorization, which argue that categories are defined contextually through communication, leading to highly divergent (i.e. path dependent) category systems. Here, we present findings from a novel study of real-time category formation to demonstrate that social processes can give rise to divergence and convergence in category systems among experimental micro-cultures, depending on the size of the social network individuals are communicating in. In this study, we used an online web platform to experimentally control the size of people’s social networks as they collaboratively categorized a novel continuum of arbitrary shapes. In dyads (N=2), communication generated highly divergent category systems, but in social networks (N=50), it led independently replicated populations to consistently converge on remarkably similar category systems. In networks, individuals were more likely to produce and adopt labels that were better at coordinating with any random person in the network, leading the same subset of categories to diffuse in separate populations. These results challenge the longstanding view that processes of social construction lead to path dependency, by showing that category coherence across cultures may be attributable to processes of social coordination and diffusion in communication networks.

Group identity has emerged as a key explanatory variable of social exclusion or economic inequality because people prefer to benefit others who are similar to them (in-group favoritism) over those belonging to a di↵erent social group. If group iden- tification were fixed, individuals belonging to disadvantaged groups would be unable to escape them. However, constructivist research has shown that identities are con- structed and can change as the product of human decision making. In this paper I study the extent to which the choice over their group identity can help individuals break free from inecient dynamics. Particularly, I look at group identity choices in a social network setting where interactions with others (both in-group and out-groups) are strategically interdependent. I provide a theoretical model and characterize equi- librium outcomes when group identities can be changed. I also test the role of group identity in di↵erent settings through laboratory experiments.

We analyze the impact of stress on risky choice in a large-scale between-subjects design with N = 196 participants. In particular, the main contribution of our work is that we are able to disentangle errors in decision making from an actual shift in preferences between our experimental treatments STRESS and NO-STRESS. Acute stress is induced using the Trier Social Stress Task for Groups (TSST-G, von Dawans et al. (2011)), risk preferences are elicited using a recent method developed by Andersson et al. (2016). Additionally, controlling for cognitive reflection (CRT; Frederick (2005), Toplack et al (2014)), we analyze whether our results are driven by different levels of cognitive ability as the driver for (increased) observed noise between treatments. Our main results show that there is no shift in risky choice towards more or less risky choices in the stress condition. Thus, we do not find a significant effect of a change in risk preferences between treatments. However, in line with previous literature, we find that, on average, a higher score in the cognitive reflection measure leads to significantly less noise in the decisions in the choice task.

Honesty can be seen as one fundamental buttress for cooperation and for economic welfare. However, market participants do not always act in an honest way. Using a natural field experiment with computer repair shops in Germany we are examining possible effects of seeking a second opinion in this credence goods market. Due to their information asymmetry, the computer mechanics have monetary incentives to defraud their customers. Our data shows evidence for a significant rise of prices when confronting the expert with an incorrect second opinion. This rise is due to an increase in the frequency of overprovision, as well as heightened prices for spare parts and working time once the expert learns the customer has already been to another shop. However, confronting the expert with a correct second opinion increases underprovision behavior, as the experts may form beliefs about the customer, and think that he is tough to serve and therefore reject him. Further the signalled information induces a higher risk for the expert to be caught overcharging or overtreating.

Only recently, the higher order risk preferences prudence and temperance have been shown to predict field behavior, such as saving and portfolio choice among the adult population in the Netherlands and Colombia (Noussair et al. 2014; Ibanez et al. 2018). Additional consequences of higher order risk preferences have been investigated theoretically without an empirical validation to date (e.g. prevention of adverse effects (Eeckhoudt and Gollier 2005), especially in – but not limited to – the health and the environmental domain).

While for risk aversion by now there is considerable evidence that the prevalence is rather comparable to that of adults and that it is not developed during adolescence (Harbaugh et al. 2002; Levin et al. 2007; Sutter et al. 2013), for the higher order risk preferences the evidence regarding their prevalence among adolescents and their development during adolescence is still scarce, despite their importance: It consists of only one study by Heinrich and Shachat (2018) investigating the classification of prudence (external margin), who fail to control for cognitive ability. As cognitive abilities have been positively linked to prudence (Braeban et al., 2018; Noussair et al., 2014), they might be important to consider, especially since the results by Heinrich and Shachat (positive relationship between age and prudence) could potentially also be explained by cognitive ability. For intensities of higher order preferences (internal margin), and temperance in general, no study has investigated the prevalence among adolescents or any age-related changes.

Moreover, predictive power for theoretically connected real-life behavior remains unstudied, with exception of the above mentioned studies.

In this study, we investigate the prevalence, age-related changes therein, and possible influence factors of (higher order) risk attitudes and patience amongst adolescents and study related field behavior of (higher order) risk preferences.

Using the method described in Schneider (2017) to elicit intensities of higher order risk preferences building on the elicitation of simple certainty equivalents that are connected to utility functions using a P-Spline approach, we study the strength of prudence and temperance among roughly 700 adolescents, aged 10 to 20, in schools in Germany. We measure cognitive ability with a matrix test and a symbol-digit correspondence test. Field behavior and demographic information are surveyed in a questionnaire targeted at the environment of adolescents, involving questions on general risk taking, planning, health and environmentally friendly behavior, and financial decision making.

We find patience and (higher order) risk preferences to decrease with age with the exception of prudence, but these findings are mainly driven by an increase in cognitive ability. For prudence, independent of cognitive abilities, we find no age effect. Females exhibit higher values of risk aversion, prudence and temperance. Risk aversion, prudence and impatience are significantly correlated with an index of general risk taking behavior, but not temperance. As predicted by theory, we find saving to be positively correlated with prudence and risky investment to be negatively correlated with temperance. Imrudence and impatience predict the risk of addiction (smartphone and drinking behavior).

Clinical decisions based on complete information are assumed to maximise expected utility. The absence of complete information on the clinical and cost-effectiveness of therapeutics introduces risk and uncertainty in clinical decision making. In this scenario, the optimal treatment pathways regarding patient management may become subjective. Ambiguity in clinical decision making arises when it is not possible to assign a probability to a particular outcome due to limited information. The maximisation of expected utility under ambiguity would, therefore, no longer be assumed to hold. This raises questions regarding clinical decision making under ambiguity. The empirical literature on decision-making under risk and uncertainty has documented ambiguity aversion for moderate likelihood gains and ambiguity seeking for low likelihood loss. In the context of health, however, little is known about how preferences are affected by ambiguity when making decisions across different outcomes. Heterogeneity in attitudes towards ambiguity may explain disparities in clinical decision making, such as treatment allocations, use of diagnostic tests and deviations from clinical guidelines.

We use data from a behavioural economic experiment which aims to assess ambiguity attitudes in two different participant groups across five outcomes of interest. The participant groups include a sample of the UK population as well as a sample of UK clinicians. A laboratory experiment is used to assess ambiguity attitudes for the UK sample following Li et al. (Management Science, 64(7), 2018) with a lab-in-the-field experiment to elicit ambiguity attitudes for UK clinicians. The laboratory and lab-in-the-field experiments both elicit matching probability using Ellsberg’s paradox design. We perform four treatments which differ in the source of uncertainty (i.e., artificial versus realistic) and in the outcome (i.e., monetary versus hypothetical health). We also perform an additional treatment which tests the participants’ choices when provided with information on conformity behaviour among their peers.

Our results extend previous findings in the domain of ambiguity by looking at a broader sample including both the general population and clinicians. The difference between physicians and the general population is particularly relevant in the case of health-related decisions. In particular, we test whether physicians are more rational and more ambiguity seeking than the general population when dealing with health-related decisions. In addition, we link ambiguity attitudes with several socio-demographic characteristics that may explain variation in attitudes towards ambiguity both within and between individuals in the two participant groups. We discuss the implications of early-phase evaluation of new therapeutics in the presence of both ambiguity aversion and ambiguity seeking behaviour for the cost-effectiveness of diagnostics and technologies.

The rise of majority-ethnic nationalism – a trend witnessed across the globe – threatens the material and symbolic status of ethnic minorities who do not share the ethnic traits associated with national identity. A common group threat is generally thought to unite threatened groups under a common banner, strengthening identification and enhancing joint efforts to achieve common goals. However, ethnic minorities do not always respond to group threats in concert. On the contrary, a threatening environment sometimes results in fragmentation and intra-group competition. Through behavioural games in Pune, India, this study aims to shed light on the diverging effects of Hindu nationalist threat on intragroup cooperation among Muslim. By randomly assigning a prime that devalues the Muslim identity according to the Hindu nationalist narrative, as well as assign status divisions within groups of Muslim participants, the study highlights the moderating effect of cross-cutting social cleavages on cooperation among threatened minorities

The increasingly popular concept of polarization is used to describe various social phenomena such as ethnic, political, and income polarization. Scholars study ethnic polarization because they assume that it is linked to conflict. So far this relationship has been investigated relying on cross-country data. Evidence is mixed but suggests that bipolarity of equally-sized ethnic groups (polarization) is the most conductive scenario for conflict. While current research concentrates on the macro level, we argue that the assumed causal link is best studied in the setting of small groups. Consequently, we study the link on the level of school classes and analyze data from the Children of Immigrants Longitudinal Survey in Four European Countries (CILS4EU). Our sample contains around 800 high school classes located in the Netherlands, Germany, England, and Sweden and we explore how the ethnic composition of school classes relates to intergroup conflict. We construct a set of treated - ethnically polarized - and a set of - non-polarized - control units and estimate the average treatment effect on the treated (ATT). Our preliminary findings suggest that there is a causal effect of ethnic polarization on the prevalence of conflict among classmates. The results also suggest that it is ethnic polarization, rather than heterogeneity per se, that is the driving force behind intergroup conflict.

Violent conflicts have negative effects on prosperity and develop- ment. Reconstruction efforts need that qualified labor force is willing to work in highly violent areas. We use a field experiment to inves- tigate the effects of life risk on sorting in the labor market. We offer comparable jobs in low and high conflict areas in Colombia to a pool of job seekers. We find that the applicant rate decreases in 12 percentage points due to life risk. Yet, the drop is similar for male and female job seekers, suggesting that if women are more risk averse than men, the difference is not large enough to have economic impacts. A salary increase helps to increase application rates to high risk jobs but does not close the gap.

The use of partisan heuristics is known to be an important explanatory factor to understand individuals’ political preferences. However, the role of negative party identification on the use of partisan heuristic remains unclear and understudied. Moreover, while some research find that political awareness increases the use of partisan cues, the causal relationship between political awareness and the use of partisan cues remains questionable. In this research we try to investigate the use of partisan heuristic for positive and negative party identification and the influence of political awareness on the use of partisan heuristics. With an experimental design based on direct democratic ballots, this research aims to show the causal relationship between party support, party position and vote intention for direct democratic proposals. Our results suggest that voters might take cues from negative and positive party identification and that political awareness might play a more ambiguous role on the use of partisan heuristics than argued in previous research.

The article investigates how exposure to gang-affiliated peers affect youth’s social behaviors and attitudes. Much of the literature finds that exposure to gangs contributes to youth’s antisocial outcomes. According to other studies, however, it can also promote prosocial behaviors. The present article reexamines this contradictory evidence, exploring potential complementarity of both reactions to gangs. Using survey of 1,944 youths from rural Colombia, I compare adolescents who are and are not in the school class with members of youth gangs. I exploit the fact that schools in rural Colombia are unsegregated. Moreover, the presence of youth gangs across these schools is linked to prior incidence of armed conflict rather than typical forms of social disadvantage. Exposure to gang-affiliated classmate can thus be considered as a quasi-random shock to affected youth. The analysis reveals gender differences in the effect of youth gang exposure. I find that girls react to male gang classmate by increased involvement in prosocial organizations. Boys, by contrast, adjust to male gangs by expressing more antisocial attitudes. The article shows that the well-documented antisocial adjustments to gangs are—population-wide—complemented by prosocial reactions, with gender being a key moderator. I discuss implications of these findings for theories of violence, gender, and neighborhood effects. I also provide suggestive evidence that my findings are applicable to the US context.

Rules play a pivotal role in creating and maintaining well-functioning societies. It seems self-evident that whether or not people follow rules has implications for their standing in groups and organizations. However, previous research has yielded mixed evidence. Some studies indicate that people who abide by rules are afforded positions of rank and influence, whereas other studies suggest that people who break rules gain influence. Building on the dominance/prestige framework of social rank, we argue that people who break rules appear high in dominance (through their assertive behavior), but low in prestige (as they do not act like role models). Thus, although rule breakers may appeal to stereotypes about leaders as assertive, in the absence of prestige, people are likely hesitant to grant leadership to rule breakers. In contrast, we argue that people who bend rules – who abide by the letter of the rule but not by its spirit – obtain the benefits of rule breakers (increased dominance) without their associated drawbacks (no reduction in prestige). Rule benders may therefore readily be granted leadership, and more so than people who abide by rules or engage in outright rule violation. Two scenario-based experiments provide support for this prediction, employing a behavioral measure of leadership granting (Study 1, N = 235) and a contextualized measure of leadership granting (Study 2, N = 480, preregistered). We discuss practical implications for the prevalence of rule bending and theoretical implications for the study of social norms in general.

We use a laboratory experiment to study the role of bystanders in bullying. We devise a simple repeated game played in groups of three subjects with one leader and two followers. At the beginning of each round, each subject receives the same initial endowment. Then the leader proposes to reallocate a certain portion of the endowment from a follower of her choice to herself. If the other follower, i.e. the bystander, agrees, the reallocation takes place. If the bystander disagrees (at zero additional cost), payoffs of all players equal to the initial endowment. This game is played repeatedly in a partner matching and with the same leader. The allocation of the role of the victim and bystander depends on leader’s choice in every round. The aim of the paper is to explore the motives of the bystander to agree with the unequal distribution. For this, we use two manipulations. First, we elicit social identity (SI) creating two teams of different colors, letting them wear different T-shirts and play a cooperative game in teams. In the treatment with different social identities, one follower is on the same team as the leader while the other follower is on the other team. In the treatment with the same SI, all players are from the same team. Second, the number of rounds played is either known (finitely repeated) or uncertain (infinitely repeated). The choices of bystanders in the last rounds of these treatments might reflect the bystander’s fear of retaliation from the side of the leader, as the bystander who disagrees with the proposed re-allocation might expect to become the victim in the next round. Our experiment shows that while the different social identity of the victim makes the bystander more likely to accept the unequal distribution, fear of retaliation has no significant effect on her choice.

Many situations require the action of a single volunteer to create a benefit for a larger group. For example, programmers provide others with open source software, interested writers publish articles on Wikipedia, or people bring wine-openers to a party in the park. These actions often happen in fairly unstructured, spontaneous and rapidly changing environments. And very naturally, individual information about personal costs and benefits, ties between actors, or the dimensions of problems entailed in the system is limited in those situations.

Maybe most importantly, we often have no or only partial information about how many other individuals are willing and able to engage. While we may have some rough feeling over the number of other potential volunteers, the exact number is usually unknown. We call this feature population uncertainty.

Volunteering situations give rise to the so-called bystander erffect - the observation that the presence of multiple potential volunteer's decreases the individual probability to act. One prominent line of reasoning or "narrative" that these defectors use in these situations is the (potentially biased belief) that somebody else will do it. In other words, decision makers might have a situation in mind where one ore more specific others - a "designated volunteer" - does the job.

As we will argue in the project, population uncertainty might hinder this mechanism because the imagined designated volunteers or specific others might just not be present. Or, alternatively, it is harder to keep up the narrative that someone else will do the job if the group and designated volunteers are not tangible. Consequently, being less able to focus on the actions of others, population uncertainty might focus decision makers on their own actions and activate a volunteering norm. Interestingly, population uncertainty might thus activate the actual norm to volunteer. Having a certain group size would then actually be a special case allowing to put blame on others.

We analyse these questions theoretically and experimentally. In particular, we exogenously generate designated volunteers by introducing heterogeneity in the costs to volunteer. Making the group size uncertain - introducing population uncertainty thus generates the possibility that these designated volunteers are not present. First results show the important impact of population uncertainty on volunteering behavior.

Social norms promote cooperation in everyday life because many people are willing to negatively sanction norm breakers at a cost to themselves. However, a norm violation may persist if only one person is required to sanction the norm breaker and everyone expects someone else to do it. Here we employ the volunteer’s dilemma game to model this diffusion of responsibility in social norm enforcement. The symmetric dilemma is a binary choice game in which all actors have the same costs of and benefits from cooperation and only one actor’s cooperation is required to provide the collective good for the group. The asymmetric dilemma differs from the symmetric game in one (strong) actor having lower costs of cooperation.

Here we experimentally test the hypothesis that the diffusion of responsibility effect decreases as a consequence of the switch from the symmetric to the asymmetric dilemma. In total, 252 subjects participated in our computerized laboratory experiment. In our experiment, we use the stealing game with a sanctioning option to emulate a situation in which a norm violation can be negatively sanctioned. That is, in a group of subjects, one randomly chosen subject can decide to steal money from the other group members (i.e. violate a social norm), who then face a volunteer’s dilemma as only one of them is required to reclaim the stolen amount for the entire group (i.e. enforce the social norm). We vary group size and symmetry in terms of the costs a subject incurs from sanctioning the thief.

Our results show a clear diffusion of responsibility effect in the symmetric dilemma, in which all group members have the same costs of sanctioning the thief. In the asymmetric dilemma, diffusion of responsibility is largely diminished but only after subject have played the game for some time.

By and large, these results are also borne out at the group level. In particular, heterogeneous groups become more effective in enforcing social norms as they manage to tacitly coordinate on the strongest subject to sanction the norm breaker alone. Our findings support the proposition that even relatively small asymmetries in observable sanctioning costs facilitate bystanders’ tacit coordination on the “strongest” individual to negatively sanction norm breakers. In other words, our results show how asymmetry can “break” the diffusion of responsibility in social norm enforcement and help to overcome the second-order free-rider problem (PLoS One, November 2018).

This paper studies the causal effect of improved housing conditions on people's risk and time preferences and cognitive function. Understanding the effects of improved housing quality is important for evaluating public housing policies as well as for understanding the impact of bad housing conditions on the choices of poor people and their ability to improve their economic situation. The data will be collected from adult participants of a unique randomized controlled trial “Rapid Re-Housing” that is currently running in Brno in the Czech Republic. In this trial, a randomly selected group of 50 homeless families were moved to municipal flats, while the 100 control families were not helped by the city and, in the majority of cases, remained in bad housing conditions.

The question about the relationship between wealth or income and risk or time preferences has been studied by many authors (e.g. Guiso and Paiella, JEEA, 2008; Tanaka et al, AER, 2010) and surveyed by Haushofer and Fehr (Science, 2014). The effect of wealth or income on cognitive abilities has been studied e.g. by Mani et al. (Science, 2013). In comparison to these studies, our treatment increases only consumption of housing services while keeping wealth or income constant. Therefore, the design of the RCT allows us to isolate the effect of one channel, better housing, through which higher wealth or income can influence preferences or cognitive abilities.

We conducted a lab-in-the-field experiment with adult members of the treatment and control group (roughly 160 subjects) in 2018. Risk and time preferences and cognitive function were measured using standard incentivized questionnaires. For measuring risk and time preferences we used standard multiple price lists adapted from Sutter et al. (AER, 2013). For measuring cognitive function, we selected the D2 test of attention.

We do not find any differences between the control and treatment groups in risk and time preferences and in their cognitive abilities. This result goes against the above-cited literature on the impact of poverty on preferences and cognitive abilities. There are several possible explanations of the zero result. First of all, our treatment changes the stream of consumption from housing services, but it does not affect the monetary wealth of our participants. The null results would be expected if the differences in preferences and abilities were due to differences in financial wealth. Second, the change in housing condition might not be large enough to drive the expected effects. In order to address this issue, we used questionnaire data about their wellbeing, which is significantly higher in the treatment group. This data is used to estimate the financial impact of the treatment. The third option is that our sample size does not provide enough power to identify the results. This issue is addressed in a power analysis.

Hyperbolic discounting behavior can arise in experiments when expected utility maximizing subjects who discount exponentially doubt the credibility of future payoffs. We show theoretically that lack of credibility introduces a present bias, as subjects internalize the uncertainty. Hence, experiments that do not ensure credibility may erroneously conclude that observed behavior is driven by hyperbolic pure time preferences, rather than the rational response to non-credible payoffs. We are currently undergoing an experiment to test the prediction of the theory.

Despite the omnipresence of inter-group conflict, little is known about the underlying heterogeneity of individuals' group-dependent social preferences and their dynamics over time. Building on research in economics, psychology, and especially the interdisciplinary parochial altruism literature, we derive a typology of group-(in)dependent social preferences from the Social Value Orientation (SVO) model.

We propose six profiles of social preferences in relation to in-group members and out-group members. Three types do not differentiate between in-group and out-group members in their preferences: Universal Altruists, Universal Egoists, and Universal Competitors, who are altruistic, egoistic, or spiteful, respectively, irrespective of others' group membership. In contrast, three other types have group-dependent preferences. These types show a greater concern for in-group members compared to out-group members. Weakly Parochial Altruists are prosocial toward in-group members and less so toward out-group members, but they do not have negative social preferences (i.e., competitiveness, aggression) toward the latter. Strongly Parochial Altruists are prosocial toward in-group members coupled with negative social preferences toward out-group members. Parochial Egoists are egoistic toward in-group members but are willing to forego some of their own gain to reduce the payoff of out-group members.

To identify the empirical prevalence of those types and the dynamics of group-dependent social preferences around and even years after a (political) conflict, we gather quota-representative, incentivized panel data from a field experiment during and after the heated 2016 Austrian presidential election. Concretely, we used the SVO Slider measure and matched subjects with both in- and out-group interaction partners (a voter of their own and the competing candidate, respectively) multiple times over the course of the conflict: in the week before the election, in the week after the election, three months after the election, and two years after the election.

We find a strong effect of the interaction partner’s group membership. In the week before the election, the average SVO angle is 24.9 degrees when matched with an in-group member and 8.8 degrees when matched with an out-group member (Cohen’s d = .8). Despite this strong effect of group membership, average in- and out-group SVO angles are remarkable stable over time, even for voters of the winning and the losing candidate separately.

At the individual level, we find considerable heterogeneity in our derived types: One week before the election, around 50% of our subjects have group-dependent social preferences. 13% of the subjects also care positively about the payoff of fellow in-group members, but are not willing to give up payoff for the benefit of out-group members (Weakly Parochial Altruists); 20% are Strongly Parochial Altruists, willing to reduce their own payoff for both the benefit of in-group members and the detriment of out-group members. 17% are Parochial Egoists, not willing to costly help in-group members but to costly hurt out-group members. 30% show a positive concern for the payoff of others, irrespective of group membership (Universal Altruists). Over time, only the share of Universal Egoists changes significantly (from 10% to 17%).

Research on social dilemmas in the production of collective goods typically focuses on cooperation problems situated within a particular social group. Factors influencing whether the dilemma is successfully solved are sought at the level of that group or of the individuals belonging to it. There is an ongoing discussion about the applicability of knowledge gained from laboratory experiments to real-life groups (e.g. Guala, 2012 and responses). For example, the limits of punishment (in the presence of retaliation, e.g. Nikiforakis, 2008; Nikiforakis & Engelmann, 2011) and reward (in the presence of reward exchange, e.g. Flache, Bakker, Mäs, & Dijkstra, 2017; Flache & Macy, 1996) as enforcers of cooperation are being explored. Additionally, there are attempts to apply the knowledge gained from laboratory experiments to increasingly practical situations (e.g. Englmaier & Gebhardt, 2016; Fehr & Leibbrandt, 2011; Kraft-Todd, Yoeli, Bhanot, & Rand, 2015).

We believe that there is an additional aspect to the generalizability of research on social dilemmas, which thus far has received little attention. This aspect is the broader social environment within which a group is embedded, specifically the possibility of overlap in group membership. In the present study, we first illustrate the relevance of the broader social structure, and overlap in group membership in particular, by giving an overview of literature outside of the field of social dilemmas which has demonstrated its importance. Then, we focus on some relevant aspects of the social dilemma literature which can serve as starting points for an investigation into the impact of the broader social structure. Finally, we take the first steps of this investigation by experimentally scrutinizing how findings from social dilemma research on monitoring of free-riders apply to multiple-group situations with overlap in group membership.

The present study makes several contributions to the literature. First, to our knowledge, this is the first experimental investigation of behavior in multiple simultaneous public good games with overlap in group membership. Second, we introduce and validate a punishment system suitable for such an experiment. Third, we investigate how imperfect monitoring of free-riders caused by multiple group membership impacts sanctioning and contribution behavior.

Our results illustrate that findings from social dilemma research which are based on single, isolated groups do not necessarily generalize to multiple groups which have some members in common. We find that the broader social context in which a group is embedded is relevant to the solution of social dilemmas within that group.

This laboratory experiment investigates whether coordination on sunspots may arise among a large number of participants (80-90 subjects), and how this coordination may be influenced by the payoff and the sunspot structures. Our setup is a bank run game where withdrawing is the safe option, but waiting is the payoff-dominant strategy. Comparing behavior in small and large groups, we find major differences that equilibrium refinements fail to predict. Coordination on sunspots never happens in large groups, while it sometimes happens in small groups. Furthermore, coordination failures are systematic in large groups as soon as coordination on the Pareto-dominant equilibrium is risky enough. In contrast, small groups may still converge to the optimal equilibrium even when the safe option is relatively more attractive.

Muslims generally report high preference for Islamic financial products over the conventional ones, yet, when a choice is available they use non-Islamic financial products. This article uses a list experiment technique to look at this demand puzzle of preferences and use of Islamic financial products by Muslim borrowers of Islamic microfinance institutions in Pakistan. Comparing the direct and indirect responses of the borrowers, we find a significant difference in the preferences and borrowing behavior. This paper highlights important aspect of the demand puzzle and guide the future research to explore the determinants of this demand puzzle.

Do futures markets have a stabilizing or destabilizing eect on commodity prices? Empirical evidence suggests that both effects are possible. We investigate this question further using a learning-to-forecast experiment with two coupled markets: a spot market and a futures market. The first exhibits negative feedback between forecasts and prices, associated with stable prices, while the second market is of the positive feedback type, which makes it susceptible to bubbles and crashes. The results show that the effect of a futures market on spot price stability changes non-monotonically with the coupling strength. When the coupling is weak the futures market has a stabilizing effect on spot prices. Under those circumstances increasing the coupling strength reduces spot price volatility. However, at larger coupling strengths this trend reverses and the effect of the futures market on spot prices becomes destabilizing.

How does memory shape individuals' financial decisions? We find experimental evidence of a self-serving memory bias for gains and losses and show that this memory bias distorts beliefs and drives investment choices. Subjects who previously invested in a risky stock are more likely to remember positive investment outcomes relative to negative outcomes than are subjects who did not invest in the stock but only observed its outcomes. Importantly, subjects do not adjust their behavior to account for the fallibility of their memory, which leads to investment mistakes. They are likely to form overly optimistic beliefs and to re-invest in the stock even when doing so reduces their expected return. The memory bias we document is relevant for understanding how people form expectations from experiences in financial markets and, more generally, for understanding household financial decision-making.

Trust has long been recognized as one of the most important ingredients of social and economic life. People’s expectation of others’ trustworthiness is the precursor of these people’s trust in others. The answer to how people form trustworthiness expectations is therefore key to answering why they (do not) trust others in particular situations. Theoretically, people’s trustworthiness can be inferred from any behavioral or contextual signs the occurrence of which is (1) correlated with these people’s trustworthiness and (2) regarded as indicative of trustworthiness by observers. Experimental research shows that generosity could be such a sign because it is correlated with trustworthiness and used to infer trustworthiness in social exchange with trust at stake. However, little is known about the prevalence and magnitude of the correlation between generosity and trustworthiness across different study designs and locations.

We conduct a meta-analysis based on experimentaldata from 25 studies in which generosity and trustworthiness were measured within subject. In these studies, generosity was measured by means of donations to charity, dictator game giving and first-mover transfers in ultimatum games; trustworthiness was measured by means of second-mover decisions in (binary) trust games, sequential prisoner’s dilemmas and second-mover transfers in investment games. We test the hypothesis that strategic generosity is a weaker sign of trustworthiness than natural generosity by comparing correlations between generosity and trustworthiness produced in two types of situations: (1) strategic situations, in which subjects know that their generosity could affect their interaction partners’ decisions and in turn affect their payoffs and (2) “natural” situations, in which subjects’ generosity cannot have such an effect by design or in which subjects are informed only later that it could.

Overall, generosity and trustworthiness are moderately correlated (r = 0.37). The correlation coefficients range from 0.03 to 0.69. In line with our hypothesis, we find that the correlation between strategic generosity and trustworthiness is smaller than the correlation between natural generosity and trustworthiness. These results substantiate that generosity can serve as a sign of trustworthiness and even more when generosity is exhibited naturally, i.e. without potential future rewards for generous acts. Our exploratory analysis reveals moreover that neither the proportion of female subjects, the use of strategy method to measure trustworthiness, nor game endowments have an effect on the correlation between generosity and trustworthiness. Our findings suggest that providing people with more opportunities to engage in and observe acts of generosity may promote trust in theirneighborhoods and society at large.

Economic sociologists have long established that trust problems give rise to market structures in which many exchanges take place within committed long-term relationships between specific trading parties. Returning to proven partners allows actors to mitigate the risk of being taken advantage of, and the preferential choice of partners who were reliable in the past provides incentives for honest behavior. We revisit the question of how trust problems affect exchange structures. Arguably, institutionalized reputations systems are a key source of trust in modern economies. They enable actors to learn from the experiences of others and thereby provide a basis for trust and trustworthiness that substitutes for the incentives in long-term relations. Do trust problems no longer shape exchange structures in the presence of technologically facilitated large-scale reputation systems? We argue that trust problems still have far-reaching implications for market structures, but that the key feature they bring about is no longer dyadic commitments. We show in a game-theoretic model and in a laboratory experiment that in the presence of reputation systems, trust problems instead lead to the emergence of high market concentration, with all buyers frequenting one or a few sellers and excluding many others. The few sellers who are lucky to serve the whole market additionally earn a sizeable premium for their good reputations.

Trusting and trustworthy environments are argued to promote collective action, as people learn to rely on their fellow citizens and believe that only few individuals will free ride. However, no study has tested the causal validity of this mechanism. To do so, we propose an experimental design that allows us to create different trusting and trustworthy conditions simply by (1) manipulating the incentive structure of an iterated binary Trust Game and (2) allowing information to flow among participants. Subjects are randomly assigned to either a low or high trusting environment, play 20 rounds of a binary Trust Game, and then participate in a one-shot binary Public Goods game. Findings indicate that, given a similar distribution of resources among subjects, trusting and trustworthy environments strongly foster the provision of public goods. This outcome is largely driven by a learning effect: we are more likely to act for the collectivity when we learn from the community to be trustful or reliable in our one-to-one interactions. The same applies in the opposite direction: we are more prone to free ride when we learn from the environment to be distrustful or unreliable in our dyadic exchanges.

The aim of this paper is to gain insight on whether and how heterogeneity affects individual and collective cooperative behaviour in CPR settings, using a Trust Game and a CPR game in a computerized laboratory experiment. This paper distinguishes between economic and sociocultural heterogeneity, where economic heterogeneity is expressed as inequality in endowments and sociocultural heterogeneity is expressed as differences in (induced) social identities. This paper will consider trust as an important mediating variable between the effect of heterogeneity on cooperation. Cooperation is measured both on the micro level (individual appropriation effort) and the macro level (total appropriation of the group). Preliminary results seem to show that general trust (revealed trust in the first round of the Investment Game) affects cooperation on the individual level positively and substantially, and that heterogeneity does not seem to play a significant role in reaching sustainable cooperation in CPR settings.

The aim of this paper is to examine whether and under what circumstances interactions in the sharing economy lead to an increase in the general perceived trustworthiness of others. In order to study this, we conducted an online field experiment among users of the home sharing platform Airbnb for which we measured participants’ trusting behavior at two points in time. Between these two phases of the experiment, a proportion of our participants had a sharing interaction facilitated by Airbnb: either as a guest, as a host, or as both. We find that sharing has a positive impact on trust, but only when participants have sufficiently many sharing interactions. Furthermore, the positive effect only exists for hosts; we do not find a significant effect for guests.