Poster Session

First NameFamily NameUniversity/institutional affiliationTitleAbstract
AvishayAicheWestern Galilee CollegeTwo stage presidential election model and the effect of the media on the election's resultsWe consider a presidential election model, with two candidates and voters who are uniformly distributed on a one dimensional policy space, as a two stage election game. At the first stage each candidate presents his political agenda. After presenting his agenda, each candidate may approach the public through different means of media in order to maximize his chances of winning the elections. The information the candidate presents in the media affect voters differently according to their political view relative to the announcement the candidate presented. We show how the candidate's belief of his strategy carried out affects his chances of winning the elections
ArnoApffelstaedtUniversity of HamburgThe Effect of (Non-) Democracy on Rule-ComplianceIn many developing countries, but also increasingly often in Western societies, one finds mistrust in political institutions related to the influence of money on political processes. Using an online experiment, we study the compliance with rules that have been implemented either by majority vote or by a corrupt voting mechanism. The rules demand people to share or to not share their income with unlucky agents in modified dictator games. Our aim is to test to what extend a corrupt institution actually leads to a decline in compliance with rules set by this institution. If we find such an effect, undermining democratic procedures might not only have moral costs but also lead to efficiency losses for governments as they will have to increasingly engage in monitoring and punishing people who do not follow rules (for example, tax laws).

The experiment is conducted online with international participants of different cultural and social backgrounds. Each experimental session implements two rounds of a modified dictator game: In each round, 100 individuals participate in a lottery to win a cash prize of Ìâå£100. Fifty percent of the participants receive 10 lottery tickets each ("receivers"), while the other fifty percent do not receive any tickets ("nonreceivers"). Before participants learn their role, they have to decide whether or not they want to give 3 out of 10 lottery tickets to a participant without tickets if they happen to be a receiver. In round 1, each participant decides privately without receiving additional information. In round 2, before participants decide anew, a non-binding "code of conduct" is implemented for all participants. The code of conduct says whether all participants should give (Rule: Give) or should not give (Rule: Don't Give). Before a rule is selected, we let all participants vote for the rule that they prefer to have implemented. Between-subject treatments vary by whether and how these votes are counted. The baseline treatment is a simple majority vote. The other treatments corrupt features of the majority vote that potentially make it a legitimate institution: One treatment introduces the possibility to accept bribes, i.e. each participant can accept a small payment in order to reverse her own vote. Another treatment requires participants to pay a fee to make their vote count, thereby violating political equality, or the one-man-one-vote principal of elections. In both cases, the rule that is finally implemented does not necessarily represent the preference of the majority of people.

We implement several technical features to ensure that our design allows us to learn more about the general nature of rule compliance. Participants' unbiased choice in round 1 and their unbiased voting behavior allows us to control for the type of participants when identifying treatment effects. We use the strategy method to elicit compliance for each participant with both rules, thus getting a more nuanced picture of individual rule compliance. We elicit (incentivized) beliefs about the behaviors of other participants in order to understand in how far rules are complied with because people have the desire to follow others. To establish this relationship causally, we exogenously manipulate beliefs by giving participants different signals about giving decisions in earlier sessions. We then use the exogenous variation to instrument for beliefs in our regression. Taking advantage of the heterogeneous online subject pool, we add a questionnaire including socio-demographic variables and questions on real-world political behavior in order to relate those to decisions in the experiment. The questionnaire is sent to the same people one week after they participated in the experiment to prevent that the answers are influenced by experiences in the previous experimental treatment. By disentangling the strategic and non-strategic components of changes in behavior in response to a social rule, we hope to be able to provide a better understanding of how results found in related studies (e.g. Dal Bo et al. 2010 or Tyran and Feld 2006) are created and mediated.

The preliminary analysis of the sessions conducted so far has produced several exciting results. Controlling for the unbiased type of participants (using round 1-choices and voting behavior), we find that corruption has a strong and significant effect for compliance with the pro-social rule (Rule: Give), but not for compliance with Rule: Don't Give. In particular, when voting is limited to those participants who pay a fee or when participants are offered a bribe to vote for the opposite rule, average compliance with Rule: Give decreases by 12% (p-values are .037 and .018, respectively) compared to the uncorrupted majority vote. The treatment effect is entirely driven by participants whose vote is directly affected by the intervention: Participants who "lost their voice" by not paying to make their vote count or by accepting the bribe are 32% and 53% less likely to follow Rule: Give, respectively (both effects are significant at 1% level and different at 5% level), while rule compliance of other participants is unaffected. Moreover, the results seem to be driven by adverse effects on the intrinsic motivation to follow a rule: They are unchanged when controlling for participants' final beliefs about the outcome of the voting mechanism. Regarding the nature of rule-compliance more generally, we find a strong causal effect of the belief about others' behavior on compliance with Rule:Ìâ Don't Give (IV-estimates show that a 10% increase in expected compliance by others increases the likelihood of own compliance by nearly 10%, p-value .012), but no significant effect of beliefs on the compliance with Rule: Give. This result confirms the popular view that anti-social behaviors are more "contagious" than pro-social behaviors.

By the time of the Behavioral Models of Politics Conference at Duke University we will be able to present results for (1) two further (corruption) treatments, (2) sessions without role uncertainty (thereby controlling for reciprocity concerns in the giving decision) and (3) sessions in which the rule is implemented together with positive (yet non-deterrent) deviation costs.
RebeccaBryanUniversity at Buffalo, SUNYStatus Update: Estimating the Impact of Social Pressure on Voter Turnout Using Social Networks Present on FacebookThis paper investigates the relationship between social media and voter turnout through an online social network's ability to apply social pressure. We theorize that social pressure, encouraging group members to participate in politics, is transmitted along ties between individuals within an online network. The data for this study was collected by matching Facebook user profiles to voter registration lists from four U.S. counties. We find that Facebook use has a positive effect on individual turnout. Specifically, users with high levels of exposure to dense Facebook networks are more likely to turnout than new users with small networks. This provides evidence that an individual's online social network plays a role in the decision to vote.
Gabriel N.Camargo-ToledoVanderbilt UniversityThe Impact of Institutions on Turnout. A Behavioral Theory.Students of political behavior have discussed the relation between turnout and electoral systems. Most agree that proportional rules are related to higher rates of turnout, but there is some debate about this. Some authors argue that this link is not consistent for newer democracies while others argue that taking into consideration other institutional features conditions this relation. Furthermore, it is also argued that this link lacks a clear theoretical mechanism of why it occurs. My work will bridge the discussion by using a different theoretical perspective, using a formal behavioral model. These kind of models have several advantages over traditional rational choice approaches of turnout and have several properties that I argue will explain the proportional-turnout link and the inconsistencies the literature has found.
CharlesChangStanford UniversityA Crisis of Trust in the Big Data EraDoes society under authoritarian rule believe in autocratsÌ¢‰â‰㢠information during and immediately after a crisis? I approach this question by examining the response of the residents of a Chinese city, Kunming, in the course of a terrorist attack on its railway station in 2014, and find that they show a decided lack of confidenceÌ¢‰â‰۝distrust rather than trustÌ¢‰â‰۝in autocratsÌ¢‰â‰㢠information, even though, superficially, it sounds helpful and reassuring. Distrust is indicated not only in opinion but, more convincingly, in movement away from the railway station even as autocrats and Party functionaries repeated their reassuring stories. My study, which is based on a unique dataset from social media and geo-tagged smartphones, shows happenings immediately before and after the attack, that is, in real time rather than retroactively, as it would be if I were to use the standard survey-and-interview technique. Theoretically, I postulate that in authoritarian societies, people are likely to distrust autocratsÌ¢‰â‰㢠information when it is loud and univocal, even in a crisis when, in the midst of uncertainty and impending chaos, a clear and authoritative voice might seem called for.
Babak RezaeeDaryakenariArizona State UniversityWhy Violence? A social movement growth model for explaining violent resistanceScholars in social science have been interested in studying violent behavior during periods of political and social conflict for quite some time, so there are a wide variety of explanations and theories in the literature on this topic. To explain violent movements, some studies have focused on state behavior, while others emphasized the structure of domestic and foreign politics. Some scholars also underscore the features of opposition movements that make a group more liable to employ violence. This study contributes to the latter discourse of understanding collective political violence by developing a social movement growth model. This theoretical framework helps us to discuss how analyzing the dynamics of capabilities possessed by a social movement improves our understanding of variations in resorting to violence across social movements.
Although mathematical modeling has helped me to model the interactions between agents in my theory, the complexity of the developed framework, due to the number of mechanisms and the changes in the agent's attributes in every period, makes using classic approaches for discussing its theoretical conclusions problematic, if not impossible. Indeed, the fact that a social movementÌ¢‰â‰ã¢s organizational evolution path affects both its and the stateÌ¢‰â‰ã¢s current period strategic behaviors is one of the core arguments in my theory. However, the theoretical implications of this model cannot be investigated adequately without utilizing computational modeling. Thus, I have relied on computational techniques to study the behavior of this complex system under different assumptions and scenarios.
JanaFreundtUniversity of HamburgThe Effect of (Non-)Democracy on Rule CompliancePeople (often severely) differ in their attitudes towards redistribution and income inequality. We rely on an online experiment to identify people‰Ûªs compliance to a non-binding rule that determines whether to share or not to share one‰Ûªs income with unlucky agents in modified dictator games. Treatments vary between subjects if social rules are implemented by a democratic procedure, i.e. majority vote, or by corrupt procedures where the majority vote has been undermined by the possibility to buy votes or to accept bribes. Taking advantage of a controlled experiment, we can isolate treatment effects while controlling for an agent‰Ûªs type and beliefs, thereby disentangling strategic and non-strategic components of changes in behavior in response to a social rule. Conducting the experiment online with an international subject pool allows for relating decisions to (self-reported) real-world political behavior, cultural background and personality traits. First results indicate that undermining democratic procedures indeed might result in lower rule compliance which is of great importance to governments and, more generally, to questions of mechanism design: In this case, undemocratic mechanisms lead to efficiency losses because governments will have to increasingly engage in incentivizing or punishing people who do not follow their rules voluntarily. This result is especially relevant in light of the widespread mistrust in political institutions related to the influence of money on political processes in developing countries as well as increasingly often in Western societies.
The experiment is conducted online with international participants of different cultural and social backgrounds. Groups of 100 individuals participate in a lottery to win a cash prize of Ìâå£100 and during the experiment they will be able to allocate lottery tickets. The experiment consists of two rounds. Round 1 is a modified dictator game. Participants know that they will be randomly selected to be dictator or receiver, where the dictators are endowed with 10 lottery tickets each and the receivers get zero tickets. Before they learn about their role, all participants have to decide whether or not to give 3 out of the 10 tickets to a randomly matched receiver in case they happen to be a dictator. In round 2, participants vote on implementing a non-binding rule for all participants that prescribes whether or not to share chances to win. The two rules are Rule: Give and Rule: Donate/Give. Subsequently, one of the rules is implemented and treatments vary by mechanism that selects a rule. For eliciting rule compliance we benefit from the strategy method. Participants are asked to make their choice to give or not give conditional on Rule: Give and Rule: Donate/Give being implemented. This allows us to elicit compliance to both rules, thus getting a more nuanced picture of individual rule compliance. After making their choices, we ask all participants to state their beliefs about how many of the other 99 participants chose to give under each rule and about how many people have voted for Rule:Give. In order to be able to establish a causal relationship between beliefs about others‰Ûª choices and own actions in round 2, we exogenously manipulate beliefs by varying the information about giving decisions of 5 selected participants from an earlier session that the participants receive after round 1. Additionally, we can define individual types from choices in round 1 and from votes. Thus, by disentangling the strategic and non-strategic components of changes in behavior in response to a social rule, we hope to be able to provide a better understanding of how effects on rule compliance are created and mediated.
In a between subject design, we vary the procedure that implements the rule: The two baseline treatments are majority vote and an exogenous implementation of a randomly selected rule. In additional treatments, we Ì¢‰âÂèÏcorruptÌ¢‰â‰㢠the majority vote such that either political equality (one-man-one-vote) or the right to express oneÌ¢‰â‰ã¢s voice are violated: More precisely, we introduce the possibility to accept bribes, the possibility to buy the right to vote or to buy additional votes. In all cases, the rule that is finally implemented not necessarily represents the preference of the majority of the people. Thus, we can not only establish an effect of the rule-setting mechanism compliance decisions but also shed light on the question which features are essential for making majority vote a legitimate procedure. Additionally, we vary the cost of deviating from the rule from zero to a non-deterrent positive cost between subjects. Thereby we compare compliance behavior for the case of a pure norm, or a default rule, to one with punishments and its interaction with the rule-setting mechanism.
Taking advantage of the heterogeneous online subject pool, we add a questionnaire including socio-demographic variables and questions on real-world political behavior in order to relate those to decisions in the experiment. The questionnaire is send to the exactly same people one week after they participated in the experiment to prevent that the answers are influenced by experiences in the previous experimental treatment.
We discuss how to include institutional effects into the social signaling model by Benabou and Tirole (2012) as affecting a person‰Ûªs (self) esteem concerns. Interestingly, the model predicts treatment effects to differ depending on whether the more prosocial or the more egoistic rule is implemented. First evidence indicates that in the majority vote treatment significantly more people follow the rules than with a corrupted majority vote. This difference is driven by behavior under the rule that prescribes to share tickets, but not observable for the rule that prescribes to keep the tickets. Regression analyses show that indeed individuals‰Ûª types and their beliefs about others‰Ûª behavior are essential for understanding giving pattern under the two rules. The data are being collected at the moment so that by the time of the conference we will be able to say much more about the nature of the legitimacy effects of the different rule-setting mechanisms.

Benabou, Roland and Jean Tirole. 2012. Laws and Norms, IZA Discussion Paper No. 6290.
XiaoliGuoFlorida State UniversityEfficient Interventions via Creating A Reciprocal RivalA dilemma facing an intervener who intends to make peace in inter- or intra-state military conflicts is to trade off the effectiveness and cost of interventions. Extant studies generally suggest that more severe measures tend to be more effective, but meanwhile more costly. Using a game-theoretical model, we provide a novel solution to improving the efficiency of interventions: to induce a reciprocal rival that only attacks if the other side strikes first. In most cases we explore, moderate penalties on the second mover in a sequential game generate such a rival, which decreases the cost of effective interventions regarding both rivals. However, when the power dynamics motivate the rival to exploit the other side's passiveness instead of coordinating for peace, moderate penalties may no work. We also find that a weaker rival is not necessarily easier to harness and sometimes its aggressiveness comes from its weakness; and interventions can work through the rival's expectation instead of implementation. Insights from our model are applied to analyze why the Marshall mission in the Chinese Civil War (1945-1949) to maintain peace and broker a coalition government failed.
ThomasJamiesonUniversity of Southern CaliforniaWhen Information Changes Preferences: Framing Effects, Public Opinion, and International Humanitarian AssistanceThis study builds on previous research that suggests that episodic frames are more likely than thematic frames to produce emotional responses that contribute to people changing their preferences. Through a survey experiment on 602 participants, I find support for the contention that episodic frames overcome people‰Ûªs preferences during periods of international crises. People treated with episodic frames increase support for both the provision of financial aid and the support of military intervention to assist victims of the crises, and they are more willing to sign petitions in support of humanitarian action than those exposed to other conditions. Contrary to expectations, thematic frames also had strong positive effects on individual attitudes towards assisting victims of the crisis, but this support did not correspond to costly actions. However, across all conditions, the effects do not endure over time, and attitudes towards humanitarian assistance declined in the two weeks between posttests. The findings have implications for the understanding of information effects, and they present an explanation for the limited duration of domestic support for government intervention during crises.
Abdul HaleemKidwaiUniversity of Massachusetts AmherstBargaining over Uncertain Resources: Is Ambiguity better for Agreement than Risk?Bargaining over resources is ubiquitous in the real-world - water reserves being fought over by different users, the surplus profits of a firm being divided between management and workers, fish stock being harvested by competing fishermen, amongst others. In a number of these cases, the size of the contested resource is uncertain. Moreover, this uncertainty might manifest itself as risk (where the probability distribution is known) or as ambiguity (where the probability distribution is unknown). This raises an important question: does the type of uncertainty, risk vs. ambiguity, have an impact on demands made on the resource? Theoretical work by Aflaki (2013) claims that it does and that under fairly plausible conditions (i.e. ambiguity-averse agents), lower demands will be made on the resource under ambiguity as compared to the case of risk. Therefore, bargaining is going to be more successful, that is, the parties will reach an agreement more often, under ambiguity than risk. Additionally, bargaining decisions depend on the consequence of failing to reach an agreement. Aflaki (2013) predicts that when the consequence is severe (the bargaining parties receive nothing from the resource), the demands made on the resource will be lower as compared to when the consequence is mild (even if bargaining parties fail to agree, they receive a portion of the resource). To the best of our knowledge, these theoretical claims have not been empirically tested. We run a stochastic Nash Demand (SND) game experiment to test these claims. The SND is a variant of the standard Nash Demand (or Divide the Dollar) game. In the standard game, the parties demand a fraction of the resource (or Dollar), if the sum of the demands is less than or equal to the resource (or Dollar), each party receives her demand. However, if the sum of demands exceeds the size of the resource, each party receives nothing (or some fraction of the resource - depending on the nature of the resource). The stochastic element in our game is the size of the resource. The bargaining parties are informed of the various possible sizes of the resource. In the risk treatment, they know the probabilities associated with these sizes but not in the ambiguity treatment. Our results show partial support for the theoretical predictions. We find, contrary to the prediction, demands are higher under ambiguity as compared to risk. We also find, in line with the theoretical prediction, that demands are lower when the consequence of bargaining failure is severe. These results highlight the necessity of distinguishing between risk and ambiguity in bargaining contexts. Furthermore, since the empirical evidence is not in agreement with the theoretical prediction of lower demands under ambiguity, it would suggest ambiguity models developed for predicting behavior in individual decision-making contexts (i.e. choosing between risky lotteries and ambiguous prospects), might not predict well in the settings with strategic interaction, such as, bargaining games.
LiKing KingCity University of Hong KongThe Effect of Majority Election vs. Proportional Election on Provision of Public Goods: An Experimental InvestigationMajority rule has been criticized for the tyranny of majority, and proportional rule has been commonly proposed as a solution. We conduct laboratory experiments to analyze the effect of social preference under the two voting rules on provision of public goods. We show that when voters exhibit social preference and public good is particularly desirable, proportional rule will lead to lower provision of public goods. Our results overturn the prior theoretical literature, where voters are assumed to be narrowly self-interested, that suggests majority voting will lead to lower provision of public goods. Subjects‰Ûª behaviors in the laboratory are closely related to their political stance in reality.
YiJyunLinUniversity of NevadaIs It a Misplaced Causality or Unbearable Lightness of Climate Change? Climate Variability-Social Conflict NexusHow does climate variability indirectly affect rates of social conflict via its impact on natural resources? Does the strength of this relationship change as a function of interaction between local climatic and geographical conditions? Does the interaction between local climate and its environment further shape the capacity and power relationship among local groups in a way that increases rates of social conflict? This topic is situated in the growing literature on the climate-conflict link, which has arrived at contradictory conclusions concerning these issues. Part of the problem is the insufficient modeling of the spatiotemporal effects of climate on conflict. This research tests the joint effects of climate, geographic conditions, natural resources, and power relationships among groups on rates of social conflict. It utilizes a global data set combining country-level aggregate data with high-resolution data which has been plotted on a grid, as they relate to local climatic and geographic conditions. Then, a three-level mixed-effect negative binomial model is employed to examine the connection between climate and rates of social conflict. Data from various resources will be used, including NOAA, SPEI, EPR, and PRIO-GRID. Preliminary analysis shows that both drought and temperature variations are significantly associated with rates of conflict. In particular, proximity to drought zones as well as connectivity to urban areas each serves as key factors that link climate variability to conflicts. Notably, although drought level does have an impact on conflicts, conflicts tend not to occur in drought areas. Instead, conflicts more often befall locations situated close to urban centers, as well as in non-drought areas because populaces need to source what their basics from the nearest closest urban center or non-drought area.
JeromeSchaferYaleWhen You DonÌ¢‰â‰ã¢t Snooze, You Lose: A Natural
Experiment on the Effect of Sleep Deprivation
on Voter Turnout and Election Results
In this article, we show that many citizens fail to vote because they are too tired. To do so, we leverage multiple approaches, including a unique natural quasi-experiment that exploits discontinuous decreases in sleep times on the eastern side of U.S. time zone boundaries. Our preferred model specification indicates that these exogenous decreases in sleep times depress county-level turnout in Congressional elections by about 2 percentage points. This effect is magnified in areas where obstacles to voting are greatest. Moreover, tiredness appears to exacerbate participatory inequality depressing turnout in low propensity communities most and push election outcomes towards Republicans. Supplementing this analysis, we conduct an observational study validating the direct relationship between tiredness and turnout. Our findings have important theoretical implications for the study of political participation. They suggest that many citizens hold the precursors to participation but lack the general, rather than expressly political, motivation to act on their intentions.
MatthewWardUniversity of HoustonReassessing Mixed Evidence in the Inequality and Democratization Debate: The Role of Political Knowledge, Authoritarianism, and Social Dominance OrientationEmpirical support for the relationship between income inequality and the impact of democratic transitions in the comparative political science literature has been mixed. Based on game theoretical modeling, two prominent theories posit either a negative linear relationship on inequality and democratization (Boix 2003) or a curvilinear relationship, with low and high levels of inequality having a negative impact on democratization (Acemoglu and Robinson 2006). These theories are based on strategic moves in a game between the elites, the middle class, and the masses. Empirical evidence of these two theories has been mixed. Research has found support for both theories, neither theory, or that the impact of inequality affects backslides into autocracy, but not the probability of democratization. I argue that these theories should be supplemented with individual behavioral and psychological considerations. Using survey and experimental data and consistent with previous research, I find that Social Dominance Orientation and Authoritarianism are highly correlated with opposition to democratic institutions, despite the high level of inequality in the societies under investigation. I also find that political knowledge is highly correlated with decreased support for the current political and economic regime in these highly unequal, undemocratic countries. Controlling for these factors at the state and individual levels, I find that the relationship between inequality and support for democracy is negative and statistically significant.
Andrei ZhirnovZhirnovBinghamton UniversityDuverger's Psychological Effect and Rolling Elections in IndiaThe plurality rule is known to create incentives that can divert the vote from the third- and lower-ranked candidates. The existence of such incentives is conducive to such outcomes only if the voters and/or elites have enough information to act according to such incentives. In this essay, I argue that campaigning provides such information, thus leading to more Duvergerian outcomes. I capitalize on the quasi-experimental settings created by the institution of rolling elections in India: in this country, voting in general elections is organized in a series of phases spread over almost a month, with different districts--even within a single state--voting on different days. I treat the timing of the district vote as the endpoint of campaigning in the district and evaluate the effect of the length of campaigning on the vote for the leading candidates, the vote for the third parties, the share of valid vote cast for the third parties, and the concentration of the vote. I find a positive, albeit diminishing, effect of longer campaigning on the leading candidates' vote and vote concentration, and a negative effect of a longer campaign period on the third parties vote.
JesseZinnClayton State UniversityAn Information-Theoretic Foundation for the Weighted Updating ModelWeighted Updating generalizes Bayesian updating, allowing for biased beliefs by weighting the likelihood function and prior distribution with positive real exponents. I provide a rigorous foundation for the model by showing that transforming a distribution by exponential weighting (and normalizing) systematically affects the information entropy of the resulting distribution. For weights greater than one the resulting distribution has less information entropy than the original distribution, and vice versa. As the entropy of a distribution measures how informative a decision maker is treating the underlying observation(s), this result suggests a useful interpretation of the weights. For example, a weight greater than one on a likelihood function models an individual who is treating the associated observation(s) as being more informative than a perfect Bayesian would.