Maj-Britt Sterba

Abstract:  This paper establishes three novel findings on fairness and redistribution by combining theory-driven experimental games with large-scale surveys in the U.S. and five European countries. First, individuals revealing egalitarian, libertarian, or meritocratic fairness preferences in experimental games show large differences in support for tax and transfer policies. Second, beliefs in merit strongly predict policy support among meritocrats, but are less predictive among non-meritocrats. Third, fairness concerns matter across income groups and political camps. Our findings challenge the assumptions that meritocratic preferences are homogeneous, that fairness is a luxury good, and that fairness is mainly a moral foundation of left voters. 

Abstract: The modern world is characterized by immense inequalities in living standards across countries, raising the question of what, if anything, we owe to each other on a global scale. One trigger for moral obligations that has been widely discussed in political philosophy is cooperation for mutual benefit. Using an experimental vignette study, we investigate whether people perceive large-scale economic cooperation for mutual benefit as a trigger for a moral responsibility to redistribute resources. Our design focuses on differences in moral responsibility depending on the level of help provided, reflecting the idea that some obligations are relationship-specific, while others are not. Our main finding is that a prior cooperative relationship significantly increases the perceived moral responsibility to redistribute resources, and this increase is more pronounced for distributions that go beyond providing for basic needs. The effect of more extensive redistribution is partly mediated by a lower perceived social distance between cooperators. Our results imply that making the cooperative nature of economic interactions more salient might broaden the scope of people’s moral concerns towards cooperators.

Abstract: When do legislators respond to voters’ support for a policy? We argue that politicians are more likely to align with voter preferences for policies within their jurisdiction, where ignoring voter support can entail political costs. Political costs refer to a politician’s loss of electoral support when their stance on a policy deviates from the majority support for that policy in the electorate. Using a randomized survey experiment with 535 German state legislators, we examine how information about voter support affects policymakers’ support for two policy reforms—one at the state level and one at the national level. Our findings support the relevance of political costs: legislators who received voter information were at least 9 percentage points more likely to support the state-level reform, while no significant effect exists for the national-level reform. By specifying the conditions for responsiveness, the paper contributes to questions of representation and accountability in legislative behavior.

Abstract:  Public beliefs and policy attitudes on contested societal issues frequently diverge from empirical evidence and scientific consensus. In a preregistered experiment with  6,558  American  participants,  we  tested  the  generalizability  of  evidence-based  AI  dialogue  as  a  tool  to  shift  beliefs  and  policy  support  across  three structurally distinct societal issues (climate change, vaccination, and economic inequality), and examined for whom effects are strongest. Participants engaged in interactive dialogues with GPT-5 targeting beliefs, policies, or both. Compared to  a  neutral  control  conversation,  AI  dialogues  shifted  beliefs  across  all  three domains,  suggesting  generalizable  persuasive  potential,  and  increased  support for politically contested policies, especially carbon and estate taxes. Estimated effects  were  strongest  among  initially  skeptical  participants,  who  are  typically hardest to reach, and those with higher trust in science. Text analyses revealed that GPT-5 adapted its approach to skeptical participants. Brief AI dialogues may  thus  function  as  a  scalable,  generalizable  tool  for  evidence-based  public communication.

Abstract: Research on social policy and target populations holds that group perceptions affect the allocation of benefits and burdens. Building on theories of social construction, we investigate which social groups legislators perceive as powerful and deserving, and how these perceptions translate into discussions of policy design. Our research design employs a novel combination of closed and open-ended questions in a field survey with over 500 elected representatives. Legislators classify recipient groups in terms of their power and deservingness and then express policy preferences for these groups. A text analysis of open-ended questions focusing on policies for people with disabilities and with addictions – where the former is perceived as more deserving than the latter – shows that group perceptions influence discussions of policy design: legislators discuss general treatment and benefits for disabled people at greater length than for addicts. Using embedding regression, we demonstrate that different perceptions of power and deservingness shift the meaning of relevant target words for both social groups. Policymakers' conceptions of recipient groups inform their policy preferences. 

Abstract: Three determining factors for economic inequality are self-chosen effort, self-chosen risk, and external circumstances. The fairness people assign to inequalities due to effort and external circumstances is widely studied. Insights on the fairness of inequalities due to self-chosen effort and self-chosen risk, however, are lacking. I study a novel experimental setting where inequality is due to a choice over effort-provision and a choice over risk-taking. While the resulting inequality is mostly seen as fair, around 10% of third-party redistribution decisions are in line with a fairness norm that only considers the choice over effort.

Abstract: Research on citizens shows that views about what constitutes a fair distribution of economic outcomes affect inequality perceptions and support for redistribution. Yet it is legislators - and their fairness preferences – who ultimately determine political responses to inequality. This study is the first to elicit the fairness preferences of politicians. We do so using a canonical experimental redistribution game, which we employ in two distinct politician samples, resulting in over 2,000 individual observations across six advanced democracies. We rely on two distinct legislator samples: Study 1 surveys locallevel politicians in the US and Norway (N=1,591). Study 2 expands the design and in face-to-face interviews surveys primarily national-level legislators across five European countries (N=501). Across both studies, politicians redistribute substantively less when inequality is due to merit than when it is due to luck. Study 2 further shows that on the individual level, 70% of politicians act as meritocrats, compared to only 50% of citizens, and that politicians’ adherence to the meritocratic fairness ideal is shared across political orientation. Differences exist in the level of inequality conservative and liberal respondents find acceptable. We also document an asymmetric representation gap: Liberal politicians are less accepting of inequality than their voters, irrespective of its source. The main patterns are highly consistent across studies and country contexts. By revealing the fairness preferences of politicians in several advanced democracies, the study contributes to a supply-side explanation on limits to political action to address inequality.