Sören Harrs

Absent high-quality online data, research questions would be constrained conceptually and in study populations. To inform the debate about online data quality, this paper provides empirical evidence that compares data quality of responses from online participants, AI agents, and human subjects in the lab.  Corresponding results support data quality on some platforms, but not others. This paper also highlights a viable path for high-quality online data in an evolving landscape: use a two-stage recruitment method to broadly recruit online subjects in a baseline study and then limit recruitment for the main study to the resulting subset of ``high quality'' subjects. 

Noise poses a threat to many empirical findings in economics. This registered report proposes a design to test whether incentives are effective and efficient in reducing noise in experimental games. In an online experiment, subjects complete identical economic games at two points in time in one of four treatment conditions: high incentives ($20), low incentives ($2), random incentives ($20 with 1 in 10 probability), or no incentives (hypothetical). The repeated observation of choices allows us to compare the level of noisy decision-making across treatments to provide causal evidence how effective incentives are in reducing noise. To evaluate the cost-efficiency of incentives, we conduct a horse race with other research methods commonly used to reduce noise in experiments and online surveys. Taken together, this design aims to inform best practices on how to reduce noise in economic research, and to shed light on the determinants of noisy decision-making.

Can voluntary contributions to public goods be motivated by identity concerns? In a theory-driven field experiment we test how positive and negative shocks to subjects’ environmental identity beliefs affect voluntary efforts for climate protection. In a real-effort task, subjects can generate donations that off-set carbon emissions. Prior to the task, we manipulate subjects’ beliefs about their environmental identity either positively or negatively compared to a control group. A negative shock to identity (“identity threat”) increases effort by about 17% compared to our control group. This effect is largest for subjects that had a strong prior environmental identity belief. We find no evidence that a positive shock to identity does affect behavior. Our results are in line with some of the main predictions from the belief-based model of identity by Bénabou and Tirole (2011). They also have implications for policy makers and NGOs that want to encourage voluntary contributions to climate protection by leveraging people’s identity concerns.

We experimentally study competitive markets with socially responsible production. Our main focus is on the producers' decision whether or not to reveal the degree of social responsibility of their product. Compared to two benchmark cases where either full transparency is enforced or no disclosure is possible, we show that voluntary and costless disclosure comes close to the full transparency benchmark. However, when the informational content of disclosure is imperfect, social responsibility in the market is significantly lower than under full transparency. Our results highlight an important role for transparent and standardized information about social externalities.