Many important decisions are made by groups because of their ability to draw on, deliberate about, and aggregate diverse perspectives. However, group decision-making can also suffer from counterproductive dynamics like herding, unearned influence, or gridlock. Given these competing forces, we study key determinants of group decision quality and test whether joint decision-making, in which members come to a shared decision after discussion, outperforms alternative decision modes. In our experiment, subjects design portfolios to maximize returns and minimize volatility, first individually and then in one of three decision modes. The modes, chosen to decompose the differences between individual and joint decision-making, are (i) Crowds, who decide alone but whose portfolios are mechanically pooled together; (ii) Panels, who communicate but whose individual portfolios are still pooled together; and (iii) Teams, who communicate and coordinate on a joint portfolio. We find that Teams outperform individuals by constructing less concentrated and less volatile portfolios. Their advantage grows when teammates hold differing market expectations. However, Teams and Panels both underperform Crowds by a similar margin, implying that communication, rather than coordination, undermines decision quality. Communication induces subjects to converge on similar portfolios, drop unique assets and sectors, and focus on returns over risk. Our findings demonstrate that joint decision-making can fail to tap the full potential of group members’ inputs, especially in settings where diversity in opinions is valuable.

Open science initiatives have gained traction in recent years. However, open peer-review practices, i.e., reforms that (i) modify the identifiability of stakeholders and (ii) establish channels for the open communication of information between stakeholders, have seen very little adoption in economics. In this paper, we explore the feasibility and desirability of such reforms. We present insights derived from survey data documenting the attitudes of 802 experimental/behavioral economists, a conceptual framework, a literature review, and cross-disciplinary data on current journal practices. On (i), most respondents support preserving anonymity for referees, but views about anonymity for authors and associate editors are mixed. On (ii), most respondents are open to publishing anonymized referee reports, sharing reports between referees, and allowing authors to appeal editorial decisions. Active reviewers, editors, and respondents from the US/Canada are generally less open to transparency reforms.

Peer review is central to the lives of researchers. We conduct a survey on improving peer review to which we received responses from over 1,400 economists. The survey is the bedrock of this article, which was written to (i) document the current state of peer review and (ii) investigate concrete steps towards improving it. We offer a snapshot of the recent submission and peer review activity of respondents, detail the difficulties they report facing, and measure their attitudes about various challenges and possible proposals to address them. We hope that this report will provide fertile ground for the development and implementation of practical solutions for improving peer review in economics.

Conducting experiments requires researchers to make numerous design choices while respecting statistical, practical, and ethical constraints. We develop the Experimental Design Assessment Tool (EDAT) to help researchers navigate this complexity when generating and evaluating evidence from experiments. EDAT is based on a comprehensive framework that distills experimental designs into 30 components and provides guidance on optimal design choices for each. We operationalize this framework with an evaluation instrument that guides evaluators through experimental designs using targeted questions and generates quantitative quality scores. We evaluate EDAT’s performance by applying it to experiments published in eight top economic journals in 2000, 2001, and 2020. Evaluators’ scores show high reliability and strongly predict citation counts, publication outcomes, and other proxies for impact and quality. This suggests that a quantitative assessment of experimental design quality is feasible. They further reveal that there cannot be a “perfect experiment” in practice due to the existence of trade-offs, such as between certain components of internal and external validity. We show how researchers navigate these trade-offs and identify the design features most valued by the profession. Finally, we document systematic improvements in quality over time and heterogeneity in design features between subfields and experiment types.

We leverage a meta-analysis to assess the accuracy of forecasts of research results relative to actual study outcomes.

We conceptualize and measure the returns to collecting forecasts of research results, and investigate the practice of eliciting them.

We conduct a robustness reproduction of Jayachandran et al. (2017) and use our findings to re-assess the benefit-to-cost ratio of Payment for Ecosystem Services programs.

In a controlled experiment with social media-like features, we study (i) whether people selectively report or misreport their successes to others and (ii) whether exposure to such biased reports harms other people’s well-being.