Deliberation has been prescribed as a cure to many of democracy's ailments, including misinformation, political apathy, and populism. We test if informed discussions regarding matters of common concern can be scaled online, without losing their democracy-enhancing qualities. In particular, we evaluate if people are persuaded by better arguments as deliberation theory intends, or by other social dynamics that are incidental to the discussion process. Our evidence stems from a large field experiment with Nepali citizens, who deliberated national development policies on an online platform or in-person. Despite notable institutional differences, both modes of deliberation caused similar opinion change and gains in knowledge, efficacy, and empathy. Our analysis of the discussion transcripts will show if these changes can be explained by `the unforced force of the better argument' as intended, or if the legitimacy of deliberation's outcomes rests on less democratic mechanisms.

What could be the role of deliberation in democracies where political patronage holds sway? We consider Nepal after the `Gen Z' protests as a case study of ongoing transition between patronage and programmatic politics. Using data from a field experiment that invited 600 Nepalis from across the country to make policy recommendations after online or in-person deliberation, we gauge if such minipublics could advance policy-making based on collective interests, or serve the interests of particular interest groups. Our analysis suggests that preference change is not driven by ascribed identities, that deliberation did not distort policy preferences, and that a majority supported policy recommendations that could feasibly guide decision-making. We discuss the replicability of these findings with respect to different functions of deliberative minipublics in a setting where patronage is a prevailing feature of elective democracy.

The world's 475 million smallholder farmers, who produce around one third of its food, are the most vulnerable to climate change. To scale the adoption of sustainable agricultural practices among them, development organizations have turned to digital advisory services. With a field experiment in Madhya Pradesh, India, this study tested two mechanisms of behavior change: one grounded in social norms, the other in agricultural economics. We developed a copy of a popular WhatsApp chatbot that delivered market price information to include sustainable land management advisory, and attempted to persuade 800 randomly selected farmers to engage with costly sustainable practices. Despite monetary incentives, study participants were not persuaded to engage with land management content. They paid no more attention to messages framed around dynamic norms or market prices than to a control message, and the limited attention we observed did not translate to engagement. These results suggest that individual digital education is insufficient to scale counter-normative behavior, casting doubt on a prominent approach to promoting sustainable agriculture.

Experimental evidence largely supports the notion that selective punishment elicits conditional cooperation, yet empirically punishment is frequently indiscriminate, not perceived as justified, and defection persists among political deviants who are punished. This paper presents results from a lab experiment that tests the effect of punishment on loyalties to rival groups in conflict.

"Loyalty trials" are common to a range of conflict settings, with consequences that range from harassment to imprisonment, torture, or death. Yet, they have received little if any attention as a general phenomenon in studies of state repression, civil war, or rebel governance, which focus on particular behaviors that authorities use to put people on trial, such as dissent, defection, and resistance. Using a computational model and data on the German Democratic Republic and the Occupied Palestinian Territories, we focus on the dynamics of "loyalty trials" held to identify enemy collaborators—the interaction between expectations, perceptions, and behavior. We use our framework to explore the conditions under which trials result in widespread defection, as in the German Democratic Republic, or in conformity as illustrated by our study of the Occupied Palestinian Territories. The polarizing nature of loyalty trials and the propensity to over- or under-identify threats to political order have notable implications for democratic and non-democratic societies alike.

Anticipating those most at-risk of being acutely malnourished significantly shapes decisions that pertain to resource allocation and intervention in times of food crises. Yet, the assumption that household behavior in times of crisis is homogeneous—that households share the same capacity to adapt to external shocks—ostensibly prevails. This assumption fails to explain why, in a given geographical context, some households remain more vulnerable to acute malnutrition relative to others, and why a given risk factor may have a differential effect across households? In an effort to explore how variation in household behavior influences vulnerability to malnutrition, we use a unique household dataset that spans 23 Kenyan counties from 2016 to 2020 to seed, calibrate, and validate an evidence-driven computational model. We use the model to conduct a series of counterfactual experiments on the relationship between household adaptive capacity and vulnerability to acute malnutrition. Our findings suggest that households are differently impacted by given risk factors, with the most vulnerable households typically being the least adaptive. These findings further underscore the salience of household adaptive capacity, in particular, that adaption is less effective for economic vis-à-vis climate shocks. By making explicit the link between patterns of household behavior and vulnerability in the short- to medium-term, we underscore the need for famine early warning to better account for variation in household-level behavior.

An evidence-driven computational modeling (EDM) framework rests on three methodological pillars: agent-based computational modeling (ABM), empirical contextualization using geographical information systems (GIS), and empirical validation. The approach is especially useful in issue areas where there is an abundance of theoretical knowledge, outcomes are driven by complex interactions between numerous factors, and it is possible to leverage empirical data to seed or validate the model. Ideally, EDM provide evidence-driven results that decision-makers can use to evaluate alternative policy options in a systematic and transparent manner.

Read More

During the cold war, civil conflict had a rural bent, which research mirrored [1]. Urban environments were traditionally viewed as undermining identifications that provide an impetus for fighting [2], too well protected as the home bases of elites and even prohibitive to rebel operations [3]. As the world population grows and increasingly clusters in urban spaces [4], we argue that conflict will be redirected — whether purposefully or unintentionally — to cities. Results from several recent studies provide substantial support for a nascent urban propensity towards conflict — an emerging urban shift.

Read More