We conduct parallel surveys of legislators and citizens in three countries to study their tolerance for corruption. In Italy, Colombia, and Pakistan legislators and citizens respond similarly to hypothetical scenarios involving trade-offs between, for example, probity and efficiency: both perceive corruption as undesirable but prevalent. These novel descriptive data further reveal that legislators generally have accurate beliefs about public opinion on corruption and understand its relevance to voters. An informational treatment updates legislators’ beliefs about public opinion. The treatment produces downward adjustments among legislators who initially overestimated citizens’ anti-corruption preferences. We also present descriptive data that tolerance of corruption is predicted by politician attributes, most notably motivations for entering politics. Finally, results reconfirm partisan bias by voters in evaluations of corruption. Overall, results suggest that barriers to effective anti-corruption policies are unlikely to lie with lack of information by legislators or by their deliberate commitment to corrupt activities
Updated: Aug 13
Co-authored with Tara Slough et al.
On what basis can we claim a scholarly community understands a phenomenon? Social scientists generally propagate many rival explanations for the phenomena that they study. How best to discriminate between or aggregate them introduces myriad questions because we lack standard tools that synthesize discrete explanations. In this paper, we assemble and test a set of approaches to the selection and aggregation of predictive statistical models representing different social scientific explanations for a single outcome: original crowd-sourced predictive models of COVID-19 mortality. We evaluate social scientists’ ability to select or discriminate between these models using an expert forecast elicitation exercise. We provide a framework for aggregating discrete explanations, including use of an ensemble algorithm (model stacking). Although the best models outperform pre-specified benchmark machine learning models, experts are generally unable to identify models’ predictive accuracy. Our findings suggest that algorithmic approaches for the aggregation of social scientific explanations can outperform human judgement or ad-hoc processes.
Updated: Aug 13
Co-authored with Saad Gulzar and Luke Sonnet
We report results of a randomized control trial conducted in Pakistan that uses Interactive Voice Response (IVR) to augment existing face-to-face communication between politicians and voters. IVR allows politicians to script questions for voters and voters to respond on cell phones. Politicians initially exhibit willingness to engage via IVR, commit- ting to recording multiple rounds. However, the intervention unexpectedly transformed from intensive to light-touch when politicians uniformly withdrew after a single round. Drawing on three dozen open-ended interviews, we find the about-face was triggered when voters misinterpreted communication as policy commitments by politicians. Politicians lack ade- quate resources to deliver services requested by voters. That the intervention backfired for politicians highlights objective constraints on service delivery in poor countries that hinder responsiveness by even well-intentioned political representatives.