Research

 

The Effects of Systematic AI- and Non-AI Image Labeling on Content Credibility

survey experiment
generative AI
Bayesian learning
information

Revise and resubmit at PNAS Nexus

A registered report testing how systematic AI- and non-AI image labeling affects perceptions of content credibility.

Manuscript available on request

April 15, 2026
 

Casual Theories of Partisan Minds

laboratory experiment
political cognition
methodology
Bayesian learning
formal model
meta-analysis

Revise and resubmit at American Journal of Political Science

Do Americans see partisanship as the source of other citizens’ policy preferences, or do they believe policy preferences determine what party others choose? These two worldviews induce divergent responses to information about others’ issue concerns: if party is a cause, then different issue concerns are correlated outcomes, but if it is a consequence, then different issues are substitute explanations. Using a laboratory experiment, I find that political independents consistently interpret others’ actions according to a policy-motivated theory, but partisans apply a double standard: they attribute co-partisans’ policy preferences to individual preference, but attribute out-partisans’ policy preferences to partisanship. However, these beliefs are not immutable; as participants learn about the variability in policy preferences within each party, they increasingly attribute others’ actions to issue-motivations.

📄 View Working Paper (PDF)

March 6, 2026
 

Voting in the Dark

electoral behavior
strategic voting
asymmetric information
media trust
formal model

How do beliefs about media partisanship influence willingness to vote without information? I present theory and experimental evidence showing that beliefs about the political alignment of news media and its audience shapes whether uninformed voters decide to abstain or vote without information. In experimental elections, I find that voters cast uninformative votes more frequently when media are misaligned than when media are nonpartisan, and least when media are aligned. I find this strategic behavior primarily among Trump voters, who are 50% more likely than Harris voters to cast uninformative votes. The asymmetrically strategic behavior results in a down-ballot advantage for the candidates favored by informed Trump voters both when media are aligned and when media are misaligned.

📄 View Working Paper (PDF)

September 29, 2025
 

Deliberative Discourse and Information Transmission Under Social Pressure

political discussion
communication
partisan norms
self-censorship
social desirability bias

Social rewards and sanctions influence what political opinions people express or withhold. Therefore, they may distort perceptions of public opinion if observers fail to account for these incentives. In two online discussion experiments, we assess how well observers (Decoders) can infer the distribution of private opinions among potential speakers (Encoders). Contrary to expectations, we find that Decoders are most accurate when our social incentives are strongest. Decoders are relatively capable of recognizing and adjusting for social desirability bias and self-censorship, compared to other sources of distortion, such as reluctance to share unoriginal opinions, which arise when we minimize peer incentives.

Working paper available on request

September 5, 2025
with Valeria Burdea
 

Consensus is a Cue for Quality

survey experiment
perceived polarization
bipartisanship
formal model

Does bipartisan support for legislation signal high policy quality or undesirable policy concessions? Existing theories of elite cues offer competing predictions. We argue that this tension arises because citizens attend to both valence and spatial dimensions of policy, but the meaning of bipartisan support depends on which is focal. We develop a social learning model with focal information processing and test its implications using two preregistered survey experiments.

📄 View Working Paper (PDF)

August 18, 2024
with Matthew Tarpey
 

The slider task: an example of restricted inference on incentive effects

experimental economics
real effort tasks
methodology
laboratory experiment

Real-effort experiments are frequently used when examining a response to incentives. For a real-effort task to be well suited for such an exercise its measurable output must be sufficiently elastic over the incentives considered. The popular slider task in Gill and Prowse (Am Econ Rev 102(1):469–503, 2012) has been characterized as satisfying this requirement, and the task is increasingly used to investigate the response to incentives. However, a between-subject examination of the slider task’s response to incentives has not been conducted. We provide such an examination with three different piece-rate incentives: half a cent, two cents, and eight cents per slider completed. We find only a small increase in performance: despite a 1500% increase in the incentives, output only increases by 5%. With such an inelastic response we caution that for typical experimental sample sizes and incentives the slider task is unlikely to demonstrate a meaningful and statistically significant performance response.

🔗 Journal of the Economic Science Association (2016)

May 14, 2016
with Felipe Araujo and others
No matching items