↓ 跳过正文

政治行为

State-sponsored agenda setting: Measuring the impact of narrative laundering

PNAS 123/36 pp. e2534920123-e2534920123 2026-09-04 Original State-sponsored agenda setting: Measuring the impact of narrative laundering Patrick L. Warren, Darren L. Linvill, Camille J. Saucier Abstract Foreign information manipulation and interference (FIMI) is an important part of the digital information ecosystem, yet its effects are poorly understood. Most research analyzing FIMI to date is descriptive, exposing efforts to erode trust, promote favorable narratives, and target specific communities. Existing work evaluating the impacts of FIMI contrasts those directly exposed to FIMI messaging (by viewing the original messages) with those who were not, finding little evidence of changes in beliefs or behavior. Rather than persuading individuals directly, we argue FIMI often operates through agenda-setting processes that shape which issues and narratives dominate public discourse. A key component of this process is narrative laundering, whereby misleading content is scrubbed of its origins and circulated into mainstream channels. This study examines the Russian-aligned Storm-1516 campaign, which uses state media, fabricated sites, and paid influencers to elevate false narratives, especially those that implicate Ukrainian President Volodymyr Zelenskyy in corrupt acts. By analyzing the dissemination of these narratives, we assess how Storm-1516 shifts the public agenda, illustrating how foreign influence can exert substantial effects on information environments even when individual-level persuasive effects may be limited.

Leveraging generative AI for causal inference with unstructured data

PNAS 123/36 pp. e2530532123-e2530532123 2026-09-03 Original Leveraging generative AI for causal inference with unstructured data Kosuke Imai, Kentaro Nakamura Abstract We introduce GenAI-Powered Inference (GPI), a statistical framework for causal inference using unstructured data, including text and images. GPI leverages open-source pretrained Generative AI (GenAI) models-such as large language models and diffusion models-not only to generate unstructured data at scale but also to extract low-dimensional representations that are guaranteed to capture their underlying structure. Applying machine learning to these representations, GPI enables estimation of causal effects while quantifying estimation uncertainty. Unlike existing approaches to representation learning, GPI does not require fine-tuning of GenAI models, making it computationally efficient and broadly accessible. We illustrate the versatility of the GPI framework through three applications: 1) estimating the effects of Chinese social media censorship while adjusting for textual confounders, 2) isolating the impact of specific image features from that of other correlated features in the same image, and 3) assessing the persuasiveness of political rhetoric. An open-source software package is available for implementing GPI.

Feminized Revival: Han Chinese Dress, Gender and Cultural Nationalism in Contemporary China

CQ — pp. 1-17 2026-09-01 Original Feminized Revival: Han Chinese Dress, Gender and Cultural Nationalism in Contemporary China Fuyun Wei Abstract Abstract Women hold disproportionate authority in China’s contemporary Hanfu revival, where Han-ness operates through implicit cultural codes rather than explicit ethnic claims. Recent scholarship highlights how Han-ness circulates as a moral, etiquette-centred grammar of “proper culture.” Drawing on 18 months of ethnography across six sites in China, this article shows how gender mediates that grammar and shapes the performances of Han-ness. Women, as arbiters of taste and propriety, enact an assertive banal nationalism through aesthetic and organizational labour. Men, marginal within the field, perform defensive overt nationalism through didactic historical narratives and moralized claims to legitimacy. To explain this pattern of identity formation and boundary making, I advance the notion of gendered everyday nationhood. The article contributes to scholarship on nationalism by showing how embodied practices sustain national belonging, and how mundane evaluation and boundary policing reproduce nationhood in everyday life.

Are Voters Myopic or Looking for Improvement?

APSR — pp. 1-18 2026-09-01 Original Are Voters Myopic or Looking for Improvement? MARKUS PRIOR, TALBOT M. ANDREWS Abstract Abstract Timing of valence outcomes plays an important role in key theories of electoral choice, and different models place different emphases on past and future outcomes. Whereas sanctioning models consider only retrospective evaluations as an accountability mechanism, selection models hold that voters use past and proposed future conditions to infer the incumbent’s ability to deliver benefits. Drawing on four surveys run on six different samples, we demonstrate that voters prefer improving conditions in both retrospection and prospection. In retrospection, this pattern is difficult to distinguish from overweighting of temporally proximate conditions, which gave rise to the notion of “the myopic voter.” Distinguishing the two processes, we find that people do not ignore early datapoints and reward improving conditions (in both retrospection and prospection). This preference for improvement does not simply reflect an uncritical heuristic for upward trends. People are not universally myopic and many genuinely value improving aggregate conditions retrospectively and prospectively.

Subversive Amplification: Online Dissent and the Limits of Information Control in China

CQ — pp. 1-18 2026-08-28 Original Subversive Amplification: Online Dissent and the Limits of Information Control in China Christian Göbel Abstract Abstract Intensified information control severely constrains digital contention in China, yet online mass events occasionally occur. Using a dataset of 274,721 Weibo messages, this study examines mobilization, coordination and grievance articulation during a two-day episode of online dissent in the aftermath of the 2022 Urumqi residential building fire, which preceded the White Paper protests. It argues that users were activated by public communication failures that transformed manifold grievances into outrage, and that the associated hashtags provided anchors to which criticism could be attached. In a process this study terms “subversive amplification,” concentrated engagement amplified the hashtags’ visibility, and with it the critical content attached to them. However, absent the unifying function of opinion leaders, grievance articulation remained fragmented and did not converge on unified demands. The case illustrates the inherent vulnerabilities of information control but also the conditions and limits of their exploitation.

Personality pairing improves human–AI collaboration

PNAS 123/35 pp. e2530627123-e2530627123 2026-08-28 Original Personality pairing improves human–AI collaboration Harang Ju, Sinan Aral Abstract Here we examine how AI agent "personalities" interact with human personalities to shape human-AI collaboration and performance. In a large-scale, preregistered randomized experiment, we paired 1,258 participants with AI agents prompted to exhibit varying levels of the Big Five personality traits. These human-AI teams produced 7,266 display ads for a real think tank, which we evaluated using 1,168 independent human raters, and a field experiment on X that generated nearly 5 million impressions. We found that human and AI personalities individually shaped ad quality and teamwork and that human-AI personality pairings directly influenced ad quality. For example, extraverted humans paired with conscientious AI produced the lowest quality ads, followed by conscientious humans paired with agreeable AI and neurotic humans paired with conscientious AI. In the field experiment, ad quality significantly influenced ad performance, measured by click-through rates and cost-per-click. Together, these results demonstrate that personality pairing can improve human-AI collaboration and performance. They also motivate future research on the complex implications of AI personalization for human-AI collaboration, teamwork, and performance.