Working Projects

  • REVERBERATE: Modelling How Preferences, Attitudes, and Attention Reverberate Between Public Opinion and Elite Policy Debate
    Abstract The traditional welfare state is under growing strain. Over the past decade, a mix of austerity, privatisation, inflation, Brexit, and the pandemic has made it increasingly difficult for governments to sustain social policies. As a result, major redistributional issues—such as wealth taxation, social care funding, and healthcare reform—have returned to the centre of political debate. Policymakers and citizens face difficult trade-offs between growth and equality, efficiency and fairness, and ideology and practicality. At the same time, trust in democratic institutions weakens, contributing to polarisation and a widening gap between citizens and political elites. Existing theories of policymaking struggle to explain how public attitudes and elite decision-making interact. This project examines how public opinion shapes political debates on redistribution, and how these debates, in turn, influence public views. It focuses on three interrelated dimensions: attention (which issues gain salience), preferences (which policies are supported), and persuasion (how actors influence one another). Core questions include how shifts in public sentiment affect government attention, what types of redistributive policies emerge, and how communication strategies mediate these dynamics. The study develops an empirically calibrated computational model of opinion dynamics, capturing interactions between citizens and elites over time. The model is grounded in multi-source data—including surveys, social media, political speeches, party manifestos, and news—and integrates methods from statistics and natural language processing for measurement, calibration, and validation. This project contributes new insights to the academic fields of social policy, political science and data science. It also has practical value: we work with political consultants and communication experts to test how our model can be used in real-world settings—to anticipate how the public might react to a proposed policy or to understand which messages are likely to resonate with different groups. By helping to bridge the gap between public opinion and political decision-making, this project supports healthier democratic processes, sheds light on how these dynamics affect the welfare resilience of different social groups—their capacity to absorb and recover from social risks. It offers tools and knowledge that can help policymakers, analysts, and civil society actors design more responsive and inclusive social policies, and ultimately strengthen trust between governments and the people they serve.
  • Measuring Attitudinal Social Cleavages Across Survey and Digital Contexts
    Abstract This project investigates how attitudinal divisions on redistributive issues are measured across survey, experimental, and digital contexts, and asks why these measures often diverge. Existing research on public opinion, political polarisation, and digital debate commonly assumes that different instruments capture a single underlying attitude with varying degrees of noise. By contrast, this project argues that disagreement across modes is itself theoretically meaningful. People do not simply reveal the same preference everywhere: they interpret issues through different schemas, express themselves differently under public and private conditions, and become digitally visible through selective processes of participation, platform affordance, and data linkage. Empirically, the project focuses on post-COVID redistributive conflict in the UK, including welfare, taxation, labour disputes, living costs, and social entitlements. It combines two stages. The first maps issue clusters, frames, and interaction patterns in digital public debate. The second links a quota-based UK survey, private preference elicitation tasks, and an embedded expression experiment that varies visibility, audience, framing, and response mode, with an additional digital-trace linkage component where feasible. This design identifies how communicative context shifts expression and how selective observability shapes the digital record. The central methodological contribution lies in a modular layered measurement architecture that treats mismatch across instruments not as residual error but as an object of inference. The model separates private orientation, public expression, observability, and channel-specific distortion, and compares this layered specification against simpler alternatives through explicit diagnostics and model comparison. In doing so, the project provides a new framework for evaluating when digital signals approximate public opinion and when they systematically misrepresent it.
  • Environmental Anomaly and Digitally Mediated Public Reactions
    Abstract This project examines how environmental anomalies—such as heatwaves and floods—shape public opinion and behavioural adaptation in digitally mediated societies. Rather than assuming direct effects of environmental shocks on attitudes, we theorise digital infrastructures and cyberspace communications as critical mediating layers that condition how anomalies are measured, interpreted, and translated into political demand and behavioural adaptiion. Advances in anomaly detection and hotspot identification show that extreme events are spatially uneven and increasingly captured through digital sensing systems (Bâra et al., 2024; Vieira Passos et al., 2024). At the same time, digital trace data reveal how public attention to such events unfolds unevenly online (Thompson et al., 2022) under the restrictions brought by digital accessibility. Research demonstrates that exposure to floods and heatwaves can shift climate beliefs and policy support (Lohmann & Kontoleon, 2023; Bergquist et al., 2025, Cologna et al 2025), yet these effects are heterogeneous and shaped by prior beliefs and information environments (Howe et al, 2019; Sambrook et al., 2021). Digital spaces play a pivotal role: extreme weather intensifies online discourse (Torricelli et al., 2023), while misinformation and polarised framing can distort attribution and influence policy preferences (Daume, 2024; Lyons et al., 2018). Integrating environmental sensing data, computational text analysis, and survey evidence, this project advances a multi-layered account of how environmental anomalies and digital mediation jointly shape public opinion and behavioural change.

Working Papers

  • Cleavage map and the focal issues: attributional divides in online redistributive discourse. Working paper.
    Abstract This paper examines how redistributive issues are framed and attributed in politically engaged online discourse. We collect data from Reddit, focusing on debates within the UKPolitics subreddit from 2020 to 2025, and focus on the discussions of council tax, universal basic income, minimum wage, income tax, national insurance, and poverty. We use BERTopic to recover the main thematic structure of the corpus, and apply large language models (LLMs) to annotate comment stance and attribution in relation to the argument advanced in the original post. The analysis shows that online redistributive discourse is organised less by a simple pro- versus anti-redistribution divide than by issue-specific attributional patterns. Government and political parties dominate users’ attribution in tax-related debates, while minimum wage discussions more often centre on employers and labour-market dynamics. These attributional patterns also vary around major policy events, with several high-salience announcements followed by shifts from partisan attribution toward institutional government blame. We further present two conversation-tree case studies that help illustrating how these broader patterns appear within nested exchanges. The paper contributes to the policy studies with a computationally informed account of how redistributive discourse is organised by attribution and issue-specific framing in a politically engaged online arena, it also shows how these patterns speak to the changing legitimacy and moral language of redistribution in the contemporary UK.
  • A Tale of Two Discourses: Social Media Discussion on Pension Reforms in China. Working paper.
    Abstract As a core redistributive policy domain, pension reform has long occupied a central place in political agendas and public debate. The retirement age adjustment in China provides a particularly illustrative case. First proposed in 2013, the reform underwent nearly a decade of deliberation before its formal announcement in late 2024. This prolonged gestation period generated sustained public attention and speculation regarding the reform’s scope, implementation, and eventual materialization. Using 184,667 Weibo posts published between 2013 and 2024 that reference retirement-age reform, this paper analyzes how China’s retirement-age reform is discussed on social media by verified institutional and individual users and assesses whether—and in what ways—state and citizen discourses diverge. We find that while both state and citizen discourses converge around macro-diagnostic issues, such as population aging and mounting pension system pressures, substantial divergence emerges on micro-prognostic issues, in which debates center on agency, responsibility, and the distribution of reform burdens. Moreover, state discourse predominantly employs technocratic and economic frames that emphasize risk pooling and system sustainability. By contrast, citizen discourse foregrounds personal choice and distributive justice, reflecting a more humanistic and socially grounded framing. This divergence in framing and underlying rationales weakens the prospects for decisive reform by limiting the emergence of a shared cognitive foundation for policy change.
  • Ting LUO, Yan WANG. 2025. The art of storytelling: Pro-regime narratives on Chinese social media and the official opinion leaders. Under Review. Available here
    Abstract This article examines how authoritarian regimes construct state narratives about politically consequential events. Building on the narrative policy framework and existing research on authoritarian propaganda, we propose two dimensions that shape narrative construction: legitimacy implications -- whether events enhance or threaten regime legitimacy, and citizen verification capacity -- the extent to which citizens can evaluate official narratives through alternative sources. Using quantitative narrative analysis of Chinese social media posts by government, state media, and celebrity accounts, we extract subject-verb-object (SVO) triplets to map dominant narrative structures across four major events. Our findings show that legitimacy implications of the event shape regime's efforts in storytelling and the beliefs highlighted in the narratives, while citizen's verification capacity could balance the strategic choice between a top-down manipulation and bottom-up responsiveness of state narratives. Together, the results reveal propaganda as a complex process of narrative construction adaptive to specific contexts, offering new insights into how dynamic storytelling sustains authoritarian resilience.
  • Avalos, C., Nick Shryane and Yan Wang. 2025. When Does Food Label Readability Matter? A Non-Equivalent Dependent Variable Design to Isolate the Content-Specific Effects on Food Choice. Preprints. Available here
    Abstract Objective: This study investigates the association between consumers’ perceived readability of Multiple Traffic Light label print size and the frequency of food consumption within the United Kingdom. We aimed to disentangle the effect of label readability from label content. Using non-equivalent dependent variables, we test if the association is specific to unhealthy convenience foods and absent for healthy or unlabeled foods. We also test for heterogeneity across consumer subgroups. Methods: Data from 8,948 adults across four waves (2012-2018) of the UK Food and You Survey were analyzed. Cumulative link ordinal logistic regressions were employed to model the association between self-reported print size readability and the consumption frequency of four product types: pre-package sandwiches and precooked meat (unhealthy, labeled), dairy (healthy, labeled), and fresh meat (unlabeled control). Results: The findings reveal a content-specific and significant dynamic relationship exclusively for pre-package sandwiches. This relationship reversed from a 9% decrease in the odds of frequent consumption in 2012 to a 4% increase by 2018. In contrast, a persistent null association was observed for pre-cooked meat, dairy, and fresh meat. Subgroup analyses for sandwiches indicated that the association with readability was strongest among less-engaged consumers. Conclusion: Empirical evidence challenges the utility of a standardised approach to food labelling, suggesting the effectiveness of label salience is contingent not just on the consumer, but on the product’s context and the content of its message.
  • Pol Rovira-Wilde, Yan WANG. Towards a Natural Language Processing Pipeline for Measuring Types of Issue-specific Polarisation using UK Party Manifestos. Working paper.
    Abstract This paper presents an NLP pipeline that measures both types of issue-specific polarisation (ideological and salience) between political parties using manifestos. Whilst current NLP methods either conflate or only measure one type of issue-specific polarisation, this study's pipeline measures both types independently, allowing one to explore the potential relationship between different types of polarisation. Furthermore, the pipeline addresses other limitations with current methods such as confounding by lexical overlap and concealing variation in polarisation across issues. To provide a framework for analysing the pipeline, its outputs are used to identify which issues are proprietal, conflictual or consensual. Using the UK Labour and Conservative parties' General Election manifestos from 2001-2024, the pipeline segments manifestos into sentence-level semantic units and then applies BERTopic so that each sentence is labelled with distinct issue topics. Issue-salience polarisation is measured by comparing normalised topic frequency, whilst ideological polarisation is measured using ManifestoBERTa embeddings and cosine dissimilarity. Results show that issues like Healthcare were consensual as they exhibited low ideological polarisation, in contrast to conflictual issues like Immigration. Additionally, the EU was especially polarised in salience during the 2010's, making it a proprietal issue despite ideological polarisation. Future enhancements could include integrating NLP tools into the pipeline that allow for a qualitative analysis of polarised issues to supplement its existing quantitative analysis.

Book

  • Yan WANG. Pension policy and governmentality in China: Manufacturing public compliance. 2022. LSE Press. Available here
    Book Blurb Economic growth is often a disruptive social process - so how has the Chinese state been able to maintain compliance from its people while at the same time pushing ahead an exceptionally rapid social and economic transformation? This book explores the question via detailed analysis of the trajectories, policy rationale, and effects of China’s pension reforms, demonstrating how statecraft shapes the ways that citizens ascribe credit and responsibility for pensions protection across themselves, the state and other actors. The book shows that China’s governmentality for manufacturing compliance is hybrid, organic, and dynamic. The targeted allocation of benefits, policy experimentation, propaganda and knowledge construction, and many other approaches are used to shape public expectations and to justify state rule. An original contribution to the study of legitimation in modern states, the analysis particularly highlights that when active counter-conduct (such as resistance) is confined, individuals may choose cognitive rebellion and falsify their public compliance.

Articles

  • Yan WANG. Building deservingness and fairness with discourse: How states construct the social legitimacy of pension schemes. forthcoming. Critical Policy Studies. Available here
    Abstract How do modern states legitimise and frame their redistributive social policies in communicating with society to promote policy changes? What kind of ‘structures of incentives’ are being promoted in their narratives? This article takes the pension reform in China in the 1990s and the 2000s as an example, using text data on corresponding official propaganda to investigate the Chinese state’s ideological repertories in framing and crafting the social legitimacy of its then-new pension designs. The results from Quantitative Text Analysis of official discourse demonstrate that the rationale for reconstructing public expectations of the redistribution of pension benefits and the allocation of welfare responsibility between the government and individuals mainly focuses on the reiteration of the principles of: ‘contribution and rewards’ and ‘rights and obligations.’ The state also built images of deserving and undeserving social groups by reconstructing the notion of fairness, blurring the distinction between merit and equity, and ultimately reshaping individual subjectivity as a self-regulated and self-motivated ‘socialised self’.
  • Xufeng ZHU, Yan WANG. Policy Experimentation as Communication with the Public: Social Policy, Shared Responsibility, and Regime Support in China. 2023. The China Quarterly. Available here
    Abstract Traditional wisdom on policy experimentation has mainly focused on central-local relations. However, scholars have paid little attention to the interaction between policy experimentation and the public. We argue that policy experimentation can be adopted by decision makers as a communication instrument with the public, facilitating the building of a social consensus regarding controversial policies. We evaluate the effects of the Chinese government’s efforts in promoting shared responsibility between the state and the individuals for the urban pension system with policy experimentation on public’s regime support. Evidence from two rounds of a nationwide survey conducted before and after the policy experiment indicates that the implementation of policy experiment has significantly contributed to citizens’ acceptance of individual welfare responsibility. Moreover, the image building of governmental responsibility via official news with varied intensity across local regions immediately consolidates the political trust of residents while posing threats to government credibility in the long run.
  • Yan WANG, Ting LUO. Politicizing for the idol: China’s idol fandom nationalism in pandemic. 2023. Information, Communication, & Society. Available here
    Abstract Chinese idol fans have been identified among the main forces in cyber nationalist activisms in recent years, acting as the nationalist fans protecting the state as an idol in response to external political shocks. Their skills in acknowledging, involving, and even reinventing the image of the state and national pride in cyber nationalist activisms do not emerge in a vacuum. This article examines how idol fans involve and reinvent the nationalist discourse in their everyday fan activities – idol promotion. We focus on the pandemic in 2020 as it provides a specific social and political context that allows us to understand better the interaction between idol fans and the state in their mundane fan activities. We construct our analysis under the computational grounded theory framework with over 6 million fan posts collected from Weibo and 11 in-depth interviews with active idol fans. Our findings show that when engaging in pandemicrelated discussion, idol fans actively borrowed official discourse on nationalism and strategically responded to key political and social events in their idol promotion activities. The idol images they built are not only positive but also nationalist. Therefore, they play not only the commercial logic commonly seen in the Japanese and Korean Kpop/idol culture but also the political logic propagated by the state in China.
  • Xiaonan WANG, Yan WANG. Too Cynical: How Stock Market in China Dismissed Anticorruption Signals. 2022. Journal of Chinese Political Science. Available here
    Abstract Political leaders in China regularly launch anti-corruption campaigns to win public support. But how are anti-corruption signals perceived? We use event study to examine the case of Xi Jinping’s anti-corruption campaign – an unprecedented effort in China to fight corruption. Contrary to expectations, we find that for the firms with connected officials later investigated, the initial anti-corruption signals – speeches from the top leadership and earlier crackdowns on other senior officials – did not decrease their stock prices. We argue that the perceived high costs of following through and repeated campaigns in the past paradoxically nurtured cynicism. We exploit the case of Zhou Yongkang and Ling Jihua – the two officials who were alleged to be involved in the power struggle and whose downfall had circulated widely since 2012. We find that when the targets of earlier crackdowns were connected to Zhou or Ling, the stock prices of the firms went down only if their connected and later investigated officials were in the same faction; the stock prices of the other firms, however, went up. We interpret the results as investors’ misperceptions of the campaign in the beginning. Our findings suggest that even real efforts in campaign-style enforcement can be dismissed.
  • Yan WANG. Pragmatism or politicism: Local officials’ decision making in policy experimentation. 2022. SSRN Available here
    Abstract It has been widely recognized that local bureaucrats are crucial actors in policy process. In policy experimentation—a popular policy instrument in social welfare areas—which heavily relies on negotiation and interaction between different sectors, local bureaucrats are the main actors that initiate the experiment plan, propose policy innovation, and implement the pilot schemes. Then what do they value when deciding on local social policies, and why would they prefer some policy-experiment schemes over others? In this research, we use two unique studies with survey experiments on municipal- and county-level government officials in China and investigate their rationale and attention allocation on social policy preferences, as well as their decision making on policy experiments. Our results show that although the instruction and support from the upper-level governments are as vital as the local initiatives, local officials are more practical than political in many scenarios of local social policy making, where under similar conditions they react more strongly to societal demands. This pragmatism is especially true in deciding the preferred pilot scheme—they place more value on financial support, local conditions, and risk environment, while the political load of the pilot schemes have relatively less leverage in changing their preference. More importantly, such a pattern is consistent across different administrative types and regional subgroups of local officials.

Book Chapters

  • Yan WANG, Yuxi Zhang. Contesting for Consensus: Social sentiment towards fellow citizens’ COVID-related behavior in China. 2023. forthcoming: in Book ”Pandemic Narratives in China and the World: Technology, Society, and Nations”, edited by Bingchun Meng, Guobin Yang, and Elaine Yuan. Available here
    Abstract COVID-19 has shifted how citizens interact profoundly. Private life is frequently dis- played in the public space and individuals are held to account should their exercise of liberty enlarges COVID-19 transmission risks. We are interested in the evolving dynam- ics among fellow citizens, especially when and how individuals react to others’ COVID-19 related actions and behavior. An extensive data set of Sina Weibo posts consisting of more than four million COVID-19 related posts provides us with a lens to answer the questions. By estimating the general sentiment of Weibo posts from January to Decem- ber 2020, as well as two in-depth case studies, we capture the information flows and discussion volumes in the public space. Combining the machine learning approach with discourse analysis, we find that the psycho-social cycle model identified in past public health emergencies and other societies during the COVID-19 pandemic also occurred in China, although demonstrating unique timing and sequence characteristics that are linked to China’s epidemic situation and policies. The all-society solidarity built at the begin- ning of 2020 was later challenged, and potentially eroded by the process of moralizing fellow citizens’ COVID-related behavior via blaming, discriminating, and scapegoating. As society lives under the pandemic for longer, fellow citizens have become more aware of problems associated with unbounded public scrutiny of private life. Such awareness and reflection, herein, encourages discussion and consensus building efforts.

    Online Appendix
    A blog about the paper

Research Grants

  • 2024-2025 Cathie Marsh Institute Seed Corn Fund (PI), University of Manchester.
  • 2024-2025 Humanities Strategic Civic Engagement Fund (Co-I), University of Manchester.
  • 2023-2024 School of Social Sciences Small Grant (PI), University of Manchester.
  • 2021-2023 Small Research Grants (Co-I), British Academy.
  • 2020-2021 Research Infrastructure and Investment Fund (RIIF), LSE Methodology.