Steven Z. Bao

Department of Sociology , THE Ohio State University

Home    Research    Teaching    Software

Research

My research lies at the intersection of political sociology, social movements, organizations, and computational methods. Across my work, I examine how collective actors acquire power, sustain collective action, and produce political change under different organizational and institutional conditions.

1. The Organizational Infrastructures of Collective Action

This line of research is built around an original dataset of approximately 40,000 formal and informal movement organizations in the United States. By linking protest participation to nonprofit tax records, funding and collaboration networks, congressional lobbying, campaign contributions, and bill-level legislative outcomes, I trace how movements organize resources, sustain coalitions, and translate organizational capacity into political influence.

  • "Power Structure and the Maintenance of Social Movement Coalitions." — drafts available [Abstract]
    • Abstract
      Social movement scholarship offers a well-developed account of how coalitions form, but a far thinner one of what becomes of partnerships after their initial joint action. I argue that coalition maintenance runs on a different currency from coalition formation: not the affinity that brings organizations together, but the distribution of power and capacity among the organizations doing the sustaining. Drawing on theories of coalition instability and resource dependence, I distinguish two objects of structural effect: the power structure of the coalitions a pair jointly enters and the power asymmetry between the partners themselves. I test their competing expectations on 10,034 recurrent organizational dyads constructed from more than 33,000 coalition protest events in the United States between 2017 and 2026. Power structure predicts the pattern of continued collaboration, including its tempo, steadiness, and volume, far better than organizational similarity does. Yet its associations do not form a single continuum of stronger or weaker maintenance. Asymmetric partnerships reconvene less often and accumulate less joint activity, but keep steadier schedules: a lower-intensity, more routinized rhythm. The weaker partner’s alternatives condition both dyadic schedule associations, buffering the tempo penalty while sharpening routinization, but leave coalition-level associations unchanged. Power, I conclude, reorganizes maintenance across dimensions and structural levels rather than uniformly sustaining or eroding it.
  • "The Three Arms of Social Movements: How Social Movement Organizations Shape Policy Outcomes." — data collection
  • "Funding the Movement: Interorganizational Resource Flows and the Division of Labor in Social Movements." — data collection

2. State, Conflict, and Contentious Politics

This line of research examines how interactions among collective actors, states, and political audiences shape political outcomes across diverse institutional settings. Across projects on protest, political violence, transitional justice, and mass-atrocity prevention, I study how states respond to challengers, how contentious actors become visible, and how can we better understand conditions under which mass atrocities occur worldwide.

  • 2026. Beck, Colin J., Eric Schoon, and Steven Bao. "All the Terrorism Fit to Print: A New Dataset on Media Coverage of Militant Groups."
       Critical Studies on Terrorism: 1-26. [Abstract] [PDF]
    • Abstract
      That news media coverage is an important aspect of terrorism has long been recognised. Yet quantitative research on the phenomenon has been limited by partial and selective data. This article introduces a new comprehensive dataset to address this shortcoming: All the Terrorism Fit to Print. The dataset features information about 1551 militant groups found in the Global Terrorism Database and reports their annual coverage rates from 1970 to 2019 in the New York Times, the Times of London, and the Associated Press. Key trends of media coverage are explored by time and space, organisational activity, key tactics, and group ideology. Regression analyses of the determinants of this English-language news coverage find that annual media coverage is affected primarily by the number of victims, the number of attacks, and whether an organisation targeted Americans.
  • 2025. Bao, Steven Z. "Tactics and Targets: Labor Protest and State Response in China."
       The Sociological Quarterly 66 (1): 3-30. [Abstract] [PDF]
    • Abstract
      Repression has long been regarded as the main response of the state when encountering contentious actions. Although recent studies point out that repression may vary by regime types and state-level concerns regarding international reputation, few analyses to date have analyzed how movement dynamics and movement targets jointly shape state responses. In this article, I analyze over 7,500 episodes of labor protests in mainland China to advance scholarship on state responses to social movements. I develop a typology of state responses, from active coercion to accommodation and even tolerance. I then explore differences in state responses based on protest tactics and whether they are against private or state-affiliated targets. My results show that state responses differ based on the tactics that movements employ. Furthermore, these effects are moderated by whether activists target state-affiliated or private entities. This research offers important insights into the dynamic relations between state and social movement, the diverse ways highly bureaucratic governments use to manage social unrest, and the conflict between maintaining legitimacy and social control in both authoritarian and democratic contexts.
  • Forthcoming. Nzitatira, Hollie N., Jack Wippell, and Steven Z. Bao. "Judging the Past: How Rwanda's Gacaca Courts Shaped Narratives of Genocide."
       In Law, Culture, and Mass Violence: Joachim J. Savelsberg's Contributions and Beyond, edited by Ryan King. Routledge. [Abstract]
    • Abstract
      Transitional justice initiatives are central to how societies respond to mass violence, and can have longstanding political, social, and cultural legacies. One such legacy is in shaping how mass violence is remembered, as court records formalize a specific account of the events. As such, the legal institutions tasked with implementing transitional justice actively shape what forms of knowledge can be produced and remembered. Drawing on Joachim Savelsberg’s theorization of the constraints of criminal legal systems, this chapter examines how Rwanda’s gacaca courts shaped collective memories of the 1994 genocide. Using court records and extensive qualitative research with judges, defendants, witnesses, survivors, and community members, we examine how four features of criminal law—the individualization of responsibility, evidentiary procedures, legal classifications, and binary judgments—structured the narratives produced through gacaca. We argue that, while gacaca created important opportunities for communities to discuss violence and assign responsibility, its legal structure simultaneously narrowed understandings of the genocide by privileging individual culpability over structural explanations, excluding some forms of suffering, and simplifying otherwise complex experiences of victimhood and participation. These dynamics highlight how the institutional logics of the law and associated legal systems shape collective memories, even within ambitious community-based transitional justice projects.
  • "What's in a Predictor? The Case(s) of Mass Killing Forecasts."
        (with Jack Wippell and Hollie N. Nzitatira) — revise and resubmit at Journal of Peace Research [Abstract]
    • Abstract
      Scholars and practitioners have increasingly turned to sophisticated quantitative models to forecast the onset of mass killings and, in so doing, identify pertinent risk factors. While these approaches have expanded early-warning capabilities, they risk obscuring the valuable insights that can be gleaned from detailed case analyses. To bridge large-N and case-based frameworks, we apply a “regression inside out” approach to three major mass killing forecasting models. By decomposing the aggregate coefficients into their underlying case-level (i.e., country) contributions, we can identify cases that drive—or, critically, contradict—general associations that researchers commonly interpret as risk factors. Our findings suggest three categories of risk factors: generalized (risk factors where most cases affirm the average association), dispersed (risk factors where various cases both affirm and detract), and concentrated (risk factors where outliers drive the association). We caution against an over-interpretation of the latter two types as consistent risk factors. We then demonstrate how a regression inside out approach can help identify “typical” and “atypical” cases for further inspection, to more thoroughly assess underlying contingencies. In short, this article shows how high-level models are more interpretable, and thus more practical, when tethered to the cases they summarize while also suggesting avenues for extending findings from these models to case-based, qualitative risk assessments of country-level situations.
  • "Reclaiming the Flag, Remaking the Nation: Symbolic Identity Struggles in 2024–2026 Protests in Serbia."
        (with Adrijana Miladinović, Anneliese Schenk-Day, and Jack Wippell) — data collection

3. The Relational Foundations of Power and Legitimacy

What are power and legitimacy, and how do collective actors acquire, exercise, and sustain them? In this line of research, I adopt a relational framework for understanding how resources become influence, how authority becomes durable, and why both depend on configurations of actors, institutions, and shared meanings.

  • "Legitimate or Valid? Evaluating Validity as a Dimension of Legitimacy."
        (with Eric Schoon and David Melamed) — under review [Abstract]
    • Abstract
      This article examines the relationship between validity and legitimacy. Validity has long been discussed in tandem with legitimacy, but the precise relationship between the two is not entirely clear: validity is variably treated as a cause, an outcome, and a constitutive element of legitimacy. Turning to contemporary theories that treat validity and propriety as constitutive components of legitimacy, we identify three distinct conceptualizations of validity—behavioral orientation, higher-order beliefs, and social facticity. We show that, under any of these conceptualizations, validity is neither necessary nor sufficient for legitimacy: each can be present where legitimacy is absent, and absent where legitimacy is present. We then argue that the three conceptualizations are not competing definitions of a single construct but three moments of a single feedback loop, observed at different time horizons: legitimacy produces compliance; observed compliance is read as evidence of consensus; and consensus that goes unquestioned hardens into taken-for-grantedness, which feeds back into assent and expectations. This model explains how the literature could plausibly describe validity as a source of legitimacy, an effect of legitimacy, and a near-synonym for legitimacy: each characterization observes a different arc of the same loop. We conclude by cataloguing the conditions under which each conceptualization of validity succeeds—and fails—as an indicator of legitimacy, and outline directions for future research.
  • "Structural Leverage of Work: The Positional Power of Occupations and Their Labor Outcomes." — drafts available [Abstract]
    • Abstract
      Workers’ power is usually understood through labor-market scarcity or collective organization. This paper examines a third account, positional power, the leverage workers derive from their location in interdependent production networks, and extends it from industries to occupations. Linking BLS input-output data to the CPS occupation-industry matrix, I construct occupation-year measures of downstream, upstream, and local positional power for 2019–2024 and relate them to individual wages, union membership, and occupational prestige. Upstream power is associated with higher wages, higher odds of union membership, and higher prestige, all surviving two-way cluster-robust inference by occupation and industry. Downstream power is associated with lower wages and prestige under simpler inference, but these fall to borderline significance under cluster-robust inference, while its negative association with union membership remains robust. Occupational location in production networks is thus a meaningful dimension of worker power, distinct from marketplace and associational power, with consequences that diverge across economic standing, associational strength, and social esteem.
  • "Configurations of Power in Social Movement." — drafts available [Abstract]
    • Abstract
      Power is central to social movement scholarship, yet it remains undertheorized: major traditions invoke resources, disruption, political opportunities, solidarity, and public displays without clearly specifying how these dimensions relate or jointly produce political outcomes. This paper makes the causal logic of movement power explicit by reconstructing the theories of power implicit in major approaches and comparing where power is located, how it is exercised, the analytical level at which it operates, and the causal process through which it generates change. This comparison reveals three problems. Existing models privilege single causal pathways as competing explanations, divide power between capacities movements possess and opportunities they encounter, and shift among actors, relations, and fields without specifying how these levels connect. I address these problems by distinguishing four forms of power: numerical power, which converts participants and resources into pressure through signaling and disruption; positional power, which exploits locations within structures of interdependence; discursive power, which reshapes the meanings that organize contention; and performative power, which influences audiences through consequential public displays. A field-level framework shows how these forms interact and how actions reshape relationships among movements, targets, and audiences. By emphasizing configurations rather than any single resource, the framework explains why seemingly powerful movements may fail while less visible movements succeed, reconciles theoretical traditions as complementary parts of a shared causal process, and offers a systematic, multilevel account of the interactive and nonlinear ways collective action produces change.
  • "Beyond Brokerage: Labor Market Intermediaries, Job Insecurity and Worker Vulnerability." — draft available [Abstract]
    • Abstract
      How do triadic relationships generate inequality? This question is increasingly important as labor staffing agencies play a growing role in managing workforce supply and demand amid shifting employment relations. Yet, their interactions with firms and workers—and their influence on labor power—remain insufficiently understood. Drawing on ethnographic fieldwork in two staffing agencies in Shanghai and daily labor price data spanning the COVID period, I examine how intermediaries shape recruitment and the organization of work under market stress. The findings show that agencies actively shift production risks onto temporary workers during periods of market turmoil. In addition, they fragment labor power by fostering competition among workers through differential bonuses and by cultivating senses of autonomy. I conclude by considering the role of brokerage in exacerbating inequality and the broader implications of these dynamics for workplace inequality in the expanding global context of flexible employment.
  • "The Hidden Arithmetic of Sociological Inquiry."
        (with Jack Wippell and Eric Schoon) — data analysis