Recent years have witnessed many strong claims that ‘diversity’ leads to more original and impactful science, which is a science-focused form of what we call “The Diversity Hypothesis.” However, what evidence supports the claim that diversity enhances scientific output or impact? This pre-registered rapid scoping review seeks to collate and evaluate the scientific evidence for the Diversity Hypothesis as a foundation for future, methodologically rigorous studies that address the Diversity Hypothesis. Guided by the Joanna Briggs Institute approach and Preferred Reporting Items for Systematic Reviews and Meta-Analyses for Scoping Reviews (PRISMA-ScR), a systematic search of three electronic databases (Web of Science, Scopus, and Ovid PsycInfo) was performed. The Mixed Methods Assessment Tool was used for quality appraisal of included articles. Studies were included if they were peer-reviewed, empirical studies that satisfied the population (‘scientists or scientific teams/labs’), concept (‘diversity of scientists or scientific teams/labs’), context (‘scientific productivity or impact’) as well as other inclusion criteria. In total, 104 articles were included and their findings pertaining to the relationship between diversity and scientific productivity and impact analysed. We estimate that only between 15
Emerging artificial intelligence (AI) technologies have recently gained significant traction beyond industry applications, increasingly permeating academic research across diverse disciplines like computer science and sociology. This widespread adoption marks a transformative shift in scholarly methodologies, particularly enhancing tasks such as extensive data analysis, literature synthesis, and theoretical modeling. The integration of AI into sociological knowledge production represents a fundamental transformation of disciplinary practice. This study examines differential AI adoption patterns across Chinese and American sociology through comparative analysis of three unstructured datasets: institutional seminars and workshops from ten top sociology departments, faculty research profiles, and national sociology conference presentations in recent years. Our findings reveal divergent trajectories as American institutions demonstrated early enthusiasm for AI-related content, while Chinese sociology, traditionally grounded in qualitative approaches, has shown accelerating adoption particularly since 2023. Each country’s national conference topics further reflect these differences. Chinese sociology combines a substantial machine-learning and intelligentisation theme with governance-centered, macro-level narratives of societal transformation, whereas American sociology pairs a dominant cluster on algorithmic systems and large language models with specialized programmes addressing inequality, prediction, and disciplinary reflexivity. These patterns reflect neither simple convergence nor persistent divergence, but rather multiple modernities in academic knowledge production shaped by distinct institutional logics and cultural contexts. The emergence of nation-specific AI tools (e.g., ChatGPT, DeepSeek) may enable distinctive methodological pathways rather than uniform global adoption. This paper highlights the potential of AI to substantially enrich knowledge production in sociology. Understanding the differential patterns of AI adoption within the discipline is essential for developing frameworks that accommodate epistemological pluralism while responsibly leveraging technological capabilities across diverse academic communities.
Criminology is often portrayed as theoretically stagnant – a view reinforced by an influential review that interpreted the modest explanatory power of published research as evidence that criminological theory had failed to advance. We argue that this diagnosis rests on the wrong yardstick. Explained variance is neither a necessary nor a sufficient indicator of theoretical progress. Drawing on a post-Popperian conception of scientific growth emphasizing the problem-solving capacity of competing research traditions, we propose evaluating criminological theory according to its ability to resolve empirical puzzles and reduce crime-related harm. We illustrate this approach with two examples. First, evidence from competing theories of desistance has increasingly converged on the view that structural turning points are better understood as facilitators and consequences of change than as exogenous catalysts, shifting attention toward agency, cognitive transformation, and reciprocal processes. Second, routine activities theory emerged in response to the puzzle of rising crime amid postwar prosperity and generated a highly productive research tradition encompassing environmental criminology, hot-spots research, and situational crime prevention. In turn, situational crime prevention translated these theoretical insights into a cumulative body of evidence demonstrating the effectiveness of opportunity-reducing interventions. These episodes suggest that theoretical progress in criminology is best understood in terms of the capacity of competing research traditions to resolve anomalies, generate fruitful lines of inquiry, and produce practical solutions. More broadly, they demonstrate that theoretically pluralistic social sciences are capable of cumulative progress even in the absence of a dominant paradigm.
Scholars have long warned that ideological homogeneity in social science can compromise the error-correcting mechanisms of scientific inquiry, but how this process operates in practice, and how fragile claims become stabilized as settled knowledge, remains insufficiently understood. In this article, I examine how field-level ideological commitments embed contestable assumptions upstream of empirical testing, shaping what can be asked, measured, and concluded. I conduct an Ideological Epistemic Analysis (IEA) of a widely cited and highly publicized study treated as definitive evidence that gender-identity-prioritizing public-accommodations policies pose no public-safety risk. I show that a priori ideological commitments, operating through category reconstitution—especially the displacement of sex as an analytically relevant category—structure the study’s framing, constrain which mechanisms can be examined, and shape interpretation before data are examined. I further document how a methodologically flawed design produced findings that were nonetheless converted into settled science through canonization bias, a process by which ideologically congenial but fragile evidence is elevated to authoritative status through citation, media uptake, and policy discourse. Rather than reflecting a curious anomaly or isolated failure, the case illuminates a broader epistemic pattern in which ideological homogeneity weakens organized skepticism and facilitates the canonization of fragile claims. I also suggest avenues for mitigating ideological epistemic distortions and promoting more robust social science.
In periodic ritual patterns, strangers exchange gestures that create a “spark” between them. This article identifies a potent form of interaction that has gone under-examined, what I term evanescent interaction, which exists in the realm of informal, momentary exchange. The article develops an ontology of the interaction which enables us to understand how it is generated as part of the interaction order of social life. While generalized as an interaction form, the phenomenon is anything but mundane. In tracing the interaction’s form and consequences, the article theorizes a way by which its micro elements create a connection to organizational life with discernible macroscopic effects. A theory of evanescent interaction is built by incorporating the ideas of personhood and social integration on the one hand with the roles of emotion and narrative on the other. While under-acknowledged and seemingly inconsequential, evanescent interaction thwarts fear and enables connection to human community.
A recent article concluded that watching a seven-minute clip from a professor’s lecture on the evolutionary psychology of human mating made participants blame a crime victim. This conclusion is not supported by the article’s own results and hypothesis tests. The conclusion is further compromised by a methodological problem with the video in the social condition. Additionally, the article lacks a coherent theory, mischaracterizes evolutionary psychology, conflates description with prescription, and dismisses objective standards of evidence.
This paper advances a theoretical contribution, a triadic framework of human-AI-society interaction, and introduces a methodological innovation to support it. We argue that AI development is not a unidirectional pipeline from design to use but a recursive system of feedback loops across three interdependent spheres: practitioner-AI (design decisions), user-AI (application practices), and society-AI (evaluation and governance). To develop this theory, we propose an LLM-enhanced computational grounded theory framework that positions LLMs as analytic instruments within a human-led workflow. The framework comprises five phases: data construction, LLM-assisted topic generation, human-LLM topic refinement, computational confirmation, and theory generation. We operationalize the framework through a case study of public discourse about ChatGPT on Twitter (now X), analyzing 584,160 English-language tweets from the first three months following the model’s release. Empirical analysis reveals 10 themes and 54 subtopics that converge into three stakeholder-oriented interaction patterns corresponding to the triadic framework. The theoretical contribution is threefold: (1) reconceptualizing human-AI interaction as irreducibly triadic rather than dyadic; (2) specifying the mechanisms of recursive feedback loops (design enables use, use generates evaluation, evaluation feeds back into design); and (3) identifying temporal asymmetries (user adaptation in hours, corporate response in days, governance in years) as a structural feature of AI sociotechnical systems. Methodologically, this study demonstrates how LLMs can support saturation assessment at scale while maintaining interpretive rigor.
This article offers the first empirical study of how professors teach contentious political subjects. Drawing on a database of millions of syllabi, we explore how three such subjects are taught: (1) racial bias in the criminal justice system; (2) the Israeli-Palestinian conflict; and (3) the ethics of abortion. We find that most professors don’t assign readings that reflect the larger scholarly debate around each issue. This practice may alienate students, undermine public trust, and compromise the civic and intellectual formation of students.
Modern bureaucratic institutions distribute not only resources, rights, sanctions, and recognition, but also time. This article develops a theory of institutional time inequality to explain how disparities in the duration between claim and resolution become a mechanism of stratification. Drawing on sociological theories of time, bureaucracy, waiting, administrative burden, and capital conversion, I argue that institutional delay is not merely an administrative inconvenience or organizational inefficiency. It is a distributive process through which life chances are shaped. The article introduces the concept of speed capital to describe the capacity to convert existing forms of economic, social, and cultural capital into accelerated institutional outcomes under conditions of differentiated temporal governance. Empirical illustrations from immigration adjudication, pretrial detention, disability benefits administration, housing waitlists, eviction proceedings, and citizenship by investment programs show how institutions allocate waiting unevenly and how advantaged actors secure faster or less damaging pathways through bureaucratic systems. The article contributes to theories of stratification by shifting attention from waiting as experience to waiting as distribution. It shows that access received too late is not equivalent to access received in time, and that inequality is reproduced not only through the unequal distribution of resources, but through the unequal distribution of institutional time itself.
Since its discovery in the 1990s, cultural omnivorousness has arguably become the most extensively researched empirical topic in the sociology of culture. This paper argues that its prominence stems from its status as the major anomaly facing the dominant theoretical program of the field: Bourdieusian sociology of cultural reproduction. Much of Bourdieu’s theorizing rests on what can be described as an emulation game model, in which elite groups choose tastes as signals for self-identification based on their inaccessibility to non-elite groups, thereby creating a cultural homology. Yet in the case of omnivorousness, elites appear to display preferences that non-elites can readily imitate. The efforts to reconcile omnivorousness with the emulation game imagery have driven much of the field’s theoretical development. These efforts have taken three principal forms, each offering an alternative to the implicit understanding of cultural consumption as the deciphering of complex codes prevalent in Bourdieu’s writings: the cultural repertoire thesis (where competence is defined by mastery of diverse vocabularies), the classification game thesis (where cultural capital lies in the ability to draw increasingly refined distinctions), and the countersignaling thesis (where higher status is expressed through the ostentatious refusal to signal). All three responses allow one to argue that a revised and more complex version of cultural homology still exists.
A foundational question in social theory concerns the mechanisms of institutional stability and change, that is, how field-wide orders are reproduced and transformed through individual-level practices. For the sociology of science, this translates to a specific question: how do the persuasive practices of researchers aggregate to shape the epistemic values of a field? Yet, systematically connecting individual-level persuasive claims to field-wide value change has remained a methodological and theoretical challenge. This paper addresses this gap by introducing the Persuasive Positioning Model (PPM), a framework that classifies the logics of persuasion into four distinct types: Empirical Superiority, Niche Construction, Methodological Virtue, and Structural Contribution. The PPM is operationalized through a case study of Artificial Intelligence (AI), using a large-scale computational analysis of 17,756 conference papers from the pre-revolution (2010–2012) and mature deep learning (2022–2024) eras. The findings document a fundamental value reconfiguration in AI, revealing a significant decline in the logic of Methodological Virtue and a corresponding rise in the logics of Niche Construction and Structural Contribution, while the logic of Empirical Superiority persisted as the field’s dominant organizing principle. These results are synthesized into a conceptual model of value reconfiguration defined by three interlocking dynamics: the Persistence of the dominant logic, the Emergence of competing counter-logics, and the resulting Pluralization of the field’s epistemic values. By connecting individual-level persuasive claims to field-wide shifts in epistemic values, the framework and the computational pipeline developed to operationalize it offer a replicable approach for studying how the values of scientific fields are constructed and reconfigured over time.
Humans built societies with transgenerational permanence, culturally discriminating identities, and defense of territories. These characteristics are shared with some other animals, suggesting that humans as social animals have a comparable biological basis. As an analog for early human social ability, modern foragers provide a model of elemental, small-scale units often called bands, which organize cooperative groups of @ 25 individuals. In these elemental groups, people knows each other intimately, but membership was not based on genetic closeness and so not a result of simple genetic selection. Rather, reciprocal cooperation favored an opportunistic association with selective advantages to the group in resource defense, social mobility, and cooperation. Early in human history, the marking of membership in such groups likely became culturally distinctive with distinguishing behavior, personal adornment, and material culture. The capability to form groups based on culture ultimately allowed humans to form societies of very large scales as represented by the modern age. Band-size units became embedded in village-size groups that could then be culturally embedded within regional polities and so forth. Human societies evolve as multi-scalar formations with each level retaining characteristics of societies– permanence, culturally discriminating identities, and defense of resources. In essence, human societies formed as societies (bands) within societies (villages) within chiefdoms, states, and empires.
Many consequential institutional changes now bear the imprint of activism without the spectacle of protest. Yet scholarship on quiet, insider, market-based, and organizational activism still lacks a clear account of when such practices should be treated as social movement action and how they relate to contentious politics. This article develops the politics of alignment as a distinct social movement regime. It argues that the politics of alignment operates when movement-linked actors build, capture, or redirect the rankings, audits, certifications, accreditation systems, investor channels, and comparable review processes through which organizations are judged, and when those devices tie movement claims to dependencies that organizations cannot easily ignore. The article narrows the concept by distinguishing it from NGO advocacy, legal mobilization, institutional activism, and private regulation in general. It identifies five conditions under which the politics of alignment should count as a social movement regime, specifies its three core mechanisms—translation, comparability, and repeated review—and places them alongside the disruptive leverage of contentious politics. Three anchor cases organize the argument: the Corporate Equality Index and transgender-inclusive benefits in large United States firms; post-Rana Plaza labor governance through the Accord, supplemented by U.S. campus anti-sweatshop campaigns around the Worker Rights Consortium; and AIDS treatment activism from ACT UP to the Treatment Action Campaign. The article shows that the politics of alignment belongs inside social movement theory, differs from contentious politics in mechanism, audience, and temporal rhythm, and can be as consequential in producing durable institutional change.
Why do ethnic markers consistently outperform economic position as bases for political coalition, even when material interests would favor cross-ethnic alliance? This article advances a theoretical framework centered on the activation problem. Drawing on the alliance detection system identified in evolutionary psychology and building upon the ecological availability framework from prior research, the current analysis argues that ethnic markers enjoy advantages in political mobilization because ethnic cues possess greater ecological availability, denser institutional infrastructure, and lower coordination costs relative to economic position. This differential cue ecology reconciles two apparently contradictory findings. Laboratory research demonstrates that racial categorization is a reversible byproduct of coalitional cognition, yet field evidence shows ethnic mobilization consistently proves easier than class mobilization. The resolution lies in recognizing that reversibility requires deliberate restructuring of the cue environment—a restructuring that modern institutional arrangements systematically favor for ethnic identities. The framework also extends beyond ethnicity to analyze how nations, religions, and other imagined communities function as activation technologies with varying cue ecologies and inherent fragmentation tendencies. Historical cases from the collapse of the Second International to the Battle of Blair Mountain illustrate the framework’s explanatory power. The analysis generates testable predictions about conditions under which economic position might overcome its activation disadvantages and addresses the deeper question of whether large-scale human solidarity can transcend the tribalistic defaults of coalitional cognition.
In social science, where exactly is the line between appropriate and inappropriate researcher conduct when it comes to a researcher’s politics? This commentary addresses the growing politicization of academic research and the blurring of long-standing (if perennially contentious) norms about appropriate scientific conduct. It enumerates my own views on acceptable political behavior on the part of researchers and a much longer list of unacceptable political behavior that violates standard research protocols. Although intended as a guide for scholars confused by the apparent discrepancy between emerging norms and long-standing instructions, this note is also an invitation to a larger conversation about what should be our research norms in today’s polarized climate, as well as how institutions should intervene, if at all.
A counter-productive division exists, both formal and informal, separating social science and natural science. Human society is rarely studied in conjunction with other known animal societies, and objections to integrating natural science and social science in this way are typically rhetorical, not substantive. Here, I make a case for naturalizing the study of society.
Relational Models Theory (RMT) has provided a remarkably durable framework for understanding how humans coordinate social life through four elementary relational models. Yet despite its success, key issues at the periphery of the theory remain underdeveloped, misunderstood, or poorly operationalized. This article addresses several of these edge cases. First, I examine the available measurements and note that they support theory testing but bring with them several challenges. One of them is that relationships frequently combine multiple relational models across domains, a complexity that RMT assumes but which classic measurement traditions may interpret as a nuisance; this is my second point. Third, I point out that because established measurement focuses on relations as units, it lacks indexing whether one is superior or subordinate in an authority ranking relation. Fourth, I highlight the neglected role of null and asocial modes of interaction, which dominate everyday urban life but are largely absent from measurement instruments and theoretical applications. Fifth, I emphasize the differentiation between legitimate authority ranking and coercive dominance. Last, I discuss the temporal limits of relational acts. Together, these points suggest that RMT’s utility lies not only in its taxonomy of relational models but also in its capacity to illuminate their mixtures, absences, and boundaries.
