From Neighborhoods to Molecules: The Selective Appropriation of Sociology in Social Epigenetics

Abstract

Social epigenetics is presented as a promising interdisciplinary avenue between the natural and social sciences to explore the links between neighborhood environments, epigenetic modifications and health outcomes. Sociological concepts and methods are mobilized, sometimes through direct collaboration between epidemiologists and sociologists, to grasp the embodiment of social inequalities. Drawing on an in-depth qualitative analysis of three epidemiological cohort studies in the United States, we offer a processual approach to the use of Chicago-style ecological research on disorganization in social epigenetic studies. We argue that a selective appropriation of sociological research operates at two different levels of study design: that of the overall cohort study in which social epigenetics research is conducted and data are obtained, and that of the social epigenetics studies. This selective appropriation represents one of the most successful attempts in social epigenetics to complexify concepts and methods for analyzing health outcomes as a product of social situations. We show that for the epidemiologists and sociologists working in this interdisciplinary space, the move to greater complexity means observing the social organization of neighborhoods from a rather narrow window of observation, but one that presumably allows them to produce robust statistical evidence for the biological embodiment of social conditions.

Introduction

For several decades, the study of the entanglements between material bodies and living conditions has attracted the interest of social and natural scientists alike (Bliss 2018; Dubois, Guaspare, and Louvel 2018; Landecker and Panofsky 2013; Larregue 2024; Meloni 2016). Epigenetics, microbiome research, and neuroscience all share the notion of a plastic, malleable body, which became the rallying point for researchers hoping to overcome an outdated nature/culture divide (Meloni, Williams, and Martin 2016, 16–17). Thus, social scientists explicitly incorporated biology into their research design and analysis, hoping to uncover the interplay between biological and social factors in the formation of social inequalities (Harris and Schorpp 2018).

Among these diverse areas of interest, social epigenetics—the study of how social experiences lead to changes in either chemical markers attached to a DNA sequence or the structure of chromatin (a complex of proteins and DNA found in the nucleus of eukaryotic cells)—is often presented as a promising interdisciplinary intersection between the natural and social sciences. While scientific and political expectations are high, public health researchers (Evans et al. 2021) and science and technology studies (STS) scholars (Niewöhner 2011) have expressed concern that studies in social epigenetics so far fail to explore the complexity of social environments and structures, and called for interdisciplinary collaborations with social scientists, as well as for community-centered research. Another group of STS scholars stress that criticism of social epigenetics draws too firm a line between reductionist and complex approaches. Instead, they point to the complexity work that researchers perform when stabilizing their experimental systems and efforts at complexifying them (Chiapperino 2024; Mann and Chiapperino 2023; Penkler 2022).

In this article, we explore how researchers integrate complexity in research areas described as “biosocial.” We study a form of interdisciplinary endeavor that is gaining momentum in social epigenetics: the borrowing of social science concepts and methods alone or in the context of collaborative efforts. This form of interdisciplinarity is found in an emerging stream of research examining the links between neighborhood environments, epigenetic modifications, and health outcomes in US-based epidemiological cohorts (see, e.g., Lei et al. 2015; Smith et al. 2017; Ward-Caviness et al. 2020). A sociological reference widely used in these studies is the research on disorganization initiated by Robert Sampson and Stephen Raudenbush in the 1990s with the Project on Human Development in Chicago Neighborhoods (PHDCN) (see especially Sampson, Raudenbush, and Earls 1997; Sampson and Raudenbush 1999). As these newer epidemiological studies reject characterizations of social epigenetics as reductionist, studying such research projects offers insight into how sociology is mobilized to complexify approaches to social environments.

In this article, we offer a processual analysis of this interdisciplinary circulation by tracing the use and appropriation of the Sampson-Raudenbush approach across time and research projects. In doing so, we examine the tensions between the pursuit of conceptual complexity and the methodological imperatives deriving from quantitative, epidemiological research. By analyzing how neighborhood conditions are conceptualized and operationalized in social epigenetics projects, we shed light on the practical labor required by, and the material constraints imposed on, interdisciplinary circulations. While interdisciplinary collaborations may be useful in rendering scientific understanding of health disparities more attuned to both social and biological realities, they do not offer a definitive remedy for reductionism.

As we shall see, the appropriation of sociological variables and notions operates at two different levels of study design, and nurtures a work on complexity of two kinds. At the level of the cohort study, sociological variables are combined with variables drawn from various disciplinary frameworks, with a view to making the epidemiological cohort an interdisciplinary research platform on health inequalities. This results in a significant reduction in the number of dimensions to be studied and of variables to be coded, when compared to the original sociological research design. Within social epigenetics research, sociological research is used to build a statistical model for the biological embodiment of social inequalities. Researchers’ conceptual and methodological work is driven by issues of data availability (both biological and social), and by the search for robust statistical links between specific neighborhood aspects and epigenetic modifications. This leads to the routine use of sociological variables whose conceptual roots are no longer questioned.

Place and Health: Social Epigenetics and the Renewed Promise of Interdisciplinarity in Epidemiology

The study of place and health has a rich and interdisciplinary legacy. In countries from the Global North, the historiography of the field usually begins with 19th-century sanitarians such as Charles Booth in England and Jane Addams in the United States, who advocated quality housing and sanitary reforms to improve the physical and mental health of urban dwellers (Deegan 1988). In the aftermath of the Second World War, the term “social epidemiology” was coined in connection with a study of the relationship between racial inequalities, neighborhoods, and mortality in the segregated US context, which was published in the American Sociological Review (Yankauer 1950). In time, this social approach to place and health was gradually replaced by a focus on individual determinants of health (Macintyre, Ellaway, and Cummins 2002, 126) and a distancing from social theory (Wemrell et al. 2016). Since the 1990s, social epidemiologists have tried to revitalize the study of the relationships between neighborhoods and health, arguing that “where one lives makes a difference to one's health” (Kawachi and Berkman 2003, 1). “Neighborhoods”—broadly defined as contexts that have both physical and social features (Diez Roux 2012)—have been presented as a rallying point for researchers from epidemiology and sociology alike willing to invigorate and re-legitimize ecological approaches to health inequalities (Macintyre, Ellaway, and Cummins 2002; Prior, Manley, and Sabel 2019).

The search for causal effects is particularly important in this regard, and social epidemiologists devoted significant conceptual and methodological work identifying causal pathways leading from neighborhoods to health (Diez Roux 2012). The combined study of social and biological pathways has attracted considerable attention within the research agenda on the embodiment of social experiences (Krieger 1999), a perspective that dates to the early 19th century in epidemiology and expanded significantly as researchers began to routinely collect and analyze biological samples (saliva, blood). One of the earliest epidemiological studies to conduct such sampling was the Framingham Heart Study, launched in the United States in 1948. However, it was not until the 1980s and 1990s that routine collection and analysis of biological samples became common for major epidemiological surveys conducted in North America and Europe. They were able not only to assess bodily responses to social environments but also to address which biological systems are impacted and with what effects on a given disease.

Foundational work has been carried out in the United States in epidemiology (Krieger 1999) and in sociology (Massey 2004), to elucidate the biological mechanisms mediating the effects of racial hypersegregation on the health of African Americans. Racially segregated neighborhoods are still an interdisciplinary meeting point for epidemiologists and sociologists investigating biosocial pathways to racial health inequalities (Roberts and Rollins 2020). However, neighborhood researchers now routinely search for biomarkers related to stress (e.g., diurnal cortisol), inflammation (e.g., C-reactive protein) and aging (e.g., telomere length) (Petteway, Mujahid, and Allen 2019). Researchers have expressed concerns that molecular evidence may confer a symbolic advantage over epidemiological evidence (Arminjon 2023), or that the search for biological proofs may cause potential escalation in what counts as strong evidence for substantiating the causes of health inequalities in public policy arenas (Roberts and Rollins 2020; Schrecker 2013).

In many countries, the research domain commonly referred to as “social epigenetics” (Louvel 2020) is regarded with hope as a research avenue with strong policy implications. This field of research purports to uncover “how social experiences affect genes and our biology,” and in doing so “may bring about changes in the various molecules that interact with DNA, determining which genes are switched on or off (National Institute on Minority Health and Health Disparities 2024). As such it has been presented as the deepest layer in the “archeology of biological embedding” (Hertzman 2012). Importantly, it is expected that studies in social epigenetics will be used in public arenas to provide biological proof of a given social harm. Interestingly, expectations of epigenetics differ not only according to national political and social contexts, but also according to the local knowledge of biological embodiment that it questions (Müller and Kenney 2021). In Australia, for example, Indigenous and non-Indigenous researchers alike may see environmental and social epigenetics as having affinities with indigenous ways of thinking about embodiment and use this to advance political and legal recognition of the intergenerational harms of colonization (Byrne and Keaney 2024; Warin, Kowal, and Meloni 2020). In other contexts, researchers and practitioners see epigenetics as a means of rethinking resilience and trauma-informed care (Müller and Kenney 2021). Finally, in the United States and Europe, social epigenetics is mostly discussed in social epidemiology as a tool to design specific policy interventions to improve health inequalities (Martin et al. 2022). However, social epidemiologists from these countries have emphasized that social epigenetics fails to meet these expectations. Methodologically, they identify the need to include more data about social life, a prospect partly made possible by the diversification of measurement tools for characterizing neighborhood settings—from census data to geocoded data (Dunn 2020). Conceptually, social determinants of health are still often operationalized as individual-level characteristics, without explicitly considering associated social stratification processes (Evans et al. 2021). In addition, research in social epigenetics may exaggerate the causal effects of certain environments because of its ignorance of the cultural and social processes that create differences among populations (Geronimus 2013). Lastly, social epidemiologists interpret the relative failure to grasp complex social environments as the result of social scientists’ lack of involvement (Evans et al. 2021; Martin et al. 2022).

STS scholars have described the “molecularization of environments,” the narrowing of culturally and socially embedded practices to individual exposures and behaviors (Niewöhner 2011) at work in social epigenetics research. They have highlighted obstacles to including complex understandings of environments in research practices, such as the need to “reduce the messiness of environmental context in a way suitable for lab work” (Niewöhner 2011, 289); institutional pressures to “sell and perform” (Penkler 2022, 6); and the view that simple stories are more compelling (Lappé, Fahey, and Hein 2022, 15). They have also stressed that research in social epigenetic studies conveys an intrinsic normativity (Chiapperino and Panese 2018). Social epigenetics provides biological explanations of harm that may stigmatize certain individuals or social groups when associated with implicit biological determinism, which is particularly problematic for racialized groups (Meloni et al. 2022; Penkler 2022). Additionally, it tends to produce narratives of individual responsibility, even when researchers intend to look at the effect of broader social forces (Lappé, Fahey, and Hein 2022). The wealth of studies that draw on social epigenetics to support the Developmental Origins of Health and Disease (DOHaD) perspective are a prominent example of a narrow understanding of the links between maternal–fetal interactions and later-life disease that directs attention to maternal responsibility (Richardson 2021). Lastly, STS scholars, just as social epidemiologists, have repeatedly called for more research collaboration to include social scientists in social epigenetics studies, and more broadly, in epidemiological DOHaD studies (Penkler 2022) and biosocial research perspectives on embodiment (Bieler and Niewöhner 2018; Rose, Birk, and Manning 2022). They also argue that engaging with communities, especially marginalized ones, is crucial to increase the scientific relevance of social epigenetics and its benefits for the involved populations (Meloni et al. 2022).

Although it is a well-established critique of social epigenetics, some scholars in the field of STS continue to question the dichotomies of simple vs. complex, reductionist vs. holistic on which it is based. Drawing on the social sciences of complexity (Law and Mol 2002), Chiapperino (2024) points out that while biosocial research aims to reduce complexity, like any scientific practice, it also implies a specific kind of complexity work that has social, technical and material dimensions. He argues that paying attention to the “panoply of complexities” and to how some researchers try to complexify their approach to biological and social complexity matters for STS scholars to cast light on what enables “alternative representations of how the ‘social’ intertwines with the ‘biological’” (Chiapperino 2024, 4). This involves going beyond a deficit-based approach (accepting that these representations will never be “thick” enough from a social science perspective) and acknowledging that the researchers involved recognize the limits of their field—in terms sometimes close to those of STS researchers—and seek to modify their practices accordingly (Louvel and Soulier 2022; Mann and Chiapperino 2023). The study of the links between structural conditions at the neighborhood level, epigenetic modifications and health (see Lei et al. 2015; Ward-Caviness et al. 2020), opens a privileged window into a kind of work on complexity that involves sociological and epidemiological research, in the context of large-scale US-based studies on social inequalities in health.

Materials and Methods

The study of neighborhood-induced epigenetic processes has been presented as a vibrant area of research in social epigenetics (Olden, Olden, and Lin 2015). However, such studies remain limited, in part due to the paucity of epidemiological surveys that have both detailed neighborhood information and DNA samples on which epigenetic analyses can be performed. A search of the Web of Science database for articles with the keywords “neighborhood” and “epigenetics” or “neighborhood” and “methylation” in the title yields 25 results, with the first article published in 2015 (search date: 23 October 2024). Almost all use data from epidemiologic studies conducted in the United States. For this article, we conducted a qualitative study of three ongoing epidemiological studies from which at least two papers have been published, on the links between neighborhood, epigenetic modifications, and health: the Family and Community Health Study (FACHS), the Detroit Neighborhood and Health Study (DNHS), and the Multi-Ethnic Study of Atherosclerosis (MESA) study.

Our analysis was restricted to articles that used the PHDCN. The rationale behind this choice is twofold: (1) the PHDCN has been one of the most influential original projects on neighborhood and development, and was instrumental in the development and framing of the three subsequent cohort studies analyzed in this article; (2) by focusing on articles that use the PHDCN, we were able to systematically compare the appropriation and diffusion of sociological concepts and methods across different areas of social epigenetics.

We traced the design of the three epidemiological studies from secondary sources (codebooks published in institutional archives, information available on websites, articles describing research design). This allowed us to learn about the theoretical perspectives underlying these studies, how they operationalize and conceptualize neighborhoods, and for what purposes they include epigenetic measures. We also conducted 13 in-depth semi-structured interviews with sociologists, epidemiologists, biostatisticians and psychologists directly involved in social epigenetic studies of neighborhoods. We prepared these interviews by reading one or two articles they had published on the topic. We conducted interviews in three stages: first, we asked about their research background and scientific training since graduate school; then about the studies they conducted on neighborhoods in social epigenetics (with specific questions about the choice of neighborhood-related variables, epigenetic measures, statistical analyses, and the reception of their study); and finally we enquired about ongoing, unpublished research. To enrich our understanding of what differentiates, or conversely connects, the study of epigenetic modifications with that of other biological markers, we conducted three additional interviews with researchers who are developing a biosocial perspective on the links between health and neighborhoods but do not focus on epigenetics. Interviews were conducted on Zoom. The study adhered to ethical standards governing research involving professionals. Potential participants received invitation emails outlining the project. All participants took part voluntarily, provided informed consent, and their data were anonymized to ensure confidentiality. The documentary sources and interview material were separately analyzed by both authors and then discussed collectively. Published articles were qualitatively analyzed in relation to their research questions, methods (including epigenetic and neighborhood measures), findings and conclusions. Additionally, the interviews were thematically coded using NVivo with help from a research assistant. The triangulation of data—articles, secondary sources and interviews—allowed us to strengthen our analytic strategy.

Findings

Integrating Sociology into the Design of Epidemiological Studies

The integration of variables derived from the PHDCN into the design of the three US cohort studies (FACHS, DNHS, and MESA) was the first stage of the interdisciplinary circulation between sociology and epidemiology that facilitated the rise of social epigenetics. We show that this integration exemplifies how the principal investigators of these cohorts mobilize references from several disciplines to form interdisciplinary research platforms. The quest for complexity (incorporating fine-grained neighborhood variables and proposing a relational approach to social situations) goes hand in hand with a pragmatic simplification of the PHDCN, whereby a limited number of variables are selected and deemed relevant.

Epidemiological Cohort Studies as Interdisciplinary Research Platforms

The FACHS, DNHS and MESA are all cohort studies, a type of longitudinal study in epidemiology in which “the participants do not have the outcome of interest to begin with. They are selected based on the exposure status of the individual. They are then followed over time to evaluate for the occurrence of the outcome of interest” (Setia 2016). Because they can help clarify the temporal sequence between exposure and outcome, cohort studies are considered an important tool to identify causal mechanisms in health research. Their architecture is flexible and builds on multiple data strategies: collection of new variables and samples, data mining of existing data sources, and data recombination.

The design of the three US cohorts we investigated has been informed by theoretical models from several social science disciplines. It is common for epidemiological research to borrow social science tools, be they methods or concepts (Fassin 2003). This multidisciplinary dialogue has been proposed as an answer to the limitations of data-driven, black-box epidemiology (Galea and Link 2013, 847). In what follows, we show that the interdisciplinary scope of the cohorts implies the need to select and simplify the loans from each discipline.

The FACHS has a dual grounding in sociology and in psychology. It was initiated at the Center for Family Research at the University of Georgia in the mid-1990s. The study

is a multi-site, longitudinal investigation of the effect of family processes, neighborhood characteristics, and other contextual factors on African American parents and their children. The study was initiated in 1996 with over 800 families, all of which contained a 5^th grader. Data has been collected from the study participants every 2–3 years and is ongoing. (University of Georgia, Center for Family Research 2025)

The study has been designed to support research on a wide range of topics, including “parenting practices, romantic relationships, school success, discrimination, racial socialization, peer affiliations, higher education, occupational achievement, religiosity, substance use, coping resources and mental health” (University of Georgia, Center for Family Research 2025). In part, this diversity reflects an effort to address a variety of factors that contribute to health and well-being among African American families in low socioeconomic settings. It is also an expression of the disciplinary affiliations and research interests of the three founders: Ronald Simons, then a professor of sociology at the University of Iowa, had conducted research on the effects of parenting and community context on child behavior problems; Steven R. Beach, then a professor of psychology at the University of Georgia, specialized in the study of marital and family orientations in clinical psychology and psychiatry; and Frederick Gibbons, then a professor of social psychology at the University of Iowa, had developed a cognitive model of adolescent risk behavior.

The Detroit Neighborhood Health Study (DNHS) includes researchers with expertise in urban health, survey research, epidemiology, and psychoneuroimmunology. It is a 5-year study funded in 2008 by the National Institutes of Health (NIH) National Institute on Drug Abuse and conducted by the Center for Social Epidemiology and Population Health at the University of Michigan School of Public Health. It is led by two Principal Investigators (PIs): Sandro Galea (then Professor of Epidemiology, Columbia University) and Allison Aiello (then Associate Professor of Epidemiology, University of Michigan School of Public Health). The purpose of the study is to “determine the predictive effects of ecological stressors, such as income distribution and residential segregation, on the development of post-traumatic stress disorder (PTSD), substance use, and other psychological and behavioral outcomes” (Aiello and Galea 2021). The 1,500 study participants were randomly recruited from Detroit's 54 city-designated neighborhoods to provide a representative sample of the city's population. The study is part of a long-standing body of research on the impact of adverse urban environments on health, particularly mental disorders, while also contributing to the expanding field of research on the “biologic explanations that may clarify the links between features of the urban environment and individual mental health [such as] changes in the immune function, gene–environment interactions and biological mechanisms” (Galea, Uddin, and Koenen 2011, 401).

Lastly, the MESA is a cohort study sponsored by the NIH National Heart Lung and Blood Institute. It involves investigators from a large range of disciplines (including epidemiology, cardiology, and biostatistics) to study the “factors that predict progression from subclinical cardiovascular disease to clinically overt cardiovascular disease,” to improve early prevention and treatment (National Heart, Lung and Blood Institute 2025). The population under study in MESA is an ethnically diverse sample of 6,814 asymptomatic men and women aged 45–84. The principal investigators present MESA as a “unique opportunity to link (this) rich individual-level information to broader social and environmental contexts” (Diez Roux et al. 2016, 353). In particular, they argue for the importance of studying the links between neighborhoods and cardiovascular risk (Diez Roux et al. 2016).

The PIs for the three cohorts have deployed a highly ambitious data collection program to achieve interdisciplinary outreach. Time constraints, however, affect the ability to collect detailed questionnaire data on all the individual and collective factors considered to play a role in the health outcomes under consideration. Investigators wish to keep participants’ burden low so as to sustain high participation and minimize attrition. This implies reducing questionnaire completion times as much as possible. In FACHS, investigators administered a series of self-report questionnaires during two two-hour home visits (Simons et al. 2002, 335), collecting information on subjects as diverse as psychological distress, personal outlook, psychological resources, negative life events, social support, and neighborhood quality. Additionally, they used US census data to calculate a measure of community socioeconomic disadvantage. In DNHS, investigators conducted a “40-min, structured telephone interview annually between 2008 and 2012 to assess perception of participants’ neighborhoods, mental and physical health status, social support, exposure to traumatic events, and alcohol and tobacco use” (Aiello and Galea 2021). Knowledge about the neighborhood was also gained through direct observation. Lastly, MESA participants are seen in clinics at one of the field centers, which are located at six universities across the United States, with an initial enrolment and baseline examination in 2000–2002, followed by five follow-up examinations. Information was collected through questionnaires, but also through medical examinations, and physical measurements. Budgetary limitations also affected the overall calibration of these epidemiological surveys, especially as they collect biological specimens (blood in FACHS from wave 6 in 2007–2008, saliva and blood in DNHS, blood and urine in MESA), which adds the significant cost of sampling, transport, storage and analysis.

Given these practical constraints, how do PIs integrate the study of neighborhoods, which is important in all three cohorts? Other epidemiological studies have drawn on a range of conceptual inputs from sociology (Prior, Manley, and Sabel 2019), including neighborhood social disorganization (with reference to foundational studies by Faris and Dunham (1939) and Shaw and McKay (1942)), neighborhood social capital (Kawachi and Berkman 2000), neighborhood disorder (applying the broken window model defined by Kelling and Wilson (1982), reviewed by O’Brien, Farrell, and Welsh (2019)), racial segregation and the ecology of inequality (Massey and Eggers 1990). As noted, in the three cohort studies analyzed in this article, the PHDCN is a key reference, albeit used differently. We show that its use is intended to make the understanding of social situations more complex, by refining knowledge about neighborhoods, and by adopting a relational approach to the social factors that influence health at this level. Integrating the PHDCN into the design of these cohorts also involves a pragmatic selection and simplification of variables and concepts.

The Pragmatic Simplification of the PHDCN

Directed by the Harvard School of Public Health and initiated in 1994, the PHDCN was

a large-scale, interdisciplinary study of how families, schools, and neighborhoods affect child and adolescent development. It was designed to advance the understanding of the developmental pathways of both positive and negative human social behaviors. In particular, the project examined the causes and pathways of juvenile delinquency, adult crime, substance abuse, and violence. At the same time, the project also provided a detailed look at the environments in which these social behaviors take place by collecting substantial amounts of data about urban Chicago, including its people, institutions, and resources. (Inter-university Consortium for Political and Social Research, PHDCN series. 2025)

From inception, the PHDCN was interdisciplinary and far-reaching, as illustrated by the oft-cited publication in Science (Sampson, Raudenbush, and Earls 1997). Co-developed by a social epidemiologist (Felton Earls), a developmental psychologist (Jeanne Brooks-Gunn), a criminologist-sociologist (Robert Sampson) and an educational researcher (Stephen Raudenbush), one of its main ambitions was to develop “ecometrics”: a fine-grained, qualitative and quantitative assessment of ecological settings (Raudenbush and Sampson 1999). On an epistemological level, the ambition to identify neighborhood social mechanisms is rooted in a relational approach to social epidemiology: emphasizing the importance of systems of interdependence and their own effects, which cannot be deduced from individual characteristics.

There were two major components to project design. First, an intensive study of Chicago neighborhoods was conducted: (1) the Systemic Social Observation (SSO) methodology was applied to assess physical and social disorder. Trained researchers visually assessed 126 neighborhood attributes on a total of 23,000 street segments in 80 neighborhood clusters; (2) public data were collected, including crime statistics and measures of concentrated disadvantage; (3) surveys were administered to 8,782 residents of 343 Chicago neighborhoods. The second component of the PHDCN was a series of longitudinal studies that “followed over 6,000 randomly selected children, adolescents, and young adults over time to examine the changing circumstances of their lives, as well as the personal characteristics, that might lead them toward or away from a variety of antisocial behaviors” (Inter-university Consortium for Political and Social Research, PHDCN series 2025).

Of particular interest to us here is that the first component of the PHDCN—the intensive study of neighborhoods, through SSO or questionnaire surveys—is adapted in FACHS and in DNHS. First, the SSO methodology is used but in a very limited way, tailored to the limited human and material resources available to implement it. In Wave 5 (2007–2008), the FACHS investigators introduced observer ratings of neighborhood social disorder. The 673 neighborhoods were rated by 67 observers trained in the use of the SSO technique (Raudenbush and Sampson 1999). Observers rated twelve items revised and selected from the PHDCN: four related to physical disorder and eight items related to social disorder. ¹ The DNHS included an in-person assessment of 138 block groups from each of Detroit's 54 municipally defined neighborhoods, conducted by a team of 20 trained assessors (Mooney et al. 2017). Second, both studies make selective use of several scales originally developed for the PHDCN (Sampson, Raudenbush, and Earls 1997) to assess perceptions of the neighborhoods through questionnaire surveys. In FACHS, as is often the case in social surveys, some measures were not imported verbatim, but rather “adapted from instruments developed for the PHDCN” (Simons et al. 2002, 336). For example, community cohesion is measured in FACHS with five items borrowed from the Social Cohesion and Trust Scale developed by Sampson, Raudenbush, and Earls (1997) and ten new items developed specifically for the study (Cutrona et al. 2000). In DNHS, ten interview questions were used to assess residents’ collective efficacy and ability to exert informal social control within the neighborhood. These questions were adopted from a study in which Sandro Galea was a co-investigator (the New York Social Environment Study, a random digit dial survey of 4,000 New York City residents conducted in 2005), which used instruments originally developed for the PHDCN.

Lastly, the MESA cohort makes a minimal use of the PHDCN, based on a small set of survey questions on social relations at the neighborhood level. This use is apparent in the MESA Neighborhood Study, an ancillary study that extended the main MESA study by collecting new data, which also began in 2000 (MESA Neighborhoods 2017). The goal was to assess dimensions of the physical, material and social environment of the neighborhood likely to affect cardiovascular health, including access to recreational resources, public transportation, healthy food, or safety, social cohesion, and interactions among neighbors. On these last two dimensions, the questions were borrowed from the PHDCN. In addition, the PHDCN was a source of inspiration when it came to measuring neighborhood characteristics in MESA. For example, additional neighborhood-level data were collected in three of the six MESA sites (Baltimore, Forsyth County, and New York City) through a separate sample of 5,988 respondents who were used as “informants” about neighborhood conditions. Here MESA researchers used the Chicago method to avoid the possibility of same-source bias, including spurious associations between self-reported neighborhood conditions and self-reported health outcomes (Diez Roux 2003).

The PHDCN had proposed a convincing blueprint to transform ecological settings into quantifiable variables, and to unravel the neighborhood-level social processes at play in health-related issues. By adopting elements of the PHDCN project, the cohort PIs put in place a methodology that gained increasing recognition in epidemiology. In particular, the use of the SSO methodology has helped to make neighborhood observations a widely used method for measuring socio-ecological inequalities, especially in North America (Schaefer-McDaniel et al. 2010). Further, from a conceptual point of view, these borrowings from the PHDCN enabled looking at neighborhoods in terms of economic and demographic characteristics, while also investigating social processes. Social processes are increasingly discussed in health research, as evidenced by the popularity of the concept of neighborhood collective efficacy, defined as “the linkage of mutual trust and the willingness to intervene for the common good” (Sampson, Raudenbush, and Earls 1997, 919), and operationalized by measurements of social control, social cohesion, and trust. It is now commonly used in epidemiological studies for a wide range of outcomes, from obesity to depression (Butel et al. 2020).

The three cohort studies, however, only integrate simplified versions of the PHDCN toolkit. This is not specific to these cohorts nor to the PHDCN, but is characteristic of the way in which epidemiological studies, which aim to answer questions from several disciplines, incorporate concepts or variables from a number of them. This simplification was done by reducing the number of dimensions considered and, for each of them, limiting the number of variables studied. Although the modalities of simplification vary greatly between the three studies, they are guided by similar practical constraints arising from financial and time investments required by researchers and participants alike.

Chicago-Style Ecological Approach to Neighborhoods in Social Epigenetics Studies

We will now argue that the PHDCN is used in social epigenetics studies to advance research agendas on social inequalities in health where researchers are less concerned with delving—conceptually and empirically—into social processes at the neighborhood level than with identifying the biological pathways by which social situations are embodied. Researchers pursuing this research agenda are moving away from the sociological questions raised in the PHDCN and turning to new issues. Furthermore, the elucidation of epigenetic mechanisms and the need to obtain solid yet straightforward statistical models require a form of scientific routinization.

A Shift Toward a Research Agenda on the Biological Embodiment of Social Inequalities in Health

Social epigenetics researchers are not engaged in in-depth theoretical or conceptual research about the social factors influencing health or behaviors at the neighborhood level, like the PHDCN researchers had been. The first reason is that most of them—sociologists, psychologists and epidemiologists, that is, 10 out of 13 researchers we interviewed—were not working on neighborhoods before delving into social epigenetics, but on various topics such as early childhood, delinquency, the family, air pollution or adolescent mental health. While working on these questions, they moved from thinking about the causes of social inequalities in health to a research program on the biological mechanisms underlying these inequalities:

When I was working on my dissertation, I just got really interested in this idea of what are the biological mechanisms that link these social factors to the kinds of health outcomes that we’re looking at in sociology? And so I just felt like there was this kind of missing piece where we were assuming the connection, but not directly looking at it. And so I started looking around for datasets that might have measures of telomere length or have stored DNA that we could use to measure telomere length. (Interview with social epidemiologist, United States, 2022)

The few researchers who had previously worked on neighborhoods turned to biomarker research for comparable reasons. For instance, a sociologist reflects on their post-doctorate trajectory, about ten years earlier:

It was difficult for me to understand how neighborhoods could affect human health so profoundly when I didn’t really understand the processes that generate poor health. So I did this additional training and sat in on disease physiology classes at the medical school. (Interview with sociologist, United States, 2022)

This change of direction toward the study of biological embodiment was facilitated by the inclusion of biological specimens in many epidemiological studies in the United States in the 2000s (Harris and Schorpp 2018) and, more recently, by the funding of post-doctoral fellowships from the NIH to train in the use of biomarkers, from which several interviewees benefited. The scientific agenda of these researchers includes further exploration of biological measurements (several have worked on other biomarkers, such as telomere length, before using epigenetic measurements) and the development of robust statistical tools to model biological pathways. The choice of cohort is presented as decisive, but less for conceptual reasons than for pragmatic ones, including biospecimen availability, access to a wide range of social variables, and the possibility of bringing together an interdisciplinary team with expertise in either social inequalities in health (social epidemiology, sociology, social psychology) or the study of biological processes (biostatistics, genetic epidemiology). The sheer availability of sought-after methylation data, which is still rare in cohort studies due to the high cost of epigenetic analyses, ultimately channels researchers to a small number of specific research infrastructures and networks, significantly limiting the range of scientific possibilities: “At the time, MESA was a natural choice because they had not only the methylation but the gene expression and they had these rich neighborhood measures, so it was probably the only study we really considered” (Interview with genetic epidemiologist, United States, 2022).

The institutional context in which social epigenetics research takes place (mostly in public health departments), the expectations of funders (the National Institutes of Health in the United States), and the goals and scope of the journals that publish this research (e.g., Clinical Epigenetics, Journal of Health and Social Behavior, Social Science & Medicine) are generally perceived as not conducive to conceptual discussions about neighborhoods. Although these journals welcome interdisciplinary work at the intersection of the biomedical and social sciences, they ultimately emphasize the implication of these studies for health, i.e., their ability to demonstrate that neighborhood conditions lead to epigenetic changes that matter for health outcomes. Thus, when asked about the peer review and publication process for their articles, only one interviewee mentioned a (minor) request for clarifying the definition of neighborhoods in the study, while all other interviewees recalled only what they described as “technical issues” related to DNA methylation analysis. In this context, even the minority of researchers with strong theoretical skills in sociology do not pursue this line of enquiry in social epigenetics: “It's difficult to see where I can develop constructs that I can then operationalize and test. There's not a lot of space to engage a very rich theoretical argument” (Interview with criminologist, United States, 2022).

The Obliteration of Sociological Concepts and the Development of Alternative Theoretical Models

Most of the researchers we interviewed know the PHDCN to be a reference study in the North American context, but perhaps in a more mundane way, they see it as a sociological reference cited in articles describing the design of the cohorts they are working on. It is a well-known fact that as theories, methods, or findings are incorporated into the scientific literature, their original sources can be gradually forgotten. This phenomenon, known as obliteration by incorporation (McCain 2014; Merton 1988, 622), can also lead to the partial depletion of the scientific content that was originally communicated. Indeed, the breadth and thickness of Sampson and Raudenbush's approach to social disorganization has been partially lost in the process of its obliteration and incorporation into the social epigenetics literature.

In fact, many researchers in the field—including sociologists—do not have a clear idea of the theoretical origins of some of the items they use:

Authors: You also had questions about the perception of crime and so, how did you come up with those questions? Are there specific works that you used that were published before yours to ask similar questions?

Interviewee: No, there weren’t. I don’t know. You know, I would love to give you a really good answer, but those questions were included in the data set years ago, before I came on. And I don’t know why they were added, but when I discovered them, I thought, this is precisely what we need. (Interview with criminologist, United States, 2022)

As obliteration occurs, theories that were once considered distinct or even opposed—and that inspired the original data collection—become conflated and synonymous. In their foundational article, Sampson and Raudenbush (1999, 604–605) presented their works as an alternative to the controversial “broken windows” approach developed by Wilson and Kelling in the early 1980s:

The “broken windows” literature sees disorder as a fundamental cause of crime (Skogan 1990, 75; Kelling and Coles 1996). The alternative hypothesis we offer is that disorder is a manifestation of crime-relevant mechanisms and that collective efficacy should reduce disorder and violence by disempowering the forces that produce both. (Sampson and Raudenbush 1999, 614)

They establish a causal scheme in which social composition (concentrated neighborhood disadvantage combined with immigrant concentration and residential stability) influences the formation of social ties and collective efforts to promote informal social control (collective efficacy), which in turn influences perceived violence and violent victimization. A few decades later, some social epigeneticists have forgotten these theoretical debates. For example, when asked about her knowledge of Sampson and Raudenbush's research, one epidemiologist responded, “So when you’re talking about Robert Sampson's work, you’re talking about broken windows?” In fact, some researchers, including those with graduate training in sociology, are not even sure which research or theoretical ideas originally inspired some of the cohort's data collection and variable construction: “[Robert Sampson] that's who I was thinking the work was probably based on, but I don’t know that for certain” (Interview with sociologist, United States, 2022).

But this obliteration is not just leading to depletion. The PHDCN variables are used to feed alternative theoretical models and stimulate conceptual reflections about biological embedding that might be only loosely related to the questions originally asked by sociological research on social disorganization. Put differently, some of the questions considered important by social epigenetics researchers are different from the questions that the PHDCN aimed to address. One example is why social epigenetics researchers are interested in neighborhood crime. The FACHS cohort study (Berg et al. 2022; Lei et al. 2015), for instance, measures the “perception of neighborhood crime” by asking respondents about the frequency of the following six criminal acts in their neighborhood: gun and knife fights, sexual assaults, robberies, drug sales, car thefts, and burglaries. They use it as an indicator of an individual's perception of danger and threat (and thus an indicator of exposure to a form of psychosocial stress) and integrate into the “neighborhood stress process framework,” which is itself an adaptation of Pearlin's “stress process model” used for understanding the links between stress and mental health (Pearlin et al. 1981). Yet this approach to neighborhood crime differs sharply from Sampson, Raudenbush, and Earls (1997), who combined measures from state and survey sources to construct a scale of violent crime at the neighborhood level and to elucidate the social pathways that lead to it.

Other variables see their analytical role shift and become more peripheral. For example, the concept of collective efficacy, which was central to Sampson and Raudenbush's processual ecological framework, has been recast as a protective factor against the effects of living in a disadvantaged neighborhood, and sometimes defined as synonymous with social cohesion. This remodeling is in line with the practical objectives pursued in social epigenetics: developing robust measures of health inequalities that might inform public policy. As researchers statistically test alternative causal factors against each other (quality of the built environment, neighborhood socioeconomic resources, social cohesion, etc.), they see collective efficacy and social cohesion as two potential buffering mechanisms or resilience factors at the neighborhood level. As a social epidemiologist working with the DNHS explains: “we often look at neighborhoods from a deficit standpoint of like bad neighborhoods, bad outcomes. But what fabrics within neighborhoods might be protective or buffer effects? And so I think this is where the social cohesion came up.” While this use of the concepts of collective efficacy and social cohesion is in line with their conceptual models (especially the stress process model), it also serves the more applied goal of helping to design public policy interventions by identifying so-called “modifying factors:” “I would imagine having an intervention that tries to rally communities together within neighborhoods so that you can build collective efficacy and cohesion and networks to hopefully buffer against some of the adversities that are faced in neighborhoods” (Interview with sociologist, United States, 2022).

Scientific Routinization

Gradual obliteration and conceptual shifting are intertwined with another ongoing phenomenon, which is scientific routinization. Routinization can be defined as the process by which a given technique becomes a “standard tool or routine procedure” (Fujimura 1996, 77), that is, when “researchers use these tools in a taken-for-granted manner in their work: they regard the tools as relatively unproblematic—and concentrate on solving new problems. Their focus is on making the problematic unproblematic, rather than vice versa” (Fujimura 1996, 77). Routinization encourages the use of fine-grained measures of social situations and facilitates cumulative findings and theoretical generalizations. After it is achieved, singular findings obtained in one place can speak about generic processes and help answer broader questions. In epidemiological studies, variables such as neighborhood “disorganization” or “social cohesion” become routine tools once the study design is achieved, meaning that researchers do not spend much time thinking about why they were constructed in a particular way or justifying their use in a particular research project or scientific article. In other words, “what has been closed down is not to be reopened otherwise no result could be achieved” (Bauer 2013, 518). There is a collective, tacit acknowledgement that these variables are not problematic: their use is considered relevant and credible enough.

This routinization process is clearly visible in the way social epigenetics researchers turn neighborhood differences into scientific variables. Far from being left to the imagination of researchers, neighborhood variables are by and large reproduced from one study to the next. For example, the way the MESA researchers chose the variables to capture the “social environment of the neighborhood” is a clear indication of how the “portability” of a tool—that is, “the qualities of simplicity and ease of movement and use” (Fujimura 1996, 215)—is conducive to its routinization. The researchers retained nine variables while others were left out (including opportunities for social interaction among neighbors): “aesthetic quality scale” (three items), “safety scale” (two items), “social cohesion scale” (four items). These variables had already been used in previous research: “those were measures that people in the research group had been using, that they had published on with other kinds of relevant health-related outcomes, and so that's why we picked them” (Interview with social epidemiologist, United States, 2022). The theoretical motivations for this selection are lost, but the chosen variables are considered appropriate for two main reasons. First, they provide a “summary construct” of the social environment (as one MESA researcher put it) that seems scientifically robust, as evidenced by the fact that previous studies conducted by the team have found it to correlate with several health outcomes, but also with telomere length, another important biomarker. Second, because of its simplicity it can be used in future studies, encouraging cumulative thinking that would be more complicated with tailored variables of neighborhood structure: “the more stuff you put in [the model], the less people can usually replicate” (interview with genetic epidemiologist, United States, 2022).

In epidemiological research, the routine use of certain variables is due to the need to focus on what is “problematic,” such as the construction of the statistical model (which and how many variables to include and adjust for, see Bauer 2013, 519) or the definition of statistical significance. Assessments of theoretical merit tend to be replaced by assessments of statistical validity. While this consideration is not specific to social epigenetics, it is important in a field of research often criticized for its lack of statistical rigor. Due to their limited sample size, studies in social epigenetics are often considered to lack the statistical power to detect small effects (Non 2021). Although this issue plays out differently depending on study design and on analytical strategy chosen to measure epigenetic modifications, researchers strive to obtain “strong” or “clear” signals from epigenetic measurements. Therefore, practical simplicity and statistical power are key factors in determining which variables to look at and which to ignore:

They did get information on things like street connectivity, availability of grocery stores. Like all kinds of different measures and I don’t remember now why we didn’t examine additional measures. I think part of it is that… When you’re doing anything with methylation, the datasets are so big and it's so complicated analytically that you want to try to make measurement of things as simple as possible. (Interview with social epidemiologist, United States, 2022)

To be clear, the routine use of certain variables does not mean that researchers never think about them, but rather that they do so in the context of issues that arise in a particular study and that are considered relevant by their scientific community. In particular, social epigenetics researchers are concerned with avoiding measurement “error” or “bias.” The limitations of subjective or self-reported measures of behavior have long been debated, and the development of new measurement techniques that aim to provide “objective” measures of neighborhood characteristics is often seen as driving scientific progress in the field, particularly by enabling comparisons across studies. Conversely, the fact that social experiences in epidemiological studies are often self-reported and therefore contextual is seen as a major limitation in studies investigating the links between neighborhoods and health: “If everything is self-report, of course you find a very clear link between neighborhood and health, definitely. Because if you perceive your neighborhood as poor and disorganized and I ask you about your health, you will say, ‘my health is bad’” (Interview with sociologist, United States, 2022). Researchers routinely divide neighborhood variables into two broad categories with these issues in mind: “objective” measures—mostly based on administrative statistics, especially at the census level, but also on neighborhood assessments by trained evaluators—and “subjective” measures, or people's perceptions of their neighborhoods, obtained through questionnaires.

This leads them to mostly rely on the SSO methodology for their neighborhood assessment in order to develop objective measures and thus make them seem more robust. As one article in social epigenetics states it, “Some of the strengths of this study include the robust assessment of objective neighborhood characteristics by trained assessors as opposed to self-report by the residents. This allowed perceptions of the neighborhood by those residing there to be separated from objective neighborhood characteristics” (Ward-Caviness et al. 2020, 6). This contrasts with Sampson and Raudenbush's (1999, 641) justification for using direct observation, as they argued that “visual cues are salient in many dimensions of social life” and that such observation is “a fundamental part of the sociological enterprise.”

Discussion and Conclusion

In this study, we have documented the integration, in social epigenetics, of Sampson and Raudenbush's Chicago-style sociological research on social disorganization conducted through the PHDCN. We have described this integration as selective appropriation and shown that it takes place through three processes: a pragmatic simplification during the design of cohort studies; an obliteration of sociological concepts; and a routine, unquestioned use of PHDCN variables and methods in the conduct of epigenetic studies.

These interdisciplinary exchanges can be interpreted in two opposed ways. On the one hand, one can emphasize that this interdisciplinary circulation comes at a cost. As sociological ideas gain power and influence by being adopted beyond the closed disciplinary circle they originate in (here, urban sociology), they also run the risk of losing the characteristics that led them to be recognized as worthy, sound social scientific contributions in the first place. For instance, the thorough systematic observation of neighborhoods carried out in Chicago gives way to a standardized, simplified list of variables deemed “good enough” to capture the effects of living conditions on health. Social epigenetics researchers put sociological concepts and methods at the service of the legitimacy of research on social inequalities in health, which they see as increasingly dependent on providing biological evidence of social disadvantage (Arminjon 2023). This interdisciplinary circulation comes at a cost for epidemiological studies as well. As they explore only a small portion of the scientific avenues opened up by the sociological research they draw on—whether for practical reasons or statistical considerations—they may overlook some key aspects in the neighborhood–health nexus. The most telling example is how these studies have largely set aside the notion of collective efficacy which Sampson and his colleagues considered fundamental. Moreover, the use of sociological concepts to answer different research questions, especially when integrated into other theoretical frameworks, changes their original meaning. This could lead to a shift in the explanation of health inequalities, with a stronger psychological register (“perception of crime” as a stressor; “collective efficacy” as a mechanism of social cohesion with buffering effects against stress). These distortions are reminiscent of those emphasized in relation to the adoption of the concept of social capital by epidemiologists. In particular, Fassin (2003) found that epidemiologists draw almost exclusively on Robert D. Putman's work, and make restrictive selection from his investigative apparatus, leading them to produce psychosocial rather than social explanations of the implementation of social capital.

In this article, while acknowledging that some parts of the PHDCN are left behind in interdisciplinary translation, we emphasize the search for complexity at work in its selective appropriation by social epigenetics researchers (Chiapperino 2024; Penkler 2022). We have shown that this search brings together—rather than divides or opposes—sociologists and epidemiologists who seek to analyze health outcomes as the product of social situations rather than individual behavior, including across class, gender, and race, with a shared belief that epidemiology can benefit from the use of variables rooted in sociological theory to describe embodiment processes leading to health inequalities. By making selective use of the PHDCN, social epigenetics researchers study neighborhood-level social mechanisms, question the importance of social relations in the formation of health inequalities, and rely for that purpose on measures deemed valid. At the same time, their work on complexity is both limited by strong constraints (cost and feasibility of data collection at the cohort level; statistical robustness of individual epigenetic studies) and directed at establishing causal links that may be presented in public arenas as compelling epigenetic evidence of the biological imprint of social situations.

Our study argues that a clear understanding of the interdisciplinary circulations between sociology and epidemiology benefits from closely analyzing the theory-method packages (Fujimura 1996) that support these circulations. This leads to examining the practical dimensions of biosocial research in epidemiology, and the institutional context in which it develops. In this regard, we have seen that selective appropriations of the PHDCN cannot be analyzed in isolation from the cohort infrastructures and the constraints they impose. Nor can it be properly understood without considering the search for biomarkers (of aging, disease, etc.), which now fulfils the role of persuasive techniques in a manner similar to quantification. Collective efficacy and neighborhood disorder penetrate social epigenetics insofar as they leave measurable molecular traces that provide plausible explanations for health disparities.

Supplemental Material

sj-docx-1-sth-10.1177_01622439251378192 - Supplemental material for From Neighborhoods to Molecules: The Selective Appropriation of Sociology in Social Epigenetics

Supplemental material, sj-docx-1-sth-10.1177_01622439251378192 for From Neighborhoods to Molecules: The Selective Appropriation of Sociology in Social Epigenetics by Julien Larregue and Séverine Louvel in Science, Technology, & Human Values