Abstract
Higher education scholars and practitioners recognize the essential role students’ relationships with others play in learning processes, yet research that examines how campuses shape social structures and their effects using relational methods appears limited and disconnected. We systematically review literature from 2010–2023 that has applied social network analysis (SNA) to the study of undergraduate students to identify the research questions, theoretical frameworks, and measures used. We observed that many studies lacked a clear alignment between the context of the study, network theory, research design, and results. Connections between theory and network measurement choices are fruitful areas for development. Based on our review’s findings indicating substantial opportunity to capitalize on SNA’s full potential to understand the social structures of campus life, we suggest future directions toward a shared project of using network approaches to advance equitable learning environment design for college students.
Keywords
Scholars of undergraduate education have long been interested in how students form community on campus and how that community, in turn, creates campus life (e.g., Horowitz, 1987; Moffatt, 1989). Student life practitioners, too, have historically focused on how students develop holistically as individuals through their interactions with others on campus (American Council on Education [ACE], 1937). The who, what, when, how, and why of students’ campus engagements are central to efforts to connect actionable education research and policies with positive influences on individual student outcomes (Pendakur et al., 2019). Yet, as Biancani and McFarland (2013) noted in their review of social network methodologies in studies of higher education over a decade ago, despite interest in key education research questions that are by definition relational and the existence of social network analytic methods to answer them, methodological individualism remained dominant in studies of college students. The promise of developing social network analysis (SNA) methods to understand how dynamic undergraduate student communities formed, changed, and affected outcomes remained largely unfulfilled at that time (Biancani & McFarland, 2013).
Social science research that uses ever more sophisticated network methods, including those that examine longitudinal network change, has burgeoned over the last two decades (Small et al., 2021). How that development has been incorporated into studies of students in postsecondary education is unclear, despite interest in relational research questions, potential theoretical alignment (Kolluri & Tierney, 2020), and the availability of network data. We review studies of undergraduate students published in peer-reviewed journals since 2010 that use social network analysis to understand whether and how research on structural and organizational arrangements offers implications for the development and maintenance of community on campus, in classrooms, and between students. We found that the number of publications featuring SNA in research on college students has indeed grown, particularly across contexts such as computer-mediated communication, face-to-face classroom networks, academic major programs, student friendship groups, and campus affinity groups created by programs or identities. Although researchers have capitalized on the availability of relational data, the focus has remained on individuals’ decisions, behaviors, and outcomes. Network measures were inconsistently connected to theory, preventing interpretation of network phenomena across the contexts and organizational boundaries contained within any one institution and subsequently across higher education studies more broadly. Because current studies’ use of SNA is rarely focused on the relational dimensions of campus life, we argue for continued work that advances our understanding of structural community dynamics on campus to support undergraduate learning and development.
The Study of Campus Life, Campus Culture, and Community
In her landmark study of undergraduate campus cultures, Campus Life (1987), historian Helen Lefkowitz Horowitz, identified how students organized themselves, contrasting the egalitarian language of undergraduate educational philosophy at the time with the intense hierarchies of insiders, outsiders, and rebels that emerged. Horowitz’s typology of campus life placed students into broad categories of community based on their relationships to each other and to institutional culture. Her work illustrated a motivating value of residential education: that students learn through their interactions with each other and with other important campus actors.
Around the same time, in his contemporary study of students’ self-organization, Coming of Age in New Jersey, Michael Moffatt (1989) identified a similar structure of insiders and outsiders. Moffatt achieved this through visualizing space and social relationships, mapping students’ connections within the physical constraints created by a residence hall. Moffatt observed that the built environment of residence halls and students’ tendency to segregate on the basis of racial homophily yielded student cultures that were, in reality, a series of subcultures resulting from both institutional constraints on the opportunities of Black students and Black students’ strategies to counteract structural exclusion. The protective nature of social network formation and maintenance for racially minoritized students has been echoed in more recent research which suggests that students build their networks to specifically buffer institutional violence (e.g., J. M. McCabe, 2016; P. D. Wicker, 2024).
Horowitz’s and Moffatt’s approaches—where students and their characteristics are given classification through their individual experience and relationship to campus social forces—is common across disciplines and fields that include higher education contexts (e.g., Stevens et al., 2008). What both studies powerfully illustrate is that the interactive ties that bind campus communities together have significant implications for the design of campus environments, and vice versa.
Notwithstanding the widespread modern interest and rich work on how relationships foster significant influence on college students’ success (Felten & Lambert, 2020; R. M. Johnson, 2022; Nunn, 2021; Silver, 2020), the relational qualities of constructs such as belonging, engagement, communication, and learning are often not explicitly modeled structurally. Methodologically, student engagement is analyzed through typologies (Hu & McCormick, 2012; Lingo & Chen, 2022) and the antecedents and effects of students’ friendships are modeled through counts and dyadic measures (e.g., Bowman & Park, 2014).
Some earlier work (e.g., Thomas, 2000; Small, 2017), a recent volume on relational sociology in education (Kolluri & Tierney, 2020), and J. M. McCabe’s notable books, Connecting in College (2016) and Making, Keeping, and Losing Friends (2025), explicitly employed a social network methodological approach in the observation and measurement of college student networks and pointed toward its generativity and ability to cope with relational complexity. We build on these earlier authors’ works and reviews, and books to observe how research articles since 2010 have explored student relationality on campus using SNA concepts and measures. It is particularly important to understand the trends in research articles because they contain less space to make theoretical arguments and display data, so the extent to which they can build on one another and contribute to a shared interdisciplinary project is essential. Campus life and campus culture are made through the complex infrastructure of interactions and relationships among individuals, and scholars of college student learning and success would benefit from a synthesis of theory and method that can elucidate how communities are formed, maintained, and abandoned over time (Biancani & McFarland, 2013).
Social Network Methods
One way to observe the community rather than the individual is to map the pattern of relationships that individuals experience on campus. Adopting a social network methodology encourages a focus on the structure of social relations and dependence in those social relations (Robins, 2015). In network studies, social structure and dependence are investigated by focusing on actors (individuals), their ties (relationships between individuals), and the networks of individuals that form around them (Wasserman & Faust, 1994). Because campuses are specific social contexts, each with its own culture, attendant roles for individuals, and resource availability (McFarland et al., 2011), network analysis is a useful tool for understanding the structures and dependencies that emerge because of influential factors such as curricular policy, peer subcultures, and students’ use of varying communication media.
While many early scholars of social networks used college students as their research sample, the implications of their work usually focused on social life and network theory development broadly (e.g., Festinger et al., 1950; Friedkin, 1978; Newcomb, 1962). Rarely was that work, or the work that followed, focused on implications for campus life. What is missing from the existing literature is a theory for how relational structure emerges, is maintained, or abandoned on campus, in classrooms, and across a student’s academic career. With a stronger theory of structure, institutions could intervene to build infrastructure that supports different kinds of relationship formation, potentially addressing inequalities that result as a byproduct of students’ self-sorting, and fostering learner communities where individuals have equal access to opportunities.
Social network studies explicitly investigate how structure in communities emerges from interdependence and the effects of the resulting infrastructure. Actors are presumed to engage in selection among a variety of courses of action that are influenced by social relationships and social structures. In this way, social network methods better approximate campus life. Structure—in the tradition of social network research—refers to the network of relationships, called network ties, between actors that afford and constrain communication and resource sharing (Wasserman & Faust, 1994). Dependence refers to the simultaneous influence of a tie between two or more actors. In studies of campus life, dependence is largely overlooked, especially in quantitative research, because traditional linear regression models operate on an assumption of independence between covariates (McFarland et al., 2011).
Only within the last decade, given the prevalence of trace data and the desire to move beyond individual explanations and deficit narratives, have scholars turned their concerted attention to the structural dimension of students’ interactions. Calls for greater attention to racial equity on campus are often paired with an assertion about the need to understand of the relational structure of undergraduate student life (e.g., Antonio, 2017). For example, Tichavakunda (2020) argued that campus racial climate was typically characterized as intangible or a set of aggregated individual perceptions when a relational approach would “understand the climate as composed of ever-shifting relations of interconnected entities and actors” (p. 99).
Coinciding with and perhaps motivated in part by continued interest in students’ sense of belonging, ecological models of development, and a renewed interest in socio-cultural learning theories, scholars of undergraduate student life have turned their attention toward the structural and relational aspects of students’ campus and classroom-based interactions (Brown, 2019; Kolluri & Tierney, 2020; J. M. McCabe, 2016; Smith, 2015, 2018). These studies seek to understand not just what students do on campus and with whom, but what patterns of relations facilitate or deter students’ learning and development. Mapping relational structure helps to answer important questions about student agency and institutional responsibility. An understanding of relational structure might allow institutions to more effectively organize educational environments and implement successful interventions with greater fidelity.
Efforts to determine how college affects students are largely driven by research that maps who students interact with, when, and in what configuration. In both the field of higher education studies (Biancani & McFarland, 2013) and across the broader field of education (Saqr et al., 2022), reviews on social network analysis focus on methodological approaches, formal roles on campus or in the classroom, or on narrow instructional contexts like MOOCs or online courses. Both the broad and narrow focus elides the role of institutional context and compositional diversity on relationship formation and may understate the complexity of factors that influence relationship formation among and between students.
In sum, our review is motivated by the need for a current characterization of where, how, and why network analytic methods are used to study college student campus life. At the core of our review of the literature are three related questions:
1) What methodological approaches to networks are used in studies of undergraduate students in higher education?
a. How do authors employ networks in their research—as independent variables, dependent variables, or descriptive phenomena?
b. How do researchers conceptualize and measure network structure?
2) What social contexts and areas of inquiry are studied in research that uses a network analytic approach to questions about student relational structure in higher education?
a. How do researchers conceptualize and measure network structure within different communities and contexts?
3) What mechanisms do researchers identify as driving network formation, maintenance, and dissolution in undergraduate student life? How do they theorize the relationships between these factors, changes in relational structure, and educational outcomes?
As this work covers significant territory across disciplines and fields, there is a critical need for research systematically reviewing the application of network theory and methods to the study of undergraduate education. Our goal is to identify shared understandings or key conflicts among theories and methods deployed so that future researchers can participate in building toward a larger project when they use network methods to study undergraduate students, communities, and institutions.
Research Frameworks and Systematic Review Methodology
Before we describe the methodology we used to conduct our systematic review, we first discuss the social network paradigm. We describe network concepts and terminology as we will deploy them in our systematic review methodology, findings, and discussion. We then situate the network paradigm in the more specific context of U.S. undergraduates, drawing on Smith and Vonhoff’s (2019) schema connecting network-based theories and measures to mechanisms of campus community formation. This conceptual framework undergirds our three research questions about students’ network-based communities and contexts, relational conceptualization and measurement, and network processes and outcomes, and it provides a structure for our subsequent analysis and discussion.
The Social Network Paradigm
The social network paradigm focuses on the relationship between any two entities and aggregations of those relationships as the units of analysis (Wasserman & Faust, 1994). At this point, numerous textbooks are available for detailed review of the method, so here we provide a very brief overview related to our literature review. A node is an individual entity, often a person or an organization, that has individual characteristics. A tie is the relationship between two nodes, which can be further characterized—often in terms of content, direction, mutuality, and strength. A network is a collection of nodes and the ties between them. Networks can be measured in two main ways: (1) based on the perceptions of a focal individual, resulting in that individual’s ego network (or egocentric network), or (2) a bounded group consisting of all the nodes and ties between them, called a sociometric (or sociocentric) network. Network-level measures relate to the number and arrangement of ties across the network, including density (“the proportion of ties present,” Robins, 2015, p. 21) and centrality (“an individual’s prominence or importance to the network structure,” Robins, 2015, p. 26). The individual’s position in the network can also be characterized relative to others, including network models expressed as forms of capital (Robins, 2015, p. 30). Data for network studies of college students can come from qualitative or quantitative self-reports in interviews or surveys, textual communications (such as LMS postings), institutional data (such as course enrollments), or trace data (such as card swipes in a dining hall). The choice of network data collection methods and network measures should answer the specific research questions and draw on theory related to the focal field of the phenomena in question as well as network-based theory about network structure and dynamics (Wasserman & Faust, 1994).
Conceptualizing Community
Though previous studies in higher education often draw on theories and conceptual frameworks that are based on social relationship patterns—such as social or cultural capital, constructivist pedagogy, student engagement, and college choice—higher education studies in its empirical literature have not laid out a systematic application of network analytic measurement to relational theories and concepts. To assist with organizing our systematic literature review and the resulting discussion of our findings, we draw on the Campus Ecological Networks Model (M. G. Brown & Smith, 2024). Briefly, the model builds on Renn and Arnold’s (2003) campus ecological perspective using the social network paradigm to better understand how students are situated within their interpersonal network patterns as influenced by the evolving opportunity structures of postsecondary institutions and the educational results of relational processes. The model provides a framework for considering how postsecondary institutions shape educational environments by delineating the relationships between students’ individual characteristics, the contents of their interactions with specific others, the resulting institutional social structure, and educational outcomes.
In particular, the model includes network-based theories of formation and change, detailed further in Smith and Vonhoff’s (2019) conceptualization of campus communities as networks. They focused on the relational mechanisms that institutional personnel draw on to help students connect with one another, hypothesizing that these mechanisms may in fact compete with one another in terms of network formation and the resulting community structure, and that networks can have negative outcomes as well as educationally beneficial ones (see Table 1). They described social mechanisms that help students bond with one another, such as balance, exchange, homophily, and proximity, as well as those that help students explore broadly, such as self-interest and contagion. They then connected these concepts with specific network measures that may be suited to further empirical research in higher education contexts. They argued that institutional personnel should intentionally consider which theoretical mechanism(s) they will draw on to help students form community, make program and policy decisions accordingly, and then measure the resulting networks and educational outcomes. Our conceptualization of community is based on these two approaches.
Theoretical mechanisms driving community formation in higher education (adapted from Smith & Vonhoff, 2019)
Research Methods: Systematic Review
Our research draws on best practices for systematic review of quantitative, qualitative, and mixed-methods research evidence (Page et al., 2021). Following the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA; Page et al., 2021), we reviewed social network studies of undergraduate students’ interactions and relationships in higher education from 2010–2023 that were published in English. Our search and selection process included the following terms using ERIC, ProQuest, and Google Scholar:
(“undergraduate” OR “college”) AND “student” AND (“social network analysis” OR "betweenness” OR “centrality” OR “homophily” OR “structural equivalence” OR “name generator” OR “position generator” OR “sociometric” OR “egocentric”)
Our initial search returned 2,938 articles. Our selection and inclusion criteria included the following:
Submitted to scholarly peer-reviewed publications
Published between 2010–2023, in English
The focal population is undergraduate students
Includes an egocentric or sociometric network of students as part of the analyses
When we subset articles to peer-reviewed scholarly journals, that number decreased to 1360. Bounding the time period to 2010–2023, restricting publications to English and limiting the studies to a focal population of undergraduates left us with 543 articles. After reviewing article content based on inclusion criteria and use of network analytics methods, a focal sample of 104 articles remained (see Figure 1). Articles that did not meet our inclusion criteria either lacked an analysis that employed social network data or reported results that were not related to students despite their inclusion in the sample. For example, while our corpus contains publications from the Swiss Student Life study (Boda, 2020; Elmer & Stadtfield, 2020; Elmer et al., 2020), we excluded two from our analytical corpus because they did not match our criteria. The first, which was featured in Social Networks, provided an overview of the study and important design considerations for data collection; it did not include an analysis (Vörös et al., 2020). We would direct readers to this article as they consider designing their own research as it is an excellent resource. The other study we did not include explored the efficacy of using Radio Frequency Identification (RFIDs) badges for capturing data about face-to-face interactions. As all the results for this study are related to the efficacy of RFIDS, we determined that the analysis was not research on students (Elmer et al., 2019). To assist the reader, we identify corpus studies by their article ID # Supplemental Appendix A and by their reference.
PRISMA diagram.
In our review for analytic corpus inclusion, we observed that while many studies use the language of network theory, very few collect network data. This was true for both qualitative and quantitative studies, where relationships were often investigated without data about specific ties, as in Morse et al.’s (2021) study of LGBT students’ socioemotional and academic adjustment to college through their peer affiliations. The authors argue for an approach that considers latent traits of students’ peer crowds, attempting to characterize the “landscape” of peer affiliation through a typology of crowds characterized as protesters, nonvocal, social, and athletic. Although this study used network language and theory in its framing, the article does not contain network data. The final 104 studies all presented either a network analysis or included network data in another form of analysis and reported results related to undergraduate students.
Data Analysis
We created a coding meta-matrix for each article where each row represented an article and each column represented an aspect of our deductive coding scheme. The columns included: article topic, theoretical or conceptual framework, research questions, research design, sample, demographic focus, equity concerns, geography, other institutional information, data sources, network data sources, network measures and operationalization, tie content, alter elicitation, network type, analytical strategy, outcome variable, substantive findings, future directions, disciplinary field of the authors, and network and community development mechanisms as identified by Smith and Vonhoff (2019) (e.g., balance, exchange, contagion); a version of this matrix that highlights data relevant to our research questions is also available in the online version of the journal (see Supplemental Appendix A). Each article was read and coded by two authors. After we completed a quarter of the corpus, we discussed our findings and negotiated rare discrepancies. We continued to meet after we worked through each additional quarter to discuss concordance in our coding. The first author then constructed tables to identify the frequency of article features, including study context, methodology and use of network measures, and theoretical mechanisms.
Positionality
Our review is informed by our experience as scholar-practitioners who are primarily interested in the relational dimensions of campus life. We care about advancing students’ holistic learning across a campus, constructing equity through relationships as a process, and understanding learning and development as outcomes with individual and community implications. Our perspective diverges from most authors we reviewed, who are interested in narrower collegiate contexts or phenomena. The critique that we offer is not about individual researchers’ goals but instead calls attention to the affordances of network analysis, which can help identify theoretical mechanisms that explain how structure and agency become mutually constitutive and foreground questions of equity and justice in educational contexts.
Findings
Some key trends in publishing about undergraduate networks emerged from our analysis. Among countries where research was conducted, 54% (56/104) of the studies we reviewed were conducted in the United States. The balance was from, in order of representation, the European Union (Germany, the Netherlands, Romania, Switzerland, Spain, Turkey; 21%), the UK (6.7%), Latin and South America (Brazil, Columbia, Mexico; 10.5%), Asia (China, Korea, Malaysia, Singapore, and Taiwan; 8.6%), and Canada and Tanzania (one study each). Two projects spanned global borders. Publication of social network studies steadily climbed from 2010–2018, increasing from three studies annually to 17. New publications featuring SNA precipitously declined from 2019 (with 11 studies) through 2023, with an average of 4.5 studies published annually. We did not observe trends across years related to either the types of methodologies represented or the type of network data collected (sociometric, egocentric, or alternative approaches). The journal Social Networks published the largest number of studies during our review period (5), followed by Innovative Higher Education (4), and International Review of Research in Open and Distributed Learning, CBE–Life Sciences Education, Physical Review Special Topics–Physics Education Research, Educational Technology & Society, and Interactive Learning Environments (3). This reflects the positioning of SNA as an innovative methodology in the field of higher education studies, the prevalence of research in STEM Education in higher education studies (Smith & Brown, 2020), and the increased interest in the use of educational technology and distance and online learning methods in undergraduate education.
Across the literature, we observed few examples of work that had as its goal advancing our understanding of relational and social structure on campus or making specific contributions to theories of interdependence related to campus life or undergraduate student learning and development. Many publications used network data to predict individual outcomes (e.g., can the number of ties a student possesses in the classroom predict their end-of-term grade) or used existing trace data about interactions among students as strong evidence for an instructional philosophy or educational technology. In general, we observed few instances where researchers studied network change, and equally rare were the studies that presented a theory or framework explaining how campus, classroom, or co-curricular communities form, persist, or dissolve. Throughout the empirical literature on undergraduate student life, a need for more processual attention to communities remains.
Conceptualizing and Measuring Networks
Our first research question asks, “What methodological approaches to networks are used in studies of undergraduate students in higher education?” To better illustrate how network methods and data are employed, we address two sub-questions related to methodology. First, we draw on Biancani and McFarland’s (2013) distinctions regarding how the networks are employed in research—as independent or dependent variables or as descriptive networks. Second, we report how researchers conceptualize and measure relationships and relational structure.
Across all studies, researchers employed a variety of social network methods, including egocentric and sociometric analytics (see Figure 2). We provide more insight into how different approaches to network design were employed by context in the next section. As an overview, sociometric studies were much more common than egocentric ones. We attribute this, in part, to the significant number of studies that used online interaction trace data as well as the studies that used an organizational boundary to define study participation. Eight studies measured what we refer to as latent networks—that is, instances of interaction where students had the potential to interact but may have not actually connected—using institutional data. These studies were always sociometric because using administrative data facilitated the measurement of whole networks based on colocation or co-enrollment. As an example, one study constructed a network from proximal swipes into a dining hall (Bowman et al., 2019, #86). This study did not, however, contain information about observed relationships. We refer to these as a separate form of analysis more accurately described as latent networks since they have the potential to be activated and could form a significant aspect of the opportunity structure on campus. A future direction for studies of this sort might be to cross-reference the latent network with the enacted campus network.
Types of social networks (n = 104).
The Role of Networks in Studies
Drawing on Biancani and McFarland’s (2013) prior review, we also explored how networks were used analytically in the study (see Figure 3). Very few studies attempted to predict either network structure or network change (n = 6), but these studies often offered significant implications for the design of postsecondary educational environments. For example, in a study of classroom connections, Ramsey et al. (2023, #99) identified the importance of repeated connections (what would be termed weighted connections in social network parlance, where actors repeatedly identify contacts across different relationship types or contexts) for identifying community across coursework. The authors observed that, when compared to a randomly generated selection of connections based on their course taking, students were more likely to purposefully seek out courses where they had an existing connection. Studies in this vein attempted to explain how relational structure emerges or is maintained.
Approaches to the use of networks in studies on undergraduate student life (n = 104).
Most studies that reported a focal outcome examined relationships between network measures and that outcome, where network measures served as independent variables (see Figure 4). We offer two primary reasons for this. First, researchers were particularly interested in finding associations between network concepts and student outcomes. Second, these network measures were often treated as proxies for other social phenomena such as behavioral engagement. The remainder of the outcomes followed along traditional areas of inquiry in research on undergraduate education: academic performance, attrition/persistence, learning strategies, dynamics of online learning environments (DOLE), and student (behavioral) engagement. One contrasting set of studies focused on membership prediction as an outcome. We differentiate these studies because they were not interested in interdependence; instead, they attempted to classify individuals by affiliation types.
Study objective (n = 99).
Network Measurement
Among enacted networks, we observed significant variation in the conceptualization and operationalization of network measures. We observed 17 broad categories of measurement across studies, spanning network measures such as centrality, density, and transitivity. Studies ranged from zero to six different measures of network properties. For example, 33 studies included at least one measure of degree and 23 included one measure of centrality (15 used in-degree and 14 used out-degree, the only two measures that were somewhat related across the corpus). In some cases, researchers created their own network measure, as Harrigan and Yap (2017, #55) did in their study of negative ties, proposing six mechanisms to test for network formation using an egocentric regression model (ERGM). Network measures were far more often derived from sociometric studies, whereas egocentric research was less likely to report measures beyond degree. This difference may also reflect the larger number of sociometric studies in our corpus. In egocentric studies, researchers were less likely to engage in inferential analysis (qualitative or quantitative), instead focusing on descriptive statistics. The exception to this approach was the handful of egocentric studies that extracted network values for inclusion in a predictive linear or logistic regression model (e.g., Whatley, 2018, #71).
There was little consistency in the measures that researchers used to explore relational structures, even when research questions had significant overlap. Given the wide variety of network measures and variability in the sophistication of network theorizing, an unclear picture emerges regarding the application of network methods. One common approach was to test different forms of centrality or density against each other (e.g., Smith & Fink, 2010, #1; Vignery, 2022, #96), such that studies effectively tested which operationalization of a concept predicted their outcome of interest, but not necessarily why it was associated.
Research Question and Data Sources Alignment in Network Studies
We observed two trends in the development of research questions connected to issues of network data and measurement sources. The first was authors who were interested in advancing a specific measure in social networks leading to clear alignment between the focal phenomenon and the social network measurement. For example, Alcock et al. (2020, #89) designed a study to explore the diffusion of undergraduate study habits in a mathematics course. The authors drew on theories of homophily and diffusion to identify how students took up different approaches to preparing for their coursework. They observed, perhaps counterintuitively, that neither individual nor collaborative study habits predicted academic performance. They acknowledged that, despite relying on the existing literature to guide their study design, accounting for individual and collaborative study habits may not have allowed them to capture “the interactions that matter” (p. 47) for learning.
The other trend of inquiry we observed involved authors who used relational data made available through administrative systems. Woolcott et al. (2019, #56) also explored retention and attainment in undergraduate mathematics courses using SNA. In their study, ties between individuals were identified using enrollment data, in comparison to Alcock et al. (2020, #89) combining direct measures of students’ relationships through social network surveys with administrative data. The authors used SNA to identify comparison groups and then calculated relative risk by cohort to identify courses in the mathematics sequence with higher odds of failure (Woolcott et al., 2019, #56). The authors offered an analytical strategy for identifying high-risk courses for potential intervention, but their approach did not engage with the opportunities presented by their social network data. This is not, in itself, a flaw. However, analyses of this sort could also account for how the structure of relationships among actors potentially explain individual and organizational processes. Enrollment patterns, given their weighted structure, as Woolcott et al. (2019, #56) demonstrated, could be used to determine the structure of opportunity for relationship formation, which may in turn be related to the kinds of academic difficulty students experience in a course.
Networks as Communities and Contexts
Our second review question asks, “What social contexts are studied in research that uses a network-analytical approach to questions about relational structure in higher education?” In our review, we observed five main postsecondary contextual themes across the corpus (see Figure 5). These included computer-mediated communication (CMC), face-to-face classroom networks, academic major programs, student friendship groups, and campus affinity groups based on programs or identities.
Study contexts (n = 104).
Computer-Mediated Communication
The largest cross-section of studies in our sample focused on how networks formed as part of online interactions in courses that were partially or completely online. We hesitate to refer to these networks as communities, because researchers rarely refer to them as such, instead referring to them as sets of communications or engagements. Author engagement with theories of learning is marginal, echoing the scholars of online coursework who have lamented the “lack of solid frameworks to explain learning in open online settings” (Joksimović et al., 2018, p. 43).
Across these studies, interest in online and computer-mediated learning encompasses a broad spectrum of computer-supported teaching approaches, including massive open online courses (MOOCs), hybrid and online courses in traditional higher education programs, and studies of forum and discussion participation in otherwise face-to-face courses. These studies are divided between a focus on how discourse emerges in the course (who responds to who, when, and about what) and an examination of technology-mediated interactions among learners, between learners and instructors, and between learners and content. In most cases, research studies lacked a theoretical orientation towards the emergence of social relationships. The primary outcomes of interest in these studies are behavioral engagement and retention, which is likely an artifact of the focus on MOOCs where retention and completion remain significant challenges.
Another facet of research in online undergraduate student communities focused on the emergence of discussion networks in online or hybrid classes. These studies attempted to map the emergent communication structure in the classroom through participation (or disengagement) in online forums and other course-based activities. These studies often offer a model for the latent structure of classroom communication, especially in online courses where forum or discussion data are used to model peer interactions. Early on, scholars framed these studies as occurring within social network site (SNS) technologies (even when the platform was a learning management system). Later the framing shifted from SNSs to explicitly naming the site of interaction as learning management systems (LMSs).
The vast majority (21 of 32) of CMC studies employed sociometric research designs. As we noted above, this is a byproduct of the ease with which researchers could capture interaction data from online learning platforms and construct a network. Researchers in educational technology and learning analytics have been calling for this approach for more than a decade (e.g., Dawson, 2010) and several scholars have taken up the call to try to characterize online classrooms through the affordances of big data about students’ behavioral engagement. Much less common were egocentric designs (4 studies) and latent designs (1 study).
Given that online courses in higher education are still emergent instructional contexts, it is perhaps unsurprising that most of the research in this area focused on developing network measures (9 of 33) and exploring student engagement online (10). The development of network measures could be as specific and idiosyncratic as identifying a measure of density in asynchronous online discussion (Ghadirian et al., 2019, #79) or as broad as the motivational orientation of learners in online environments (Rienties et al., 2014, #18).
Less common were studies of academic performance and student learning, making CMC studies dissimilar from other work we reviewed. We attribute this, in part, to the developing literature on student social networks in online learning contexts. As researchers develop a more advanced approach for studying online classroom dynamics using network measures (and investigating networks as an outcome, which accounted for three of the studies we reviewed), they will, we expect, eventually turn their attention to more in-depth studies of performance, persistence, and learning.
As a result, the network process most often studied in this area of the literature focused on the dynamics of online learning environments (DOLE), with a handful of studies exploring other aspects of network formation such as community in the classroom, relational structures, and peer selection. We differentiate between DOLE and other forms of social networking more commonly studied in the SNA literature, because in DOLE studies there was generally not an analogous offline process. DOLE studies focused on either interactions or processes that are fundamentally shaped by their location within online digital learning platforms, as in Toikkanen and Lipponen’s (2011, #104) study of networked learning. In that study, ties were composed of read and reply posts and a two-mode network (where interactions and events are included as nodes in the network) that defined relationships between individuals based on their participation within a thread. Even offline networks focused on discourse within a face-to-face classroom would differ significantly from the phenomena the authors sought to map here, because interactions were separated by conversation threads within the platform, were asynchronous, and lacked social cues.
Among CMC-focused studies was a strand of work that focused on the use of online social networking for pedagogy development. These studies used data from social network sites or from learning management systems to observe the formation of networks, their connectivity (using measures like density or indexes of interactions developed by the researchers), and occasionally present a typology of learner roles. These studies are difficult to compare because they focus on such varying instructional contexts, ranging from small online classroom seminar courses to large MOOCs. The result of this low consensus is that findings are difficult to connect across contexts, even when the underlying processes of discourse generation are relatively similar.
Face-to-Face Pedagogy as a Driver of Networks
A smaller literature focused on classroom community formation. These studies provided some insight into what communities look like in different face-to-face instructional contexts. As in studies of online and CMC learning contexts, face-to-face studies were also often driven by a focus on connecting in-class network measures to individual-level outcomes like performance and persistence. The characterization of classrooms as communities is a potentially fruitful direction for research applying SNA in either face-to-face or online contexts.
Similar to the studies of online learning contexts, research on face-to-face pedagogy often focused on specific iterations of instruction or pedagogy that were characterized as (inter)active. Departing from models of “sage on the stage” instruction, these studies covered a broad cross-section of active learning strategies such as equity pedagogy (Sengupta-Irving, 2014, #23) and group-based learning (Rienties & Tempelaar, 2018, #64). The focus in these studies was rarely on how active instructional techniques shaped classroom community structure but instead tended to use the number of students’ connections to study partners, friends, or other known individuals (or a similar measure of network connectivity like degree or density) to predict individual-level outcomes such as academic performance or persistence. This work, despite collecting information about ego or sociometric structure, generally used network measures either as a proxy for or a mediator of classroom interaction. In this way, network measures stand in for student engagement in the classroom—a behavioral variable rather than a relational explanation. Still, at the core of nearly all these designs was an acknowledgement that the focal pedagogy relied on shared activity.
Face-to-face studies nearly all employed sociometric designs, with one study using an egocentric approach and one study employing a latent network. In contrast to CMC studies, however, nearly all the research on face-to-face classes focused on either predicting academic performance or identifying the relationship between network features and students’ learning strategies. Because most of these studies were oriented around exploring specific pedagogical strategies, the focus on learning and performance is intuitive. The remainder of the studies explored network outcomes in some form, either investigating the impact of membership in a group (1), identifying network measures as an outcome (2), or describing the network that resulted from an instructional process (1).
In this area of research, the focal network processes were diffuse. A third of the studies focused on relational structure (4), for example, studies of group- or project-based learning or active learning strategies that required interaction among participants. Researchers sought to understand how relationships formed in the context of a service-learning course (Teymuroglu, 2013, #15), how students adopted leadership roles in a capstone design course (Novoselich & Knight, 2018, #39), and how students became behaviorally engaged in a “learning by doing” physics course (Williams et al., 2019, #80). A third of the studies focused on peer selection, most often investigating who individuals were choosing to interact with (e.g., Christopoulos, 2016, #42; Fjelkner-Pihl, 2022, #95). Finally, the remaining third of studies were distributed across group functioning, classroom community formation, boundary crossing in learning, and online network dynamics attached to a face-to-face course.
There is mixed evidence that, in classrooms using interactive pedagogies, aspects of social network structure are related to course outcomes. Although it is a widely held belief among student affairs practitioners that behavioral engagement and connections on campus promote learning, development, and retention, the articles we reviewed that tested the hypothesis of network density against academic performance offered inconsistent results. It is possible that density is not the correct measure for all network-oriented conceptions of engagement, and a number of the authors in studies we reviewed raised the question of whether we are observing the right interactions that matter for learning (e.g., Alcock et al., 2020, #89; Bruun & Brewe, 2013, #14; Rienties & Tempelaar, 2018, #18).
Academic Majors and Programs as Networks
Compared with classroom networks, a much smaller group of studies focused on students within academic majors and programs. These were less studies of community formation, per se, and more examinations of how the programmatic boundary offered an opportunity to study other related phenomena. For example, researchers explored students’ academic major awareness among community college students in general education courses (Baker, 2018, #59), students in Biology, Computer Science, and/or Math majors and their persistence (Ramsey et al., 2023, #99), and the role of demographic diversity in academic success in engineering (A. M. Johnson, 2019, #76).
Among the few studies focused on academic major programs (7), the reported research designs were split between egocentric (1), sociometric (4), and latent networks (1). Study outcomes varied among major awareness (1), characterizing a network (2), developing and testing a network measure (1), student well-being (1) and persistence (1). All three studies that focused on network processes investigated relationship formation through students’ social and academic integration (Benbow & Lee, 2022, #93; Stadtfeld et al., 2019, #77) and experiences during social distancing (Elmer et al., 2020, #87). In general, the lack of research on academic majors and programs as relational arrangements surprised us, given the prevalence of administrative data about group membership and the significant interest in students’ major choice process. The fact that this process is primarily studied as an individual behavior suggests an important future direction for research and theory. How might relational structure inform major choice and persistence, for instance?
Friendship Matters
A little more than 10% of studies in our review focused on how students formed non-academic peer relationships. These studies highlighted how social network analysis can be used to examine the complications of community, in terms of both positive and negative outcomes. This research was often pegged to the exploration of friendship formation within specific identity-focused relationships (e.g., cisgender women in sororities, Becker et al., 2018, #68; gay men’s friendship formation, Morris, 2018, #70; white students’ friendships in residential education, Mark & Harris, 2012, #8), although they did not generally offer implications for theory building focused on socio-structural properties. As in prior research, network data was often used to predict behavioral outcomes. Unlike prior studies, the focus was often on non-academic behavior such as binge drinking (Lorant & Nicaise, 2015, #31), disclosure of sexual assault (Dworkin et al., 2016, #43), and interracial integration (Ribeiro et al., 2019, #83). These studies provided evidence of social influence on behavioral engagement as well as relationship formation within particular campus subcultures.
The design and research focus of friendship studies did differ considerably from the academic focused networks we described above. For friendship studies, sociometric networks again were the primary research design (9 of 14), which meant that research tended to focus on friendships within a clearly defined organizational, programmatic, or social boundary. Less common were egocentric studies (4) and latent networks (1). Research focused on different forms of behavioral engagement (4), including on-campus involvement and network features (e.g., membership (1), network description (2), and network measure development and testing (4)). Nearly all these studies described a network process: either peer selection (3), an outcome we would expect in studies of friendship, and relational structure (5). In the case of peer selection and relational structure, studies focused on different types of important alters and multiplex relationships.
Student Affinity Groupings and Related Phenomena
The final group of studies were loosely connected in their focus on students who had sorted into affinity groups, either by institutional programming mechanisms such as participation in an exchange program (Ugurlu, 2016, #44) or their own preferences, such as in co-curricular spaces like an academic support center (Brewe et al., 2012, #10). Nearly two-thirds of these studies relied on students’ group membership or their participation in a campus program. These studies usually had a focal phenomenon related to co-curricular or extracurricular engagement.
Among this group, studies were nearly split between sociometric designs (12) and egocentric networks (11), with a few that drew on latent network data (4). We refer to these studies as having “opportunistic boundaries” because they tended to focus on short-term contexts where individuals came to be in relationships because of either the research design or a self-selection mechanism. Studies of student mobility across nation states used administrative data to place students into cohorts of migrators (Yin & Yeakey, 2019, #85), for example. Academic support resources like a physics learning center (Brewe et al., 2012, #10), STEM camps (Reding et al., 2017, #58), college mentoring programs (Ahn, 2010, #2), or campus clubs (Karimi & Matous, 2018, #75) were some of the institutionally defined contexts in which students formed relationships. Relationships were shaped by a combination of institutional design and individual preferences. In the literature on contexts, students’ relationships were often assumed to only occur in a singular context—an online or face-to-face classroom, an academic major program, or among a friendship group. The studies that employ opportunistic boundaries highlight how students’ relational decision-making occurs across multiple contexts. This contrasted with friendship studies, which, even when they employed egocentric methods, tended to focus on bounded contexts or ignored contexts for a focus on relationships.
Research topics included binge drinking, sexual health, and students’ mental health outcomes. Public health scholars, for example, asked questions about exercise habits (Patterson & Goodson, 2018, #75) within sororities because organizational membership created a natural social boundary. The implicit network mechanism in this case was contagion, although researchers rarely attempted to enumerate how social processes occurred through network properties.
Research on opportunistic boundaries is best illustrated by the handful of latent network studies. Bowman et al. (2019, #86) attempted to predict retention and graduation based on social networks derived from when a student used a card to swipe into a dining hall. Students in this study were presumed to be friends if they frequently swiped into the dining hall within a one-minute window. The authors observed significant relationships between retention in each subsequent year of college and graduation from the institution with their measure of co-dining behavior. Bowman et al.’s (2019, #86) work highlights how the varying contexts across which potentially influential relationships from on campus merits further consideration. The focus on classrooms and campus programs divorced from the larger context in which they occur potentially understates the important role that environments play in shaping learning and retention outcomes. Bowman et al.’s findings suggest that the design of campus environments creates a structure of opportunity through which students make relationships that in turn may impact their learning, development, and persistence.
Use of Network Theory
Our third research question asks, “What mechanisms do researchers identify as driving network formation, maintenance, and dissolution in undergraduate student life? How do they theorize the relationships between these factors, changes in relational structure, and educational outcomes?” With a few exceptions, studies generally exhibited a lack of network theory integration when applying network analytics to their areas of inquiry. Network theory was often positioned as a future direction.
The network theory development we observed was diffuse (see Figure 6). Researchers developed theoretical mechanisms to explain network dynamics in online learning (13), peer selection and influence (14), relational structures as an influence on student experiences in face-to-face contexts (16), and a similar body of literature on how relational structure forms and is maintained in face-to-face contexts (13).
Focal network phenomena in studies of undergraduate student life (n = 63).
The group of studies focused on the relational dimensions of online classroom participation were often informed by external contingencies of enforcement where participation was incentivized through points. These gradebook requirements shaped the structure of a classroom network. Scholars often then explored the structure of the network they could trace through students’ behavioral engagement. These studies offered examples of how students learn through online coursework and were less likely to draw on existing social network hypotheses. Much of the focus was on what might more accurately be referred to as “interaction patterns” (Lu & Churchill, 2014, #17). These studies did not share much in terms of institutional or classroom context or content, nor were shared methodologies common.
Research on disentangling influence from selection mechanisms focused exclusively on face-to-face contexts and the tie content that researchers investigated was most often multiplex, in sharp contrast to other studies which were more interested in a singular relationship type (e.g., Sengupta-Irving, 2014, #23). These studies remained focused on individual outcomes, though, so drawing propositions for network theory is difficult. In fact, across these studies we can observe that while selection and influence remain a lingering question in studies of learning, development, and college student behavioral engagement, when researchers fail to collect information about multiplex relationships, their implications are naturally limited.
This challenge is illustrated by the research studies that attempted to theorize the impact of relational structure on student outcomes. This work often used participation data (Lamm et al., 2018, #65) or administrative data (Sun et al., 2018, #61) to identify ties between students. This sort of design provides insight into one context, but relational structures cross contexts and span boundaries. These studies speak to network dynamics in discrete campus spaces and times. But theorizing about structure in contexts that are isolated from students’ other campus engagements limits the utility of results. For latent networks, this is in part because their observations do not measure an actual interaction, more often reflect a colocation (e.g. Karimi & Matous, 2018, #75; Zhu et al., 2013, #17). Researchers might consider how they could engage in studies of multiplex relationships that address context-spanning interactions.
Three studies of relational structure offer an example of what context-spanning research could look like in that they compare students’ networks (e.g., Brown, 2019, #81; Mamas, 2018, #66; Teymuroglu, 2013, #15). These studies explore the relationship of friendship networks to in-classroom group formation. By comparing students’ in-class and out-of-class relationships, researchers can start to identify testable hypotheses for future studies. Teymuroglu (2013, #15) observed that courses which used project-intensive pedagogical designs did not necessarily stimulate out-of-class friendships—a finding echoed by Brown (2019, #81) in research on first-year students during peer instructions in physics classrooms. Mamas (2018#66) also explored the relationship between friendship dynamics and group work in undergraduate courses and observed that in more advanced coursework students brought existing friendships into the classroom through which they formed groups. These three studies, which elicit network data combined with administrative or trace data, focused on multiplex relationships and network comparisons and identified important implications for classroom and campus learning design. This kind of implication can only be derived across studies that considered multiplex relationships, which was rare.
Emphasis on Bonding Mechanisms
Most of the studies in our review belong to the bonding strand of community inquiry laid out by Smith and Vonhoff (2019). Latent network examples relied on mechanisms of proximity (or propinquity, in the social network literature), although a small group of studies that considered the role of residential learning environments also operated under an assumption of proximity as a formative influence on structural relationships (e.g., Smith, 2015, #34, 2018, #72).
Homophily (where individuals gravitate to each other based on shared characteristics or preferences) was a frequently observed and powerful influence, although most of the studies started from an assumption of homophily and sought to assess its prevalence (Mark & Harris, 2012; Woolcott et al., 2017, #55). While homophily might explain something about network formation and maintenance, researchers rarely extended these explanations to help readers understand how community is affected by homophilous sorting. For example, homophily in dominant-group relationship formation might limit individual students’ access to opportunities in and out of the classroom, providing resources and experiences to students who have the easiest time forming and maintaining new relationships. Students who seek out densely knit peer relationships that are relatively closed to the environment would naturally be excluded (J. M. McCabe, 2016).
For the studies that employed some of the network concepts that Smith and Vonhoff (2019) detailed, we observed little connection to the theories that would motivate their network measurement choice. Density, for example, was a proxy for behavioral engagement among students in courses, especially in the case of CMC studies. Rather than employ and theorize network measures as network phenomena, they are too often treated as substitutes for the individualist measures that already limit our understanding of community on-campus and in the classroom.
Although little research we reviewed made the specific connections to community purpose in their theorizing, several studies did fit into the broad categories of purpose and connected theoretical mechanisms offered by Smith and Vonhoff (2019). While these studies did not specifically investigate those mechanisms, we highlight how they might have approached questions about community purpose. For example, in Identifying Gatekeepers in Online Learning Networks, Gursakal and Bozkurt (2017, #51) used measures of betweenness centrality in a MOOC as a predictor of gatekeeping behavior. This was achieved through a network census where individuals with high levels of interconnectivity informed the network’s information flow. The authors argued that gatekeepers in each of their case study courses had the potential to control the flow of information in the classroom but did not explore why this occurred.
Smith and Vonhoff (2019) give us a potential answer. Individuals acting out of self-interest in a network who construct relational networks to foster their own self-interest are simultaneously driving community formation via exploration mechanisms, in contrast to the more-often studied bonding mechanisms of homophily and propinquity. In collegiate contexts, these mechanisms may be conceptually associated with particular network measures—bonding with density and bridging with power-based notions of centrality. The network, because of an individual actor engaging in bridging of information flows, takes on an unbalanced structure. In this same study, we can see evidence of exchange (or resource sharing) and potentially contagion in that other actors over time may attempt to engage in self-interested interactions to achieve their own goals. That this is an online learning network also reinforces the mechanisms that M. G. Brown and Smith (2024) our have highlighted regarding relationship formation in campus ecological networks. Modality, frequency of communication, the “presence” of individuals and the absence of social cues all inform how the network forms in substantially different ways relative to face-to-face or blended interactions. Researchers who take a network mechanism-focused approach to the study of community in research that uses SNA could significantly advance our understanding of college student community processes.
Summary
Our findings present an initial portrait of SNA use in research on undergraduate student life and learning, suggesting significant opportunities for future research. In terms of research design, sociometric studies were more common, using a wide variety of measurements. While a handful of studies sought to explicitly advance our understanding of relational structure, more often given the proliferation of trace data, we observed researchers capitalizing on available data resources to ask recurring questions in new ways. The opportunity to address both individual- and community-level outcomes exists, and work building on this literature could better attend to broader community concerns.
Overall, we observed themes in the literature related to five main contexts. The emphasis on computer-mediated communication (CMC) as a context for inquiry was most likely an artifact of the large number of media evaluation studies and trace-data–focused studies that were popular during this period, reflecting the rapidly changing campus media ecology and the shift toward online modalities of instruction (M. G. Brown & Smith, 2024). Second, studies examined the formation and change of face-to-face classroom networks, often as the result of interactive teaching strategies. The substantial interest in active learning served as the primary motivation for many of these studies, and students’ active learning partnerships were often hypothesized to be predictive of course-level outcomes (e.g., Novoselich & Knight, 2018, #39; Putnik et al., 2016, #41). Third, the smallest aspect of the literature looked at relational structure within academic major programs. Many of these studies focused on a specific curricular or co-curricular experience, such as a first-year seminar course or participation in programs for veterans, but their inquiry was driven by an interest in relationships within an academic major program (e.g., Benbow & Lee, #93). In this way, classroom studies and academic major program studies were related. Fourth, researchers focused on the role of friendship groups in student outcomes on campus. This literature featured the majority of implications for campus environment and program design in postsecondary education (e.g., Mark & Daniels, 2012, #9; Ribiero et al., 2019, #83).
Lastly, across the literature, we observed research that used opportunistic boundaries created by affinity groups, such as academic programs like learning communities (Smith, 2018, #72) or the community college transfer process (Yücel et al., 2022, #97), to examine a related phenomenon believed to have a social component. This research often focused on existing bounded campus communities such as fraternities and sororities (e.g., Patterson & Goodson, 2018, #74) or campus-based programs designed to serve subpopulations, such as international students (McFaul, 2016, #45; Ugurlu, 2016, #44). This category highlights the potential of network studies that cross student contexts.
In terms of theory use and development, we observed a need for more research using SNA that spans organizational contexts in undergraduate student life. The reliance on administrative data for network construction naturally constrained inquiry into the contexts for which data interactions or traces were available. Similarly, the emphasis on bonding mechanisms provides insight into the factors that shape students’ relationship formation and maintenance. However, a significant opportunity exists for researchers to investigate exploration mechanisms that might better explain self-interested sorting behavior and the unbalanced structures of campus life through which challenges to retention and persistence emerge.
Constructing the Networks of Campus Life: Discussion and Implications
To understand the construction of campus life, we require research that systematically explores how structure informs individual and group functioning. Our review of research articles employing social network analysis to study aspects of collegiate student life highlights the ongoing tension between individualistic and community-oriented processes and outcomes in education. Most of the research included in our review’s corpus focused primarily on the first of our guiding questions—the associations of relational structures with individual outcomes—rather than the third, which has to do with the ways institutions shape these community environments and where the key implications for practice lie. Even when researchers employ methods that take an explicitly socio-structural approach, the pull to explain and optimize individual outcomes is strong. A focus on relational phenomena across contexts, in alignment with concerns about holistic learning experiences and social equity as a process, should build toward a shared conversation about student learning and development that crosses disciplinary boundaries.
Networks as a process and an outcome (Biancani & McFarland, 2013) align with relational theories of education (Kolluri & Tierney, 2020); however, when studies focus on narrow higher education contexts and employ network measures devoid of motivating theory, a shared lexicon among scholars of college students does not develop. Network boundary specification and the content of network ties remain thorny methodological issues in relation to educational theory. Many relational theories underline the role of power in social relationships, which structurally oriented network researchers are well positioned to take up but have historically avoided. Practically, the current body of work may lead those shaping classroom and co-curricular contexts to focus narrowly on their own settings, rather than understanding how multiplex relationships and campus-wide relational structure shape students’ opportunities and learning.
Stronger connections between theory and analysis would address many of the shortcomings we observed, although several overarching thorny issues remain in research on campus life. We discuss these issues later: the meaning of network structure and ties, the current uses and absences of theory, contexts and boundary-making, and network measurement. Interwoven in our discussion are the evergreen challenges of conducting SNA studies in higher education, including specialized training that is not routinely embedded in graduate programs and the need for original data collection requiring high response rates. Despite these challenges, studies that consider multiplex ties across multiple contexts offer paths toward theoretical alignment and clearer implications for the design of campus environments and learning experiences. The interconnectedness of our findings discussed above highlights how researchers’ decisions about focal relational phenomena, campus context, network measurement, and analysis should not be considered in isolation from one another. Continued work toward aligning substantive theory with network analytic choices would greatly strengthen the collective contributions of innovative researchers from a variety of disciplines applying SNA to higher education research and practice.
The Meaning of Network Structure and Network Tie Content
We observed three approaches to the use of social network data in research on undergraduate students. First, and most frequently, researchers used networks (or network data) as a proxy for a structural influence on individual outcomes. In this strand of work, scholars struggled with identifying meaningful relationships that have a theoretically substantive and statistically significant association with outcomes. Often, findings were counterintuitive given prior research, as in studies assuming that higher levels of connection (or degree) did not predict academic performance or retention. Study design choices prevented researchers from disentangling the moderating impact of interdependence. Rather than exploring how network structure impacts individuals, these studies treated relationships as individualized measures, failing to account for how groups, teams, or community functioning informs individual behavior. One promising approach in this literature involved studies that accounted for multiple contexts to better understand how students in a situated ecological network might make choices given the constraints of the relational structures that surround them (e.g., M. G. Brown & Smith, 2024).
In contrast to Biancani and McFarland’s (2013) framework, which identifies networks as independent or dependent variables, a considerable number of studies explored networks as interventions. In the digital and face-to-face strands of classroom literature, researchers explored processes meant to form and maintain relationships among communities of learners. In these studies, the presence of a tie was often considered evidence of the intervention’s efficacy, but scholars highlighted a similar unanswered question to research on networks as predictors: How do we identify meaningful interactions? For example, interactions mapped by discourse approaches to online classroom participation may reflect another phenomenon—one that is less organically social and more to LMS design and assessment choices such as participation points. This work also raises a corollary question: What can we understand from interactions to extrapolate toward structure? Answering both questions would advance research and practice and provide insight into how campuses might address inequality enacted through peer relationships.
The smallest group of studies focused on networks as outcomes. These studies provided insight into the kinds of relationships students are trying to form, although they were often limited by their focus on singular relationships or environmental contexts. Investigating multiplex relationships, comparing networks to each other, and tracing networks over time could help uncover how interdependent campus ecologies emerge and change.
Our review does suggest that researchers are interested and engaged in developing means to observe network content, indicating that the potential for advancing theory and practice using social network methodology has momentum. Any single study using SNA methods naturally faces pragmatic design limitations, in that researchers must delineate contextual boundaries, actor sets, tie content, and theoretical motivation. Clarity about and alignment of these choices would facilitate a shared agenda across studies and contribute to a generative conversation about the making of campus life.
Uses and Absences of Theory
We encourage scholars using SNA to offer network-theoretic approaches to the study of student opportunity, equity, learning, and development that are connected to substantive problems in student success and to postsecondary contexts. As a result, we might uncover new mechanisms that inform access to resources and experiences on campus that have important implications for inequality in opportunities and outcomes. Theories of inequality could be advanced by closer attention to the network as an opportunity structure, rather than focusing on institutional actor types and their interactions with students. In this paradigm, research questions shift from individual characteristics to the specification (and educational design) of structures and systems.
In much of the work we reviewed that engaged the theoretical meaning of measures, scholars carried forward untested assumptions from prior work (e.g., dense networks have advantages over sparse ones) when network phenomena are not well-defined in postsecondary contexts. In other network research, scholars have identified the potential ceiling on certain types of relationships or interactions that an individual can regularly engage in. Similar research on undergraduate campus experiences could help scholars identify whether encouraging denser networks is a solution to overcoming isolation or whether the sparse networks that students are connecting to are sufficient, if only they were configured in ways that provide benefits relevant to the challenges of undergraduate education.
Homophily and propinquity remained powerful social forces, especially in residential and face-to-face education where individuals can interact and sort based on visible social characteristics. That these are durable influences on the structure of peer relationships in higher education is not surprising. Still, the field and postsecondary educators would benefit from a systematic study of how homophily influences relationship formation beyond dyads with particular attention paid to when homophily is a force for buffering institutional violence and when homophily reproduces inequitable access to opportunities for undergraduate students. Recent research on Black women’s experiences in STEM, for example, suggests that students construct complex multiplex relationships that span contexts to support and inform their success (P. Wicker et al., 2023). Social network methods that consider multiple contexts and multiplex relationships offer an ideal opportunity to consider both the ideological and structural antecedents of powerful social forces such as racism (Kolluri & Tichavakunda, 2023). Oppression is always structured, and to fully address it, educators need tools that can diagnose how structure comes to be. The marginalizing forces of campus culture–that is, the interactions and exclusions that animate students’ experiences of social isolation and minoritization–are enacted through relational structures.
Relatedly, we were surprised in our review by the lack of inquiry into how power functions in the studies we reviewed. For example, despite the widespread interest in social capital in education broadly and in higher education specifically (Dika & Singh, 2002), we observed only two studies in the corpus that investigated how social capital functioned on campus using network methods (Ahn, 2010, #3; Okpych & Gray, 2021, #92) although other research built on the concept (e.g., “gay capital”, Morris, 2018, #69). It was also rare for researchers to take up experiences of minoritization and marginalization as relational phenomena using social network data. There is evidence from social network research in higher education, for example, that racially minoritized students build close-knit communities to buffer inhospitable classroom environments (Brown, 2019; #81) and campus (J. M. McCabe, 2016) environments. Transgender and non-binary students also develop close-knit communities to similar ends on campus (M. G. Brown & Smith, 2024; Nicolazzo, 2016). Yet, more recent research suggests that Black and Indigenous women experience difficulty forming the kind of relationships that support their well-being on campus (P. D. Wicker, 2024). If relationships offer protective or sustaining mechanisms, then they also need to be investigated as a source of the violence that students experience.
Network theory draws our attention to these sources of inequality in information and support sharing, access to resources, and the effects of social isolation (McDonald et al., 2024). Research that uses network methods without taking up these concerns limits our understanding of how networks inform access and equity on campus. This work could advance research that considers campus climate and address how it is operationalized through relationships on campus. Negative ties or relationships that negatively impact students’ experiences were less common and primarily connected to public health outcomes, despite increasing evidence that college student friendships can have detrimental impacts on student success and sense of belonging (J. M. McCabe, 2023; Thelamour et al., 2019). As a recent example, Henderson (2024) used SNA to observe how Asian American, Pacific Islander, and underrepresented racial minorities in engineering design courses were less likely to have their ideas enacted in design discussions illustrating how classroom environments become chilly through relational processes.
If we believe that both knowledge and identity are socially constructed, then the relational dimensions of cognition and identity development should be further conceptualized. Complementing position-based developmental models or those that look at increasing cognitive complexity, researchers could incorporate students’ interpersonal relationships, the things they talk about in varying contexts with particular people, and how their sense-making is shaped by (and shapes that of) specific others over time. Such an approach requires longitudinal engagement in network data collection, which was rare in the work that we reviewed.
Our review suggested that one of the challenges researchers may have in employing theory, and the resulting lack of structural and relational implications, is potentially because many of the studies that we reviewed, especially those focused on online learning contexts, were not designed to capture network data (and therefore were not potentially driven by questions about relational processes), but rather constructed network data from available trace data. This approach suggests a lack of systematic inquiry into social processes. These studies often elide questions of justice or inequality because they capture little to no data about the learners in their social context. Some researchers may be using network data when it is already present and convenient, rather than to answer pressing questions of community development and outcomes. In contrast to relatively simple trace data collection and analysis, the latter may require more complicated mixed-methods, or at least connections to other data sources.
Collegiate Network Contexts and the Challenge of Network Boundaries
Rarely did we observe research that was interested in either communities or nested ecological contexts. Often, researchers focused on a singular context, usually tied to a course or campus involvement niche, with less consideration for how other aspects of context might inform both the formation of relationships and the outcomes they sought to better understand. Both the CMC and face-to-face pedagogy strands of network literature conceptualized the classroom as a bubble divorced from the campus context and from students’ other social and academic connections. The focus in both cases was almost exclusively on how in-class connections predicted within-course outcomes. This approach curtailed the opportunity to include mediating or moderating impacts of other campus-based connections and off-campus relationships as explanations for how in-class networks form. Work within the other three contexts—friendship groups, academic major programs, and affinity-based networks—all appeared to be driven by the same fundamental challenge of social network analysis. Researchers sought a clearly defined boundary around which a social network could be modeled and then proceeded with questions that often had to do with either the evaluation of a program or individual behavioral outcomes (especially negative behavior such as binge drinking or disordered eating). Less common were studies that tried to develop measures of community structure or cohesion through the observation of an opportunistic boundary. Those few studies that did focus on structure tended to offer implications for network science and analysis, rather than campus life or student learning and development.
What we can observe from these studies is that studies of relational structure in higher education should more explicitly consider the interaction modality they seek to document. DOLE studies, studies of online interactions that rely on existing SNA frameworks like community formation, and studies of face-to-face interactions must wrestle with the increasingly digital nature of students’ relationships (M. G. Brown & Smith, 2024). Ecological researchers have accounted for this shift by revising their models to explicitly account for virtual and physical environments, with an acknowledgement that these contexts overlap and inform each other (Navarro & Tudge, 2023). Yet we only observed one study among the CMC research that attempted to explore students’ awareness of their peer relationships online (Lin et al., 2015, #29). Future research should investigate the relationships between physical and virtual contexts, and students’ awareness of this overlap to facilitate the design of learning and developmental experiences across modalities.
In the common current approach—which emphasizes individual-level outcomes—research lacks insight into community structure. Theory-driven research questions could influence the choice of focal context and methodological approach (sociometric, egocentric, or latent), as well as methods of eliciting and collecting data about alters. Theories of learning and development should consider the role of the network boundary on shaping interaction. Increased attention to boundary spanning and boundary objects in learning (including among some of this review’s studies, such as Rienties & Tempelaar, 2018, #64) illustrate how this approach could be used to understand interaction and to conceptualize technology-mediated learning contexts. Doing so would help researchers and practitioners who do work that recurs across contexts and whose work is affected by the broader institutional ecology.
Network Conceptualization, Data, and Measurement
Building on our finding about the need for more explicit use of network measurement, we would also call for work that takes up questions of alignment between a network-based theory in a higher education context (or network theory paired with inquiry-specific relational theory) and a network-based measure. For example, in SNA studies, density is often positively associated with individual-level outcomes because it assumes positive tie content and indicates more or better support or information or interaction—what does density mean in these contexts? Current research does not approach this question. What is the content of the ties, why is it appropriate to measure density, and what does it mean within a theory that approaches outcomes from a social constructivist perspective? Tie content itself is also often undertheorized. Researchers should consider the meaning of posting, friendship, communication, and particular types of support as connected to their phenomena of interest and as how they then contribute to broader network measures. For example, density of negative interactions should yield different outcomes than positive ones.
Next, we would encourage researchers considering the use of SNA for research on campus life to design research with questions that align with network measures. How does network structure and student social position change in different instructional contexts or amidst different policy implementations? Understanding what measures like density, centrality, and homophily mean in-context for how networks influence student outcomes would be a fruitful next step, which would inevitably require a mixed-methods design.
Current research questions most often focus on an individual’s position in the network as related to an individual outcome. A second, complementary set of research questions could examine community-based phenomena that draw on sociometric measures. Existing studies that seek to explain resulting network structures could approach this goal. What is an “ideal” structure of a community—one that facilitates the goals of teaching and learning and has the capacity to address inequality and injustice? How does that community come to be and who is involved in making it possible? To what extent can educators design this community with intentionality? Researchers like Horowitz (1987), Moffatt (1989), and J. M. McCabe (2016, 2025) have approached this question from the perspective of students’ social worlds, and such study would be enriched by investigation of networks spanning social and academic interactions.
Finally, in terms of network data, opportunities to examine social structure (or proxies for social structure) have never been greater, given the amount of trace and administrative data higher education institutions produce. Studies that use such data present a unique opportunity for connecting to educational practice and to theory development. Some studies have not yet made use of the full potential of such data. For example, it may be that a student has posted to an LMS and subsequently earned a certain grade, but the content of the posting and actual human exchanges could be examined to better answer questions about social structure’s relationship to learning. We suggest that further studies should connect trace data to qualitative and egocentric studies that help us understand what such data might mean to the humans in their offline worlds. The recent trend towards use of data on campus because it is available, without meaningful engagement about its purpose and representations, is concerning (Brown & Klein, 2023). Mixed-methods approaches can also help mitigate the potential dangers of atheoretical and decontextualized network data mining, but instead produce studies appropriately grounded in theory and context.
Implications for Future Practice
The potential for network analytic approaches to inform how practitioners facilitate the formation and maintenance of community, along with the associated benefits on campus, is enormous. The lessons this research could generate about how and why networks form, and what they can do for students and the campus writ large, are a logical extension of the current focus on individual psychological concepts such as sense of belonging (Nunn, 2021) and relational concepts such as campus climate (Tichavakunda, 2020). Aligning theories in use with structural analytical approaches produces direct implications for daily practice and policy.
Our review indicates important implications for practitioners of college student learning, development, and success. In the research we reviewed, students’ peer and social networks were sometimes latent or invisible to them. Encouraging students to proactively map their networks, identify structural holes, or consider the role of their networks in achieving individual objectives could help practitioners connect students and develop policies and programs that maintain healthy student ecologies. Similar work with instructors, especially in online learning contexts (e.g., Dawson, 2010), could have a transformative impact on pedagogy. Rather than approaching students’ peers through individual characteristics, typologies, or imagined organizational boundaries, educators may benefit from a structural perspective on how, where, and with whom students are connected across campus cultures.
Network approaches could be embedded intentionally into existing assessment and institutional research practices. Researchers already use trace data from course schedules, LMS engagement, card swipes, residence hall living arrangements, and membership rosters that can be formatted and analyzed as networks. Institutions could examine whether their policies and programs effectively connect students within and across campus microsystems. Such analyses may also help faculty and staff think beyond their most immediate campus context to better understand their students’ experiences holistically.
The making of campus life is about understanding how students’ patterns of relationships and interactions across campus contexts shift and shape their learning, well-being, degree completion, and ultimately their lives and communities beyond college. If scholars choose to adopt a social network theoretical perspective aligned with their analytical approach, we might observe a transformative shift in how we design education research, curricula, pedagogy, and assessment systems in U.S. undergraduate education. Individualist methods reflect the individualist purposes that higher education is often put in the service of—individual learning, meritocracy, job preparation, and personal economic success. Network approaches could speak to those outcomes, but they also offer a powerful antidote to those interested in the structure of power, collectives, community, and ultimately, the value of human relationships. Social network approaches, when aligned with research questions, theoretical frameworks, measurement, and data, continue to hold untapped promise for creating more generative and equitable campus communities.
Supplemental Material
sj-csv-1-rer-10.3102_00346543261438444 – Supplemental material for The Threads of Campus Life: A Review of the Literature on the Use of Social Network Analysis in Research Articles on Undergraduate Students
Supplemental material, sj-csv-1-rer-10.3102_00346543261438444 for The Threads of Campus Life: A Review of the Literature on the Use of Social Network Analysis in Research Articles on Undergraduate Students by Michael Brown, Rachel Smith, Jennifer Tipton and Gabrielle Leitner in Review of Educational Research
Footnotes
Authors
MICHAEL BROWN is an associate professor in the Center for the Study of Higher and Postsecondary Education in the Marsal Family School of Education at the University of Michigan, Ann Arbor; email:
RACHEL SMITH is associate professor of student affairs and higher education at Iowa State University, Ames, IA; email:
JENNIFER TIPTON is the assistant director of the First Scholars Office at Iowa State University, Ames, IA; email:
GABRIELLE LEITNER is a Registrar Specialist I in the Office of the Registrar where She currently supports Certifications and Eligibility processes at Iowa State University, Ames, IA; email:
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
