Diverse Disconnectedness: Homophily, Social Capital Inequality, and Student Experiences in Law School

Abstract

Law school students are encouraged frequently to “network.” However, depending on demographic categories, they may have access to differently resourced social networks in law school. In this article, we draw from our mixed-methods research to explore this diversity of experience, its limitations of access, and the possible network inequalities that may limit the value of legal education to diverse students across different institutional contexts. Using survey and network data (N = 744), collected during the fall of 2019 from three law schools, as well as supplementary interview data (N = 55), we examined students’ social networks, the structures of these relationships, and their associations with law school satisfaction. We find that, while students tended to cluster based on shared characteristics (that is, race, gender, sexual identity, political orientation, religion, and age) and contexts (that is, type of program, section assignments), these emergent clusters produced disparities in satisfaction across racial categories. Homophilous networks were tied to satisfaction for Black and White students, but the same embeddedness was associated with lower satisfaction with law school for Asian and Latinx students. These results provide grounds for rethinking how diversity matters in law school and its implications for marginalized students’ experience and success.

Law school cultures in the United States encourage students to engage in “networking”—that is, forming connections expected to promote success. This reflects the perception, supported by research on the legal profession, that social relationships play a central role in legal careers. Building and maintaining professional networks can lead to more opportunities, better performance, and higher organizational satisfaction in the legal profession (Marmaros and Sacerdote 2002; Kay and Hagan 2003; Dinovitzer 2006; Dinovitzer and Garth 2007; Harper 2008; Kay and Wallace 2009; Deo and Griffin 2011; Casciaro, Gino, and Kouchaki 2016). Moreover, a rich and growing literature has noted that alumni networks of elite universities and colleges are important for lawyers’ subsequent careers (Gargiulo and Benassi 2000; Kim 2009; Woodson 2014; Dawe 2018) and professional success (Dezalay and Garth 1996a, 1996b; Dinovitzer 2006; Khan 2012; Rivera 2015; Deo, Lazarus-Black, and Mertz 2019; Bodamer 2020; Chavez 2020). However, we know little about the ways in which social networks might matter for outcomes in law school.

Law school communications and marketing materials, not surprisingly, are filled with messages about the value of social networks but generally focus on hierarchical relationships—the connections of students to faculty, staff, and alumni (see, for example, Indiana University Maurer School of Law 2022). Less is known about students’ peer relationships and the ways in which these networks are organized along the lines of students’ demographic identities. Social ties are rarely distributed evenly across demographic categories, and, even if they are, the resources that students might mobilize from them are not necessarily equally beneficial. Particularly, students from a range of demographic categories—race, gender, sexuality, national status, political orientation, religion, age, and class—may have differential access to social capital, resulting in varied, unequal experiences, tracks, or pathways even within seemingly shared environments (Wilkins and Gulati 1996). Furthermore, different educational environments might have different kinds of structures that foster and constrain social connections for students with historically marginalized identities in law, but these organizational differences are rarely studied across comparative institutional contexts. To address these myriad dynamics and phenomena, our research pursues a comparative, multilevel consideration of a combination of circumstances—diversity of experiences, differential access, and resulting “network inequalities”—which may produce inequities in the value of legal education for different kinds of diverse students.

In this article, we take up the issue of social networks among law students by focusing on three interrelated empirical issues. First, utilizing a unique multi-method dataset of students’ social ties, we explore what law school social networks look like and the types of social capital associated with these network structures. Specifically, we explore patterns of social networks among first-year juris doctor of law (JD) and master of laws (LLM) students across three differently ranked and situated law schools, each feeding into the same large, cosmopolitan market for legal services in the United States. Second, we examine whether students’ demographic identities are linked to clustering in these networks, which is consistent with the notion of homophily based on shared identities and network closure. We utilize exponential random graph models (ERGMs), an inferential network method, to estimate associations between students’ demographic identities and the presence of ties. Finally, we investigate whether students’ personal networks are important for their experiences in law school, as measured by law school satisfaction, by employing regression modeling and qualitative interview data. We focus on satisfaction because it is an important marker for understanding organizational inequalities in the legal profession (see, for example, Lempert, Chambers, and Adams 2000; Dinovitzer and Garth 2007; Hagan and Kay 2007; Silver and Watkins 2012; Chambers 2019). Our social network data enables us to introduce a meso-level perspective to phenomena, which is typically limited to a micro-level (individual students) or macro-level (law schools as organizations) analysis.

Our findings have several interrelated implications for thinking about inclusivity in educational environments and the intergroup variations that predicate student experiences within them. Below, our results show significant patterns of homophily—the tendency of “birds of a feather to flock together” (McPherson, Smith-Lovin, and Cook 2001)—based on different demographic categories (that is, race, gender, sexualities, international status, political orientation, religion, age, and class) and triadic closure or the tendency of a friend of a friend to become a friend (the two combine to produce homophilous clusters in these law school networks). The striking racial segregation that we observe in these networks falls in line with what other social network scholars have found for undergraduate and graduate students (see, for example, Wimmer and Lewis 2010; McCabe 2016; Rethemeyer and Ryu 2020) and resonates with the argument that intra-racial socialization might be crucial for law students responding to an otherwise hostile environment (Pan 2017).

However, beyond these expected measures of homophily and segregation, our data offer new insights for considering the implications of embedded segregation on disparate law student satisfaction. Particularly, we find that, while racially homophilous networks are associated with lower law school satisfaction for Asian and Latinx students, embeddedness with same-race peers increases satisfaction for White and, importantly, Black students. For White students, who are in the demographic majority, the finding that in-group bonding is beneficial tracks the logic that, as the normative category of an ideal law student, they benefit from social exclusion and holding their resources within their own networks. Our findings for Black students call for more critical analysis. On the one hand, it might extend Yung-Yi Pan’s (2017) theory of “incidental racialization” to suggest that these marginalized students find a sense of belonging with each other and that this social exclusion from the dominant student group—unlike their other racially diverse peers—heightens their sense of satisfaction. At the same time, their distance from the normative category of law student (especially when compared to their other racially diverse peers) might suggest that they do not seek such assimilation and, therefore, are not dissatisfied when it does not occur. It is this variation in satisfaction amongst racially diverse students that is predicted by the nature of disconnection from the dominant group that we call “diverse disconnectedness.” We suggest this heterogeneity in networking—between different categories of students, their networks, and their law schools overall—can inform our understandings about students’ racialized experiences as well as the institutional responses that might nourish their growth across a range of contexts.

Background

Following Pierre Bourdieu (1986), who extended the concept of capital beyond economic resources to include networks of relationships channeling information, resources, and even culture, there has been a line of scholarship analyzing how social and cultural capital produces advantages in elite professional work (see, for example, Gorman 2015; Reid 2015; Rivera 2015). Social capital refers to resources and support accessed through social networks. Cultural capital, alternatively, captures the notion that aspects of culture—symbols, practices, repertoires, and tastes—can be used to accumulate social capital and to display fitness with organizational contexts. In Bourdieu’s account, economic capital can be converted to social and cultural capital within a given institutional context to socially reproduce hierarchy. When demographically different actors access the same institution, their ability to gain from the structure, once within them or after, is differentially predicated on their cultural resources enabling them to fit into organizations and their social networks channeling resources and support (Gorman 2015).

Research on lawyers’ careers consistently support these observations about social and cultural capital. Regarding the former, John Hagan and Fiona Kay (1995) show how, despite the changing recruitment practices that have hired more women at entry, equality was compromised by network structures disadvantaging women, leading to fewer promotions, less respect, and lower retention. Women lawyers are more disadvantaged in terms of building social connections in the legal profession; these social capital inequalities partially accounted for disparities in promotions to partner (Kay and Hagan 1998) and work satisfaction (Kay and Wallace 2009). This denial of opportunity has, over time, left women less likely to trust the institutions in which they are embedded (Kay and Hagan 2003) and, consequently, more likely to leave the spaces that they find inhospitable. Similarly, as Swethaa Ballakrishnen (2022) reveals in their data on Muslim lawyers, religious identities and their social reception shape professional navigation across legal institutions, and comparative research consistently finds career disparities between racially diverse (Kay 2018) and religiously diverse lawyers (Dinovitzer 2006) are partially accounted for by inequalities in social capital.

Regarding cultural capital, some individuals possess symbolic advantages within institutional contexts that enhance their likelihood of being selected and receiving greater rewards (Dezalay and Garth 1996a, 1996b, 2021; Mertz 2007; Gomez and Perez-Perdomo 2018; Ballakrishnen 2021). For example, scholarship theorizing identities within the legal profession (see, for example, Wilkins 1998; Sommerlad 2007; Webley et al. 2016) has revealed cultures that “despite change, remain largely unwelcoming to outsiders, inducing assimilation or exit” (Sommerlad 2007, 194). New entrants with pre-existing cultural capital, such as high socioeconomic status, elite education, and historically normative demographic identities (for example, White men), have distinct advantages (see, for example, Payne-Pikus, Hagan, and Nelson 2010; Headworth et al. 2016). Bryant Garth and Joyce Sterling (2018) argue that the lack of partners of color in large law firms, despite increases in minority pipelines from law schools, is primarily attributable to comfort and fit based on class-based symbolic practices. While those from privileged backgrounds, regardless of race, find synergies with large law firm cultures that allow for ease of navigation, those most likely to stay are lawyers (mostly White and male) who had the “right” to enter these spaces.

Cultural and Social Capital in Law Schools

We apply these insights about social and cultural capital in the legal profession to the study of law student networks. Connections in law school—to other students, faculty, staff, alumni, potential employers, and others in the profession—may act as important network structures of social capital that channel resources, reciprocity, and support (Burt 2005; Glanville and Bienenstock 2009), thereby mobilizing better law school experiences (for example, satisfaction) as well as concrete rewards (for example, grades and jobs). But little is known about how social capital is held within, and dispersed across, demographic categories of students in law school. Studying law students, especially in professional schools, can offer crucial insights into the ways in which social capital initially gets organized by cultural capital and accumulated before entry into the profession. This focus on the roots of relational inequalities may provide insights about the emergence of cumulative advantage in professional careers (DiPrete and Eirich 2006).

Specifically, we examine how demographic identities influence who is connected with whom in law schools—that is, how cultural capital influences the acquisition of social capital. Law schools have become diverse, with increasing representation from historically marginalized identities (that is, women, LGBTQ students, students of color, and international students). In turn, for many of these students, navigating a professional identity has required an internal recalibration about the role of their own identities and their salience with their external ambitions (Bliss 2017; Ballakrishnen 2023). These internal negotiations with their environments have implications for how they interact with others in the community. Pan (2017) suggests that students of color engage in “incidental racialization,” forming ties with similar others in response to exclusive White networks that isolate the former. Similarly, international and Muslim law students, attuned to their outsider status, find that connecting with similar others helps them to navigate their sometimes alienating school environments (Ballakrishnen and Silver 2019; Ballakrishnen 2022). Regarding markers of class, students with relatives who are lawyers have access to insider knowledge and connections to the legal profession; it follows then that this cultural capital can lead to the formation of ties between advantaged students. Similarly, students from highly ranked undergraduate colleges and universities may tend to engage in social closure based on their elite credentials. Taken together, we consider the importance of several types of demographic categories (race, gender, sexualities, international status, political orientation, religion, age, and class) in relation to who students connect with and expected patterns of homophily—the tendency for “birds of a feather to flock together”—which is partially in response to social closure by students who possess the “right” demographic identities.

We also examine the importance of organizational factors for structuring social connections in law school. Organizational factors, such as program (that is, LLM and JD) and JD section assignments (which generally organize first-year (1L) JD students in their required courses and typically do not encompass LLMs), could induce proximity even among students without shared demographic attributes and thus offer the opportunity for routine social interaction (Feld 1981). As other research has suggested (Park et al. 2011; Silver 2013; Ballakrishnen and Silver 2019), program and section distinctions between LLM and JD students create strong incentives for students to form and maintain ties within (rather than across) programs because of the ways in which they are structured and executed. Organizational factors may promote clustering based on program distinctions (JD versus LLM students) but may also increase diversity in students’ personal networks by promoting within-section ties that cut across demographic identities.

Differential Access to Social Capital

It is this intuition—that different categories of actors might respond to seemingly neutral institutions in different and socially interdependent ways—that roots our research. For instance, although homophilous clusters seem similarly insular, they might house different resources and have the capacity to engender advantages differently based on a range of factors. Similarly, these resources could produce a social trap, where the pursuit of individual-level advantages leads to collectively undesirable outcomes (for example, segregation). Students from similar college backgrounds, for example, or those who have lawyers in their extended kinship networks, may possess cultural capital that produces resource-rich ties at the individual level, but these very resources could produce and promote inequalities in the larger networks in which these students participate. Relatedly, students of color might have resource rich ties that serve them at the individual level (for example, close friendships with like others in student groups), but this could isolate them even further within the larger law school community.

Existing literature on social networks in other contexts highlights two forms of social capital that exist in interpersonal relationships: capital that helps with bonding among similarly situated peers in dense networks (Coleman 1988), mentioned earlier, and capital that helps with bridging across broader networks with peers who might be more heterogenous (Burt 2005). Each of these forms of social capital promote different kinds of advantages for individuals usually predicated on different kinds of social connections. Particularly, while “bonding” is associated with dense network structures that serve to promote trust, shared identities, and norm enforcement, “bridging” reflects brokerage structures that position actors at the nexus of diverse information and resource flows (Burt 2005; Paik and Navarre-Jackson 2011).

Following the extant literature (for a review, see Shin 2021), we utilize network measures to operationalize bonding and bridging social capital. Specifically, we use measures of homophily and triadic closure to tap bonding social capital and popularity as a measure of bridging social capital. The notion of capital implies a scarce resource that is likely to be distributed unequally. Network inequalities may be produced by implicit bias, prejudice, and discrimination, thereby creating, sustaining, or even exacerbating inequalities across demographic categories (McDonald 2011). This also suggests that there may be differential access to social capital based on demographic categories. Specifically, we focus on racial disparities in social capital as well as on intra-racial differences.

Social Capital and Student Satisfaction

Student satisfaction, defined as subjective assessments of students’ experiences and outcomes at school, has long been considered an important outcome in higher education research (Elliott and Shin 2002).¹ Law school satisfaction is regularly used by law schools as a global measure of student experiences (Quintanilla 2018; Quintanilla and Erman 2020; Green et al. 2022; Petzold 2022).² It is positively related to student interaction with other students and with faculty (Korobova and Starobin 2015, 75; Kuh 2019; LSSSE 2019, 5; Quintanilla 2018; Wong and Chapman 2023) and, consequently, is relevant to this research where the focus is on students’ social networks. Studies have found lower satisfaction among students of color and students with historically marginalized identities (Fan, Luchok, and Dozier 2021). Wan Hoong Wong and Elaine Chapman (2023) found that interaction among students was positively associated with student satisfaction in higher education; further, their work “suggests that student-student interaction could be the most critical form of interaction in terms of student satisfaction levels” (18). Analyzing the Law School Survey of Student Engagement (LSSSE), Victor Quintanilla and Sam Erman (2020) found that students of color had weaker ties with other students and faculty, thereby attenuating their satisfaction with law school.³ Despite this research, little is known about the network structures contributing to this racial gap in law school satisfaction.

We expect that measures of social capital will be associated with law school satisfaction. To the extent that bridging social capital provides students with diverse information flows and resources, we expect that network popularity will be associated with law school satisfaction. Moreover, because bonding social capital is linked to trust and support, we expect that higher levels of homophily and triadic closure will be associated with law school satisfaction. However, bonding social capital also may have a “dark side” (Kwon and Adler 2014): dense social networks with few outside ties can constrain individuals to a limited, isolating social world (Beasley 2012). To the extent that students prefer greater inclusion but are disconnected due to social closure, we expect to see a negative association between bonding social capital and law school satisfaction. Research suggests that students of color, in particular, benefit from interracial relationships (Woodson 2023). Indeed, this article maps how disconnectedness—the flip side of bonding social capital—impacts students differently, especially if we consider the ways in which they experience satisfaction within their environments.

This idea of diversity in disconnectedness adds to these findings of racial homophily and differential experience by mapping how embeddedness in social relationships might explain the ways in which students have distinct patterns of distance away from normative institutions and identities. Just as students’ demographic characteristics might impact their patterns of social clustering, segregation, and experience (see, for example, Quintanilla 2018; Quintanilla and Erman 2020), so might their differential distance and lack of connection from normative structures and students. Differences in satisfaction, predicated on the specificities of this deviation among different kinds of homophilous groups, may offer new ways to think about the larger literature on bridging and bonding. In turn, these findings could have important implications for what we think of as organizational responses to particular kinds of minority actors and their experience of law school.

Data and Methods

Data were drawn from the first wave of the Student Experiences in Law School Study (SELSS), a multimethod project undertaken by an interdisciplinary research team with expertise and experience in law and sociology. The first wave of the project, conducted in October and November 2019, consisted of 744 web-survey responses and fifty-five in-depth interviews of beginning law students enrolled in JD and LLM programs at three law schools that were in or near a single large city in the United States. For the first wave, we invited all first-year JD and LLM students at the three law schools to take our survey. A subset of web-survey respondents, varying by gender, race, national status, program, and school, were then invited to participate in follow-up interviews. Participants received fifteen dollars per completed response (both surveys and interviews were compensated) and also were entered into a raffle. Recruitment included email messages and in-person efforts to attract participation. The overall first wave response rate for the web survey was 64 percent, which suggests we were successful in generating a high-level of participation in these student cohorts.

Participating law schools were selected for their commonalities as well as differences. Because all three schools are proximate to the same large city, their law students frequently sought employment opportunities in this legal market during and after law school. We anticipated that proximity to the same major legal market would reveal students’ connections to specific local organizations and individuals, which could reflect potential disparities in access. The schools differed in terms of selectivity, US News & World Report ranking, size, affiliation, and LLM programs. We labeled each school with a numeric designation, reflecting their relative law school ranking: School 1 was ranked in the top twenty-five in the 2019 US News & World Report law school rankings; School 2 was in the 25–100 group; and School 3 was unranked (US News & World Report 2020). Included in the three were public and private law schools. Most LLM students at Schools 1 and 2 were international, whereas School 3’s (much smaller) LLM programs were marketed to lawyers in the region. The law schools included in this research are not the only law schools feeding into the same legal market.

The survey instrument asked students about their demographic characteristics (for example, race, gender); primary romantic or sexual partners; social connections with other students, faculty, staff, and alumni; membership in student groups; prior experience and family background in law; law school experiences; and career aspirations. Since the first wave was fielded in students’ first semester of law school, grades were not available. Surveys also provided respondents with a comprehensive roster of their combined first-year JD and LLM classmates and asked them to nominate up to twenty students with whom they regularly discussed school or socialized. We used these data to track nominations sent and received among survey respondents (that is, who they nominated as well as who nominated them), which in turn provides the data for our network analysis. Interviews focused on experiences and thoughts about networking and social connectedness.

In this article, our focus is on students’ initial entry to law school, their community-building during this period, and its relationship to satisfaction. Our analysis focused on students’ social networks shortly after the start of law school as reported by the first wave and interview respondents. As such, these data captured students’ formative connections with one another prior to receiving their fall semester grades. Complete respondent data for independent variables were available for approximately 690 respondents. To maximize reported network ties included in our social network analysis (SNA), we imputed values for respondents with missing data on just one or two variables. The analytical sample with imputation consisted of 737 respondents with complete information regarding nominations sent and received and excluded students who were nominated by respondents but were not survey respondents themselves (n = 334).⁴ Table 1 presents descriptive statistics for respondent-level variables across programs and law schools.

Table 1. Descriptive statistics for respondent-level variables (n=737)

Estimates are percentages and means (sd in parentheses)

Dependent Measures

To examine students’ school-based networks and their overall experience in law school, we focused on two outcomes: (1) ties between respondents, which indicated for every pair of respondents whether one nominated the other and vice versa⁵ and (2) law-school satisfaction, which was based on a single item (how satisfied are you with your legal education?) and answered on a five-point scale (1 = very dissatisfied; 5 = very satisfied).⁶

Independent Measures

Individual Characteristics

We utilized measures of respondent-level demographic identities, including binary variables for racial categories (Asian, Black, Latinx, and White),⁷ national origin (whether or not students indicated they were international), sexual identity (heterosexual versus lesbian/gay/bisexual/queer/other), gender (with women and nonbinary combined due to small numbers for the latter), religion (Christian, Jewish, other or no religion), class, political orientation, and age. Due to the limited number of cases in the other-religion category, we could not use a more fine-grained measure of religion. We assessed class differences using two binary variables: the first tapped whether respondents had “immediate or extended family members” who were lawyers; the second indicated whether respondents attended, based on US News & World Report’s 2020 rankings, a top twenty-five undergraduate college or university in the United States. Political orientation was a summated scale (alpha = 0.74) of two questions on five-point scales (1 = very conservative; 5 = very liberal) about political beliefs on social and economic/fiscal issues. These measures were included as respondent-level predictors in our regression models and as homophily measures in our ERGMs.

Organizational Characteristics

We also generated binary variables to indicate program distinctions (that is, JD or LLM) within the law schools as well as for different 1L section assignments (as proxies for shared classes) across JD students. Program distinctions and section assignments were respondent-level predictors in our regressions and homophily measures in our ERGMs.

Social Capital Measures

We utilized racial homophily and triadic closure as measures for bonding social capital and popularity as a measure for bridging social capital. To examine the level of racial homophily around each respondent, we utilized a modified version of a network homophily measure known as the E-I index (for a review, see Bojanowski and Corten 2014), which has been used previously to examine the demographic composition of personal networks in higher education and firms (see, for example, Lee, Chung, and Park 2018), but it has yet to be used to study personal networks in law schools. For each respondent’s personal network, we calculated the numerator as the number of ties with same-race alters minus the number of ties with different-race alters and divided this quantity by the total number of alters in the respondent’s personal network. This index ranged from –1 to 1, with the former indicating respondents with personal networks completely comprised by different-race alters and the latter completely with same-race alters. Following the notion that a friend of a friend tends to become a friend, we assessed the extent to which triads were closed (that is, formed a triangle). We included an overall measure of triadic closure in our inferential network models⁸ and a respondent-level measure of triadic closure in our regression models. For each respondent’s personal network, as defined by those (alters) nominating or being nominated by the former, we calculated the proportion of alters that were connected to one another, thereby closing the triad with the respondent. Lastly, respondents’ popularity was calculated as the number of network nominations received.⁹ In our inferential network model, we used a geometrically weighted version of number of nominations received.

Control Measures

Finally, we included several control variables in our models. We employed measures for whether respondents were partnered (that is, married or cohabiting), had children, were currently employed, and had joined student organizations. Each of these variables might differently affect respondents’ availability to connect with other students and law school satisfaction. In our inferential network models, we included control variables for edges (that is, the overall density of each law school’s network) and mutuality (reciprocated ties between pairs of respondents); these two controls are necessary for obtaining accurate estimates of more complex network structures, such as homophily, triadic closure, and popularity.

Modeling

In the following sections, we focus on three analyses to make sense of students’ social connections. First, we used descriptive network visualizations to examine broad patterns of homophily and students’ tendencies to make communities. These visualizations recognize the analytical intervention of networks as meso-level relational data that contribute to research on legal education, which has focused primarily on individual- (micro) or organizational- (macro) level data. Next, we estimated ERGMs, an inferential network model designed to examine the presence of ties and that can estimate whether tendencies for homophily, triadic closure, and popularity can account for observed networks (Morris, Hancock, and Hunter 2008). In our ERGMs, we included homophily measures for demographic identities and organizational structure, which allowed us to estimate whether students with shared characteristics were more likely to form ties with those that shared their identities. We also included measures for triadic closure, popularity, and the above-mentioned controls.¹⁰ Finally, we considered how differences in personal networks, particularly seen through the lenses of bonding and bridging social capital, relate to variation in students’ law school satisfaction. To model law school satisfaction, we employed ordinary least squares regression and included the above-mentioned demographic identities and organizational characteristics (program and section), bonding social capital (racial homophily and triadic closure), bridging social capital (popularity), and control variables for children, intimate partnerships, employment, the number of student organization affiliations, school dummies, and popularity. We contextualized our main quantitative findings drawing from our in-depth interviews with students across the three schools.

Interview Data

To complement the quantitative survey and network data, we examined our interview data to illustrate patterns identified in the quantitative modeling. Interviews were conducted with fifty-five survey respondents at the end of the first wave of survey administration. Prospective interviewees initially were identified from survey respondents to oversample for students of color, students who identified as LGBTQ, and students who indicated they were international, while ensuring a sample of interviewees across the three schools. Interviewers reached out to prospective interviewees by email to invite them to participate in an interview and to schedule a time to talk. Most of the first wave interviews were conducted in person. Interviews were conducted in the first wave with fifteen students each at Schools 1 and 3 and twenty-five students at School 2. Most of the interviews were with JD students. We conducted interviews with fifteen LLM students, nearly all of whom were enrolled in two of the schools; at one school, we interviewed only one LLM student. Thirteen of the LLMs identified as international.¹¹ The first wave interviews lasted approximately one hour; all interviews were recorded. Recordings were transcribed, and pseudonyms were assigned to each interviewee. Pseudonyms were derived to reflect interviewees’ shared identities and names, including using names common to an international interviewee’s home country and, for students using American names, following that choice. Transcripts were reviewed by research team members who conducted the interviews and their research assistants to identify emerging themes from the interviews and surveys. The analysis of interview transcripts was shaped and oriented toward findings emerging from the ERGMs and the regression models.

Results and Findings

We first evaluated the representativeness of our samples by comparing descriptive statistics, presented in Table 1, against the American Bar Association’s (ABA) mandatory disclosures made by each school for race and gender. According to the ABA, White students comprised between a majority and two-thirds of the first-year JD class across the three schools (ABA 2020). In the SELSS data, White/other students accounted for 60 percent of JD respondents at School 1, 67 percent at School 2, and 69 percent at School 3. Latinx students were 12–17 percent of JD respondent pools; Black students, 4–8 percent of JD respondents at the three schools; and Asian students, 10–20 percent of the schools’ JD respondents. Similarly, men comprised between 35 and 50 percent of the 1L population at the schools (ABA 2020); the SELSS samples had similar statistics for gender, with men accounting for 30–47 percent of respondents at each school. Taken together, comparing the SELSS data with the ABA’s mandatory disclosures suggests that the former approximated the latter, at least in terms of gender and race, for the aggregate of the schools. Differences between the ABA’s mandatory disclosure and our respondent pool indicates fewer diverse respondents for School 3 compared to their proportional representation at that school.

While each of the three law schools had LLM programs, their student demographics were not disclosed through the ABA’s mandatory disclosures. Table 1 shows that the demographic composition of LLM respondents was quite different compared not only to JD students but also across schools. School 1’s LLM respondents were primarily Asian (55 percent) or Latinx (26 percent), international (87 percent), and women (61 percent). At School 2, most LLM respondents were international (98 percent), Asian (48 percent), White (47 percent), and women (61 percent). In contrast, the small number of LLM respondents at School 3 were comprised of students who were domestic (92 percent), White (62 percent), and men (62 percent).

Descriptive Analyses

To get a bird’s-eye perspective of the network structure of law school, we first created network visualizations for each school (see Figures 1–3). These visualizations employed an algorithm where “nodes” (circles representing students) were simultaneously repelled from one another but held together by “springs” (directed arrows showing when one student nominated another).¹² The top and bottom panels for each school have identical orientations, but, in the top panels, the colors reflect degree program for LLMs and sections for JD students. In the bottom panels, the colors reflect race, as reported by respondents.

Figure 1. School 1 network structure by race and program/section.

Figure 2. School 2 network structure by race and program/section.

Figure 3. School 3 network structure by race and program/section.

Looking first at School 1, LLMs (in orange) are clustered at the bottom of Figure 1 in two large groups. At the top of the figure, JDs are clustered, and many appear to be organized around sections (blue, purple, green, and, to a lesser extent, red). Relatively fewer ties connect the two LLM clusters to the JD cluster. This finding was corroborated by Lola Garcia, a Brazilian LLM at School 1, who explained that she does not mix with JDs “very much. But a little bit. I have some classes with 1Ls. They are really competitive here in the US, because they need the grade, and they need to find jobs. So, this competitive thing that they have here, we don’t have in Brazil. It is a little bit different.”

The bottom panel of Figure 1 highlights that the two LLM clusters appear to be segregated by race, reflecting national origin. The larger LLM cluster is comprised of Asian international students, whereas the smaller one is populated by Latinx international students. In the top panel of the figure, White students in the large JD cluster appear to be particularly distant from the international LLM students. Most Black JD students are located near the center of this cluster, while Asian and Latinx JD students appear closer to the Asian and Latinx LLM clusters, respectively. This may reflect the international JD population of School 1. Relatedly, law school organizations can ameliorate the divisions structured by law school degree programs and sections. As Linda Martinez, an LLM at School 1 explained, although her social interactions were still mostly with LLMs, student organizations offered chances to meet JD students: “There is an organization called … I forgot the name. … But … it involves Latin Americans, including JDs.” Overall, Figure 1 highlights the importance of organizational affiliations (that is, through section assignment and through variations in program) and racial homophily for structuring students’ relationships in School 1.

The network visualization for School 2, presented in Figure 2, shows a similar pattern where law school social networks are organized largely by program and section as well as by race. In the top panel of Figure 2, clustering by JD sections, compared to School 1, appears to be more pronounced. In interviews with School 2 respondents, sections also emerged as an important source of relationships. Cody, a 1L at School 2, commented that he could identify at least one person who was “definitely a close friend…. Just through him sitting next to me in contracts; you know, we talk a lot, and we can relate on things. He’s cool. Monday [in October] was the first time we hung out outside of school.”

Similar to School 1, School 2’s LLM students appear segregated from JDs and seem to organize into two smaller clusters associated with race and, relatedly, national origin. However, connections between LLM and JD students appear to be less frequent in School 2 compared to School 1, which may be related to the greater proportion of international JDs in School 1 where we saw that shared international status facilitated connections between JD and LLM students (Silver and Ballakrishnen 2018; ABA 2020). Dasha, a School 2 LLM, suggested that the in-class opportunity to meet friends still tilts toward differences in degree programs by commenting that “in our classes we have also JD students. Sometimes we work together, but not much as with LLM students.” The bottom panel of Figure 2 shows that White and Asian LLM students appear to be segregated from one another at School 2. Interestingly, two of the JD sections appear to be racially heterogeneous, including Black, Asian, Latinx, and White students, whereas the remaining JD section is primarily White and Latinx. These patterns suggest that JD section assignments may attenuate the strong tendency for racial homophily among students in our data.

School 3, in contrast, shows a markedly different pattern with even more pronounced section-based clustering and minimal clustering by race. The top panel of Figure 3 shows three large clusters, which are organized by section. These patterns of homophily, we believe, were a function of several factors specific to the school, such as a small number of LLM students in our data, the higher proportion of White students, and a significant commuter population (making the centrality of their section assignments more important than any other sorting that might have led to networks). Differences in commuting patterns can affect relationships, complicated by course scheduling that may bring only particular sections into the building on certain weekdays, for example. Other factors possibly affecting networks relate to the use of space in the law school and how community meeting areas are designed and used. These factors may interact with one another and can affect relationships, as Kyla, a School 3 JD student commented: “I live across the street from the law school, too. So, I don’t feel a need to stay there until 10:00 p.m. and study. A lot of them do, because they live pretty far. I guess in that sense I’m a little bit different.” She also commented that proximity has affected her friendships: “Two of the people that I’ve met and become really good friends with was not necessarily because of the law school. We live in the same apartment building, too.” Another student, Luke Miller, had the opposite experience: he described the five people he studies with as friends: “But almost not where it’s social. Just in the law school aspect. I live so far up north; I can’t really ask them to go hang out with me and grab lunch. It’s too out of the way for me, and I’m sure it’s out of the way for them, too, to come to my side of town…. The people that I study with live within 10–15 minutes from here. And I’m almost an hour by public transportation.” The bottom panel appears to show that there are no obvious patterns of clustering by race. Taken together, all three figures suggest that students’ ties are strongly organized by program and section distinctions, and, where the institutional conditions permit—as they did in Schools 1 and 2—same-race ties are more likely to form than interracial connections.

Our statistical analyses confirm and extend our initial understandings about the nature of clustering in the network visualization data.¹³ As Table 2 shows, law school degree programs and section assignments play a substantial role in organizing social ties among students, but there are some organizational differences. Social networks also appear to be organized along the lines of race, international status, and religion at Schools 1 and 2 but not at School 3. Homophily based on age appears to be present at all three schools, and graduating from a top twenty-five college is salient at School 1. In contrast, the assortativity coefficients for gender, sexual identity, lawyers in the family, and political orientation are mostly close to zero, suggesting that homophily based on these characteristics is less important at the three law schools.

Table 2. Assortativity coefficients by school

Critically, law school structures, which divide students into programs and JD students into sections that stay together for all or substantially all academic work throughout the first semester, explain the relational dynamics apparent in the network visualizations. At the same time, policies that distinguish on the basis of degree program and result in differences in classes for LLMs also explain the separateness of LLMs and JDs at Schools 1 and 2. These structural factors are joined at two of the schools by substantial homophily based on race and national origin, and these forces interact in important ways: the organizational structures in place in JD programs may also be acting as a critical intervention that disrupts or decreases demographically based homophily by encouraging ties that cut across demographics. While these results suggest the centrality of particular organizing forces, we now turn to ERGMs to examine the underlying factors driving these observed patterns. ERGMs allow us to control for multiple homophily variables simultaneously as well as to estimate the effects of triadic closure (that is, the tendency of a friend of a friend to become a friend), which promotes the formation of homogeneous clusters.

Determinants of Student Ties

Table 3 reports results from the ERGMs, which examined whether triadic closure and homophily based on demographic identities and organizational structure predicted the presence of ties between students. The estimated coefficients here represent log-odds, so exponentiating them obtains odds ratios (ORs) for the presence versus absence of ties. The ERGM results confirm the importance of organizational affiliations and demographic homophily for shaping the social networks of law students as they appear in the visualizations and assortativity analyses. Across all three schools, degree program and section assignments were statistically significant and positive. We observed statistically significant homophily effects for all JD sections at the three schools. Within-section ties were significantly more likely to occur, with odds ratios ranging from 2.9 (e ^1.08) to 6.4 (e ^1.86) times more likely. It is noteworthy that these odds ratios appear to be larger in School 3 than for the other schools, consistent with the differences noted above in the discussion of network visualizations. Similarly, ties between LLM students were 2.1 (e ^.74) and 7.7 times (e ^2.04.) more likely than cross-section/program ties at Schools 1 and 2, respectively. This pattern suggests that School 3’s network is strongly organized by JD section assignments.¹⁴

Table 3. ERGMs predicting ties by law school

Notes:

*** p < 0.001; ** p < 0.01; * p < 0.05

Note: Not shown are controls for edges, mutuality, being partnered, children, employment status, and the number of club affiliations (sd in parentheses).

These findings also align with comments of our interviewees. For example, Veronica, a 1L JD at School 3, explained the significance of sections at her school:

Veronica: One thing that’s really difficult—and I know a lot of us have been talking about—we’re split into cohorts. Sections. I forget that other sections even exist, because they’re never around when we’re around.

Q: The timings are different?

Veronica: Yes. Even though we have a class of 400, I feel like there’s only 104 of us—because that is how many people I’m in a section with. It has been a lot easier; we have two classes out of our five that are much smaller groups. So, it’s been nice to actually meet friends in the smaller groups.

Similarly, it was common for School 2’s interviewees to describe making friends through classes: Teresa Herrera, a 1L at School 2, explained that

there are a lot of people in my section who are just great. Like, really fun-loving people and really helpful and understanding. I made a few friends. We just got to know each other through classes. I made one friend on the first day, and then she made another friend and then we started hanging out. Then she had another friend. So, it’s just kind of like through classes, through hanging out, through meeting someone to eat lunch with.

In short, organizational affiliations play a central role in structuring students’ connections with one another—a commonality likely reflected at most law schools but where room for difference also exists among schools. Since JD sections bring together students with a diverse range of demographic characteristics, they appear to be critical for disrupting the influence of demographic homophily.

In terms of racial homophily measures, all coefficients were statistically significant at Schools 1 and 2. For example, White/other (e. ²⁷ = 1.3 OR), Asian (e ^.49 = 1.6 OR), Black (e ^1.43 = 4.2 OR), and Latinx (e ^.94 = 2.6 OR) students at School 1 were all more likely to form same-race ties compared to interracial relationships. Terrell, a multiracial (Black and Latinx) JD student at School 1 was illustrative of this pattern: he commented that his close friendship group revolved around four other students: he was in an affinity group with one, in another affinity group with another, and described the other two as being “closer in age” to him (all being older than the average student at School 1). In School 3, only the coefficient for White students was statistically significant, but the lack of significance for Black and Latinx students, as we discussed previously, may be attributable not only to weaker racial homophily but also to the small proportions of students of color in our sample. In addition to the presence of racial homophily in Schools 1 and 2, Table 3 also shows that, in School 1, both international and domestic students were likely to have homophilous ties, but only the latter was significant at School 2.

Moreover, the interview data suggest that racial homophily was supported by organizations and events in law school. Students of color from School 1, for example, revealed in their first-wave interviews that many of their close networks were comprised of people of the same race they met at orientation. Susan Yang, for example, a JD at School 1, described having met a couple of people who she saw as friends “[a]t an event before orientation,” but she described the process of creating relationships as slower in law school, compared to college, “because we’re in different sections or something. Not so easy. Maybe because I’ll see them once or twice a week.” Another JD at School 1, Jennifer Chen, also met same-race friends at a preorientation event that another incoming 1L had organized: “The friends that I made there … and then we kind of created a group. And then we added on more people as school went on. It was nice, because they were people from different sections. Otherwise, I would never see them.” Affinity groups at School 1 also function as a bridge across sections or class cohorts. Myra Khan, another School 1 JD student, described meeting 2Ls and 3Ls through

student groups. I signed up for a lot of different mentorship programs … every affinity group has a mentorship program…. I have also realized that the best networking experiences I’ve had have been at events that are tailored to people or groups that I affiliate with. So, like, I went to a diversity one. I went to an OutLaw one. I went to a Muslim one this past weekend. Those smaller groups, where it feels like there is a real effort to try to build community—and it feels a lot more genuine.

We previously reported that assortativity coefficients were low for most of the remaining covariates, but the ERGM results suggest that homophily was still relevant for other demographic characteristics. All gender coefficients were statistically significant across the three schools. For example, men and women in School 1 were, respectively, 36 (e ^.31) and 23 (e ^.21) percent more likely to have homophilous ties compared to cross-gender ties. We also observe that LGBTQ students were 1.6 (e ^.47) and 1.4 times (e ^.36) more likely to have homophilous ties in Schools 1 and 3, whereas straight students were 1.2 times (e ^.15) more likely to have homophilous ties in School 1. Carla, a JD student at School 2, described being LGBTQ as anchoring her relationships both in and outside of law school: “It’s a lot easier to make friends that way, because we kind of all huddle together. Because that’s sort of … not an artificial bond, but it’s a leg up. And it’s a very close community. The community shares a lot of trauma. A lot of mutual experiences. That helps a lot, in terms of making friends. … But again, most of my social circle is very much very gay in general.”

In School 1, students who have lawyers in the family and those graduating from highly ranked colleges were, respectively, 1.3 (e ^.23) and 1.2 (e ^.16) times more likely to have homophilous ties, but those students without lawyers in the family or who graduated from lower ranked undergraduate schools were not. This pattern in School 1 indicates that there is some social closure among students who seem to have more cultural capital, but it is not obvious why similar patterns do not appear in the analyses for Schools 2 and 3, but perhaps the more elite the school, the more such markers of status matter. Similarly, network coordinates were mapped by political beliefs and religion. Particularly, political differences were associated with lower odds of ties in Schools 1 and 2, which suggests that the converse—political homophily—was, perhaps unsurprisingly, associated with the presence of ties. Ties between Christian law students were more likely than interreligious ties at Schools 1 and 2, and Jewish students were more likely to be connected to one another across all three schools. Finally, ties were more likely to occur between students who were close in age across all three schools.

In Table 3, we also analyzed triadic closure, which captures the tendency for group formation, such as when student A and student B are both friends with student C, A and B are likely to also become friends with each other, as well as popularity. All three schools (b = 1.27, 1.38, and 1.18) show statistically significant coefficients for triadic closure. In the context of extensive homophily, triadic closure will tend to amplify homophily, such as those shown previously by section affiliations and race in the network visualizations. In contrast, the popularity coefficient was only statistically significant in School 1 (b = –0.44, p < 0.05). In short, the tendencies for racial homophily and triadic closure are producing racial segregation in law school networks.

Racial Differences in Social Capital

Our analysis, so far, shows that racial homophily and triadic closure were key network mechanisms giving rise to the law school networks that we observed in Figures 1–3, but this raises questions about whether there were racial disparities in access to bonding (homophily and triadic closure) and bridging social capital (popularity). For each respondent, we calculated the level of racial homophily, the level of triadic closure, and their popularity based on their immediate network (that is, the students to whom respondents were connected). Figure 4 shows boxplots of our social capital measures by race. As shown in the left-most boxplot, Black and Latinx students had the lowest median levels of racial homophily, as indicated by the bold line, in their immediate networks. Black students also had lowest median level of triadic closure and popularity, whereas Latinx students had the highest triadic closure and popularity. These findings suggest that Black students, who are under-represented in law school, tend to be embedded in social networks with both lower bonding and bridging social capital.

Figure 4. Social capital measures by race

This finding is not unlike other research that reinforces the role of internal bonding for minority students in higher education. The popularity of Latinx students is consistent with prior research that has found they tend to be friends with both Black and White students (Moody 2001). In contrast, the incidental racialization that Pan (2017) finds among law students and their student groups might help shed some light on the bonding that gave Black students seemingly different rewards from their peers. Across the schools studied, Black students were initially part of a strong organizing student group, which was not the same for other minority student groups.

Student Satisfaction

Next, we explored how race factors into the relational dynamics of law school and their association with satisfaction. We examined the extent to which racial differences in bonding social capital were associated with students’ satisfaction with their law school experiences. Table 4 presents regressions of students’ satisfaction with law school on demographic, organizational, network, control measures, and binary indicators for schools, which allowed for a pooled analysis. We included an interaction between respondents’ race and the racial homophily of their personal networks to whether associations between respondents’ race and satisfaction differed by the racial homophily of their networks. This table only displays statistically significant coefficients from regressions of the full sample and a subsample consisting of JD students only.

Table 4. Regression of law school satisfaction

Notes:

*** p < 0.001; ** p < 0.01; * p < 0.05.

The reference category for race in this table is Asian. This table displays statistically significant coefficients. Not shown are controls for LLM program, international status, gender, lawyers in the family, graduating from a top twenty-five college, religion, age, triadic closure, being partnered, employment status, and the number of club affiliations (sd in parentheses).

For the full sample, the main effects for all race categories and interactions for Black and White students were statistically significant, showing that associations between race and law school satisfaction depended on the extent to which respondents were embedded in racially homophilous networks. Popularity was positively associated with law school satisfaction (b = 0.02, p < 0.05), but we observed no effect for triadic closure. Importantly, the model for JD students only showed a similar pattern for race, racial homophily, and popularity with only the main effects for Latinx and racial homophily becoming nonsignificant. This suggests that our results were not driven by the inclusion of LLM students. Taken together, these results support the notion that bonding and bridging social capital matters for students’ satisfaction with law school and that the former, as measured by racial homophily, depended on respondents’ race.

In addition, several demographic and control variables were statistically significant in the full and JD-only models. Holding more liberal political views was associated with higher law school satisfaction in both the full and JD-only models. Separately, comparisons across schools showed that School 3 had slightly lower satisfaction only in the full model (b = –0.22, p < 0.05). Importantly, LGBTQ students in the JD only model (b = –.19, p < 0.05) had lower satisfaction. None of the other variables, including other demographic characteristics, program and section assignments, and control measures were statistically significant in either model.

To facilitate interpretation of the interactions in Table 4, Figure 5 shows how predicted levels of law school satisfaction in relation to the level of racial homophily in respondents’ immediate network depended on race for both the full sample and the JD subsample. The top figure shows that racially homophilous networks are associated with lower law school satisfaction for Asian and Latinx students. In contrast, law school satisfaction increases for White and Black students when they are embedded in racially homophilous networks. The pattern for Black students suggests that embeddedness in racially homophilous networks, especially against the backdrop of the larger populations of majority White students,¹⁵ facilitates a critical mass for bonding and social support. For White students, who comprise the majority of JD students and of our respondents, the results are discouraging but familiar: segregated social worlds dominated by White students are associated with increased law school satisfaction. The bottom figure, which displays predicted levels of satisfaction for JD students only, shows an almost identical pattern.

Figure 5. Predicted law school satisfaction by racial homogeneity and race.

Our interviews offer context to understanding these racially homophilous networks within the student body. Scott Lee, a 1L at School 1, explained that his friendship group is predominantly Asian American, like him: “Most of my friends—most of the people I talk to the most—I think just happen to be Asian American. I guess … it wasn’t like a conscious choice or anything. I did join … the Asian student association. But we haven’t had that many events yet. I think it was more just we got along. We just ended up sitting next to each other.” In responding to a question about a sense of fit in the law school, Scott referred to the same community: “[E]ven before I started—I felt like the Asian student organization was pretty proactive. I just reached out. They sounded welcoming, back in April,” suggesting that Scott, like Carla’s “very gay” networks, viewed access to same-race friends as contributing to at least some part of his satisfaction with law school. While racial homophily in the social networks of Asian students appear to play an important role in providing social support, the resulting segregation of students of color from the majority White population also appears to detract from their law school experiences. This may suggest that increased bridging social capital in the networks of Asian and Latinx student could promote greater law school satisfaction.

At the same time, comments of Asian students in their qualitative interviews stressed competition as both a negative in the law school environment and something they tried to avoid. Jennifer Chen, for example, who commented earlier about same-race friends, explained that “law school is such a competitive environment, having ambition to be, like, at the forefront of what you present to other people can make other people feel uncomfortable or diminished.” Eman Hussain described it as “people are a lot more anxious than I thought. I have met a lot of really nice people, but, at the same time, I do try to disconnect a little and go home, and not do so many things with the other 1Ls. Just because it becomes really anxiety-inducing.” Myra Khan described law school as “very stressful, Very insular…. It’s competitive.” Susan Yang focused on grades as related to competition:

[T]here are certain moments where I am aware of it [referring to competition]. Like, when people make jokes, Or I made cookies yesterday and so I brought them to class. Someone stared at me; did you drug these? Because we have a practice exam on Wednesdays. Who would do that? But the fact that people think about that takes me—I guess because we’re in law school. So, there are those moments where I’m like—oh, yeah; there are people here who are super-competitive. And also, people care so much about grades here. So, so much. Which is understandable, but not something I was used to, I think.

While we cannot be certain that this sort of emphasis in perception explains the relationship of homophily and satisfaction for Asian students, it suggests an area for further inquiry in our work.

Latinx students had an analogous dynamic to Asian students in the survey data analysis of homophily and satisfaction. In interviews, two themes emerged that may relate. First, Latinx LLM students described a division with the LLM community between Asian students and other students. Their networks were not so much focused on other Latinx students as distinct from Asian student networks—not from animosity but, rather, from what they perceived as differences in demeanor and comfort with English, among other things. Second, very little about affinity groups emerged from the Latinx JD interviews. Overall, these data also point toward the need for additional research.

In contrast, for Black students in our interview data, student groups were almost part of the backdrop of law school. One student, Marquis, a 1L at School 3, almost did not think to include these networks when we asked him about how he made many of his friends: “I forgot, that [student groups] is probably a big component. I do meet a lot of students through student groups. There is the Black Law Student Association here.” Later in the same conversation, before admitting that most of his close friends were either who he sat next to in class, or people from race-based student group associations who were in his friend group, he shared: “[T]here are some pretty big student organizations here. That is how I meet a lot of students.” Another School 3 student, Danielle, explained her dual strategy for networking as involving both the law student group focused on criminal law—because she hoped to work in that field—and the Black Law Student Association—in order to get to know “people who are similar to me, as far as … economic; you know, how we grew up, where they are from.” For Black students, then, alongside chance and physical proximity of the first-year classroom, student affinity groups were significant to building their social networks.

At first glance, it might seem like Scott Lee, and Marquis have similar kinds of homophily within their friend groups. For Scott, the predominantly Asian American student group was mostly accidental, something that was not a “conscious choice or anything” but, instead, classmates who “just ended up sitting next to each other.” Although it is possible that the students that “sounded welcoming” before he joined law school were part of why he was satisfied as a 1L, its resonance given our larger network data about variations in satisfaction offers us new ways to think about Scott’s disconnectedness from networks dominated by White students. It is possible that Scott stumbled into these sticky networks of fellow in-group students and that this passive entry into these networks limited the chances he had for making more varied ties and, in turn, his satisfaction.

Discussion

Our initial motivation for this project was to observe what law students’ networks looked like and the bridging and bonding social capital coursing through them. Particularly, we were interested in how diversity of experiences, differential access, and resulting social capital inequalities might affect the value of legal education for different categories of students. We expected that students’ tendencies to embed themselves in homophilous networks, thereby forming bonding social capital grounded in shared identities, might help us understand their law school experiences and socialization. In turn, we imagined that these patterns would tell us something about the way in which diverse groups experience the high-status environments that socialize them into the profession.

Following these preliminary intuitions, our data allowed us to observe law school networks and to examine their implications for producing disparities in student experience. The finding that homophily is relevant for determining network ties confirms prior research on the legal profession, which suggested that law schools might offer structural incentives for marginalized students to bond with one another (Pan 2017), even if such bonding and the resulting segregation does not always result in positive outcomes (Quintanilla and Erman 2020). At the same time, our data suggest that there are diverse ways in which students experience this disconnectedness that might result from in-group networking. For Black students, the local community of like peers acted as a buffer against the predominant “White space” of law school, but, as shown in our analysis, many do not have access to this bonding social capital (Capers 2021). Although homophilous networks were tied to satisfaction with law school for Black students, the same embeddedness was associated with lower satisfaction for Asian and Latinx students. This finding of diverse disconnectedness—that Black students benefited by being disconnected from White-dominated parts of the network, while others were negatively affected by similar isolation—is a crucial extension to our understandings of law student experience because it clarifies that, although homophily might impact a range of student groups, not all groups suffer equally.

These mixed-methods data allows us to observe experiences in legal education that might have otherwise been missed had we only used individual survey or interview data to ask students about their networks and satisfaction. Striking in our interview data was how students across social categories spoke about friends and groups that shared their social characteristics, a finding that scholars like Pan (2017) have found in their research on the importance of student communities in law student contexts. But, informed by our network findings, it was clear that not all students came to be embedded in these networks in the same way. These results highlight how cultural capital, linked to demographic identities and markers of class, influences social capital or how students fit into law school in terms of their social networks.

If we were only tracing student narratives, the accounts between Marquis and Scott might not be all that different. In fact, Marquis’s admission that there were some “pretty big student groups” that were a “big component” in determining “how he met a lot of students” sounds like it might be in line with Scott’s remarks. But reading it alongside our network data more generally about patterns in Black student networks and satisfaction, the particularity of these accounts becomes even more relevant. Even though it might have been his main network, it seemed to be so much a part of his background that Marquis did not even feel the need to expressly mention it until prompted, whereas Scott’s lack of “conscious choice” in making his friend groups perhaps tracks a different kind of homophily experience. It is this patterned divergence even within seemingly similar structures of in-group homophily that we term diverse disconnectedness. Without the network data, these interviews might have not offered any substantive distinctions, but the mixed methods offer us a more layered perspective about diversity in relationships, distance from law school structures, and student satisfaction. Reading individual accounts alongside the larger patterns in homophily offer us new strains of connection and isolation to pay attention to.

This finding of diverse disconnectedness also helps inform our broader understandings of the ways in which network bonding (Coleman 1988) and bridging (Burt 2005) intersect with homophily based on demographic student categories. First, while racial homophily in the social networks of Asian and Latinx students appear to play an important role in providing social support, the resulting segregation of students of color from the majority White population appears to detract from their law school experience. These findings could reflect dissatisfaction among Asian and Latinx students with feeling segregated, presumably because they seek to—and are capable of—being more adjacent to a majority or normative identity or reflect an alternative explanation such as the presence of fewer resources in their homophilous networks. Regardless, increased bridging social capital in the networks of Asian and Latinx student could promote greater law school satisfaction. Second, our data indicate that White students, who constitute much of the core of student networks, benefit personally from cultural capital, creating homogenous, closed networks of White students, but this “categorical closure” produces the collectively unsatisfactory outcome of racial segregation, which is contrary to the inclusion of diversity. All these diverse and distinct ways in which the experience of marginalized students is disconnected from structures in law school is important because it offers a framework of thinking about diversity in more concrete terms rather than as an ambiguous problem that needs to be fixed with performative posturing.

Diverse disconnectedness also captures the diversity in the kinds of disconnectedness that students feel across different institutional contexts and the ways in which organizational structures are determinative in structuring student networks and experiences. In addition to homophily across demographic identities where we would expect such bonding (for example, race, international status, sexual orientation, religion, and age), these patterns also vary across different schools and within programs and sections in each school. For example, in School 1, racial homophily was particularly important; one reason for this might have been the school’s approach to structuring student orientation programs that allow for network formation around affinity and identity groups. In contrast, race-clustered networks in School 2, which was described in interviews as having a slightly less elaborate student-group structure, were less obvious. The networks around sections in School 3, which was described in interviews as having a significant commuter-student influence, were most revealing of the ways in which organizational structures and cultures produce student networks. School 3 interviewees emphasized their living arrangements and locations—whether or not they were close to the school—as factors shaping their involvement in school activities and opportunities to interact with peers, as described above by Kyla.

Similarly, across schools, the distinctions between respondents in JD and LLM programs were stark. As we have shown in earlier work (Ballakrishnen 2011; Silver 2011; Silver and Ballakrishnen 2018; Ballakrishnen and Silver 2019), these distinctions between groups were to be expected, revealing socialization tracks, which LLM students are often eager to cross to have a more “authentic” law school experience. Still, the racial homophily in networks, alongside our interviews, confirm another trend from our earlier work, which is that, despite “crossing tracks,” student networks are likely to differentiate on international and domestic student identities. Beyond homophily, variation in satisfaction speak to the interaction of individual and organizational factors in these data. School 3, which had lower student satisfaction even after controlling for demographic factors and social networks, highlights the importance of organizational factors. These results are consistent with the larger network literature that suggests that local in-group networks may act as oases and buffers for marginalized students, fostering community. However, these closed ties also limit their capacities for more diverse law school experiences, which our data show are associated with higher satisfaction for some groups.

In addition to organizational structures of the programs and the resources within the schools themselves, student networks were influenced by the ways in which they were assigned to their classrooms. Our findings suggest that program discrepancies and assignments into different law school sections may be critical for disrupting the tendency of ties to occur among individuals with shared demographic characteristics, such as race, gender, sexualities, class or status, political views, religion, and age. Just as in Scott Lee’s account above, many of our respondents mentioned finding their first friends in law school because they were sitting next to them in a first-year class. It is worth noting that the law school with the highest proportion of White respondents in the response pool to our sample—School 3—has a law school network that is most strongly reflective of first-year section assignments,¹⁶ whereas section assignment in the school with the most racially diverse respondent pool in our data—School 1—appears to be somewhat less important compared to race and national origin. This is consistent with the notion that students are motivated to form same-race relationships, even if it involves forming same-race relationships with students who are from a different JD section.

Our data reveal nuanced differences in the ways in which marginalized students experience similar environments and the kinds of fixes that might serve their interests but which cannot be solved or theorized within broader categories of blanket diversity and inclusion. While these analyses do not allow us to definitively confirm whether these are associated with the strength of student affinity groups, placement of student sections, or other factors, they nevertheless offer a new way to think about administrative incentives to allocate resources for student network integration and diversification where possible. These findings also offer insight into rethinking our institutional commitments to inclusive student environments. If homophily is likely to help different students differently, and if the operationalization of diverse disconnectedness is likely to be school specific, perhaps having more opportunities for different kinds of interactions in the first year might be important while trying to gauge how individual schools can best support students. Key to this unpacking is having better systematic checks on student experiences and distinctions rather than assuming that the same fixes might work, much less similarly, for all students. While desegregating network opportunities might serve Asian and non-heterosexual students, for example, Black students might need institutional support to reinvest more soundly into their relatively closed networks to foster their satisfaction.

It is important to note several limitations. First, this analysis was cross-sectional and focused on racial homophily and student satisfaction at the very beginning of their law school careers. We have yet to entirely account for the impact of first-semester grades, the churn of students’ networks over time, and the role of student groups in sustaining these relationships. Second and, here, our focus is on racial homophily, future research could examine whether alternative network structures are also important for student outcomes. Third, this article focuses only on law school satisfaction, which is central for assessing students’ perceptions about legal education, but there are likely correlated outcomes, such as students’ perceived belonging, that are deserving of attention. We address both issues in forthcoming research. Lastly, although a strength of our article is a comparison of three very different schools, as opposed to being based on a single school, we note that our findings may not be generalizable to all law schools in the United States.

Overall, our analysis suggests that network resources may be accessed by different kinds of marginalized actors in different ways and that particular networks might also be protecting and problematizing different students differently. These connections cement inequalities that are already present and persistent, and homophilous bonding and diverse bridging do not serve everyone equally. Bonds can make it harder to bridge to more resource-laden environments, while bridging itself can bring different advantages depending on the group in question. Our research also suggests that thinking about diversity as a one-size-fits-all model is not helpful. Rather, what is needed is a model that itself is diverse and allows consideration of intra- and inter-group difference. These data also inform our understanding of capital navigation and negotiation within elite networks and reminds law schools that they must conceive of equality as an ongoing, dynamic project instead of a one-shot deal at entry injection, given the importance of law school socialization and all that follows from it.

Conclusion

Our core findings center on the importance of racial homophily in law school social networks and how these network structures differentially impact Black, Asian, Latinx, and White students’ satisfaction with their law schools. One implication of this study is the notion that students, particularly those who are demographically in the minority, are seeking to “find their people”—friends and confidantes that provide support and advice as students navigate their education. It is quite possible that segregation in social networks is an unintended emergent outcome. Students presumably are seeking to connect with diverse others as opposed to being siloed (Ballakrishnen and Silver 2019). Indeed, the students who appear to maintain larger social networks appear to be more satisfied. In short, the bonding social capital that we observed in law school, while fulfilling immediate needs for social support, may cascade to form larger clustered communities that ultimately detract from the law school experience and student satisfaction.

Finding the balance between having social networks that promote social support and those that allow students to connect across different categories of diversity is no doubt difficult. Law school is hard, and it is no surprise that students seek comfort and support from communities that sync in terms of identity and experience. But seeking comfort may also yield isolation from the general gestalt of law school. It is in this tension between a supportive and homologous environment, on the one hand, and a diverse and complex environment, on the other, that we observe diverse disconnectedness and inequality. Different students balance this tension differently, reflecting both their own demographics as well as the structure of their law school, and this melds into continuing patterns of hierarchical sorting that continue beyond graduation and into the profession.

Our analyses of the interactive and layered relationships that comprise law school networks can offer schools a way to think about the kinds of institutional responses that could generate resource-rich ties and environments for students who most need them but are possibly blocked from accessing them. Law schools have at least one mechanism that can generate ties across demographic categories, which is the role of sections; topically focused student groups and co-curricular activities also may mitigate the pull of demographics as students proceed in law school. At the same time, our analysis points to new ways for schools to think about student dynamics and inclusion as a living process, which they have a hand in designing to enable more equitable experiences for their students.