Professional social connections are associated with student science identity in a research-based field biology course

Abstract

Undergraduate field courses provide students with valuable opportunities to learn in and from the natural environment. Prior research shows that field courses can improve students' cognitive and affective outcomes and encourage them to engage in social interaction with their classmates and professors. Despite frequent documentation of the ability for field courses to promote social relationships, no prior studies have characterized student interaction in this context or its influence on student outcomes. To better understand student interactions, we used social network analysis to characterize how and with whom students formed communities in a semester-long field course with an integrated research project. We found that students form small, tightly connected communities composed of professional and personal contacts with whom they maintain contact throughout the semester. We asked students what they discussed when talking about their research project with their professional and personal contacts. We found that students explained their research to members of their professional and personal communities. Discussions with professional communities largely focused on guidance related to the technical aspects of their research projects and personal communities focused on personal support. We also examined student research project performance and science identity. Bootstrapped multivariate linear modeling indicated that neither student demographic factors nor the degree to which a student discussed their research with their personal or professional communities influenced their research project grades. However, the strength of the professional but not personal connections was associated with science identity. Based on these findings, we offer evidence-based recommendations to field course instructors on how best to structure student interaction.

INTRODUCTION

Field courses, where students leave the classroom to learn outdoors through observing and describing natural phenomena (Mogk & Goodwin, 2012), are a staple in undergraduate life science education (Shinbrot et al., 2022). Participation in field courses is associated with a wide array of positive student outcomes that promote retention in ecology and evolutionary biology (EEB) undergraduate degree programs (Beltran et al., 2020; Mogk & Goodwin, 2012; O'Connell et al., 2021, 2022; Shinbrot et al., 2022; Shortlidge et al., 2021). In the cognitive domain, previous research indicates that students who participate in these courses exhibit greater academic achievement than their nonparticipating peers or those who engage in equivalent, laboratory-based ecology experiences (Easton & Gilburn, 2012; Scott et al., 2012). Further, participation in field courses can promote students' knowledge growth, spatial reasoning, and familiarity with the scientific process (Mason et al., 2018; Riggs et al., 2009). They also can support students' affective development (O'Connell et al., 2021, 2022; Shinbrot et al., 2022; Treibergs et al., 2022). For example, field courses can promote developments in students' motivation (Scott et al., 2019) and science self-efficacy (Beltran et al., 2020), and can further increase students' confidence to engage in fieldwork (Peacock et al., 2018). These courses also help students develop skills outlined in the Four-Dimensional Ecology Education framework, such as emphasizing the significance of human–environment interactions, collaboration, and broader disciplinary practices (Klemow et al., 2019).

In addition to bolstering student outcomes, field courses can promote meaningful social interactions between students (Boyle et al., 2007; Cotton, 2009). Students engage in formal and informal social interactions with their instructors and classmates and can form lasting relationships that extend beyond the course (Hart et al., 2011; Mason et al., 2018). Moreover, shared peer experiences during field courses can promote a sense of community and help students become integrated within a community of peers in the discipline (Fedesco et al., 2020; Murphy, 2001; Race et al., 2021; Streule & Craig, 2016).

While studies have documented the connective power of field courses, to our knowledge, no prior studies have formally investigated how students interact in field courses and whether these interactions relate to outcomes. Guided by education theory, we used a suite of statistical techniques to examine social interaction in an undergraduate field biology course and how they relate to student science identity and performance.

Theoretical framework: Theory of Undergraduate Socialization

This research is situated within the Theory of Undergraduate Socialization (Weidman, 2006), which considers a student's background as they enter college learning contexts (inputs), the mechanisms of inter- and intrapersonal interaction that students engage in within formal and informal settings (environment), and changes in students' academic performance and affect (outcomes) (Figure 1). This theory is particularly fitting for research on field-based coursework, as field experiences are important for students' socialization into field-intensive disciplines (Nunez et al., 2021; Streule & Craig, 2016).

FIGURE 1. Theory of Undergraduate Socialization, as adapted for studying students' interactions in a campus-based undergraduate field course. Inputs refer to students' backgrounds, including their beliefs, predispositions, demographic background, and values that they enter normative contexts with, which may then influence or be influenced by the interactions they have with their professional and personal communities in the environment. People of Color (PoC) identity includes racial and/or ethnic identities long marginalized and excluded from EEB disciplines (e.g., Latine, Asian, Black identities) (Bowser et al., 2012; U.S. Department of Education, 2019). Environment refers to the institutional and course culture, course structure, and other organizational factors wherein students interact and includes normative contexts. Outcomes involve the changes students may experience through the socialization process, which are theoretically influenced by both the inputs that students entered the environment with, as well as the interactions students engage in within the environment itself (Weidman, 2006).

The consideration of inputs in the Theory of Undergraduate Socialization—such as students' backgrounds—is salient to research on students' interactions and outcomes in EEB undergraduate degree programs (Figure 1). For example, research on these programs shows that students who identify as People of Color (PoC)—who have been historically minoritized in STEM on the basis of their racial and ethnic identity—can feel a sense of exclusion among their predominantly White peers (O'Brien et al., 2020). In addition, race, ethnicity, gender identity, and financial income can impact the recreational outdoor experience of students and their interest in doing fieldwork (Larson et al., 2011; Lee et al., 2001; Morris et al., 2020). Given that students who have less experience with the outdoors may feel less comfort, belonging, and aptitude during undergraduate field courses (O'Brien et al., 2020; Zavaleta et al., 2020), it is imperative to consider the influence of student inputs on their outcomes.

The center of the model, which focuses on the environment, considers normative contexts or settings where students are exposed to teaching, experiences, ideas, perspectives, and values (Figure 1; Weidman, 1989). The model posits that college normative contexts (e.g., classrooms, dorms) promote certain norms, values, and beliefs that are either explicitly written or spoken or implicit procedures that constitute what is often referred to as “the hidden curriculum” (Snyder, 1971). Students can come to understand these aspects of normative contexts through interactions with their professional (e.g., practitioners, classmates) and personal communities (e.g., family, friends) to achieve greater social integration in the community (Weidman, 1989). Field biology courses are a normative context wherein students learn about fieldwork practices, norms, culture, and language through social interactions with their professional and personal communities (Mogk & Goodwin, 2012). Through these interactions, students may experience shifts in their identification with field biology and knowledge in the domain (Streule & Craig, 2016).

Normative contexts can facilitate formal and informal student interactions with members of their professional communities, such as classmates and educators, or members of their personal communities, such as family and friends (Weidman, 2006; Weidman et al., 2014). In campus-based field courses, even in short-field activities without intentional structuring for social interaction, students communicate and build rapport with their classmates, instructors, and course staff (Peacock et al., 2018). In addition, field courses often ask students to engage in inquiry-based or discovery-based research activities to strengthen their knowledge of the scientific process (e.g., Flaherty et al., 2017; Laungani et al., 2018). These activities are designed to promote student collaboration with their peers and instructors (Corwin et al., 2014) and discussion with others outside of the classroom (Hanauer & Hatfull, 2015). Despite these opportunities promoting student interaction, little is known about the degree, frequency, or topics of these interactions. As such, this study aims to clarify the socialization processes of students' interactions centered around their course-based field research projects (hereafter referred to as “research projects”).

The end goal of the socialization process is to augment student outcomes as they become novice professional practitioners, accruing the knowledge, skills, and dispositions held by disciplinary professionals (Figure 1). Researchers and educators have established that field courses can develop these outcomes in students as they promote an engaging and social environment where students gain proficiency and become acculturated to the discipline (Mogk & Goodwin, 2012; Streule & Craig, 2016). While the Theory of Undergraduate Socialization considers many types of outcomes (e.g., knowledge, skills, dispositions, identity), this study focuses on two specific outcomes: academic performance and science identity.

The first outcome, academic performance, is an important predictor of retention (Redmond-Sanogo et al., 2016). Previous work shows that participation in undergraduate field courses can enhance students' academic performance and retention (Beltran et al., 2020). The second outcome is student science identity. A person's identity is a context-dependent social construct that regards who an individual views themselves to be (Brown, 2004; Gee, 2000). A developed science identity can promote student persistence in college and graduate degree programs and increase intent to pursue a career in STEM (Chemers et al., 2011; Kuchynka et al., 2019). Science identity involves the interplay of four domains or subconstructs: (1) competence or beliefs in the ability to understand science; (2) performance or the ability to successfully carry out science tasks; (3) interest or the inclination to learn science; and (4) recognition or the feeling that others recognize oneself as a science person (Carlone & Johnson, 2007; Hazari et al., 2010).

As identity is socially constructed, an individual's interactions with others can influence students' identity development in various ways (Risman, 2004). For instance, students' social interactions with others, such as receiving recognition from faculty and STEM-majoring peers or family, are essential for science identity production (Dou et al., 2019; Rodriguez et al., 2019; Thompson & Jensen-Ryan, 2018). Although prior studies describe the ability of field courses to bolster students' development of science identity (Beltran et al., 2020), how social interactions within a field course can shape or influence student science identity remains an open question.

Research questions

This work describes how the interactions students have with members of their professional and personal communities centered on their course research relate to their research project performance and science identity. We employed student surveys, social network analysis (SNA), descriptive statistics, and bootstrapped multivariate linear modeling to answer the following research questions:

1. With whom do students form communities when discussing their course-based field research project and how do students' interactions with their community members change over time?
2. What topics do students discuss with professional and personal contacts and do these topics differ depending on the type of community?
3. How does student engagement with professional and personal contacts relate to their science identity and research project performance?

MATERIALS AND METHODS Contextual information Institutional and course context

During the Fall 2020 semester, we collected data in an introductory field biology course at a research-intensive institution located in the Northeastern United States. Introductory Field Biology is an entry-level course that aims to provide students with a more developed understanding of the natural world by facilitating student engagement in field-based exercises and field research. This 15-week course includes a weekly 1-h lecture session and semiweekly 3-h field laboratory sessions. Specific learning outcomes of the course included: (1) an ability to recognize ecosystem types; (2) species identification knowledge of the flora and fauna of the northeast; (3) an understanding of field research methods in a variety of ecological disciplines; (4) an ability to collaborate with peers; and (5) an understanding of field research.

To accomplish the learning outcomes, students completed the research project. Within the first two weeks of the semester, students were asked about their scientific interests (e.g., ornithology, stream ecology) through a brief survey. Subsequently, students were grouped into research pairs based on their scientific interests, designed experimental or observational protocols, submitted a research proposal, and were given critical feedback from teaching staff within the first three weeks of the semester. Between weeks 4 and 12, students collected data through fieldwork to address their research questions. Depending on their research questions, student teams conducted their research at one or multiple field sites. These field sites were located either on campus or in the wooded areas and streams of the northern temperate broadleaf forests that surround the campus. Students regularly received feedback from instructors through debriefing sessions after returning from the field. At the end of the semester, students analyzed their data and presented their findings through a research report and virtual presentation delivered to the entire class. Notably, students were required to return to their permanent residences in late November due to the COVID-19 pandemic.

Participant recruitment

The participants in this study represent a convenience sample of undergraduate students (n = 36) enrolled in two sections of an undergraduate campus-based field course titled “Introductory Field Biology” (Table 1). All students were majors in Environment and Sustainability.

TABLE 1 Self-reported demographic characteristics of students enrolled in “Introductory Field Biology” (n = 36^a).

Category	Whole course (%)
Class standing
First year	2.94
Sophomore	52.94
Junior	38.24
Senior	5.88
Race/ethnicity
People of Color	29.41
White students	70.59
Gender identity
Woman	79.41
Man	20.59
College generational status
First generation	32.35
Continuing generation	67.65
Previous research experience
No prior research experience	88.24
Field-based research experience	2.94
Laboratory-based research experience	8.82

^aOne student did not fill out the course demographic survey (Appendix S1) and was omitted from frequency statistics.

Survey instruments and measures

We employed a quasi-experimental research approach to understand student interactions centered around their research project and how those interactions influenced students' academic performance and affective outcomes. To collect information on these processes, we implemented multiple surveys. All students enrolled in the Introductory Field Biology course consented to take part in the research, which has been approved by the Cornell University Institutional Review Board under the exempt protocol no. 2001009364.

Name generator network survey

Students completed a name generator survey (Appendix S1) at weeks 6, 11, and 15 of the 15-week semester. This approach allowed us to gain a representative account of students' social interactions and networking behaviors over the semester. Name generator surveys typically include two types of questions that probe the names of those with whom an individual interacted (i.e., name generators) and information on those an individual interacted with (i.e., name interpreters) (Perry et al., 2018). The name generator survey question asked students to list the names of people they had discussed the research project with over the two weeks prior to the survey implementation. In addition, students completed a series of name interpreter questions asking about:

1. the roles (e.g., classmate, family member) of their contacts;
2. how frequently (e.g., once, 5+ times) they talked with their listed contacts about their research project;
3. the specific topic (e.g., explaining the research project, requesting materials) of their conversations with their listed contacts surrounding their research project. Students were allowed to specify the topics of conversation with their contacts by choosing from a closed-response list of topics that are commonly associated with student research projects (e.g., methods, design, and analysis of the research project). See Appendix S1 for more details.

Science Identity Scale

Given their theoretical connections (Figure 1), we were interested in understanding the relationship between social interaction and science identity. We asked students to complete the Science Identity Scale (Estrada et al., 2011; Appendix S1) within and alongside the broader Persistence in the Sciences Survey (PITS) (Hanauer et al., 2016). While students completed the entire PITS, only data from the Science Identity Scale were analyzed given our research focus and theoretical framework. There is validity evidence for the use of the Science Identity Scale and the broader PITS in undergraduate STEM populations to measure student science identity (Estrada et al., 2011; Hanauer et al., 2016). In the Science Identity Scale, students are asked to rate their agreement on a five-point scale ranging from strongly disagree (1) to strongly agree (5), with five items concerning their view and sense of themselves as scientists. To obtain a composite measure of science identity for each student, we took the sum of student responses to the five items for a maximum science identity score of 25.

Research project grades

Through the research project, students were required to design and perform a research project, operationalize content knowledge gained in the course to complete the project, identify and read relevant scientific literature, collaborate with other students, and engage in scientific visualization and communication. Because of the holistic nature of this assignment, we used students' final research project grades as the academic performance outcome. The research project comprised 30% of students' final grades in the course and included the research project proposal (5% of the grade); attendance to in-class data review meetings (5% of the grade); measures of peer review, teamwork, and self-reflection (20%, graded for completion); and the whole-class presentation and a final research report (70% of the grade). Detailed rubrics (Appendix S1) were used to score the research project and the final research report.

Analysis Social network analysis

SNA is a set of statistical techniques used to investigate the relationships between individuals in a social system (Borgatti et al., 2009). We employed SNA to study the influence of students' interactions with members of their professional and personal communities on their research project performance and science identity. First, we used students' responses to the name generator network survey to create visualizations of the social system at each time point. In these visualizations, people are represented by nodes, and the interactions between them are represented by edges (see Table 2 for definitions and Figure 2 for illustrations). Given the nature of the data collected, wherein students listed individuals they interacted with, each of the networks produced is directional (i.e., ties flow from one node to another) and weighted (i.e., ties are weighted based on self-reported frequency of interaction). All network analyses presented in this manuscript were conducted using the package igraph (Csardi & Nepusz, 2006) in the R statistical computing environment (R Core Team, 2021, version 4.1.0).

TABLE 2 Description of applied network metrics.

Term	Definition
Node	Symbol (Figure 2a) represents an individual in a social system.
Edges	Symbols (Figure 2b) represent interactions between individuals in a social system. Edges in directed networks are depicted as arrows to indicate the directionality of the interaction (i.e., who initiated the interaction).
Network density (D)	A measure (Figure 2c) of social cohesion within a network (Borgatti et al., 2018). D can be expressed as the no. observed edges versus the no. possible edges. D ranges from 0 to 1, where D = 1 suggests complete social cohesion.
Network modularity (Q)	Measures the strength of the division of a network into “modules” (Figure 2d) or subcommunities (Newman, 2006). Q ranges from −1 to 1, where Q = 0 means the network is indivisible into subcommunities. Q > 0 means there are more edges within and fewer between subcommunities than expected due to random chance. Q < 0 means there are fewer edges within and more between subcommunities than expected due to random chance.
Indegree	The sum of ties directed toward a node. Indegree (Figure 2e) is often considered a measure of social influence or popularity within the network (Borgatti & Brass, 2019).
Outdegree	The sum of ties emanating from a node towards others. Outdegree (Figure 2f) is considered a measure of sociality or help-seeking behavior (Borgatti et al., 2018).
Instrength	The sum of weighted ties directed toward a node. Instrength (Figure 2g), while similar to indegree, provides a measure of the frequency at which a certain node is sought out by others (Borgatti et al., 2018).
Outstrength	The sum of weighted ties emanating from a node. Outstrength (Figure 2h), while similar to outdegree, provides a measure of the frequency at which a node reached out to others (Borgatti et al., 2018).

FIGURE 2. Illustrations of applied network metrics. Panels (a)–(h) provide an illustrative example of each of the network terms and measures described in Table 2.

We calculated a series of network measures (Table 2, Figure 2) to characterize the network at each time point and understand each student's overall connectivity within the network. The first of these measures is network density (D), which often provides an overall measure of the social cohesion in a network. That is, density describes the degree of connection in a social system, with more connected social networks being more socially cohesive (Borgatti et al., 2018).

To understand the structure of the community—and, thus, how students organized within the social network—we calculated the network modularity (Q) (Table 2, Figure 2). It is a global measure that indicates the degree to which a network assorts into smaller communities (Newman, 2006). We used a Louvain community detection algorithm to calculate modularity, which iteratively removes and adds nodes into different communities until there is no improvement to the graph modularity (Blondel et al., 2008).

We also calculated a series of network metrics to determine the ways individual students engaged with their professional and personal communities. Among these are the directional variants of degree centrality and strength. Degree centrality is a commonly used network measure that describes “the size of an individual's network” and refers to the number of unique individuals with whom a student interacted (Borgatti & Brass, 2019, p. 11). In directed networks, two different measures of degree centrality are used to understand each node's positionality in the network—indegree and outdegree (Table 2, Figure 2). Node strength—considered to be a weighted version of degree centrality (Borgatti & Everett, 2006)—describes how frequently students interacted with members of their professional and personal communities (Freeman et al., 1991). Like degree centrality, node strength is divided into the directed variants: instrength and outstrength (Table 2, Figure 2). We used the students' self-reported contact frequency with their listed contacts to weigh the edges and calculate node strength. While studies in education have focused on characterizing the presence or absence of social connections on outcomes (e.g., Bruun & Brewe, 2013; Williams et al., 2019), we expand this work to also include the influence of the strength of students' social connections.

In alignment with the Theory of Undergraduate Socialization (Weidman, 1989), we categorized the contacts named by students as belonging either to their professional community (e.g., classmates, teaching assistants [TAs], instructors) or personal community (e.g., nonclassmate friends, family). An additional third category of “other” was added to capture members of students' communities who did not fit any of the professional and personal roles listed on the survey. If a student listed a contact as “other” on the name generator network survey but did not provide further information on the role of this individual, the contact was categorized as “other.” While most respondents did not expand on the roles of contacts that they listed as “other,” those who did mentioned discussing their course research with a therapist and an academic advisor. These categories allowed us to parse our measures of outdegree and outstrength into “professional” and “personal” variants to capture the influence of these interactions on student outcomes. Notably, as students were the only individuals who completed the name generator network surveys, indegree and instrength only consider students' interactions with their classmates and, thus, only capture “within-class” social influence.

Descriptive statistics to understand the reason for contact and frequency of interaction

We employed descriptive statistics to understand the reasons students engaged with their professional and personal communities. First, we extracted the topics of conversation and contact roles from students' responses to the name generator network surveys across the three time points. These responses were then aggregated by contact type and topic of interaction. We took a similar approach to understand how students' frequency of interaction with their professional and personal communities changed over time, with the exception that contacts named by students were categorized as professional (e.g., classmates, TAs, instructors) or personal (e.g., nonclassmate friends, family) community members. To represent these interactions, we constructed balloon plots using the function ggballoonplot of the R package ggpubr (Kassambara, 2020).

Bootstrapped multiple linear regression

Multiple linear regression can be used to understand how students' demography, environment, and outcomes are related (Theobald & Freeman, 2014). We controlled for student demography (i.e., inputs) through the inclusion of variables for: (1) college generational status, where students whose parents achieved a bachelor's degree or higher were considered as continuing generation; (2) gender identity, although the survey provided opportunities for students to identify as gender expansive, all student identified within the gender binary; and (3) PoC identity, where students who did not identify as White were considered as PoC. We aggregated PoC students in our analysis for two reasons. The first is to preserve the rigor and validity of our analyses due to our sample size. Secondly, we grouped students as PoC in alignment with previous research and data highlighting the historical exclusion that all PoC have faced in EEB and other fieldwork-intensive disciplines (Beltran et al., 2020; Bernard & Cooperdock, 2018; O'Brien et al., 2020). For example, EEB undergraduate degree programs remain composed of predominantly White students (O'Brien et al., 2020), and fewer than 10% of undergraduate degrees in EEB were conferred to students identifying as PoC over the past two decades, even though they enroll in EEB programs at similar rates (Beltran et al., 2020; U.S. Department of Education, 2019).

We constructed four separate models to understand how student demography (inputs) and the interactions centered around their research projects (environment) influence academic performance and science identity (outcomes). Two of these models are degree-based and are designed to understand the impact of interaction with unique individuals over the semester on their science identity and academic performance on the research project. The remaining models are strength-based and are intended to investigate how the number of contacts students listed and/or the frequency of interaction they maintained with their listed contacts influenced the outcomes of interest.

Of note is that linear regression assumes that observations are independent of each other and that the data are normally distributed (Nimon, 2012). Social network data violate these independence assumptions (Wasserman & Faust, 1994). As such, we implemented bootstrapped multiple linear regression, a type of nonparametric modeling, to determine how students' interactions with their professional and personal communities impacted their science identity and final grades. All models were constructed and fit using lme4 (Bates et al., 2015) and bootstrapped using the boot.pval (Thulin, 2021) and car (Fox & Weisberg, 2019) packages.

RESULTS With whom do students form communities when discussing their course-based field research project and how do students' interactions with their community members change over time?

The course social network shows that the professional community of students, their classmates, and instructors began and remained highly connected throughout the semester (Figure 3a). Additionally, students also interacted with friends and family that make up their personal communities, who are located on the periphery of the network because they typically interact with only one student in the course. We found that the field course social network density (D) remained consistent over the semester (Figure 3a), indicating that the overall level of social cohesion was established early and remained steady over the semester. However, the modularity of the network (Q) gradually increased throughout field instruction (Figure 3a). A Q > 0 indicates that there are more connections within the detected communities than between communities than expected due to random chance (Newman, 2006). In this course, students gradually formed densely connected communities of professional and personal contacts with whom they talked about their research (Figure 3a). Further, the total number of reported interactions (early = 220, middle = 174, late = 161) and the average number of interactions reported per student (early = 6.286, middle = 5.273, late = 4.879) both decreased throughout the semester. These results suggest that students interacted with fewer individuals about their research over the course of the semester but interacted more often.

FIGURE 3. Field course student social interaction over time. Sociograms represent the social system early (weeks 4–6), in the middle (weeks 9–11), or late (weeks 13–15) in the semester. (a) Nodes (circles) represent the study participants or contacts named by the study participants (professional, personal, or other). The professional community includes students, instructors, or teaching assistants. The personal community includes students' friends and family. Edges (lines) are directional and have an arrow that signifies which nodes were in contact with each other. The network density (D) and network modularity (Q) are included in each sociogram. (b) Student reported contact frequency across each time point. Percentages are normalized by contact frequency and the professional or personal community to which the reported contact belongs.

To understand how frequently students reached out to their professional and personal communities, we analyzed the contact frequency self-reported by students for each of their reported interactions. Students engaged in research-centered interactions with members of their professional and personal communities throughout the semester at differing frequencies, depending on the time point (Figure 3b). Early in the semester, students were spending most of their time defining the goals of the research project, securing access to appropriate resources and field sites, and planning experiments. In this period, students mainly reached out to members of their professional and personal communities once or twice. However, as students began their fieldwork through to the end of the semester when they disseminated their research, they engaged in more frequent interactions with their professional and personal communities (Figure 3b). Students rarely talked with members who they indicated as filling “other” roles, and when they did, most did not provide further information on their roles.

What topics do students discuss with their community members and do these topics differ depending on the type of community?

The topics of students' interaction events were similar across both professional and personal community members but differed in key and complementary ways (Figure 4). Of the 332 professional community interactions reported by students, most were centered around topics such as explaining their research, discussing the technical aspects of their research, and talking about their experiences in the field. Students reported more interactions with their research partner than with their nonpartner classmates (Figure 4). Similarly, students interacted with the instructors more than they did with the TAs in the course.

FIGURE 4. Role of contact by the interaction topics. This balloon plot shows the frequency of students' communications by the professional and personal role of the contact. The balloons (circles) in the plot are sized by the number of reported interactions (x-axis) for a given topic (y-axis). The topics of students' communications were not mutually exclusive, students could list several topics for a single interaction. Interactions with “other” contacts (n = 12) were filtered out. RP stands for “research project.”

Students engaged with their personal community in 213 interactions. Of these interactions, students' discussions focused on explaining their research project. However, students also provided their friends and family with an overview of their experiences in the field and sought personal support related to their research project. Students reported more communication with their friends than with their family across all conversation topics.

How does student engagement with their community members relate to their science identity and research project performance?

The Theory of Undergraduate Socialization (Figure 1) posits that students' interactions with their professional and personal communities are, among other factors, influential to students' cognitive and affective outcomes (Weidman, 2006). To investigate these interactions, we used bootstrapped multivariate linear regression to describe the influence of students' interactions with their professional and personal communities on students' science identity and research project grades, controlling for student background (e.g., demographics).

We found that professional outstrength—but not outdegree—was associated with a higher science identity. Strength-based network measures focus on the weight of the ties—or how frequently students reach out to unique individuals—rather than the number of unique individuals a student talks to throughout the semester, which is explained by degree-based network measures (Borgatti & Everett, 2006). In the strength-based model for science identity, our results indicate that students' science identity, as measured by the Science Identity Scale (Estrada et al., 2011), increased by 0.545 SD for every 1 SD increase in a student's professional community outstrength. Personal community outstrength did not influence students' science identity scores (Table 3). When interpreting the strength- and degree-based models together, we conclude that students' frequent interaction with those in their professional communities (i.e., other students, instructors, TAs) is associated with a higher science identity.

TABLE 3 Bootstrapped standardized linear models describing relationships between student interactions and outcomes.

Outcome measure	Science identity		Research project grades
	Degree-based	Strength-based	Degree-based	Strength-based
Intercept	0.357 [−0.484, 1.155]	0.331 [−0.379, 1.117]	−0.361 [−1.207, 0.362]	−0.280 [−1.125, 0.414]
Gender identity	−0.308 [−1.282, 0.715]	−0.290 [−1.269, 0.679]	0.348 [−0.572, 1.294]	0.370 [−0.521, 1.367]
PoC identity	−0.175 [−1.171, 0.659]	−0.490 [−1.335, 0.394]	0.368 [−0.540, 1.212]	0.332 [−0.570, 1.184]
College generational status	−0.398 [−1.284, 0.463]	−0.111 [−1.034, 0.733]	0.065 [−0.691, 0.814]	−0.094 [−1.039, 0.748]
Professional community outdegreea	0.288 [−0.125, 0.721]	…	−0.133 [−0.555, 0.267]	…
Professional community outstrengtha	…	0.545* [0.042, 1.082]	…	−0.211 [−0.722, 0.280]
Personal community outdegreea	0.056 [−0.300, 0.426]	…	−0.033 [−0.391, 0.299]	…
Personal community outstrengtha	…	−0.206 [−0.639, 0.214]	…	0.110 [−0.332, 0.540]
Within-class indegreea	0.223 [−0.206, 0.630]	…	0.114 [−0.296, 0.513]	…
Within-class instrengtha	…	−0.109 [−0.530, 0.349]	…	0.016 [−0.381, 0.410]

Note: Values are β with 95% CI in brackets.

Abbreviation: PoC, People of Color.

^aNote that the values corresponding to these variables were converted to Z scores to ease interpretation.

^*p < 0.05 in boldface.

When controlling student inputs, neither the degree nor strength of students' social interactions with their professional or personal communities were associated with academic performance as measured by their grades on the research project. It is possible that the lack of variance in students' research project grades could have resulted in a ceiling effect on academic performance, as most students received high grades (mean = 90.53, SD = 4.31). In addition, we found no significant influence of the degree or strength of students' social interactions on a modified “research project grade” variable, which includes only the grades for the final research report and presentation (i.e., filtering out the influence of the collaboration, attendance, and self-reflection) (see Appendix S1).

DISCUSSION

Undergraduate field courses engage students in immersive outdoor learning experiences where they learn scientific practices while interacting with peers and instructors (Mogk & Goodwin, 2012; Peacock et al., 2018; Shinbrot et al., 2022; Shortlidge et al., 2021). To better understand student social behavior in campus-based field courses, we explored the following questions: (1) With whom do students form communities when discussing their course-based field research project and how do students' interactions with their community members change over time? (2) What topics do students discuss with professional and personal contacts and do these topics differ depending on the type of community? (3) How does student engagement with professional and personal contacts relate to their science identity and research project performance? We found that students formed social connections related to their research project early in the semester with professional and personal contacts. These communities gradually became more modular as students interacted with fewer contacts more frequently. Further, we found that students received slightly different yet complementary support from their professional and personal communities through the analysis of self-reported conversation topics. Finally, our results indicate that the strength of professional interactions is positively associated with students' science identity scores.

Students interact with their professional and personal communities throughout the semester

Our data are well-aligned with prior research on course-based undergraduate research experiences (CUREs), which show that students interact with their professional communities, such as students in their research groups, classmates, instructors, and TAs (Corwin et al., 2015; Deveau et al., 2020; Esparza et al., 2020). While students in our study predominantly interacted with their research partner when discussing their course-based research, we also found that students discussed their projects with their other classmates (Figure 3a). Several studies have shown that field courses encourage students to engage in social interaction with their peers and instructors, which can promote a sense of community (Boyle et al., 2007; Cotton, 2009; Mason et al., 2018; Peacock et al., 2018; Race et al., 2021).

This research is the first to illustrate that students engage in substantial field biology research-centered interaction with their personal communities (Figure 3a). These results align with prior studies on student outcomes in molecular biology (Hanauer & Hatfull, 2015), public health (Olimpo et al., 2019), and neuroscience (D'Arcy et al., 2019) CUREs, in which students discussed their course research with family, friends, and the broader scientific community. Notably, these connections can prove valuable in creating productive partnerships for community-centered research, wherein students form or make use of existing personal connections through engagement and community intervention. Prior research has shown that, through community-engaged CUREs, students experience gains in their understanding of the communal nature of research and the relevance of science to their communities while expanding their social networks (Malotky et al., 2020). Further, it is through interactions with their personal communities that students may grow in their knowledge of the domain as they find accessible ways to explain their research (Hanauer & Hatfull, 2015) and may receive perceived recognition as a scientist—a subconstruct of science identity (Rodriguez et al., 2019).

Students reported fewer interactions over time as the number of small, densely connected communities—or groups of students and their professional/personal contacts—became more common in the network (Figure 3a). Research on physics student social networks similarly found that, even though the number of interactions and average number of ties reported per student decreased throughout the semester, the frequency and influence of these interactions on students' academic performance in the course increased (Williams et al., 2019). The findings likely do not suggest a decrease in the interactivity of students in the field course over time but rather imply the existence of a selection effect, where students may strategically change who they interact with over time in ways that best fit their academic and social needs. Further, students' frequency of interaction with their professional and personal communities increased throughout the semester (Figure 3b). Altogether, these results suggest that as students decided which professional and personal contacts to discuss their research with, they also engaged with these contacts more frequently.

Student interactions with their professional and personal communities have commonalities, with key differences

Social interactions with peers and instructors in field courses are intended to provide students with opportunities to adopt the language and common practices of field scientists (Posselt & Nuñez, 2022; Streule & Craig, 2016). We found that students discussed many aspects of their research project with their professional communities, indicating that the research project encouraged students to practice scientific discourse (Figure 4). Therefore, students' engagement with their professional communities may further their acculturation into the research field.

In courses that offer research experience, instructors and TAs often fill the role of intensive research mentors to students (Heim & Holt, 2019; Hensel, 2018). During undergraduate research experiences, students indicate that they receive instrumental guidance from their mentors in the form of research-related technical guidance and clarity on the broader relevance and importance of their work (Thiry & Laursen, 2011). Thus, these prior studies align well with our finding that students procure technical guidance from their professional communities during research-centered interactions (Figure 4).

Notably, prior studies on undergraduate research (Aikens et al., 2016) and higher education (Mishra, 2020; Nelson, 2019) indicate that students also receive psychosocial support from their mentors through personal and emotional support. While students reached out to TAs and instructors for personal support on occasion, they mainly sought personal support from their research partners. Students who engage in field courses can gradually develop collaborative relationships with their classmates and build rapport by participating in shared experiences (Race et al., 2021; Treibergs et al., 2022). Therefore, it is possible that students built enough rapport to seek personal support from their research partners.

Undergraduate students also procure academic support (e.g., advice) and psychosocial support (e.g., emotional support) from members of their personal communities (McCabe, 2016; Mishra, 2020; Roksa & Kinsley, 2019). In this study, students predominantly explained their research, discussed their experiences in the field, and sought personal support from their family and friends (Figure 4). These findings are well-aligned with the research on undergraduate student sociality, wherein students contact parents and commiserate with friends about prominent and ongoing assignments and activities or during stressful periods (Cutrona et al., 1994; Mishra, 2020; Nelson, 2019). It is possible that the research project elicited a similar response from our students, wherein they predominantly sought personal support from their personal community relative to their professional community (Figure 4). Moreover, it may—at least partially—explain the increase in students' number of reported interactions with their personal communities in the final weeks of the semester when final analyses, research reports, and presentations were due.

Professional social connections—but not personal connections—are positively associated with science identity

Using bootstrapped multivariate linear regression, we found that only the strength of students' interactions with their professional community was associated with an increased science identity (Table 3). These results demonstrate the effectiveness of field courses in supporting the development of strong and productive social relationships between students, their classmates, and instructors (Mason et al., 2018). Strong ties—connections that are characterized by more frequent interaction—are critical for their role in transferring deeper social resources (Krackhardt, 1992). As such, the strength of the relationships that students build with their professional communities during a field course may be critical for the development of positive affective outcomes. Furthermore, prior studies showcase the importance of students' social interactions with others, illustrating how critical social recognition is for students' science identity production (Hazari et al., 2010; Rodriguez et al., 2019; Thompson & Jensen-Ryan, 2018). Given the positive relationship between professional outstrength and science identity, it is possible that students gained a sense of scientific recognition and acknowledgment of their capabilities as scientists through their instructors, classmates, and TAs. Future research should adopt a qualitative approach to understand the types of social resources that students receive through their professional community interactions. For example, such a study could employ semistructured student interviews, asking them why their professional community interactions were helpful to their learning and when interactions with their instructors, TAs, and classmates were crucial to their success.

Despite the personal support offered by members of students' personal communities, we did not find any relationship between students' interaction with their personal community and their science identity (Table 3). Previous studies have shown that students who talked to their families about science during elementary and middle school identified more strongly with science (Dou et al., 2019; Dou & Cian, 2021). Given the topics of students' interactions with their personal communities, it is possible that students' interactions mainly served to be a sounding board as they debriefed about their experiences during fieldwork.

In this study, none of the social interaction variables (e.g., strength, degree) influenced academic performance on the research project (Table 3). Individuals who are central in social networks are hypothesized to receive more social support and resources due to their high connectivity (Lin, 1999). In the physics education literature, students' measures of strength and degree are predictive of higher grades (Bruun & Brewe, 2013; Williams et al., 2019). However, other work has found that, depending on the context, a higher centrality does not necessarily translate to the mobilization of the resources provided by their beneficial social position to earn a higher grade (Joksimović et al., 2016). One reason why our initial hypothesis based on the theoretical framework and the cited literature is different from the outcome is the narrow spread in research project grades. To clarify the relationship between sociality and cognitive outcomes in field courses, future research could include assessments of core concepts (e.g., ecology) and competencies (understanding of the nature of science) that are aligned with field course learning goals. For field courses in EEB, instructors and researchers can implement assessments such as the Ecology and Evolution—Measuring Achievement and Progression in Science (EcoEvo-MAPS) (Summers et al., 2018) and the Biology Lab Inventory of Critical Thinking in Ecology (Eco-BLIC) (Heim et al., 2023) to measure cognitive outcomes.

LIMITATIONS

While SNA is an effective way to understand patterns in a social system, it presents certain limitations. First, while surveys are the standard for collecting social network data, they can be limited by recall bias, where participants forget interactions or names of peers, list only certain types of interactions, or list interactions with certain individuals (Borgatti et al., 2018; Brewer, 2000; Marsden, 2005). For instance, participants may be more likely to list routine interactions (e.g., with people they see regularly), interactions with close contacts, or interactions with people who belong to certain groups (Bell et al., 2007; Brashears & Quintane, 2015; Brewer et al., 2005; Wright & Pescosolido, 2002). Second, modularity algorithms are sensitive to network size, wherein larger networks will present larger modularity values (Good et al., 2010). This outcome is unlikely in this study because each time point represents a cross-section of the same network primarily containing the same nodes and edges. In addition, we would expect modularity to decrease as the network size decreased. However, we observed that network modularity increased as the network size decreased, suggesting that the network size had little to no effect on the network statistics.

It is also important to acknowledge that the results in this study are from a campus-based undergraduate field course at a university during the COVID-19 pandemic. In a typical year, students complete their fieldwork in this course before the end of November, given lowering temperatures and seasonal change. In 2020, students followed this same schedule, although all students at this university left campus and returned to their permanent residences in late November. To account for the limitations that the COVID-19 pandemic placed on in-person social interaction, students were instructed to list contacts with whom they interacted both in-person and online/over the phone. Even in a time of social distancing, it is encouraging that students were able to achieve substantial engagement with their communities even though students may not have had the opportunity to experience the full interactive and social benefits characteristic of field instruction (Mason et al., 2018; Peacock et al., 2018). It is possible that the instructional changes may, at least partially, explain students' increased engagement with their personal communities and the increased network modularity late in the semester. Still, we are hesitant to state that our findings are explained by the shift to online classes/meetings, as students' interaction with members of their professional communities also increased late in the semester.

In addition, by aggregating our input variables (i.e., college generational status, PoC identity), we risk homogenizing the diverse experiences of STEM students (Bhatti, 2021; Teranishi et al., 2020). We recognize that within these categories there are important and meaningful differences that could be explored with future research on large and diverse samples. Such research would improve our understanding of diverse student outcomes and allow researchers and practitioners to design field courses to empower all students.

Lastly, field courses can take on a variety of forms including campus-based, residential, and destination field courses (Mogk & Goodwin, 2012). The present study focused on student behavior and outcomes in a campus-based field course. Although student and instructor social behavior is similarly described across residential, destination, and campus-based field courses (Mason et al., 2018; McLaughlin & Johnson, 2006; Peacock et al., 2018; Race et al., 2021; Shaulskiy et al., 2022), it is possible that instructors who teach noncampus-based field courses may observe different results given the differences in pedagogical structure between these modalities. Future research should take a multi-institutional approach across field course types to understand the relationships between students' interactions and their cognitive and affective outcomes. The insights from such a study would allow instructors to better design field courses to leverage their connective power and further enhance student outcomes in the domain.

CONCLUSIONS AND PEDAGOGICAL RECOMMENDATIONS

Our work characterizes the social interaction reported in campus-based field course settings. Further, this research shows that student interaction during course-based research can provide critical forms of support and further their enculturation in the discipline. Based on our research, we recommend that faculty who teach field courses—especially those who provide students with the opportunity to engage in research in a campus-based setting—structure specific periods during the course to talk with students about their assignments. When instructors schedule time to meet with students, it promotes improved knowledge and affective development (Micari & Pazos, 2012; Vogt, 2008). During these meetings, instructors can aid students by answering their questions about the course material and research projects and asking about students' experiences as they engage in field research. In addition, field course instructors can also implement assignments, like the research project in this course, that support student collaboration.

While interacting with their personal communities did not influence student outcomes (Table 3), the topics of students' science discourse indicate the potential for science communication. Hanauer and Hatfull (2015) posit that if students are talking to their friends and family about the research they are conducting in their courses, then they are engaging in a form of science communication. Science communication can be a way to build trust in science and enhance the public perception of scientific work (Brownell et al., 2013). Thus, if most students are talking to their personal communities about their research, instructors can integrate opportunities for students to engage in science communication with their personal communities as part of the field course curriculum. For example, instructors could provide opportunities for students to involve their personal communities in research-related milestones, such as allowing them to invite family members and friends to end-of-semester research presentations (Rodriguez et al., 2019) or teaching about effective science communication to nonexperts (see Wack et al., 2021, for framework). By implementing such strategies, students can practice communicating the broader relevance and importance of their work to those closest to them. In addition, by teaching students about effective science communication, we may also help students gain perceived recognition and emotional support from their personal community (Alpert et al., 2009; Rodriguez et al., 2019).

AUTHOR CONTRIBUTIONS

David Esparza led the conception, funding acquisition, experimental design, data collection efforts, analysis, interpretation of results, writing, and editing of the manuscript. Michelle K. Smith supported funding acquisition, experimental design, interpretation, and editing of the manuscript.

ACKNOWLEDGMENTS

We thank the students who participated in the present research. Data collection efforts were supported by Drs. Kira Treibergs and Marc Goebel. Aspects of the analyses were made possible through collaboration with Dr. Matt Thomas (Cornell Statistical Consulting Unit). We are indebted to the Cornell Discipline-based Education Research group for their thoughtful comments on the manuscript and Lillian Senn for her insight and feedback on the research. This material is based upon work supported by the National Science Foundation Graduate Research Fellowship Program under Grant No. DGE—2139899 and the Cornell Laboratory of Ornithology Athena Fund for Graduate Studies.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

DATA AVAILABILITY STATEMENT

The modeling dataset (Esparza, 2023a), node and edge lists (Esparza, 2023b), and the social network analysis dataset (Esparza, 2023c) are available from Figshare: https://doi.org/10.6084/m9.figshare.21717500.v2, https://doi.org/10.6084/m9.figshare.21824733.v2, and https://doi.org/10.6084/m9.figshare.21718079.v2, respectively. These datasets are anonymized in accordance with the guidelines described in protocol no. 2001009364 approved by the Cornell University Institutional Review Board. Any requests for additional information pertaining to the data should be directed to the corresponding author.