How Linked Interdisciplinary CUREs Impact Student Social Networks and Thinking About the Value of Other Disciplines

Abstract

To address global environmental and health problems, scientists must work across disciplinary boundaries. The Science of Team Science is a field of study that examines the processes by which effective scientific teams operate across disciplines. We incorporated strategies from team science into undergraduate courses to help students develop an appreciation of other disciplines and to learn how to create productive science research teams. We then explored student team interactions and interdisciplinary thinking in three course-based undergraduate research experiences (CURE) in three different disciplines that were linked by complementary research questions in the same system. Through writing prompts scored with the Interdisciplinary Science Rubric, students demonstrated an intermediate level of understanding of the importance of interdisciplinary teams. Social network analysis revealed evidence of students learning from and building trust with students in the other CUREs. This study highlights the benefits of integrating concepts across disciplines in science, technology, engineering, and mathematics (STEM) education to better prepare undergraduates for modern STEM careers.

INTRODUCTION

Addressing complex global environmental and health-related challenges requires productive collaboration between professionals from multiple disciplines. There have been widespread calls to increase interdisciplinary work in higher education in order to prepare students for these workplace demands (AAAS, 2009; Santangelo et al., 2021). However, those calls are often met with acknowledgments that interdisciplinary instruction is difficult due to higher education structure, funding regulations, and publication limitations (Hubbard and Gregory, 2011; Opoku Agyeman et al., 2019; Crichton et al., 2022; Telling, 2023). Finding creative solutions to integrate instruction with ideas from team science (Börner et al., 2010; Billerbeck, 2021), however, can help prepare students to work effectively in teams with students from other disciplines while working within the structural limitations of institutions. The Science of Team Science is a field of study that examines the processes by which scientific teams communicate, collaborate, and function productively (National Research Council, 2015). Strategies from team science can help students develop the appreciation of other disciplines, strong communication skills, and trust that are necessary for professional science teams to effectively accomplish desired outcomes (Feitosa et al., 2020; Kilty and Burrows, 2022).

Lessons from team science particularly focus on creating productive science research teams and could help enrich undergraduate research experiences. The educational benefits of participation in undergraduate research experiences (URE) are well-documented in academic literature (Lopatto, 2009; Bangera and Brownell, 2014; Castillo and Estudillo, 2015; Krim et al., 2019). One frequently utilized method of incorporating UREs into the curriculum is through the development of CUREs (Auchincloss et al., 2014; Bangera and Brownell, 2014; Krim et al., 2019). CUREs have demonstrated impacts on helping students improve professional skills (Auchincloss et al., 2014; Krim et al., 2019). However, there are few CUREs that explicitly include a focus on team science, using interdisciplinary collaboration from diverse scientific fields to solve present-day problems (Hall et al., 2018). This study examines the effectiveness of promoting interdisciplinary interactions amongst students in three linked course-based undergraduate research experiences (CUREs). By infusing team science instruction into interdisciplinary CUREs (ID-CURE), we tested whether students can gain the interdisciplinary thinking and skills for collaboration that will better prepare them for entry into the STEM workforce.

There are multiple definitions for interdisciplinary research across team science (Austin et al., 2008), education (Tripp and Shortlidge, 2019; 2020; Tripp et al., 2020), and organizational psychology (Collin, 2009) literatures. In this paper, we adopt the working definition of the term interdisciplinary that was proposed by Tripp and Shortlidge (2019) based on a survey of science faculty and published works. “Interdisciplinary science is the collaborative process of integrating knowledge/expertise from trained individuals of two or more disciplines—leveraging various perspectives, approaches, and research methods/methodologies—to provide advancement beyond the scope of one discipline's ability” (Tripp and Shortlidge, 2019, p. 5).

Interdisciplinary Science Framework

We used the Interdisciplinary Science Framework (Tripp and Shortlidge, 2019; Tripp et al., 2020; Tripp and Shortlidge, 2020) as our theoretical basis. The Interdisciplinary Science Framework was developed to encourage educators to engage undergraduate students in interdisciplinary thinking (Tripp and Shortlidge, 2019). In this study, instead of utilizing the framework to inform course design, it was utilized to assist researchers in the identification of instances in which student participants worked across disciplines or showed an appreciation for other disciplines.

The Interdisciplinary Science Framework was developed by incorporating constructs from previous education research theories that addressed interdisciplinarity and environmental sustainability. Their five initial constructs included disciplinary grounding (basic understanding of contributing disciplines), advancement through integration (how the disciplines integrate to advance the solution of a question toward a common goal), disciplinary humility (developing a mindset that is infused with humility, inclusivity, and respect for other disciplinary epistemologies), different research methods (the ability to know and be able to use a range of possible methods from different disciplines), and collaboration across disciplines (working with people from other disciplines) (Boix Mansilla and Duraisingh, 2007; Öberg, 2009; Byrne et al., 2016). Through the process of rigorously testing the reliability and validity of a rubric to assess this framework, Tripp and Shortlidge (2020) later reduced the constructs to four: objective (why different disciplines need to work together), disciplinary grounding (understanding of the contributing disciplines themselves), integration (how to combine information from different disciplines to solve a problem), and broader awareness (ability to communicate the broader impacts of the work). Together, these four criteria serve as the theoretical framing for our study. They can be assessed in a variety of ways, including an evaluation of student products, writing prompts, and student interactions with people from other disciplines during a shared research project.

Social Network Theory

One way to study student interactions, within and across disciplines, is with social network analysis (SNA). Social network theory states that relationships between people and/or organizations are central to the transfer of resources and information within a network. Relationships between individual participants in the network may support or inhibit this transfer (Borgatti and Halgin, 2011). Although many theories utilized in education research attribute differences in the performance of students to their unique characteristics, social network theory suggests that relationships among network participants can have a negative and/or positive effect on individual behavior and performance. Therefore, an individual's contributions must be examined in the context of their social network, which includes individuals, interactions with others, and an overall network structure. These interactions amongst individuals provide paths through which they may affect each other directly and indirectly. A fundamental tenet of social network theory is that a student's position within the network influences their support, information, and resources, and thus also the capacity of their team to perform their work effectively (Borgatti and Halgin, 2011; Poulsen et al., 2024). Social network theory offers a useful conceptual framework and accompanying methods for describing and analyzing the structure of the social system formed within the linked interdisciplinary CUREs (ID-CURE) to understand how team interactions may influence each other (Daly, 2012).

SNA is often utilized to study the structure of teams and visualize interactions between individuals in a given community (Digital Promise, 2018). In a social network, individuals (in this study, individual teams) are represented by a point called a node, and their interactions are represented by a directional arrow called a tie (some SNA studies also refer to this as an edge; we will refer to it as a tie). Figure 1 shows a hypothetical social network of three teams, Alpha, Beta, and Gamma. In this network, there were two interactions between team Alpha and team Beta, so their tie is thick. However, teams Gamma and Alpha only interacted once, so their tie is thinner. Teams Beta and Gamma did not interact, so they do not have a tie between them. Networks can be viewed at multiple levels, including the individual node level, dyad level (properties of and ties between two nodes), and the whole-network level (e.g., size and density; Borgatti and Halgin, 2011). The strength of a network can be characterized based on the overall density, or total number of ties (Digital Promise, 2018). For this study, teams were the unit of interest, and SNA allows us to use teams as nodes to evaluate the degree of connectedness between student teams within and across disciplines (i.e., within and across linked CUREs).

FIGURE 1.

An example social network for individuals Alpha, Beta, and Gamma.

Although social networks reflect the overall social system formed within and across these ID-CUREs, the characteristics of individuals, including their social identity, may impact their relationships within the network and their placement within the network. For example, exclusionary pressures and individuals’ preferences led racial minorities and women to interact more frequently with individuals of the same race or gender in one academic cohort than did members of majority groups (Mehra et al., 1998). Also, lab courses with less structure and higher levels of inquiry have been found to reinforce gender disparities by encouraging students to sort themselves into stereotypical gender roles (e.g., males using scientific equipment and women completing managerial work) (Doucette et al., 2020). However, in other circumstances, members of a minority group have been found to be equally well connected to the center of the network as were members of a majority group (Leonard et al., 2008). In addition, student gender identity did not influence who students viewed as most proficient in CURE investigations (Esparza et al., 2023). Therefore, the impact of social identity on one's placement and connections in a social network can vary. In this study, we focused on teams as our main unit of analysis. As our student teams were made up of individuals with a variety of social identities, we could not evaluate the role of social identities on these social networks.

This study addresses the following research questions (RQs):

• RQ1: How does information flow between teams within, and between, the three disciplines (i.e., the three linked CUREs)?

• RQ2: After participating in these linked ID-CUREs, do students demonstrate proficiency at interdisciplinary collaboration as measured by the four constructs from the IDSR? Do the three disciplines differ in their performance on these metrics?

• RQ3: How connected is the network of teams in these linked CUREs, both within and between disciplines?

MATERIALS AND METHODS

Descriptions of ID-CUREs

This study takes place at a large, R1 (Carnegie Classifications|Definitions, n.d.) university in the Southeastern United States. This study was approved as exempt from institutional review board (IRB) review by the supporting institution. The linked ID-CUREs were three upper-level undergraduate courses: Microbial Ecology (biology), Groundwater Hydrology (environmental engineering), and Hydrogeology and the Environment (geology).

Students that met the discipline-specific prerequisite courses self-selected to register for a CURE. Student demographics acquired from university course databases were well-aligned with institutional means (Table 1), though students identified as female were slightly less represented in the CUREs than the institution due to the skewed gender distribution in some of the participating disciplines. We have data from 36 undergraduate students who participated in the ID-CUREs during spring 2023, 15 undergraduate students in the biology course, 11 in the engineering course, and 10 in the geology course.

TABLE 1.

Participant demographics

Demographic	ID-CUREs (N=36)	Institution
Female	17 (47%)	61%
PEER	9 (25%)	27%
Pell Eligible	15 (42%)	45%

Note: Demographics collected utilizing institutional records. Gender was collected on a binary scale. PEER was defined as persons historically excluded due to ethnicity or race. In this study, the PEER designation includes individuals who are Black or African American, Hispanic or Latino/a/x, or identify with two or more races/ethnicities.

All three courses were four-credit-hour courses designed to align with the five components of a CURE—science practices, discovery, relevance, collaboration, and iteration—that have been previously recommended (Auchincloss et al., 2014). This study takes place in the third implementation of these ID-CUREs. Students were enrolled in a single course within their discipline and were arranged into teams of three to six undergraduate students within their courses.

Approximately once per month in the 4-mo semester, students from all three courses met jointly during their normal class time (the classes were scheduled with overlapping meeting times). Additionally, there was a Microsoft Teams site on which students were encouraged to communicate, share data, and ask questions across the three linked ID-CUREs. Instructors and teaching assistants (TAs) for each of the three courses were also available to students in the other courses, further promoting the sharing of knowledge between the disciplines.

Students in all three courses were working on a collaborative research project focused on a stream near the campus of the participating institution. The overarching scientific research goal was to evaluate how the microbial community and hydrology impacted the effectiveness of a regenerative stormwater conveyance as a sediment and nutrient treatment best management practice. Each course addressed different parts of the large research question and then shared information across courses. These collaborations reinforced the goals of the ID-CUREs and allowed the students to work on their portions of the research project throughout the semester.

Team Science Interventions

Each CURE had an instructor and a graduate teaching assistant. ID-CURE faculty and TAs received multiple hours of training in team science. Faculty participated in a 2-d team science workshop through the American Institute of Biological Sciences covering topics such as the importance of interdisciplinary teams, characteristics of effective teams, strategies for ensuring shared goals and objectives, promoting productive communication, and dealing with conflict. Teaching assistants attended three workshops with project leaders throughout the semester, for a total of ∼5 h. This curriculum was designed to help the teaching assistants promote the development of team science competencies in their CURE students. In addition to these training activities, there were multiple sessions for CURE faculty and teaching assistants to share their experiences. Our team science training materials are available online (https://digitalmarket.ecu.edu/teamscience/).

In each of the courses, team science instruction and activities were implemented to improve students’ knowledge, skills, and attitudes (KSA) toward working in teams and across disciplines (Hall et al., 2018). For example, developing team competencies was articulated to the students as an explicit learning objective of these CUREs, and team-building activities were implemented early in the semester. Students were encouraged to exchange contact information and to communicate with students in the linked CUREs to accomplish their research tasks, build trust, and model the collaboration of interdisciplinary teams in real-world projects. In addition, students were expected to produce two important tools designed from Team Science principles to promote effective teamwork: team communication plans and research plans (Vance-Chalcraft et al., 2023).

The team communication plan was completed at the beginning of the semester, as soon as the students were assembled into research teams. When completing these documents, the students had to reflect and decide on what they felt were important principles for effective and inclusive communication, what mode of communication they wanted to use, the frequency of communication and time to respond, and how they would handle conflicts that arose (Vance-Chalcraft et al., 2023). If a team member, instructor, or teaching assistant noticed signs of communication challenges throughout the semester, they had the team review their communication plans and revise as necessary. Instead of intervening, the instructors encouraged the students to follow their developed process (or create a new one) to resolve their conflict.

The research plan was a living document that students added information to throughout the semester (usually weekly). At the beginning of the semester, they had to articulate as a team what they saw as the overall research goal or question, and their hypothesis. They would then regularly add to the document to reflect on their progress, determine next steps, and assign tasks and deadlines for each team member. Teams were instructed on the importance of making sure each person had a role each week and using the research plan as a tool for accountability and for ensuring that the team members all shared the same mental model of their project.

Data Sources

Student Products.

The linked ID-CUREs used Microsoft Teams to allow students to share data, files, and post questions to each other throughout the semester. In addition, students were given a variety of research deliverable assignments to meet the learning objectives of the individual course. All student teams gave a research presentation at the end of the semester as one of the meetings with all three disciplines. One researcher performed content analysis on these course artifacts from all three courses, including the Microsoft Teams site and final presentation deliverables (all other deliverables differed across the three courses). Content analysis is best utilized to assist researchers in reducing large amounts of qualitative data and identifying core consistency and meaning in the data (Patton, 2002). We evaluated artifacts to determine whether there was any evidence of communication between student research teams, particularly between groups in different disciplines within the ID-CUREs. The Microsoft Teams messages between the disciplines were counted to track instances of communication, including who requested information and who provided information across the courses. Analysis was confirmed by two additional researchers familiar with the project and qualitative analysis techniques. This analysis was used to verify that information did flow between the disciplines and was not intended for more substantive conclusions, as the communication was generally very focused on requesting and providing specific pieces of information (e.g., site photos, data on a specific process) across disciplines.

Writing Prompts.

As part of their final exam, the ID-CURE students responded to a writing prompt that asked them to describe a recommended team approach for dealing with a real-world ecological problem. All three ID-CUREs completed a prompt designed by the authors of the Interdisciplinary Science Rubric (IDSR) (text of the prompt is available in Supplementary Material A as “course C” from Tripp and Shortlidge, 2020). The writing prompt provided a scenario in which they were to imagine they were expert scientists hired by a city to create a long-term plan for using green methods to clean up a waterfront area that is contaminated from an oil spill. Students were specifically asked to include the intended approach to the task and what tools, techniques, procedures, individuals, and groups needed to be involved in the process. They were provided with the instructions and student version of the IDSR rubric when the writing prompt was assigned, to mirror the procedures the original authors used (Tripp and Shortlidge, 2020). The restoration project discussed in the writing prompt was not taught or discussed explicitly in class. However, the themes presented were similar to the research project students participated in as part of the CUREs.

The validated instructor version of the IDSR (Table 2 in Tripp and Shortlidge, 2020) was used to score student responses to the writing prompt to determine student thinking about the value of working with others in varying disciplines to solve problems. As in Tripp and Shortlidge (2020), each item was assigned a score between zero and three, in which zero is considered naïve, one is novice, two is intermediate, and three is mastery. High-quality responses to this prompt include an explicit discussion of the involvement of professionals from multiple background disciplines and their role in the project. Additionally, students must discuss ways to facilitate collaboration between these experts from different disciplines (Tripp and Shortlidge, 2019). Scores were analyzed for all four subscales on the rubric: Objective, Disciplinary Grounding, Integration, and Broader Awareness.

To validate the use of the IDSR in our student population, think-aloud interviews (n = 3) were conducted with undergraduate students from biology and engineering who were not participants in the spring 2023 ID-CUREs. These interviews took place online and were recorded to generate a transcript used for qualitatively coding the responses. During the interview, students were asked to read the writing prompt and IDSR rubric and then describe how they would approach crafting their response. After outlining their approach, interviewees were asked to describe what would be necessary to include in their response in order to receive a score of three for each of the IDSR criteria. Additionally, interview participants were asked if there was anything they found confusing regarding either the writing prompt or IDSR. After the transcripts were reviewed for accuracy, a codebook was developed deductively to evaluate whether the students had a rational essay approach and if their interpretation of the IDSR rubric was in alignment with the research team. Two of the transcripts were coded by two members of the research team to establish interrater agreement. A kappa score of 0.879 was achieved, which indicates near-perfect agreement between the two researchers (Cohen, 1960). The remaining transcript was then coded by one of the researchers. These student interview methods mirror those used by the original rubric authors (Supplementary Material B in Tripp and Shortlidge, 2020).

To become familiar with using the IDSR rubric to score student writing prompt responses, two researchers scored student writings with a similar format from a prior semester. Then, a subset of the student responses to the writing prompts was scored (four from each discipline) by both researchers to obtain interrater agreement. A kappa score of 0.671 was achieved, indicating substantial agreement between researchers (Cohen, 1960). The researchers discussed their interpretation of the rubric's criteria in detail and created notes for scoring. One researcher then scored the remaining student responses. Kruskal–Wallis tests were used to determine whether there were significant differences in the scores among constructs (objective, disciplinary grounding, integration, broader awareness) or by discipline (biology, engineering, geology) for each of the constructs.

Social Network Surveys.

SNA was used to examine how many connections were formed between teams in and across the courses. To gather social network data, surveys were administered to CURE students at two points in the semester. The presurvey was administered within the first week of the CUREs to establish a baseline social network. At the end of the CURE experience, a second, similar postsurvey was administered to all CURE participants. Survey items were printed on the columns of a grid, with the names of all ID-CURE participants printed along the rows and potential modes of interactions as the columns. CURE participants were instructed to mark “yes” when another participant fulfills a survey item. Three networks were constructed based on two survey items: “Who do you trust?” (the trust network) and “Who did you learn from?” (the learn network), respectively. The trust item was asked on both the pre-and post-surveys, while the learn item was only asked on the post-, thus resulting in a total of three networks constructed. We did not collect pre-learning data because we were focused on who an individual learned from in this particular ID-CURE setting. Trust and learn items were selected for SNA as they best exemplify the tenets necessary to build strong teams, as discussed in the team science aspects of the CUREs (Börner et al., 2010).

Following the SNA survey data collection, individual student responses were condensed into their teams to form valued adjacency matrices that assess the prevalence of interactions at the team level, so each node in the network represents a team in one of the ID-CUREs. As the responses from each student were aggregated into their respective teams, the interaction between one individual from one team with an individual from another team would establish a tie, or connection, between these teams. The network ties were then weighted (as shown by the thickness of an arrow) based on the number of undirected connections that existed between two teams. The networks and their characteristics were examined using SNA software called UCINET 6 (Borgatti et al., 2002). Networks were drawn using the network visualization tool NetDraw that comes with the UCINET 6 software. Network metrics (i.e., density of ties between teams) and E–I indexes were calculated.

The density of a social network is a metric used to demonstrate how interconnected individuals are within a given network. It is calculated by dividing the total number of ties present within a network over those possible if all individuals within the network were connected. This produces a value between 0 and 1, in which a value of 0 would represent a network with no connections and 1 in which all individuals were fully connected to one another. Network density can also be used as a tool for comparing networks across time or multiple contexts. Networks that are more dense have a greater proportion of individuals who are connected with one another. Increased density is thought to correspond to greater information flow within a group of individuals and an increased ability to build knowledge by working to understand complex tasks or information (Obstfeld, 2005; Himelboim, 2017). Higher levels of knowledge sharing within a network have been positively associated with a team's overall performance (Henttonen et al., 2013).

The E–I index (Krackhardt and Stern, 1988) is a measure of homophily, or how closely tied similar (as opposed to dissimilar) individuals are within a network. This measure evaluates the balance of internal and external ties within a network on the basis of a predetermined group assignment or macrostructure. This method allows researchers to determine how bound the nodes within a network are to their assigned structure or whether connections extend throughout the entirety of a network. The E–I index can be calculated based on the equation below at several levels of network organization, including the whole network, group, or individual level. Because any tie is viewed as an instance of connection, the directionality of the tie is ignored. If a group had an E–I index of –1, this would indicate that all connections are internal and there are no external connections to any other group. With an E–I index of 1, all connections within the group are externally directed, and there are no interactions between group members. An E–I index of 0 would indicate that the number of internal and external ties is perfectly balanced, and the given group interacts both with its members as well as groups outside of their assigned structure.

In this study, the E–I index was used to see if student teams in the ID-CUREs were making connections across disciplines (E–I index closer to 1) or solely within their discipline (E–I index closer to –1). Each individual team was grouped by discipline (course) for the purpose of calculating the E–I indexes. Internal ties were considered to be connections between teams in the same discipline, while external ties were connections between teams from different disciplines. Valued adjacency matrices were used to calculate the E–I indexes to take into account the frequency of the interactions between the teams. The E–I indexes were calculated for both the whole network (all three linked CUREs) and by discipline (each CURE separately) to provide a more comprehensive view of the network, evaluate whether the level of homophily within the network was driven by any one discipline, and compare information flow on a multidisciplinary level. Because social network measures are not independent from one another as they evaluate measures between individuals or groups, traditional statistical analyses cannot be performed to determine whether there is a significant level of internal or external connection among the members of the network. As a result, permutation tests were performed on the whole network in RStudio (Posit Team, 2024) to determine whether there was a significantly greater or lower level of homophily within the disciplines than would otherwise be predicted by chance.

RESULTS

Analysis of Student Products

Analysis of posts on the Microsoft Teams page and student final presentations indicated that there was communication between the disciplines, primarily by students of one discipline asking individuals from another discipline if they had certain information (e.g., details of certain site features, data on stream flow) and another discipline providing (or not) that information. Specifically, students and instructors from geology took pictures of site locations and shared them on the Teams page, many of which were used in the final presentation. The biology students relied heavily on data collected by the geology students and information (figures, descriptions, etc.) from the engineering students. For example, site and watershed descriptions were present in all three disciplines’ presentations (Figure 2), even though biology students were not involved in data collection at the field site as part of their coursework. A shared field laboratory planning document on Microsoft Teams further demonstrated that all field data were regularly collected by the geology class and then shared within a folder that could be accessed by all courses. Questions regarding these data were resolved within a data-sharing notes file in which teams from all disciplines made requests for specific pieces of data, which were shared by individuals from either geology or engineering.

FIGURE 2.

An example of student collaboration is evident in a student team's final presentation. This example is from a biology presentation where the image and site descriptions are based on information gathered from collaboration with the other disciplines. Image converted to grayscale for publication purposes.

Counts of the instances of interteam communication between the disciplines indicated which disciplines were requesting information from the others. Biology students initiated the greatest number of conversations with other disciplines, with a total of seven external messages to engineering and six to geology. These messages most commonly were regarding questions related to the field data collected and took place through direct messaging to geology and engineering teams on their individual Microsoft Teams channels. Engineering initiated four conversations with biology students and one with geology to share and discuss data related to the study site. The geology students did not initiate any conversations with any other disciplines; however, both engineering and biology directly messaged the geology teams on their respective channels to request field data. Additionally, there is no evidence that students from the engineering or geology students relied on data or information from the biology students directly.

Writing Prompt Scores

The results of the think-aloud interviews indicated that all of the student participants were able to construct a response to the writing prompt that was rational and in alignment with that of the research team. For each of the rubric categories on the IDSR, all interviewees demonstrated comprehension of the corresponding criteria and were able to successfully explain how they would incorporate the criteria into their response. Their interpretation of the criteria on the IDSR was in alignment with the expectations of the research team. No aspects of the writing prompt or IDSR were noted to be especially confusing by any think-aloud participant. Additionally, all of the participants indicated that they valued interdisciplinary work and felt the assignment was unique and useful for synthesizing interdisciplinary knowledge.

Table 2 presents exemplars from student essays for each construct. The three criteria for the Objective construct include Purpose, Approach, and Credibility. The Purpose criteria (1.1) required that the essay include a description of the site (highlighted in bold), the problem (highlighted in underlined text), and then identify the task (highlighted in italics). In the exemplar, these three elements are: a contaminated waterfront, soil contamination with alkyl halides, and reducing the soil contamination to use the site as a community park and garden. For the Approach criteria (1.2), a mastery score required multiple steps with specific procedures. The Credibility criteria (1.3) required students to use sources that were relevant to the problem or the task. Students often cited course materials, but the mastery score required citing relevant peer-reviewed research.

TABLE 2.

Student essay exemplars for rubric criteria

Construct	Criteria	Exemplars from student essays
Objective	1.1 Purpose: Description of the site (bolded), the problem (underlined), and the task (italics).	A coastal Louisiana town has a large waterfront area that they would like to convert into a community park with a large vegetable garden. However, due to the town's large amounts of industrial runoff, there is a high amount of oil along the shoreline and in surrounding soil samples. After analyzing the soils in the shoreline and around the shoreline, the town also found large amounts of alkyl halides, an organic compound not suitable for plant growth. With the help of myself and other scientific experts in various fields, we would like to reduce this contamination along the shoreline and in the soils so that the coastal Louisiana town may build their community park with the vegetable garden. [BIOL; 3.0]
	1.2 Approach: Formulate a plan that clearly outlines your approach (steps/procedures) to solve this problem.	Therefore, with the help of various team members we would like to begin by constructing an oil-degrading bacterial consortium, to remove the oil from the shoreline and surrounding soil samples (Márquez-Rocha et al., 2001). Furthermore, to remove the alkyl halides from the soil by injecting said soil with a compound called Phanerochaete chrysosporium, which is a fungus capable of organic breakdown (Kennedy et al., 1990). In this case, the organic breakdown of alkyl halides. [BIOL; 3.0]
	1.3 Credibility: Use peer-reviewed articles and/or other supporting information that are relevant to the problem/task.	Kennedy, D. W., Aust, S. D., & Bumpus, J. A. (1990). Comparative biodegradation of alkyl halide insecticides by the white rot fungus, Phanerochaete chrysosporium (BKM-F-1767). Applied and Environmental Microbiology, 56(8), 2347–2353. Márquez-Rocha, F.J., Hernández-Rodrí, V. & Lamela, M.T. Biodegradation of Diesel Oil in Soil by a Microbial Consortium. Water, Air, & Soil Pollution 128, 313–320 (2001). [BIOL; 3.0]
Disciplinary Grounding	2.1 Disciplines/Experts: Include two or more disciplines and/or experts in your approach to the problem/task. (Bolded) 2.2 Disciplinary Reasoning: Meaningfully explain the reasoning behind the use of each discipline and/or expert. (Underlined) 2.3 Methods and Tools: Include techniques/procedures/tools from contributing disciplines and/or experts. (Italicized)	A large workforce will be needed to make this operation successful and possible. To begin the process, A civil engineer will be hired to decide on the best type of dam for the site and on the most suitable materials to be used. Following this, a team of builders will begin to install the dam in the appropriate location. The specific tools and materials used will be based on the kind of dam being built. Next, a large excavation team will be needed to remove much of the harmful soil in the area. This team will correspond with the bioremediation team to ensure that the excavation will not hinder or remove important soil used for the plant life. This team will include site managers, heavy equipment operators, and a cleanup crew. Large machinery (Hydraulic shovels) will be used in the excavation process. Next, gardeners and plant life specialists will be hired to find what plant life will be best to help treat the area as well as go forward to plant them in the surrounding area. Next, specialized stormwater management companies will be involved. These organizations have the expertise and equipment necessary to design and construct stormwater conveyance systems that meet local and federal regulations and effectively manage stormwater runoff. [GEOL; 3.0]
Integration	3.1 Leveraging Disciplines/Experts: Specifically address how each discipline's and/or expert's contribution (knowledge/methods) will build off one another to effectively address the problem/task in a way that one contribute cannot.	One specific series of events where I could see all members being utilized would be the following: The soil scientists will collect soil and water samples. After conducting testing, they understand they are dealing with a significant oil amount. After consulting with biologists, they together evaluate wetland options, sorbent recycled materials, as well as natural straw and hay options, which are natural, biodegradable, and lightweight. Together, they determine that polypropylene, a floating, woven plastic polymer, which has proven high removal rates for large oil spills, would be a feasible solution. Polypropylene is a hydrophobic fiber, commonly installed as floating mats or along shorelines, and has higher adherence to oil than sand (Murray, et.al, 2020). The environmental engineers would use their knowledge of common sorption practices to design the quality of these woven mats to implement a useful solution. The construction managers would be consulted to not only do the physical labor, but to establish a designing schedule so the rest of the team knows when they will be required to report on their parts of the project and inform the community of the intended opening date. The ecologists will plan for how to minimize the disturbance on the surrounding ecosystem as well as minimize the carbon footprint of the project. They will also consult with the environmental engineers to conduct a disposal and, if possible, a recycling process for mats that meet their maximum removal abilities. [ENVE; 3.0]
	3.2 Collaboration: Include and explain two or more ways to build community and respect among different disciplinary team members (e.g., establishing common ground and language, overcoming different perspectives, etc.).	Creating a comprehensive remediation plan will require all members of the multidisciplinary team to collaborate and to inform other members of their specific expertise. The lack of knowledge from some disciplines will be made up with knowledge of others. The team will need to establish a common understanding of their specific roles in the project. They will need to communicate openly and often about their progress as well as current obstacles. [ENVE; 2.0] To foster successful partnerships between the disciplinary team members, we will set regular meetings (bi-weekly or monthly) to update each other on our progress in the project. In addition, open lines of communication will be maintained using messaging apps. At the beginning of the project, we will create a communication plan outlining the mode of communication, frequency of communication, and rules of communication. We will also create a plan to address conflict and come to an effective compromise. An interdisciplinary research plan with a timeline and task list will also be formed to ensure all members are aware of what each team's role is in the project. Both the communication and research plans are to be living documents that will be updated throughout the course of the project. [BIOL; 3.0]
Broader Awareness	4.1 Societal Impact: Include why your solution is locally and more broadly relevant to society and what/who will be affected (e.g., economics, politics, social, health, etc.).	It is expected that the park will be returned to natural conditions, with a large domain of the park suitable for agricultural growth with the utilization of these methods. These changes will benefit the entirety of the community by creating a more sanitary living environment and functioning as a source for agriculture. At a greater scale, this transformation will serve as an example at a nationwide scale by illustrating how important water quality and ecologically friendly methods of renovation can transform a community, with the hope that other communities will adopt similar methods. [GEOL; 3.0]
	4.2 Limitations: Forecast possible limitations of your plan and provide resolutions.	Some possible limitations within this plan include microbes/plants not surviving, and industrial runoff contaminants persisting within the site. The remediation of the crude oil relies on the microbes and plants flourishing in their environment so they can carry out their natural processes, while also remediating. In the event of cold weather or a natural disaster, these organisms could die, thus stopping remediation. To combat this, the project will be implemented during the summer months to avoid the cold. The weather will also be forecast out several months to anticipate any storm events. If the environmental engineering team cannot produce a green solution to the industrial runoff contaminants, the contaminants can be located and excavated. This solution will greatly increase cost and labor, but will exterminate the contaminants. [ENVE; 3.0]

Note: Constructs and criteria are from the final version of IDSR rubric (only the formatting criterion was not included here for research purposes) (Tripp and Shortlidge, 2020). Exemplars come from student responses to the writing prompt in the linked ID-CUREs. The discipline of the student and the score they received on this criterion is included, in addition to a segment of their text response.

The Disciplinary Grounding construct included the criteria of Disciplines/Experts, Disciplinary Reasoning, and Methods and Tools. The scoring of the Disciplines/Expert criteria (2.1) was simply a count of two or more disciplines that were relevant and specific. The Reasoning criteria (2.2) required students to provide an accurate rationale for including each discipline they named. The Methods and Tools criteria (2.3) had to be present and accurate for all disciplines.

The Integration construct has two criteria: Leveraging Disciplines/Experts (3.1) and Collaboration (3.2). A response was considered to demonstrate mastery of the first criterion (Leveraging Disciplines/Experts) if they described, in detail, how each discipline's and/or expert's contribution would build off one another to effectively address the problem in a way that could not be done by one discipline/expert alone. The Collaboration criterion requires a clear identification of tools to build collaboration, such as a communication plan, conflict resolution strategies, regular meetings, and research or project plans with clear expectations.

The Broader Awareness construct has two criteria: Societal Impact (4.1) and Limitations (4.2). The Societal Impact criterion requires specific details on the impact on the local community. In this case, some of the impacts could be on agriculture and the potential to serve as a national model for a green solution to industrial runoff contaminants. Finally, student responses for the Limitations criterion meet the Mastery level if they identify potential limitations, such as microbes/plants not surviving due to cold weather.

The mean scores and SD for the IDSR writing prompt (each criterion scored from zero to three) are presented in Table 3. Mean scores revealed that students were achieving an intermediate level of understanding in the Objective and Disciplinary Grounding categories (indicated by a mean of 2 on the IDSR scale), and a novice level of understanding in the Integration and Broader Awareness categories (mean of 1 on the IDSR scale). Kruskal–Wallis test results indicated there was a significant difference in the student scores among the four constructs (χ²[3] = 49.47, p < 0.001). The mean rank scores were 99.83 for the Objective construct, 92.54 for the Disciplinary Grounding construct, 55.04 for the Integration construct, and 42.58 for the Broader Awareness construct. There was no significant difference in the scores for each construct among the three disciplines (all p>0.05). The order of mean construct score was consistent across all disciplines, with Objective > Disciplinary Grounding > Integration > Broader Awareness. The total IDSR scores also did not differ significantly by discipline (χ²[2] = 3.79, p = 0.151), but there was a trend for the biology scores to be higher than the engineering scores (multiple comparison p = 0.052 with Bonferroni correction requiring p < 0.017 for significance).

TABLE 3.

Student IDSR scores (means with SDs in parentheses) across all three disciplines

	Biology (N=15)	Engineering (N=11)	Geology (N=10)
	IDSR Rubric (Mean Score (SD))
Objective	2.5 (0.58)	2.4 (0.47)	2.5 (0.39)
Disciplinary Grounding	2.5 (0.47)	2.2 (0.83)	2.0 (1.01)
Integration	1.6 (0.92)	1.0 (0.81)	1.4(1.27)
Broader Impacts	1.2 (0.88)	0.7 (0.61)	1.3 (0.92)
Overall Mean	2.0 (0.51)	1.6 (0.37)	1.7 (0.64)

Note: N indicates the sample size.

SNA

Social networks were created based on SNA survey questions, forming trust and learn networks. The trust item was on both the pre- and post-CURE surveys, and the data from both networks are presented (Figure 3). However, the learn question was only asked on the post-survey as the pre-surveys were administered at the start of the first day of class before students interacting as classmates, so there is only one network for that survey item (Figure 4). After network formation, the whole-network density was calculated as well as E–I indexes (Table 4). E–I indexes are presented as both whole-network values and by discipline to assess if relationships formed differentially between the classes.

FIGURE 3.

Trust social networks in which nodes (representing student teams) are shaped and colored by discipline; biology is a blue triangle, engineering is a red circle, and geology is a black square. Tie thickness indicates the frequency of connections between the two teams, with a thicker tie indicating more connections between the two teams. (A) Network formed from the presurveys given at the beginning of the courses. (B) Network formed from the postsurveys administered at the end of each course.

FIGURE 4.

Learn social network in which the nodes are shaped and colored by discipline, biology is a blue triangle, engineering is a red circle, and geology is a black square. Tie thickness indicates the frequency of connections between the two teams, with a thicker tie indicating more connections between the two teams. Data on whom someone learned from were only collected on the postsurveys administered at the end of each course.

TABLE 4.

Social network measures

	Whole-network measures		Discipline-specific E–I indexes
Network	Density	E–I Index	Biology (N=4)	Engineering (N=3)	Geology (N=2)
Pre-Trust	0.250	−0.66	−0.80	−0.79	0
Post-Trust	0.444	−0.18	0.02	−0.71	−0.25
Learn	0.597	0.37	0.43	0.55	−0.30

Note: Whole-network density and mean valued E–I indices for the whole network and by discipline. N indicates sample size (number of student teams).

Whole-network densities were calculated to reflect how many connections existed between teams in all three linked CUREs compared with the maximum possible number of connections. In the trust network, overall density nearly doubled between the pre-and post-CURE networks (0.250–0.444; Table 4). The learn network demonstrated the greatest density of connections between teams, followed by the post-survey trust network, and then the pre-survey trust network (Table 4).

The E–I index of –0.66 for the pre-trust network indicates a high degree of homophily within the disciplines, with biology (–0.80) and engineering (–0.79) having a greater preference for internal (within discipline) ties and geology (0) demonstrating an equal number of internal and external connections. A permutation test (p < 0.001) demonstrated that the overall pre-trust network exhibited a greater level of homophily than would be predicted by chance. In the overall post-trust network, the disciplines became less homophilous (–0.18), with biology indicating a greater number of interdisciplinary connections (0.02) while trust ties remained predominately internal within engineering (–0.71) and geology (–0.25) classes. However, the level of homophily was still significantly greater than would otherwise have been predicted by chance (p < 0.001).

The whole-network E–I index for learn network (0.37) (post-only) indicates that students across all of the classes had a greater number of learning ties to individuals who were outside of their discipline than within their discipline. The discipline-specific learn network E–I indexes exhibit ties both within and across disciplines. The E–I indices for biology (0.43) and engineering (0.55) indicate that these disciplines had a greater proportion of external ties associated with who the students in these disciplines learned from. However, geology teams learned from other teams primarily within their discipline (–0.30). A permutation test on the learn network was nonsignificant (p = 0.244) and did not find that there was a strong structural preference for interdiscipline or intradiscipline connections.

DISCUSSION

This project intentionally introduced students to team science, including the importance of interdisciplinary science. The team science tools that were integrated into the class (i.e., the communication and research plans) introduced students to best practices from the team science literature and provided structure for collaboration and communication, within and between teams. Analysis of student products and SNA provides evidence of communication between teams in the linked ID-CUREs. The student writing prompts reflect that students are internalizing and valuing team science principles, even if they are not always experts in the way in which they apply those principles.

Student Thinking About the Importance of Interdisciplinary Science

The writing prompt scored with the IDSR rubric (Tripp and Shortlidge, 2020) resulted in similar overall scores for students in each discipline. Specifically, students exhibited similar levels of 1) thinking about the objectives associated with having multiple disciplines work together, 2) grounding in their own discipline, 3) the ability to integrate across disciplines, and 4) the ability to communicate the broader impacts of their work. Students received similar levels of team science instruction in each of the three linked CUREs that is reflected by their statistically similar writing prompt scores.

Students performed best on objective and disciplinary grounding subscores. That pattern of results is similar to a preliminary evaluation of writing prompt data in an early (COVID era) iteration of the ID-CUREs (Vance-Chalcraft et al., 2023). However, integration was not the lowest score here as it was previously. This change is perhaps due to an increased level of in-person collaboration between students during this study compared with the prior study that occurred during a time in which there were still some COVID-19 restrictions. Instead, student rubric scores being ordered Objective > Disciplinary Grounding > Integration > Broader Awareness is in line with the perceived difficulty of each concept presented by the rubric designers (Tripp et al., 2020; Tripp and Shortlidge, 2020) and the literature surrounding the implementation of these concepts into STEM courses (Spelt et al., 2015).

We could not find other published papers with IDSR scores to compare with our results, outside of the original Tripp and Shortlidge (2020) implementation. Our student scores were lower overall than those reported by the original authors. Without data from additional institutions or more detailed results from the original study (e.g., by criterion rather than one overall score), it is difficult to hypothesize why our scores are lower. We encourage others to use the writing prompts and IDSR to evaluate interdisciplinary learning so we can better understand whether differences in our student population, our course implementation, or our scoring resulted in lower total scores on the IDSR.

Mean IDSR scores corresponded to a novice level of student thinking in the Integration and Broader Awareness categories, indicating that the students’ responses did not include the characteristics of responses that would be expected from an expert or even an intermediate in the field. As students scored more highly on the Objective and Disciplinary Grounding constructs, they appeared to understand the purpose and reasoning behind including multiple disciplines, but struggled with the how and why when presented with a situation different from their research project. The Integration and Broader Awareness categories require sophisticated thinking about how other disciplines should integrate and collaborate together. Although the students integrated information across disciplines and collaborated across disciplines in these ID-CUREs, they may have struggled to see how to translate those types of interactions in a nonclass setting. Including more examples of other complex projects and how interdisciplinary teams are formed and successful in the real world may help students advance in these areas (Lyall et al., 2015).

Student Formation of Interdisciplinary Connections

SNA showed an increase in connections between teams, within and across disciplines, over the semester. Students developed an increased number of connections between disciplines, indicating that though students may not have had day-to-day interactions with people in the other disciplines of the ID-CUREs, they were still valuing the information and perspectives the other disciplines were bringing to the overall project.

SNA revealed evidence of interdisciplinary connections in the trust network and the learn networks, with the most interdisciplinary connections being in the learn network. The discipline-specific E–I indexes from the learn network provide further evidence of communication patterns seen within the student product analysis, with the biology and engineering courses primarily receiving information that was sourced from geology. Biology and engineering teams were more likely to indicate they trusted and learned from other disciplines than were geology students at the end of the semester. Therefore, it may be necessary for students to use data and information from other disciplines in order to develop trust and feel like they are learning from other disciplines. Being the source of data may not be sufficient.

The learn network had the highest density and the most exhibited cases of interdisciplinary connections. Team-based literature suggests that instances of learning from coworkers are the most common type of workplace interaction (Herkenhoff et al., 2024). Additionally, the trust network exhibited a slight reduction in homophily (tendency to interact with one's own discipline) and increased interdisciplinary connections over the course of the semester. Although learning interactions can take place in one instance, trust is built over time (Costa et al., 2017). Longer-term research would be needed to determine whether network density would continue to increase and whether more interdisciplinary connections would form between teams in ID-CUREs that lasted multiple semesters. A logical hypothesis would be that the density of learn networks would plateau sooner than the density of trust networks.

LIMITATIONS

The three ID-CUREs in this study had a total of 36 participants in a total of nine teams. More participants, particularly more teams in each individual discipline, may help reveal statistical differences between the disciplines. In addition, we were not able to look at how student identity influences a student's placement within a network or their interactions within and across teams due to small sample sizes within any one identity group. We did not analyze data from a comparison group because the ID-CUREs are distinct from other courses, making comparisons challenging. These three courses were each taught in only one section per year, and previous iterations of these courses did not collect the same types of social network data. We focused our analyses on how the three disciplines compared with each other and how team responses changed from the beginning to the end of the semester.

Also, the pre-SNA survey did not contain a question about who the students had learned from in prior experiences. Therefore, our analysis of the learn network is limited to who the students felt they had learned from in this particular ID-CURE course context. We cannot determine if the students who were most tightly connected in the post-learn network had pre-existing learning ties before entering these ID-CUREs.

An additional limitation is that it is difficult to determine the validity of individual student responses in SNA. Recommendations in the literature are to pair SNA data with other data sources (Wald, 2014) and to use data collection methods that minimize the likelihood that outliers skew the study (Längler et al., 2019). These suggestions were followed in this study by our use of multiple data types.

IMPLICATIONS AND CONCLUSION

Although most scientists agree that solving our world's most pressing problems requires interdisciplinary collaboration, few undergraduate programs have found ways to break out of disciplinary silos and train their students how to work effectively across disciplines. Our novel approach of combining the evidence-based lessons from team science and the high-impact practice of undergraduate research experiences provides one avenue for increasing interdisciplinary training for undergraduate students. Also, by using the IDSR to evaluate the level of interdisciplinary thinking in our students, we provide a rare example of the implementation of this validated and well-constructed rubric.

The results of this study suggest that including explicit conversation about the importance of interdisciplinary expertise in courses led to these students exhibiting a level of thinking about the importance of forming connections across disciplines that was just below expert level. When asked to apply these principles to a novel situation, however, students struggled with identifying real-world opportunities and broader awareness of these connections. We did not find significant disciplinary differences between the students’ thinking about and implementation of connections across disciplines, despite each course having differing levels of direct communication. Geology students perceived themselves as having learned less from other disciplines, based on their role mainly as sources of data to be shared rather than the recipients of shared data. Implementing discussions of team science and the value of interdisciplinary interactions into courses, as well as having students use data from other disciplines, may help prepare students for the real-world problems STEM professionals face and better train undergraduates for future careers.

The structure of these ID-CUREs is unique and may be difficult to implement at some institutions. However, instructors interested in incorporating similar ideas into their course design can have students share data from previous semesters of other courses or include real-world examples within their course materials. Simply acknowledging the interdisciplinary nature of STEM fields may be beneficial in breaking down the siloing that commonly occurs within higher education. The team science tools here can be used to enhance interdisciplinary thinking, but can also be beneficial for creating more productive teams even within a single discipline. Also, the use of writing prompts and the IDSR can help instructors evaluate the effectiveness of efforts to add interdisciplinary components to their courses.

This study provides an example of the use of SNA to analyze interactions amongst student teams in a CURE. The E–I index provides an additional means of examining interactions within and between teams or disciplines. Future work could continue to explore the use of SNA in education research (Grunspan et al., 2014), particularly involving student interactions with others in group or team-based settings. Additionally, the analysis in this study focused primarily on teams as a whole; shifting the perspective to individual students and/or TAs and instructors would likely provide additional insight into the mechanisms of team interactions.