Modeling Feedback Dynamics in Learning Studios: A Causal-Loop Analysis of an Undergraduate Mechanical Engineering Curriculum

Research Paper Recommended citation: Alrizqi, M., & Godwin, A. (2025). Modeling Feedback Dynamics in Learning Studios: A Causal-Loop Analysis of an Undergraduate Mechanical Engineering Curriculum. In Kangaslampi, R., Langie, G., Järvinen, H.-M., & Nagy, B. (Eds.), SEFI 53rd Annual Conference. European Society for Engineering Education (SEFI), Tampere, Finland. DOI: 10.5281/zenodo.17631226. This Conference Paper is brought to you for open access by the 53rd Annual Conference of the European Society for Engineering Education (SEFI) at Tampere University in Tampere, Finland. This work is licensed under a Creative Commons Attribution-NonCommercial-Share Alike 4.0 International License. Modeling Feedback Dynamics in Learning Studios: A Causal-Loop Analysis of an Undergraduate Mechanical Engineering Curriculum Mohammed A. Alrizqi a,1, Allison Godwin b a Sibley School of Mechanical and Aerospace Engineering, Cornell University, Ithaca, United States, 0000-0001-6034-8314 b Robert Frederick Smith School of Chemical and Biomolecular Engineering, Cornell University, Ithaca, United States, 0000-0002-0741-3356 Keywords: Systems Thinking, Causal Loop Diagrams, Experiential Learning, Feedback Loops, Collaborative Learning ABSTRACT This study adopts a systems perspective to examine how a newly introduced “Learning Studios” (LS) model in a mechanical and aerospace engineering (MAE) school influences undergraduate engineering experiences. LS are cross-course labs that engage multiple concepts and learning objectives, embedded throughout all three years of the curriculum. We conducted semi-structured interviews with eight mechanical engineering students at a U.S. R1 institution to capture their perceptions of hands-on, real-world engineering tasks. Drawing on the Expectancy-Value-Cost framework, we developed our qualitative study and then employed inductive thematic analysis to identify key factors, including motivation, personal growth, belonging, and engineering identity. Subsequently, we developed causal loop diagrams (CLDs) to visualize how these factors interact and shape feedback loops in students’ academic and social environments. Findings indicate that LS experiences foster reinforcing cycles: mastery of practical tasks boosts self-confidence, which, in turn, enhances engagement and strengthens peer networks. Academic and social integration also foster a sense of belonging that further sustains motivation and reduces the risk of attrition. However, the diagrams reveal balancing loops when academic challenges strain students’ resilience. Mapping these dynamics underscores the value of a holistic systems approach in engineering education. By illuminating both positive feedback loops and potential bottlenecks, our study provides actionable insights for designing interventions that support students’ holistic development, from technical competence to the formation of a professional identity. 1 Corresponding Author M. A. Alrizqi [email protected] 1 INTRODUCTION 1.1 Background Engineering education operates within a dynamic system where multiple interconnected factors shape student success. A system is defined as “a group of interacting, interrelated, or interdependent components that form a complex and unified whole."(Anderson & Johnson, 1997, p. 2). Considering these characteristics, engineering education can be viewed as a complex system (Monat et al., 2022). Despite this understanding, traditional research has often approached engineering education (EE) in isolation (Sigahi & Sznelwar, 2022). A systems approach is essential as it allows for a more holistic and integrated understanding of complexities and interconnected levels (Walters & Litchfield, 2015). As a result, there has been increasing interest in the past several years for a systems approach to tackle the challenges of complex systems in engineering education research (EER). For example, Sigahi et al. (2022) explained how complexity theory and systems thinking can transform EER by providing frameworks to address interdisciplinary challenges and interdependent factors in engineering curricula (Sigahi & Sznelwar, 2022). Dhukaram et al., (2016) argued that higher education interacts with a complex ecosystem of political, economic, technological, and organizational factors, necessitating systems thinking for modeling and policy analysis. This understanding is key in education. Arnold (2018) argued that academic and social factors in college experiences are closely linked and should be addressed together for effective enhancement. Failing to address feedback on these aspects can lead to ineffective interventions. Furthermore, when universities implement changes like new curricula without considering such system dynamics, they may encounter limited success. To address these challenges, a few researchers have applied a systems dynamics approach—specifically, causal loop diagrams (CLDs)—in EER. CLDs, which function similarly to concept maps, visually depict how a change in one factor, like motivation, can influence another, like interest in a major (Sterman, 2000). Once a detailed causal loop diagram is created, systemic issues can be modeled and strategies for change identified (Vanasupa et al., 2008). CLDs allow researchers to visualize key variables and their relationships by linking them with arrows that indicate causality and designating loops as either reinforcing or balancing (Zhang et al., 2012). This visualization helps uncover hidden feedback structures, offering a more holistic perspective on complex educational challenges (Dhukaram et al., 2016). For example, a student’s identity can affect their motivation, which in turn influences their engagement and performance, and this feedback can impact their sense of belonging (Godwin et al., 2016; Godwin & Kirn, 2020). As discussed by several researchers, CLDs facilitate the visualization of how feedback loops can generate complex behaviors in social systems (Anderson & Johnson, 1997; Walters & Litchfield, 2015). These behaviors include self-reinforcing improvements in virtuous cycles (e.g., peer collaboration ➔ improved problem-solving ➔ stronger collaboration) or balancing mechanisms (e.g., academic pressure ➔ stress ➔ reduced performance). For education stakeholders, mapping these relationships between factors can highlight new areas for research and complicate the current EER landscape, thereby facilitating effective change. Although still emerging, CLDs and system mapping are gaining traction in EER. For example, one study demonstrated how CLDs capture interdependencies in socialecological systems, reinforcing their value in modeling education dynamics (Gray et al., 2019). Another study applied CLDs to faculty motivation, finding 13 feedback loops that affect teaching innovation and noting that many change initiatives fail due to ignored system-wide interconnections (Cruz-Bohorquez et al., 2024). These studies illustrate how CLDs offer a holistic lens on educational complexity, guiding more effective interventions. Despite these advancements, a significant gap remains in the holistic modeling of the undergraduates' experience using a systems approach. In this study, we bridge the gap by evaluating engineering students' experiences in a newly designed lab, LS, through a dynamic system approach, CLDs. Our aim in this paper is to discuss how we used CLDs to explore student experiences and offer this approach as a potential way to consider complex, interrelated data in engineering education. 1.2 Learning Studios Model In the Fall of 2022, the mechanical and aerospace engineering (MAE) department at a U.S. Northeastern R1 University introduced Learning Studios (LS) to enhance hands-on, real-world learning across multiple courses. Unlike traditional projectbased learning, LS integrates fully operational engineering systems (e.g., combustion engines, drones, forklifts) into the core curriculum, ensuring all students gain progressive, hands-on experience rather than limiting it to electives. These LS are spread across the entire three years of the MAE curriculum and integrate crosscutting concepts across courses in these applications. This approach is unique— most programs that integrate hands-on labs do so with a single course or system. The LS approach bridges theory and practice using four key components: real-world engineering systems, advanced analysis tools, simplified models, and structured learning modules. 1.3 Objective and Research Questions In this paper, we use a pilot of student interview data translated into CLDs to capture the multifaceted impact of LS on students’ experiences. This work can enhance our understanding of how and why interventions like LS are beneficial, providing a system-level study that complements traditional educational assessments. Specifically, we ask two questions: 1. What are the causal links between student engagement with LS, shifts in selfperception, and community feelings, and how do these affect their learning journey? 2. What are the main driving factors for the main themes emerging from the study? 2 METHODOLOGY Our research employs an interpretive, qualitative research design combined with a systems-thinking analytical approach. 2.1 Data Source The interview protocol utilized the Expectancy-Value-Cost (EVC) model (Alrizqi et al., 2025; Barron & Hulleman, 2015). This framework suggests that students' choices in achievement are shaped by their expectations for success and the personal value they associate with tasks, encompassing intrinsic, attainment, and utility values, as well as associated costs. The semi-structured interview protocol was crafted in adherence to recognized qualitative research methodologies (Alrizqi et al., 2025; Borrego et al., 2009; Cridland et al., 2015; Kallio et al., 2016). Such interviews offer benefits by having a balance between structured questions and the flexibility to dive into newly emerging topics (Kallio et al., 2016). This approach also mirrors the recommendations of a previous study, which found that interviews are an effective way to elicit rich qualitative data for systems dynamics models (Luna-Reyes & Andersen, 2003). Also, this was employed in similar system dynamics studies (e.g., El Halabi et al., 2012) to gather insights on key variables and relationships. 2.2 Participant Selection and Sampling Strategy A purposive sampling method was used to select participants with firsthand experience in LS. Selection criteria included enrolment in the MAE degree program, engagement in at least one LS, and stratification by class standing, gender, and race to ensure diverse perspectives. Eight mechanical engineering students met these criteria and voluntarily participated; this sample size is appropriate for a qualitative preliminary exploration. Appendix 1 presents the demographics of the participants. Participants chose their pseudonyms; however, if they did not provide one, the researcher assigned one to them. 2.3 Data Collection and Analysis We piloted the interview protocol with one participant to ensure data quality and make refinements based on feedback. Each semi-structured interview was then conducted and lasted approximately 45 minutes. Each interview took place in a private setting to ensure confidentiality and was audio-recorded with the participant's consent. The Institutional Review Board (IRB) approved the study, ensuring compliance with ethical research standards (Kallio et al., 2016). Furthermore, an inductive thematic analysis was conducted following Braun and Clarke's Braun & Clarke (2006) six-phase framework. This approach was chosen because it allows themes to emerge directly from the data, which is essential when exploring under-researched areas (Patton, 2014; Saldaña, 2013). These six steps are (1) Familiarization with the transcripts, (2) generating initial codes (using MAXQDA software), (3) searching for potential themes, (4) refining themes to ensure accuracy and to ensure coder consistency, (5) defining and naming themes, and (6) organizing into a coherent narrative. 2.4 Causal Loop Diagram Development After we coded the data, we built the CLDs. This step involved transforming the qualitative data into a visual systems map illustrating feedback structures. 2.4.1 Identifying Variables First, we identified the main variables that would appear as nodes in the CLDs. These corresponded closely to the themes, subthemes, and factors identified through the thematic analysis. For example: • A theme like “Belonging and Supportive Community” is represented as a variable “Sense of Belonging.” • A theme like “Enhanced Motivation and Confidence” was split into two related variables: “Motivation” and “Self-Confidence,” since those appeared as distinct concepts in the data but were part of one theme. 2.4.2 Mapping Causal Links Next, we revisited the interview data to map out causal relationships between these variables. We followed the approach suggested by Newberry and Carhart (2023), which involved a “data source reference table" for traceability (Newberry & Carhart, 2023, p. 12). For instance, Sarah said, “we would just talk about… courses we want to take in the future, or… just give advice to each other. This made me feel more belong[ing] to mechanical.” From such a statement, we can derive a causal link: Peer Support → Sense of Belonging→ Mechanical Engineer Identity. From the literature on engineering identity, we connected Mechanical Engineering Identity → Retention (Godwin & Kirn, 2020). We documented each such link and noted its polarity (positive or negative influence). A positive link (“+”) means that as one variable increases/decreases, the other tends to increase/decrease. A negative link (“–”) means an inverse relationship where an increase in one leads to a decrease in the other, and vice versa. In the example above, peer support had a positive impact on belonging, which in turn had a positive effect on identity and retention. 2.4.3 Constructing the Diagram Once we had a list of variables and their pairwise causal links, we started constructing CLDs using Vensim™. We first clustered related variables, often corresponding to the thematic groups from the analysis. To enhance clarity, we segmented our overall model into three sub-diagrams, similar to the method used by Halabi et al. (2012) (El Halabi et al., 2012). 3 RESULTS 3.1 Overview of Themes Our findings highlight three key themes where LS positively impacted students’ experiences: (1) Motivation, (2) Personal Growth, (3) Sense of Belonging, and (4) Identity as Mechanical Engineers. Participants reported personal development through their hands-on experiences. Interviews revealed that mastering courses fostered a sense of belonging in mechanical engineering, while struggles led to alienation. Moreover, students honed their skills and experienced personal growth and community by engaging with real-world applications. Findings show that LS was vital in building student confidence, helping them overcome feelings of inadequacy, and enhancing their satisfaction with learning. Recognizing these impacts is essential for creating effective educational environments and practices that enhance collaborative and hands-on learning, which are critical for developing engineering skills and fostering professional growth. 3.2 Causal Loop Diagrams (CLDs) In our case, the three interconnected sub-CLDs were: (1) Learning and Personal Growth, (2) Social Dynamics and Community, and (3) Motivation and Career Aspirations. Subsetting the models made it simpler to discuss specific loops. Each sub-CLD captured a subset of variables and their corresponding feedback loops, which pertained to a specific facet of the student experience. 3.2.1 Experiential Learning and Self-Reinforcement (Figure 1) This CLD illustrates a reinforcement loop that examines how hands-on experience leads to a self-reinforcing cycle. It demonstrates that practical experiences and a sense of competence have a significant influence on the learning process, leading to a deeper understanding and increased engagement. For example, Charlle explained her experience by saying, “It was fun to tear down an engine and look at all the parts. Definitely helped me understand how engine worked a lot better,” showing how direct manipulation of a physical system can bridge abstract theory and practical application. She continued, " Now I feel comfortable working in a lab…I can tackle anything,” indicating that iterative successes bolster self-efficacy. Another student, Zaher, described how applied tasks sparked intrinsic motivation, Self-Determination Theory’s key element (Deci & Ryan, 1985). He said, “Comparing theory to real Otto cycle data was fascinating.” These insights show that LS encouraged repeated hands-on successes that cultivate student success and drive further engagement with complex engineering challenges The participants' descriptions illustrate Kolb’s (Kolb et al., 2001)Experiential Learning and Bandura’s Self-Efficacy (Bandura, 1978) theories. Specifically, it aligns with Kolb’s four-stage learning cycle (concrete experience, reflective observation, abstract conceptualization, and active experimentation) by demonstrating how continuous, hands-on practice reinforces learning through reflection and theorybuilding. Fig 1. This reinforcing loop (R1) illustrates how hands-on tasks contribute to deeper conceptual understanding, which enhances confidence and engagement. This, in turn, ultimately increases motivation and enjoyment in the learning process ( R2), reinforcing students' sense of competence. 3.2.2 Resilience Through Challenge-Support Dynamics (Figure 2) The second CLD examines how academic challenges interact with peer relationships and institutional support to foster resilience. It illustrates how peer support can help individuals overcome challenges, thereby enhancing their resilience and, in turn, their personal growth. This aligns with Tinto (1975) emphasis on academic integration—while normalizing struggle in line with Dweck (2006) growth mindset, thus motivating sustained engagement in rigorous engineering contexts. For example, students’ struggles—such as Zaher’s comment, “Statics made me rethink my path”—reflect Tinto’s emphasis on academic integration as essential for retention. Learning communities’ collaborative environments help counteract possible disengagement by fostering social integration. As Zaher observed, “Everyone cheers each other on—it’s not cutthroat,” exemplifying how normalized struggle and peer support reframe challenges as learning opportunities, aligning with Dweck’s growth mindset framework. These findings highlight how well-structured lab spaces such as LS can mitigate the risk of attrition by integrating challenge and support. 3.2.3 Collaborative Identity and Belonging (Figure 3) This CLD centers on how teamwork in LS fosters a professional identity and sense of belonging. It also emphasizes the social component of the learning environment, such as Collaborative projects, in shaping the mechanical engineering identity. Group tasks, such as troubleshooting fluid systems, reflect legitimate peripheral participation, where learners gain competence through shared practice. Sarah’s statement, “Helping others with their projects made me feel like part of the team,” and Charlle’s reflection, “Seeing peers succeed made me proud to be aMechE,” both underscore how shared practice and mutual support foster mechanical engineering identity. It also demonstrates how recognition by others reinforces an individual's engineering identity (Godwin et al., 2016). Exclusion from teams, however, can introduce a balancing loop, as noted by Kwami: “I was discouraged when I didn’t get into a project team.” This highlights the importance of fostering institutional belonging to prevent potential disengagement. However, Kwami’s eventual resilience, “That definitely reinforced my positive attitude Fig 2. exhibits how academic challenges interact with peer/instructor support in a balancing loop (B) that mitigates reduced belonging. A reinforcing loop (R1) shows how overcoming challenges builds resilience, leading to personal growth and student persistence (R2), rei nforcing engagement towards the mechanical engineering,” illustrates how situated learning experiences promote the co-construction of engineering identity. Overall, these observed dynamics align with theories such as Communities of Practice (Lave & Wenger, 1991) Vygotsky’s Social Constructivism (VYGOTSKY, 1978), and Social Identity Theory (Tajfel & Turner, 1979), which suggests that LS serve as communities of practice that continually reinforce a sense of membership, shared purpose, and professional pride. 4 DISCUSSION Across the three models, several systemic insights become apparent. Two of those CLDs' loops are reinforcing, indicating that LS can create virtuous cycles—positive feedback loops in which each gain fuels the next. Students frequently describe upward trends as starting out uncertain, then gaining skills, forming friendships, growing in confidence, and ending highly motivated. This result does not mean the process is universal or automatic; balancing loops (e.g., academic challenges) can impede the positive cycles. However, the fact that we identified common reinforcing loops suggests that well-designed interventions can push a student’s experience into a positive, self-sustaining trajectory—what systems theory calls an “attractor state,” a stable pattern that the system naturally tends to maintain. Our findings align strongly with established educational theories and prior research. For instance, the importance of community and belonging in driving engagement aligns with Tinto’s model of student retention, as demonstrated in Model 2 (section 3.2.2). Recent studies (Smith et al., 2021; Walton & Cohen, 2011) Additionally, interventions targeting a sense of belonging can enhance academic outcomes. Our qualitative data provide a causal narrative for how these interventions may work (belonging → engagement → motivation → achievement). Similarly, the role of hands-on learning in boosting motivation resonates with active learning literature (Freeman et al., 2014). Our CLDs outline the mechanism; hands-on successes lead to enjoyment and motivation, which in turn lead to deeper engagement. The social dimension is also captured, revealing how peer support fuels a sense of belonging, which in turn reinforces engagement and enhances the ability to face academic challenges. A key methodological advantage of CLD-based qualitative analysis is its ability to make indirect causal pathways and temporal dynamics explicit. In our model, the primary intervention—hands-on studio experience—has a direct, positive impact on Fig 3.This reinforcing loop (R) shows how collaborative projects enhance peer recognition, strengthening engineering identity, belonging, and motivation, which drives further engagement.