Using Reflexive Principlism to Investigate Longitudinal Changes in Engineering Students’ Perceptions of Ethics

Recruitment and selection of interview participants

In the fall of 2015, we recruited 757 first-year engineering students to complete an initial survey for our study. These participants came from four US universities (Arizona State University, Brigham Young University, the Colorado School of Mines, and Purdue University), representing diverse geographic locations and university types (including both public and private, religious and non-religious, and research-intensive and primarily undergraduate institutions). This first survey sample was gender and ethnically diverse. Special efforts were taken to recruit students who were both likely and not likely to receive enhanced exposure to considerations of social and ethical responsibility during their degree programs. All necessary human subjects research approvals were obtained from our institutions prior to the research. Students indicated their gender on a demographic question included with the survey; these same genders are reflected in the pronouns used to refer to participants in this paper. Additionally, our research team selected pseudonyms for all student participants.

In the spring of 2015, we recruited 113 participants from our survey sample to participate in a semi-structured interview during their first year of university studies (T1). We purposefully sampled interviewees to include students whose survey responses indicated prior experience with, or future intent to participate in, various activities and experiences that could potentially impact how they perceive ethics and social responsibility, for example, service learning, volunteering, honors programs, formal religious instruction, internships, and other work experiences. We also sought to stratify our sample to reflect demographic diversity, especially in terms of engineering major and gender. We followed up with these initial interviewees in the fourth year of their studies (Spring 2019), and 33 of these students participated in a second interview (T2). These pairs of longitudinal interviews form the data set discussed in this paper.

Among the 33 longitudinal participants, 8 were from Brigham Young University, 12 were from the Colorado School of Mines, and 13 were from Purdue University. No interview data was included from Arizona State University due to challenges recruiting prior interviewees from this university in the fourth year of the project. In terms of gender, 20 of the interviewees selected the “male” option and 13 selected the “female” option at the time of the second interview. Regarding race/ethnicity, 27 participants identified as white, five as Asian, and one as mixed race. The interviewees ranged in age from 21 to 24 in the T2 interview. One participant had recently graduated at the time of their second interview, and all others expected to graduate in the current or next calendar year. The participants were in various engineering degree programs, including: agricultural/biological (1 student), biomedical (3), chemical/biochemical (2), civil (3), electrical/computer (5), information technology/computer science (2), materials/metallurgical (2), mechanical (11), and manufacturing/industrial/systems (4).

Interview procedures

We conducted pairs of interviews with all 33 participants. The two interviews took place three years apart (in Spring 2016 and Spring 2019). Each interview was 45–60 minutes long, conducted face-to-face and audio recorded. Participants received $20 gift cards for completing the first interview and another $20 gift card for completing the second interview. Following the completion of each interview, the audio recordings were transcribed by a third-party transcription service. The transcripts were then reviewed for accuracy and anonymized by the research team. Pseudonyms were assigned to each participant following the first interview and used again in the second interview transcription.

The semi-structured interview protocols used in both interviews included questions related to 1) general definitions of ethics and engineering ethics, 2) experiences (past, present, and future) that may shape students’ ethical perspectives and sense of social responsibility, 3) ethical climate, 4) ethical scenarios, and 5) select survey items identified as promising for further probing. The first and second interview protocols were very similar, though the questions used in the second interview were reworded to account for the fact that the participants were fourth year engineering students by the time the second interviews took place. For further reference, the complete T2 interview protocol is presented in Howland, Kim, et al. (2022).

Exploratory data analysis

The research approach and findings presented in this paper grew out of earlier efforts to use a narrative case study approach to understand how one’s student’s views of ethics and social responsibility changed over time (Claussen et al., 2021). While the narrative case study served as an effective way to explore the changes in a single participant, it proved difficult to scale and apply to all 33 participants due to the diversity of perspectives across the entire pool and the feasibility of drafting and presenting large numbers of narratives. Thus, we needed a method that could illuminate more nuanced changes in each participant (which our quantitative analysis was largely unable to do, as noted above) while avoiding the risk that too much focus on individual cases might obscure patterns shared among multiple participants.

To better understand the changes across time and across participants, we worked as a team of four researchers (all co-authors on this paper). Two researchers were involved with the project from its inception, and had conducted interviews (both T1 and T2) and participated in the data analysis throughout the project. A third researcher was involved since the midpoint of the project, and had conducted a portion of the second interviews and contributed to many aspects of the data analysis. The fourth researcher joined the project after all interviews were completed and data analysis was underway. Thus, the majority of us were very familiar with the interview transcripts, having read many of them in full for previous analysis efforts. Finally, a fifth co-author who was involved with this project from the beginning joined our team meetings in the later stages of the analysis, serving as a further check on the quality of our findings.

As mentioned previously, we initially used narrative analysis to understand the longitudinal changes in one participant’s views (Claussen et al., 2021). To expand our analysis to all 33 interview pairs, we divided up the transcript pairs among the four researchers primarily involved in the data analysis. Given the limitations of frameworks used in previous studies, we adopted an inductive approach, reading through the pairs of transcripts in full and identifying themes that emerged to describe the changes we were seeing across time. Thus, our unit of analysis was the 33 transcript pairs (rather than 66 individual transcripts). This allowed us to be attentive to changes over time while also accounting for how individual participants talked about engineering ethics in each of their interviews.

Our analysis was initially iterative and inductive: after each of us read through a set of transcript pairs, we created a draft set of themes. After comparing students’ T1 and T2 responses from both interviews and an initial round of open coding on a subset of interviews, there were a few common themes but also enough differences that it was difficult to synthesize results. We realized that the greatest changes we were observing in the data centered on how the participants talked about engineering ethics and social responsibility. We collectively observed an increase in the specificity with which they applied ethical principles to their professional and academic contexts, and an increase in their attention to balancing between multiple and sometimes conflicting principles. In addition, we saw changes in the principles that were mentioned in interviews, both in terms of which specific ethical principles were discussed and the frequency with which they were talked about. These observations in turn led us to adopt the reflexive principlism framework (Beever & Brightman, 2016), as described in more detail below.

Use of reflexive principlism in this study

Reflexive principlism is described by Beever and Brightman (2016) as “an approach to ethical decision making that focuses on internalizing a reflective and iterative process of specification, balancing, and justification of four core ethical principles in the context of specific cases” (p. 275). The four principles, nonmaleficence, beneficence, autonomy, and justice, are derived from an interpretation of previous work in biomedical ethics by Beauchamp and Childress (2009). Reflexive principlism posits that through the ethical reasoning processes of specification, balancing, and justification, the relative value of each of these ethical principles is determined, which allows the individual to weigh and prioritize these four ethical principles according to the specifics of a situation (Beever & Brightman, 2016). The goal is for ethical reasoning to move from “a reflective, conscious, process” to one that is “internalized and applied in a reflexive, ‘second-nature’ way to unique situations” (p. 282).

In previous research, the reflexive principlism framework has been presented as a course-level intervention (Hess et al., 2019), analytical lens for student interviews in the context of an engineering design course (Corple et al., 2020), and framework for analyzing student responses to a case-based assessment exercise (Hess et al., 2014). Whereas Corple et al. (2020) used reflexive principlism as a framework to examine how students engaged with ethical decision-making throughout the design of a community-based project, we used reflexive principlism to examine how students’ engagement with engineering ethics more generally changed over time. Because our interview protocol was not developed with reflexive principlism in mind, we used the open-ended questions about engineering ethics from our interviews to identify illustrative examples of the principles and students’ ethical reasoning using the following definitions presented by Beever and Brightman (2016):

• Nonmaleficence: Avoiding causing harm

• Beneficence: Providing benefits to society or others

• Autonomy: Respecting the agency of individuals in decision making

• Justice: Distributing risks, benefits, and costs equitably among all individuals.

In addition to these four principles, Beever and Brightman’s framework (2016) includes two ethical reasoning mechanisms or processes that were relevant in our data: balancing and specification. The authors define balancing as, “adjudicating conflicts among principles in the context of a particular case,” (p. 282) and specification as, “reducing the indeterminateness of general norms to give them increased action guiding capacity, while retaining the moral commitments in the original norm” (p. 282). Finally, though less relevant in our data set, Beever and Brightman include a third process, justification, which they defined as “the process by which the coherence and completeness of an ethical-reasoning decision is evaluated… Both a general method of coherence and also a pedagogical stage in the process of decision-making” (p. 283). It is through these ethical reasoning mechanisms that the value of each principle is determined because the framework does not place the principles in a hierarchy. Reflexive principlism allows the individual to give weight to the principles according to the specifics of a situation.

Deductive data analysis

We next undertook further analysis of our data using a more deductive approach, while also being mindful of how reflexive principlism might constrain our analysis and findings. We returned to the pairs of interview transcripts, working to identify whether aspects of the reflexive principlism framework were present, absent, or emerging in each of the transcripts (Table A-1 of the appendix). In addition, we looked for illustrative examples of each aspect to see how it described (or failed to describe) the changes in our participants over time. The emerging designation was used when a student’s response touched on one of the principles or aspects of ethical reasoning present in reflexive principlism, but did not reach a level of articulation and clarity which would allow us to definitively claim that these aspects were present in the students’ responses. For example, the participant Parson vaguely referenced the principle of beneficence in his T1 interview when he described his motivation to study biomedical engineering: “I wanted to be cutting edge, so [biomedical engineers] do a lot of development of new techniques and stuff like that, and having all my work count towards helping someone and actually benefiting someone’s life is really cool to me.” Parson mentioned that his work will benefit someone, but only as a brief follow-up to a point about innovation. The principle of beneficence was emerging in his response but not in a way that was robust enough to claim that Parson clearly demonstrated beneficence. Further examples of emerging aspects of reflexive principlism can be found in Table A-2 of the appendix.

Although we were able to find evidence of balancing and specification, we found that it was often challenging to identify the justification aspect of reflexive principlism because our interview questions allowed students to share their ethics experiences and general interpretations of engineering ethics without fully explicating their underlying reasoning. In one section of the interviews, the students were asked to explain the reasoning behind their choice in an ethics scenario which could potentially point towards to justification; however, the multiple-choice nature of the ethical scenarios did not give students the opportunity to develop an independent response and justification. Rather, they developed a post hoc justification for their selection.

In addition, the interview questions did not provide adequate insights regarding students’ reflectivity and reflexivity. Although the entire interview was a reflective experience by nature, isolating specific instances of reflection was challenging if not impossible. Reflexivity is another aspect of the framework that was difficult to identify in the study. As Beever & Brightman (2016) argue, reflexivity involves cultivating habits and intuitions, like an ethics reflex, that help individuals make sound ethical decisions. While some students are arguably quicker to recall an experience with ethics or respond to certain prompts, we could not make a confident claim about students’ ethical reflexivity.

We next turn to our findings, focused on the four principles and three processes introduced above. We begin with a high-level overview, and then delve into specific illustrative examples.

Ethical principles

This section is organized by each of the four ethical principles outlined in reflexive principlism. For each principle, we describe patterns we observed in the number of interviewees who referenced that ethical principle and provide examples of how those interviewees engaged with the principle. When relevant, we also describe changes in how individual participants relied on that ethical principle over time.

Specification

In addition to varied patterns in prevalence among the four principles, there were variations in how students discussed the principles in line with Beever and Brightman’s definition of specification as “reducing the indeterminateness of general norms to give them increased action guiding capacity, while retaining the moral commitments in the original norm” (2016, p. 282).

Increased attention to balancing across principles

One of the most notable changes we saw in our participants’ responses over time was an increase in both the prevalence and depth of their references to balancing across principles, especially balancing between nonmaleficence and beneficence. In T1, 10 (30%) participants showed evidence of balancing, and this number increased to 26 (79%) of participants in T2. These changes were easiest to see when we compared the entire T1 and T2 transcripts. Thus, in this section, we highlight two students, Parson and Beatrice, who showed increased attention to balancing across principles. By focusing on only two students, we are able to better illustrate the multiple instances in their interviews that reflect changes over time in their balancing. It should be noted the explicit use of the terms balance or balancing was not required to be considered as part of the students’ ethical reasoning; however, we have selected quotes that include the term in order to better highlight balancing within student responses.

As a first-year student, Parson talked in the abstract about ethics as maximizing benefits and reducing harm: “Ethics, I feel, is doing what’s right for other people, kind of seeing the effects of what you’re doing, and even if it’s harder, taking the path that will benefit the most people, and reduce hurt.” More specifically, he described a situation where he was forced to account for balancing interests and needs in the context of an engineering course project focused on using a bioprinter to print organs:

Here, Parson wrestled with how to balance beneficence (helping other people by creating 3D printed organs quickly) with nonmaleficence (avoiding harming them with a poorly constructed organ).

As a fourth-year student in T2, Parson spoke in more detail about the balancing required while he was doing biomedical research with rodents in order to design and test a medical device to relieve pain. He described the challenges of creating an animal testing protocol to test pain relief “which is a particularly difficult thing to do because you have to make something go through a painful experience to try and relieve it.” He said he needed to write a protocol “that was not excessive in the way that it applied pain, but also direct in that it did apply pain so that we can relieve it.” Parson appreciated the need to balance minimizing the pain to the rodent (nonmaleficence) while also achieving the research required to develop and test the device he and his collaborators were creating to ultimately benefit future patients (beneficence). He went on to explain that the balancing extended beyond this, since his lab did quite a bit of animal research, and he realized the importance of maintaining their reputation for performing ethical animal testing and of getting their proposed research protocol approved in the first place.

Beatrice provided another example of increased attention to balancing over time. In her T1 interview, Beatrice mentioned balancing only once, in response to an interview question which specifically asked, “How do you engineers balance duties to the public and to their employers?” Her reply was rather vague:

Beatrice responded to the question about balancing, but with no reference to an ethical principle beyond that of providing “better lifestyle, better living” to the public (beneficence).

However, in T2, Beatrice gave multiple examples that showed her understanding of the importance of balancing across ethical principles, drawing many of these examples from her experience interning in a structural engineering firm. As one example, when asked early in the interview about examples of situations where engineers face ethical situations in their work, Beatrice talked about how engineers in her company had to balance the practical sides of working with clients with the desire to ensure that designs are sufficiently safe:

Here, Beatrice talked about how conservative safety tolerances can actually harm a client, when they require work (like removing a beam in a house that is nearly complete) that may not actually be needed for a safe design. In effect, she was balancing between nonmaleficence—avoiding designs that lack structural integrity—and beneficence—getting a project completed without unnecessary delays. There also seemed to be hints of the principle of autonomy at work in Beatrice’s example, as she wrestled with who was making the decision (likely the engineers or builder rather than the homeowner) and how much to defer to professional standards.

Later in the interview, she discussed the need to balance creating a good design in an efficient manner:

Here, along with multiple other times in her T2 interview, we saw Beatrice drawing from her professional experience at the structural engineering firm to offer examples of where she engages in ethical balancing in her everyday work as an engineer.

Development beyond reflexive principlism

In addition to increased specification that we observed over time in our participants’ discussions of the four ethical principles in the reflexive principlism framework—especially that of nonmaleficence—we also found that there was an increase in participants’ specification of ethical norms beyond those contained in the framework. In most cases, these additional norms came from the students’ T2 interviews, where the students relied on their lived experiences from their four years of engineering studies to provide explicit and detailed explanations of how an ethical principle might be applied.

One student, Beverley, demonstrated increased specification in her tackling of how engineers are to guard confidential or secure data. In both her T1 and T2 interviews, she spoke about maintaining security protocols during defense-related engineering work, which enabled us to closely observe changes in her language and perspectives over time. In T1, Beverley referred to the Air Force base near her hometown, and how the engineers who work there are very careful to maintain security protocols: “They know, and I know, that they can’t tell me what they do because it’s in their contract that they can’t release any details or whatever.” She doesn’t portray this as a positive or negative attribute of the engineers, nor does she speculate about how they might actually put the norm into practice.

In T2, Beverley talked about the exact same example—that of the importance of confidentiality when working on defense-related work—but her own personal experience working as a defense contractor helped her realize the benefits of these security guidelines:

Concerns about data security and confidentiality present in Beverly’s response did not fit neatly into the principles defined in reflexive principlism. An argument could be made that confidentiality could be part of nonmaleficence, but confidentiality alone does not prevent harm. Confidentiality could potentially conflict with some ethical principles depending on the context of the engineering project, such as whether a whistleblower should expose unethical practices. Beverly linked confidentiality to national defense, but without any more details about the projects, it was difficult to connect other ethical principles.

In the case of Pete, his specification increased through his experiences in another culture. Pete expressed in his T1 interview that he “generally [thinks his] way is the right way, just in general,” and that he enjoyed arguing for his viewpoints in a discussion-based course he took. In his T2 interview, he consistently referenced living in Mexico for eight months as part of a global engineering program. Throughout the interview, Pete recalled experiences in Mexico that expanded his understanding of other cultural values:

Pete’s attitude changed from T1 to T2, shifting from defending and debating his own beliefs with others to respecting how culture can affect the ethical beliefs of people. Instead of insisting on the correctness of his own morals and beliefs, there was evidence that he increased his specificity through a learned respect for the autonomous beliefs of the host culture. This example also connects to balancing because he had to balance what he believed is right with what the surrounding culture elieved is right. Pete’s example could potentially also be related to the principle of autonomy because he wanted to respect the autonomy of a culture to make their own decisions and have their own values; however, as defined by reflexive principlism, autonomy is focused on individual decision making. Individual decisions are influenced by the culture the person is situated in, but Pete’s discussion focused on respecting the culture as a whole and did not address the interplay between individual and cultural values.

Notable findings

Nonmaleficence was the most frequently cited ethical principle in both the first and second interviews. However, there were instances where a student referenced this principle in the first interview but not the second, and vice versa. Because many students referenced nonmaleficence in one but not both interviews, we hypothesize that nonmaleficence may be an aspect of ethics that is near-universal for these engineering students, and does not necessarily differentiate one participant’s views in time from another’s. This consistent attention to nonmaleficence among our interviewees also raises the question of how engineering students develop an understanding that nonmaleficence is important to engineering ethics. Without explicit instruction around the principle, even first-year students seem to grasp its importance, even if it is not expressed in particularly sophisticated or nuanced ways. Further, we saw many instances where students would cite the importance of safety in engineering designs. We see safety as an important aspect of nonmaleficence, but not one that is synonymous with it.

The heightened attention that our participants seemed to have to questions of safety raises the question of whether this particular concern comes at the expense of other engineering ethics concepts and principles. Our results echo the findings from Rulifson and Bielefeldt’s longitudinal study of students’ views of socially responsible engineering, in which their participants maintained the importance of safety over time but decreased in their concerns toward improving society or addressing poverty (2019). Our participants’ attention to safety, coupled with their overall emphasis on nonmaleficence, leads us to conclude that, in formal engineering ethics education, discussions of nonmaleficence should be rebalanced to make time and space for discussing other, less familiar ethical principles. This conclusion aligns with Ausman et al.’s recent work on the idea of “undone ethics,” in which they argue that engineering students often possess a tacit understanding of the importance of safety, but lack the ability to navigate the multitude of nuances involved in engineering ethics (2022). Because of this, the authors argue for explicit instruction on the other dimensions of ethics that exist in engineering contexts.

Compared to nonmaleficence, beneficence was cited by students less frequently and in less precise ways, though we did see some changes over time towards more detailed discussions of it in select T2 participants. Despite social-good messaging that has emerged around engineering in the last few decades (e.g., National Academy of Engineering, 2004; National Academy of Engineering, 2008), students seem to still not be primarily concerned with benefits to others as they think about engineering ethics. We wonder if this is possibly because they see helping or benefiting others as an important part of their work, but do not equate such concerns with ethics. This finding from our student interviewees may also mirror a pattern we see in engineering as a whole, which emphasizes avoiding harm rather than improving welfare or benefiting the public. This is exemplified in the preamble of the National Society of Professional Engineer’s Code of Ethics, which states that the services provided by engineers must be, “dedicated to the protection [emphasis added] of the public health, safety, and welfare” (NSPE, 2019, p. 1), underlining that health, safety and welfare are things to be preserved (nonmaleficence) rather than increased (beneficence). There is also the potential that unlike nonmaleficence, where the tangible impacts like safety may be easier for students to conceive, beneficence is a more abstract concept. What a better society looks like, and the steps taken by engineers to get there, are not as concrete as safety imagery related to nonmaleficence.

Both autonomy and justice were notably less present in student interviews than the principles of nonmaleficence and beneficence. When open-ended interview questions asked students to define engineering ethics, autonomy and justice were never the primary focus and were often absent from their responses. Even on questions with language more in line with these principles, like a question regarding engineers’ and users’ responsibility for how technology is used, most students focused on user safety and unintended risks rather than the autonomy of user decisions. When students did reference justice or autonomy, it was often on a micro or personal level rather than a macro or societal level application. In order words, they were somewhat attuned to justice and autonomy as it applies to them, but less so for end users or society. The distinction between how justice and autonomy impact the students on a personal level, as opposed to a societal level, is reflective of the microethics and macroethics framing presented by Herkert (2005).

The focus on personal impacts of justice and autonomy may have been influenced by the phrasing of interview questions, as mentioned previously when discussing justice, but that would not explain the overall disengagement from these principles compared to nonmaleficence and beneficence. Students’ relative lack of attention to autonomy and justice may be because these are more abstract concepts than nonmaleficence and even beneficence, and are likely less prevalent in the messaging or culture students receive from faculty, industry, and the broader culture about the responsibilities of engineers (Leydens & Lucena, 2017; Smith & Lucena, 2021). Here again it may be much easier to imagine the impact of a bridge collapsing (nonmaleficence) than to consider the equitable distribution of risks and benefits associated with building a bridge in a specific location and context (justice). As a result, students may not be accustomed to considering issues of autonomy and justice, nor practiced in applying them to ethical situations.

The increase in specification and balancing in student responses across the interviews was often associated with their personal experience in engineering, such as an internship or design course. In T1, the students typically lacked the real-world experience needed to identify the specifics of an engineering ethical dilemma and their language reflected their lack of personal experience. In addition to a lack of specificity observed in T1, the students often described their principles in vague language rooted in a general sense of right and wrong. Examples of ethical engineering and ethical behavior were not grounded in specific engineering problems but rather students’ broad sense of engineering as a field. By T2, many of the students had engineering-related experience, gained through formal education or work/internship experiences, to reflect on and to help guide how their values connect to engineering. The students with engineering experience in T2 were able to reference these experiences when prompted and relate them to engineering ethics. When reflecting on their experiences, the students were able to specify conflicting principles and stakeholders more clearly than in T1. While it is expected that students’ language would become more mature from T1 to T2, the explicit connections students made to engineering suggest that the increase in specificity is not purely the result of individual, general growth from their years in a university setting, but is potentially also influenced by their engineering experiences. The specific engineering experiences varied between each student, but the authentic experiences were important for developing their personal ethical framework within the engineering space. Over the course of their engineering education, the students were socialized to an engineering environment and were able to reflexively reference applicable experiences when prompted. Prior to their engineering studies, most of the interviewed students lacked the experience needed to integrate more technical language into their understanding of engineering ethics. Our results suggest that through their engineering educations, some students were able to utilize their exposure to engineering language and situations to discuss engineering ethics in a more detailed and nuanced manner as compared to when they were first-year students. These observations strongly align with our previously published work, which offered additional insights about what students report learning about ethics through workplace experiences (Howland, Kim, et al. 2022).

Similarly, the students’ experiences in engineering—especially in industry or research settings—increased their ability to balance conflicting ethical values. When exposed to real engineering problems, such as testing a biomedical device or adjusting a building design, students have to address a diverse set of stakeholder needs that may be in conflict. Repeated experience making decisions with inherent tension helps students learn how to articulate and balance the ethical principles involved. Even when two principles may contradict each other, such as Parson’s example of harming a lab animal to develop a device that will help people, the experience of making these real-world ethical decisions increases a student’s ability to acknowledge multiple positions and find middle ground.

Implications of this work

This paper is part of a larger project which aimed to use diverse quantitative and qualitative measures to gain a multi-faceted understanding of engineering students’ perceptions of engineering ethics and social responsibility and how those views changed over time (Zoltowski et al., 2020). The process of analyzing our longitudinal, qualitative data revealed many of the attendant methodological challenges associated with analyzing data related to engineering ethics. Difficulties associated with quantitatively measuring students’ understanding of ethics and qualitatively interpreting students’ understandings of ethics are likely two sides of the same coin.

As discussed in the introduction to this paper, engineering education has long struggled with measuring competencies including communication, lifelong learning, and the ability to work effectively in teams (Cruz et al., 2020; Shuman et al., 2005). We wonder if engineering ethics is difficult to measure—and its associated research data difficult to analyze—for many of the same reasons: the lack of a shared and commonly agreed-upon definition or understanding of what is meant by engineering ethics and social responsibility, and a lack of validity and reliability evidence (Cruz et al., 2020). Research on engineering ethics may benefit from a reliance on multiple frameworks rather than just a single approach, as we originally set out to do in our larger research project. However, identifying what those multiple frames could be and then identifying what framework(s) to use in the associated analysis of specific data sources are challenges that remain, as we saw in the analysis described here.

Despite these challenges, the findings presented in this paper further our understanding of how engineering students conceptualize ethics and social responsibility. Prior work used reflexive principlism to understand how engineering students engage in ethical decision making in a service-learning design course (Corple et al., 2020). In comparison to our work, Corple and co-authors approached their analysis from the assumption that all engineering work—especially that which is service-focused—is fundamentally about beneficence, with the other principles of autonomy, justice, and nonmaleficence acting to specify or mediate beneficence. Their results showed that student design teams’ interactions with users and project partners were the primary factor that influenced student awareness of, and engagement with, ethical principles. Depending on the team and project context, evidence of autonomy, nonmaleficence, and justice was found in multiple stages of the design process. Given the context of our study which examined ethical and social responsibility more generally, we did not begin with a similar assumption situating beneficence as fundamental to answering ethical questions and as being more salient than others. We wonder if the relative lack of justice and autonomy that was present in our data compared to Corple’s may be because our context was more general, and not situated in a service-learning design context with engagement with users and project partners, where they may be more relevant and prevalent.

Finally, our findings have educational implications. They show that engineering students may be prone to relying on nonmaleficence more than any other ethical principle, pointing to a need for engineering ethics instructors to emphasize beneficence, autonomy and justice (while at the same time perhaps limiting the existing disproportionate attention to nonmaleficence). Explicitly introducing students to these four ethical principles, as proposed by Corple et al. (2020), may also make students more attuned to them when ethical situations arise. Finally, we suggest that balancing and specification are important parts of ethical reasoning that might also benefit from direct instruction, though our work finds that some students seem to develop these skills even in the presumed absence of such instruction.

Limitations of the reflexive principlism framework

While reflexive principlism provided a useful lens for examining how students’ understanding of ethics changed over time (in terms of both the ethical principles students referred to and the ethical reasoning processes they engaged in), our approach also highlighted some of its limitations. First, reflexive principlism focuses on only four principles of ethical engineering when there are likely principles beyond those four that are relevant to engineering ethics. As mentioned previously, confidentiality and cultural relativism, two principles which participants in our data set referenced, do not fit neatly into the four principles originally derived from biomedical ethics. There are also aspects of engineering work that the four principles do not address such as design feasibility, budget constraints, professionalism, and community norms. We could argue over each aspect of engineering work and how it fits into one of the four principles of the framework, but that runs the risk of ignoring some nuances in order to force an alignment. Beever and Brightman (2016) argue that the four principles are comprehensive and unified because they converge together with coherence, but they fail to elaborate on the details of this convergence or how it looks in practice. The four principles present in the framework are robust, but there may be room to add more principles and expand the framework as the application extends beyond biomedical ethics and is applied to engineering ethics. Formosa, Wilson, and Richards (2021), for example, propose “explicability” as a fifth principle for cybersecurity ethics, underscoring the need for intelligibility and transparency in this particular domain of technology.

Other limitations may stem from our emergent decision to use reflexive principlism, resulting in potential gaps between our interview protocol and the subsequent analysis. While reflexive principlism was identified as a relevant and applicable conceptual framework to apply to these data, there were some aspects of the framework that the interview protocol was better at eliciting than others. For example, as mentioned previously, the justice-focused question in the interview did not align with the definition of justice presented in reflexive principlism and provided very little usable data as a result. More exact questions about the principles or ethical reasoning used by our interviewees, rooted in their established definitions, could potentially prompt students to respond in kind. Additionally, the aspects of reflection and reflexivity from the framework were difficult to discern from the interview questions. The students were reflecting back on their engineering ethics experiences in the interview, but more critical reflective questions could be asked to better highlight these aspects of the framework.

Reflexive principlism was developed as a pedagogical framework, where students are expected to develop the ethics reflex through repeated and iterative practice. Since we are using the framework as an analytical tool, the framework requires more methodological creativity to measure and evaluate reflexivity that is not present in a post-hoc application of the framework. The reflexivity of responses would be a difficult aspect to measure in interviews and may require supplemental measurement in future studies, such as evaluating the use of filler words or recording the time it took students to respond to questions. Even an analysis into the structure of student responses may not reveal the reflexive nature of student responses due to the limitations of the interview format. The question-and-answer nature of interviews prompts a direct response instead of letting the student incorporate ethics reflexively, without intervention.

While the decision to use reflexive principlism as a theoretical framework only at the analysis phase presents some limitations, the applicability of the framework speaks to its overall strength. The interviews were not designed with reflexive principlism as the theoretical basis, but aspects of the framework were still clear in student responses. The framework was also useful for addressing aspects of the student responses, such as the increased specificity, in ways we struggled to articulate with other analytic approaches. Our emergent decision to use reflexive principlism to analyze our interview data allowed us to more accurately probe which ethical principles and processes students described without the biases that explicitly asking about specific dimensions of the framework may have introduced.