2.1 Research Contexts and Participants
Participants were undergraduate engineering students enrolled in the course [blinded], at [blinded]. [blinded] is a US-Chinese educational institute founded in 2006 and located in [blinded]. It offers BS, MS, and PhD degrees in engineering, and has programs in mechanical engineering and electrical and computer engineering. Engineering education at the [blinded] is modeled on that at [blinded], which explains its focus on ethics education for engineering students. Like [blinded], to partially fulfill ABET student outcomes related to ethics, [blinded] offers [blinded]. [blinded] is a required, two-credit-hour course that students typically take during their junior or senior years, and it is unique in several ways.
Most curricula in engineering ethics take a top-down, micro-ethical approach, beginning with professional codes and/or normative ethical theories that are then applied to ethical issues facing individual engineers, which arise in case studies focusing mainly on disasters that have taken place in the Western world (Herkert, 2001; Hess & Fore, 2018; Polmear, Bielefeldt, Knight, Canney, & Swan, 2019; Van de Poel & Royakkers, 2011). However, such approaches are problematic for several reasons.
First, professional codes and technical guidelines can vary by country (AlZahir & Kombo, 2014) and – as explained above – engineering is evermore global. Educators can no longer assume a shared cultural tradition – or country of origin or destination for work – between themselves and their students (Clancy & Zhu, 2022). Next, there is widespread disagreement regarding which (if any) normative ethical theories are correct: Even after thousands of years, professional ethicists continue to disagree about which normative ethical theories should be used for ethical analysis (Greene, 2014; Luegenbiehl, 2010). Third, normative ethical theories used in engineering ethics education tend to come from the Western cultural tradition, including consequentialism, deontology, and virtue ethics. In recent years, attempts have been made to diversify the number and nature of ethical theories and traditions used in engineering and technology ethics, including ones from Asia and Africa (Verharen et al., 2021; Wong & Wang, 2021). However, this simply raises the first two issues again, regarding the national and cultural background/destinations of students enrolled in engineering ethics courses, and which normative theories are correct/should be used in engineering ethics. Finally, there is a growing consensus that engineering ethics education should move beyond case studies involving engineering disasters with a focus on micro-ethical issues alone, those involving disasters that result from decisions by and the behaviors of individual engineers. Instead, education should focus on “aspirational ethics,” cases where engineers have done the right thing (Harris, 2008; Harris et al., 2018), as well macro-ethical issues, for example, public policies, government actions, and corporate social responsibility (Polmear et al., 2019; Van de Poel & Royakkers, 2011; Zhu et al., 2022). [blinded] was developed and implemented to address these issues.
Rather than a top-down, micro-ethical approach using professional codes and/or normative ethical theories that are then applied to ethical issues that arise in case studies focusing mainly on disasters in the Western world, [blinded] takes a bottom-up, micro- and macro-ethical approach, beginning with case studies that focus on a broader range of global topics, reasoning to topic- and discipline-specific ethical principles and values on that basis. (A detailed description of this course, and factors shaping its development, can be found in [blinded].) This course is distinctive in its approach to engineering ethics education. It responsive to different technologies and cultural traditions, but it is not rooted in any one technology or tradition. This approach is the most appropriate for an educational institution such as [blinded], where students come from and go on to further study and work in countries and companies throughout the world. A thorough discussion of engineering ethics education in China would lead well beyond the current study and paper, but the interested reader is encouraged to consult the growing literature on the topic (Cao, 2015; Clancy, Zhu, & Tang, n.d.; Fan, Zhang, & Xie, 2015; M. J. Murphy, 2016; Tang, Zhang, & Yang, 2017; Wang & Yan, 2019; Zeng & Resnik, 2010; H. Zhang & Davis, 2018; Zhu, 2010).
Ultimately, 99 students were included as study participants (female = 29; mean age = 21.3), of whom none were US citizens. Data collection occurred in two waves, during the Fall and Summer offerings of [blinded]. The number of participants who completed the survey, consented to have their responses used for research purposes, passed attention/earnestness checks embedded in the MFQ, and whose pre- and post-course surveys matched can be found in Table 1.
Table 1. Participant numbers
|
Semester
|
Beginning
|
End
|
Completed and matched
|
|
Completed
|
Consented
|
Completed
|
Consented
|
|
Fall
|
79
|
68 (86%)
|
88
|
84 (95%)
|
28
|
|
Summer
|
127
|
115 (90%)
|
132
|
126 (95%)
|
71
|
|
Total
|
206
|
183 (88%)
|
220
|
210 (95%)
|
99
|
The discrepancy between the number of participants who completed the survey at the beginning versus the end of the semester results from the fact that registration at [blinded] is open the first two weeks of the semester, such that students dropped out and enrolled in the class after the first day of class and before the last day of class, when surveys were distributed and completed.
To ensure the sample quality, relatively stringent criteria were used to include responses: Only the response of participants who responded to all survey items and correctly – in other words, filling in only one response – were included. According to MFQ protocols, responses should be excluded for answering 3 or above on the “math” catch question, and 2 or below on the “good” catch question. Pre- and post-course responses were joined using a coded id, and only the responses of those participants who completed a pre- and post-course study survey were maintained.
2.1.1 Citizenship and language
None of the participants were US citizens. 1 came from Africa, and the rest identified their region of origin as China, Korea, or Japan. None of the participants were native-English speakers, but 68 participants had taken the TOEFL (Test of English as a Foreign Language), and the mean score of those participants was 103.4. Of undergraduate students who have taken the TOEFL, this score falls in approximately the 85th percentile (EST, 2018). These results are typical of students in the [blinded]. In 2017, the [blinded] conducted a survey of undergraduate students who took the TOEFL, finding the mean score was 102.45 (n = 186; SD = 6.19). As a result, all participants in this sample have high-level English-language proficiency. (For comparison, the average TOEFL score of test takers from China is 79 (EST, 2018).)
This results from the fact the official language of the [blinded] is English, and all study participants received immersive English-language instruction. Again, since students from the [blinded] go on to study and work in international engineering environments, where English is typically the language used, and since language can affect ethical decision-making (Chan, Gu, Ng, & Tse, 2016; Costa et al., 2019; Keysar, Hayakawa, & An, 2012), the assessment and educational instruction associated with this study took place in English.
2.2 Procedure and measures
Data collection occurred in two waves, first during the fall 2019 offering of [blinded] and second during the summer 2020 offering of [blinded]. During the first wave of data collection, participants completed a paper version of the survey, which was handed out at the beginning of the first day of class. During the second wave of data collection, participants completed a digital version of the survey, which they accessed through a link or QR code also provided at the beginning of the first day of class. The same respective procedures were followed to collect post-course data, which occurred on the last day of class.
In all cases, participants were given 45 minutes to complete the survey. A brief description of the nature of and motivations for the research was given, and participants were required to consent to have their responses used for research purposes. Such work was exempt from securing IRB approval at [blinded], and only the results of participants who consented to have their responses used for research purposes were included in this study. The survey consisted in three parts: the 1. ESIT, 2. MFQ, and 3. demographic items.
2.2.1 ESIT
The ESIT is a neo-Kohlbergian instrument, an engineering-and-science-specific variant of the DIT (Defining Issues Test)/DIT2 (Rest et al., 2000; Rest, Narvaez, Thoma, & Bebeau, 1999), developed and validated by Jason Borenstein and colleagues (Borenstein et al., 2010). It presents participants with six ethical dilemmas related to engineering and/or science. Each scenario is followed by a choice of different ways to resolve the dilemma, as well as twelve considerations that could be relevant to that choice. Participants are asked to rate the relevance of each consideration and then pick the four they think are the most important. Each of these considerations corresponds to one of three different “schemas,” ways of thinking about matters of right and wrong: 1. the preconventional schema consists in reasoning based on self-interest; 2. the conventional schema consists in reasoning based on authority and social norms; 3. the postconventional schema consists in reasoning based on universal principles (Borenstein et al., 2010; Rest, Narvaez, Thoma, et al., 1999).
The more postconventional considerations one picks in the top four, the higher one’s P score, indicative of the prevalence of postconventional reasoning. This measure was designed, in part, to assess the prevalence of postconventional reasoning. The prevalence of preconventional and conventional reasoning is determined in this same manner. An additional measure of ethical reasoning used by the ESIT is the N2 score. The N2 score indicates the prevalence of postconventional relative to preconventional reasoning – not only that participants use postconventional reasoning but also that they do not use preconventional reasoning. On this view, reasoning based on universal principles related to justice would be the most developed/advanced and, therefore, the most ethical, while reasoning based on authority and social norms, and self-interest, would be less developed/advanced.
Higher levels of education, age, and more politically liberal views have been associated with higher P and N2 scores on the DIT and DIT2 (Dong, 2011; Rest, Narvaez, Thoma, et al., 1999). On average, US citizens/native-English speakers score higher on these measures (Borenstein et al., 2010; Canary et al., 2012), while East Asians tend to score higher on measures of preconventional and conventional reasoning (Hwang, 2012a).
2.2.3 MFQ
The MFQ is associated with MFT (Moral Foundations Theory) and presents participants with two sets of statements. For the first set of statements, participants decide how important each would be when deciding whether something is right or wrong, the “relevance” subscale. For the second set of statements, participants indicate their levels of agreement, the “judgment” subscale (Graham et al., 2011). Each statement corresponds to one of five different “moral foundations,” ways of conceiving matters of right and wrong, concerned with different kinds of behaviors and considerations. These are care-harm, fairness-cheating, loyalty-betrayal, authority-subversion, and sanctity-denigration, where caring for others is good and harming them is bad, acting fairly is good and cheating is bad, and so on (Graham et al., 2011). Care and fairness are called the “individualizing” foundations, since they are associated with virtues aimed at protecting individuals, whereas loyalty, authority, and sanctity are called the “binding” foundations, since they are associated with virtues aimed at binding individuals into groups (Graham et al., 2011). Higher mean scores on items corresponding to each of the foundations indicate the relative preference given to these foundations and their associated intuitions.
Those who identify as politically conservative and those from East-Asian cultures tend to care about all the foundations, whereas those who identify as politically liberal and those from Western cultures prioritize the individualizing foundations (Graham, Haidt, & Nosek, 2009; Graham, Meindl, Beall, Johnson, & Zhang, 2016; Graham et al., 2011; Kim, Kang, & Yun, 2012; Y. Zhang & Li, 2015). Such insights can contribute to developing more psychologically realist theories of ethics, concerned with how people actually think about matters of right and wrong rather than merely how they should (Ancell, Steenbergen, Flanagan, & Martin, 2014; Flanagan, 2017). For example, as a pluralist theory of ethical reasoning, MFT helps to explain how different, competing goods can conflict, resulting in the kinds of conflicts of interests that are central to engineering ethics and other forms of professional ethics (Harris et al., 2018; Van de Poel & Royakkers, 2011; Zhu et al., 2022).
Although moral foundations have been likened to dispositions – collections of (relatively) invariant traits, similar in nature to personality types (Haidt, 2012) – little research has explored if or how moral foundations change over time (Graham et al., 2011; Hatemi, Smith, Alford, Martin, & Hibbing, 2015). Of the work that has been done, the methods used and conclusions drawn have been contested (Haidt, 2017). No research of which the authors are aware has explored the effects of education on moral foundations.
2.2.4 Relations between the ESIT and MFQ
As the foregoing makes clear, the ESIT and MFQ draw on two different conceptual models of ethical decision-making and moral judgments. While the ESIT is based on a model of ethical decision-making involving neo-Kohlbergian schemas, the MFQ is based on a model of moral judgments involving social intuitions. Although the ESIT and MFQ have not been used together, the DIT2 – the neo-Kohlbergian measure on which the ESIT is based – and MFQ have (Baril & Wright, 2012; Glover et al., 2014).
These studies found evidence of positive relations between ethical reasoning and the individualizing foundations, and negative relations between ethical reasoning and the binding foundations: P and N2 scores on the DIT2 were positively related to mean individualizing foundation scores on the MFQ, and they were negatively related to mean binding foundation scores. These relations are likely because the DIT2/neo-Kohlbergian model conceives of ethical reasoning as applying universal principles related to justice and care, and the individualizing foundations concern intuitions about fairness and care. By contract, the binding foundations concern intuitions about loyalty, authority, and sanctity and, according to the DIT2/neo-Kohlbergian model, principles associated with loyalty, authority, and sanctity belong to conventional reasoning.
2.2.5 Hypotheses and planned analyses
In this study, MFQ and ESIT scores were treated as outcome variables, and education and demographic information were treated as input variables. Since relatively few studies have used the ESIT (Borenstein et al., 2010; Canary et al., 2012; Kerr, Brummel, & Daily, 2016) – and none have used the ESIT in conjunction with the MFQ – this study was largely exploratory in nature. Nevertheless, based on previous work, to conduct analyses and present results, the following hypotheses were posed:
1. It was hypothesized that students in this sample would receive lower N2 scores on the ESIT than those in (Borenstein et al., 2010), since the participants in this sample were non-US citizens, and non-US citizens have been found to receive lower N2 scores (Borenstein et al., 2010; Canary et al., 2012).
2. Since previous research found evidence for the effects of pre-course/-study ethics education on ESIT P and N2 scores (Borenstein et al., 2010), it was hypothesized that participants with pre-course/-study ethics education would receive higher P and N2 scores than those without and, by extension, that students would receive higher P and N2 scores after completing [blinded].
3. It was hypothesized that higher mean scores on the individualizing foundations and lower mean scores on the binding foundations would be associated with higher P and N2 scores on the ESIT, based on prior work using the MFQ and the DIT2, a neo-Kohlbergian instrument like the ESIT (Baril & Wright, 2012; Glover et al., 2014).
Since previous research has not explored the effects of education on moral foundations, no hypotheses were made regarding its effects, although this was a point of interest as well, the results of which are reported below.