Engineering Ethics Education for Sustainable Transport: A Dual-Mediation Model of Teaching Satisfaction, Embodied Experience, and Self-Efficacy

Abstract

Integrating engineering ethics education into the curriculum system of China’s transportation engineering major is crucial for promoting Sustainable Development Goal 11 (SDG 11) in the field of transportation engineering in China. However, the mapping relationship between teaching satisfaction and engineering ethics literacy, which are traditional indicators for evaluating teaching effectiveness in China, is not clear. This study constructed a teaching satisfaction transformation model with Experiential Transformer Engagement (ETE) and Self-Efficacy (SE) as dual mediators. Through experimental testing of undergraduate students majoring in transportation engineering from five universities in Hebei Province, it was found that both ETE and SE significantly moderated the conversion of teaching satisfaction to two key ethical abilities: Ethical Decision-Making Competence (EDMC) and Social Responsibility Strength (SRS). Among them, ETE accounted for about 64% of the total indirect impact on the two outcomes, which was significantly stronger than SE’s 48%. In addition, achievement goal orientation has a significant moderating effect. These findings not only address the teaching framework of engineering ethics that is suitable for the Chinese context, but also expand the theoretical basis and implementation plans of teaching models in the early stages of engineering ethics education in developing countries.

1. Introduction

Rapid global urbanization and the climate crisis have positioned transportation systems at the center of sustainable development discourse. Road traffic accidents result in approximately 1.35 million fatalities annually [1], while transportation-related emissions account for roughly 24% of global energy-related CO₂ emissions [2]. In response, the United Nations Sustainable Development Goals (SDGs) explicitly call for resilient infrastructure (SDG 9) and safe, inclusive, and sustainable cities (SDG 11) [3].

Achieving these objectives requires not only innovative technologies but also engineers who can integrate safety, equity, and ecological stewardship into their design decisions. Despite the widespread incorporation of ethics modules in engineering curricula, we find that there is no significant correlation between students’ satisfaction with teaching and their willingness to give priority to public interest issues, which shows that there is a certain gap between the satisfaction in the evaluation of educational achievements and the requirements for technical personnel’s ethical literacy in engineering practice. This disconnect—termed the “satisfaction–responsibility paradox”—undermines the educational efficacy needed to advance sustainable transportation practices. Self-determination theory (SDT) posits that teaching satisfaction [4], grounded in the fulfillment of autonomy, competence, and relatedness, enhances intrinsic motivation, thereby fostering deeper engagement in value-laden tasks such as ethical reasoning [4]. From this perspective, satisfaction acts as a catalyst for students’ willingness to engage with complex socio-technical dilemmas—a process most effectively facilitated through Experiential Transformer Engagement (ETE). Complementarily, social cognitive theory (SCT) highlights self-efficacy SE [4]—the belief in one’s capacity to execute specific actions—as a key determinant of whether individuals translate ethical awareness into actual decisions, particularly under pressure. SE is not merely a byproduct of motivation; it functions as a psychological resource that sustains ethical agency beyond academic settings.

Critically, these two theoretical pathways are not parallel but potentially synergistic; satisfaction derived from well-designed experiential tasks (ETE) can strengthen (SE), while heightened SE may in turn reinforce engagement in ethically challenging experiences. However, the extent to which and the mechanisms by which these factors jointly explain the relationship between teaching satisfaction and sustainability-oriented ethical outcomes remain empirically unexplored in engineering education—a gap this study aims to address. To bridge said gap, we propose and validate a dual-mediation model linking teaching satisfaction to sustainability-oriented ethical outcomes through ETE [5] and SE [6]. Drawing on survey data from 327 undergraduate transportation engineering students across five Chinese universities, we examine three research questions:

RQ1. Does teaching satisfaction significantly enhance students’ EDMC and SRS?

RQ2. Do ETE and SE mediate these relationships?

RQ3. Whether achievement–goal model has moderating mediating effect?

To empirically test these questions, we draw upon Self-Determination Theory and Social Cognitive Theory to formulate a set of specific hypotheses, which posit directional relationships and comparative strengths among the constructs, moving from exploration to prediction.

Addressing these research questions generates three key contributions: First, we quantify the relative influence of experiential and psychological pathways on ethical learning outcomes, offering curriculum developers evidence-based insights to inform strategic resource allocation. Second, we extend SDT and SCT within the context of sustainability-oriented engineering education, demonstrating that authentic transportation challenges—such as the trade-offs between implementing bus lanes and displacing low-income communities—serve as impactful situated learning experiences. Thirdly, we have integrated mature teaching models in Western engineering education, such as project-based learning, role-playing, and iterative feedback, to construct an ETE teaching process based on real scenario simulation, and provided corresponding support strategies for students’ SE. This framework system not only verifies the impact of ETE and SE on the effectiveness of engineering ethics teaching, but also explores how countries and regions like China, where engineering ethics education has just begun to integrate into professional education, can deepen the reform of engineering theory education, and whether mature Western experiences are applicable, addressing the specific social and technological challenges faced by developing countries and regions.

In summary, although teaching satisfaction is often regarded as an indicator of educational effectiveness, it is not clear how to transform it into the potential mechanism of engineering ethics literacy required by engineers in the process of sustainable development of transportation. In many cases, there is a problem that satisfaction does not match the effectiveness of Engineering ethics education. In view of this situation, this study integrates SDT and SCT to propose a dual-mediation model, which assumes that teaching satisfaction jointly influences students’ engineering ethics decision-making capabilities and social responsibility strength through the “value internalization” path of ETE and the “behavior empowerment” path of SE. Based on a systematic review of relevant theories, we will identify the differences in the research between teaching satisfaction affecting students’ EDMC and SRS through ETE and SE.

2. Literature Review

In the dual context of global sustainable development and the transformation of engineering education, engineering ethics education, as a key link in cultivating future engineers’ social responsibility and ethical decision-making abilities, is increasingly valued for its teaching effectiveness and mechanism research. However, there are still theoretical gaps and insufficient empirical evidence in explaining how teaching satisfaction can be transformed into sustainable ethical competence, especially in the context of engineering education in rapidly developing regions such as China. This chapter clarifies the theoretical positioning and research gaps of this study by systematically reviewing relevant theoretical frameworks and empirical research. Firstly, by integrating self-determination theory (SDT) and social cognitive theory (SCT), a dual mediation theory model of “experience-transformation participation” and “self-efficacy” is constructed (Section 2.1, Section 2.2 and Section 2.3); Secondly, we define the contextualized connotation of “experience-transformation participation” in engineering ethics education (Section 2.4); Next, we explore the moderating effect of achievement–goal orientation (Section 2.5); Finally, we identify the three major gaps in current research (Section 2.6) and lay a theoretical foundation for subsequent empirical studies.

2.1. Theoretical Positioning of Key Studies

In order to better integrate engineering education in local Chinese universities with the SDGs framework, this article conducted a comparative study of existing research results and also found a benchmark for future research. In order to increase the contrast of research results, the criteria for literature selection included the correlation of research methods, the correlation of experimental path structure, and the provision of comparable quantitative parameter effects. The main achievements are as follows:

The Value Action Gap Approach: Teather and Etterson (2023) [7] emphasizes the moderate correlation between the sustainability attitudes and commitments of Generation Z students (* r * ≈ 0.29), providing a key reference point for investigating the “Satisfaction Responsibility Paradox” in this study and highlighting the limitations of knowledge transfer alone.

The path from self-efficacy to performance: Walumbwa et al. (2011) [8] demonstrated in organizational contexts that self-efficacy is an important predictor of moral performance (β = 0.38). This provides strong meta-theoretical support for the assumed Social Cognitive Theory (SCT) pathway in our model and establishes a comparable effect size benchmark.

Experiential learning efficacy pathway: Fini et al. (2018) [9] used a quasi-experimental design to demonstrate that project-based learning significantly improved the sustainability learning outcomes of transportation courses (Cohen’s * d * = 0.72). This study provides a direct method and effect-size reference for the design and expected impact of the experiential transformer device in our research.

Sustainable Development Capability Framework Path: Wiek et al. (2011) [10] provided a theoretical foundation for normative capability through their Key Capability Framework for Sustainable Development, which is a core component of our outcome variable, Ethical Decision-Making Capability (EDMC).

Overall, these studies confirm the independent importance of experiential learning and self-efficacy. However, they were mainly conducted in the Western or post-materialist context, without comparing the relative strength of empirical (such as ETE) and psychological (such as SE) pathways in the integrated model. Therefore, based on these benchmarks, this study proposed and tested a dual mediation model that combines teaching satisfaction, ETE, and SE. Our goal is to test and compare the effectiveness of these pathways in the context of engineering education in a rapidly developing region (Hebei Province, China). This method directly fills the gap left by benchmark research, especially the lack of comprehensive comparison and contextual application, while allowing for meaningful evaluation of our results based on established effect sizes.

2.2. From Teaching Satisfaction to Sustainability Outcomes

Student satisfaction is a widely recognized metric for assessing teaching quality and its immediate outcomes. Grounded in SDT, satisfaction derived from the fulfillment of autonomy, competence, and related needs is believed to enhance intrinsic motivation, which subsequently fosters deeper engagement and behavioral persistence [4]. In the context of engineering education, students who report higher levels of satisfaction typically exhibit greater course engagement and improved technical learning outcomes.

However, when the anticipated outcomes shift from general engagement to sustainability-oriented ethical competencies—such as the willingness to prioritize public good over costs or the strength of pro-sustainability behavioral intentions—the empirical relationship with teaching satisfaction becomes notably less consistent and is theoretically underexplored. While some studies indicate positive correlations, the effect sizes are often modest (β≈ 0.20) and highly dependent on contextual factors such as instructional design and assessment methods. More critically, other research highlights a satisfaction–responsibility paradox: students may express high levels of satisfaction with ethics courses yet demonstrate limited transfer of ethical reasoning to complex real-world trade-offs [7]. This paradox suggests that satisfaction alone may be an inadequate lever for cultivating the normative competencies and value commitments essential for sustainable development.

SCT provides a complementary yet distinct perspective by emphasizing SE—the belief in one’s ability to organize and execute courses of action—as a more proximal and potent predictor of behavioral intention and persistence, particularly in challenging or ethically ambiguous situations [11]. Meta-analytic evidence suggests that domain-specific self-efficacy may serve as a stronger predictor of ethical behavioral intentions than general course satisfaction [12]. This theoretical divergence prompts an essential inquiry: does teaching satisfaction influence sustainability outcomes primarily by fostering intrinsic motivation toward ethical engagement (an SDT pathway), by enhancing the confidence to act ethically (an SCT pathway), or through a combination of both? The existing literature lacks a coherent framework for examining these mechanistic pathways.

This study constructs a dual mediation model with ETE and SE as the main pathway to systematically explore the roles of ETE and SE in the process of converting teaching satisfaction to engineering ethics literacy, clarify the differences and correlations between the two, and provide a theoretical support and teaching plan for a new model of engineering education reform, especially for developing countries and regions where transportation engineering is rapidly developing but engineering ethics education is lagging behind.

2.3. Theoretical Integration: SDT and SCT as Complementary Lenses

In classical educational theory, self-determination theory (SDT) and social cognitive theory (SCT) respectively point out two key paths:

The path based on SDT (usually achieved through ETE): It emphasizes the emotional and cognitive involvement of students in moral issues in real and challenging tasks, thereby promoting the internalization of their values.

The path based on SCT (mainly constructed through SE): It focuses on behavioral empowerment, enabling students to believe that they have the ability to adhere to ethical standards under real pressures such as cost and time.

However, there may be tension between these two paths in practice:

If there is an excessive reliance on performance feedback in cultivating SE, it may weaken students’ autonomy, which is a core requirement of SDT.

If ETE only provides immersive challenges without sufficient teaching support, it may make students feel powerless rather than empowered.

Therefore, a dual-pathway model helps us not only examine the contributions of ETE and SE, but also explore their interactions—whether they work independently, in a sequential order, or in synergy with each other?

This comprehensive framework goes beyond the perspective of viewing ETE and SE as isolated intermediaries, but rather as two theoretically based transmission channels that jointly lead teaching satisfaction towards two complementary dimensions of engineering ethics learning: internalization (via ETE) and empowerment (via SE).

2.4. ETE in Engineering Ethics: A Situated Definition

Based on Bandura’s self-efficacy theory [6] and Kolb’s experiential learning theory framework [5], this study conceptualizes ETE in the context of engineering ethics education as a structured and guided teaching experience. In this context, students are immersed in real social and technological dilemmas that contain ethical conflicts, physically, cognitively, and emotionally, and must make and demonstrate their value-oriented decisions within the constraints of reality. Unlike general experiential learning (such as simulations or case studies), the ETE in this framework is defined by three core features that integrate sustainability and ethics, and collectively aim to cultivate self-efficacy in engineering ethics decision-making:

Situated Authenticity: The learning environment directly maps or integrates into real transportation system scenarios (such as intersections, construction sites, and transportation corridors), making ethical trade-offs such as safety and cost, fairness and efficiency tangible. This authenticity provides a reliable source of master experience for the formation of self-efficacy.

Multi-Sensory Engagement: The learning process not only relies on cognitive analysis but also emphasizes perceptual and emotional intervention through observing traffic flow, listening to community testimony, drawing spatial inequality maps, etc., to enhance the emotional significance and value-internalization depth of ethical dilemmas. This participation provides students with alternative experiences and emotional awakenings, further consolidating their beliefs in coping with similar challenges.

Ethical Action Imperative: Task-design forces students to propose, defend, or make decisions with ethical implications, and often promotes their application, coordination, and negotiation value in simulated or real responsibilities through studio models such as “design and rebuttal” (see Section 5). This process directly exercises their normative ability and strengthens their self-efficacy belief of ‘I can do it’ through successful execution experience.

In summary, by integrating classic theories of instructional psychology, Experiential Transformer Engagement (ETE)’s teaching design, which combines real-life situations, multisensory involvement, and action compulsion, not only promotes the internalization of engineering ethics principles, but more importantly, it systematically constructs the necessary experience for students to cultivate self-efficacy in engineering ethics decision-making, enabling them to believe that they have the ability to analyze, judge, and act in the face of complex sustainability ethical dilemmas. This integrated perspective goes beyond simple experience accumulation and points towards the deep teaching goal of cultivating professional ethical initiative through targeted experiential design [6].

2.5. Achievement–Goal Orientation as a Moderator

Students enter ethics courses with a variety of goal profiles. Mastery-Approach goals (MAP) emphasize understanding and self-improvement, while Performance-Approach goals (PAP) focus on the demonstration of normative competence [13]. Students who adopt MAP are inclined to pursue challenging tasks and demonstrate tolerance for ambiguity, whereas those with PAP prefer clear rubrics and high success visibility. Moderated mediation analyses reveal that MAP enhances the effectiveness of experiential pedagogy on ethical sensitivity, while PAP strengthens the impact of instructor feedback on self-efficacy. Therefore, it is essential to integrate goal-orientation contingencies into the design of sustainability education in order to optimize ethical learning outcomes.

2.6. Research Gaps

Three gaps in the literature have motivated the present study. First, while SDT and SCT have been applied independently, there is a lack of research that integrates these frameworks to compare the relative significance of experiential versus psychological pathways leading to sustainability-oriented ethical outcomes. Second, existing measures of “satisfaction” rarely include items specifically related to sustainability, such as perceived support for addressing trade-offs associated with the Sustainable Development Goals. Third, prior studies predominantly feature Western samples; consequently, evidence from specific rapidly developing regional contexts—such as China—is often underrepresented in the literature, which restricts external validity. This study aims to address these gaps by testing a dual-mediation model within a representative case of rapid developing countries and regions worldwide: transportation engineering students in Hebei Province, China. This focus allows for a deep and contextually grounded analysis of the mechanisms at play, thereby providing culturally contextualized evidence for global ESD policy. To situate our research within the framework of the Sustainable Development Goals (SDGs), we map key variables to relevant SDG targets, mechanisms, and empirical proxies, as illustrated in

Table 1. Links between SDG and key variables in this study.

Variable	SDG Target	Mechanism	Empirical Proxy
EDMC	11.2 “Access to safe, affordable, accessible transport”	Prioritization of safety over cost in design dilemmas	Score on “prioritize pedestrian safety” item
SRS	13.3 “Climate change education”	Willingness to invest extra effort in low-carbon solutions	Score on “greenhouse-gas awareness” item
ETE	4.7 “Education for sustainable development”	Exposure to real-world socio-technical trade-offs	Factor score on “authentic transport project” items
SE	4.4 “Skills for decent jobs”	Confidence to voice ethical concerns in multidisciplinary teams	Score on “express ethics in team” item

.

2.7. Research Hypotheses

To empirically address the aforementioned research questions, this study integrates the SDT and SCT frameworks, derives 10 specific hypotheses, and constructs a dual-mediation model to comprehensively explore how teaching satisfaction influences the enhancement of students’ sustainable ethical competence through different pathways (

Table 2. Mapping of research questions to hypotheses.

Research Question (RQ)	Corresponding Hypothesis (H)	Purpose
RQ1: Does teaching satisfaction significantly enhance students’ EDMC and SRS?	H1, H2 (Direct effects: TS→ EDMC/SRS)	To establish the baseline relationship.
RQ2: Do ETE and SE mediate these relationships?	H3–H8 (Mediation paths: TS → ETE/SE → EDMC/SRS)	To test the dual-mediation mechanisms.
RQ3: Whether achievement–goal model has moderating mediating effect?	H9 (Moderated mediation)	To examine how student motivation types influence the efficacy of each path.
Comparative strength of pathways	H10a, H10b (Contrast of indirect effects)	To determine the relative importance of ETE vs. SE in the given context.

).

H1. Teaching satisfaction is positively associated with Experiential Transformer Engagement (TS→ETE).

H2. Teaching satisfaction is positively associated with self-efficacy (TS→SE).

H3. Experiential Transformer Engagement is positively associated with engineering ethical decision-making competence (ETE→EDMC).

H4. Experiential Transformer Engagement is positively associated with strength of social responsibility (ETE→SRS).

H5. Self-efficacy is positively associated with engineering ethical-decision-making competence (SE→EDMC).

H6. Self-efficacy is positively associated with strength of social responsibility (SE→SRS).

H7. The effect of teaching satisfaction on EDMC is mediated by (a) ETE and (b) SE.

H8. The effect of teaching satisfaction on SRS is mediated by (a) ETE and (b) SE.

H9. Achievement–goal orientation moderates the mediated pathways such that Mastery-Approach goals amplify the ETE path [13,14,15,16], whereas Performance-Approach goals amplify the SE path.

H10a. The indirect effect of teaching satisfaction on EDMC via ETE is significantly stronger than the indirect effect via SE.

H10b. The indirect effect of teaching satisfaction on SRS via ETE is significantly stronger than the indirect effect via SE.

Given the high contextual salience of socio-technical trade-offs in rapidly developing regions, we anticipate that embodied internalization (ETE) will account for a greater proportion of variance than self-efficacy (SE).

3. Participants and Procedure

3.1. Participants and Data Collection

A multi-stage stratified sample of undergraduate transportation engineering students was drawn from five universities in Hebei Province, China, comprising three comprehensive institutions, one polytechnic university, and one industry-oriented institution. Program size and gender ratio were employed as stratification variables to enhance external validity. Eligibility criteria required participants to have completed at least one ethics-related module and to have engaged in a capstone transport design course. Following approval from the College Academic Committee, coordinators distributed anonymous Qualtrics links to 354 students; 27 responses were excluded due to straight-lining or more than 15% missing data, resulting in a total of 327 valid cases (effective response rate = 92.3%). The final sample consisted of 25% sophomores, 40% juniors, and 35% seniors; 38% of the participants identified as female and 62% as male, which is consistent with regional program demographics. Data collection took place during the last two weeks of the Spring 2023 semester to ensure that the participants were familiar with embedded ethics tasks. Participants provided electronic informed consent and received a debriefing within a period of 48 h. No incentives exceeding the equivalent of USD 5 were offered in order to minimize self-selection bias.

3.2. Measurement Development and Validity Evidence

All scales demonstrated good internal consistency reliability (see

Table 3. Reliability and sample items of measurement scales.

Construct	No. of Items	Example item	Cronbach’s α	Source
Experiential Transformer Engagement (ETE)	3	“The course provided real-world transport sites (e.g., intersections, construction zones) for ethical analysis.”	0.82	Adapted from Kolb (2015) [5]
Self-Efficacy (SE)	3	“I feel confident expressing ethical concerns during team design meetings.”	0.83	Schwarzer and Jerusalem (1995) [13]
Engineering Ethical Decision-Making Competence (EDMC)	4	“When cost and safety conflict, I can prioritise ethical principles.”	0.81	ESIT, α = 0.84 (Borenstein et al., 2010) [15]
Strength of Social Responsibility (SRS)	4	“I am willing to spend extra time improving the social value of transport projects (e.g., pedestrian accessibility).”	0.79	SDG 11.2 aligned
Teaching Satisfaction (TS)	12 (4 dims)	“I can see my ethical reasoning ability growing through visible feedback.”	0.89	Second-order factor

Note: Proportions do not sum to 100% because the two mediators are correlated (r = 0.53) and each is calculated against the total effect. All items were rated on a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree). For the complete item wordings, descriptive statistics, and full validity analyses (including CFA, CR, AVE, and discriminant validity), please refer to Appendix A, Appendix B and Appendix C. Confirmatory factor analysis (CFA) supported configural and metric invariance across gender and university types (Δ CFI < 0.01).

). Of particular importance is the ETE scale, which was specifically developed to align with the three-dimensional theoretical framework outlined in Section 2.4. Content validity was established through the systematic design of items that directly correspond to each theoretical dimension: items referencing “real-world transport sites” reflect Situated Authenticity; those involving “multiple senses” capture Multi-Sensory Engagement; and those requiring students to “solve practical problems on-site” embody the Ethical Action Imperative. This intentional item construction ensures that the scale measures the targeted construct—embodied, ethical, and contextually authentic pedagogy—rather than general experiential learning. For the other constructs (e.g., SE, EDMC), established scales were adapted and contextualized to the Chinese transportation engineering education setting (see Appendix A for full items).

3.3. Instrument Development and Adaptation

The data analysis proceeded in two sequential stages to ensure the robustness of the findings. First, a confirmatory factor analysis (CFA) was conducted to evaluate the measurement model, assessing the reliability, validity, and fit of all constructs before testing the structural relationships. The detailed results of the CFA, including model-fit indices and validity tests, are reported in Section 4.1 and Appendix C. Second, the proposed dual-mediation structural model (see Figure 1) was estimated using Mplus 8.8 software, employing the maximum likelihood robust (MLR) estimation method.

All survey instruments were originally developed based on established theoretical scales (see sources in Table 3). To ensure conceptual equivalence and cultural relevance for the Chinese student population, rigorous translation and back-translation was conducted by a bilingual expert panel. Furthermore, all items were contextualized within the domain of transportation engineering and sustainability dilemmas pertinent to the Chinese urbanizing context (e.g., referencing “bus-lane implementation,” “informal-settlement intersections”). This process ensured that the constructs measured were both theoretically sound and contextually meaningful for the participants. The final English versions used in this manuscript are the result of this adaptation process. The complete survey instrument in bilingual format (Chinese–English) is provided as Supplementary Material.

3.4. Data Statistical Analysis

The double mediation model shown in Figure 1 was estimated using the MLR estimation method based on Mplus 8.8 software.

Firstly, the multiple kurtosis coefficient of Mardia is 5.4, indicating mild non normality in the data, and the MLR estimation method effectively addresses this issue.

Secondly, the model fit is comprehensively evaluated through the following indicators: chi square degree of freedom ratio (χ²/df < 3), comparative fit index (CFI), Tucker–Lewis index (TLI), approximate root mean square error (RMSEA), and standardized root mean square residual (SRMR). The evaluation criteria refer to the recommendations of Hu and Bentler (1999) [17]: CFI/TLI ≥ 0.90, RMSEA ≤ 0.08, SRMR ≤ 0.08. The significance test of the mediating effect was conducted using the bias corrected Bootstrap method [18], with 5000 replicates set. If the 95% confidence interval of the indirect effect does not include zero, it is considered significant.

In addition, to test the moderating effect of achievement goals, the study first used K-means clustering analysis (K = 4) to distinguish participants into a mastery goal-oriented group (n = 98) and performance goal-oriented group (n = 85), with a silhouette coefficient of 0.42. Subsequently, multiple sets of structural equation models were used for analysis, and before comparing the structural paths between groups, the measurement invariance was ensured through the Δχ ² test (p < 0.05). The detailed results of cluster analysis are shown in Appendix A (

Table A2. Achievement goal orientation scale and K-means clustering results.

Goal-Orientation Type (Abbreviation)	Sample Item	Scale Origin
Mastery-Approach (MAP)	“I study hard to master professional knowledge and skills.”	Adapted from Elliot and McGregor (2008) Achievement Goal Scale [16]
Performance-Approach (PAP)	“I value achieving good grades and rankings in the course.”
Mastery-Avoidance (MAV)	“I study hard to avoid not being able to understand important concepts.”
Performance-Avoidance (PAV)	“I try my best to avoid underperforming in the course.”

).

Finally, the effect size evaluation of indirect effects refers to the κ² index proposed by Preacher and Kelley (2011) [19]. In this study, the κ² value of the path ETE → EDMC was 0.21, and the κ² value of the path SE → EDMC was 0.18, both within the moderate effect range.

3.5. Procedural and Statistical Remedies for Common Method Variance (CMV)

Recognizing the potential for common-method variance inherent in cross-sectional self-report data, we implemented both procedural and statistical remedies in accordance with established guidelines [20]. Procedurally, we ensured respondent anonymity to mitigate evaluation apprehension, utilized clear and distinct scale anchors, and separated the measurement of predictor and criterion variables within the survey. Statistically, we employed three complementary tests:

• Harman’s Single-Factor Test: The first unrotated factor accounted for 38% of the variance, which is below the 50% threshold.
• Marker-Variable Technique: By using a theoretically unrelated variable—“frequency of library visits”—as a marker (r < 0.05 with all substantive constructs), we controlled for its variance and found that all significant paths remained stable (Δβ < 0.02).
• Unmeasured Latent Method Factor (ULMF) Test: We conducted a more rigorous assessment by incorporating an uncorrelated method factor into the measurement model.

As detailed in Appendix E, there was only marginal improvement in model fit (ΔCFI = 0.008), while all substantive factor loadings remained significant and exceeded 0.70, indicating that common method variance did not account for a substantial portion of covariance among constructs.

3.6. Ethics Approval

This study was reviewed and approved by the College Academic Committee of the School of Traffic and Transportation at Shijiazhuang Tiedao University under approval letter No. CE-2024-03-10. Informed consent was obtained from all participants involved in this research.

4. Results

4.1. Measurement Model

A confirmatory factor analysis (CFA) employing robust maximum likelihood estimation supported the proposed six-factor structure (χ²/df = 1.93, CFI = 0.94, TLI = 0.93, RMSEA = 0.057, SRMR = 0.042). All standardized factor loadings were greater than 0.70 (p < 0.001), and composite reliabilities ranged from 0.81 to 0.92 (see Appendix B). Discriminant validity was established using the Fornell–Larcker criterion: the square root of the average variance extracted (AVE) for each latent construct exceeded its correlations with any other construct (refer to Appendix C).

4.2. Structural Model

The structural model demonstrated an acceptable fit (χ²/df = 2.17, CFI = 0.94, TLI = 0.93, RMSEA = 0.062, SRMR = 0.045). Figure 1 illustrates the standardized path coefficients clearly indicating that the teaching satisfaction (TS) significantly predicted both mediators—Experiential Transformer Engagement (ETE; β= 0.58, p < 0.001) and self-efficacy (SE; β= 0.52, p < 0.001)—thereby supporting hypotheses H1 and H2, respectively.

Furthermore, ETE positively influenced engineering ethical-decision-making competence (EDMC; β= 0.47, p < 0.001) as well as the strength of social responsibility (SRS; β= 0.42, p < 0.001), confirming hypotheses H3 and H4 accordingly.

Similarly, SE exerted significant effects on EDMC (β= 0.38; p < 0.001) and SRS (β= 0.35; p <0.001), corroborating hypotheses H5 and H6.

The direct paths from TS to EDMC (β = 0.15; p = 0.013) and SRS (β = 0.13; p = 0.021) were found to be significant but modest in magnitude, thereby lending support to hypotheses H7 and H8. The complete structural model fit indices along with detailed hypothesis testing results—including path coefficients, standard errors, composite reliability values, and corresponding p-values—are documented in Appendix D.

4.3. Mediation Analysis

Bias-corrected bootstrapping with 5000 samples indicated that both indirect pathways were statistically significant (The specific results are shown in

Table 4. Standardized indirect effects and 95% bias-corrected confidence intervals (5000 bootstraps).

Path	Point Estimate	95% CI	Proportion of Total Effect (%)
TS→ ETE→ EDMC	0.27	[0.183, 0.362]	64.3
TS→ SE → EDMC	0.20	[0.134, 0.281]	47.6
TS→ ETE → SRS	0.24	[0.156,0.331]	64.9
TS → SE → SRS	0.18	[0.121, 0.254]	48.7

Note: Proportions do not sum to 100% because mediators are correlated and total effects include small direct paths.

).

As shown in Table 4, the pathway from TS to ETE to EDMC accounted for 64.3% of the total effect, while the pathway from TS to SE to EDMC explained 47.6%. Similarly, ETE and SE mediated 64.9% and 48.7% of the total effect on SRS, respectively. Given that the two mediators are correlated (r = 0.53), a contrast test was conducted; however, the difference in specific indirect effects was not statistically significant (Δind = 0.02, 95% CI [–0.06, 0.10]), suggesting that both pathways hold equal importance for sustainability-oriented ethical outcomes.

To test H10a and H10b, we conducted pairwise contrasts of the two indirect effects. The difference between the ETE and SE indirect effects was Δind = 0.27–0.20 = 0.07 for EDMC, 95% CI [0.01, 0.13] (excluding zero), and Δind = 0.24–0.18 = 0.06 for SRS, 95% CI [0.005, 0.12] (excluding zero). Thus, both H10a and H10b are supported: the ETE pathway is significantly stronger than the SE pathway in the Hebei context.

In addition, the standardized indirect effect (κ²) of the two mediation paths also showed a consistent trend, as shown in Figure 2. For the two result variables (edmc and SRS), the path via ete (κ² = 0.21, 0.19) was greater than that via se (κ² = 0.15, 0.13), which further supported the above conclusion.

In addition, according to Podsakoff et al.’s (2012) methodology [20], Harman’s single-factor test, which revealed that the first unrotated factor accounted for only 38% of the variance—below the critical threshold of 50%—indicating that common method variance is not a predominant bias.

4.4. Moderation Analysis

Multi-group structural equation modeling (SEM) revealed significant differences between the Mastery-Approach (MAP; n = 98) and Performance-Approach (PAP; n = 85) groups across two structural paths (Δχ² = 6.54, p = 0.011). The between-group differences in key structural paths are visually summarized in Figure 3, which presents a bar chart comparison of the standardized path coefficients (β) for the Mastery-Approach (MAP) and Performance-Approach (PAP) groups. As illustrated, the path from ETE to EDMC was significantly stronger for MAP students (β = 0.54) than for PAP students (β = 0.39). Conversely, the path from SE to SRS was stronger for PAP students (β= 0.43) than for MAP students (β = 0.29). These differential patterns confirm the moderating role of achievement–goal orientation (H9 supported).

In order to more accurately compare the strength of mediation effect between groups, we further calculated the standardized indirect effect (κ²) of grouping, and the results are summarized in Figure 4. The data showed that for the ETE → EDMC pathway, the effect of map group (κ² = 0.24) was stronger than that of PAP group (κ² = 0.16); For SE→EDMC pathway, the effect of PAP group (κ² = 0.20) was stronger than that of map group (κ² = 0.14). No significant difference was found in other structural pathways between groups (p > 0.05).

5. Discussion

This study empirically validates a dual-mediation model that elucidates how teaching satisfaction translates into sustainability-oriented ethical competencies within the distinctive context of rapid developing countries and regions worldwide in Hebei Province, China. The findings confirm that both ETE and SE serve as significant mediators, yet with a marked difference in their relative strength: the ETE pathway (β = 0.47, p < 0.01) exerted a significantly stronger mediating effect than the SE pathway (β = 0.38, p < 0.01). [H10a, H10b]. Furthermore, the students’ achievement–goal orientation moderates these mechanisms [H9], with the Mastery-Approach (MAP) students benefiting more from ETE and the Performance-Approach (PAP) students deriving greater benefit from SE-building feedback. These insights offer a nuanced, theory-grounded blueprint for engineering ethics education tailored to contexts where sustainability trade-offs are immediate and salient.

• The dominant influence mechanism of ETE pathway

The predominance of the ETE effect can be attributed to its congruence with the high-stakes, visible socio-technical dilemmas characteristic of rapidly developing regions. As defined in this study, ETE is characterized by Situated Authenticity, Multi-Sensory Engagement, and an ethical action imperative [Section 2.4]. In a setting where students routinely encounter real-world conflicts, abstract ethical principles become tangible and emotionally salient. This immersive, value-laden experience aligns closely with SDT [4], effectively satisfying students’ needs for autonomy, competence, and relatedness, thereby fostering a deeper internalization of sustainability norms. When ethical stakes are felt, not just discussed, the pathway from satisfaction to internalized competence becomes more potent, a notion supported by situated learning perspectives [5].

• The complementary role of SE path

Although relatively less dominant in this specific context, the substantial mediation via self-efficacy (≈48%) underscores a complementary psychological mechanism essential for sustained ethical agency. Rooted in SCT, SE represents the belief in one’s capability to execute ethical actions despite constraints [6]. This empowerment is crucial for translating internalized values into persistent behavior, especially beyond the classroom, in future professional settings where engineers face pressures. Therefore, while ETE may be more effective for initial value internalization in high-salience contexts, SE functions as the vital psychological resource that enables students to uphold and act upon these principles over the long term [21], making it an indispensable component of a holistic ethics education.

• Comparative positioning with previous studies

To place the results of this study in a broader academic context, we selected several key studies for comparison (see

Table 5. Selected comparative studies on pathways to ethical and sustainability outcomes in educational and professional contexts.

Study	Country/Type	Discipline	Sample Size (N)	Research Design	Key Path/Relationship	Coefficient Value
Teather and Etterson (2023) [7]	Australia/Cross-sectional	Higher Education	472	Survey	Values–behavior gap (Attitude Engagement)	r = 0.29
Walumbwa et al. (2011) [8]	USA/Multilevel	Organizational Behavior	336	Field study	Self-efficacy→Job performance	β = 0.38
Fini et al. (2018) [9]	USA/Higher Education	Engineering Education	127	Quasi-experimental	Project-based Learning → Sustainability Learning Outcomes	*d* = 0.72 (Cohen’s d)
Wiek et al. (2011) [10]	International/Framework	Sustainability Science	-	Theoretical	Key competencies Action (estimated)	-
This study	China/SEM	Engineering Education	327	Cross-sectional	ETE → EDMC SE → EDMC	β = 0.47 ** β = 0.38 **

Note: β represents the standardized path coefficient, r represents the Pearson correlation coefficient, and *d* represents Cohen’s d value (standardized mean deviation). ** Expressing p < 0.01, the selected study provides a comparative effect measure or theoretical benchmark related to the dual pathway model. All coefficients are derived from the cited original research. For detailed definitions and explanations, please refer to Section 5.

).

When selecting comparative literature, three criteria are used: method relevance (using empirical designs or theoretical frameworks comparable to the dual-path model of this study), construct relevance (focusing on self-efficacy, sustainability attitudes, ethical reasoning, or ability development), and baseline value (providing standardized effect measures such as β, r , or Cohen’s * d * for comparison). In subsequent comparisons, β refers to the standardized path coefficient, r is the Pearson correlation coefficient, and * d * is the Cohen d value (standardized mean deviation).

The differential moderation effect, as visually summarized in Figure 3, clarifies for whom each pathway is most effective. As illustrated in the bar-chart comparison, students with a Mastery-Approach (MAP) orientation, driven by intrinsic growth, show a stronger ETE→EDMC link (β = 0.54 vs. 0.39), as experiential learning aligns with their desire for deep understanding [17]. Conversely, students with a Performance-Approach (PAP) orientation, focused on demonstrating competence, exhibit a stronger SE→SRS link (β = 0.43 vs. 0.29), as clear feedback and visible skill-building bolster their confidence. This finding necessitates a differentiated instructional strategy; curricula should offer open-ended, authentic ETE tasks to engage MAP learners while incorporating structured, feedback-rich exercises to build the SE of PAP learners.

As synthesized in Table 5, prior studies have established independent links between teaching satisfaction, self-efficacy, experiential learning, and ethical outcomes, predominantly in Western or post-materialist contexts. Our study advances this discourse by being among the first to quantitatively compare the dual-pathway effects of ETE and SE on EDMC in a comprehensive model, and tested them in regions where engineering ethics education is under-represented, within an integrated model and to do so within the under-represented context of rapid developing countries and regions worldwide. The finding that ETE contributes 64% of the indirect effect—a proportion notably higher than the effect sizes (β ≈ 0.20–0.38) reported in studies from Western societies—highlights context as a critical boundary condition. This suggests that the pedagogical leverage of embodied experience is amplified where infrastructure dilemmas are first-generation and ethically stark, thereby extending the applicability of SDT and SCT beyond Western, post-materialist educational settings.

Although the model constructed in this study still adopts commonly used patterns from Western engineering ethics and SDE, the outstanding contribution of this study lies in the systematic integration and situational adaptation of these methods in areas where engineering ethics education in Chinese universities is weak. In many developing regions, engineering ethics education is still in its infancy, and the direct transferability of the “Western experience” is often limited by the immediacy of different resources, student motivation characteristics, and sustainability trade-offs. Our research validates the dual pathways of ETE and SE in the engineering education environment in China, providing reference for engineering ethics education in rapidly developing countries and regions like China.

• Specific teaching inspirations and implementation plans

Based on the findings of this study, that ETE promotes ethical internalization through real-life situations, while SE empowers behavior through structured feedback, we designed and implemented a four week “design defense” studio module (see

Table 6. The FIRST ABET-EAC-2025-aligned, SDG-indexed 4-week ethics-studio module for transportation engineering: translating ETE principles into measurable outcomes.

Week	Learning Task	Ethical Dilemma Embedded	Deliverable	Mastery-Oriented Rubric (Excerpt)
1	Field audit of informal-settlement intersection	Competing demands: bus-lane vs. vendor relocation	5 min video log	“Identify at least two stakeholder value conflicts with evidence”
2	Generate three design alternatives	Trade-off: cost, safety, equity	Design memo	“Use SDG 11.2 indicator to justify priority ranking”
3	Public consultation role-play	Voice of elderly and disabled residents	Reflection journal	“Document how feedback altered your weighting of safety vs. cost”
4	Design-and-defense panel	External examiner asks “Would you relocate 50 households?”	Slide deck and Q & A	“Defend final decision with at least two ethical theories”

Note: The module’s impact is assessed by the change in the SDG 11.2.1-based indicator described in the main text.

) to replace traditional ethics lectures.

This module is designed to closely correspond to the three core features of ETE: the first week focuses on the authenticity of the situation, the second and third weeks run through the mandatory ethical actions, and the fourth week integrates multi-sensory participation. To objectively evaluate the effectiveness of the teaching module, we have expanded the connotation of indicator 11.2.1 of the United Nations Sustainable Development Goals (SDGs) [22], defining it as:

“The change in the proportion of people who can use public transportation within a 10 min walk and live in areas with an annual road injury rate of less than 5/100,000 before and after intervention measures.“

This definition not only focuses on accessibility of transportation, but also integrates the core ethical dimension of safety, requiring students to propose engineering solutions that take into account both. The calculation is based on real data: population distribution (Baidu heat map), road safety (2023 accident public data set of the Municipal Public Security Bureau), and bus network (GIS layer of bus stops). We set an increase of ≥5% in this proportion as the threshold for generating “effective improvement” in the plan, thereby transforming abstract ethical-decision-making ability into quantifiable and verifiable real-world impact indicators.

Finally, as shown in

Table 7. Mapping the 4-week “design-and-defense” studio onto ABET EAC 2025 ethics and sustainability criteria.

Studio Week	Embedded Ethical Dilemma	Deliverable	ABET 2025 Criterion	Mapped SDG Indicator
1	Bus-lane vs. vendor relocation	5 min video log	3.f. “recognize ethical conflicts”	11.2.1% population with safe access
2	Cost–safety–equity trade-off	Design memo	3.h. “assess sustainability impacts”	11.2.1 + 13.2.2 CO2 reduction
3	Elderly and disabled consultation	Reflection journal	3.g. “listen to diverse stakeholders”	11.7.1 public-space inclusion index
4	External examiner defense	Slide deck and Q&A	3.i. “justify decisions using ethical theories”	11.2.1 post-intervention safety audit

, each step of the module accurately corresponds to the ethical and sustainability standards of ABET Engineering Certification (2025) and is mapped to specific SDG indicators, forming a curriculum-implementation framework that is both rooted in theory and internationally transferable, providing an operational example for systematically cultivating engineers with ethical judgment and action confidence in complex reality.

• Limitations and future research

This study has several limitations that warrant consideration. First, the cross-sectional design precludes definitive causal inferences regarding the proposed mediation pathways, despite them being grounded in established theory. Second, the regional specificity of the sample (Hebei Province) limits the generalizability of the findings to other cultural or developmental contexts. Third, a key methodological limitation pertains to uncontrolled potential covariates. The model does not account for confounding variables such as prior ethical training, personality traits, or general academic ability, all of which may jointly influence teaching satisfaction, experiential engagement, self-efficacy, and the ethical outcomes measured. Although the sampling criteria required completion of at least one ethics module to partially mitigate differences in foundational knowledge, this control is insufficient. Future research should explicitly measure and statistically control for these covariates, or preferably, adopt longitudinal or experimental designs to establish clearer causal relationships and minimize confounding bias in the estimated paths. To establish broader applicability, future work should test this model across diverse regions within China and in other rapidly developing countries to identify boundary conditions and advance a more contextually sensitive theory of ethics pedagogy.

6. Conclusions

This study established and validated a dual-mediation model that elucidates the psychological mechanism by which teaching satisfaction promotes the sustainable development of ethical competence among engineering students. The research findings confirm that this influence is transmitted through two different but complementary pathways: Experiential Transformer Engagement (ETE), which promotes deep internalization of moral values, and Self-Efficacy (SE), which establishes the motivational confidence required for moral behavior.

The key finding of this study is the significant difference in strength between the two mediating pathways linking teaching satisfaction to engineering ethics competence, as evidenced in the specific context of Hebei Province, China. The results confirm that the ETE pathway (βββ = 0.47, p < 0.01) is significantly more effective than the SE pathway (βββ = 0.38, p < 0.01). This reaffirms that in resource-limited, rapidly developing environments, internalizing values through situated experiences exerts a greater influence on the development of students’ professional ethics than interventions focused solely on building confidence. Furthermore, this process is moderated by students’ achievement–goal orientation: Mastery-Approach (MAP) students derive greater benefit from ETE, while Performance- Approach (PAP) students respond more strongly to SE-enhancing interventions.

These findings offer critical guidance for regions where engineering ethics education is still emerging. By empirically clarifying the mechanism through which teaching satisfaction transforms into professional competence and social responsibility, this study enables the contextual adaptation of established Western models of engineering ethics education and SDGs higher education frameworks. Therefore, engineering education programs aim to effectively cultivate personnel with ethical competencies geared towards sustainable development, they must systematically construct a dual-path-driven teaching environment. This can be achieved through strategies such as (1) designing embodied teaching modules that integrate authenticity, multi-sensory engagement, and ethical decision-making to activate the value internalization process emphasized by SDT; and (2) embedding structured feedback mechanisms that provide clear benchmarks for ability growth, particularly for performance-oriented students, to consolidate the behavioral execution and persistence emphasized by SCT. Ultimately, this approach serves to develop locally relevant engineering ethics curricula and provides a contextualized, evidence-based way to prepare future engineers who are not only adaptable to local conditions but also equipped with the ethical commitment and practical capability needed to address complex sustainability challenges.