Integrating Sustainability Learning in Thai Higher Education: Impacts on Graduate Students’ Knowledge, Attitudes, and Engagement

The world is currently experiencing a ‘polycrisis’ comprising climate change, heightened social disparities, and fragile economic systems, which call for urgent systemic responses [1]. Once an idealistic goal, sustainable development has now become an imperative for survival, with many higher educational institutions positioned to lead the transformation.

The United Nations Educational, Scientific, and Cultural Organization (UNESCO) has frequently advocated that higher education institutions shift toward becoming incubators of sustainability literacy [2], which requires blending knowledge, skills, and attitudes for addressing complex global issues [3].

In an effort to embrace these sustainability goals, Thailand has implemented its national Bio‑Circular‑Green (BCG) Economy model [4], which prioritizes sustainable growth. Thai universities have also begun incorporating the Sustainable Development Goals (SDGs) into their strategic plans [5], but unfortunately, gaps remain between policy and practice. One such gap is the reliance of instructors and their courses on traditional lectures, with students rarely engaging in hands‑on, community‑based learning that would connect sustainability concepts to real‑world problems [6, 7]. Studies have shown that integrated learning for global citizenship can build knowledge, values, and attitudes. However, few Thai studies have tested a formal model linking curriculum design, student engagement, and sustainability outcomes using robust quantitative methods.

Despite growing interest in Education for Sustainable Development (ESD) in Thailand [8, 9], several gaps remain unaddressed. First, no empirical study has tested a structural model linking Sustainability Learning Integration (SLI) – defined as the deliberate combination of sustainability content, experiential activities, and structured reflection – to graduate students' sustainability knowledge (SSK) and attitudes (SSA) with student engagement as a potential mediator [10-12]. Second, existing Thai studies on sustainability education have predominantly employed qualitative methods or simple pre-post design without a control group, limiting their ability to establish causal pathways or generalize findings. Third, while international research has demonstrated positive effects of engagement on learning outcomes, no study has validated these relationships within the Thai higher education context using structural equation modeling (SEM). Consequently, university administrators and curriculum planners lack evidence-based guidance on whether to invest limited resources in curriculum redesign, faculty training, or experiential learning infrastructure. The present study addresses these gaps by: (a) proposing and testing a theoretically grounded SEM model (H1–H6), (b) using a large sample (n = 520) from six Thai universities, and (c) employing a mixed-methods sequential explanatory design to both quantify effects and explain underlying mechanisms.

3.1 Research design

A mixed‑methods sequential explanatory design was employed [28]. First, quantitative data were collected from a large sample of graduate students to test the SEM. Second, qualitative interviews were conducted with a subset of participants to explain and elaborate on the quantitative results [29].

3.1.1 Connection between quantitative and qualitative phases

Following the sequential explanatory design [28, 29], the quantitative phase was conducted first. Data from 520 graduate students were analyzed using SEM to test the hypothesized relationships (H1--H6). The results of the quantitative phase directly informed the qualitative phase in three specific ways.

First, the SEM results revealed that SE significantly mediated the relationships between SLI and both outcome variables (SSK and SSA). Therefore, interview questions were designed to probe how and why engagement functions as a mediator. Participants were asked to describe specific learning activities that increased their behavioral, emotional, or cognitive engagement (see Appendix A, Questions #1, #3, #6).

Second, the quantitative data showed variation in engagement levels across participants. Using a purposive sampling strategy, 24 students were selected from three engagement categories: high engagement (n = 8, top 20% of SE scores), medium engagement (n = 8, middle 40-60%), and low engagement (n = 8, bottom 20%). This stratification allowed the research team to compare how SLI experiences differed across engagement levels.

Third, the standardized path coefficients indicated that the SLI→SSK pathway (β = 0.62) was slightly stronger than the SLI→SSA pathway (β = 0.57). The qualitative interviews therefore included questions exploring why knowledge acquisition might be more directly influenced by SLI than attitude change. Interviewers used Question #4 from Appendix A ("Did you notice any changes in your attitudes or perspectives toward sustainability issues? How?"), followed by the specific probe: "Did your beliefs about sustainability change as much as your understanding? Why or why not?"

All qualitative interview questions (see Appendix A for the full 12-question protocol) were derived directly from these quantitative findings. Follow-up probes were used flexibly to elicit deeper responses. This ensured that the qualitative phase served to explain, elaborate on, and contextualize the statistical results rather than simply collecting parallel data.

3.2 Participants and sampling

The target population comprised graduate students (master's and doctoral) enrolled in six public universities in Thailand who had participated in sustainability-related courses or activities [30]. A multistage sampling procedure was employed to ensure representativeness and replicability, consisting of four sequential stages.

Stage 1 (Region selection). Four geographic regions of Thailand were purposively selected to capture regional diversity: Bangkok Metropolitan Region (central), Northern Region, Northeastern Region, and Southern Region. These regions represent distinct socioeconomic and environmental contexts relevant to sustainability education.

Stage 2 (University selection). From each of the four regions, six public universities were randomly selected from the Office of the Higher Education Commission's list of accredited institutions. A total of six universities participated (two from Bangkok, and one from each of the other three regions). University names are withheld to protect participant confidentiality.

Stage 3 (Course identification). Within each university, all graduate-level courses offered during the second semester of the 2024 academic year were reviewed by the research team. Courses were eligible for inclusion if the syllabus explicitly contained at least three of the following SLI elements: (a) SDG-related content covered across multiple weeks, (b) project-based learning addressing real-world sustainability problems, (c) community engagement activities, (d) structured reflective assignments (e.g., journals, portfolios), and (e) interdisciplinary sustainability projects. A total of 42 courses met these criteria across the six universities.

Stage 4 (Student sampling). From each eligible course, students were randomly sampled using a simple random sampling method (lottery technique). A total of 520 students agreed to participate and completed the survey, representing a response rate of 78.2% (520 of 665 invited). The sample included students from four faculties: Education (n = 136, 26.15%), Business Administration/Economics (n = 133, 25.58%), Science and Technology (n = 128, 24.62%), and Social Sciences/Humanities (n = 123, 23.65%).

Qualitative subsample. For the qualitative phase, 24 students were purposively selected from the quantitative sample using stratified purposive sampling. Participants were selected based on their student engagement (SE) scores, with eight students drawn from each of three categories: high engagement (top 20% of SE scores, >4.2 on a 5-point scale), medium engagement (middle 40-60%, SE scores between 3.5 and 4.0), and low engagement (bottom 20%, SE scores <2.9). This stratification was designed to capture a range of experiences and to understand how engagement levels shape the SLI→SSK and SLI→SSA pathways. All 24 invited students agreed to participate in interviews.

3.3 Instruments

All quantitative measures were first developed in Thai. Subsequently, they were translated into English and then back-translated into Thai to ensure semantic equivalence. Five experts assessed content validity, and each item scored above 0.80 on the Index of Item‑Objective Congruence (IOC) [31]. Cronbach's alpha exceeded 0.85 for each scale, indicating acceptable internal consistency.

3.3.1 Sustainability learning integration scale

The SLI scale was developed de novo for the present study. No existing scale was directly adapted. Item generation was guided by two theoretical frameworks: Kolb's ELT [19, 20] and Mezirow's TLT [22, 23]. The operational definition of SLI was constructed as the deliberate combination of three components: (a) sustainability content integration (explicit coverage of SDGs and sustainability concepts), (b) experiential learning processes (project-based, community-engaged, or problem-based activities), and (c) structured reflection opportunities (journals, discussions, or portfolios).

An initial pool of 24 items was generated based on this definition. Items were reviewed by five experts in sustainability education and educational measurement for content validity. After eliminating redundant or ambiguous items, 12 items remained, each with an IOC score above 0.80. The final 12-item scale covered three sub-dimensions (four items each): content integration (e.g., "SDGs were explicitly discussed in the course"), learning process (e.g., "We worked on real-world sustainability problems"), and experiential activities (e.g., "I participated in a community project related to sustainability") [32]. Students answered on a five-point Likert scale from "strongly disagree" (1) to "strongly agree" (5).

All items were measured on a five-point Likert scale ranging from 'strongly disagree' (1) to 'strongly agree' (5).

3.3.2 Student engagement scale

The authors took 15 items from Fredricks [24] and adapted them for our context. The items measured behavioural engagement (“I actively participated in class discussions”), emotional engagement (“I felt excited about sustainability topics”), and cognitive engagement (“I tried to relate what I learned to my own life”). Again, a five‑point Likert scale was used.

3.3.3 Sustainability knowledge test

The research team put together 25 multiple‑choice questions. They covered SDGs, systems thinking, climate change, social equity, and the circular economy. Reliability was acceptable – the Kuder‑Richardson 20 coefficient was 0.82.

3.3.4 Sustainability attitudes scale

Ten items measured values and beliefs. Example statements: “I believe that protecting the environment is more important than economic growth” and “I am willing to change my lifestyle to reduce my carbon footprint” [33].

3.3.5 Interview protocol

For the qualitative part, we used semi‑structured questions. The goal was to explore how SLI influenced students’ understanding, feelings, and actions regarding sustainability. A typical question was: “Can you describe a specific learning activity that changed the way you think about sustainability?”

3.4 Data collection

Quantitative data were collected during the second semester of the 2024 academic year. With permission from university ethics committees, surveys were administered online (via Google Forms) and in paper format, taking approximately 30 minutes. Participation was voluntary and anonymous. Qualitative interviews (45‑60 minutes each) were conducted via video call, audio‑recorded, and transcribed verbatim.

3.5 Data analysis

Quantitative data were analysed using SPSS for descriptive statistics, and LISREL 9.10 was used for SEM. The model fit was assessed using χ²/df, CFI, GFI, SRMR, and RMSEA [34]. Mediation effects were examined using bootstrapping (5,000 resamples). Qualitative data were analysed using thematic analysis, with codes and themes identified inductively from the transcripts.

This study examined how integrating sustainability learning into graduate courses affects SSK, SSA, and SE, with student engagement as a potential mediator. The results strongly support the proposed model and offer several theoretical and practical insights for sustainable development planning in higher education.

5.1 Direct effects of Sustainability Learning Integration on knowledge, attitudes, and engagement

The finding that SLI had a strong positive effect on SSK (β = 0.62, p < .001) and SSA (β = 0.57, p < .001) confirms and extends prior international research [15-16, 36] while providing the first empirical validation of such relationships in the Thai higher education context [37]. The magnitude of these effects is noteworthy: SLI explained 58% of the variance in knowledge and 52% of the variance in attitudes. The course design requires students to study sustainability materials throughout the term while engaging in experiential learning opportunities in real-world community contexts [38]. In practical terms, this means that students who experienced well-designed SLI – combining content, experiential activities, and reflection – scored substantially higher on both outcome measures than those who did not [39].

From a theoretical perspective, these results align with Kolb's Experiential Learning Theory [19, 20]. Kolb argued that learning is most effective when learners cycle through concrete experience, reflective observation, abstract conceptualization, and active experimentation. The SLI scale in this study explicitly measured all four components. Therefore, the strong SLI→SSK pathway suggests that sustainability knowledge is not simply transmitted from instructor to student but is actively constructed through direct engagement with real-world problems. These findings challenge traditional lecture-dominated approaches still common in many Thai universities [6, 7].

The slightly weaker effect of SLI on attitudes (β = 0.57 versus 0.62 for knowledge) is also theoretically meaningful. Mezirow's Transformative Learning Theory [22, 23] posits that attitude and value change are more difficult to achieve than knowledge acquisition because they require learners to critically examine and potentially abandon deeply held assumptions. In the Thai cultural context [37], where hierarchical relationships and saving face are highly valued, challenging one's own worldview may be particularly uncomfortable. The qualitative findings support this interpretation: students who reported attitude change almost always described an emotionally charged experience (e.g., "I almost cried," "I broke down") rather than a purely intellectual one.

The strong direct effect of SLI on SE (β = 0.68, p < .001) indicates that how sustainability is taught matters as much as what is taught. Students became more engaged – behaviourally, emotionally, and cognitively – when courses incorporated experiential learning and structured reflection. This finding has practical implications for faculty development: training instructors to design and facilitate experiential learning activities may yield higher returns on investment than simply adding sustainability content to existing lectures.

5.2 The mediating role of student engagement

The significant mediation effects (SLI → SE → SSK: indirect β = 0.24, p < .01; SLI → SE → SSA: indirect β = 0.20, p < .01) demonstrate that engagement is not merely a desirable outcome but a mechanism through which SLI produces learning gains. In other words, SLI works partly because it increases student engagement, and increased engagement then leads to better knowledge and attitudes.

This finding is consistent with Fredricks' three-dimensional model of engagement [24] and with prior research showing engagement as a mediator of innovative teaching practices [25-27]. However, the partial nature of the mediation (direct effects remained significant even after accounting for engagement) suggests that SLI also influences knowledge and attitudes through other pathways not captured in this model. Possible alternative mechanisms include increased self-efficacy, development of sustainability identities, or shifts in moral reasoning. Future research should explore these possibilities.

For sustainable development planning, the mediation finding has a clear implication: universities should monitor student engagement as an early indicator of sustainability learning outcomes. Engagement is easier to measure in real-time (e.g., through classroom observations, learning analytics, or brief surveys) than knowledge or attitude change, which typically require end-of-term assessments. Low engagement scores mid-semester could trigger instructional interventions before it is too late.

5.3 Implications for Thailand's Bio-Circular-Green economy model and sustainable development goals

Thailand's national BCG Economy model [4] prioritizes sustainable growth but has been criticized for remaining at the policy level without sufficient implementation in educational settings [5, 40]. The present study provides empirical evidence that specific, actionable curriculum designs – not just policy statements – can produce measurable sustainability learning outcomes.

Specifically, the finding that community-based projects and structured reflection were most effective (from qualitative data) suggests that Thai universities should allocate resources to: (a) developing partnerships with local communities and NGOs to provide authentic project sites, (b) training faculty in experiential and reflective pedagogies, and (c) revising curriculum approval processes to require explicit SLI elements rather than generic sustainability statements.

These actions align directly with SDG 4 (Quality Education), which calls for ensuring all learners acquire knowledge and skills needed for sustainable development, and SDG 12 (Responsible Consumption and Production), which emphasizes education for sustainable lifestyles [41]. The model's high explained variance (R² = 0.58 for knowledge, 0.52 for attitudes) indicates that well-designed SLI can achieve substantial progress toward these goals even within a single semester.

5.4 Comparison with prior Thai studies

Previous Thai research on sustainability education has been predominantly qualitative [17, 18] or used simple pre-post designs without control groups. While these studies provided valuable initial insights, they could not establish causal pathways or generalize findings beyond small samples. The present study advances the field by: (a) testing a theory-driven SEM model, (b) using a large sample (n = 520) across multiple universities and disciplines, and (c) employing a mixed-methods design that both quantifies effects and explains underlying mechanisms.

The qualitative findings echo Oe et al. [17], who found that in Japan community-based projects and reflective journals increased sustainability awareness. However, the present study goes further by demonstrating that these effects are partially mediated by engagement and that engagement itself is predicted by SLI. This suggests that prior Thai studies may have underestimated the importance of engagement as an intermediate outcome.

5.5 Limitations acknowledged in interpretation

Several limitations should be considered when interpreting these findings. First, the cross-sectional design prevents strong causal claims. Although the SEM results are consistent with causal pathways hypothesized from theory, longitudinal data would provide stronger evidence. Second, all measures were self-reported except the knowledge test, raising the possibility of common method bias. Third, the sample included only graduate students from six public universities; findings may not generalize to undergraduate or vocational students, or to private universities. Fourth, the qualitative sample (n = 24) was drawn from the same population as the quantitative sample, limiting the diversity of perspectives. Future research should address these limitations through longitudinal designs, multi-informant measures (e.g., peer or instructor ratings of engagement), and more diverse samples.