Scale validation of the “Key Competences in Sustainability” Indicator Tool (KCiS-IT) among business and economics students in seven European HEIs Open Access

Global resource pressures have intensified the need for more strategic and sustainable practices. Despite international commitments, only 16% of the Sustainable Development Goal targets are currently on track for 2030, with the remaining 84% showing limited progress or regression (Sachs et al., 2024). This uneven progress, led by Nordic and BRICS (i.e. Brazil, the Russian Federation, India, China and South Africa) countries, while poorer fall behind, highlights the long-term investment challenge of sustainable development, which partly hinges on reforming global financial structures and redistributing resources more equitably (United Nations [UN], 2023). Tackling complex sustainability challenges requires a shift in mindset toward systems thinking and an integrated approach to environmental, social and economic issues (Liu et al., 2015; United Nations Educational, Scientific and Cultural Organization [UNESCO], 2023). This transformation extends beyond policy change; it calls for a cultural shift in education. Higher education plays a pivotal role in preparing learners with sustainability competences that align with societal transitions, institutional strategies and evolving workforce demands (Finnveden and Schneider, 2023; Michel et al., 2024). In response, critical competencies such as systems thinking and anticipatory thinking are increasingly being integrated into the curriculum via whole-institution approaches and novel pedagogies (Duarte et al., 2023; Hyytinen et al., 2023; Lozano et al., 2022; Rieckmann, 2018; Talley and Hull, 2023).

This study is grounded in the growing imperative to integrate key sustainability competences into higher education. Although several frameworks have defined these competences (e.g. Bianchi et al., 2022; Wiek et al., 2011, 2016), validated assessment tools remain limited, particularly those applicable across diverse educational contexts (Michel et al., 2024). Developing such tools is crucial for translating theoretical conceptions into practice, supporting empirical research and guiding curriculum design and evaluation. In response to ongoing calls from international organisations such as the European Commission (Bianchi, 2020; Bianchi et al., 2022), UN (2023), UNESCO (2017, 2023) and researchers (e.g. Annelin and Boström, 2023; Brundiers et al., 2021; Filho et al., 2020; Michel et al., 2024), this study addresses three key aims. First, it develops the “Key Competences in Sustainability” Indicator Tool (hereinafter KCiS-IT), which draws on the concept of Integrated Problem-Solving Competence (hereinafter IPSC) (Wiek et al., 2011, 2016). Second, it evaluates the construct validity of the tool’s competence measures and its meta-competence, IPSC. Third, it assesses concurrent validity by examining how IPSC relates to students’ pro-environmental norms and behaviour using data from business and economics students across seven Higher Education Institutions (HEIs) in five European countries. As a result, this work contributes to a validated, evidence-based assessment tool that may help educators and policymakers track and promote the development of key sustainability competences in higher education.

This literature review outlines the working definitions, existing frameworks and core sustainability competences that underpin this research. It also highlights the research questions and hypotheses examined in this study.

Sustainability is broadly defined as “prioritising the needs of all life forms and of the planet by ensuring that human activity does not exceed planetary boundaries” (Bianchi et al., 2022, p. 12). Sustainability competences, in turn, refer to “complexes of knowledge, skills and attitudes that enable successful task performance and problem-solving related to real-world sustainability, challenges and opportunities” (Wiek et al., 2011, p. 204). The KCiS are grounded in a convergence of educational, psychological and systems theories that emphasise the capacity of individuals to navigate complexity, uncertainty and socio-ecological challenges. Rooted in competence theory (Mulder, 2014), these frameworks define competences not merely as knowledge or skills, but as integrated capabilities that enable responsible and effective action. These competences are inherently interdisciplinary and action-oriented, aligning with constructivist learning theories and the goals of Education for Sustainable Development (ESD), which promote critical reflection, learner agency and transformative change (Rieckmann, 2018; UNESCO, 2017). Collectively, these theoretical underpinnings underscore the importance of equipping learners not just with knowledge but with the holistic capacities needed to co-create sustainable futures.

Numerous sustainability competence frameworks exist, albeit differing in their theoretical foundations and methods of development. Normative models such as UNESCO’s (2017, 2023) ESD framework and the European Commission’s GreenComp (Bianchi et al., 2022) were developed through literature reviews and stakeholder consultations. Other frameworks, like Rieckmann’s (2012), draw on empirical methods – specifically, a Delphi study involving educators from 15 countries identified six core competences that conceptually overlap with those of Wiek et al. (2016). Similarly, Lozano et al. (2017) used a combination of literature synthesis and expert input to propose 12 competences relevant to higher education. Among these, the most cited is the KCiS conceptual framework by Wiek et al. (2011, 2016). This framework draws on systems thinking, futures literacy and transformative learning theory to propose five interrelated core competences:

1. systems thinking competence;

2. anticipatory or futures thinking competence;

3. values thinking or normative competence;

4. strategic thinking or action-oriented competence; and

5. collaboration or interpersonal competence – integrated under a meta-competence of integrated problem solving.

Following a Delphi study with 14 international experts (Brundiers et al., 2021), the five original competences in the KCiS framework were reaffirmed, but two additional competences were proposed – intrapersonal and implementation. Subsequently, Redman and Wiek (2021) updated the KCiS framework to incorporate seven competences and the mega-competence of integrated problem solving. However, the inclusion of these two competences is not without its fair share of criticism (Bianchi, 2020). The same experts in the Brundiers et al. (2021) study could not reach consensus on their categorisation. Furthermore, some of them had serious reservations on both of them. They argued that implementation competence does not fit the cognitive orientation framework of KCiS but would require a hands-on approach that goes beyond the mission of HEIs. Additionally, intrapersonal competence is more of a mindset than a competence. We share the same sentiment that intrapersonal competence is a personal capacity of being, while implementation competence requires a context that can never be manifested in a university setting.

While most of these aforementioned frameworks are conceptual or exploratory, those developed through empirical methods often rely on qualitative approaches, which limits the ability to quantitatively assess their psychometric properties. Additionally, frameworks tailored to specific groups, such as entrepreneurs (Lans et al., 2014), would-be entrepreneurs (Ploum et al., 2018) or corporate social responsibility managers (Wesselink et al., 2017), may lack broader applicability to other disciplines (Annelin and Boström, 2023).

In light of these critiques, this study adopts the original KCiS framework (Wiek et al., 2011, 2016). To this end, this study will specifically evaluate the psychometric properties of the measures pertaining to those five core competences that are observable and applicable within traditional higher education settings.

Many researchers across disciplines often fail to report the reliability and validity of their scales (Lakshmi and Mohideen, 2013) or use single-item measures (Babin and Svensson, 2012), reducing assessment quality and study accuracy (Tavakol and Dennick, 2011). Reliability refers to the repeatability and consistency of a measure, while validity is the extent to which a measure accurately represents what it is intended to conceptually measure (Bezzina and Saunders, 2014). When examining the psychometric properties of measurement scales based on structural equation modelling (SEM), it is generally recommended that researchers establish composite reliability (a measure of the internal consistency of the measures), the factor structure (to test whether the data fit the hypothesised measurement model) via confirmatory factor analysis (CFA) and convergent/discriminant validity (Cheung et al., 2024). Convergent validity is the extent to which measures of the same construct share a high proportion of variance, while discriminant validity is the extent to which a construct is “unique and captures some phenomenon other measures do not” (Hair et al., 2010, p. 710).

This study aims to investigate the construct validity of the KCiS-IT measures using the IPSC concept (Wiek et al., 2011, 2016). This reflects an important gap in the literature, since a validated instrument measuring key competences using the IPSC framework is lacking. Hence, the first objective of this study is to empirically evaluate and seek support for its five-factor model, as well as to analyse its convergent and discriminant validity. Therefore, our first research question is:

Furthermore, theoretical frameworks in sustainability education emphasise that addressing complex sustainability challenges requires the combined application of multiple sustainability competences. As such, IPSC can be conceptualised as a single broader dimension that accounts for the correlations among a set of first-order factors, better known as a second-order factor (Brown, 2015). It reflects the interconnection and co-activation of key competences (Redman and Wiek, 2021; Wiek et al., 2011). This approach aligns with the view of IPSC as a meta-competence that emerges from effective use of more specific first-order competences in practice (Brundiers et al., 2021). Having a single broader dimension has many statistical advantages. Apart from the fact that highly correlated factors might fail to produce a “clean” first-order factor structure, a second-order model reduces complexity, is more parsimonious (consumes less degrees of freedom) resulting in better model fit, can avert confounding effects like multicollinearity in multidimensional model structures and increases the bandwidth of content covered by the construct (Hair et al., 2019; Sarstedt et al., 2022). Hence, the second objective of this study is to empirically evaluate and seek support for the second-order factor of the KCiS-IT, as well as to analyse its convergent and discriminant validity. Therefore, our second research question is the following:

In the presence of empirical support for the IPSC, the third objective of the study is to examine how the IPSC behaves in relationships with theoretically related constructs. This would provide robust evidence of concurrent validity, another important psychometric property when validating a new scale (Clark and Watson, 2019). Concurrent validity involves testing whether a new scale correlates in expected ways with conceptually related constructs measured concurrently, without implying conceptual overlap (Cohen et al., 2011). There is both theoretical and empirical support for a relationship between pro-environmental orientations and behaviours. Competences such as systems thinking and action-oriented competence have been linked to the development of normative orientations and sustainability-related behaviours (Redman and Wiek, 2021; Wiek et al., 2011). It also reflects the broader sustainability education literature, which sees competences as enabling informed action and value-driven behaviour (UNESCO, 2017). Building on this rationale, we used an adaptation of the Norm-Activation Model (Schwartz, 1977; Schwartz and Howard, 1981), which posits that prosocial behaviour is expected to follow from personal norms that reflect a moral obligation to perform or refrain from specific actions. In our conceptual model (see Figure 1), the single broader dimension of the sustainable competences (IPSC) will replace awareness of consequences and ascription of responsibility (see Figure 1).

Figure 1 exhibits the hypothesised mediational pathway.

In view of the above, our third research question is:

Studies have revealed that as individuals develop sustainable competences, their environmental norms related to sustainability become stronger (Nguyen et al., 2021; Steg et al., 2005). Additionally, interventions through environmental education programmes have contributed to the activation of personal norms (Hungerford and Volk, 1990; UNESCO, 2017). Therefore, we postulate that:

There is strong empirical evidence that personal environmental norms directly influence conservation behaviour (Kaiser et al., 2005). They have also been linked to specific pro-environmental actions such as recycling, reducing waste or energy conservation (Harland et al., 1999; Onwezen et al., 2013). Meta-analyses further suggest that personal norms are the key psychological determinants of pro-environmental behaviour (e.g. Bamberg and Möser, 2007). Therefore, we postulate that:

Various studies have also reported how sustainable competences, once developed, improve environmental behaviours such as sustainable consumption, conservation practices and/or waste management (e.g. Barth et al., 2007; Mochizuki and Fadeeva, 2010; Rieckmann, 2012). Other studies extended this direct effect from educational settings to professional and personal settings (Evans, 2019). Therefore, we postulate that:

Finally, studies have provided empirical evidence that personal norms play a crucial role in translating sustainability competences into actions such as energy conservation/reduction and recycling (e.g. Bamberg et al., 2007; Helferich et al., 2023; Schumacher and Reuter, 2021). Therefore, personal norms bridge the gap (Steg and Vlek, 2009) and strengthen the relationship between sustainable competences and pro-environmental behaviour. Therefore, we postulate that:

The following section outlines the methodological approach used to develop and validate the KCiS-IT. It details the sampling procedures, measurement tools and data analysis techniques employed in the study.

The study targeted business and economics students hailing from seven universities in Poland, Malta, Czech Republic, Kosovo and Spain that were participating in a bid for a student internationalisation project funded by the Polish National Agency for Academic Exchange (NAWA). The selection of universities across different countries was intended to capture institutional-level insights within a multi-country context, not as representative of all students in each respective HEI or country.

Ethical clearance was sought from the University of Malta’s Faculty of Economics, Management and Accountancy Research Ethics Committee (Application ID: FEMA/2023/00850), and consent was obtained from the respective deans. Following that, the faculty managers issued an email with a weblink to an online questionnaire, requesting their undergraduate and postgraduate students to participate in our study.

Hence, the sampling method utilised in this study was the census-based voluntary sampling approach. According to Bryman (2016), this approach maximises inclusive reach, upholds ethical standards by respecting autonomy and informed consent and is more suited for small, accessible populations. However, voluntary response may have introduced bias by attracting individuals with strong views or distinct characteristics, while non-response could limit statistical power and generalisability. Confidentiality and anonymity were guaranteed.

Before data collection, we calculated the minimum sample size required for this study for CFA and SEM modelling. For CFA, using the widely accepted 10:1 ratio of “participants to measured variables” (Hair et al., 2010) would require 250 participants. For SEM, we used the inverse square root method. Assuming a significance level of 5%, a power level of 80% and a minimum path coefficient of 0.2 to be significant (Hair et al., 2022), a minimum sample of 154 observations is necessary. Hence, we established a minimum sample size of 250 for our study.

We collected 713 complete responses between 7 and 21 December 2023. This exceeded the a priori minimum sample size of 250. The participants hailed from the Wroclaw University of Economics and Business, Poland (25.5%), University of Opole, Poland (17.3%), University of “Haxhi Zeka,” Kosovo (14.9%), University of Malta, Malta (14.6%), Prague University of Economics and Business, Czech Republic (9.8%), Mendel University in Brno, Czech Republic (9.3%), and University of Valencia, Spain (8.6%). The respondents were mostly female (52.5%), followed an undergraduate (76.3%) programme (23.7%), had experienced work opportunities (55.3%) but no internship opportunities (57.6%). Their mean age was 23.0 years (SD = 6.58), with ages ranging from 18 to 58. We were unable to compare sample and population distributions because the participating HEIs did not offer demographic data. It is also worth mentioning that none of the participants took part in programmes that were explicitly or systematically designed to develop core competences in sustainability. As a result, their responses should reflect their prior awareness or informal exposure to sustainability, rather than the outcome of planned instructional intervention initiatives.

The questionnaire elicited demographic information about the respondents’ gender, age, university, level of study and whether they had work or internship opportunities. The respondents were also requested to respond to the 24 Likert-type statements, which are exhibited in the Supplementary File (Appendix 1).

The KCiS scale consists of 15 items, with three items per construct: anticipatory (A1–A3), strategic thinking (S1–S3), values thinking (V1–V3), systems thinking (ST1–ST3) and collaboration (C1–C3). Five items for values thinking and collaboration were taken from Ploum et al. (2018), three strategic thinking items were adapted from Torres and Augusto (2017) with a sustainability focus, and the remaining seven were developed for this study based on the working definitions of Wiek et al. (2011, 2016) and guidance from Annelin and Boström (2023) and Bianchi (2020). Cronbach’s alpha coefficients were 0.81 (anticipatory), 0.80 (values thinking), 0.85 (strategic thinking), 0.78 (collaboration), 0.83 (systems thinking) and 0.92 for the overall scale.

The Pro-environmental Behaviour (PEB) scale includes six items adapted from the “Employees” Environmental Behaviour’ scale (Chou, 2014). Items were revised for contextual relevance and to avoid assumptions about employment, car ownership or chopstick use. This scale has two facets: Saving Energy (SE1–SE3) and Reducing Waste (RW1–RW3). Cronbach’s alpha was 0.81 for Saving Energy, 0.78 for Reducing Waste and 0.83 for the overall scale.

Finally, the Personal Environmental Norms (PEN) scale (N1–N3) includes three items adapted from Chou (2014), avoiding direct reference to energy saving or waste reduction to prevent overlap with PEB items. Cronbach’s alpha was 0.84.

Sustainability competence items were rated on a 5-point agreement scale (1 = strongly disagree to 5 = strongly agree), while PEN and PEB items used a 5-point frequency scale (1 = never to 5 = most of the time).

Before the main study, a pilot study with 272 HEI students was conducted and analysed using exploratory factor analysis (EFA) with maximum likelihood estimation, Promax rotation (an oblique rotation method that allows factors to be correlated rather than independent) and the latent root criterion (where only factors with eigenvalues greater than 1 are retained). EFA identified five factors (eigenvalues > 1), explaining 66.10% of the variance. Sampling adequacy was high (Kaiser–Meyer–Olkin = 0.92), and Bartlett’s test confirmed factorability (⁠$χ2$= 3773.00, df = 276, p < 0.001). However, the initial pattern matrix revealed overlap: values, collaboration and systems thinking items loaded on one factor, with low loadings for V1–V3 and high cross-loadings for V1 and V2. Pro-environmental behaviour items also loaded weakly. This suggested an eight-factor model (five for the competences, one for personal environmental norms and two for pro-environmental behaviour), which is justified when testing established theory (Hair et al., 2010). A subsequent EFA with eight a priori factors revealed a clean factor structure with strong loadings (>0.60) and no significant cross-loadings (>0.30). These factors explained 76.88% of the variance (see Supplementary File – Appendix 2).

For RQ1, we tested a five-factor CFA model with five factors (Model 1a) and a competing three-factor model (Model 1b), as EFA had grouped values thinking, collaboration and systems into one competence. Model fit was assessed using the chi-square test (noting its sensitivity to large samples, Hair et al., 2010), the Normed chi-square (⁠$χ2$/df : ≤ 3 good, ≤ 5 acceptable), the confirmatory fit index (CFI: ≥ 0.95 good, ≥ 0.90 acceptable) and the root mean square error of approximation (RMSEA: ≤ 0.05 good, ≤ 0.08 acceptable) as recommended by Byrne (2016). Error terms with modification indices greater than 20 were covaried to improve model fit. We compared models using the chi-square difference test. Convergent validity requires composite reliability (CR) and average variance explained (AVE) to exceed 0.70 and 0.50, respectively. Discriminant validity requires the maximum shared variance (MSV) to be smaller than the AVE and the square root of the AVE to exceed inter-construct correlations.

For RQ2, we generated a second-order CFA (Model 2) with two second-order factors (IPSC and pro-environmental behaviour) and one first-order factor (personal environmental norms). We followed the same goodness-of-fit procedures outlined for RQ1.

Before addressing RQ3, we tested common method bias (CMB) using the common latent factor (CLF) method (Kock et al., 2021). In the presence of CMB, we used CBM-adjusted composite variables when conducting mediation analysis (SEM) in IBM SPSS AMOS (v29). Bias-corrected factor scores adjust for measurement error, offering more reliable estimates than raw scores or unadjusted estimates; this is crucial for mediation, where bias can distort direct and indirect effects (Hair et al., 2010). In Model 3, IPSC (X) predicted pro-environmental behaviour (Y) via personal environmental norms (M), whilst controlling for demographics. A significant direct effect (p ≤ 0.05) supported the hypothesis. Indirect effects were tested using the bias-corrected percentile method with 5,000 bias-corrected bootstraps and 95% confidence intervals (CIs); significance was inferred when the CI excluded zero. Mediation is partial if both direct and indirect effects are significant, and full if only the indirect path is significant (Byrne, 2016).

This section presents the findings addressing the research questions, detailing the analysis and key outcomes of the study.

The five-factor CFA (Model 1a) produced a good fit (⁠$χ2$ = 233.73, df = 80, p < 0.01; $χ2$/df = 2.93, CFI = 0.98, RMSEA = 0.05), while the competing three-factor CFA (Model 1b) produced a poor fit to the data (⁠$χ2$ = 819.67, df = 87, p < 0.01; $χ2$/df = 9.42, CFI = 0.88, RMSEA = 0.11). The chi-square difference test confirmed that Model 1a had a significantly better fit (Δ$χ2$ = 0.585.94, Δdf = 1, p < 0.01). Since Model 1a did not produce any modification indices greater than 20, we proceeded to obtain descriptive statistics and convergent/discriminant validity measures of its scales (see Table 1). There were no convergent and discriminant validity concerns (Cheung et al., 2024). However, it is worth noting that anticipatory competence barely met the criteria for discriminant validity.

Therefore, we concluded that five measures of the KCiS-IT demonstrated adequate construct validity.

The CFA model comprising two distinct second-order factors for IPSC and pro-environmental behaviour and a first-order factor for personal environmental norms (Model 2) produced an adequate fit to the data (⁠$χ2$ = 775.67, df = 244, p < 0.01; $χ2$/df = 3.17, CFI = 0.94, RMSEA = 0.06). An improvement in model fit was obtained when the strategic thinking and collaboration error terms were covaried (Δ$χ2$ = 41.29, Δdf = 1, p < 0.01). The resulting Model 2b produced a relatively good fit (⁠$χ2$ = 734.38, df = 243, p < 0.01; $χ2$/df = 3.02, CFI = 0.95, RMSEA = 0.05). Table 2 provides descriptive statistics and convergent/discriminant validity measures for the variables. All correlations were in the correct theoretical direction, and no concerns regarding convergent and discriminant validity emerged (Cheung et al., 2024). Therefore, we concluded that IPSC was producing the associations among the first-order factors, providing evidence of strong construct validity.

Figure 2 exhibits the CFA output with standardised loadings. The second-order factor loadings were strong, ranging from 0.81 to 0.91.

The CFA model with the CLF (Model 2c) produced a good fit to the data (⁠$χ2$ = 716.62, df = 242, p < 0.01; $χ2$/df = 2.96, CFI = 0.95, RMSEA = 0.05). The chi-square significance test (Δ$χ2$ = 17.76, Δdf = 1, p < 0.01) confirmed an improved fit with the CLF, thereby confirming the presence of CMB. Therefore, we generated CMB-adjusted composite variables.

The mediation model (Model 3) was just-identified (df = 0), meaning it perfectly reproduced the observed covariances, rendering fit indices irrelevant (Byrne, 2016). Table 3 exhibits the regression weights for the structural paths (Model 3).

Table 3 reveals that after controlling for the demographic variables, IPSC produced a positive and significant direct effect on PEN, while PEN produced a positive and significant direct effect on PEB. This supported H1 and H2. Furthermore, IPSC produced a positive and significant direct effect on PEB, thereby supporting H3.

To release some degrees of freedom, four insignificant paths (i.e. p > 0.05) were eliminated. The parsimonious model (Model 3b) produced an excellent fit to the data (⁠$χ2$ = 2.81, df = 4, p = 0.59; $χ2$/df = 0.70, CFI = 1.00, RMSEA = 0.00) (see Figure 3).

Table 4 shows that the indirect effect was statistically significant. Therefore, PEN partially mediated (strengthened) the relationship between IPSC and PEB. This supported H4.

This section presents the study’s key findings in the context of existing research, highlighting their theoretical and practical implications.

The study provided empirical support for the construct validity of the KCiS measures, both at the individual level of competences and at the aggregated level via the over-arching mega competence, IPSC. Furthermore, the IPSC provided theoretically meaningful relationships in a hypothesised model that explored pro-environmental norms and behaviour. More specifically, IPSC produced a direct effect on personal environmental norms, while personal environmental norms produced a direct effect on pro-environmental behaviour. Additionally, IPSC produced both direct and indirect effects via personal environmental norms on pro-environmental behaviour. This provided empirical support for the concurrent validity of the IPSC.

This study sought to address a gap in the literature by proposing and validating a KCiS-IT based on the sustainability research and problem-solving competence framework (Wiek et al., 2011, 2016). After identifying three indicators for each of the five competences in this framework, this study provides empirical evidence of construct validity and concurrent validity of its measures, both as first-order factors and as a second-order factor through IPSC. This is in line with calls highlighting the need to create valid tools to assess competences in sustainability across educational levels and sectors (Annelin and Boström, 2023; Bianchi et al., 2022; Filho et al., 2020; UNESCO, 2017, 2023).

The analysis of individual competences (RQ1) revealed two key points for discussion. First, while most studies highlight values as the driver of other sustainability competences (Annelin and Boström, 2023), Molderez and Fonseca (2018) found collaboration as the most influential, suggesting that students must first learn to collaborate before developing other competences. In our study, strategic thinking emerged as the leading competence – it had the highest mean score and strongest standardised loading with IPSC in the second-order CFA. Although the KCiS-IT targets students in HEIs, this may reflect our sample of business and economics students, for whom strategic thinking is naturally central (Boyles, 2022). Second, while correlations among competences were relatively high, none exceeded thresholds that would suggest a lack of discriminant validity (Cheung et al., 2024). However, anticipatory competence was borderline. Similarly, Ploum et al. (2018) found that collaboration competence barely met discriminant validity among would-be entrepreneurs and argued that this is probably due to its broad applicability. High correlations can reduce the unique variance captured by each factor, decreasing CFA’s ability to differentiate between constructs (Kyriazos and Poga, 2023). In such cases, rather than clustering competences as suggested by Lans et al. (2014), we recommend using a second-order factor to account for shared variance and avoid redundancy.

A second important finding of this study is that the five KCiS measures can be represented by a second-order factor (IPSC), thereby improving potential usability (Brown, 2015). The use of higher-order constructs has grown significantly in recent years, particularly with the rising complexity of theories and cause–effect models in the social sciences (Sarstedt et al., 2022). A second-order construct is more parsimonious, overcomes the issue of multicollinearity (which is known to reduce discriminant validity) among first-order factors and captures the phenomenon in a broader sense (Hair et al., 2019).

The third important finding of this study is related to empirical evidence of concurrent validity. The results of this study are encouraging since they provide evidence that the KCiS-IT has theoretical meaningfulness as a general measure of sustainability competences for HEI students. The internal mechanics of this indicator tool follow relatively well hypothetical pathways and lends weight to previous claims that:

• sustainable competences produce direct effect on personal norms (Hungerford and Volk, 1990; Nguyen et al., 2021; Steg et al., 2005) and environmental behaviour (Barth et al., 2007; Mochizuki and Fadeeva, 2010);

• personal norms have a direct effect on pro-environmental behaviour (Bamberg and Möser, 2007; Harland et al., 1999; Kaiser et al., 2005; Onwezen et al., 2013); and

• personal norms mediate (strengthen) the relationship between sustainable competences (IPSC) and pro-environmental behaviours (Bamberg et al., 2007; Helferich et al., 2023; Schumacher and Reuter, 2021).

Overall, our findings suggest that IPSC is a composition of sustainability competences that can help shape perceived sustainable behaviours by influencing HEI students’ belief system. Since such competences are amenable to change, they have practical relevance for HEI settings.

This study offers both theoretical and practical contributions for researchers and educators interested in assessing sustainability competences.

From a theoretical perspective, this validated scale addresses a critical gap in the sustainability education literature, which has lacked a standardised instrument for measuring key sustainability competences among students. By capturing both individual competences and an overarching second-order construct – representing integrated sustainability competence – this tool enables more nuanced analysis. Modelling the construct as a higher-order factor aligns with established psychometric practice and is supported by empirical literature for its theoretical and statistical robustness (e.g. Sarstedt et al., 2022). This allows researchers and educators to assess both students’ general sustainability competence and the relative strength of its component dimensions. In addition, theory building necessitates robust measures, which accurately assess the constructs underlying the phenomenon. Therefore, this study makes a theoretical contribution by providing a valid and reliable measurement tool.

From a practical perspective, a validated tool that assesses key sustainable competences has multiple applications in higher education, policy and management. The KCiS-IT enables university tutors to identify what students know and use this evidence to adapt and develop curricula accordingly. This would not only help students gain essential sustainability competences but would also lead to more efficient teaching methods, targeted learning and meaningful feedback (Savage et al., 2015), resulting in improved educational outcomes (Lozano et al., 2017, 2022; Wiek et al., 2011). Sustainability competences are pivotal for preparing students (tomorrow’s leaders) with the mindset needed to address global challenges. By fostering an IPSC, sustainable interventions can empower students as agents of change. Alongside strategic thinking, anticipatory, systems thinking and collaborative skills for sustainability, learners also nurture values for responsible action in local and global contexts. From a management perspective, this validated tool can help organisations diagnose workforce capabilities in sustainability. Identifying employee competences allows for targeted training and professional development. Aligning sustainability with business goals can also help mitigate risk, enhance reputation and improve effectiveness (Schulte and Knuts, 2022). For policymakers, this tool offers measurable data to inform policies on education, training and sustainable strategies. It also enables comparisons across sectors or institutions and supports the creation of competence benchmarks aligned with sustainability goals, contributing to evidence-based policymaking (Bianchi, 2020).

This study has several limitations. First, its cross-sectional design limits causal interpretation, unlike longitudinal or experimental designs. Second, it relied on self-reported behaviour, which may differ to some extent from actual behaviour. Third, differences across universities were not examined as this would warrant a separate study with a much larger sample. Fourth, the study utilised students from one academic discipline across seven HEIs in five countries. As such, we cannot confirm, nor rule out, the generalisability of the findings to other HEIs or academic disciplines. Finally, the competence set utilised may not be exhaustive.

Future studies could replicate this study across various other disciplines and regions, adopt longitudinal designs and explore additional competences to build a more comprehensive model. Further investigation is also needed on the impact of embedding sustainability into higher education and on identifying and sharing effective practices to enhance their social and educational value.

It is hoped that as more educational institutions embed sustainability across curricula and adopt evidence-based tools like the KCiS-IT, they will help cultivate a generation equipped to drive societal transformation towards a more equitable, resilient and sustainable future, positioning education as a catalyst for sustainable development.

The authors thanks the deans and faculty managers of the participating HEIs for their support during data collection, and we are grateful to all participants who completed the online questionnaire.

The supplementary material for this article can be found online.