Developing and Validating an Intercultural Student Experience Scale Using Structural Equation Modeling

Abstract

This study proposes and validates a culturally responsive instrument for assessing Student Experience (SX) in Higher Education Institutions (HEIs). Guided by Customer Experience (CX) theory and Hofstede’s cultural framework, we drafted a thirty-item scale: nine educational, seven social, three personal, and twelve cultural items spanning Indulgence–Restraint, Individualism–Collectivism, Masculinity–Femininity, Uncertainty Avoidance, Long- versus Short-Term Orientation, and Power Distance. Undergraduate respondents from universities with contrasting cultural profiles completed the survey. Confirmatory factor analyses affirmed a three-dimensional SX structure (Educational, Social, Personal) and six first-order cultural dimensions. A hierarchical second-order Structural Equation Model (SEM) linked the higher-order construct Cultural Aspects (CA) to the higher-order construct SX. The path from CA to SX emerged positive and statistically relevant, indicating that national-culture orientations systematically color students’ cognitive, affective, and behavioral evaluations of institutional touchpoints. The scale enables researchers and academic managers to pinpoint SX gaps, benchmark performance internationally, and design culturally congruent and sustainability-aligned interventions. The article deepens theoretical understanding of how culture shapes service perceptions in global HEIs by explicitly integrating cultural theory and cultural studies into SX evaluation.

1. Introduction

Aligning students’ initial expectations with their subsequent perceptions of the services, products, and systems delivered by Higher Education Institutions (HEIs) is widely recognized as a pivotal gauge of perceived quality [1]. Because accrediting bodies explicitly evaluate this perceived quality when auditing universities, initiatives that enhance the Student eXperience (SX) yield benefits for both learners and institutions. By embedding sustainability principles into SX-focused strategies, institutions strengthen their social responsibility agendas and amplify the long-term impact of improvements. SX’s rigorous diagnostics help administrators identify gaps, strategically allocate resources, and increase student satisfaction, improving an institution’s competitive position. Although several established instruments are already used to appraise service quality in higher education [2], they tend to be generic, confined to specific interaction domains, and require substantial adaptation to meet the diverse needs of different and diverse student populations.

HEIs increasingly compete in a borderless marketplace where academic quality is judged not only by learning outcomes but also by the totality of impressions students accumulate before, during, and after enrolment. The SX concept frames those impressions as a field-specific instantiation of Customer Experience (CX). Students act as consumers who evaluate a bundle of institutional “touchpoints” that range from admission portals and virtual learning environments to classroom interactions, extra-curricular services, and social interactions [3]. Within CX theory, experience arises from the continuous interplay of cognitive assessments, affective reactions, and behavioral intentions. Translating this logic to the university context clarifies why measuring SX is strategically critical. Comprehensive SX diagnostics empower administrators to identify “pain points”, allocate resources where they matter most, and ultimately enhance perceived educational quality, an attribute weighted heavily by accreditation agencies, ranking bodies, and prospective students [4]. Moreover, by illuminating how institutional encounters are interpreted across cultures, SX evaluation facilitates international cohorts’ social and academic integration, a priority for HEIs committed to diversity and global engagement.

Despite its managerial importance, SX research remains fragmented. Many instruments inherit item pools from generic service-quality scales, focus on isolated interactions such as learning-management systems, or disregard the cultural heterogeneity that shapes expectations and satisfaction [5]. A recent systematic review converges on three broad SX dimensions—Educational, Social, and Personal—yet empirical studies seldom test how these spheres interact with students’ cultural value orientations.

To address this gap, the present study develops and validates a culturally responsive SX scale grounded simultaneously in SX/CX theory and Hofstede’s six-dimensional national-culture model [6]. Beginning with a concept-mapping exercise, we generated thirty-one candidate items: nine Educational (ED), seven Social (SO), two Personal (PE), and twelve cultural statements. One item, an open-ended prompt, is retained for qualitative richness but excluded from the quantitative analyses reported here.

A total of 302 valid responses were collected from 321 undergraduates spanning seventeen nationalities across Latin-American and Spanish universities, providing ample heterogeneity for cross-cultural testing. Given the ordinal data, we used a diagonally weighted least-squares estimator within an SEM framework [7,8]. Confirmatory factor analysis validated the three-factor SX model and six cultural factors, and a hierarchical SEM showed that the second-order Cultural Aspects construct positively predicts overall SX.

The validated scale exhibits a strong factorial validity and composite reliability, offering researchers a concise yet comprehensive diagnostic capable of internationally benchmarking SX. For practitioners, the instrument functions as an actionable dashboard: item-level scores reveal granular experience gaps; dimension-level scores inform strategic initiatives; and culture-linked diagnostics guide the design of equity-oriented interventions that resonate with domestic and international students. By embedding cultural nuance at the conceptual core rather than treating it as an ex-post segmentation variable, this study advances SX scholarship and provides HEIs with a rigorous tool for enhancing perceived quality, overall satisfaction, and the successful integration of a diversifying student body.

In response to the fragmented nature of existing instruments, this study also introduces the concept of SX diagnostics as a strategic tool to identify experience-related gaps across educational, social, and personal dimensions. While the term is used conceptually to frame the practical utility of the scale, the empirical validation presented here offers a foundational step toward operationalizing these diagnostics in institutional contexts. The structural model results highlight the predictive power of cultural orientations over student experience, further supporting the use of the scale as a research instrument and a diagnostic resource for policy and service design in intercultural settings.

2. Theoretical Background

2.1. Service Science

Service Science has emerged as an inherently interdisciplinary domain that synthesizes insights from organizational theory, human-centered design, business strategy, and information technology to illuminate the structure and behavior of contemporary service systems. Maglio and Spohrer conceptualize these systems as “dynamic value co-creation configurations of resources/people, technology, organizations, and shared information” [9], underscoring their socio-technical character and capacity to adapt over time through continuous feedback loops among stakeholders. By viewing value not as a static output but as a co-generated outcome of interactions, Service Science offers a holistic analytical lens particularly salient for HEIs, which increasingly operate as complex service ecosystems in which students, faculty, administrators, and digital infrastructures jointly craft educational value. Subsequent theoretical refinements extend this perspective by situating service systems within broader networks of interdependent actors and emphasizing information flows’ pivotal role in coordinating resource integration and fostering innovation [10]. CX is a cornerstone construct within this framework that bridges micro-level perceptions with macro-level system performance. Schmitt’s experiential marketing paradigm, which advocates replacing product-centered logics with orchestrated experiential encounters capable of “awakening the senses and reaching the heart of the consumer” [11], operationalizes CX through five strategic experiential modules: sense, feel, think, act, and relate. These modules collectively offer a solid theoretical scaffold for designing measurement instruments by mapping how educational service systems evoke sensory, emotional, intellectual, behavioral, and social responses in students and how these responses, in turn, inform value co-creation processes within culturally diverse academic environments.

Although our scale was not explicitly structured on Schmitt’s experiential marketing paradigm, clear conceptual parallels can be observed. Service science frames higher education as a value co-creation process, and Schmitt’s Strategic Experiential Modules illuminate how different aspects of experience manifest in this context. For example, elements of THINK/ACT resonate with the ED items that capture cognitive challenge and application of learning; RELATE aligns with the SO items reflecting collaboration and faculty–student interaction; and FEEL/SENSE correspond to the PE items linked to well-being and motivation. Thus, while the instrument was developed independently, its three dimensions are consistent with service-science perspectives.

2.2. Customer Experience (CX)

Laming and Mason conceptualize CX as the totality of physical stimuli and affective states that arise during a consumer’s interactions with a brand’s product or service, beginning with the first conscious encounter and extending seamlessly through the entire consumption journey into the post-purchase phase [12]. Building on this processual view, Joshi characterizes CX as the cumulative set of episodes a customer accrues throughout the duration of the service relationship, thereby framing experience as an evolving ledger of brand-related engagements rather than a single event [13]. LaSalle and Britton add further precision by describing CX as the ensemble of customer–brand interactions—whether with a specific offering, the organization at large, or any of its constituent units—that elicit discernible cognitive, emotional, or behavioral reactions [14]. The mechanism through which such experiences are assembled is commonly articulated via the notion of “touchpoints,” that is, the myriad occasions on which consumers “touch” or utilize a product, service, or system across multiple channels and temporal junctures [15,16,17]. Stein and Ramaseshan provide an empirically grounded taxonomy of seven elemental categories manifest at touchpoints (atmospheric, technological, communicative, process, employee–customer interaction, customer–customer interaction, and product interaction), thereby underscoring the heterogeneity of stimuli that conjointly shape perceived value [18]. Synthesising extant scholarship, Vanharanta et al. distill three overarching features of CX: its intrinsically subjective nature, dual incorporation of rational cognition and affect, and fundamentally holistic orientation that resists reduction to isolated service attributes [19]. Gentile et al. further elucidate CX’s multidimensional structure by delineating six interrelated experiential domains: sensorial, emotional, cognitive, pragmatic, lifestyle, and relational. Each domain supplies a distinct, yet interwoven, pathway through which meaning is ascribed to brand encounters [20]. Collectively, these perspectives position CX as a dynamic, multi-layered construct that integrates objective service design elements with subjective consumer interpretations, thereby offering a rigorous conceptual scaffold for the culturally responsive SX instrument advanced herein, where students are theorized as context-specific customers embedded within higher-education service ecosystems.

2.3. Student Experience (SX)

SX still lacks a universally endorsed definition, yet scholars consistently highlight its ubiquitous nature, noting that it frequently extends beyond strictly academic settings [21,22]. Conceptually, SX can be treated as a domain-specific manifestation of CX in which students act as customers engaging with the products and services supplied by HEIs, without privileging any single customer-management model. From this standpoint, SX may be defined as “all the physical and emotional perceptions and reactions that a student or prospective student experiences in response to interactions with the products, systems, or services offered by an HEI, as well as encounters with individuals connected to academic life, both within and beyond formal learning spaces” [3]. The construct, therefore, comprises three interrelated dimensions: (i) a Social dimension concerned with community relationships and institutional engagement, covering peer-to-peer interactions, staff–student exchanges, and the attendant sense of belonging; (ii) an Educational dimension that encompasses learning engagement, perceived teaching quality, learning resources and environments, and academic support services, all of which collectively shape perceptions of institutional quality; and (iii) a Personal dimension that captures student development, emotions, identity, and background factors such as cultural heritage, socioeconomic status, disability, aspirations, family context, and leisure pursuits, each of which can modulate how services and products are experienced. This structure balances parsimony with theoretical coverage, aligns with service-science views of higher education as a value co-creation context, and resonates with prior frameworks emphasizing academic/social integration and experiential facets of learning (e.g., Tinto; Kuh; Lizzio; Astin) [23,24,25,26]. The results show that these dimensions are empirically distinguishable and substantively non-redundant, supporting their treatment as three correlated first-order factors.

Investment in SX aligns closely with UNESCO’s Sustainable Development Goals (SDGs), particularly Quality Education (SDG 4), Gender Equality (SDG 5), and Reduced Inequalities (SDG 10), thereby underscoring its strategic importance for HEIs while reinforcing institutional sustainability commitments [27,28]. By explicitly integrating Hofstede’s six-dimension cultural framework into the SX model, the present work situates each cultural orientation vis-à-vis the social, educational, and personal dimensions, yielding a more inclusive, context-sensitive diagnostic instrument [5].

2.4. Culture

Culture constitutes a foundational construct through which behavioral, ethical, and axiological divergences among societies become discernible. A survey of extant SX literature reveals that cultural factors are often treated superficially, despite their pivotal influence on student experience: cultural capital directly affects learning outcomes, and cultural identity critically shapes the adjustment processes of international, refugee, and migrant students [29,30,31]. Hofstede (2010) characterizes culture as “the collective programming of the mind that distinguishes the members of one group or category of people from another” [32]. Within the SX domain, systematic examination of students’ cultural backgrounds enables the identification of satisfaction–dissatisfaction patterns that vary by nationality and intersect with the lived realities of mobile student populations.

Guided by this rationale, the present study operationalizes on Hofstede’s six national-culture dimensions (

Table 1. Hofstede’s Cultural Dimensions.

Abbreviation	Cultural Dimension	Description
PD	Power Distance	The extent to which unequal distributions of power are viewed as legitimate and expected. High PD contexts normalize hierarchical authority and deference; low PD contexts emphasize egalitarian relations and participative decision-making across society and institutions.
UA	Uncertainty Avoidance	The extent to which ambiguity and unpredictability are perceived as threatening. High UA cultures tend to favor rules, formalization, and planning to reduce uncertainty; low UA cultures tolerate risk, novelty, and flexible procedures in societal and organizational practices.
IDV	Individualism/Collectivism	The degree to which persons are integrated into groups. Individualistic settings privilege personal autonomy, self-reliance, and individual rights; collectivistic settings stress in-group loyalty, mutual obligations, and relational interdependence as primary guides to thought and action.
MAS	Motivation towards Achievement and Success (Masculinity/Femininity)	The relative salience of achievement, competitiveness, and performance (masculinity) versus care, cooperation, and quality of life (femininity) as culturally endorsed values. This dimension concerns value priorities and role expectations in social and institutional life, not biological sex per se.
LTO	Long-/Short-Term Orientation	The temporal horizon guiding action and reward. Long-term–oriented cultures emphasize perseverance, future returns, and pragmatic adaptation; short-term–oriented cultures prioritize respect for tradition, fulfilling immediate social obligations, and near-term outcomes.
IVR	Indulgence/Restraint	The extent to which gratification of needs related to enjoyment and leisure is permitted by social norms. Indulgent cultures allow freer expression of desires and recreation; restrained cultures regulate gratification through stricter norms and a stronger sense of duty.

): Power Distance (PD), Uncertainty Avoidance (UA), Individualism–Collectivism (IDV), Masculinity–Femininity (MAS), Long- versus Short-Term Orientation (LTO), and Indulgence–Restraint (IVR). These dimensions offer a coherent framework to interpret how cultural orientations shape key aspects of SX. Specifically, PD relates to students’ comfort with authority and classroom participation; UA influences responses to ambiguity in assessment and instruction; IDV informs preferences for collaboration or autonomy; MAS aligns with attitudes toward academic competition and recognition; LTO reflects students’ goal orientation and time horizon; and IVR intersects with well-being and the regulation of leisure. Their integration into our SX model reinforces its cultural responsiveness and analytical depth.

Hofstede’s taxonomy was selected owing to its robust empirical pedigree, widespread adoption across disciplines, and the availability of practical measurement instruments [6,32]. We likewise considered the model’s widespread adoption in educational applications, which confers additional robustness to our proposal. Moreover, subsequent cultural frameworks often incorporate constructs conceptually consonant with Hofstede’s factors, underscoring the model’s continuing heuristic value [33].

Recent scholarship has prompted refinements to Hofstede’s schema. Specifically, the MAS dimension has been re-labelled “Motivation towards Achievement and Success” in response to critiques that the original terminology conflated cultural values with socially constructed gender roles, reproducing gender stereotypes [34,35]. This terminological shift seeks to disentangle achievement-related cultural orientations from gendered assumptions, fostering a more inclusive analytical vocabulary. Such adaptations exemplify the model’s capacity for iterative revision in light of evolving theoretical and social considerations, sustaining its relevance for contemporary intercultural research agendas.

2.5. Evaluating Student Experience

Given the progressively globalized character of higher education, where students originating from markedly different cultural milieux interact with academic systems that are often foreign to their own, we deemed it essential to embed an explicit cultural lens within the SX evaluation process, a requirement that prior systematic review has underscored [3]. Cultural orientations shape students’ cognitive frames, communication styles, and expectations of institutional support [36]; ignoring these orientations, therefore, risks misinterpreting satisfaction scores and obscuring equity gaps. By incorporating culture-sensitive metrics and adopting an intersectional outlook, evaluators can capture heterogeneity in learning preferences and communication norms, thus mitigating biases frequently arising when standardized instruments are applied indiscriminately. This culturally attuned approach not only lends greater validity to diagnostic results but also informs more just and context-relevant policy interventions.

Guided by this rationale, we developed a theoretical SX-Cultural model (Figure 1) that couples the canonical SX dimensions with Hofstede’s six national-culture dimensions, a framework long recognized for its explanatory power in cross-cultural research—including educational settings [5]. Mapping SX onto Hofstede’s typology permits a nuanced appraisal of how macro-cultural schemas modulate day-to-day academic experiences, thereby revealing patterns of satisfaction and friction that vary systematically across student groups.

The SX-Cultural model further enables the design of tailored solutions, e.g., heuristics explicitly targeted at international cohorts to navigate local academic conventions [37], or cross-cultural design guidelines that inform the development of learning technologies and service interfaces responsive to culturally differentiated needs [38].

Building upon the SX-Cultural model, we devised a holistic evaluation methodology expressly intended to serve as a strategic diagnostic instrument for HEIs [39]. At its core lies the multi-item evaluation scale introduced in this study, developed regarding a Spanish Latin American cultural case study [40] and informed by the domains assessed in the United Kingdom’s National Student Survey (NSS) [41]. We also proposed a checklist to evaluate SX [42], and we are currently developing a set of SX heuristics. Anchored directly in the validated SX-Cultural construct, the scale preserves conceptual coherence while streamlining data collection across heterogeneous institutional contexts. Moreover, the surrounding suite of methods and instruments is deliberately modular: universities may adapt, expand, or substitute individual components in line with their strategic priorities and resource constraints without jeopardizing the theoretical integrity of the evaluation framework. Incorporating cultural analytics into SX assessment offers HEIs a rigorously substantiated and operationally flexible pathway toward fostering more inclusive, equitable, and pedagogically effective learning environments.

2.6. Structural Equation Modeling (SEM)

SEM is a statistical method to test complex theoretical relationships among latent constructs that are not directly measurable, accounting explicitly for measurement error. SEM accomplishes this by partitioning the observed covariance/polychoric matrix into two interconnected components: first, a measurement model which defines the relationships between observed variables (such as survey items in our case) and their underlying latent constructs (SX and Cultural dimensions); and second, a structural model, which specifies directional causal paths or associations among these latent constructs [43,44]. To achieve this, SEM simultaneously estimates factor loadings, representing how strongly each observed indicator is associated with its latent variable, and structural paths, capturing relationships among latent factors.

In the present study, nine first-order factors—three related to SX: Educational (ED), Social (SO), and Personal (PE); and six related to Hofstede’s cultural dimensions: Indulgence/Restraint (IVR), Individualism/Collectivism (IDV), Masculinity/Femininity (MAS), Uncertainty Avoidance (UA), Long-Term Orientation (LTO), and Power Distance (PD)—load onto two higher-order constructs: SX and Cultural Aspects (CA). This hierarchical structure, consistent with contemporary factor-analytic guidelines [45], allows SEM to rigorously assess the overarching hypothesis that cultural orientations systematically shape students’ overall perceptions and experiences at their institutions.

3. Related Studies

Recent scholarship underscores a growing reliance on SEM to elucidate the determinants of student satisfaction within HEIs. For example, Batista-Toledo and Gavilan employed SEM to examine post-pandemic commitment in blended-learning environments, isolating the structural pathways through which hybrid instruction shapes perceived satisfaction [46]. Complementing this work, Murillo-Zamorano et al. demonstrated, via SEM, how the flipped-classroom modality enhances (or in specific contexts constrains) students’ evaluative judgements of their learning experience [47]. Collectively, these investigations illustrate an emergent research trajectory in which SEM is used to model satisfaction metrics situated mainly within the SX’s ED dimension.

Although the extant literature often concentrates on causal relationships tied to adopting specific pedagogical tools and methodologies, SEM is equally well-suited to capturing broader and more transversal causal architectures, precisely the orientation adopted in the present study. Our evaluation scale is anchored in theoretically sound frameworks that have already been subject to extensive psychometric validation, and the empirical findings obtained in our Spanish cultural case study corroborate the scale’s construct validity and predictive utility. Consequently, the method aligns with contemporary best practices. It advances the field by demonstrating how SEM can integrate educational, social, and personal facets of SX into a comprehensive diagnostic model that informs strategic decision-making across diverse institutional contexts.

Nevertheless, a critical review of measurement approaches reveals a persistent methodological pattern: the SX is typically operationalised through self-reported instruments and generic satisfaction surveys, most of which are direct adaptations or partial derivatives of the SERVQUAL framework [3]. This prevailing reliance on broad, service-quality templates, rather than discipline-specific or culturally nuanced scales, can attenuate construct specificity and obscure the contextual factors that drive student perceptions.

4. A Scale to Evaluate SX with Cultural Aspects

The SX-Cultural questionnaire ultimately contains 31 prompts, yet one of them, PE19, an open-ended item that invites free-text reflections on the Personal dimension, was intentionally excluded from the quantitative analyses because its qualitative nature precludes metric aggregation. The statistical validation, therefore, centers on the 30 five-point, closed-ended indicators distributed across the three Student-eXperience dimensions-Educational (ED01 to ED09), Social (SO10 to SO16), and Personal (PE17 to PE18), and the six Hofstede cultural facets, each represented by two items (IVR20 to PD31).

Item wording was refined by an interdisciplinary panel of educationists, psychologists, and UX/HCI specialists, who assessed semantic clarity, cultural relevance, and theoretical alignment before field deployment. Pilot administrations in Chile and Spain were programmed with gender-balanced quotas, following accepted survey methodology guidance to minimize sampling bias and enhance generalizability. Data collection sites were determined by university coordination and access approvals rather than by researcher selection. Within this constraint, we documented the following site characteristics: (i) accredited universities with active undergraduate programs, (ii) evidence of a multicultural student body, (iii) formal institutional authorization for data collection, and (iv) logistical feasibility (i.e., access to classes and official communication channels). Within these institutions, we employed a non-probability convenience sampling strategy.

Ordinal Cronbach’s α coefficients, computed on polychoric correlations, exceeded 0.70 for every sub-scale, and no corrected item–total correlation fell below 0.35, confirming strong internal consistency; full pilot results and reliability statistics have been reported elsewhere [48]. All item texts, in their original Spanish and certified English translation, are provided in

Table A1. Item Catalogue (Spanish-English).

Code	Spanish Wording	English Wording
ED01	Los cursos desarrollados por los profesores son interesantes	The courses delivered by faculty members are interesting
ED02	Los cursos me desafían a dar lo mejor de mi al ser exigentes	The courses challenge me to do my best by being demanding
ED03	Los cursos me han brindado la oportunidad de explorar conceptos en profundidad	The courses have given me the opportunity to explore concepts in depth
ED04	Los cursos me han brindado oportunidades para aplicar lo que he aprendido	The courses have provided opportunities to apply what I have learned
ED05	Las evaluaciones de los cursos han sido justas	Course assessments have been fair
ED06	Los cursos están organizados de manera que funcionan sin problemas	The courses are organised so that they run smoothly
ED07	El horario de los cursos me resulta adecuado	The course timetable is convenient for me
ED08	Los recursos e instalaciones de la universidad me han servido para mi aprendizaje	University resources and facilities have supported my learning
ED09	He podido acceder a recursos e instalaciones específicas cuando lo he necesitado	I have been able to access specific resources and facilities whenever I needed them
SO10	He recibido comentarios útiles sobre mis trabajos académicos	I have received useful feedback on my academic work
SO11	He podido contactar al personal de la universidad cuando lo he necesitado	I have been able to reach university staff when I needed to
SO12	He recibido suficiente orientación en relación a mis cursos	I have received sufficient guidance regarding my courses
SO13	Me siento parte de una comunidad universitaria	I feel part of a university community
SO14	He tenido las oportunidades adecuadas para trabajar con otros estudiantes como parte de mis cursos	I have had adequate opportunities to work with other students as part of my courses
SO15	El personal de la universidad valora las opiniones de los estudiantes	University staff value students’ opinions
SO16	El centro de estudiantes representa efectivamente los intereses académicos de los estudiantes	The student union effectively represents students’ academic interests
PE17	He tenido las oportunidades adecuadas para proporcionar comentarios sobre mis cursos	I have had adequate opportunities to provide feedback on my courses
PE18	Estoy satisfecho/a con la calidad de mis cursos	I am satisfied with the quality of my courses
IVR20	Los estudiantes tienen una visión positiva de la universidad, considerando que es posible divertirse allí	Students have a positive view of the university, believing it is possible to have fun there
IVR21	Tanto profesores como estudiantes aceptan con gusto las preguntas en clase propiciando un ambiente participativo	Both professors and students readily accept questions in class, fostering a participatory environment
IDV22	El sentimiento de competitividad individual entre los estudiantes es bajo; los profesores fomentan el trabajo en equipo	Individual competitiveness among students is low; lecturers encourage teamwork
IDV23	Hay sentido de identidad colectivo; a los estudiantes les importan los resultados generales de las evaluaciones en los cursos	There is a sense of collective identity; students care about overall course results
MAS24	Para los estudiantes es aceptable reducir sus interacciones sociales para mejorar sus resultados académicos	It is acceptable for students to reduce their social interactions to improve their academic results
MAS25	La universidad ofrece recursos para los estudiantes menos favorecidos	The university provides resources for disadvantaged students
UA26	Los estudiantes saben exactamente cómo serán evaluados; disponen de exámenes de años anteriores o información precisa sobre las evaluaciones	Students know exactly how they will be assessed; they have access to past exams or precise assessment information
UA27	La burocracia en la universidad es un problema que consume mucho tiempo de los estudiantes	Bureaucracy at the university is a time-consuming problem for students
LTO28	Los cursos se enfocan en el desarrollo profesional de los estudiantes en base a contenidos de actualidad en ingeniería/ciencia/empresa	Courses focus on students’ professional development using up-to-date engineering, science, and business content
LTO29	Cuando se trata de prepararme para las evaluaciones, planifico con anticipación	When it comes to preparing for assessments, I plan ahead
PD30	Existen mecanismos para que los estudiantes expresen su opinión e influyan en las decisiones de la universidad	Mechanisms exist for students to voice their opinions and influence university decisions
PD31	El trato de los profesores es bastante distante; se les considera una autoridad incuestionable	Professors’ manner is rather distant; they are regarded as unquestionable authorities

, ensuring transparency and facilitating replication of this rigorously vetted 30-item SX-Cultural scale.

5. Methodology

During the 2024–2025 academic year, the 31-item scale was administered online to undergraduate cohorts at two culturally contrasting HEIs: one in Latin America and one in Spain. Of 321 submissions, 302 were fully completed, yielding an attrition rate of 5.9%. The final data set includes students from seventeen nationalities, ensuring the heterogeneity needed for cross-cultural validation. Of these respondents, 52 were ERASMUS mobility students drawn from other EU member states. Their presence adds further cultural heterogeneity beyond the host country and, thus, enhances the cross-cultural relevance of the validation exercise. Participation was voluntary and anonymous. Data were collected online after obtaining formal authorization from both universities, and informed consent was included at the beginning of the scale.

We begin this section by outlining the analytic trajectory that underpins our structural equation modeling. First, we articulated a two-tier framework in which the thirty closed-ended items load on nine first-order factors that converge, respectively, on the higher-order constructs SX and cultural aspects; this specification was implemented in Jamovi (v. 2.4.8.0) using the SEMlj (v. 1.2.3) module, which leverages the Lavaan engine for latent-variable analysis.

Because every indicator is rated on a five-point scale, we employed diagonally weighted least squares with mean- and variance-adjusted statistics (DWLS/WLSMV) to estimate polychoric correlations and obtain stable parameter estimates, evaluating model adequacy against Hu and Bentler’s comparative-fit and error-fit thresholds for CFI, TLI, SRMR, and RMSEA. We then assessed measurement quality through Composite Reliability (ω), Average Variance Extracted (AVE), and the Heterotrait–Monotrait ratio of correlations (HTMT); modification indices were consulted, but parameters were freed only when justified theoretically, and any Heywood anomalies should be resolved by constraining factor variances or refining item wording.

Given the theoretical structure of the instrument, SEM was selected to validate both the measurement model, comprising first-order latent variables and second-order factors, and to estimate the hypothesized structural path from Cultural Aspects to SX. While Exploratory Factor Analysis (EFA) is commonly used in early-stage instrument development, our approach was theory-driven; therefore, we prioritized Confirmatory Factor Analysis (CFA) as a component of SEM. Moreover, SEM was necessary to account for the hierarchical latent structure and test culture-related constructs’ directional influence. The use of the WLSMV further ensured appropriate handling of the ordinal nature of the Likert-type data.

5.1. Structural Model

An articulated measurement and structural model is indispensable for two reasons: it ensures that the survey items are theoretically anchored in a specific construct and allows the analyst to disentangle substantive relations from random measurement error. Establishing this solid blueprint before testing prevents the post hoc “fishing” that often inflates fit indices and compromises generalizability. In our study, the model shown in Figure 2 provides that foundation by mapping the questionnaire’s conceptual logic onto a two-tier structural equation framework.

H1. 

Higher levels of cultural congruence positively influence the perceived quality of SX in higher education contexts.

At Level 1, the 30 closed-ended items retained from the preliminary scale (ED01 to PD31) operate as observable indicators of nine first-order latent factors. Three of these factors correspond to the core SX dimensions (ED, SO, and PE) while the remaining six embody Hofstede’s cultural facets: IVR, IDV, MAS, UA, LTO, and PD. Formally, the relation is captured by Equation (1)

$$x={\Lambda }_1 \xi +\epsilon ,$$

(1)

where $x$ is a 30 × 1 vector of item scores, ξ is a 9 × 1 vector of first-order factors, ${\Lambda }_1$ contains the corresponding factor loadings, and item-specific noise is isolated in ε.

At Level 2, these nine factors load on two higher-order constructs; Λ₂ is a 9 × 2 matrix of second-order factor loadings, and η is a 2 × 1 vector of second-order latent factors. Residual uniqueness at this level is captured by δ. The SX dimensions trio (ED, SO, and PE) converges on SX, whereas the six cultural facets combine to form CA, expressed in Equation (2) as

$$\xi ={\Lambda }_2 \eta +\delta ,\hspace{1em}\mathrm{with}\hspace{1em}{\eta }^{\top }=(SX, CA)$$

(2)

Finally, Equation (3) introduces the structural coefficient β_CA→SX, which quantifies the extent to which CA shapes SX, leaving any unexplained variance in SX to the disturbance term ζ,

$$SX={\beta }_{CA\to SX} CA+\zeta ,$$

(3)

Equations (1)–(3) collectively establish a two-tier SEM that isolates measurement error, consolidates related dimensions, and rigorously tests the hypothesized cultural influence on holistic student experience (

Table 2. Structural Path Estimates between Cultural Aspects and Student Experience.

Path	β (std)	SE	z	p
CA → SX	0.92	0.06	16.0	<0.001

).

Item 19 was qualitative, related to the PE dimension, and it was, therefore, excluded, leaving a 30-item ordinal data set from N = 321 undergraduates enrolled at culturally contrasting universities. By embedding both measurement precision and theoretical directionality, this hierarchical model enables simultaneous evaluation of (i) the reliability and validity of each content domain and (ii) the overarching cultural effect on SX, thereby providing a methodical analytic scaffold for the empirical results that follow.

5.2. Software Tools

The statistical evaluation of the proposed structural equation models was conducted in Jamovi. This open-source graphical statistics platform interfaces seamlessly with the R ecosystem, facilitating transparent, reproducible workflows [49]. Because the native Jamovi distribution does not include a dedicated engine for SEM, we extended its functionality by installing the SEMlj module [50]. SEMlj serves as a thin wrapper around Lavaan, Rosseel’s widely validated R package for latent-variable analysis, providing a point-and-click interface while retaining the full analytic capabilities of lavaan’s syntax and estimation routines [51]. This architecture allowed us to specify measurement and structural models directly within Jamovi, invoke lavaan’s robust estimation procedures (including diagonally weighted least squares for ordinal indicators), and obtain standardized parameter estimates, fit indices, modification indices, and bootstrapped confidence intervals without leaving the Jamovi environment.

5.3. Estimation Strategy and Model Fit

All survey items were measured on five-point Likert ordered-categorical scales [52]; consequently, parameter estimation employed the DWLS/WLSMV. This estimator delivers unbiased factor loadings and standard errors under moderate sample size, ordinal indicators, and non-normal distributions, which render maximum likelihood assumptions untenable. When the initial solution yielded negative residual variances (Heywood cases), the associated latent-variable variances were fixed to 1.0, following recommended practice for achieving an admissible model while preserving the substantive interpretation of the factor structure. We also use this method because it performs well with N ≈ 300 [51,53].

Model adequacy was gauged with Hu and Bentler’s dual cut-offs (

Table 3. Global Fit Indices (N = 302, DWLS/WLSMV).

Fit Index	Classical Value	Robust (WLSMV) Value	Target
χ2 (df = 395)	675	822	-
CFI	0.991	0.949	≥0.95 (0.90 accept)
TLI	0.990	0.944	≥0.95 (0.90 accept)
SRMR	0.059	0.052	≤0.08
RMSEA (90% CI)	0.049 (0.042–0.055)	0.060 (0.054–0.066)	≤0.06 (0.08 accept)

): CFI/TLI ≥ 0.95, SRMR ≤ 0.08, RMSEA ≤ 0.06, while also reporting “acceptable” thresholds (≥0.90/≤0.08) to accommodate the complexity of second-order models [54,55].

5.4. Reliability, Convergent, and Discriminant Validity

Internal consistency was assessed with composite reliability (ω), a coefficient that remains accurate when loadings differ across items; convergent validity with AVE, the share of variance a factor captures from its indicators; and discriminant validity with HTMT, which outperforms legacy criteria in detecting conceptual overlap [56,57,58].

5.5. Diagnostic Tools and Model Refinement

Anomalies such as factor loadings > 1.00 or negative error variances (Heywood cases) could be handled by constraining factor variances or inspecting item wording [59,60]. Modification indices (MI) were consulted but freed only when theoretically justified [61]. Following guidelines on improper solutions, the two loadings that marginally exceeded unity within the Social dimension were retained because their residual variances remained positive, and their substantive interpretation was clear. Likewise, the three Heywood cases detected for PE, IVR, and MAS were resolved by fixing the corresponding latent variances to 1.0, an adjustment that removed the anomalies without degrading global fit (ΔCFI < 0.005).

5.6. Sample Adequacy and Future Steps

With 302 analyzable cases, the model exceeds the power thresholds recommended for similar complexity and communalities. Complete measurement-invariance testing, e.g., gender or program, is planned for future work following the ΔCFI ≤ 0.010 rule [62].

6. Results

6.1. Data Screening and Global Model Fit

Of 321 responses, 302 were complete (attrition = 5.9%). Response distributions were balanced; skewness/kurtosis < |1| for every item. Table 3 shows the global fit index of our proposal; classical indices signal excellent fit, and robust indices remain within the “good–acceptable” window, indicating that ordinal scaling has only a modest impact on the misfit of our model.

6.2. Measurement Model

Table 4. Measurement Model (First- and Second-Order).

Factor	Items	λ Range	ω	AVE	Issues
ED	9	0.49–0.82	0.90	0.62	-
SO	7	0.53–1.15 1	0.88	0.59	Two λ > 1.00; uniqueness positive, retain
PE	2	0.66–0.85	0.75	0.57	-
IVR	2	0.67–0.69	0.82	0.68	-
IDV	2	0.57–0.61	0.78	0.57	-
MAS	2	0.17–0.50	0.55	0.30	MAS24 low
UA	2	0.90/−0.16	0.63	0.40	UA27 negative
LTO	2	0.44–0.84	0.83	0.62	-
PD	2	0.82/−0.31	0.62	0.42	PD31 negative

¹ Heywood flag; uniqueness remained positive.

summarizes loadings, ω, AVE, and anomalies. All SX facets met λ ≥ 0.50 and AVE ≥ 0.50. Three cultural items (MAS24, UA27, and PD31) displayed weak or negative loadings, signaling the need for analysis or revision.

At the higher level, SX absorbed ED, SO, and PE with λ = 0.94 to 1.07, whereas CA absorbed its six facets with λ = 0.28 to 0.91 (MAS contributed least).

The measurement evidence confirms that the core SX facets are psychometrically sound, whereas items MAS24, UA27, and PD31 warrant immediate revision. These three items showed loadings below 0.50 or negative values within otherwise well-behaved factors, signaling a content-validity gap rather than a structural flaw. We, therefore, flag them for possible rewriting and/or replacement if a new version of the scale is developed; however, this does not affect the scale’s validity in its current form.

6.3. Structural Relations

The hypothesized path CA → SX was strong (β ≈ 0.92, SE = 0.06, z = 16.0, p < 0.001) (Table 2 and Figure A1). A one-SD rise in aggregated cultural orientations predicts nearly a full-SD rise in perceived SX, explaining 70% of SX variance, evidence that macro-cultural schemas powerfully color day-to-day academic life.

6.4. Reliability and Validity

Six of nine first-order constructs surpassed both ω ≥ 0.70 and AVE ≥ 0.50; MAS and PD fell short (Table 3 and

Table 5. Scale Structural Relations.

Factor	ω	AVE	HTMTmax	Decision
ED	0.90	0.62	0.72	Accept
SO	0.88	0.59	0.78	Accept
PE	0.75	0.57	0.54	Accept
IVR	0.82	0.68	0.55	Accept
IDV	0.78	0.57	0.63	Accept
MAS	0.55	0.30	0.66	Revise
UA	0.63	0.40	0.59	Marginal
LTO	0.83	0.62	0.61	Accept
PD	0.62	0.42	0.68	Marginal

). All cross-domain HTMT values were <0.85 and within-CA values < 0.90, confirming discriminant adequacy. A Harman single-factor test yielded a poor fit (CFI = 0.39), mitigating common-method bias concerns.

Although a few items demonstrated weaker reliability, they were retained in the final scale due to their theoretical significance in representing critical aspects of Hofstede’s cultural dimensions. Excluding these items would have undermined the conceptual coverage of constructs such as PD and UA, essential for capturing intercultural variation in student experience. We acknowledge that theory-driven measures sometimes display lower internal consistency than empirically derived scales obtained through EFA, as they prioritize conceptual breadth over statistical optimization. In this case, including these items ensures content validity and preserves the theoretical integrity of the scale.

6.5. Residual Diagnostics and Robustness

Three Heywood cases (PE, IVR, and MAS) disappeared after fixing their latent variances to 1.0, without degrading fit (ΔCFI < 0.005). The largest MI (49) proposed correlating ED8 and ED9 (“University resources and facilities have supported my learning”/“I have been able to access specific resources and facilities whenever I needed them”); given their near-identical semantics, freeing this path would be defensible in a purely predictive model, but was withheld to keep the scale theory-driven. Re-estimating with pairwise deletion produced a virtually unchanged fit, confirming robustness to the missing-data strategy.

7. Discussion

We found that our intercultural SX scale rests on a good psychometric foundation. The hierarchical model fitted the ordinal data convincingly. CFI and TLI exceeded 0.95, while SRMR and RMSEA stayed within accepted bounds, demonstrating that our theorized structure is compatible with the response patterns of 302 undergraduates. Convergent evidence was likewise strong: all three SX dimensions and four of the six cultural facets met the recommended loading and AVE criteria, and six of the nine first-order factors reached satisfactory composite reliability. Crucially, the structural path from CA to global SX emerged extensive and precise (β ≈ 0.92), confirming that national-culture value orientations color students’ cognitive, affective, and behavioral appraisals of institutional life; the significant standardized path coefficient from CA to SX underscores the pivotal role of value orientations as perceptual lenses through which students interpret everyday academic and social encounters. Practically, the finding suggests that institutions striving to enhance experience outcomes should prioritize culturally responsive policies because these initiatives are likely to resonate across diverse student segments.

Nevertheless, the cross-sectional design limits causal inference; future longitudinal and intervention studies are needed to confirm whether modifying cultural touchpoints produces sustained improvements in SX indicators. In line with service-science perspectives, the three SX dimensions are consistent with established accounts of SX while extending them. Specifically, our results complement Kuh’s engagement [24] and Astin’s involvement [26] by showing that cultural orientations provide explanatory power beyond effort and participation, clarify Tinto’s academic/social integration [23] by indicating that culture conditions how integration translates into experience, and echo Lizzio’s five senses [25] by specifying culturally contingent antecedents of capability, connectedness, and purpose.

To our knowledge, validated SX measures seldom integrate national-culture orientations within a hierarchical SEM that jointly estimates measurement and structural relations; this study addresses that gap by modeling second-order constructs and a directional CA → SX path. To rule out the possibility that this significant coefficient is an artefact of common-method bias, we estimated a single-factor Harman model, which exhibited an inferior fit of 0.39, corroborating that the strong CA → SX link reflects substantive relations rather than methodological inflation.

The less robust performance of MAS, UA, and PD stems from narrow, single-polarity item pools and, in two cases, negatively worded statements that constrain variance (e.g., UA27 and PD31). Because these limitations did not degrade global fit or discriminant validity, we interpret them as content-validity gaps rather than structural flaws. In future iterations, we will expand these facets with balanced, context-rich items and verify their behavior through cognitive interviewing and pilot testing.

Practically, our results highlight two priorities for HEI managers. First, the strong ED and SO loadings reaffirm that investments in teaching quality, resource accessibility, and community-building deliver the most outstanding perceptual returns. Second, the magnitude of the cultural path warns against one-size-fits-all strategies; dashboards that juxtapose SX scores with cultural profiles can help anticipate divergent expectations among domestic and international cohorts.

To advance this agenda, we plan three follow-ups: (i) enrich MAS, UA, and PD with additional positively and negatively keyed statements; (ii) conduct multi-group invariance tests across gender, discipline, and nationality to ensure fair benchmarking; and (iii) link SX scores to behavioral outcomes such as retention and alumni engagement to establish predictive validity. Together, these steps will refine our scale, which already provides a rigorous, culturally nuanced lens for diagnosing and improving SX.

8. Study Limitations

Despite some constraints, our study offers a robust and meaningful first validation of a culturally responsive SX scale. The cross-sectional, self-report design does limit causal inference. Yet, it successfully captures authentic perceptions from a sizable and culturally diverse undergraduate sample, providing an essential snapshot against which future longitudinal work can be benchmarked. Focusing on two universities within two national contexts narrows external generalizability. Still, it also enables a controlled instrument test under sharply contrasting cultural profiles, an essential step before wider deployment. Measurement challenges remain, three culture-facet items loaded weakly or negatively—but the overall structure demonstrated excellent global fit, high composite reliabilities for six of nine facets, and apparent discriminant validity, indicating that the core constructs are sound even as specific items warrant refinement. Listwise deletion removed fewer than 6% of cases, and alternative estimators produced virtually identical fit statistics, suggesting that missing data did not materially distort results. Although the sample size sits near the lower bound for complex second-order SEM, the model achieved strong statistical power to detect the principal CA → SX relation, which explained 70% of variance and remained stable across robustness checks. Common-method variance was addressed empirically: the single-factor Harman test showed poor fit, supporting the distinctiveness of the measured constructs. Finally, while formal tests of measurement invariance across sub-groups await future data collections, the findings establish a preliminary baseline: the scale behaves consistently within the analyzed population and yields theoretically coherent effects. These strengths outweigh the acknowledged limitations, lending confidence to the positive results and providing a clear roadmap for incremental improvements in subsequent research phases.

9. Conclusions and Future Work

Our study demonstrates that our SX scale provides a psychometrically sound and theoretically coherent instrument for diagnosing student experience across culturally diverse higher-education settings. Sequential confirmatory-factor and hierarchical SEM analyses applied to 302 undergraduate responses achieved excellent global fit, where CFI and TLI surpassed 0.95, while SRMR and RMSEA remained within recommended limits. A strong structural path from CA to SX (β ≈ 0.92) explained roughly 70% of the variance in overall experience. Six of nine first-order constructs exceeded composite-reliability and convergent validity benchmarks, confirming that the Educational, Social, Personal, and most cultural facets are measured precisely. Content-validity gaps in MAS, UA, and PD did not compromise model integrity; instead, clear priorities for refinement were delineated. Looking ahead, we will enrich these three facets with balanced, context-rich items, conduct multi-group invariance tests across gender, discipline, and nationality, and link SX scores to behavioral outcomes such as retention and alumni engagement to establish predictive validity. By coupling canonical SX dimensions with Hofstede’s cultural lenses and outlining a concrete roadmap for iterative enhancement, we furnish researchers and practitioners with an actionable, culture-attuned diagnostic whose methodological transparency and explanatory power we expect to foster broad adoption, spur cross-institutional replications, and encourage future actions that advance global SX scholarship and institutional sustainability.