Validating a Psychometric Instrument for Assessing Students’ Digital Skills: A Latent Variable Approach for Policy-Oriented Educational Research

Validating a Psychometric Instrument for Assessing Students’ Digital Skills: A Latent Variable Approach for Policy-Oriented Educational Research | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Validating a Psychometric Instrument for Assessing Students’ Digital Skills: A Latent Variable Approach for Policy-Oriented Educational Research Emil Frashëri, Marinela Teneqexhi This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7289357/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract This study presents the development and validation of a 24-item instrument designed to assess students’ digital skills, an essential competency in modern education. Grounded in a robust conceptual framework, the instrument captures key dimensions of digital literacy and was tested using cross-sectional data alongside advanced latent variable modeling techniques. The analytical methods applied included Confirmatory Factor Analysis (CFA), Exploratory Structural Equation Modeling (ESEM), bifactor-CFA, and bifactor-ESEM. Among these, ESEM yielded the best model fit, offering a nuanced representation of the multidimensional structure of students’ digital skills. The model demonstrated strong psychometric properties, including high internal consistency, solid construct validity, and measurement invariance across gender. Predictive validity was also confirmed through significant associations with relevant educational outcomes. These findings support the instrument’s application in large-scale assessments and policy initiatives aimed at improving digital literacy. The validated framework provides a foundation for evidence-based decisions in curriculum design, teacher training, and educational planning, and is well-suited for integration into broader SEM framework-based research. Applied Statistics Educational Philosophy and Theory Digital skills assessment Educational policy Exploratory Structural Equation Modeling (ESEM) Instrument development Psychometric validation Figures Figure 1 Figure 2 1. Introduction The integration of information and communication technologies has shifted education toward digital knowledge practices, aligned with both the potential and limitations of technology (Ramos et al., 2023). As higher education transitions from traditional to technology-driven teaching, it faces significant challenges (Raji et al., 2023). Developing students’ digital skills is now a critical priority, with schools playing a central role in preparing youth for the demands of the 21st century. Embedding digital competencies into curricula fosters creativity, innovation, and adaptability in a digital society (Erwin & Mohammed, 2022). This integration is increasingly recognized as essential for preparing a future-ready workforce (Kryukova et al., 2022; Sotelo-Núñez et al., 2024). Kayyali (2024) emphasizes that digital literacy should be a core, not complementary, component of all academic programs. These competencies, developed during higher education, form a vital foundation for students’ future professional success (López, 2023). The European Union has paid great attention to improve the level of digital competences and skills at all stages of education and training, for all segments of the population (Council of the European Union, 2018). The State of the Digital Decade 2024 1 report highlighted the need for European Union countries to strengthen cooperation on more ambitious collective efforts, prioritizing investment in digital education and skills, to achieve the Digital Decade targets and objectives, as a necessity for ensuring future economic prosperity. A European Union policy initiative in support of digital education ecosystem, enhancing digital skills for the digital transformation, is the Digital Education Action Plan (2021–2027) 2 . 1.1 Digital transformation in Albanian education system Albania, as part of the Western Balkans, is gradually progressing toward EU digitalization standards, though its education system remains in the early stages of digital transformation. Higher education continues to rely heavily on traditional teaching methods, hindered by infrastructure limitations and limited access to digital tools. Until recently, ICT education was limited to lower secondary levels, with informatics offered optionally in upper secondary schools (Miço & Zaçellari, 2020). Digital skill levels in Albania remain significantly below the EU average. Only 24% of individuals aged 16–74 had basic digital skills in 2021, and just 4% had above-basic skills (OECD, 2024). In this context, the development of digital and ICT competencies in Albanian public education is seen as a necessity (UNESCO, 2017; European Commission, 2021). Recognizing this gap, Albania’s “Digital Agenda Strategy 2022–2026” promotes ICT integration across all education levels. Recent reforms include coding courses in primary schools and revised ICT curricula. European Training Foundation (ETF) and the Joint Research Centre (JRC) of the European Commission in the framework of the Digital Agenda for the Western Balkans have launched SELFIEforTeachers (SfT ), a measurement tool for assessing and improving primary and secondary teachers' digital literacy. Despite these efforts, tangible progress in digital inclusion remains limited. 1.2 The problematic situation created and the need for intervention Educators face a growing challenge: the digital skills of graduating students often fall short of what employers expect, limiting both student success and workforce readiness (Feijao et al., 2021; Kayyali, 2024). Studies show that digital competence remains low among both teachers and students, even among so-called “digital natives” (Gallardo-Echenique et al., 2015). This gap calls for urgent updates to digital education programs. To design effective interventions, educators need reliable tools to assess digital skills using sound measurement methods (Siddiq et al., 2016; Laanpere, 2019; Helsper et al., 2020). However, rapid technological change has made it difficult to define and measure these competencies consistently (Sillat et al., 2021). A dynamic, future-focused framework must be envisioned, one through which students are empowered with adaptable digital skills to lead and thrive in a rapidly evolving, knowledge-driven world. 2. Conceptualization of digital skills Over the past two to three decades, a range of conceptual terms has been employed to describe students’ engagement with digital technologies, including ICT competences, digital literacy, media literacy, and 21st-century skills (Sánchez-Caballé et al., 2020). These terms reflect the evolving nature of digital competence, which was initially understood as a set of basic technical skills (Richter et al., 2001; Selber, 2004). However, contemporary perspectives recognize digital competence as a multidimensional construct encompassing technical abilities, cognitive skills, and socio-emotional attributes, such as critical thinking, problem-solving, and adaptability, essential for meaningful participation in a knowledge-based society (Van Laar et al., 2017; Fan & Wang, 2022; Widowati et al., 2023). This conceptual expansion has led to the development of various frameworks tailored to specific educational and professional contexts. Voogt and Roblin (2012), through a comparative analysis of eight major frameworks, identified a convergence around core digital competencies. Among these, the European Commission’s DIGCOMP framework has gained prominence for its comprehensive and adaptable structure. Developed by the Joint Research Centre Institute for Prospective Technological Studies (JRC-IPTS), DIGCOMP synthesizes insights from 15 ICT literacy models and international assessments such as PISA and PIAAC (Ferrari, 2013; Vuorikari et al., 2016; Carretero et al., 2017). The framework defines five competence areas, Information, Communication, Content Creation, Safety, and Problem Solving, across five proficiency levels, allowing for nuanced assessment and targeted development. Subsequent revisions of DIGCOMP (e.g., DigComp 2.0 and 2.1) have enhanced its applicability across diverse sectors and learner groups, reflecting the increasing need for context-sensitive approaches to digital competence. These developments underscore the importance of a flexible, future-oriented framework capable of equipping individuals with adaptable digital skills in response to the dynamic demands of the digital age. 2.1. Students’ digital skills assessment. The background for the empirical research Ma and Ismail (2025) conducted a comprehensive study, combining bibliometric analysis and systematic review, to examine the landscape of research on digital competencies within the education sector. Their findings indicate that digital competencies remain a relatively recent but rapidly expanding area of inquiry, with significant potential for further exploration. Although scholarly interest in digital competencies in higher education has grown markedly, particularly since 2021, there remains no unified conceptualization of what constitutes students’ digital competencies (Ilomäki et al., 2014; Spante et al., 2018) or digital skills (Barbazan et al., 2021; Paz et al., 2021; Perdomo et al., 2020). For the purposes of the present study, digital skills are defined in accordance with the International Telecommunication Union (ITU, 2018) as “ the ability to use ICTs in ways that help individuals to achieve beneficial, high-quality outcomes in everyday life for themselves and for others, and to reduce potential harm associated with more negative aspects of digital engagement ”. This definition has been widely adopted in recent literature (e.g., van Deursen et al., 2016; Helsper & van Deursen, 2018; Haddon et al., 2020; Helsper et al., 2020), and serves as a robust conceptual foundation for assessing digital skills, particularly when aligned with the European Commission’s DIGCOMP framework. The present study employs the youth Digital Skills Indicator (yDSI), developed by Helsper et al. (2020), as the primary measurement instrument. The yDSI is a cross-nationally validated tool designed to assess digital skills among young people across four core dimensions: (1) technical and operational skills, (2) information navigation and processing, (3) communication and interaction, and (4) content creation and production. Each dimension captures both functional and critical aspects of digital engagement, reflecting the multifaceted nature of digital competence in contemporary society. Although the yDSI is conceptually grounded in the DIGCOMP framework (European Commission, 2020), it excludes the “problem-solving and safety” dimension. This exclusion is based on evidence from the literature suggesting that elements of this domain are already embedded within the other four dimensions. The yDSI scale comprises 24 items, carefully developed and validated through cognitive interviews, exploratory and confirmatory factor analyses, measurement invariance testing, and performance-based assessments. Special attention was given to avoiding common pitfalls in self-assessment tools, such as social desirability and acquiescence bias, by piloting item wording and response formats to ensure strong psychometric properties. The yDSI emphasizes the importance of equipping young individuals with foundational skills across all four dimensions to enable full and meaningful participation in a digital society. Unlike measures of digital self-efficacy, which focus on perceived competence, the yDSI prioritizes actual skill acquisition as a more reliable predictor of positive digital engagement. The instrument draws on both academic literature and grey literature sources (e.g., Cortesi et al., 2020), and has been successfully implemented in large-scale initiatives such as the EU-funded ySKILLS project under Horizon 2020 (Helsper et al., 2020). As Helsper et al. (2020) pointed out, yDSI can be adopted in other projects with young people. 2.2 Types of methodologies used to measure digital skills Over the past two decades, digital skills have been assessed using a range of methods, including direct, indirect, and hybrid approaches. Among these, performance-based assessments, such as PIAAC (OECD) and ICILS (NCES), are considered the most valid, as they involve authentic, task-based interactions with digital environments (Darling-Hammond & Adamson, 2010). However, their high cost, context specificity, and labor-intensive nature limit their scalability for large-scale studies (Vuorikari et al., 2025; van Deursen & van Dijk, 2010). Consequently, indirect methods, particularly self-reported questionnaires, are more commonly used due to their efficiency and ease of administration. These tools have evolved through three generations to address issues such as social desirability bias and overestimation of competence (Ballantine et al., 2007; Porat et al., 2018). The first-generation Digital Skills Indicator (DSI) and its updated version, DSI 2.0 (Vuorikari et al., 2022), laid the groundwork for improved psychometric design. The second-generation Internet Skills Scale (ISS) by van Deursen et al. (2016) introduced more behaviorally anchored items. The third-generation Youth Digital Skills Indicator (yDSI), developed by Helsper et al. (2020), further refines this approach by focusing on four core dimensions: technical and operational skills, information navigation, communication and interaction, and content creation. Excluding problem-solving and safety due to conceptual overlap, the yDSI offers a validated, cross-national tool for assessing youth digital skills, emphasizing actual ability over perceived self-efficacy. 2.3 Research Aims Research on digital competencies and skills has expanded considerably over the past two decades, with a notable acceleration in recent years. While much of the existing literature has focused on the theoretical conceptualization of digital skills, relatively limited attention has been devoted to the development of empirical models that can inform practical applications. As a degree of conceptual consensus has emerged, there is now a pressing need to advance models that explain the interaction between operationalized indicators of digital skills, particularly in light of their multidimensional and hierarchical structure. This shift from conceptual clarity to empirical modeling holds important implications for policy. Robust, evidence-based models are essential for designing targeted interventions that address specific skill gaps across different learner populations. The present study contributes to this effort by assessing university students’ functional digital skills using the short version of the Youth Digital Skills Indicator (yDSI) (Helsper et al., 2020). By applying latent variable modeling techniques, this study aims to uncover the structural relationships among functional digital skill dimensions, offering insights that can support the development of more effective, data-driven digital education policies. 2.4 Measurement scale assessment When developing a quantitative instrument, it is essential to rigorously assess its psychometric quality, particularly in terms of reliability and validity (Scholtes et al., 2011). Even if an instrument has been previously validated, re-evaluating its properties in new contexts remains best practice (Bandalos, 2018). In recent years, researchers across various countries have increasingly employed latent variable models to assess digital competencies in both pre-university and higher education contexts. These models frequently utilize Exploratory Factor Analysis (EFA), as well as first-order and, occasionally, second-order Confirmatory Factor Analysis (CFA). However, as noted by Marsh and colleagues (Marsh, Hau, & Grayson, 2005; Marsh, 2007; Marsh et al., 2010), when measuring multifactorial constructs with factors represented by at least five items each, the restrictive independent clusters model (ICM) of CFA often fails to achieve acceptable model fit. To address this, newer approaches have been proposed that account for the fallibility of items as indicators of single latent dimensions and the hierarchical structure of the constructs being measured. Among these, Exploratory Structural Equation Modeling (ESEM), bifactor-CFA (B-CFA), and bifactor-ESEM (B-ESEM) have emerged as valuable tools for addressing construct-relevant multidimensionality. Although still limited, the application of B-CFA and B-ESEM in the assessment of digital skills is gaining traction in the literature. 2.4.1 Modeling Latent Structures in Digital Skills Assessment Exploratory Factor Analysis (EFA) is a data-driven technique used to identify the underlying latent dimensions that explain the covariance among observed variables. It allows for free estimation of factor loadings and cross-loadings to derive a simple, interpretable structure (Brown, 2015). While suitable for exploratory purposes, EFA lacks confirmatory power, limiting its application in complex, theory-driven research (Morin et al., 2013; Marsh et al., 2009). Confirmatory Factor Analysis (CFA), introduced by Jöreskog (1969), is a theory-driven approach that tests how well the data fit a hypothesized factor structure. It assumes items load on a single latent factor, with cross-loadings fixed to zero (Marsh, 2007). Although CFA offers advantages in model parsimony, error correction, and measurement invariance (Morin et al., 2020), its restrictive assumptions can lead to inflated factor correlations and compromised discriminant validity in multidimensional constructs (Marsh et al., 2009; Schmitt & Sass, 2011). Exploratory Structural Equation Modeling (ESEM), developed by Asparouhov and Muthén (2009), integrates EFA within the SEM framework, allowing for cross-loadings, while retaining confirmatory capabilities. ESEM models are more flexible and typically yield better fit and more realistic factor correlations than CFA (Marsh et al., 2014). They are particularly useful for assessing instruments with conceptually related dimensions, such as digital competencies (Morin et al., 2016a, 2016b). Bifactor-CFA (B-CFA) models allow each item to load on a general factor and one specific factor, enabling simultaneous assessment of general and domain-specific variance (Reise et al., 2010). However, by constraining cross-loadings to zero, B-CFA may overestimate general factor loadings and distort model interpretation (Murray & Johnson, 2013). Bifactor-ESEM (B-ESEM) addresses these limitations by combining the bifactor structure with ESEM’s flexibility, allowing cross-loadings to approach, but not equal, zero through orthogonal target rotation. This approach better captures the hierarchical and multidimensional nature of constructs and avoids parameter inflation (Morin et al., 2016b). B-ESEM also supports model stability and the development of higher-order models (Marsh et al., 2020). 2.4.2 Modeling Interdependent Dimensions of Digital Skills The latent construct of digital skills assessed in this study comprises four interrelated dimensions. Both the Youth Digital Skills Indicator (yDSI; Helsper et al., 2021) and the DigComp 2.2 framework (Vuorikari et al., 2022) emphasize that these dimensions are not discrete but functionally interconnected. The yDSI provides both theoretical and empirical justification for a four-dimensional model, highlighting the critical interdependencies among the dimensions. Similarly, DigComp 2.2 outlines the conceptual and practical links across digital competence areas. Among these, Technical and Operational Skills serve as a foundational layer, enabling the development of other competencies. Information Navigation and Processing Skills depend on technical fluency and, in turn, support Communication and Interaction Skills . Effective digital communication requires not only platform proficiency but also the ability to interpret and disseminate information. Finally, Content Creation and Production Skills represent a culmination of the other three, requiring technical tools, information literacy, and communication strategies for effective digital content development. Empirical evidence supports these interdependencies. For instance, Pongrac et al. (2025) found significant positive correlations among the four digital skill dimensions. Therefore, statistical models used to assess this construct must account for these interrelations. Exploratory Structural Equation Modeling (ESEM) is particularly well-suited for this purpose, as it naturally accommodates cross-loadings and provides a more realistic representation of overlapping constructs, especially in educational research (Sillat et al., 2021). Recent studies (e.g., Dierendonck, 2023) have demonstrated that ESEM often yields superior model fit compared to traditional CFA or bifactor-CFA models when applied to multidimensional educational constructs. Furthermore, ESEM can be extended to hierarchical structures, such as hierarchical ESEM (H-ESEM) or bifactor-ESEM (B-ESEM). The B-ESEM model is especially valuable for modeling digital skills because: It incorporates cross-loadings, allowing for a more comprehensive assessment of theoretically grounded skill dimensions (Runge et al., 2023). It enables the examination of how general and specific skill dimensions relate to students’ actual technology use. 2.4.3 Exogenous Variables as Predictors of the Digital Skills Construct Academic procrastination is the tendency to delay academic tasks or exam preparation without valid justification (Solomon & Rothblum, 1984). The Academic Procrastination Scale Short Form (Yockey, 2016), a five-item unidimensional measure, is widely regarded as a reliable and valid tool. Recent studies indicate a negative relationship between digital competence and academic procrastination, suggesting that students with stronger digital skills are less prone to procrastinate (Fayda-Kınık, 2023; Yuan et al., 2024). Self-efficacy , as defined by Bandura (1977), refers to individuals’ beliefs in their ability to organize and execute actions to achieve specific goals. In education, academic self-efficacy reflects students’ confidence in their learning and academic success. The three-item General Self-Efficacy Scale Short Form (Beierlein et al., 2013) demonstrates strong psychometric properties, including unidimensionality and cross-linguistic validity. Research consistently shows a positive association between academic self-efficacy and digital competence across various skill domains (Falma & Putra, 2025; Javier-Aliaga et al., 2024). Affinity for technology refers to an individual’s interest, comfort, and confidence in engaging with digital tools (Franke et al., 2019). The four-item ATI-S scale (Wessel et al., 2019) reliably measures this construct across gender and age groups. Research shows that students with higher technology affinity tend to excel in digital skill domains, including programming and technical problem-solving (Jokisch et al., 2025; Cabezas-González et al., 2023). 3. Methods 3.1 Participants and Procedure Data for this study were collected between January and May 2025 from students at Fan S. Noli University of Korça, located in the southeastern region of Albania. The survey instrument was primarily based on the short version of the Youth Digital Skills Indicator (yDSI; Helsper et al., 2020). The core of the instrument consisted of 24 items measuring functional digital skills across four interrelated dimensions: Technical and Operational Skills (TO), Information Navigation and Processing Skills (INP), Communication and Interaction Skills (CI), and Content Creation and Production Skills (CCP). Responses were recorded using a six-point Likert scale: 0 = I do not understand what you mean by this ; 1 = Not at all true of me ; 2 = Not very true of me ; 3 = Neither true nor untrue of me ; 4 = Mostly true of me ; 5 = Very true of me . An additional response option, I do not want to answer , was coded as 99 and treated as a missing value. To assess predictive validity, the questionnaire included items for three exogenous latent variables: academic procrastination (5 items), self-efficacy (3 items), and technology affinity (4 items, including two reverse-coded), all rated on a five-point Likert scale (1 = Strongly disagree to 5 = Strongly agree ). Prior to data collection, the questionnaire was reviewed by ICT and linguistics experts to ensure content validity and accurate translation into Albanian. Following institutional approval, the survey was distributed via the university’s official website using Google Forms. Completion time was approximately 10 minutes. A total of 603 students participated (Mean age = 22.07 years; 69.3% female). The sample size closely approximated the recommended minimum of 630 participants, as calculated using the Soper (2025) a priori sample size calculator for structural equation modeling, based on an anticipated effect size of 0.15 and a desired statistical power level of 0.80 (4 latent variables and 24 observed variables). 3.2 Psychometric Modeling and Analytical Approach To assess the temporal stability of the instrument, a test–retest procedure was conducted with a randomly selected subsample of 30 students. Participants completed the self-report questionnaire twice, with a three-week interval between administrations. The Spearman’s rank correlation coefficient between the two time points was high (ρho = 0.794), indicating strong test–retest reliability and suggesting that the instrument yields consistent results over time. To examine the factorial structure of students’ functional skills, the 24-item scale was analyzed using a series of latent variable models: Confirmatory Factor Analysis (CFA), Exploratory Structural Equation Modeling (ESEM), bifactor-CFA (B-CFA), and bifactor-ESEM (B-ESEM), following the comparative framework of Morin et al. (2016a). This approach enables the identification of construct-relevant psychometric multidimensionality. CFA applied the restrictive Independent Cluster Model (ICM), allowing items to load only on their designated factors, with all cross-loadings fixed to zero. ESEM, using oblique TARGET rotation (Asparouhov & Muthén, 2009), allowed estimation of primary loadings while constraining cross-loadings to be as close to zero as possible. Bifactor-CFA modeled a general digital skills factor alongside four specific dimensions, assuming orthogonal factors and no cross-loadings. Bifactor-ESEM extended this by combining the bifactor structure with ESEM’s flexibility, allowing cross-loadings to be freely estimated but constrained as close as possible to zero, offering a more realistic representation of the hierarchical and multidimensional nature of digital skills. 3.3 Data Analysis Data analysis was performed using Mplus Version 8.11 (Muthén & Muthén), with each of the four competing models estimated independently. Given the six-category ordinal nature of the digital skills items and their approximately normal distribution (see Table 1, Appendix), model robustness was evaluated using two recommended estimators: the Robust Maximum Likelihood (MLR), which treats ordinal data as continuous and applies Satorra-Bentler (Satorra & Bentler, 1994) corrections (Rhemtulla et al., 2012) ) to adjust for kurtosis, which is a primary concern in non-normal data, and the Weighted Least Squares Mean and Variance adjusted (WLSMV), which treats data as categorical and uses polychoric correlations. Following Finney and DiStefano (2013), WLSMV is particularly suitable for ordinal data with five or fewer categories and performs reliably with sample sizes exceeding 200 (Beauducel & Herzberg, 2006; Bandalos, 2014; Flora & Curran, 2004). Model fit was evaluated using standard indices: CFI and TLI (incremental fit), and RMSEA with 90% CI and SRMR (absolute fit), following established thresholds (Hu & Bentler, 1999; Marsh et al., 2004). CFI and TLI values ≥ .90/.95 and RMSEA ≤ .08/.05, along with SRMR ≤ .08, indicate acceptable to excellent fit. Due to its sensitivity to sample size and model misspecification, the chi-square test was interpreted alongside other indices (Marsh et al., 2005). For ordinal data, SRMR has been shown to outperform RMSEA (Shi et al., 2020). Although ordinal variables are inherently non-normally distributed (Kaplan, 2009), skewness and kurtosis values for all items were below 1 in absolute terms, indicating only minor deviations from normality (West et al., 1995; Hair et al., 2010). According to Muthén and Kaplan (1992), under conditions of approximate normality, minimal distortion is expected in model estimation. When sample sizes are adequate, both Maximum Likelihood (ML) and Weighted Least Squares (WLS) estimators perform reliably. All competing models demonstrated good to excellent fit across multiple indices (see Table 1). The only fit index that approached concern for categorical data was the RMSEA, with values ranging from .075 to .090, near the conventional cutoff of .08, indicating acceptable model fit. Mplus Version 8.11 employs a modified computation of the SRMR (Asparouhov & Muthén, 2018), for which values below .08 are considered indicative of good fit. Next, we compared four competing models to evaluate their relative fit. Specifically, comparisons were made between ESEM and CFA, B-CFA and CFA, and B-ESEM and ESEM, using two robust estimators. Following the framework proposed by Morin et al. (2016b), model comparisons were based on differences in goodness-of-fit indices and measurement quality indicators. Table 1. Model fit evaluation of four competing models with two different estimators MLR estimator for continuous data Model df CFI TLI RMSEA 90% CI SRMR AIC BIC SABIC Meets Criteria M1 CFA 540.5 246 .954 .949 .045 (.040 .050) .033 37527.4 37870.2 37622.6 Yes M2 ESEM 413.8 186 .965 .948 .045 (.039 .051) .021 37446.3 38052.9 37614.8 Yes M3 B-CFA 484.9 228 .960 .952 .043 (.038 .049) .029 37463.0 37885.0 37580.2 Yes M4 B-ESEM 366.8 166 .969 .948 .045 (.039 .051) .017 37352.1 38046.5 37544.9 Yes WLSMV estimator for categorical data M1 CFA 1267.6 246 .981 .978 .083 (.079 .088) .026 - - - No/Yes M2 ESEM 1086.5 186 .983 .975 .090 (.085 .095) .019 - - - No/Yes M3 B-CFA 990.5 228 .986 .983 .075 (.070 .080) .024 - - - Yes M4 B-ESEM 803.2 166 .988 .980 .080 (.075 .086) .015 - - - Yes - Chi-square; df - degrees of freedom; TLI - Tucker-Lewis Index; CFI - Comparative Fit Index; RMSEA - Root Mean Square Error of Approximation [90%CI]; SRMR - Standardized Root Mean Square Residual; AIC - Akaike Information Criterion; BIC - Bayes Information Criterion; SABIC - Adjusted Bayes Information Criterion. For nested models, differences smaller than .01 in CFI and TLI, and less than .015 in RMSEA and SRMR, are generally considered indicative of equivalent model fit (Chen, 2007; Cheung & Rensvold, 2002). When using the MLR estimator, models with lower AIC, BIC, and aBIC values are preferred. As shown in Table 2, the ESEM model demonstrated a relatively better fit to the data compared to the CFA model. Table 2. Comparison of four competing models with two different estimators MLR Model comp. Δ χ2 Δ df Δ CFI Δ TLI Δ RMSEA Δ SRMR Δ AIC Δ BIC Δ aBIC Meets criteria M2 vs. M1 -126.710 -60 .011 -.001 .000 -.012 -81.060 182.660 -7.830 No/Yes M3 vs. M1 -55.580 -18 .006 .003 -.002 -.004 -64.360 14.760 -42.400 No M4 vs. M2 -46.970 -20 .004 .000 .000 -.004 -94.250 -6.350 -6.350 No WLSMV Model comp. Δ χ2 Δ df Δ CFI Δ TLI Δ RMSEA Δ SRMR Meets criteria M2 vs. M1 -181.070 -60 .002 -.003 .007 -.007 No M3 vs. M1 -277.040 -18 .005 .005 -.008 -.002 No M4 vs. M2 -283.330 -20 .005 .005 -.010 -.004 No Although the ESEM model showed slightly better fit than the CFA model, the differences were not substantial across all indices. Improvements were primarily observed in SRMR, AIC, aBIC, and to a lesser extent, in RMSEA, while BIC favored the more parsimonious CFA model. Except for the CFI difference under the MLR estimator, most fit index differences fell within the thresholds for model equivalence (Chen, 2007; Cheung & Rensvold, 2002). However, the interpretation of model fit differences remains a topic of ongoing debate (Murray & Johnson, 2013), and should be complemented by an examination of parameter estimates. As shown in Table 3, the ESEM model yielded lower factor loadings and inter-factor correlations compared to CFA, with mean reductions of 7.06% and 6.59%, respectively, across estimators. The largest decrease in item loadings was observed for the INP factor (12.71%), while the smallest was for CCP (3.53%). A similar pattern was observed in the inter-factor correlations. Based on the practical recommendations of van Zyl and ten Klooster (2022), and as supported by Morin et al. (2020), the ESEM model was retained for further analysis. These authors suggest that even if ESEM does not show better overall fit than CFA, it should still be preferred when it produces lower factor correlations, as this indicates better discriminant validity between the constructs. Table 3. Factor loadings and factor correlations comparison for CFA and ESEM models (continuous data) Items TO CFA/ESEM Items INP CFA/ESEM Items CI CFA/ESEM Items CCP CFA/ESEM to1 .657/.497 inp1 .758/.541 ci1 .853/.668 ccp1 .845/.773 to2 .829/.767 inp2 .840/.467 ci2 .870/.807 ccp2 .849/.786 to3 .866/.911 inp3 .825/.600 ci3 .902/.892 ccp3 .851/.846 to4 .861/.932 inp4 .708/.809 ci4 .889/.962 ccp4 .829/.824 to5 .678/.565 inp5 .750/.827 ci5 .747/.674 ccp5 .863/.883 to6 .822/.775 inp6 .745/.794 ci6 .826/.741 ccp6 .805/.752 Mean .786/.741 Mean .771/.673 Mean .848/0.791 Mean .840/.811 % decrease -5.644 -12.711 -6.743 -3.530 Grand Mean .811 /.754 % decrease -7.063 Factor correlations CFA/ESEM TO Diff. % INP Diff. % CI Diff. % CCP TO 1 INP .795/.728 -8.428 1 CI .751/.719 -4.261 .764/.661 -13.482 1 CCP .749/.728 -2.804 .808/.749 -7.302 .784/.751 -4.209 1 Mean decrease % -5.164 -10.392 -4.209 Grand Mean decrease % -6.588 In the ESEM model, the mean factor loadings across the four dimensions of the Digital Skills Indicator (DSI) ranged from .673 to .811, suggesting well-defined latent constructs. When treating variables as continuous, both CFA and ESEM models demonstrated strong composite reliability, with values ranging from .898 to .939 for CFA and from .838 to .920 for ESEM, well above the recommended threshold of .70 (see Table 4). These results indicate excellent internal consistency, suggesting that each set of items reliably measures its corresponding latent factor. In the ESEM model, two sub-constructs, Communication and Interaction (CI) and Content Creation and Production (CCP), demonstrated particularly strong internal cohesion, with composite reliability (CR) values of .912 and .920, respectively. Table 4. Results of Composite Reliability, Convergent and Discriminant validity for CFA and ESEM models with continuous data. CR (> .7) AVE (> .5) Factors CFA (continuous data) TO INP CI CCP .908 .624 TO .790* .898 .597 INP .795 .772* .939 .721 CI .751 .764 .849* .935 .707 CCP .749 .808 .784 .841* CR (> .7) AVE (> .5) Factors ESEM (continuous data) TO INP CI CCP .886 .576 TO .759* .838 .473 INP .728 .688* .912 .637 CI .719 .661 .798* .920 .659 CCP .728 .749 .751 .812* Note . CR-Composite Reliability; AVE- Average Variance Extracted; * Square Root of AVE Regarding construct validity, both CFA and ESEM models showed satisfactory convergent and discriminant validity. For the CFA model, all Average Variance Extracted (AVE) values exceeded the recommended threshold of .50, indicating that over half of the variance in each item was explained by its respective factor. In the ESEM model, all AVE values also exceeded .50, except for the Information Navigation and Processing (INP) factor, which had an AVE of .473, slightly below the ideal cutoff, but still within an acceptable range depending on the context. To assess discriminant validity, the square root of each AVE was compared to the inter-factor correlations. According to Fornell and Larcker (1981), discriminant validity is supported when the square root of a factor’s AVE is greater than its highest correlation with any other factor. This criterion was met for all sub-dimensions except INP, which showed overlapping variance with Technical and Operational (TO) and Content Creation and Production (CCP) in both CFA and ESEM models, indicating potential discriminant validity concerns for this sub-dimension. When accounting for the categorical nature of the indicators, the four latent dimensions of the measurement scale demonstrated strong internal consistency in both the CFA and ESEM models. Composite reliability (CR) values ranged from .924 to .953 for CFA and from .886 to .934 for ESEM, all exceeding the recommended threshold of .70 (see Table 5). Additionally, all AVEs for the four sub-dimensions in both models were above .50, supporting the presence of convergent validity. Discriminant validity was confirmed for both the CFA and ESEM models using the Fornell-Larcker criterion. This finding supports the conceptual and statistical distinctiveness of the four latent sub-constructs, lending credibility to the measurement models. A minor exception was observed in the ESEM model for the Information Navigation and Processing (INP) factor, where the square root of its AVE (.756) was slightly lower than its correlation with the Content Creation and Production (CCP) factor (.771). In terms of measurement quality, all factor loadings exceeded the recommended threshold of .35 (see Table 2, Appendix), ranging from .467 (INP2) to .962 (CI4), and were statistically significant ( p < .001). These results indicate that the items are strongly associated with their respective latent constructs. Cross-loadings were minimal (< .30), with most below .10, supporting excellent discriminant validity. According to van Zyl and ten Klooster (2022), cross-loadings below .50 are acceptable in ESEM models, further confirming the robustness of the factor structure. Table 5. Results of Composite Reliability, Convergent and Discriminant validity for CFA and ESEM models with categorical data CR (> .7) AVE (> .5) Factors CFA (categorical data) TO INP CI CCP .932 .697 TO .835* .924 .672 INP .787 .820* .953 .771 CI .753 .771 .878* .950 .759 CCP .752 .809 .804 .871* CR (> .7) AVE (> .5) Factors ESEM (categorical data) TO INP CI CCP .918 .657 TO .811* .886 .572 INP .738 .756* .923 .670 CI .719 .683 .818* .934 .705 CCP .730 .771 .762 .840* Note. CR-Composite Reliability; AVE- Average Variance Extracted; * Square Root of AVE Additionally, the R-squared values for the scale items ranged from .426 to .822, reflecting a healthy proportion of explained variance. These values suggest that the underlying construct, Digital Skills Indicator (DSI), accounts for a substantial portion of each item's variance, reinforcing the model’s strong psychometric properties and construct validity. All item uniqueness values fell between .10 and .90, indicating that each item contributed meaningfully to the factor structure without being redundant or overly error-prone. During the model comparison phase, the bifactor-CFA model (M3) also outperformed the standard CFA model (M1). However, the primary focus was on evaluating M4 against M2. In comparing the bifactor-ESEM (M4) and ESEM (M2) models, all three information criteria (AIC, BIC, and aBIC) consistently favored the more complex M4 model, with lower values. Notably, AIC and aBIC favored the more complex models across all comparisons, while BIC tended to support the more parsimonious alternatives. Although M4 demonstrated marginally superior global fit indices, the differences (see Table 2) remained within conservative thresholds, indicating statistical equivalence. Given this, model parsimony was prioritized. Importantly, the ESEM model exhibited very low cross-loadings (all < 0.10), suggesting limited support for a dominant general factor, a core assumption of bifactor models. Therefore, its added complexity lacked sufficient theoretical and empirical justification. These findings support the selection of the ESEM model (M2) as the more appropriate and parsimonious solution (see Figure 2). A critical aspect of validating the short DSI scale involved testing for measurement invariance across gender to determine whether the instrument functions equivalently for male and female respondents. Without such invariance, comparisons of digital skills across gender could be biased or invalid. Measurement invariance is essential for ensuring the theoretical validity of a construct across groups (van de Vijver & Tanzer, 2004). To assess this, a series of increasingly restrictive invariance tests were conducted following established procedures (Meredith, 1993; Millsap, 2011). These tests were applied to the retained ESEM model and included six levels: Configural invariance (same factor structure across groups) Metric invariance (equal factor loadings) Scalar invariance (equal intercepts) Strict invariance (equal residuals) Latent variance–covariance invariance Latent mean invariance (enabling valid group comparisons of latent means). As shown in Table 6, all models demonstrated acceptable to good fit across gender groups, supporting the conclusion that the DSI scale operates equivalently for male and female students. Table 6. Measurement invariance sequential models’ fit indices and differences Model χ² (df) CFI TLI RMSEA 90%CI SRMR ΔCFI ΔTLI ΔRMSEA ΔSRMR Configural invariance 814.116 (372) .937 .906 .063 (.057 .069) .027 Weak (λ) invariance 824.431 (452) .947 .935 .053 (.047 .058) .033 .010 .029 -.010 .006 Strong (λ, ν) invariance 873.182 (472) .942 .933 .053 (.048 .059) .034 -.005 -.002 .000 .001 Strict (λ, ν, δ) invariance 923.983 (496) .939 .932 .054 (.048 .059) .045 -.003 -.001 .001 .011 Latent var-cov (λ, ν, δ, ξ/φ) invariance 935.828 (506) .938 .933 .053 (.048 .059) .051 -.001 .001 -.001 .006 Latent mean (λ, ν, δ, ξ/φ, η) invariance 942.929 (510) .938 .933 .053 (.048 .059) .053 .000 .000 .000 .002 To evaluate each level of measurement invariance, the Satorra–Bentler scaled chi-square difference test (Satorra & Bentler, 2010) was used to compare nested models. This test is considered more robust than the traditional chi-square difference test, particularly under conditions of non-normality. As shown in Table 3 (Appendix), the results supported metric (weak) invariance, but not scalar (strong) or strict invariance. According to Byrne et al. (1989), establishing metric invariance is sufficient for comparing structural relationships across groups, such as factor loadings and covariances. Although the Satorra–Bentler (S-B) scaled chi-square difference test adjusts for non-normality, it is not without limitations. As noted by Dimitrov (2010) and Maïano et al. (2015), the test remains sensitive to small sample sizes, which can compromise the reliability of the scaling correction and increase the risk of Type I errors. Additionally, the computation of the scaling correction factor (SCF) varies across statistical software (e.g., LISREL, EQS, Mplus), and incorrect application may result in inadmissible test statistics, such as negative values or incorrect degrees of freedom (Bryant & Satorra, 2012). For these reasons, researchers are encouraged to supplement the S-B test with alternative goodness-of-fit indices (GFIs), which are generally less sensitive to sample size, model complexity, and baseline model fit. According to Chen (2007), for sample sizes greater than 300, the following thresholds indicate acceptable levels of invariance: Metric (weak) invariance: ΔCFI ≤ .010, ΔRMSEA ≤ –.015, ΔSRMR ≤ .030; Scalar and strict invariance: ΔCFI ≤ .010, ΔRMSEA ≤ –.015, ΔSRMR ≤ .010. Cheung and Rensvold (2002) similarly recommend ΔCFI ≤ .010 as a criterion for supporting invariance. In the metric invariance model (see Table 6), item loadings were constrained to be equal across gender. The model showed good fit: χ² (372) = 824.431, CFI = .947, TLI = .935, RMSEA = .063 [90% CI: .057 .069], SRMR = .027. The S-B scaled chi-square difference test comparing the metric and configural models was non-significant: Δχ² = 56.15, Δdf = 82, p = .98, supporting metric invariance. Additionally, changes in fit indices (ΔCFI = .010; ΔTLI = .029; ΔRMSEA = –.010; ΔSRMR = .006) were within acceptable thresholds, allowing valid comparisons of factor loadings across gender. For scalar invariance, both item loadings and intercepts were constrained. The model fit remained good: χ² (472) = 873.182, CFI = .942, TLI = .933, RMSEA = .053 [90% CI: .048 .059], SRMR = .034. However, the S-B test indicated a significant difference from the metric model: Δχ² = 55.99, Δdf = 20, p < .001, suggesting a lack of scalar invariance. In contrast, changes in fit indices (ΔCFI = –.005; ΔTLI = –.002; ΔRMSEA = .000; ΔSRMR = .001) supported scalar invariance, indicating that latent means could still be compared. Strict invariance was tested by additionally constraining residual variances. The model fit remained good: χ² (496) = 923.983, CFI = .939, TLI = .932, RMSEA = .054 [90% CI: .048, .059], SRMR = .045. The S-B test showed a significant difference from the scalar model: Δχ² = 47.51, Δdf = 24, p < .005, suggesting a lack of strict invariance. However, changes in fit indices (ΔCFI = –.003; ΔTLI = –.001; ΔRMSEA = .001; ΔSRMR = .011) remained within acceptable limits, supporting strict invariance. Finally, models testing latent variance–covariance and latent mean invariance also demonstrated good fit. The S-B chi-square difference tests were non-significant ( p > .05), and changes in fit indices (ΔCFI = –.001/.000; ΔTLI = .001/.000; ΔRMSEA = –.001/.000; ΔSRMR = .006/.002) did not exceed recommended thresholds (ΔCFI and ΔTLI < .010; ΔRMSEA < .015). These results support the conclusion that the ESEM factor structure is invariant across gender, enabling meaningful comparisons of structural relationships and latent means between male and female students. In the next phase of analysis, a hierarchical Exploratory Structural Equation Model (H-ESEM) was estimated to examine the presence of a higher-order digital skills factor underlying the four first-order dimensions. Unlike B-ESEM, which separates general and specific variance to distinguish broad from narrow constructs, H-ESEM allows observed variables to contribute indirectly to a higher-order factor through their respective first-order dimensions. This approach avoids imposing a direct general-factor influence on each item, offering greater flexibility in modeling the hierarchical structure of the construct. In H-ESEM, first-order factors remain connected to their observed indicators while also serving as indicators of a second-order latent factor. This structure captures both the distinctiveness of the four digital skills dimensions and their shared variance, without enforcing rigid hierarchical constraints. The model thus reflects the dynamic interplay between specific and general digital competencies. To estimate the H-ESEM model, non-standardized factor loadings from the original ESEM solution were used as starting values. One indicator per factor was constrained to retain its original loading, and all factor variances were fixed to one for model identification. The four first-order factors, Technical and Operational Skills, Information Navigation and Processing, Communication and Interaction, and Content Creation and Production, were specified as indicators of a higher-order digital skills factor (H-FACTOR). Table 7 presents the standardized path coefficients from each first-order factor to the higher-order construct. Table 7. Standardized factor loadings from H-ESEM model with WLSMV estimator H-FACTOR BY Estimate S.E. Est./S.E. Two-Tailed P-Value TO .882 .016 56.383 .000 INP .909 .016 56.215 .000 CI .818 .015 53.869 .000 CCP .831 .018 47.209 .000 All standardized loadings from the first-order factors to the higher-order digital skills factor (H-FACTOR) were high (all > .80) and statistically significant ( p < .001), indicating that the general digital skills construct strongly explains variance in each of the four sub-dimensions. This supports the presence of a hierarchical structure, where a broad digital skills factor underlies the more specific dimensions. Among the sub-dimensions, Information Navigation and Processing (INP) had the highest loading (.909), suggesting it is most strongly associated with the general factor. Communication and Interaction (CI) had the lowest loading (.818), though still well within the range considered strong. These results confirm that the H-FACTOR is a robust overarching construct, with each sub-dimension contributing meaningfully to it. The proportion of variance explained by the H-FACTOR was substantial across all first-order factors, ranging from 66.8% for CI to 82.6% for INP. Additionally, all four latent dimensions were highly inter-correlated and strongly associated with the higher-order factor (see Table 4, Appendix), further supporting the hierarchical model. These findings suggest that while the dimensions are conceptually distinct, they are closely related and collectively represent a unified construct of digital skills. To address the complexity of the measurement model while examining structural relationships with external predictors, we adopted the ESEM-within-CFA (EWC) approach. This method mitigates the convergence issues often encountered in traditional ESEM models when applied within complex structural frameworks. EWC enables the estimation of cross-loadings while preserving the structural clarity and interpretability characteristic of CFA. It is particularly well-suited for models incorporating predictors or covariates (Morin & Asparouhov, 2018; Marsh & Alamer, 2024). In our analysis, the EWC model was estimated with three external predictors, procrastination , self-esteem , and technology affinity , targeting their effects on the four latent dimensions. This approach enhances model stability and interpretability by allowing predictors to exert more controlled influence over the factor structure. For comparative purposes, Table 5 (Appendix) presents traditional fit indices for both the ESEM and EWC models with predictors, estimated using two robust estimators. The ESEM model showed excellent fit with the WLSMV estimator (CFI, TLI, SRMR), though RMSEA was marginally acceptable. With the MLR estimator, RMSEA indicated excellent fit, while other indices remained within good to excellent range. The EWC model with predictors showed acceptable to good fit using MLR, and good to excellent fit with WLSMV, including a notable RMSEA improvement. Predictor Effects on EWC Dimensions Self-esteem (SE) emerged as the only significant positive predictor (p < .001) of the TO and CI dimensions across both estimators and under conditions of fixed and free factor variances (see Table 8). Procrastination (PR) and Affinity for Technology (AF) were not significant predictors of TO. However, procrastination had a significant negative effect on INP (p < .05), while self-esteem had a significant positive effect on the same dimension (p < .05). Affinity for Technology was a marginally significant predictor of INP, with a stronger effect observed under the WLSMV estimator (Figure 1, Appendix). Table 8. The impact of three predictors on four latent factors of EWC with fixed and free factor variances for two different robust estimators MLR WLSMV Fixed variance Free variance Fixed variance Free variance Factors Estimates p-value Estimates p-value Estimates p-value Estimates p-value TO ON PR -.024 .736 -.022 .758 -.023 .752 -.021 .769 SE .610 .000 .613 .000 .626 .000 .623 .000 AF -.114 .456 -.121 .433 -.135 .374 -.134 .377 INP ON PR -.159 .031 -.160 .031 -.179 .005 -.182 .005 SE .343 .012 .337 .014 .344 .005 .330 .007 AF .301 .057 .306 .053 .302 .044 .315 .036 CI ON PR -.097 .123 -.096 .129 -.090 .194 -.088 .201 SE .659 .000 .662 .000 .723 .000 .725 .000 AF -.103 .478 -.108 .454 -.176 .255 -.181 .242 CCP ON PR -.145 .036 -.145 .037 -.155 .020 -.155 .020 SE .381 .002 .378 .002 .397 .001 .391 .001 AF .241 .100 .243 .097 .233 .120 .239 .111 Procrastination (PR) was a significant negative predictor of CCP (p < .05), while self-esteem (SE) had a significant positive effect (p .05). The four EWC dimensions exhibited distinct prediction patterns, supporting their conceptual distinctiveness. This is further reinforced by moderate to high inter-factor correlations (Table 6, Appendix), none exceeding .85, indicating acceptable discriminant validity and suggesting the factors capture unique constructs. The results indicate that the four EWC dimensions are related yet distinct, aligning with expectations for a multidimensional construct such as digital skills. 4. Discussion Factor Analysis (CFA) and bifactor models , ESEM demonstrated superior model fit, greater parsimony, and stronger theoretical alignment, all while preserving the complexity inherent in multidimensional constructs. These findings corroborate prior research by Morin et al. (2020) and Gegenfurtner (2022) , who emphasized the advantages of ESEM in modeling complex psychological constructs. Furthermore, this study extends the work of Vuorikari et al. (2022) by offering a rigorous empirical implementation of the recommendations proposed in Vuorikari et al. (2025) , which explicitly advocate for the use of ESEM to more accurately capture the overlapping dimensions of digital competence. As such, this research contributes to the growing body of evidence supporting the adoption of flexible, multidimensional modeling approaches in digital skills assessment. The ESEM model demonstrated strong psychometric properties, including high internal consistency, satisfactory convergent and discriminant validity, and minimal cross-loadings, indicating a well-defined factorial structure. The hierarchical ESEM (H-ESEM) further confirmed the presence of a higher-order digital skills factor, with all four sub-dimensions, Technical and Operational Skills (TO), Information Navigation and Processing (INP), Communication and Interaction (CI), and Content Creation and Production (CCP), loading significantly onto the general factor. Although it was not the primary objective of this study, we recognize the potential value of further exploring the multidimensional and hierarchical structure of digital skills through the application of a bifactor-ESEM model (Howard et al., 2016; Alamer, 2022). Such an approach would offer complementary insights to those obtained from the higher-order ESEM (H-ESEM) model, contributing to a more nuanced and accurate understanding of the construct. In particular, the use of ancillary bifactor indices (Rodriguez et al., 2016), such as the Explained Common Variance (ECV), Omega Hierarchical (ωH), and Percentage of Uncontaminated Correlations (PUC), could provide more definitive evidence regarding the degree of unidimensionality versus multidimensionality inherent in the digital skills construct. Such insights would not only enhance the theoretical understanding of digital competence but also inform the appropriate use of total versus subscale scores in applied settings within broader Structural Equation Modeling (SEM) frameworks, especially in studies aiming to examine predictors or outcomes of digital competence. To validate the robustness of the measurement and structural models, parameter estimates were compared using both MLR and WLSMV estimators. Results across estimators were largely consistent, with only minor variations in model fit indices and path coefficients. Model comparisons (e.g., M2 vs. M1) showed negligible changes in CFI, TLI, RMSEA, and SRMR, suggesting that model refinements did not yield substantial improvements in fit. This consistency supports the stability of the model regardless of estimation method, revealing that parameter estimates or fit indices are not sensitive to estimator choice. ESEM models demonstrated improved discriminant validity through reduced factor correlations, while measurement invariance testing across gender revealed metric invariance, allowing for valid comparisons of factor loadings between male and female students. Although scalar and strict invariance were not fully supported by chi-square difference tests, alternative fit indices suggested acceptable model stability, reinforcing the robustness of the instrument across demographic groups. This study offers a complementary contribution to existing approaches in digital skills assessment, particularly the Youth Digital Skills Indicator (yDSI) developed by Helsper et al. (2020). The yDSI combined performance-based tasks with composite scoring and was rigorously tested for measurement invariance across diverse demographic and cultural groups, demonstrating strong psychometric robustness. Its validated items have proven effective for large-scale, cross-national research. However, its reliance on composite scoring constrains its ability to fully capture the multidimensional and hierarchical structure of digital competence. In contrast, the present study employs latent factor scores derived from an Exploratory Structural Equation Modeling (ESEM) framework. This approach enables a more nuanced representation of digital skills by modeling both general and domain-specific competencies, while accounting for the interdependence among skill dimensions. The ESEM model demonstrated weak, strong, and strict measurement invariance across gender, thereby supporting valid comparisons of both latent means and observed scores. This level of psychometric rigor strengthens the empirical foundation of the digital skills construct and directly responds to Helsper et al.’s call for more precise and equitable assessment tools. By transitioning from manifest to latent modeling and establishing full measurement invariance, the current study enriches the yDSI framework and provides a robust platform for future cross-group and longitudinal analyses in digital competence research. The ESEM-within-CFA (EWC) model with external predictors provided further evidence of construct validity. Self-efficacy emerged as a consistent and significant positive predictor across all four digital skill dimensions, while academic procrastination negatively predicted INP and CCP. Affinity for technology showed marginal effects, suggesting that attitudinal variables may influence digital engagement differently. Although the sample size in this study was determined through an a priori power analysis using Soper’s (2025) calculator—based on anticipated effect sizes and a desired statistical power—this approach, while methodologically sound, may not fully account for potential sources of model bias inherent in complex latent variable models. As such, we acknowledge the value of more advanced techniques for evaluating model adequacy. Specifically, although not the primary focus of this study, we contend that the use of Monte Carlo simulations (Muthén & Muthén, 2002) represents a more holistic and rigorous strategy. Monte Carlo methods enable researchers to assess estimation accuracy, model fit, and potential sources of bias under varied conditions by simulating data based on realistic assumptions. This approach facilitates the evaluation of factor loadings, standard errors, factor correlations, residual variances, confidence interval coverage rates, and statistical power. By identifying conditions under which model estimates may become biased or unstable, Monte Carlo simulations can enhance the validity, reliability, and generalizability of findings in structural equation modeling. 4.1 Theoretical implications This study provides robust evidence for a psychometrically sound instrument to assess students’ digital skills, reinforcing the conceptualization of digital skills as a multidimensional construct. This structure aligns with contemporary frameworks (e.g., DigComp, ISTE 3 ) and advances them by providing empirical validation through robust psychometric modeling. Our findings demonstrate that Exploratory Structural Equation Modeling (ESEM) outperforms traditional CFA model in capturing the complexity of digital skills. ESEM’s allowance for cross-loadings enhances construct validity and discriminant precision, making it a preferred approach for future psychometric research in education and technology. The significant associations between digital skills and self-esteem, procrastination, and technology affinity highlight the interdisciplinary nature of digital competence. This positions digital skills not just as technical abilities but as psycho-social constructs, influenced by motivation, self-concept, and behavioral tendencies. These findings align with contemporary theories of digital literacy, which emphasize both technical proficiency and socio-cognitive engagement (Belshaw, 2012; Feola, 2016). 4.2 Practical Implications The findings of this study offer actionable insights for enhancing digital skills education through curriculum design, assessment strategies, and institutional policy. The validated hierarchical model of digital competence provides a robust foundation for evidence-based educational planning. Specifically, the results support the following implications: a. Curriculum Design The H-ESEM results provide strong empirical support for a hierarchically structured digital competence framework. These findings confirm and extend previous research emphasizing the foundational role of technical and information-processing skills in digital competence development (e.g., UNESCO, 2011; Vuorikari et al., 2022; Mattar et al., 2022). Curriculum designers are encouraged to leverage these insights to develop sequenced, evidence-based learning pathways that prioritize foundational skills while scaffolding the development of communication and content creation capabilities, ensuring a coherent progression from foundational to higher-order competencies. b. Assessment and Intervention The validated instrument enables the diagnosis of specific skill gaps, monitoring of student progress over time, and early identification of learners at risk of digital exclusion. This supports the development of targeted interventions and personalized learning pathways. c. Institutional Policy and Practice The use of psychometrically sound assessment data can inform strategic decisions related to infrastructure investment, teacher training, and curriculum reform. Furthermore, the demonstrated measurement invariance across gender, promotes equity and inclusion, ensuring that digital competence assessments are fair and accessible to all learners. Educators can also be empowered through professional development initiatives that help them interpret digital skills data and apply insights to instructional planning. For educators, the instrument offers a practical and scalable tool for assessing digital literacy at the classroom, school, or system level. Its alignment with established frameworks (e.g., DigComp, UNESCO ICT-CFT 4 ) makes it suitable for integration into national assessments and policy-driven initiatives aimed at enhancing digital competence. 5. Conclusions This study provides robust empirical evidence for the validity and reliability of a short-form psychometric instrument designed to assess students’ digital skills. The ESEM model, supported by hierarchical and predictive analyses, offers a statistically sound and theoretically grounded framework for measuring digital competence in higher education contexts. The validated instrument is suitable for large-scale assessments, policy evaluations, and integration into broader structural models. It can inform curriculum development, teacher training, and digital inclusion strategies, particularly in under-resourced educational systems like Albania’s, where digital transformation is still in its early stages. 5.1. Limitations Despite the strengths of the study, several limitations should be acknowledged: (1) Sample Specificity : The data were collected from a single university in Albania, which may limit the generalizability of the findings to other educational contexts or countries. (2) Self-Report Bias : The use of self-reported measures, although carefully designed to minimize bias, may still be subject to social desirability and overestimation of skills. (3) Cross-Sectional Design : The study’s cross-sectional nature precludes causal inferences and limits the ability to assess changes in digital skills over time. (4) Limited External Predictors : While three predictors were included, other relevant variables (e.g., socioeconomic status, prior ICT training) were not examined. 5.2. Future Research Directions To build on the current findings, future research should consider the following directions: (1) Cross-Cultural Validation : Replicating the study in diverse educational and cultural settings to test the instrument’s cross-national applicability and measurement invariance. (2) Longitudinal Studies : Implementing longitudinal designs to track the development of digital skills over time and assess the impact of educational interventions. (3) Performance-Based Measures : Combining self-report instruments with performance-based assessments to triangulate findings and enhance validity. (4) Expanded Predictive Models : Including a broader range of exogenous variables (e.g., digital access, motivation, institutional support) to better understand the determinants of digital competence. (5) The assessment of critical digital skills , measured through knowledge-based items in addition to functional digital skills, can be significantly enhanced through the application of both the full B (ESEM) and Set B (ESEM) models. This dual-model approach allows for a more comprehensive evaluation of digital competence by capturing both foundational and higher-order cognitive dimensions. (6) Policy Impact Evaluation : Using the instrument in policy evaluation frameworks to assess the effectiveness of national digital education strategies. Declarations Participant Consent Statement: All participants provided informed consent prior to their involvement in the study. The research protocol, including the consent procedure, was reviewed and approved by the relevant institutional ethics committee. Participation was voluntary, and data were collected anonymously to ensure confidentiality. References Alamer, A. (2022). Exploratory structural equation modeling (ESEM) and bifactor ESEM for construct validation purposes: Guidelines and applied example. Research Methods in Applied Linguistics , 1, 100005. https://doi.org/10.1016/j.rmal.2022.100005 Ala-Mutka, K. (2011). Mapping Digital Competence: Towards a Conceptual Understanding . Luxembourg: Publications Office of the European Union. JRC Technical Report. Asparouhov T, Muthén B. Exploratory structural equation modeling. Struct Equat Model. (2009) 16:397–438. doi: 10.1080/10705510903008204 Asparouhov, T., & Muthén, B. (2018). SRMR in Mplus. Unpublished manuscript. https://www.statmodel.com/download/SRMR2.pdf Bandalos, D. L. (2014). Relative performance of categorical diagonally weighted least squares and robust maximum likelihood estimation. Structural Equation Modeling: A Multidisciplinary Journal, 21 (1), 102–116. https://doi.org/10.1080/10705511.2014.859510 Bandalos, D. L. (2018). Measurement theory and applications for the social sciences . The Guilford Press. Barbazan, D., Ben, Khaddouja., & Montes, C. (2021). La competencia digital docente en educación superior: estado del arte en España y Latinoamérica [Digital Competence for Teachers in Higher Education: State of the Art in Spain and Latin America]. Etic@net. Revista Científica Electrónica de Educación y Comunicación En La Sociedad Del Conocimiento, 21(2), 267-282. https://doi.org/10.30827/eticanet.v21i2.20837 Beauducel A, Herzberg PY. (2006). On the Performance of Maximum Likelihood Versus Means and Variance Adjusted Weighted Least Squares Estimation in CFA. Structural Equation Modeling 13(2): 186-203, DOI: 10.1207/s15328007sem1302_2. Beierlein, C., Kemper, C. J., Kovaleva, A., & Rammstedt, B. (2013). Short Scale for Measuring General Self-efficacy Beliefs (ASKU). Methods, Data, Analyses, 7 (2), 251–278. https://doi.org/10.12758/mda.2013.014 Belshaw, D. A. J. (2012). What is digital literacy? A pragmatic investigation (Doctoral dissertation, Durham University). Durham E-Theses. https://etheses.dur.ac.uk/3446/ Bryant, F. B., & Satorra, A. (2012). Principles and practice of scaled difference chi-square testing . Structural Equation Modeling: A Multidisciplinary Journal , 19 (3), 372–398 . https://doi.org/10.1080/10705511.2012.687671 Byrne, B. M., Shavelson, R. J., & Muthén, B. (1989). Testing for equivalence of factor covariance and mean structures: The issue of partial measurement invariance. Psychological Bulletin, 105, 456-466. Cabezas-González, M., Casillas-Martín, S., & García-Valcárcel Muñoz-Repiso, A. (2023). Theoretical models explaining the level of digital competence in students. Computers, 12 (5), 100. https://doi.org/10.3390/computers12050100 Carretero, S., Vuorikari, R., & Punie, Y. (2017). DigComp 2.1: The Digital Competence Framework for Citizens with eight proficiency levels and examples of use. Publications Office of the European Union. EUR 28558 EN. https://doi.org/10.2760/38842 Chen, F. F. (2007). Sensitivity of goodness of fit indexes to lack of measurement. Structural Equation Modeling, 14, 464–504. http://dx.doi.org/10.1080/10705510701301834 Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of fit indexes for testing measurement invariance. Structural Equation Modeling, 9, 233–255. http://dx.doi.org/10.1207/S15328007SEM0902 5 Cortesi, S., Gasser, U., & Hasse, A. (2020). Youth and digital citizenship+ (plus): Understanding skills for a digital world . Berkman Klein Center for Internet & Society, Harvard University. https://cyber.harvard.edu/publication/2020/youth-and-digital-citizenship-plus Darling-Hammond, L., & Adamson, F. (2010). Beyond basic skills: The role of performance assessment in achieving 21st century standards of learning . Stanford Center for Opportunity Policy in Education (SCOPE). https://edpolicyinca.org/publications/beyond-basic-skills[1](https://cyber.harvard.edu/people/scortesi) Dierendonck C (2023) Measuring the classroom level of the Dynamic Model of Educational Effectiveness through teacher self-report: development and validation of a new instrument. Front. Educ. 8:1281431. doi: 10.3389/feduc.2023.1281431 Dimitrov, D. M. (2010). Testing for Factorial Invariance in the Context of Construct Validation. Measurement and Evaluation in Counseling and Development , 43 (2), 121-149. https://doi.org/10.1177/0748175610373459 (Original work published 2010) Erwin, K., & Mohammed, S. (2022). Digital Literacy Skills Instruction and Increased Skills Proficiency. International Journal of Technology in Education and Science, 6(2), 323-332. https://doi.org/10.46328/ijtes.364. European Commission. (2020). Digital Competence Framework for Citizens (DigComp 2.1): The Digital Competence Framework . Publications Office of the European Union. https://education.ec.europa.eu/focus-topics/digital-education/action-plan/Updating-the-European-Digital-Competence-Framework European Commission. (2021). Albania 2021 Report. Accompanying the document Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions 2021 Communication on EU Enlargement Policy. Strasbourg 19.10.2021 Retrieved from https://ec.europa.eu/neighbourhood-enlargement/albania-report-2021_en Falma, F. O., & Putra, A. H. (2025). Correlation between academic self-efficacy and digital competence: An empirical study on first-year students. Journal of Counseling, Education and Society, 6 (1), 1–9. https://jurnal.iicet.org/index.php/jces/article/download/5403/pdf Fan, C., & Wang, J. (2022). Development and validation of a questionnaire to measure digital skills of Chinese undergraduates. Sustainability, 14(6), 3539. https://doi.org/10.3390/su14063539 Fayda-Kınık, F. Ş. (2023). The impact of digital competences on academic procrastination in higher education: A structural equation modeling approach . Pegem Journal of Education and Instruction, 13(3), 25–35. https://files.eric.ed.gov/fulltext/EJ1387897.pdf Feijao, C., Flanagan, I., van Stolk, C. and Gunashekar, S. 2021. The global digital skills gap. Feola, E. I. (2016). Digital literacy and new technological perspectives. Universal Journal of Educational Research, 4 (9), 2174–2180. https://doi.org/10.13189/ujer.2016.040929 Ferrari, A. (2013). DIGCOMP: A framework for developing and understanding digital competence in Europe. Publications Office. Finney, S.J., & DiStefano, C. (2013). Non-normal and categorical data in structural equation modeling. In G.R. Hancock & R.O. Mueller (Eds.), Structural equation modeling: A second course, 2 -nd Edition (pp. 439–492). Charlotte, NC: Information Age Publishing. Flora, D. B., & Curran, P. J. (2004). An empirical evaluation of alternative methods of estimation for confirmatory factor analysis with ordinal data. Psychological Methods, 9 (4), 466–491. https://doi.org/10.1037/1082-989X.9.4.466 [2] (https://psycnet.apa.org/record/2014-05296-010) Fornell, C., and Larcker, D. F. (1981). Evaluating structural equation models with unobservable variables and measurement error. J. Mark. Res. 18, 39–50. doi: 10.2307/3151312 Franke, T., Attig, C., and Wessel, D. (2019). A personal resource for technology interaction: Development and validation of the Affinity for Technology Interaction (ATI) scale. Int. J. Hum. Comput. Interact. 35, 456–467. doi: 10.1080/10447318.2018.1456150 Gallardo-Echenique, E. E., De Oliveira, J. M., Marqués-Molias, L., Esteve-Mon, F., Wang, Y., & Baker, R. (2015). Digital competence in the knowledge society. MERLOT Journal of Online Learning and Teaching, 11(1), 1–16. Gegenfurtner, A. (2022). Bifactor exploratory structural equation modeling: A meta-analytic review of model fit. Frontiers in Psychology, 13 , 1037111. https://doi.org/10.3389/fpsyg.2022.1037111 Haddon, L., Cino, D., Doyle, M. A., Livingstone, S., Mascheroni, G., & Stoilova, M. (2020). Children’s and young people’s digital skills: A systematic evidence review. Leuven, KU Leuven: ySKILLS. Available at: https://zenodo.org/record/4274654#.X-pMceSWysc Hair, J. F., Anderson, R. E., Babin, B. J., and Black, W. C. (2010). Multivariate data analysis: A global perspective, vol. 7. Upper Saddle River, NJ: Pearson Helsper, E. J., Schneider, L. S., van Deursen, A. J. A. M., & van Laar, E. (2021). The youth Digital Skills Indicator: Report on the conceptualisation and development of the ySKILLS digital skills measure . University of Twente. https://research.utwente.nl/en/publications/the-youth-digital-skills-indicator-report-on-the-conceptualisation Helsper, E. J., van Deursen, A. J. A. M., & Eynon, R. (2020). Digital skills, digital inequalities and the digital divide: Theories and evidence . In J. A. G. M. van Dijk & A. J. A. M. van Deursen (Eds.), Digital Skills: Unlocking the Information Society (pp. 21–42). Springer. https://link.springer.com/chapter/10.1007/978-3-031-46929-9_3 Helsper, E.J., Schneider, L.S., van Deursen, A.J.A.M., & van Laar, E. (2020). The youth Digital Skills Indicator: Report on the conceptualisation and development of the ySKILLS digital skills measure. KU Leuven, Leuven: ySKILLS. Howard, J. L., Gagné, M., Morin, A. J. S., Wang, Z. N., & Forest, J. (2016). Using bifactor exploratory structural equation modeling to test for a continuum structure of motivation. Journal of Management , 1–59. https://www.statmodel.com/download/Bifactor-ESEM.pdf Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria vs new alternatives. Structural Equation Modeling, 6, 1–55. http://dx.doi.org/10.1080/10705519909540118 Ilomäki L, Paavola S, Lakkala M, Kantosalo A. Digital competence—An emergent boundary concept for policy and educational research. Education and Information Technologies. 2014; 21(3):655–679. International Telecommunication Union (ITU) (2018). Measuring the Information Society report, Volume 1. Geneva, Switzerland: ITU Publications. Available at: www.itu.int/en/ITU-D/Statistics/Documents/publications/misr2018/MISR-2018-Vol-1-E.pdf. Javier-Aliaga, D., Silva Neyra, O. R., Calizaya-Milla, Y. E., & Saintila, J. (2024). Academic self-efficacy and digital competence in a sample of university students. Contemporary Educational Technology, 16 (4), ep540. https://doi.org/10.30935/cedtech/15601 Jokisch, C., Schlimbach, R., Strzebkowski, M., Hobert, S., Kurpejovic, H., & Schumann, M. (2025). Empirical insights on the impact of technology affinity on programming skill development through an adaptive game-based learning system . Proceedings of the 58th Hawaii International Conference on System Sciences. https://hdl.handle.net/10125/109435 Joreskog KG. A general approach to confirmatory maximum likelihood factor analysis. Psychometrika. (1969) 34:183–202. doi: 10.1007/BF02289343 Kaplan, D. (2009). Structural Equation Modeling: Foundations and Extensions (2nd Ed.). Newbury Park, CA: SAGE Publications. https://doi.org/10.4135/9781452226576 Kayyali, M. (2024). Digital Literacy in Higher Education: Preparing Students for the Workforce of the Future. IJISC., 11(01): 53-73. Kryukova, N. I., Chistyakov, A. A., Shulga, T. I., Omarova, L. B., Tkachenko, T. V., Malakhovsky, A. K., et al. (2022). Adaptation of higher education students’ digital skills survey to Russian universities. Eurasia J. Mathem. Sci. Technol. Educ. 18:em2183. doi: 10.29333/ejmste/12558 Laanpere, M. (2019). Recommendations on assessment tools for monitoring digital literacy within UNESCO’s digital literacy global framework. Montreal, Quebec, UNESCO Institute for Statistics, 2019. https://unesdoc.unesco.org/ark:/48223/pf0000366740 López, J. C. (2023). Competencias digitales en la educación superior. Horizontes Rev. Invest. Ciencias Educ. Superiorn 7, 1548–1563. doi: 10.33996/revistahorizontes.v7i29.612 Ma, H., Ismail, L. Bibliometric analysis and systematic review of digital competence in education. Humanit Soc Sci Commun 12, 185 (2025). https://doi.org/10.1057/s41599-025-04401-1 Maïano, C., Morin, A. J. S., & Mascret, N. (2015). Psychometric properties of the short form of the physical self-description questionnaire in a French adolescent sample. Body Image, 12, 89–97. https://doi.org/10.1016/j.bodyim.2014.10.005 Marsh HW, Guo J, Dicke T, Parker PD, Craven RG. Confirmatory factor analysis (CFA), exploratory structural equation modeling (ESEM), and set-ESEM: optimal balance between goodness of fit and parsimony. Multivariate Behav Res. (2020) 55:102–19. doi: 10.1080/00273171.2019.1602503 Marsh HW, Morin AJ, Parker PD, Kaur G. Exploratory structural equation modeling: An integration of the best features of exploratory and confirmatory factor analysis. Annu Rev Clin Psychol. (2014) 10:85–110. doi: 10.1146/annurev-clinpsy-032813-153700 Marsh HW, Muthén B, Asparouhov T, Lüdtke O, Robitzsch A, Morin AJ, et al. Exploratory structural equation modeling, integrating CFA and EFA: application to students’ evaluations of university teaching. Struct Equ Model. (2009) 16:439–76. doi: 10.1080/10705510903008220 Marsh, H. W. (2007). Application of confirmatory factor analysis and structural equation modeling in sport/exercise psychology. In G. Tenenbaum & R. C. Eklund (Eds.), Handbook of on sport psychology (3rd ed., pp. 774 - 798). New York: Wiley. Marsh, H. W., & Alamer, A. (2024). When and how to use set‐exploratory structural equation modelling to test structural models: A tutorial using the R package lavaan . British Journal of Mathematical and Statistical Psychology . Advance online publication. https://doi.org/10.1111/bmsp.12336 Marsh, H. W., Hau, K.-T., & Grayson, D. (2005). Goodness of fit in structural equation models. In A. Maydeu-Olivares & J. J. McArdle (Eds.), Contemporary psychometrics: A festschrift for Roderick P. McDonald (pp. 275–340). Lawrence Erlbaum Associates Publishers Marsh, H. W., Wen, Z., & Hau, K.-T. (2004). Structural equation models of latent interactions: Evaluation of alternative estimation strategies and indicator construction. Psychological Methods, 9 (3), 275–300. https://doi.org/10.1037/1082-989X.9.3.275 Marsh, H.W., Lüdtke, O., Muthén, B., Asparouhov, T., Morin, A. J. S., Trautwein, U., & Nagengast, B. (2010). A new look at the big-five factor structure through exploratory structural equation modeling. Psychological Assessment, 22, 471-491. Mattar, J., Ramos, D. K., & Lucas, M. R. (2022). DigComp-based digital competence assessment tools: Literature review and instrument analysis. Education and Information Technologies, 27 , 10843–10867. https://doi.org/10.1007/s10639-022-11034-3 Meredith, W. (1993). Measurement invariance, factor analysis and factor invariance. Psychometrika, 58, 525-544. Miço, H., & Zaçellari, M. (2020). “Online learning as a means toward achieving an adaptable right to education in Albania”. Knowledge-International Journal, 2020, 42(2) 293–298. ISSN online 2545-4439 https://ikm.mk/ojs/index.php/KIJ/article/view/4672/4459 Millsap, R. E., & Meredith, W. (2004, May). Factorial invariance: Historical trends and new developments. Paper presented at the Factor Analysis at 100 Conference, Chapel Hill, NC. Morin AJS, Asparouhov T. Estimation of a Hierarchical Exploratory Structural Equation Model (ESEM) Using ESEM-Within-CFA. Montreal, QC: Substantive Methodological Synergy Research Laboratory (2018). Morin AJS, Myers ND, Lee S. Modern factor analytic techniques: bifactor models, exploratory Exploratory Structural Equation Modeling (ESEM),19 structural equation modeling and bifactor-ESEM. In: Tenenbaum G, Eklund RC, editors. Handbook of Sport Psychology, 4th Edition, Vol. 2. New York, NY: Wiley Publishers (2020). p. 1044–73. doi: 10.1002/9781119568124.ch51 Morin, A. J. S., Arens, A. K., & Marsh, H. W. (2016a). A bifactor exploratory structural equation modeling framework for the identification of distinct sources of construct-relevant psychometric multidimensionality. Structural Equation Modeling: A Multidisciplinary Journal, 23 (1), 116–139. https://doi.org/10.1080/10705511.2014.961800 Morin, A. J. S., Marsh, H. W., & Nagengast, B. (2013). Exploratory structural equation modeling. In Hancock, G. R., & Mueller, R. O. (Eds.). Structural equation modeling: A second course (2nd ed., pp. 395-436). Charlotte, NC: Information Age Publishing, Inc. Morin, A. J. S., Myers, N. D., & Lee, S. (2020). Modern factor analytic techniques: Bifactor models, exploratory structural equation modeling (ESEM), and bifactor‐ESEM. In G. Tenenbaum, R. C. Eklund, & N. Boiangin (Eds.), Handbook of sport psychology: Exercise, methodologies, & special topics (4th ed., pp. 1044–1073). John Wiley & Sons. https://doi.org/10.1002/9781119568124.ch51 Morin, A.J.S., Arens, A.K., Tran, A., & Caci, H. (2016b). Exploring sources of construct-relevant multidimensionality in psychiatric measurement: A tutorial and illustration using the Composite Scale of Morningness. International Journal of Methods in Psychiatric Research, 25, 277–288. http://doi.org/10.1002/mpr.1485 Murray, A.L., & Johnson, W. (2013). The limitations of model fit in comparing the bi-factor versus higher-order models of human cognitive ability structure. Intelligence, 41, 407-422. Muthén, B., & Kaplan, D. (1992). A Comparison of Some Methodologies for the Factor Analysis of Non-Normal Likert Variables: A Note on the Size of the Model. British Journal of Mathematical and Statistical Psychology, 45, 19-30. https://doi.org/10.1111/j.2044-8317.1992.tb00975.x Muthén, L. K., & Muthén, B. O. (2002). How to use a Monte Carlo study to decide on sample size and determine power. Structural Equation Modeling, 9 (4), 599–620. https://doi.org/10.1207/S15328007SEM0904_8 Muthén, L.K., & Muthén, B.O. (1998–2025). Mplus (Version 8.11) [Computer software]. Los Angeles, CA: Muthén & Muthén. OECD (2024), Western Balkans Competitiveness Outlook 2024: Regional Profile, Competitiveness and Private Sector Development, OECD Publishing, Paris, https://doi.org/10.1787/170b0e53-en. Paz, D., Pontarolo, E., Santos, G., & Bernartt, M. (2021). Análisis de conteúdo sobre as definições de competência, competência digital e competência digital docente [Content analysis on the definitions of competence, digital competence and teacher digital competence]. Trabalho & Educação, 30(2), 207-225. https://doi.org/10.35699/2238-037x.2021.32817 Perdomo, B., González-Martínez, O. A., & Barrutia Barreto, I. (2020). Competencias digitales en docentes universitarios: una revisión sistemática de la literatura. Innova Scientific . https://dialnet.unirioja.es/descarga/articulo/7559036.pdf Pongrac, D., Alić, M., & Cafuta, B. (2025). Digital competences of digital natives: Measuring skills in the modern technology environment . Informatics, 12 (1), Article 23. https://doi.org/10.3390/informatics12010023 Porat, E., Blau, I., & Barak, A. (2018). Measuring digital literacies: Junior high-school students’ perceived competencies versus actual performance. Computers and Education, 126, 23–36. https://doi.org/10.1016/j.compedu.2018.06.030 Raji, N. A. S., Busson-Crowe, D. A., and Dommett, E. J. (2023). University-wide digital skills training: a case study evaluation. Educ. Sci. 13:333. doi: 10.3390/educsci13040333 Ramos, A., Casado, E., and Sevillano, U. (2023). Evaluating digital literacy levels among students in two private universities in Panama. South Florida J. Dev. 4, 1869–1886. doi: 10.46932/sfjdv4n5-003 Reise, S. P., Moore, T. M., & Haviland, M. G. (2010). Bifactor models and rotations: Exploring the extent to which multidimensional data yield univocal scale scores. Journal of Personality Assessment, 92 (6), 544–559. https://doi.org/10.1080/00223891.2010.496477 Rhemtulla, M., Brosseau-Liard, P. É., & Savalei, V. (2012). When can categorical variables be treated as continuous? A comparison of robust continuous and categorical SEM estimation methods under suboptimal conditions. Psychological methods, 17, 354-373. Richter, T., Naumann, J., & Groeben, N. (2001). The computer literacy inventory (INCOBI): An instrument for the assessment of computer literacy and attitudes toward the computer in university students of the humanities and the social sciences. Psychologie in Erziehung und Unterricht, 48(1), 1–13. Rodriguez, A., Reise, S. P., & Haviland, M. G. (2016). Evaluating bifactor models: Calculating and interpreting statistical indices. Psychological Methods, 21 (2), 137–150. https://doi.org/10.1037/met0000045 Runge, I., Lazarides, R., Rubach, C., Richter, D., & Scheiter, K. (2023). Teacher-reported instructional quality in the context of technology-enhanced teaching: The role of teachers’ digital competence- related beliefs in empowering learners. Computers & Education, 198, 104761.https://doi.org/10.1016/j.compedu.2023.104761. Sánchez-Caballé, A., Gisbert-Cervera, M., & Esteve-Mon, F. (2020). The digital competence of university students: A systematic literature review. Revista de Psicologia, Ciències de l'Educació i de l'Esport, 38(1), 63–74. https://bit.ly/3lkkh3D Satorra, A., & Bentler, P. M. (1994). Corrections to test statistics and standard errors in covariance structure analysis. In A. von Eye & C. C. Clogg (Eds.), Latent variables analysis: Applications for developmental research (pp. 399–419). Sage Publications Satorra, A., & Bentler, P. M. (2010). Ensuring positiveness of the scaled difference chi-square test statistic. Psychometrika, 75 (2), 243–248. https://doi.org/10.1007/s11336-009-9135y[1](https://publications.jrc.ec.europa.eu/repository/bitstream/JRC101254/jrc101254_digcomp%202.0%20the%20digital%20competence%20framework%20for%20citizens.%20update%20phase%201.pdf) Savalei, V. (2021). Improving fit indices in structural equation modeling with categorical data. Multivariate Behavioral Research, 56(3), 390-407. Schmitt, T.A., & Sass, D.A. (2011). Rotation criteria and hypothesis testing for exploratory factor analysis: implications for factor pattern loadings and interfactor correlations. Educational & Psychological Measurement, 71, 95-113. Scholtes, V. A. B., Terwee, C. B., & Poolman, R. W. (2011). What makes a measurement instrument valid and reliable? Injury , 42 (3), 236–240. https://doi.org/10.1016/j.injury.2010.11.042 Selber, S. A. (2004). Reimagining the functional side of computer literacy. College Composition and Communication, 55(3), 470–503. doi:10.2307/4140696. Shi, D., Maydeu-Olivares, A., & Rosseel, Y. (2020). Assessing fit in ordinal factor analysis models: SRMR vs. RMSEA. Structural Equation Modeling: A Multidisciplinary Journal, 27(1), 1-15. Siddiq, F., Hatlevik, O. E., Olsen, R. V., Throndsen, I., & Scherer, R. (2016). Taking a future perspective by learning from the past: a systematic review of assessment instruments that aim to measure primary and secondary school students’ ICT literacy. Educational Research Review, 19, 58–84. https://doi.org/10.1016/j.edurev.2016.05.002 Sillat, L. H., Tammets, K., & Laanpere, M. (2021). Digital competence assessment methods in higher education: A systematic literature review. Education in Science, 11, 402. https://doi.org/10.3390/educsci11080402 Solomon, L., & Rothblum, E. (1984). Academic procrastination: Frequency and cognitive-behavioral correlates. Journal of Counseling Psychology, 31, 503–509. http://dx.doi.org/10.1037/0022-0167.31.4.503. Soper, D. S. (2025). A-priori sample size calculator for structural equation models [Software]. Available from https://www.danielsoper.com/statcalc Sotelo-Núñez, A. C., Herrera Rojas, J. J., Herrera Rojas, M. Z., and López-Regalado, O. (2024). Competencia digital en estudiantes universitarios: una revisión sistemtica. Horizontes 8, 1781–1800. doi: 10.33996/revistahorizontes.v8i34.833 Spante M, Hashemi SS, Lundin M, Algers A. Digital competence and digital literacy in higher education research: Systematic review of concept use. Cogent Education. 2018; 5(1):1–21. https://doi.org/10.1080/2331186X.2018.1519143 UNESCO. (2017). Albania. Educational policy review; Issues and Recommendations. Paris, April 2017, p. 24. Retrieved from https://unesdoc.unesco.org/ark:/48223/pf0000247993 UNESCO. (2011). ICT competency framework for teachers . United Nations Educational, Scientific and Cultural Organization. https://unesdoc.unesco.org/ark:/48223/pf0000213475 van de Vijver, F., & Tanzer, N. K. (2004). Bias and equivalence in cross-cultural assessment: An overview. European Review of Applied Psychology, 54, 119-135. van Deursen, A. J. A. M., & van Dijk, J. A. G. M. (2010). Internet skills and the digital divide . New Media & Society , 12(8), 1239–1259. https://doi.org/10.1177/1461444809355564 van Deursen, A. J. A. M., Helsper, E. J., & Eynon, R. (2016). Development and validation of the Internet Skills Scale (ISS). Information Communication & Society, 19(6), 804–23. doi:10.1080/1369118x.2015.1078834. van Laar, E., van Deursen, A. J. A. M., van Dijk, J. A. G. M., & de Haan, J. (2017). The relation between 21st-century skills and digital skills or literacy: A systematic literature review. Computers in Human Behavior, 72, 577–588. doi:10.1016/j.chb.2017.03.010 van Zyl LE and ten Klooster PM (2022) Exploratory Structural Equation Modeling: Practical Guidelines and Tutorial With a Convenient Online Tool for Mplus. Front. Psychiatry 12:795672. doi: 10.3389/fpsyt.2021.795672 Voogt, J., & Roblin, N. P. (2012). A comparative analysis of international frameworks for 21st century competences: Implications for national curriculum policies. Journal of Curriculum Studies, 44(3), 299–321. doi:10.1080/00220272.2012.668938 Vuorikari, R., Punie, Y., Carretero, S., & Van den Brande, G. (2016). DigComp 2.0: The Digital Competence Framework for Citizens. Update Phase 1: The Conceptual Reference Model . Luxembourg: Publications Office of the European Union. https://doi.org/10.2791/11517 Vuorikari, R., Castaño Muñoz, J., Punie, Y., & Redecker, C. (2022). DSI 2.0: Towards a more impactful Digital Skills Indicator . Publications Office of the European Union. https://data.europa.eu/doi/10.2760/47949[1](https://link.springer.com/book/10.1057/9781137437037) Vuorikari, R., Jerzak, N., Karpinski, Z., Pokropek, A., & Tudek, J. (2022). Measuring digital skills across the EU: Digital Skills Indicator 2.0 (EUR 31193 EN). Publications Office of the European Union. https://doi.org/10.2760/897803 Vuorikari, R., Kluzer, S., & Punie, Y. (2022). DigComp 2.2: The Digital Competence Framework for Citizens – With new examples of knowledge, skills and attitudes (EUR 31006 EN). Publications Office of the European Union. https://doi.org/10.2760/115376 Vuorikari, R., Pokropek, A., & Castaño Muñoz, J. (2025). Enhancing digital skills assessment: Introducing compact tools for measuring digital competence. Technology, Knowledge and Learning . https://doi.org/10.1007/s10758-025-09825-x Wessel, D., Attig, C., and Franke, T. (2019). “ATI-S-An ultra-short scale for assessing affinity for technology interaction in user studies,” in Proceedings of the mensch und computer 2019 (MuC’19), eds F. Alt, A. Bulling, and T. Döring (New York, NY: Association for Computing Machinery), 147–154. doi: 10.1145/3340764.3340766 West, S. G., Finch, J. F., & Curran, P. J. (1995). Structural equation models with nonnormal variables: Problems and remedies. Thousand Oaks, CA: Sage. Widowati, A., Siswanto, I., and Wakid, M. (2023). Factors affecting students’ academic performance: self-efficacy, digital literacy, and academic engagement effects. Int. J. Instr. 16, 885–898. doi: 10.29333/iji.2023.16449a Yockey, R. D. (2016). Validation of the Short Form of the academic procrastination scale. Psychological Reports, 118(1), 171-179. Yuan, X., Rehman, S., Altalbe, A., Rehman, E., & Shahiman, M. A. (2024). Digital literacy as a catalyst for academic confidence: Exploring the interplay between academic self-efficacy and academic procrastination among medical students . BMC Medical Education, 24, Article 1317. https://doi.org/10.1186/s12909-024-06329-7 Footnotes 2024 State of the Digital Decade package ( https://digital-strategy.ec.europa.eu/en/policies/2024-state-digital-decade-package ) https://education.ec.europa.eu/focus-topics/digital-education/action-plan International Society for Technology in Education. (2016). ISTE standards for students . ISTE. https://www.iste.org/standards/iste-standards-for-students UNESCO ICT Competency Framework for Teachers (ICT-CFT) Additional Declarations The authors declare no competing interests. Supplementary Files ecwategoricalfreepred.pdf1.pdf Standardizes parameters of free variance EWC with three predictors with WLSMV Appendix.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7289357","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":495275857,"identity":"6a158b4d-f640-44ba-bf4b-aabfd51365dd","order_by":0,"name":"Emil Frashëri","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABCklEQVRIie3RMUsDMRTA8RcC75akt6ZkuK+Q0qn0tF/lSuGmQp2cC4XXqXTtoPgtiuPJwU3RueJgi+CkUBfh6KCRdlEM6uaQP4Q8Aj/eEIBQ6D/GCwA2BsBo4qbi8CrcafgIZgciquwzwZ8IqKH5HYlRbl7qy3TUgOGrPrHpKJ6O2fqZIPGRJkVtLW3eIbhe6sUq7yxswVvnBC3yEHNPoBmVBtlsqcW2NEZlqCUB85Fehbyu6c0gF497kqyjnSM97xZEVJIKdwvUYvWxBZA70vcRVSF2JQ2ME+2usLlRtj9pnt2ogY/EhPy2pmOTXDxs7kSVmnhaXm2fTtOjuYd81/6b/gBCoVAo9LV3KAtO4m0IXTYAAAAASUVORK5CYII=","orcid":"https://orcid.org/0009-0003-1062-8432","institution":"Fan S. Noli University of Korce","correspondingAuthor":true,"prefix":"","firstName":"Emil","middleName":"","lastName":"Frashëri","suffix":""},{"id":495275858,"identity":"1bb10418-d831-4557-9b44-d9b0e472ccc1","order_by":1,"name":"Marinela Teneqexhi","email":"","orcid":"","institution":"Fan S. Noli University of Korce","correspondingAuthor":false,"prefix":"","firstName":"Marinela","middleName":"","lastName":"Teneqexhi","suffix":""}],"badges":[],"createdAt":"2025-08-04 09:25:16","currentVersionCode":1,"declarations":{"humanSubjects":true,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":true,"humanSubjectConsent":true,"humanSubjectClinicalTrial":true,"humanSubjectCaseReport":true,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-7289357/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7289357/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":88588083,"identity":"a1c91a58-2647-4b4a-85e6-c3ccfd49c5e6","added_by":"auto","created_at":"2025-08-08 04:59:25","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":170837,"visible":true,"origin":"","legend":"\u003cp\u003eIntegrated and Interdependent Digital Competencies in Education\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eNote.\u003c/em\u003e Original figure created by the authors with visualization support from AI\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7289357/v1/aa6a5fba5ea9bbb156e6b28a.jpeg"},{"id":88588087,"identity":"4b9af494-9c54-40bd-a364-42040aebf845","added_by":"auto","created_at":"2025-08-08 04:59:25","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":303410,"visible":true,"origin":"","legend":"\u003cp\u003eESEM standardizes parameter estimates with WLSMV\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7289357/v1/80f85abf42c0388912556e86.png"},{"id":88591076,"identity":"16efb210-0059-456f-8655-ec74aa38c2d5","added_by":"auto","created_at":"2025-08-08 05:38:48","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1514756,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7289357/v1/599c67a4-bcec-45ab-82e9-52a632c721a9.pdf"},{"id":88588084,"identity":"36fa5c40-5567-4bc3-9158-4a6d50a7be6b","added_by":"auto","created_at":"2025-08-08 04:59:25","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":263253,"visible":true,"origin":"","legend":"\u003cp\u003eStandardizes parameters of free variance EWC with three predictors with WLSMV\u003c/p\u003e","description":"","filename":"ecwategoricalfreepred.pdf1.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7289357/v1/cd14e550ed71070ae0d5776b.pdf"},{"id":88590800,"identity":"eeeabe53-f5d1-4721-bb44-6dab5715b706","added_by":"auto","created_at":"2025-08-08 05:38:23","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":399770,"visible":true,"origin":"","legend":"","description":"","filename":"Appendix.docx","url":"https://assets-eu.researchsquare.com/files/rs-7289357/v1/8717c9b1c56799283dcd7aab.docx"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eValidating a Psychometric Instrument for Assessing Students’ Digital Skills: A Latent Variable Approach for Policy-Oriented Educational Research\u003c/p\u003e","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eThe integration of information and communication technologies has shifted education toward digital knowledge practices, aligned with both the potential and limitations of technology (Ramos et al., 2023). As higher education transitions from traditional to technology-driven teaching, it faces significant challenges (Raji et al., 2023). Developing students\u0026rsquo; digital skills is now a critical priority, with schools playing a central role in preparing youth for the demands of the 21st century. Embedding digital competencies into curricula fosters creativity, innovation, and adaptability in a digital society (Erwin \u0026amp; Mohammed, 2022). This integration is increasingly recognized as essential for preparing a future-ready workforce (Kryukova et al., 2022; Sotelo-N\u0026uacute;\u0026ntilde;ez et al., 2024). Kayyali (2024) emphasizes that digital literacy should be a core, not complementary, component of all academic programs. These competencies, developed during higher education, form a vital foundation for students\u0026rsquo; future professional success (L\u0026oacute;pez, 2023).\u0026nbsp;The European Union has paid great attention to improve the level of digital competences and skills at all stages of education and training, for all segments of the population\u0026nbsp;(Council of the European Union, 2018).\u0026nbsp;The State of the Digital Decade 2024\u003ca href=\"#_ftn1\" name=\"_ftnref1\" title=\"\"\u003e\u003c/a\u003e\u003csup\u003e1\u003c/sup\u003e report highlighted the need for European Union countries to strengthen cooperation on more ambitious collective efforts, prioritizing investment in digital education and skills, to achieve the Digital Decade targets and objectives, as a necessity for ensuring future economic prosperity. A European Union policy initiative in support of digital education ecosystem, enhancing digital skills for the digital transformation, is the Digital Education Action Plan (2021\u0026ndash;2027)\u003ca href=\"#_ftn2\" name=\"_ftnref2\" title=\"\"\u003e\u003csup\u003e\u003c/sup\u003e\u003c/a\u003e\u003csup\u003e2\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e1.1 Digital transformation in Albanian education system\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eAlbania, as part of the Western Balkans, is gradually progressing toward EU digitalization standards, though its education system remains in the early stages of digital transformation. Higher education continues to rely heavily on traditional teaching methods, hindered by infrastructure limitations and limited access to digital tools. Until recently, ICT education was limited to lower secondary levels, with informatics offered optionally in upper secondary schools (Mi\u0026ccedil;o \u0026amp; Za\u0026ccedil;ellari, 2020). Digital skill levels in Albania remain significantly below the EU average. Only 24% of individuals aged 16\u0026ndash;74 had basic digital skills in 2021, and just 4% had above-basic skills (OECD, 2024). In this context,\u0026nbsp;the development of digital and ICT competencies in Albanian public education is seen as a necessity (UNESCO, 2017; European Commission, 2021). Recognizing this gap, Albania\u0026rsquo;s \u0026ldquo;Digital Agenda Strategy 2022\u0026ndash;2026\u0026rdquo; promotes ICT integration across all education levels. Recent reforms include coding courses in primary schools and revised ICT curricula.\u0026nbsp;European Training Foundation (ETF) and the Joint Research Centre (JRC) of the European Commission in the framework of the Digital Agenda for the Western Balkans have launched SELFIEforTeachers (SfT\u003cstrong\u003e),\u0026nbsp;\u003c/strong\u003ea measurement tool for assessing and improving primary and secondary teachers\u0026apos; digital literacy. Despite these efforts, tangible progress in digital inclusion remains limited.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e1.2 The problematic situation created and the need for intervention\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eEducators face a growing challenge: the digital skills of graduating students often fall short of what employers expect, limiting both student success and workforce readiness (Feijao et al., 2021; Kayyali, 2024). Studies show that digital competence remains low among both teachers and students, even among so-called \u0026ldquo;digital natives\u0026rdquo; (Gallardo-Echenique et al., 2015). This gap calls for urgent updates to digital education programs. To design effective interventions, educators need reliable tools to assess digital skills using sound measurement methods (Siddiq et al., 2016; Laanpere, 2019; Helsper et al., 2020). However, rapid technological change has made it difficult to define and measure these competencies consistently (Sillat et al., 2021). A dynamic, future-focused framework must be envisioned, one through which students are empowered with adaptable digital skills to lead and thrive in a rapidly evolving, knowledge-driven world.\u003c/p\u003e\n\u003cdiv id=\"ftn2\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003c/div\u003e"},{"header":"2. Conceptualization of digital skills","content":"\u003cp\u003eOver the past two to three decades, a range of conceptual terms has been employed to describe students\u0026rsquo; engagement with digital technologies, including ICT competences, digital literacy, media literacy, and 21st-century skills (S\u0026aacute;nchez-Caball\u0026eacute; et al., 2020). These terms reflect the evolving nature of digital competence, which was initially understood as a set of basic technical skills (Richter et al., 2001; Selber, 2004). However, contemporary perspectives recognize digital competence as a multidimensional construct encompassing technical abilities, cognitive skills, and socio-emotional attributes, such as critical thinking, problem-solving, and adaptability, essential for meaningful participation in a knowledge-based society (Van Laar et al., 2017; Fan \u0026amp; Wang, 2022; Widowati et al., 2023). This conceptual expansion has led to the development of various frameworks tailored to specific educational and professional contexts. Voogt and Roblin (2012), through a comparative analysis of eight major frameworks, identified a convergence around core digital competencies. Among these, the European Commission\u0026rsquo;s DIGCOMP framework has gained prominence for its comprehensive and adaptable structure. Developed by the Joint Research Centre Institute for Prospective Technological Studies (JRC-IPTS), DIGCOMP synthesizes insights from 15 ICT literacy models and international assessments such as PISA and PIAAC (Ferrari, 2013; Vuorikari et al., 2016; Carretero et al., 2017). The framework defines five competence areas, Information, Communication, Content Creation, Safety, and Problem Solving, across five proficiency levels, allowing for nuanced assessment and targeted development. Subsequent revisions of DIGCOMP (e.g., DigComp 2.0 and 2.1) have enhanced its applicability across diverse sectors and learner groups, reflecting the increasing need for context-sensitive approaches to digital competence. These developments underscore the importance of a flexible, future-oriented framework capable of equipping individuals with adaptable digital skills in response to the dynamic demands of the digital age.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e2.1.\u0026nbsp;\u003c/em\u003e\u003cem\u003eStudents\u0026rsquo; digital skills assessment. The background for the empirical research\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eMa and Ismail (2025) conducted a comprehensive study, combining bibliometric analysis and systematic review, to examine the landscape of research on digital competencies within the education sector. Their findings indicate that digital competencies remain a relatively recent but rapidly expanding area of inquiry, with significant potential for further exploration. Although scholarly interest in digital competencies in higher education has grown markedly, particularly since 2021, there remains no unified conceptualization of what constitutes students\u0026rsquo; digital competencies (Ilom\u0026auml;ki et al., 2014; Spante et al., 2018) or digital skills (Barbazan et al., 2021; Paz et al., 2021; Perdomo et al., 2020). \u0026nbsp; For the purposes of the present study, digital skills are defined in accordance with the International Telecommunication Union (ITU, 2018) as \u0026ldquo;\u003cem\u003ethe ability to use ICTs in ways that help individuals to achieve beneficial, high-quality outcomes in everyday life for themselves and for others, and to reduce potential harm associated with more negative aspects of digital engagement\u003c/em\u003e\u0026rdquo;. This definition has been widely adopted in recent literature (e.g., van Deursen et al., 2016; Helsper \u0026amp; van Deursen, 2018; Haddon et al., 2020; Helsper et al., 2020), and serves as a robust conceptual foundation for assessing digital skills, particularly when aligned with the European Commission\u0026rsquo;s DIGCOMP framework. The present study employs the youth Digital Skills Indicator (yDSI), developed by Helsper et al. (2020), as the primary measurement instrument. The yDSI is a cross-nationally validated tool designed to assess digital skills among young people across four core dimensions: (1) technical and operational skills, (2) information navigation and processing, (3) communication and interaction, and (4) content creation and production. Each dimension captures both functional and critical aspects of digital engagement, reflecting the multifaceted nature of digital competence in contemporary society. Although the yDSI is conceptually grounded in the DIGCOMP framework (European Commission, 2020), it excludes the \u0026ldquo;problem-solving and safety\u0026rdquo; dimension. This exclusion is based on evidence from the literature suggesting that elements of this domain are already embedded within the other four dimensions. The yDSI scale comprises 24 items, carefully developed and validated through cognitive interviews, exploratory and confirmatory factor analyses, measurement invariance testing, and performance-based assessments. Special attention was given to avoiding common pitfalls in self-assessment tools, such as social desirability and acquiescence bias, by piloting item wording and response formats to ensure strong psychometric properties. The yDSI emphasizes the importance of equipping young individuals with foundational skills across all four dimensions to enable full and meaningful participation in a digital society. Unlike measures of digital self-efficacy, which focus on perceived competence, the yDSI prioritizes actual skill acquisition as a more reliable predictor of positive digital engagement. The instrument draws on both academic literature and grey literature sources (e.g., Cortesi et al., 2020), and has been successfully implemented in large-scale initiatives such as the EU-funded ySKILLS project under Horizon 2020 (Helsper et al., 2020).\u0026nbsp;As Helsper et al. (2020) pointed out, yDSI can be adopted in other projects with young people.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e2.2 Types of methodologies used to measure digital skills\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eOver the past two decades, digital skills have been assessed using a range of methods, including direct, indirect, and hybrid approaches. Among these, performance-based assessments, such as PIAAC (OECD) and ICILS (NCES), are considered the most valid, as they involve authentic, task-based interactions with digital environments (Darling-Hammond \u0026amp; Adamson, 2010). However, their high cost, context specificity, and labor-intensive nature limit their scalability for large-scale studies (Vuorikari et al., 2025; van Deursen \u0026amp; van Dijk, 2010). Consequently, indirect methods, particularly self-reported questionnaires, are more commonly used due to their efficiency and ease of administration. These tools have evolved through three generations to address issues such as social desirability bias and overestimation of competence (Ballantine et al., 2007; Porat et al., 2018). The first-generation \u003cem\u003eDigital Skills Indicator\u003c/em\u003e (DSI) and its updated version, DSI 2.0 (Vuorikari et al., 2022), laid the groundwork for improved psychometric design. The second-generation \u003cem\u003eInternet Skills Scale\u003c/em\u003e (ISS) by van Deursen et al. (2016) introduced more behaviorally anchored items. The third-generation \u003cem\u003eYouth Digital Skills Indicator\u003c/em\u003e (yDSI), developed by Helsper et al. (2020), further refines this approach by focusing on four core dimensions: technical and operational skills, information navigation, communication and interaction, and content creation. Excluding problem-solving and safety due to conceptual overlap, the yDSI offers a validated, cross-national tool for assessing youth digital skills, emphasizing actual ability over perceived self-efficacy.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e2.3 Research Aims\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eResearch on digital competencies and skills has expanded considerably over the past two decades, with a notable acceleration in recent years. While much of the existing literature has focused on the theoretical conceptualization of digital skills, relatively limited attention has been devoted to the development of empirical models that can inform practical applications. As a degree of conceptual consensus has emerged, there is now a pressing need to advance models that explain the interaction between operationalized indicators of digital skills, particularly in light of their multidimensional and hierarchical structure. This shift from conceptual clarity to empirical modeling holds important implications for policy. Robust, evidence-based models are essential for designing targeted interventions that address specific skill gaps across different learner populations. The present study contributes to this effort by assessing university students\u0026rsquo; functional digital skills using the short version of the Youth Digital Skills Indicator (yDSI) (Helsper et al., 2020). By applying latent variable modeling techniques, this study aims to uncover the structural relationships among functional digital skill dimensions, offering insights that can support the development of more effective, data-driven digital education policies.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e2.4 Measurement scale assessment \u0026nbsp;\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eWhen developing a quantitative instrument, it is essential to rigorously assess its psychometric quality, particularly in terms of reliability and validity (Scholtes et al., 2011). Even if an instrument has been previously validated, re-evaluating its properties in new contexts remains best practice (Bandalos, 2018). In recent years, researchers across various countries have increasingly employed latent variable models to assess digital competencies in both pre-university and higher education contexts. These models frequently utilize Exploratory Factor Analysis (EFA), as well as first-order and, occasionally, second-order Confirmatory Factor Analysis (CFA). However, as noted by Marsh and colleagues (Marsh, Hau, \u0026amp; Grayson, 2005; Marsh, 2007; Marsh et al., 2010), when measuring multifactorial constructs with factors represented by at least five items each, the restrictive independent clusters model (ICM) of CFA often fails to achieve acceptable model fit. To address this, newer approaches have been proposed that account for the fallibility of items as indicators of single latent dimensions and the hierarchical structure of the constructs being measured. Among these, Exploratory Structural Equation Modeling (ESEM), bifactor-CFA (B-CFA), and bifactor-ESEM (B-ESEM) have emerged as valuable tools for addressing construct-relevant multidimensionality. Although still limited, the application of B-CFA and B-ESEM in the assessment of digital skills is gaining traction in the literature.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e2.4.1 Modeling Latent Structures in Digital Skills Assessment\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eExploratory Factor Analysis (EFA)\u003c/em\u003e is a data-driven technique used to identify the underlying latent dimensions that explain the covariance among observed variables. It allows for free estimation of factor loadings and cross-loadings to derive a simple, interpretable structure (Brown, 2015). While suitable for exploratory purposes, EFA lacks confirmatory power, limiting its application in complex, theory-driven research (Morin et al., 2013; Marsh et al., 2009).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eConfirmatory Factor Analysis (CFA),\u003c/em\u003e introduced by J\u0026ouml;reskog (1969), is a theory-driven approach that tests how well the data fit a hypothesized factor structure. It assumes items load on a single latent factor, with cross-loadings fixed to zero (Marsh, 2007). Although CFA offers advantages in model parsimony, error correction, and measurement invariance (Morin et al., 2020), its restrictive assumptions can lead to inflated factor correlations and compromised discriminant validity in multidimensional constructs (Marsh et al., 2009; Schmitt \u0026amp; Sass, 2011).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eExploratory Structural Equation Modeling (ESEM),\u003c/em\u003e developed by Asparouhov and Muth\u0026eacute;n (2009), integrates EFA within the SEM framework, allowing for cross-loadings, while retaining confirmatory capabilities. ESEM models are more flexible and typically yield better fit and more realistic factor correlations than CFA (Marsh et al., 2014). They are particularly useful for assessing instruments with conceptually related dimensions, such as digital competencies (Morin et al., 2016a, 2016b).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eBifactor-CFA (B-CFA)\u003c/em\u003e models allow each item to load on a general factor and one specific factor, enabling simultaneous assessment of general and domain-specific variance (Reise et al., 2010). However, by constraining cross-loadings to zero, B-CFA may overestimate general factor loadings and distort model interpretation (Murray \u0026amp; Johnson, 2013).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eBifactor-ESEM (B-ESEM)\u003c/em\u003e addresses these limitations by combining the bifactor structure with ESEM\u0026rsquo;s flexibility, allowing cross-loadings to approach, but not equal, zero through orthogonal target rotation. This approach better captures the hierarchical and multidimensional nature of constructs and avoids parameter inflation (Morin et al., 2016b). B-ESEM also supports model stability and the development of higher-order models (Marsh et al., 2020).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e2.4.2 Modeling Interdependent Dimensions of Digital Skills\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe latent construct of digital skills assessed in this study comprises four interrelated dimensions. Both the Youth Digital Skills Indicator (yDSI; Helsper et al., 2021) and the DigComp 2.2 framework (Vuorikari et al., 2022) emphasize that these dimensions are not discrete but functionally interconnected. The yDSI provides both theoretical and empirical justification for a four-dimensional model, highlighting the critical interdependencies among the dimensions. Similarly, DigComp 2.2 outlines the conceptual and practical links across digital competence areas.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAmong these, \u003cem\u003eTechnical and Operational Skills\u003c/em\u003e serve as a foundational layer, enabling the development of other competencies. \u003cem\u003eInformation Navigation and Processing Skills\u003c/em\u003e depend on technical fluency and, in turn, support \u003cem\u003eCommunication and Interaction Skills\u003c/em\u003e. Effective digital communication requires not only platform proficiency but also the ability to interpret and disseminate information. Finally, \u003cem\u003eContent Creation and Production Skills\u003c/em\u003e represent a culmination of the other three, requiring technical tools, information literacy, and communication strategies for effective digital content development.\u003c/p\u003e\n\u003cp\u003eEmpirical evidence supports these interdependencies. For instance, Pongrac et al. (2025) found significant positive correlations among the four digital skill dimensions. Therefore, statistical models used to assess this construct must account for these interrelations. Exploratory Structural Equation Modeling (ESEM) is particularly well-suited for this purpose, as it naturally accommodates cross-loadings and provides a more realistic representation of overlapping constructs, especially in educational research (Sillat et al., 2021).\u003c/p\u003e\n\u003cp\u003eRecent studies (e.g., Dierendonck, 2023) have demonstrated that ESEM often yields superior model fit compared to traditional CFA or bifactor-CFA models when applied to multidimensional educational constructs. Furthermore, ESEM can be extended to hierarchical structures, such as hierarchical ESEM (H-ESEM) or bifactor-ESEM (B-ESEM). The B-ESEM model is especially valuable for modeling digital skills because:\u003c/p\u003e\n\u003col start=\"1\" type=\"1\"\u003e\n \u003cli\u003eIt incorporates cross-loadings, allowing for a more comprehensive assessment of theoretically grounded skill dimensions (Runge et al., 2023).\u003c/li\u003e\n \u003cli\u003eIt enables the examination of how general and specific skill dimensions relate to students\u0026rsquo; actual technology use.\u003c/li\u003e\n\u003c/ol\u003e\n\u003ch3\u003e\u003cstrong\u003e\u003cem\u003e2.4.3 Exogenous Variables as Predictors of the Digital Skills Construct\u003c/em\u003e\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003e\u003cem\u003eAcademic procrastination\u003c/em\u003e is the tendency to delay academic tasks or exam preparation without valid justification (Solomon \u0026amp; Rothblum, 1984). The Academic Procrastination Scale Short Form (Yockey, 2016), a five-item unidimensional measure, is widely regarded as a reliable and valid tool. Recent studies indicate a negative relationship between digital competence and academic procrastination, suggesting that students with stronger digital skills are less prone to procrastinate (Fayda-Kınık, 2023; Yuan et al., 2024).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eSelf-efficacy\u003c/em\u003e, as defined by Bandura (1977), refers to individuals\u0026rsquo; beliefs in their ability to organize and execute actions to achieve specific goals. In education, academic self-efficacy reflects students\u0026rsquo; confidence in their learning and academic success. The three-item General Self-Efficacy Scale Short Form (Beierlein et al., 2013) demonstrates strong psychometric properties, including unidimensionality and cross-linguistic validity. Research consistently shows a positive association between academic self-efficacy and digital competence across various skill domains (Falma \u0026amp; Putra, 2025; Javier-Aliaga et al., 2024).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eAffinity for technology\u003c/em\u003e refers to an individual\u0026rsquo;s interest, comfort, and confidence in engaging with digital tools (Franke et al., 2019). The four-item ATI-S scale (Wessel et al., 2019) reliably measures this construct across gender and age groups. Research shows that students with higher technology affinity tend to excel in digital skill domains, including programming and technical problem-solving (Jokisch et al., 2025; Cabezas-Gonz\u0026aacute;lez et al., 2023).\u003c/p\u003e"},{"header":"3. Methods","content":"\u003cp\u003e\u003cem\u003e3.1 Participants and Procedure\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eData for this study were collected between January and May 2025 from students at Fan S. Noli University of Kor\u0026ccedil;a, located in the southeastern region of Albania. The survey instrument was primarily based on the short version of the Youth Digital Skills Indicator (yDSI; Helsper et al., 2020). The core of the instrument consisted of 24 items measuring functional digital skills across four interrelated dimensions: Technical and Operational Skills (TO), Information Navigation and Processing Skills (INP), Communication and Interaction Skills (CI), and Content Creation and Production Skills (CCP).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eResponses were recorded using a six-point Likert scale: 0 = \u003cem\u003eI do not understand what you mean by this\u003c/em\u003e; 1 = \u003cem\u003eNot at all true of me\u003c/em\u003e; 2 = \u003cem\u003eNot very true of me\u003c/em\u003e; 3 = \u003cem\u003eNeither true nor untrue of me\u003c/em\u003e; 4 = \u003cem\u003eMostly true of me\u003c/em\u003e; 5 = \u003cem\u003eVery true of me\u003c/em\u003e. An additional response option, \u003cem\u003eI do not want to answer\u003c/em\u003e, was coded as 99 and treated as a missing value.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo assess predictive validity, the questionnaire included items for three exogenous latent variables: academic procrastination (5 items), self-efficacy (3 items), and technology affinity (4 items, including two reverse-coded), all rated on a five-point Likert scale (1 = \u003cem\u003eStrongly disagree\u003c/em\u003e to 5 = \u003cem\u003eStrongly agree\u003c/em\u003e).\u003c/p\u003e\n\u003cp\u003ePrior to data collection, the questionnaire was reviewed by ICT and linguistics experts to ensure content validity and accurate translation into Albanian. Following institutional approval, the survey was distributed via the university\u0026rsquo;s official website using Google Forms. Completion time was approximately 10 minutes. A total of 603 students participated (Mean age = 22.07 years; 69.3% female). The sample size closely approximated the recommended minimum of 630 participants, as calculated using the Soper (2025) a priori sample size calculator for structural equation modeling, based on an anticipated effect size of 0.15 and a desired statistical power level of 0.80 (4 latent variables and 24 observed variables).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e3.2 Psychometric Modeling and Analytical Approach\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eTo assess the temporal stability of the instrument, a test\u0026ndash;retest procedure was conducted with a randomly selected subsample of 30 students. Participants completed the self-report questionnaire twice, with a three-week interval between administrations. The Spearman\u0026rsquo;s rank correlation coefficient between the two time points was high (\u0026rho;ho = 0.794), indicating strong test\u0026ndash;retest reliability and suggesting that the instrument yields consistent results over time.\u003c/p\u003e\n\u003cp\u003eTo examine the factorial structure of students\u0026rsquo; functional skills, the 24-item scale was analyzed using a series of latent variable models: Confirmatory Factor Analysis (CFA), Exploratory Structural Equation Modeling (ESEM), bifactor-CFA (B-CFA), and bifactor-ESEM (B-ESEM), following the comparative framework of Morin et al. (2016a). This approach enables the identification of construct-relevant psychometric multidimensionality.\u003c/p\u003e\n\u003cp\u003eCFA applied the restrictive Independent Cluster Model (ICM), allowing items to load only on their designated factors, with all cross-loadings fixed to zero. ESEM, using oblique TARGET rotation (Asparouhov \u0026amp; Muth\u0026eacute;n, 2009), allowed estimation of primary loadings while constraining cross-loadings to be as close to zero as possible. Bifactor-CFA modeled a general digital skills factor alongside four specific dimensions, assuming orthogonal factors and no cross-loadings. Bifactor-ESEM extended this by combining the bifactor structure with ESEM\u0026rsquo;s flexibility, allowing cross-loadings to be freely estimated but constrained as close as possible to zero, offering a more realistic representation of the hierarchical and multidimensional nature of digital skills.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e3.3 Data Analysis\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eData analysis was performed using Mplus Version 8.11 (Muth\u0026eacute;n \u0026amp; Muth\u0026eacute;n), with each of the four competing models estimated independently. Given the six-category ordinal nature of the digital skills items and their approximately normal distribution (see Table 1, Appendix), model robustness was evaluated using two recommended estimators: the Robust Maximum Likelihood (MLR), which treats ordinal data as continuous and applies Satorra-Bentler (Satorra \u0026amp; Bentler, 1994) \u0026nbsp;corrections (Rhemtulla et al., 2012) ) to adjust for kurtosis, which is a primary concern in non-normal data, and the Weighted Least Squares Mean and Variance adjusted (WLSMV), which treats data as categorical and uses polychoric correlations. Following Finney and DiStefano (2013), WLSMV is particularly suitable for ordinal data with five or fewer categories and performs reliably with sample sizes exceeding 200 (Beauducel \u0026amp; Herzberg, 2006; Bandalos, 2014; Flora \u0026amp; Curran, 2004). Model fit was evaluated using standard indices: CFI and TLI (incremental fit), and RMSEA with 90% CI and SRMR (absolute fit), following established thresholds (Hu \u0026amp; Bentler, 1999; Marsh et al., 2004). CFI and TLI values \u0026ge; .90/.95 and RMSEA \u0026le; .08/.05, along with SRMR \u0026le; .08, indicate acceptable to excellent fit. Due to its sensitivity to sample size and model misspecification, the chi-square test was interpreted alongside other indices (Marsh et al., 2005). For ordinal data, SRMR has been shown to outperform RMSEA (Shi et al., 2020).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAlthough ordinal variables are inherently non-normally distributed (Kaplan, 2009), skewness and kurtosis values for all items were below 1 in absolute terms, indicating only minor deviations from normality (West et al., 1995; Hair et al., 2010). According to Muth\u0026eacute;n and Kaplan (1992), under conditions of approximate normality, minimal distortion is expected in model estimation. When sample sizes are adequate, both Maximum Likelihood (ML) and Weighted Least Squares (WLS) estimators perform reliably. All competing models demonstrated good to excellent fit across multiple indices (see Table 1).\u003c/p\u003e\n\u003cp\u003eThe only fit index that approached concern for categorical data was the RMSEA, with values ranging from .075 to .090, near the conventional cutoff of .08, indicating acceptable model fit. \u003cem\u003eMplus\u003c/em\u003e Version 8.11 employs a modified computation of the SRMR (Asparouhov \u0026amp; Muth\u0026eacute;n, 2018), for which values below .08 are considered indicative of good fit. Next, we compared four competing models to evaluate their relative fit. Specifically, comparisons were made between ESEM and CFA, B-CFA and CFA, and B-ESEM and ESEM, using two robust estimators. Following the framework proposed by Morin et al. (2016b), model comparisons were based on differences in goodness-of-fit indices and measurement quality indicators.\u003c/p\u003e\n\u003cp\u003eTable 1. Model fit evaluation of four competing models with two different estimators\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" align=\"\" width=\"591\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"11\" style=\"width: 591px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMLR\u0026nbsp;\u003c/strong\u003eestimator for continuous data\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 54px;\"\u003e\n \u003cp\u003eModel\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 54px;\"\u003e\n \u003cp\u003e\u003cimg width=\"14\" height=\"17\" src=\"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAA4AAAARCAMAAADaFm2tAAAAAXNSR0IArs4c6QAAAGlQTFRFAAAAAAAAADpmADqQAGa2OgAAOgA6OgBmOma2OpDbZgAAZgA6ZjoAZjqQZmZmZrb/kDoAkDpmkLb/kNv/tmYAtmY6tpCQtrZmttu2tv//25A625CQ27a22//b2////7Zm/9uQ//+2///bFqsyLgAAAAF0Uk5TAEDm2GYAAAAJcEhZcwAADsQAAA7EAZUrDhsAAAAZdEVYdFNvZnR3YXJlAE1pY3Jvc29mdCBPZmZpY2V/7TVxAAAAYklEQVQYV42OSRKAIAwEgztuqIAKqAT+/0g9abiZW1dmugbg30XNMvNFPQclkuZ6UTwScgYXgNAXgJUEzxh7um4fzok44rw1VKFy6gidpE89UrTcEhHWxnL3jm0FYJlu/Uw3LO0EtqM1878AAAAASUVORK5CYII=\" alt=\"image\"\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 36px;\"\u003e\n \u003cp\u003edf\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 42px;\"\u003e\n \u003cp\u003eCFI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 42px;\"\u003e\n \u003cp\u003eTLI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 78px;\"\u003e\n \u003cp\u003eRMSEA\u003c/p\u003e\n \u003cp\u003e90% CI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48px;\"\u003e\n \u003cp\u003eSRMR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003eAIC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003eBIC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003eSABIC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003eMeets Criteria\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 54px;\"\u003e\n \u003cp\u003eM1\u003c/p\u003e\n \u003cp\u003eCFA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 54px;\"\u003e\n \u003cp\u003e540.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 36px;\"\u003e\n \u003cp\u003e246\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 42px;\"\u003e\n \u003cp\u003e.954\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 42px;\"\u003e\n \u003cp\u003e.949\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 78px;\"\u003e\n \u003cp\u003e.045\u003c/p\u003e\n \u003cp\u003e(.040 .050)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48px;\"\u003e\n \u003cp\u003e.033\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e37527.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e37870.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e37622.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 54px;\"\u003e\n \u003cp\u003eM2\u003c/p\u003e\n \u003cp\u003eESEM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 54px;\"\u003e\n \u003cp\u003e413.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 36px;\"\u003e\n \u003cp\u003e186\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 42px;\"\u003e\n \u003cp\u003e.965\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 42px;\"\u003e\n \u003cp\u003e.948\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 78px;\"\u003e\n \u003cp\u003e.045\u003c/p\u003e\n \u003cp\u003e(.039 .051)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48px;\"\u003e\n \u003cp\u003e.021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e37446.3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e38052.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e37614.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 54px;\"\u003e\n \u003cp\u003eM3\u003c/p\u003e\n \u003cp\u003eB-CFA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 54px;\"\u003e\n \u003cp\u003e484.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 36px;\"\u003e\n \u003cp\u003e228\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 42px;\"\u003e\n \u003cp\u003e.960\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 42px;\"\u003e\n \u003cp\u003e.952\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 78px;\"\u003e\n \u003cp\u003e.043\u003c/p\u003e\n \u003cp\u003e(.038 .049)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48px;\"\u003e\n \u003cp\u003e.029\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e37463.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e37885.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e37580.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 54px;\"\u003e\n \u003cp\u003eM4\u003c/p\u003e\n \u003cp\u003eB-ESEM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 54px;\"\u003e\n \u003cp\u003e366.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 36px;\"\u003e\n \u003cp\u003e166\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 42px;\"\u003e\n \u003cp\u003e.969\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 42px;\"\u003e\n \u003cp\u003e.948\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 78px;\"\u003e\n \u003cp\u003e.045\u003c/p\u003e\n \u003cp\u003e(.039 .051)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48px;\"\u003e\n \u003cp\u003e.017\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e37352.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e38046.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e37544.9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"11\" style=\"width: 591px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWLSMV\u003c/strong\u003e estimator for categorical data\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 54px;\"\u003e\n \u003cp\u003eM1\u003c/p\u003e\n \u003cp\u003eCFA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 54px;\"\u003e\n \u003cp\u003e1267.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 36px;\"\u003e\n \u003cp\u003e246\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 42px;\"\u003e\n \u003cp\u003e.981\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 42px;\"\u003e\n \u003cp\u003e.978\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 78px;\"\u003e\n \u003cp\u003e.083\u003c/p\u003e\n \u003cp\u003e(.079 .088)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48px;\"\u003e\n \u003cp\u003e.026\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003eNo/Yes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 54px;\"\u003e\n \u003cp\u003eM2\u003c/p\u003e\n \u003cp\u003eESEM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 54px;\"\u003e\n \u003cp\u003e1086.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 36px;\"\u003e\n \u003cp\u003e186\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 42px;\"\u003e\n \u003cp\u003e.983\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 42px;\"\u003e\n \u003cp\u003e.975\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 78px;\"\u003e\n \u003cp\u003e.090\u003c/p\u003e\n \u003cp\u003e(.085 .095)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48px;\"\u003e\n \u003cp\u003e.019\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003eNo/Yes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 54px;\"\u003e\n \u003cp\u003eM3\u003c/p\u003e\n \u003cp\u003eB-CFA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 54px;\"\u003e\n \u003cp\u003e990.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 36px;\"\u003e\n \u003cp\u003e228\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 42px;\"\u003e\n \u003cp\u003e.986\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 42px;\"\u003e\n \u003cp\u003e.983\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 78px;\"\u003e\n \u003cp\u003e.075\u003c/p\u003e\n \u003cp\u003e(.070 .080)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48px;\"\u003e\n \u003cp\u003e.024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 54px;\"\u003e\n \u003cp\u003eM4\u003c/p\u003e\n \u003cp\u003eB-ESEM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 54px;\"\u003e\n \u003cp\u003e803.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 36px;\"\u003e\n \u003cp\u003e166\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 42px;\"\u003e\n \u003cp\u003e.988\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 42px;\"\u003e\n \u003cp\u003e.980\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 78px;\"\u003e\n \u003cp\u003e.080\u003c/p\u003e\n \u003cp\u003e(.075 .086)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48px;\"\u003e\n \u003cp\u003e.015\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 60px;\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"11\" style=\"width: 591px;\"\u003e\n \u003cp\u003e\u003cimg width=\"14\" height=\"17\" src=\"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAA4AAAARCAMAAADaFm2tAAAAAXNSR0IArs4c6QAAAGlQTFRFAAAAAAAAADpmADqQAGa2OgAAOgA6OgBmOma2OpDbZgAAZgA6ZjoAZjqQZmZmZrb/kDoAkDpmkLb/kNv/tmYAtmY6tpCQtrZmttu2tv//25A625CQ27a22//b2////7Zm/9uQ//+2///bFqsyLgAAAAF0Uk5TAEDm2GYAAAAJcEhZcwAADsQAAA7EAZUrDhsAAAAZdEVYdFNvZnR3YXJlAE1pY3Jvc29mdCBPZmZpY2V/7TVxAAAAYklEQVQYV42OSRKAIAwEgztuqIAKqAT+/0g9abiZW1dmugbg30XNMvNFPQclkuZ6UTwScgYXgNAXgJUEzxh7um4fzok44rw1VKFy6gidpE89UrTcEhHWxnL3jm0FYJlu/Uw3LO0EtqM1878AAAAASUVORK5CYII=\" alt=\"image\"\u003e\u0026nbsp;- Chi-square; df - degrees of freedom; TLI - Tucker-Lewis Index; CFI - Comparative Fit Index; RMSEA - Root Mean Square Error of Approximation [90%CI]; SRMR - Standardized Root Mean Square Residual; AIC - Akaike Information Criterion; BIC - Bayes Information Criterion; SABIC - Adjusted Bayes Information Criterion.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eFor nested models, differences smaller than .01 in CFI and TLI, and less than .015 in RMSEA and SRMR, are generally considered indicative of equivalent model fit (Chen, 2007; Cheung \u0026amp; Rensvold, 2002). When using the MLR estimator, models with lower AIC, BIC, and aBIC values are preferred. As shown in Table 2, the ESEM model demonstrated a relatively better fit to the data compared to the CFA model.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 2. Comparison of four competing models with two different estimators\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"616\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd colspan=\"10\" valign=\"bottom\" style=\"width: 542px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMLR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\n \u003cp\u003eModel comp.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003e\u0026Delta; \u0026chi;2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026Delta; df\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 45px;\"\u003e\n \u003cp\u003e\u0026Delta; CFI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026Delta; TLI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 69px;\"\u003e\n \u003cp\u003e\u0026Delta; RMSEA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 65px;\"\u003e\n \u003cp\u003e\u0026Delta; SRMR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\n \u003cp\u003e\u0026Delta; AIC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\n \u003cp\u003e\u0026Delta; BIC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\n \u003cp\u003e\u0026Delta; aBIC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 52px;\"\u003e\n \u003cp\u003eMeets criteria\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\n \u003cp\u003eM2 vs. M1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003e-126.710\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 38px;\"\u003e\n \u003cp\u003e-60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 45px;\"\u003e\n \u003cp\u003e.011\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 46px;\"\u003e\n \u003cp\u003e-.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 69px;\"\u003e\n \u003cp\u003e.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 65px;\"\u003e\n \u003cp\u003e-.012\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\n \u003cp\u003e-81.060\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\n \u003cp\u003e182.660\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\n \u003cp\u003e-7.830\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 52px;\"\u003e\n \u003cp\u003eNo/Yes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\n \u003cp\u003eM3 vs. M1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003e-55.580\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 38px;\"\u003e\n \u003cp\u003e-18\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 45px;\"\u003e\n \u003cp\u003e.006\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 46px;\"\u003e\n \u003cp\u003e.003\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 69px;\"\u003e\n \u003cp\u003e-.002\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 65px;\"\u003e\n \u003cp\u003e-.004\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\n \u003cp\u003e-64.360\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\n \u003cp\u003e14.760\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\n \u003cp\u003e-42.400\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 52px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\n \u003cp\u003eM4 vs. M2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003e-46.970\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 38px;\"\u003e\n \u003cp\u003e-20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 45px;\"\u003e\n \u003cp\u003e.004\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 46px;\"\u003e\n \u003cp\u003e.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 69px;\"\u003e\n \u003cp\u003e.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 65px;\"\u003e\n \u003cp\u003e-.004\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\n \u003cp\u003e-94.250\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\n \u003cp\u003e-6.350\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\n \u003cp\u003e-6.350\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 52px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd colspan=\"10\" valign=\"bottom\" style=\"width: 542px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWLSMV\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\n \u003cp\u003eModel comp.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003e\u0026Delta; \u0026chi;2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 38px;\"\u003e\n \u003cp\u003e\u0026Delta; df\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 45px;\"\u003e\n \u003cp\u003e\u0026Delta; CFI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 46px;\"\u003e\n \u003cp\u003e\u0026Delta; TLI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 69px;\"\u003e\n \u003cp\u003e\u0026Delta; RMSEA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 65px;\"\u003e\n \u003cp\u003e\u0026Delta; SRMR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 52px;\"\u003e\n \u003cp\u003eMeets criteria\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\n \u003cp\u003eM2 vs. M1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003e-181.070\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 38px;\"\u003e\n \u003cp\u003e-60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 45px;\"\u003e\n \u003cp\u003e.002\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 46px;\"\u003e\n \u003cp\u003e-.003\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 69px;\"\u003e\n \u003cp\u003e.007\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 65px;\"\u003e\n \u003cp\u003e-.007\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 52px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\n \u003cp\u003eM3 vs. M1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003e-277.040\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 38px;\"\u003e\n \u003cp\u003e-18\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 45px;\"\u003e\n \u003cp\u003e.005\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 46px;\"\u003e\n \u003cp\u003e.005\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 69px;\"\u003e\n \u003cp\u003e-.008\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 65px;\"\u003e\n \u003cp\u003e-.002\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 52px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 74px;\"\u003e\n \u003cp\u003eM4 vs. M2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003e-283.330\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 38px;\"\u003e\n \u003cp\u003e-20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 45px;\"\u003e\n \u003cp\u003e.005\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 46px;\"\u003e\n \u003cp\u003e.005\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 69px;\"\u003e\n \u003cp\u003e-.010\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 65px;\"\u003e\n \u003cp\u003e-.004\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 54px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 52px;\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eAlthough the ESEM model showed slightly better fit than the CFA model, the differences were not substantial across all indices. Improvements were primarily observed in SRMR, AIC, aBIC, and to a lesser extent, in RMSEA, while BIC favored the more parsimonious CFA model. Except for the CFI difference under the MLR estimator, most fit index differences fell within the thresholds for model equivalence (Chen, 2007; Cheung \u0026amp; Rensvold, 2002). However, the interpretation of model fit differences remains a topic of ongoing debate (Murray \u0026amp; Johnson, 2013), and should be complemented by an examination of parameter estimates.\u003c/p\u003e\n\u003cp\u003eAs shown in Table 3, the ESEM model yielded lower factor loadings and inter-factor correlations compared to CFA, with mean reductions of 7.06% and 6.59%, respectively, across estimators. The largest decrease in item loadings was observed for the INP factor (12.71%), while the smallest was for CCP (3.53%). A similar pattern was observed in the inter-factor correlations.\u003c/p\u003e\n\u003cp\u003eBased on the practical recommendations of van Zyl and ten Klooster (2022), and as supported by Morin et al. (2020), the ESEM model was retained for further analysis. These authors suggest that even if ESEM does not show better overall fit than CFA, it should still be preferred when it produces lower factor correlations, as this indicates better discriminant validity between the constructs.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 3. Factor loadings and factor correlations comparison for CFA and ESEM models (continuous data)\u003c/p\u003e\n\u003cdiv align=\"\"\u003e\n \u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"601\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 61px;\"\u003e\n \u003cp\u003eItems TO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003eCFA/ESEM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 63px;\"\u003e\n \u003cp\u003eItems INP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003eCFA/ESEM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 63px;\"\u003e\n \u003cp\u003eItems CI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003eCFA/ESEM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 63px;\"\u003e\n \u003cp\u003eItems CCP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 88px;\"\u003e\n \u003cp\u003eCFA/ESEM\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 61px;\"\u003e\n \u003cp\u003eto1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.657/.497\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003einp1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.758/.541\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003eci1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.853/.668\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003eccp1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.845/.773\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 61px;\"\u003e\n \u003cp\u003eto2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.829/.767\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003einp2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.840/.467\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003eci2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.870/.807\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003eccp2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.849/.786\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 61px;\"\u003e\n \u003cp\u003eto3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.866/.911\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003einp3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.825/.600\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003eci3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.902/.892\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003eccp3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.851/.846\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 61px;\"\u003e\n \u003cp\u003eto4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.861/.932\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003einp4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.708/.809\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003eci4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.889/.962\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003eccp4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.829/.824\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 61px;\"\u003e\n \u003cp\u003eto5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.678/.565\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003einp5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.750/.827\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003eci5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.747/.674\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003eccp5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.863/.883\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 61px;\"\u003e\n \u003cp\u003eto6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.822/.775\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003einp6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.745/.794\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003eci6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.826/.741\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003eccp6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.805/.752\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 61px;\"\u003e\n \u003cp\u003eMean\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e.786/.741\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 63px;\"\u003e\n \u003cp\u003eMean\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e.771/.673\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 63px;\"\u003e\n \u003cp\u003eMean\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e.848/0.791\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 63px;\"\u003e\n \u003cp\u003eMean\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e.840/.811\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 61px;\"\u003e\n \u003cp\u003e% decrease\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e-5.644\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 63px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e-12.711\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 63px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e-6.743\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 63px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e-3.530\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"8\" valign=\"top\" style=\"width: 601px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eGrand Mean\u003c/strong\u003e \u003cstrong\u003e.811 /.754 \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; % decrease \u0026nbsp;-7.063\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"8\" valign=\"top\" style=\"width: 601px;\"\u003e\n \u003cp\u003eFactor correlations \u0026nbsp; \u0026nbsp; CFA/ESEM\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 61px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003eTO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 63px;\"\u003e\n \u003cp\u003eDiff. %\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003eINP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 63px;\"\u003e\n \u003cp\u003eDiff. %\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003eCI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 63px;\"\u003e\n \u003cp\u003eDiff. %\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003eCCP\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 61px;\"\u003e\n \u003cp\u003eTO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 63px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 63px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 63px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 61px;\"\u003e\n \u003cp\u003eINP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.795/.728\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003e-8.428\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 63px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 63px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 61px;\"\u003e\n \u003cp\u003eCI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.751/.719\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003e-4.261\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.764/.661\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003e-13.482\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 63px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 61px;\"\u003e\n \u003cp\u003eCCP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.749/.728\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003e-2.804\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.808/.749\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003e-7.302\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e.784/.751\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003e-4.209\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 149px;\"\u003e\n \u003cp\u003eMean decrease %\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e-5.164\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e-10.392\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 63px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e-4.209\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 88px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"8\" valign=\"top\" style=\"width: 601px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eGrand Mean decrease %\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e-6.588\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eIn the ESEM model, the mean factor loadings across the four dimensions of the Digital Skills Indicator (DSI) ranged from .673 to .811, suggesting well-defined latent constructs. When treating variables as continuous, both CFA and ESEM models demonstrated strong composite reliability, with values ranging from .898 to .939 for CFA and from .838 to .920 for ESEM, well above the recommended threshold of .70 (see Table 4).\u003c/p\u003e\n\u003cp\u003eThese results indicate excellent internal consistency, suggesting that each set of items reliably measures its corresponding latent factor. In the ESEM model, two sub-constructs, Communication and Interaction (CI) and Content Creation and Production (CCP), demonstrated particularly strong internal cohesion, with composite reliability (CR) values of .912 and .920, respectively.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 4. Results of Composite Reliability, Convergent and Discriminant validity for CFA and ESEM models with continuous data.\u003c/p\u003e\n\u003cdiv align=\"\"\u003e\n \u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"422\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003eCR\u003c/p\u003e\n \u003cp\u003e(\u0026gt; .7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003eAVE\u003c/p\u003e\n \u003cp\u003e(\u0026gt; .5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eFactors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" valign=\"top\"\u003e\n \u003cp\u003eCFA (continuous data)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eTO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eINP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCCP\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.908\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.624\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eTO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.790*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.898\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.597\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eINP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.795\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.772*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.939\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.721\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.751\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.764\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.849*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.935\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.707\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCCP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.749\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.808\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.784\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.841*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"7\" valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003eCR\u003c/p\u003e\n \u003cp\u003e(\u0026gt; .7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003eAVE\u003c/p\u003e\n \u003cp\u003e(\u0026gt; .5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eFactors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" valign=\"top\"\u003e\n \u003cp\u003eESEM (continuous data)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eTO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eINP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCCP\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.886\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.576\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eTO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.759*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.838\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.473\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eINP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.728\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.688*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.912\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.637\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.719\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.661\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.798*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.920\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.659\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCCP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.728\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.749\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.751\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e.812*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"7\" valign=\"top\" style=\"width: 422px;\"\u003e\n \u003cp\u003e\u003cem\u003eNote\u003cstrong\u003e.\u0026nbsp;\u003c/strong\u003eCR-Composite Reliability; AVE-\u003c/em\u003e\u003cem\u003e\u0026nbsp;\u003c/em\u003e\u003cem\u003eAverage Variance Extracted;\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n \u003cp\u003e\u003cem\u003e* Square Root of AVE\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eRegarding construct validity, both CFA and ESEM models showed satisfactory convergent and discriminant validity. For the CFA model, all Average Variance Extracted (AVE) values exceeded the recommended threshold of .50, indicating that over half of the variance in each item was explained by its respective factor. In the ESEM model, all AVE values also exceeded .50, except for the Information Navigation and Processing (INP) factor, which had an AVE of .473, slightly below the ideal cutoff, but still within an acceptable range depending on the context.\u003c/p\u003e\n\u003cp\u003eTo assess discriminant validity, the square root of each AVE was compared to the inter-factor correlations. According to Fornell and Larcker (1981), discriminant validity is supported when the square root of a factor\u0026rsquo;s AVE is greater than its highest correlation with any other factor. This criterion was met for all sub-dimensions except INP, which showed overlapping variance with Technical and Operational (TO) and Content Creation and Production (CCP) in both CFA and ESEM models, indicating potential discriminant validity concerns for this sub-dimension.\u003c/p\u003e\n\u003cp\u003eWhen accounting for the categorical nature of the indicators, the four latent dimensions of the measurement scale demonstrated strong internal consistency in both the CFA and ESEM models. Composite reliability (CR) values ranged from .924 to .953 for CFA and from .886 to .934 for ESEM, all exceeding the recommended threshold of .70 (see Table 5). Additionally, all AVEs\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003efor the four sub-dimensions in both models were above .50, supporting the presence of convergent validity.\u003c/p\u003e\n\u003cp\u003eDiscriminant validity was confirmed for both the CFA and ESEM models using the Fornell-Larcker criterion. This finding supports the conceptual and statistical distinctiveness of the four latent sub-constructs, lending credibility to the measurement models. A minor exception was observed in the ESEM model for the Information Navigation and Processing (INP) factor, where the square root of its AVE (.756) was slightly lower than its correlation with the Content Creation and Production (CCP) factor (.771).\u003c/p\u003e\n\u003cp\u003eIn terms of measurement quality, all factor loadings exceeded the recommended threshold of .35 (see Table 2, Appendix), ranging from .467 (INP2) to .962 (CI4), and were statistically significant (\u003cem\u003ep\u003c/em\u003e \u0026lt; .001). These results indicate that the items are strongly associated with their respective latent constructs. Cross-loadings were minimal (\u0026lt; .30), with most below .10, supporting excellent discriminant validity. According to van Zyl and ten Klooster (2022), cross-loadings below .50 are acceptable in ESEM models, further confirming the robustness of the factor structure. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 5. Results of Composite Reliability, Convergent and Discriminant validity for CFA and ESEM models with categorical data\u003c/p\u003e\n\u003cdiv align=\"\"\u003e\n \u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"422\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003eCR\u003c/p\u003e\n \u003cp\u003e(\u0026gt; .7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003eAVE\u003c/p\u003e\n \u003cp\u003e(\u0026gt; .5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eFactors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" valign=\"top\"\u003e\n \u003cp\u003eCFA (categorical data)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eTO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eINP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCCP\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.932\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.697\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eTO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.835*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.924\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.672\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eINP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.787\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.820*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.953\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.771\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.753\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.771\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.878*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.950\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.759\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCCP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.752\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.809\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.804\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.871*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"7\" valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003eCR\u003c/p\u003e\n \u003cp\u003e(\u0026gt; .7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003eAVE\u003c/p\u003e\n \u003cp\u003e(\u0026gt; .5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"2\" valign=\"top\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eFactors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" valign=\"top\"\u003e\n \u003cp\u003eESEM (categorical data)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eTO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eINP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCCP\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.918\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.657\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eTO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.811*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.886\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.572\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eINP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.738\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.756*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.923\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.670\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.719\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.683\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.818*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.934\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.705\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCCP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.730\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.771\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.762\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\"\u003e\n \u003cp\u003e.840*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"7\" valign=\"bottom\"\u003e\n \u003cp\u003e\u003cem\u003eNote. CR-Composite Reliability; AVE-\u003c/em\u003e\u003cem\u003e\u0026nbsp;Average Variance Extracted;\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n \u003cp\u003e\u003cem\u003e* Square Root of AVE\u003c/em\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eAdditionally, the R-squared values for the scale items ranged from .426 to .822, reflecting a healthy proportion of explained variance. These values suggest that the underlying construct, Digital Skills Indicator (DSI), accounts for a substantial portion of each item\u0026apos;s variance, reinforcing the model\u0026rsquo;s strong psychometric properties and construct validity. All item uniqueness values fell between .10 and .90, indicating that each item contributed meaningfully to the factor structure without being redundant or overly error-prone.\u003c/p\u003e\n\u003cp\u003eDuring the model comparison phase, the bifactor-CFA model (M3) also outperformed the standard CFA model (M1). However, the primary focus was on evaluating M4 against M2. In comparing the bifactor-ESEM (M4) and ESEM (M2) models, all three information criteria (AIC, BIC, and aBIC) consistently favored the more complex M4 model, with lower values. Notably, AIC and aBIC favored the more complex models across all comparisons, while BIC tended to support the more parsimonious alternatives. Although M4 demonstrated marginally superior global fit indices, the differences (see Table 2) remained within conservative thresholds, indicating statistical equivalence. Given this, model parsimony was prioritized. Importantly, the ESEM model exhibited very low cross-loadings (all \u0026lt; 0.10), suggesting limited support for a dominant general factor, a core assumption of bifactor models. Therefore, its added complexity lacked sufficient theoretical and empirical justification. These findings support the selection of the ESEM model (M2) as the more appropriate and parsimonious solution (see Figure 2).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eA critical aspect of validating the short DSI scale involved testing for \u003cem\u003emeasurement invariance\u0026nbsp;\u003c/em\u003eacross gender to determine whether the instrument functions equivalently for male and female respondents. Without such invariance, comparisons of digital skills across gender could be biased or invalid. Measurement invariance is essential for ensuring the theoretical validity of a construct across groups (van de Vijver \u0026amp; Tanzer, 2004).\u003c/p\u003e\n\u003cp\u003eTo assess this, a series of increasingly restrictive invariance tests were conducted following established procedures (Meredith, 1993; Millsap, 2011). These tests were applied to the retained ESEM model and included six levels:\u003c/p\u003e\n\u003col start=\"1\" type=\"1\"\u003e\n \u003cli\u003e\u003cem\u003eConfigural invariance\u003c/em\u003e (same factor structure across groups)\u003c/li\u003e\n \u003cli\u003e\u003cem\u003eMetric invariance\u003c/em\u003e (equal factor loadings)\u003c/li\u003e\n \u003cli\u003e\u003cem\u003eScalar invariance\u003c/em\u003e (equal intercepts)\u003c/li\u003e\n \u003cli\u003e\u003cem\u003eStrict invariance\u003c/em\u003e (equal residuals)\u003c/li\u003e\n \u003cli\u003e\u003cem\u003eLatent variance\u0026ndash;covariance invariance\u003c/em\u003e\u003c/li\u003e\n \u003cli\u003e\u003cem\u003eLatent mean invariance\u003c/em\u003e (enabling valid group comparisons of latent means).\u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003eAs shown in Table 6, all models demonstrated acceptable to good fit across gender groups, supporting the conclusion that the DSI scale operates equivalently for male and female students.\u003c/p\u003e\n\u003cp\u003eTable 6. Measurement invariance sequential models\u0026rsquo; fit indices and differences\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"592\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 124px;\"\u003e\n \u003cp\u003eModel\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e\u0026chi;\u0026sup2; (df)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 35px;\"\u003e\n \u003cp\u003eCFI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 30px;\"\u003e\n \u003cp\u003eTLI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 74px;\"\u003e\n \u003cp\u003eRMSEA\u003c/p\u003e\n \u003cp\u003e90%CI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48px;\"\u003e\n \u003cp\u003eSRMR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 52px;\"\u003e\n \u003cp\u003e\u0026Delta;CFI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 51px;\"\u003e\n \u003cp\u003e\u0026Delta;TLI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 63px;\"\u003e\n \u003cp\u003e\u0026Delta;RMSEA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 58px;\"\u003e\n \u003cp\u003e\u0026Delta;SRMR\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 124px;\"\u003e\n \u003cp\u003eConfigural invariance\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e814.116\u003cbr\u003e\u0026nbsp;(372)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 35px;\"\u003e\n \u003cp\u003e.937\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 30px;\"\u003e\n \u003cp\u003e.906\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 74px;\"\u003e\n \u003cp\u003e.063\u003c/p\u003e\n \u003cp\u003e(.057 .069)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48px;\"\u003e\n \u003cp\u003e.027\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 52px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 51px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 63px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 58px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 124px;\"\u003e\n \u003cp\u003eWeak (\u0026lambda;) invariance\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e824.431\u003cbr\u003e\u0026nbsp;(452)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 35px;\"\u003e\n \u003cp\u003e.947\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 30px;\"\u003e\n \u003cp\u003e.935\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 74px;\"\u003e\n \u003cp\u003e.053\u003c/p\u003e\n \u003cp\u003e(.047 .058)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48px;\"\u003e\n \u003cp\u003e.033\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 52px;\"\u003e\n \u003cp\u003e.010\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 51px;\"\u003e\n \u003cp\u003e.029\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 63px;\"\u003e\n \u003cp\u003e-.010\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 58px;\"\u003e\n \u003cp\u003e.006\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 124px;\"\u003e\n \u003cp\u003eStrong (\u0026lambda;, \u0026nu;) invariance\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e873.182\u003cbr\u003e\u0026nbsp;(472)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 35px;\"\u003e\n \u003cp\u003e.942\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 30px;\"\u003e\n \u003cp\u003e.933\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 74px;\"\u003e\n \u003cp\u003e.053\u003c/p\u003e\n \u003cp\u003e(.048 .059)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48px;\"\u003e\n \u003cp\u003e.034\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 52px;\"\u003e\n \u003cp\u003e-.005\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 51px;\"\u003e\n \u003cp\u003e-.002\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 63px;\"\u003e\n \u003cp\u003e.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 58px;\"\u003e\n \u003cp\u003e.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 124px;\"\u003e\n \u003cp\u003eStrict (\u0026lambda;, \u0026nu;, \u0026delta;) invariance\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e923.983\u003cbr\u003e\u0026nbsp;(496)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 35px;\"\u003e\n \u003cp\u003e.939\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 30px;\"\u003e\n \u003cp\u003e.932\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 74px;\"\u003e\n \u003cp\u003e.054\u003c/p\u003e\n \u003cp\u003e(.048 .059)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48px;\"\u003e\n \u003cp\u003e.045\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 52px;\"\u003e\n \u003cp\u003e-.003\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 51px;\"\u003e\n \u003cp\u003e-.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 63px;\"\u003e\n \u003cp\u003e.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 58px;\"\u003e\n \u003cp\u003e.011\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 124px;\"\u003e\n \u003cp\u003eLatent var-cov (\u0026lambda;, \u0026nu;, \u0026delta;, \u0026xi;/\u0026phi;) invariance\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e935.828\u003cbr\u003e\u0026nbsp;(506)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 35px;\"\u003e\n \u003cp\u003e.938\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 30px;\"\u003e\n \u003cp\u003e.933\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 74px;\"\u003e\n \u003cp\u003e.053\u003c/p\u003e\n \u003cp\u003e(.048 .059)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48px;\"\u003e\n \u003cp\u003e.051\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 52px;\"\u003e\n \u003cp\u003e-.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 51px;\"\u003e\n \u003cp\u003e.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 63px;\"\u003e\n \u003cp\u003e-.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 58px;\"\u003e\n \u003cp\u003e.006\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 124px;\"\u003e\n \u003cp\u003eLatent mean (\u0026lambda;, \u0026nu;, \u0026delta;, \u0026xi;/\u0026phi;, \u0026eta;) invariance\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 57px;\"\u003e\n \u003cp\u003e942.929\u003cbr\u003e\u0026nbsp;(510)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 35px;\"\u003e\n \u003cp\u003e.938\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 30px;\"\u003e\n \u003cp\u003e.933\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 74px;\"\u003e\n \u003cp\u003e.053\u003c/p\u003e\n \u003cp\u003e(.048 .059)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 48px;\"\u003e\n \u003cp\u003e.053\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 52px;\"\u003e\n \u003cp\u003e.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 51px;\"\u003e\n \u003cp\u003e.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 63px;\"\u003e\n \u003cp\u003e.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 58px;\"\u003e\n \u003cp\u003e.002\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eTo evaluate each level of measurement invariance, the Satorra\u0026ndash;Bentler scaled chi-square difference test (Satorra \u0026amp; Bentler, 2010) was used to compare nested models. This test is considered more robust than the traditional chi-square difference test, particularly under conditions of non-normality. As shown in Table 3 (Appendix), the results supported metric (weak) invariance, but not scalar (strong) or strict invariance. According to Byrne et al. (1989), establishing metric invariance is sufficient for comparing structural relationships across groups, such as factor loadings and covariances.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAlthough the Satorra\u0026ndash;Bentler (S-B) scaled chi-square difference test adjusts for non-normality, it is not without limitations. As noted by Dimitrov (2010) and Ma\u0026iuml;ano et al. (2015), the test remains sensitive to small sample sizes, which can compromise the reliability of the scaling correction and increase the risk of Type I errors. Additionally, the computation of the scaling correction factor (SCF) varies across statistical software (e.g., LISREL, EQS, Mplus), and incorrect application may result in inadmissible test statistics, such as negative values or incorrect degrees of freedom (Bryant \u0026amp; Satorra, 2012). For these reasons, researchers are encouraged to supplement the S-B test with alternative goodness-of-fit indices (GFIs), which are generally less sensitive to sample size, model complexity, and baseline model fit. According to Chen (2007), for sample sizes greater than 300, the following thresholds indicate acceptable levels of invariance: Metric (weak) invariance: \u0026Delta;CFI \u0026le; .010, \u0026Delta;RMSEA \u0026le; \u0026ndash;.015, \u0026Delta;SRMR \u0026le; .030; Scalar and strict invariance: \u0026Delta;CFI \u0026le; .010, \u0026Delta;RMSEA \u0026le; \u0026ndash;.015, \u0026Delta;SRMR \u0026le; .010. Cheung and Rensvold (2002) similarly recommend \u0026Delta;CFI \u0026le; .010 as a criterion for supporting invariance.\u003c/p\u003e\n\u003cp\u003eIn the metric invariance model (see Table 6), item loadings were constrained to be equal across gender. The model showed good fit: \u0026chi;\u0026sup2; (372) = 824.431, CFI = .947, TLI = .935, RMSEA = .063 [90% CI: .057 .069], SRMR = .027. The S-B scaled chi-square difference test comparing the metric and configural models was non-significant: \u0026Delta;\u0026chi;\u0026sup2; = 56.15, \u0026Delta;df = 82, \u003cem\u003ep\u003c/em\u003e = .98, supporting metric invariance. Additionally, changes in fit indices (\u0026Delta;CFI = .010; \u0026Delta;TLI = .029; \u0026Delta;RMSEA = \u0026ndash;.010; \u0026Delta;SRMR = .006) were within acceptable thresholds, allowing valid comparisons of factor loadings across gender.\u003c/p\u003e\n\u003cp\u003eFor scalar invariance, both item loadings and intercepts were constrained. The model fit remained good: \u0026chi;\u0026sup2; (472) = 873.182, CFI = .942, TLI = .933, RMSEA = .053 [90% CI: .048 .059], SRMR = .034. However, the S-B test indicated a significant difference from the metric model: \u0026Delta;\u0026chi;\u0026sup2; = 55.99, \u0026Delta;df = 20, \u003cem\u003ep\u003c/em\u003e \u0026lt; .001, suggesting a lack of scalar invariance. In contrast, changes in fit indices (\u0026Delta;CFI = \u0026ndash;.005; \u0026Delta;TLI = \u0026ndash;.002; \u0026Delta;RMSEA = .000; \u0026Delta;SRMR = .001) supported scalar invariance, indicating that latent means could still be compared.\u003c/p\u003e\n\u003cp\u003eStrict invariance was tested by additionally constraining residual variances. The model fit remained good: \u0026chi;\u0026sup2; (496) = 923.983, CFI = .939, TLI = .932, RMSEA = .054 [90% CI: .048, .059], SRMR = .045. The S-B test showed a significant difference from the scalar model: \u0026Delta;\u0026chi;\u0026sup2; = 47.51, \u0026Delta;df = 24, \u003cem\u003ep\u003c/em\u003e \u0026lt; .005, suggesting a lack of strict invariance. However, changes in fit indices (\u0026Delta;CFI = \u0026ndash;.003; \u0026Delta;TLI = \u0026ndash;.001; \u0026Delta;RMSEA = .001; \u0026Delta;SRMR = .011) remained within acceptable limits, supporting strict invariance.\u003c/p\u003e\n\u003cp\u003eFinally, models testing latent variance\u0026ndash;covariance and latent mean invariance also demonstrated good fit. The S-B chi-square difference tests were non-significant (\u003cem\u003ep\u003c/em\u003e \u0026gt; .05), and changes in fit indices (\u0026Delta;CFI = \u0026ndash;.001/.000; \u0026Delta;TLI = .001/.000; \u0026Delta;RMSEA = \u0026ndash;.001/.000; \u0026Delta;SRMR = .006/.002) did not exceed recommended thresholds (\u0026Delta;CFI and \u0026Delta;TLI \u0026lt; .010; \u0026Delta;RMSEA \u0026lt; .015). These results support the conclusion that the ESEM factor structure is invariant across gender, enabling meaningful comparisons of structural relationships and latent means between male and female students.\u003c/p\u003e\n\u003cp\u003eIn the next phase of analysis, a hierarchical Exploratory Structural Equation Model (H-ESEM) was estimated to examine the presence of a higher-order digital skills factor underlying the four first-order dimensions. Unlike B-ESEM, which separates general and specific variance to distinguish broad from narrow constructs, H-ESEM allows observed variables to contribute indirectly to a higher-order factor through their respective first-order dimensions. This approach avoids imposing a direct general-factor influence on each item, offering greater flexibility in modeling the hierarchical structure of the construct. In H-ESEM, first-order factors remain connected to their observed indicators while also serving as indicators of a second-order latent factor. This structure captures both the distinctiveness of the four digital skills dimensions and their shared variance, without enforcing rigid hierarchical constraints. The model thus reflects the dynamic interplay between specific and general digital competencies.\u003c/p\u003e\n\u003cp\u003eTo estimate the H-ESEM model, non-standardized factor loadings from the original ESEM solution were used as starting values. One indicator per factor was constrained to retain its original loading, and all factor variances were fixed to one for model identification. The four first-order factors, Technical and Operational Skills, Information Navigation and Processing, Communication and Interaction, and Content Creation and Production, were specified as indicators of a higher-order digital skills factor (H-FACTOR). Table 7 presents the standardized path coefficients from each first-order factor to the higher-order construct.\u003c/p\u003e\n\u003cp\u003eTable 7. Standardized factor loadings from H-ESEM model with WLSMV estimator\u003c/p\u003e\n\u003cdiv align=\"\"\u003e\n \u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eH-FACTOR BY\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eEstimate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eS.E.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 61px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eEst./S.E.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 58px;\"\u003e\n \u003cp\u003eTwo-Tailed\u0026nbsp;P-Value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eTO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e.882\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e.016\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 61px;\"\u003e\n \u003cp\u003e56.383\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 58px;\"\u003e\n \u003cp\u003e.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eINP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e.909\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e.016\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 61px;\"\u003e\n \u003cp\u003e56.215\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 58px;\"\u003e\n \u003cp\u003e.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eCI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e.818\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e.015\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 61px;\"\u003e\n \u003cp\u003e53.869\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 58px;\"\u003e\n \u003cp\u003e.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 84px;\"\u003e\n \u003cp\u003eCCP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e.831\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 54px;\"\u003e\n \u003cp\u003e.018\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 61px;\"\u003e\n \u003cp\u003e47.209\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 58px;\"\u003e\n \u003cp\u003e.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eAll standardized loadings from the first-order factors to the higher-order digital skills factor (H-FACTOR) were high (all \u0026gt; .80) and statistically significant (\u003cem\u003ep\u003c/em\u003e \u0026lt; .001), indicating that the general digital skills construct strongly explains variance in each of the four sub-dimensions. This supports the presence of a hierarchical structure, where a broad digital skills factor underlies the more specific dimensions.\u003c/p\u003e\n\u003cp\u003eAmong the sub-dimensions, Information Navigation and Processing (INP) had the highest loading (.909), suggesting it is most strongly associated with the general factor. Communication and Interaction (CI) had the lowest loading (.818), though still well within the range considered strong. These results confirm that the H-FACTOR is a robust overarching construct, with each sub-dimension contributing meaningfully to it.\u003c/p\u003e\n\u003cp\u003eThe proportion of variance explained by the H-FACTOR was substantial across all first-order factors, ranging from 66.8% for CI to 82.6% for INP. Additionally, all four latent dimensions were highly inter-correlated and strongly associated with the higher-order factor (see Table 4, Appendix), further supporting the hierarchical model. These findings suggest that while the dimensions are conceptually distinct, they are closely related and collectively represent a unified construct of digital skills.\u003c/p\u003e\n\u003cp\u003eTo address the complexity of the measurement model while examining structural relationships with external predictors, we adopted the ESEM-within-CFA (EWC) approach. This method mitigates the convergence issues often encountered in traditional ESEM models when applied within complex structural frameworks. EWC enables the estimation of cross-loadings while preserving the structural clarity and interpretability characteristic of CFA. It is particularly well-suited for models incorporating predictors or covariates (Morin \u0026amp; Asparouhov, 2018; Marsh \u0026amp; Alamer, 2024). In our analysis, the EWC model was estimated with three external predictors, \u003cem\u003eprocrastination\u003c/em\u003e, \u003cem\u003eself-esteem\u003c/em\u003e, and \u003cem\u003etechnology affinity\u003c/em\u003e, targeting their effects on the four latent dimensions. This approach enhances model stability and interpretability by allowing predictors to exert more controlled influence over the factor structure. For comparative purposes, Table 5 (Appendix) presents traditional fit indices for both the ESEM and EWC models with predictors, estimated using two robust estimators.\u003c/p\u003e\n\u003cp\u003eThe ESEM model showed excellent fit with the WLSMV estimator (CFI, TLI, SRMR), though RMSEA was marginally acceptable. With the MLR estimator, RMSEA indicated excellent fit, while other indices remained within good to excellent range. The EWC model with predictors showed acceptable to good fit using MLR, and good to excellent fit with WLSMV, including a notable RMSEA improvement.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003ePredictor Effects on EWC Dimensions\u003c/em\u003e\u003cbr\u003e\u0026nbsp;Self-esteem (SE) emerged as the only significant positive predictor (p \u0026lt; .001) of the TO and CI dimensions across both estimators and under conditions of fixed and free factor variances (see Table 8). Procrastination (PR) and Affinity for Technology (AF) were not significant predictors of TO. However, procrastination had a significant negative effect on INP (p \u0026lt; .05), while self-esteem had a significant positive effect on the same dimension (p \u0026lt; .05). Affinity for Technology was a marginally significant predictor of INP, with a stronger effect observed under the WLSMV estimator (Figure 1, Appendix).\u003c/p\u003e\n\u003cp\u003eTable 8. The impact of three predictors on four latent factors of EWC with fixed and free factor variances for two different robust estimators\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 45px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMLR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 45px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eWLSMV\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 22px;\"\u003e\n \u003cp\u003eFixed variance\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 22px;\"\u003e\n \u003cp\u003eFree variance\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 22px;\"\u003e\n \u003cp\u003eFixed variance\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 22px;\"\u003e\n \u003cp\u003eFree variance\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003eFactors\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003eEstimates\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003ep-value\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003eEstimates\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003ep-value\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003eEstimates\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003ep-value\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003eEstimates\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003ep-value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTO\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003eON\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003ePR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.736\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.758\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.752\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.021\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.769\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003eSE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.610\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.613\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.626\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.623\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003eAF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.114\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.456\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.121\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.433\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.135\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.374\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.134\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.377\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eINP\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003eON\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003ePR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.159\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.031\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.160\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.031\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.179\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.005\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.182\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.005\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003eSE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.343\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.012\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.337\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.014\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.344\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.005\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.330\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.007\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003eAF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.301\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.057\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.306\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.053\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.302\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.044\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.315\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.036\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003eON\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003ePR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.097\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.123\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.096\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.129\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.090\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.194\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.088\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.201\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003eSE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.659\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.662\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.723\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.000\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.725\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003eAF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.103\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.478\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.108\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.454\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.176\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.255\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.181\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.242\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCCP\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003eON\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003ePR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.145\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.036\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.145\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.037\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.155\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.020\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e-.155\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.020\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003eSE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.381\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.002\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.378\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.002\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.397\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.391\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003eAF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.241\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.100\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.243\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.097\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.233\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.120\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 12px;\"\u003e\n \u003cp\u003e.239\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 10px;\"\u003e\n \u003cp\u003e.111\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eProcrastination (PR) was a significant negative predictor of CCP (p \u0026lt; .05), while self-esteem (SE) had a significant positive effect (p \u0026lt; .005). Affinity for Technology (AF) was not a significant predictor (p \u0026gt; .05). The four EWC dimensions exhibited distinct prediction patterns, supporting their conceptual distinctiveness. This is further reinforced by moderate to high inter-factor correlations (Table 6, Appendix), none exceeding .85, indicating acceptable discriminant validity and suggesting the factors capture unique constructs. The results indicate that the four EWC dimensions are related yet distinct, aligning with expectations for a multidimensional construct such as digital skills.\u003c/p\u003e"},{"header":"4. Discussion","content":"\u003cp\u003e\u003cstrong\u003eFactor Analysis (CFA)\u003c/strong\u003e and \u003cstrong\u003ebifactor models\u003c/strong\u003e, ESEM demonstrated superior model fit, greater parsimony, and stronger theoretical alignment, all while preserving the complexity inherent in multidimensional constructs.\u003c/p\u003e\n\u003cp\u003eThese findings corroborate prior research by \u003cstrong\u003eMorin et al. (2020)\u003c/strong\u003e and \u003cstrong\u003eGegenfurtner (2022)\u003c/strong\u003e, who emphasized the advantages of ESEM in modeling complex psychological constructs. Furthermore, this study extends the work of \u003cstrong\u003eVuorikari et al. (2022)\u003c/strong\u003e by offering a rigorous empirical implementation of the recommendations proposed in \u003cstrong\u003eVuorikari et al. (2025)\u003c/strong\u003e, which explicitly advocate for the use of ESEM to more accurately capture the overlapping dimensions of digital competence. As such, this research contributes to the growing body of evidence supporting the adoption of flexible, multidimensional modeling approaches in digital skills assessment.\u003c/p\u003e\n\u003cp\u003eThe ESEM model demonstrated strong psychometric properties, including high internal consistency, satisfactory convergent and discriminant validity, and minimal cross-loadings, indicating a well-defined factorial structure. The hierarchical ESEM (H-ESEM) further confirmed the presence of a higher-order digital skills factor, with all four sub-dimensions, Technical and Operational Skills (TO), Information Navigation and Processing (INP), Communication and Interaction (CI), and Content Creation and Production (CCP), loading significantly onto the general factor. Although it was not the primary objective of this study, we recognize the potential value of further exploring the multidimensional and hierarchical structure of digital skills through the application of a bifactor-ESEM model (Howard et al., 2016; Alamer, 2022). Such an approach would offer complementary insights to those obtained from the higher-order ESEM (H-ESEM) model, contributing to a more nuanced and accurate understanding of the construct. In particular, the use of ancillary bifactor indices (Rodriguez et al., 2016), such as the Explained Common Variance (ECV), Omega Hierarchical (\u0026omega;H), and Percentage of Uncontaminated Correlations (PUC), could provide more definitive evidence regarding the degree of unidimensionality versus multidimensionality inherent in the digital skills construct. Such insights would not only enhance the theoretical understanding of digital competence but also inform the appropriate use of total versus subscale scores in applied settings within broader Structural Equation Modeling (SEM) frameworks, especially in studies aiming to examine predictors or outcomes of digital competence.\u003c/p\u003e\n\u003cp\u003eTo validate the robustness of the measurement and structural models, parameter estimates were compared using both MLR and WLSMV estimators. Results across estimators were largely consistent, with only minor variations in model fit indices and path coefficients. Model comparisons (e.g., M2 vs. M1) showed negligible changes in CFI, TLI, RMSEA, and SRMR, suggesting that model refinements did not yield substantial improvements in fit. This consistency supports the stability of the model regardless of estimation method, revealing that parameter estimates or fit indices are not sensitive to estimator choice.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eESEM models demonstrated improved discriminant validity through reduced factor correlations, while measurement invariance testing across gender revealed metric invariance, allowing for valid comparisons of factor loadings between male and female students. Although scalar and strict invariance were not fully supported by chi-square difference tests, alternative fit indices suggested acceptable model stability, reinforcing the robustness of the instrument across demographic groups.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThis study offers a complementary contribution to existing approaches in digital skills assessment, particularly the Youth Digital Skills Indicator (yDSI) developed by Helsper et al. (2020). The yDSI combined performance-based tasks with composite scoring and was rigorously tested for measurement invariance across diverse demographic and cultural groups, demonstrating strong psychometric robustness. Its validated items have proven effective for large-scale, cross-national research. However, its reliance on composite scoring constrains its ability to fully capture the multidimensional and hierarchical structure of digital competence.\u003c/p\u003e\n\u003cp\u003eIn contrast, the present study employs latent factor scores derived from an Exploratory Structural Equation Modeling (ESEM) framework. This approach enables a more nuanced representation of digital skills by modeling both general and domain-specific competencies, while accounting for the interdependence among skill dimensions. The ESEM model demonstrated weak, strong, and strict measurement invariance across gender, thereby supporting valid comparisons of both latent means and observed scores.\u003c/p\u003e\n\u003cp\u003eThis level of psychometric rigor strengthens the empirical foundation of the digital skills construct and directly responds to Helsper et al.\u0026rsquo;s call for more precise and equitable assessment tools. By transitioning from manifest to latent modeling and establishing full measurement invariance, the current study enriches the yDSI framework and provides a robust platform for future cross-group and longitudinal analyses in digital competence research.\u003c/p\u003e\n\u003cp\u003eThe ESEM-within-CFA (EWC) model with external predictors provided further evidence of construct validity. Self-efficacy emerged as a consistent and significant positive predictor across all four digital skill dimensions, while academic procrastination negatively predicted INP and CCP. Affinity for technology showed marginal effects, suggesting that attitudinal variables may influence digital engagement differently.\u003c/p\u003e\n\u003cp\u003eAlthough the sample size in this study was determined through an a priori power analysis using Soper\u0026rsquo;s (2025) calculator\u0026mdash;based on anticipated effect sizes and a desired statistical power\u0026mdash;this approach, while methodologically sound, may not fully account for potential sources of model bias inherent in complex latent variable models. As such, we acknowledge the value of more advanced techniques for evaluating model adequacy. Specifically, although not the primary focus of this study, we contend that the use of Monte Carlo simulations (Muth\u0026eacute;n \u0026amp; Muth\u0026eacute;n, 2002) represents a more holistic and rigorous strategy. Monte Carlo methods enable researchers to assess estimation accuracy, model fit, and potential sources of bias under varied conditions by simulating data based on realistic assumptions. This approach facilitates the evaluation of factor loadings, standard errors, factor correlations, residual variances, confidence interval coverage rates, and statistical power. By identifying conditions under which model estimates may become biased or unstable, Monte Carlo simulations can enhance the validity, reliability, and generalizability of findings in structural equation modeling.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e4.1 Theoretical implications\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThis study provides robust evidence for a psychometrically sound instrument to assess students\u0026rsquo; digital skills, reinforcing the conceptualization of digital skills as a multidimensional construct. This structure aligns with contemporary frameworks (e.g., DigComp, ISTE\u003ca href=\"#_ftn1\" name=\"_ftnref1\" title=\"\"\u003e\u003c/a\u003e\u003csup\u003e3\u003c/sup\u003e) and advances them by providing empirical validation through robust psychometric modeling. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eOur findings demonstrate that Exploratory Structural Equation Modeling (ESEM) outperforms traditional CFA model in capturing the complexity of digital skills. ESEM\u0026rsquo;s allowance for cross-loadings enhances construct validity and discriminant precision, making it a preferred approach for future psychometric research in education and technology.\u003c/p\u003e\n\u003cp\u003eThe significant associations between digital skills and self-esteem, procrastination, and technology affinity highlight the interdisciplinary nature of digital competence. This positions digital skills not just as technical abilities but as psycho-social constructs, influenced by motivation, self-concept, and behavioral tendencies. These findings align with contemporary theories of digital literacy, which emphasize both technical proficiency and socio-cognitive engagement (Belshaw, 2012; Feola, 2016).\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e4.2 Practical Implications\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe findings of this study offer actionable insights for enhancing digital skills education through curriculum design, assessment strategies, and institutional policy. The validated hierarchical model of digital competence provides a robust foundation for evidence-based educational planning. Specifically, the results support the following implications:\u003c/p\u003e\n\u003cp\u003e\u003cem\u003ea. Curriculum Design\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe H-ESEM results provide strong empirical support for a hierarchically structured digital competence framework. These findings confirm and extend previous research emphasizing the foundational role of technical and information-processing skills in digital competence development (e.g., UNESCO, 2011; Vuorikari et al., 2022; Mattar et al., 2022). Curriculum designers are encouraged to leverage these insights to develop sequenced, evidence-based learning pathways that prioritize foundational skills while scaffolding the development of communication and content creation capabilities, ensuring a coherent progression from foundational to higher-order competencies.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eb. Assessment and Intervention\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe validated instrument enables the diagnosis of specific skill gaps, monitoring of student progress over time, and early identification of learners at risk of digital exclusion. This supports the development of targeted interventions and personalized learning pathways.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003ec. Institutional Policy and Practice\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe use of psychometrically sound assessment data can inform strategic decisions related to infrastructure investment, teacher training, and curriculum reform. Furthermore, the demonstrated measurement invariance across gender, promotes equity and inclusion, ensuring that digital competence assessments are fair and accessible to all learners. Educators can also be empowered through professional development initiatives that help them interpret digital skills data and apply insights to instructional planning.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFor educators, the instrument offers a practical and scalable tool for assessing digital literacy at the classroom, school, or system level. Its alignment with established frameworks (e.g., DigComp, UNESCO ICT-CFT\u003ca href=\"#_ftn2\" name=\"_ftnref2\" title=\"\"\u003e\u003c/a\u003e\u003csup\u003e4\u003c/sup\u003e) makes it suitable for integration into national assessments and policy-driven initiatives aimed at enhancing digital competence.\u003c/p\u003e"},{"header":"5. Conclusions","content":"\u003cp\u003eThis study provides robust empirical evidence for the validity and reliability of a short-form psychometric instrument designed to assess students\u0026rsquo; digital skills. The ESEM model, supported by hierarchical and predictive analyses, offers a statistically sound and theoretically grounded framework for measuring digital competence in higher education contexts. The validated instrument is suitable for large-scale assessments, policy evaluations, and integration into broader structural models. It can inform curriculum development, teacher training, and digital inclusion strategies, particularly in under-resourced educational systems like Albania\u0026rsquo;s, where digital transformation is still in its early stages.\u003c/p\u003e\u003cdiv id=\"Sec20\" class=\"Section2\"\u003e\u003ch2\u003e5.1. Limitations\u003c/h2\u003e\u003cp\u003eDespite the strengths of the study, several limitations should be acknowledged:\u003c/p\u003e\u003cp\u003e(1) \u003cem\u003eSample Specificity\u003c/em\u003e: The data were collected from a single university in Albania, which may limit the generalizability of the findings to other educational contexts or countries.\u003c/p\u003e\u003cp\u003e(2) \u003cem\u003eSelf-Report Bias\u003c/em\u003e: The use of self-reported measures, although carefully designed to minimize bias, may still be subject to social desirability and overestimation of skills.\u003c/p\u003e\u003cp\u003e(3) \u003cem\u003eCross-Sectional Design\u003c/em\u003e: The study\u0026rsquo;s cross-sectional nature precludes causal inferences and limits the ability to assess changes in digital skills over time.\u003c/p\u003e\u003cp\u003e(4) \u003cem\u003eLimited External Predictors\u003c/em\u003e: While three predictors were included, other relevant variables (e.g., socioeconomic status, prior ICT training) were not examined.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec21\" class=\"Section2\"\u003e\u003ch2\u003e5.2. Future Research Directions\u003c/h2\u003e\u003cp\u003eTo build on the current findings, future research should consider the following directions:\u003c/p\u003e\u003cp\u003e(1) \u003cem\u003eCross-Cultural Validation\u003c/em\u003e: Replicating the study in diverse educational and cultural settings to test the instrument\u0026rsquo;s cross-national applicability and measurement invariance.\u003c/p\u003e\u003cp\u003e(2) \u003cem\u003eLongitudinal Studies\u003c/em\u003e: Implementing longitudinal designs to track the development of digital skills over time and assess the impact of educational interventions.\u003c/p\u003e\u003cp\u003e(3) \u003cem\u003ePerformance-Based Measures\u003c/em\u003e: Combining self-report instruments with performance-based assessments to triangulate findings and enhance validity.\u003c/p\u003e\u003cp\u003e(4) \u003cem\u003eExpanded Predictive Models\u003c/em\u003e: Including a broader range of exogenous variables (e.g., digital access, motivation, institutional support) to better understand the determinants of digital competence.\u003c/p\u003e\u003cp\u003e(5) \u003cem\u003eThe assessment of critical digital skills\u003c/em\u003e, measured through knowledge-based items in addition to functional digital skills, can be significantly enhanced through the application of both the full B (ESEM) and Set B (ESEM) models. This dual-model approach allows for a more comprehensive evaluation of digital competence by capturing both foundational and higher-order cognitive dimensions.\u003c/p\u003e\u003cp\u003e(6) \u003cem\u003ePolicy Impact Evaluation\u003c/em\u003e: Using the instrument in policy evaluation frameworks to assess the effectiveness of national digital education strategies.\u003c/p\u003e\u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003eParticipant Consent Statement:\u003c/p\u003e\n\u003cp\u003eAll participants provided informed consent prior to their involvement in the study. The research protocol, including the consent procedure, was reviewed and approved by the relevant institutional ethics committee. Participation was voluntary, and data were collected anonymously to ensure confidentiality.\u003c/p\u003e\n"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAlamer, A. (2022). Exploratory structural equation modeling (ESEM) and bifactor ESEM for construct validation purposes: Guidelines and applied example. \u003cem\u003eResearch Methods in Applied Linguistics\u003c/em\u003e, 1, 100005. https://doi.org/10.1016/j.rmal.2022.100005\u003c/li\u003e\n\u003cli\u003eAla-Mutka, K. (2011). \u003cem\u003eMapping Digital Competence: Towards a Conceptual Understanding\u003c/em\u003e. Luxembourg: Publications Office of the European Union. JRC Technical Report.\u003c/li\u003e\n\u003cli\u003eAsparouhov T, Muth\u0026eacute;n B. Exploratory structural equation modeling. Struct Equat Model. (2009) 16:397\u0026ndash;438. doi: 10.1080/10705510903008204\u003c/li\u003e\n\u003cli\u003eAsparouhov, T., \u0026amp; Muth\u0026eacute;n, B. (2018). SRMR in Mplus. Unpublished manuscript. https://www.statmodel.com/download/SRMR2.pdf\u003c/li\u003e\n\u003cli\u003eBandalos, D. L. (2014). Relative performance of categorical diagonally weighted least squares and robust maximum likelihood estimation. \u003cem\u003eStructural Equation Modeling: A Multidisciplinary Journal, 21\u003c/em\u003e(1), 102\u0026ndash;116. https://doi.org/10.1080/10705511.2014.859510\u003c/li\u003e\n\u003cli\u003eBandalos, D. L. (2018). \u003cem\u003eMeasurement theory and applications for the social sciences\u003c/em\u003e. The Guilford Press.\u003c/li\u003e\n\u003cli\u003eBarbazan, D., Ben, Khaddouja., \u0026amp; Montes, C. (2021). La competencia digital docente en educaci\u0026oacute;n superior: estado del arte en Espa\u0026ntilde;a y Latinoam\u0026eacute;rica [Digital Competence for Teachers in Higher Education: State of the Art in Spain and Latin America]. Etic@net. Revista Cient\u0026iacute;fica Electr\u0026oacute;nica de Educaci\u0026oacute;n y Comunicaci\u0026oacute;n En La Sociedad Del Conocimiento, 21(2), 267-282. https://doi.org/10.30827/eticanet.v21i2.20837\u003c/li\u003e\n\u003cli\u003eBeauducel A, Herzberg PY. (2006). On the Performance of Maximum Likelihood Versus Means and Variance Adjusted Weighted Least Squares Estimation in CFA. Structural Equation Modeling 13(2): 186-203, DOI: 10.1207/s15328007sem1302_2.\u003c/li\u003e\n\u003cli\u003eBeierlein, C., Kemper, C. J., Kovaleva, A., \u0026amp; Rammstedt, B. (2013). Short Scale for Measuring General Self-efficacy Beliefs (ASKU). \u003cem\u003eMethods, Data, Analyses, 7\u003c/em\u003e(2), 251\u0026ndash;278. https://doi.org/10.12758/mda.2013.014\u003c/li\u003e\n\u003cli\u003eBelshaw, D. A. J. (2012). \u003cem\u003eWhat is digital literacy? A pragmatic investigation\u003c/em\u003e (Doctoral dissertation, Durham University). Durham E-Theses. https://etheses.dur.ac.uk/3446/\u003c/li\u003e\n\u003cli\u003eBryant, F. B., \u0026amp; Satorra, A. (2012).\u003cem\u003e \u003c/em\u003ePrinciples and practice of scaled difference chi-square testing\u003cem\u003e. \u003c/em\u003eStructural Equation Modeling: A Multidisciplinary Journal\u003cem\u003e, \u003c/em\u003e\u003cstrong\u003e19\u003c/strong\u003e(3), 372\u0026ndash;398\u003cem\u003e. \u003c/em\u003ehttps://doi.org/10.1080/10705511.2012.687671\u003c/li\u003e\n\u003cli\u003eByrne, B. M., Shavelson, R. J., \u0026amp; Muth\u0026eacute;n, B. (1989). Testing for equivalence of factor covariance and mean structures: The issue of partial measurement invariance. Psychological Bulletin, 105, 456-466. \u003c/li\u003e\n\u003cli\u003eCabezas-Gonz\u0026aacute;lez, M., Casillas-Mart\u0026iacute;n, S., \u0026amp; Garc\u0026iacute;a-Valc\u0026aacute;rcel Mu\u0026ntilde;oz-Repiso, A. (2023). Theoretical models explaining the level of digital competence in students. \u003cem\u003eComputers, 12\u003c/em\u003e(5), 100. https://doi.org/10.3390/computers12050100\u003c/li\u003e\n\u003cli\u003eCarretero, S., Vuorikari, R., \u0026amp; Punie, Y. (2017). DigComp 2.1: The Digital Competence Framework for Citizens with eight proficiency levels and examples of use. Publications Office of the European Union. EUR 28558 EN. https://doi.org/10.2760/38842\u003c/li\u003e\n\u003cli\u003eChen, F. F. (2007). Sensitivity of goodness of fit indexes to lack of measurement. Structural Equation Modeling, 14, 464\u0026ndash;504. http://dx.doi.org/10.1080/10705510701301834\u003c/li\u003e\n\u003cli\u003eCheung, G. W., \u0026amp; Rensvold, R. B. (2002). Evaluating goodness-of fit indexes for testing measurement invariance. Structural Equation Modeling, 9, 233\u0026ndash;255. http://dx.doi.org/10.1207/S15328007SEM0902 5\u003c/li\u003e\n\u003cli\u003eCortesi, S., Gasser, U., \u0026amp; Hasse, A. (2020). \u003cem\u003eYouth and digital citizenship+ (plus): Understanding skills for a digital world\u003c/em\u003e. Berkman Klein Center for Internet \u0026amp; Society, Harvard University. https://cyber.harvard.edu/publication/2020/youth-and-digital-citizenship-plus\u003c/li\u003e\n\u003cli\u003eDarling-Hammond, L., \u0026amp; Adamson, F. (2010). \u003cem\u003eBeyond basic skills: The role of performance assessment in achieving 21st century standards of learning\u003c/em\u003e. Stanford Center for Opportunity Policy in Education (SCOPE). https://edpolicyinca.org/publications/beyond-basic-skills[1](https://cyber.harvard.edu/people/scortesi)\u003c/li\u003e\n\u003cli\u003eDierendonck C (2023) Measuring the classroom level of the Dynamic Model of Educational Effectiveness through teacher self-report: development and validation of a new instrument. Front. Educ. 8:1281431. doi: 10.3389/feduc.2023.1281431\u003c/li\u003e\n\u003cli\u003eDimitrov, D. M. (2010). Testing for Factorial Invariance in the Context of Construct Validation. \u003cem\u003eMeasurement and Evaluation in Counseling and Development\u003c/em\u003e, \u003cem\u003e43\u003c/em\u003e(2), 121-149. https://doi.org/10.1177/0748175610373459 (Original work published 2010)\u003c/li\u003e\n\u003cli\u003eErwin, K., \u0026amp; Mohammed, S. (2022). Digital Literacy Skills Instruction and Increased Skills Proficiency. International Journal of Technology in Education and Science, 6(2), 323-332. https://doi.org/10.46328/ijtes.364.\u003c/li\u003e\n\u003cli\u003eEuropean Commission. (2020). \u003cem\u003eDigital Competence Framework for Citizens (DigComp 2.1): The Digital Competence Framework\u003c/em\u003e. Publications Office of the European Union. https://education.ec.europa.eu/focus-topics/digital-education/action-plan/Updating-the-European-Digital-Competence-Framework\u003c/li\u003e\n\u003cli\u003eEuropean Commission. (2021). Albania 2021 Report. Accompanying the document Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions 2021 Communication on EU Enlargement Policy. Strasbourg 19.10.2021 Retrieved from https://ec.europa.eu/neighbourhood-enlargement/albania-report-2021_en\u003c/li\u003e\n\u003cli\u003eFalma, F. O., \u0026amp; Putra, A. H. (2025). Correlation between academic self-efficacy and digital competence: An empirical study on first-year students. \u003cem\u003eJournal of Counseling, Education and Society, 6\u003c/em\u003e(1), 1\u0026ndash;9. https://jurnal.iicet.org/index.php/jces/article/download/5403/pdf\u003c/li\u003e\n\u003cli\u003eFan, C., \u0026amp; Wang, J. (2022). Development and validation of a questionnaire to measure digital skills of Chinese undergraduates. Sustainability, 14(6), 3539. https://doi.org/10.3390/su14063539\u003c/li\u003e\n\u003cli\u003eFayda-Kınık, F. Ş. (2023). \u003cem\u003eThe impact of digital competences on academic procrastination in higher education: A structural equation modeling approach\u003c/em\u003e. Pegem Journal of Education and Instruction, 13(3), 25\u0026ndash;35. https://files.eric.ed.gov/fulltext/EJ1387897.pdf\u003c/li\u003e\n\u003cli\u003eFeijao, C., Flanagan, I., van Stolk, C. and Gunashekar, S. 2021. The global digital skills gap.\u003c/li\u003e\n\u003cli\u003eFeola, E. I. (2016). Digital literacy and new technological perspectives. \u003cem\u003eUniversal Journal of Educational Research, 4\u003c/em\u003e(9), 2174\u0026ndash;2180. https://doi.org/10.13189/ujer.2016.040929\u003c/li\u003e\n\u003cli\u003eFerrari, A. (2013). DIGCOMP: A framework for developing and understanding digital competence in Europe. Publications Office.\u003c/li\u003e\n\u003cli\u003eFinney, S.J., \u0026amp; DiStefano, C. (2013). Non-normal and categorical data in structural equation modeling. In G.R. Hancock \u0026amp; R.O. Mueller (Eds.), Structural equation modeling: A second course, 2\u003csup\u003e-nd\u003c/sup\u003e Edition (pp. 439\u0026ndash;492). Charlotte, NC: Information Age Publishing.\u003c/li\u003e\n\u003cli\u003eFlora, D. B., \u0026amp; Curran, P. J. (2004). An empirical evaluation of alternative methods of estimation for confirmatory factor analysis with ordinal data. \u003cem\u003ePsychological Methods, 9\u003c/em\u003e(4), 466\u0026ndash;491. https://doi.org/10.1037/1082-989X.9.4.466 [2] (https://psycnet.apa.org/record/2014-05296-010)\u003c/li\u003e\n\u003cli\u003eFornell, C., and Larcker, D. F. (1981). Evaluating structural equation models with unobservable variables and measurement error. J. Mark. Res. 18, 39\u0026ndash;50. doi: 10.2307/3151312\u003c/li\u003e\n\u003cli\u003eFranke, T., Attig, C., and Wessel, D. (2019). A personal resource for technology interaction: Development and validation of the Affinity for Technology Interaction (ATI) scale. Int. J. Hum. Comput. Interact. 35, 456\u0026ndash;467. doi: 10.1080/10447318.2018.1456150\u003c/li\u003e\n\u003cli\u003eGallardo-Echenique, E. E., De Oliveira, J. M., Marqu\u0026eacute;s-Molias, L., Esteve-Mon, F., Wang, Y., \u0026amp; Baker, R. (2015). Digital competence in the knowledge society. MERLOT Journal of Online Learning and Teaching, 11(1), 1\u0026ndash;16.\u003c/li\u003e\n\u003cli\u003eGegenfurtner, A. (2022). Bifactor exploratory structural equation modeling: A meta-analytic review of model fit. \u003cem\u003eFrontiers in Psychology, 13\u003c/em\u003e, 1037111. https://doi.org/10.3389/fpsyg.2022.1037111\u003c/li\u003e\n\u003cli\u003eHaddon, L., Cino, D., Doyle, M. A., Livingstone, S., Mascheroni, G., \u0026amp; Stoilova, M. (2020). Children\u0026rsquo;s and young people\u0026rsquo;s digital skills: A systematic evidence review. Leuven, KU Leuven: ySKILLS. Available at: https://zenodo.org/record/4274654#.X-pMceSWysc\u003c/li\u003e\n\u003cli\u003eHair, J. F., Anderson, R. E., Babin, B. J., and Black, W. C. (2010). Multivariate data analysis: A global perspective, vol. 7. Upper Saddle River, NJ: Pearson\u003c/li\u003e\n\u003cli\u003eHelsper, E. J., Schneider, L. S., van Deursen, A. J. A. M., \u0026amp; van Laar, E. (2021). \u003cem\u003eThe youth Digital Skills Indicator: Report on the conceptualisation and development of the ySKILLS digital skills measure\u003c/em\u003e. University of Twente. https://research.utwente.nl/en/publications/the-youth-digital-skills-indicator-report-on-the-conceptualisation\u003c/li\u003e\n\u003cli\u003eHelsper, E. J., van Deursen, A. J. A. M., \u0026amp; Eynon, R. (2020). \u003cem\u003eDigital skills, digital inequalities and the digital divide: Theories and evidence\u003c/em\u003e. In J. A. G. M. van Dijk \u0026amp; A. J. A. M. van Deursen (Eds.), \u003cem\u003eDigital Skills: Unlocking the Information Society\u003c/em\u003e (pp. 21\u0026ndash;42). Springer. https://link.springer.com/chapter/10.1007/978-3-031-46929-9_3\u003c/li\u003e\n\u003cli\u003eHelsper, E.J., Schneider, L.S., van Deursen, A.J.A.M., \u0026amp; van Laar, E. (2020). The youth Digital Skills Indicator: Report on the conceptualisation and development of the ySKILLS digital skills measure. KU Leuven, Leuven: ySKILLS. \u003c/li\u003e\n\u003cli\u003eHoward, J. L., Gagn\u0026eacute;, M., Morin, A. J. S., Wang, Z. N., \u0026amp; Forest, J. (2016). Using bifactor exploratory structural equation modeling to test for a continuum structure of motivation. \u003cem\u003eJournal of Management\u003c/em\u003e, 1\u0026ndash;59. https://www.statmodel.com/download/Bifactor-ESEM.pdf\u003c/li\u003e\n\u003cli\u003eHu, L., \u0026amp; Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria vs new alternatives. Structural Equation Modeling, 6, 1\u0026ndash;55. http://dx.doi.org/10.1080/10705519909540118\u003c/li\u003e\n\u003cli\u003eIlom\u0026auml;ki L, Paavola S, Lakkala M, Kantosalo A. Digital competence\u0026mdash;An emergent boundary concept for policy and educational research. Education and Information Technologies. 2014; 21(3):655\u0026ndash;679.\u003c/li\u003e\n\u003cli\u003eInternational Telecommunication Union (ITU) (2018). Measuring the Information Society report, Volume 1. Geneva, Switzerland: ITU Publications. Available at: www.itu.int/en/ITU-D/Statistics/Documents/publications/misr2018/MISR-2018-Vol-1-E.pdf.\u003c/li\u003e\n\u003cli\u003eJavier-Aliaga, D., Silva Neyra, O. R., Calizaya-Milla, Y. E., \u0026amp; Saintila, J. (2024). Academic self-efficacy and digital competence in a sample of university students. \u003cem\u003eContemporary Educational Technology, 16\u003c/em\u003e(4), ep540. https://doi.org/10.30935/cedtech/15601\u003c/li\u003e\n\u003cli\u003eJokisch, C., Schlimbach, R., Strzebkowski, M., Hobert, S., Kurpejovic, H., \u0026amp; Schumann, M. (2025). \u003cem\u003eEmpirical insights on the impact of technology affinity on programming skill development through an adaptive game-based learning system\u003c/em\u003e. Proceedings of the 58th Hawaii International Conference on System Sciences. https://hdl.handle.net/10125/109435\u003c/li\u003e\n\u003cli\u003eJoreskog KG. A general approach to confirmatory maximum likelihood factor analysis. Psychometrika. (1969) 34:183\u0026ndash;202. doi: 10.1007/BF02289343\u003c/li\u003e\n\u003cli\u003eKaplan, D. (2009). Structural Equation Modeling: Foundations and Extensions (2nd Ed.). Newbury Park, CA: SAGE Publications. https://doi.org/10.4135/9781452226576\u003c/li\u003e\n\u003cli\u003eKayyali, M. (2024). Digital Literacy in Higher Education: Preparing Students for the Workforce of the Future. IJISC., 11(01): 53-73.\u003c/li\u003e\n\u003cli\u003eKryukova, N. I., Chistyakov, A. A., Shulga, T. I., Omarova, L. B., Tkachenko, T. V., Malakhovsky, A. K., et al. (2022). Adaptation of higher education students\u0026rsquo; digital skills survey to Russian universities. Eurasia J. Mathem. Sci. Technol. Educ. 18:em2183. doi: 10.29333/ejmste/12558\u003c/li\u003e\n\u003cli\u003eLaanpere, M. (2019). Recommendations on assessment tools for monitoring digital literacy within UNESCO\u0026rsquo;s digital literacy global framework. Montreal, Quebec, UNESCO Institute for Statistics, 2019. https://unesdoc.unesco.org/ark:/48223/pf0000366740\u003c/li\u003e\n\u003cli\u003eL\u0026oacute;pez, J. C. (2023). Competencias digitales en la educaci\u0026oacute;n superior. Horizontes Rev. Invest. Ciencias Educ. Superiorn 7, 1548\u0026ndash;1563. doi: 10.33996/revistahorizontes.v7i29.612\u003c/li\u003e\n\u003cli\u003eMa, H., Ismail, L. Bibliometric analysis and systematic review of digital competence in education. \u003cem\u003eHumanit Soc Sci Commun\u003c/em\u003e 12, 185 (2025). https://doi.org/10.1057/s41599-025-04401-1\u003c/li\u003e\n\u003cli\u003eMa\u0026iuml;ano, C., Morin, A. J. S., \u0026amp; Mascret, N. (2015). Psychometric properties of the short form of the physical self-description questionnaire in a French adolescent sample. Body Image, 12, 89\u0026ndash;97. https://doi.org/10.1016/j.bodyim.2014.10.005\u003c/li\u003e\n\u003cli\u003eMarsh HW, Guo J, Dicke T, Parker PD, Craven RG. Confirmatory factor analysis (CFA), exploratory structural equation modeling (ESEM), and set-ESEM: optimal balance between goodness of fit and parsimony. Multivariate Behav Res. (2020) 55:102\u0026ndash;19. doi: 10.1080/00273171.2019.1602503\u003c/li\u003e\n\u003cli\u003eMarsh HW, Morin AJ, Parker PD, Kaur G. Exploratory structural equation modeling: An integration of the best features of exploratory and confirmatory factor analysis. Annu Rev Clin Psychol. (2014) 10:85\u0026ndash;110. doi: 10.1146/annurev-clinpsy-032813-153700\u003c/li\u003e\n\u003cli\u003eMarsh HW, Muth\u0026eacute;n B, Asparouhov T, L\u0026uuml;dtke O, Robitzsch A, Morin AJ, et al. Exploratory structural equation modeling, integrating CFA and EFA: application to students\u0026rsquo; evaluations of university teaching. Struct Equ Model. (2009) 16:439\u0026ndash;76. doi: 10.1080/10705510903008220\u003c/li\u003e\n\u003cli\u003eMarsh, H. W. (2007). Application of confirmatory factor analysis and structural equation modeling in sport/exercise psychology. In G. Tenenbaum \u0026amp; R. C. Eklund (Eds.), \u003cem\u003eHandbook of on sport psychology\u003c/em\u003e (3rd ed., pp. 774 - 798). New York: Wiley. \u003c/li\u003e\n\u003cli\u003eMarsh, H. W., \u0026amp; Alamer, A. (2024). \u003cem\u003eWhen and how to use set‐exploratory structural equation modelling to test structural models: A tutorial using the R package lavaan\u003c/em\u003e. \u003cem\u003eBritish Journal of Mathematical and Statistical Psychology\u003c/em\u003e. Advance online publication. https://doi.org/10.1111/bmsp.12336\u003c/li\u003e\n\u003cli\u003eMarsh, H. W., Hau, K.-T., \u0026amp; Grayson, D. (2005). Goodness of fit in structural equation models. In A. Maydeu-Olivares \u0026amp; J. J. McArdle (Eds.), \u003cem\u003eContemporary psychometrics: A festschrift for Roderick P. McDonald\u003c/em\u003e (pp. 275\u0026ndash;340). Lawrence Erlbaum Associates Publishers\u003c/li\u003e\n\u003cli\u003eMarsh, H. W., Wen, Z., \u0026amp; Hau, K.-T. (2004). Structural equation models of latent interactions: Evaluation of alternative estimation strategies and indicator construction. \u003cem\u003ePsychological Methods, 9\u003c/em\u003e(3), 275\u0026ndash;300. https://doi.org/10.1037/1082-989X.9.3.275\u003c/li\u003e\n\u003cli\u003eMarsh, H.W., L\u0026uuml;dtke, O., Muth\u0026eacute;n, B., Asparouhov, T., Morin, A. J. S., Trautwein, U., \u0026amp; Nagengast, B. (2010). A new look at the big-five factor structure through exploratory structural equation modeling. Psychological Assessment, 22, 471-491.\u003c/li\u003e\n\u003cli\u003eMattar, J., Ramos, D. K., \u0026amp; Lucas, M. R. (2022). DigComp-based digital competence assessment tools: Literature review and instrument analysis. \u003cem\u003eEducation and Information Technologies, 27\u003c/em\u003e, 10843\u0026ndash;10867. https://doi.org/10.1007/s10639-022-11034-3\u003c/li\u003e\n\u003cli\u003eMeredith, W. (1993). Measurement invariance, factor analysis and factor invariance. Psychometrika, 58, 525-544.\u003c/li\u003e\n\u003cli\u003eMi\u0026ccedil;o, H., \u0026amp; Za\u0026ccedil;ellari, M. (2020). \u0026ldquo;Online learning as a means toward achieving an adaptable right to education in Albania\u0026rdquo;. Knowledge-International Journal, 2020, 42(2) 293\u0026ndash;298. ISSN online 2545-4439 https://ikm.mk/ojs/index.php/KIJ/article/view/4672/4459\u003c/li\u003e\n\u003cli\u003eMillsap, R. E., \u0026amp; Meredith, W. (2004, May). Factorial invariance: Historical trends and new developments. Paper presented at the Factor Analysis at 100 Conference, Chapel Hill, NC.\u003c/li\u003e\n\u003cli\u003eMorin AJS, Asparouhov T. Estimation of a Hierarchical Exploratory Structural Equation Model (ESEM) Using ESEM-Within-CFA. Montreal, QC: Substantive Methodological Synergy Research Laboratory (2018).\u003c/li\u003e\n\u003cli\u003eMorin AJS, Myers ND, Lee S. Modern factor analytic techniques: bifactor models, exploratory Exploratory Structural Equation Modeling (ESEM),19 structural equation modeling and bifactor-ESEM. In: Tenenbaum G, Eklund RC, editors. Handbook of Sport Psychology, 4th Edition, Vol. 2. New York, NY: Wiley Publishers (2020). p. 1044\u0026ndash;73. doi: 10.1002/9781119568124.ch51\u003c/li\u003e\n\u003cli\u003eMorin, A. J. S., Arens, A. K., \u0026amp; Marsh, H. W. (2016a). A bifactor exploratory structural equation modeling framework for the identification of distinct sources of construct-relevant psychometric multidimensionality. \u003cem\u003eStructural Equation Modeling: A Multidisciplinary Journal, 23\u003c/em\u003e(1), 116\u0026ndash;139. https://doi.org/10.1080/10705511.2014.961800\u003c/li\u003e\n\u003cli\u003eMorin, A. J. S., Marsh, H. W., \u0026amp; Nagengast, B. (2013). Exploratory structural equation modeling. In Hancock, G. R., \u0026amp; Mueller, R. O. (Eds.). Structural equation modeling: A second course (2nd ed., pp. 395-436). Charlotte, NC: Information Age Publishing, Inc.\u003c/li\u003e\n\u003cli\u003eMorin, A. J. S., Myers, N. D., \u0026amp; Lee, S. (2020). Modern factor analytic techniques: Bifactor models, exploratory structural equation modeling (ESEM), and bifactor‐ESEM. In G. Tenenbaum, R. C. Eklund, \u0026amp; N. Boiangin (Eds.), \u003cem\u003eHandbook of sport psychology: Exercise, methodologies, \u0026amp; special topics\u003c/em\u003e (4th ed., pp. 1044\u0026ndash;1073). John Wiley \u0026amp; Sons. https://doi.org/10.1002/9781119568124.ch51\u003c/li\u003e\n\u003cli\u003eMorin, A.J.S., Arens, A.K., Tran, A., \u0026amp; Caci, H. (2016b). Exploring sources of construct-relevant multidimensionality in psychiatric measurement: A tutorial and illustration using the Composite Scale of Morningness. International Journal of Methods in Psychiatric Research, 25, 277\u0026ndash;288. http://doi.org/10.1002/mpr.1485\u003c/li\u003e\n\u003cli\u003eMurray, A.L., \u0026amp; Johnson, W. (2013). The limitations of model fit in comparing the bi-factor versus higher-order models of human cognitive ability structure. Intelligence, 41, 407-422.\u003c/li\u003e\n\u003cli\u003eMuth\u0026eacute;n, B., \u0026amp; Kaplan, D. (1992). A Comparison of Some Methodologies for the Factor Analysis of Non-Normal Likert Variables: A Note on the Size of the Model. British Journal of Mathematical and Statistical Psychology, 45, 19-30. https://doi.org/10.1111/j.2044-8317.1992.tb00975.x\u003c/li\u003e\n\u003cli\u003eMuth\u0026eacute;n, L. K., \u0026amp; Muth\u0026eacute;n, B. O. (2002). How to use a Monte Carlo study to decide on sample size and determine power. \u003cem\u003eStructural Equation Modeling, 9\u003c/em\u003e(4), 599\u0026ndash;620. https://doi.org/10.1207/S15328007SEM0904_8\u003c/li\u003e\n\u003cli\u003eMuth\u0026eacute;n, L.K., \u0026amp; Muth\u0026eacute;n, B.O. (1998\u0026ndash;2025). \u003cem\u003eMplus\u003c/em\u003e (Version 8.11) [Computer software]. Los Angeles, CA: Muth\u0026eacute;n \u0026amp; Muth\u0026eacute;n.\u003c/li\u003e\n\u003cli\u003eOECD (2024), Western Balkans Competitiveness Outlook 2024: Regional Profile, Competitiveness and Private Sector Development, OECD Publishing, Paris, https://doi.org/10.1787/170b0e53-en.\u003c/li\u003e\n\u003cli\u003ePaz, D., Pontarolo, E., Santos, G., \u0026amp; Bernartt, M. (2021). An\u0026aacute;lisis de conte\u0026uacute;do sobre as defini\u0026ccedil;\u0026otilde;es de compet\u0026ecirc;ncia, compet\u0026ecirc;ncia digital e compet\u0026ecirc;ncia digital docente [Content analysis on the definitions of competence, digital competence and teacher digital competence]. Trabalho \u0026amp; Educa\u0026ccedil;\u0026atilde;o, 30(2), 207-225. https://doi.org/10.35699/2238-037x.2021.32817\u003c/li\u003e\n\u003cli\u003ePerdomo, B., Gonz\u0026aacute;lez-Mart\u0026iacute;nez, O. A., \u0026amp; Barrutia Barreto, I. (2020). Competencias digitales en docentes universitarios: una revisi\u0026oacute;n sistem\u0026aacute;tica de la literatura. \u003cem\u003eInnova Scientific\u003c/em\u003e. https://dialnet.unirioja.es/descarga/articulo/7559036.pdf\u003c/li\u003e\n\u003cli\u003ePongrac, D., Alić, M., \u0026amp; Cafuta, B. (2025). \u003cem\u003eDigital competences of digital natives: Measuring skills in the modern technology environment\u003c/em\u003e. \u003cstrong\u003eInformatics, 12\u003c/strong\u003e(1), Article 23. https://doi.org/10.3390/informatics12010023\u003c/li\u003e\n\u003cli\u003ePorat, E., Blau, I., \u0026amp; Barak, A. (2018). Measuring digital literacies: Junior high-school students\u0026rsquo; perceived competencies versus actual performance. Computers and Education, 126, 23\u0026ndash;36. https://doi.org/10.1016/j.compedu.2018.06.030\u003c/li\u003e\n\u003cli\u003eRaji, N. A. S., Busson-Crowe, D. A., and Dommett, E. J. (2023). University-wide digital skills training: a case study evaluation. Educ. Sci. 13:333. doi: 10.3390/educsci13040333\u003c/li\u003e\n\u003cli\u003eRamos, A., Casado, E., and Sevillano, U. (2023). Evaluating digital literacy levels among students in two private universities in Panama. South Florida J. Dev. 4, 1869\u0026ndash;1886. doi: 10.46932/sfjdv4n5-003\u003c/li\u003e\n\u003cli\u003eReise, S. P., Moore, T. M., \u0026amp; Haviland, M. G. (2010). Bifactor models and rotations: Exploring the extent to which multidimensional data yield univocal scale scores. \u003cem\u003eJournal of Personality Assessment, 92\u003c/em\u003e(6), 544\u0026ndash;559. https://doi.org/10.1080/00223891.2010.496477\u003c/li\u003e\n\u003cli\u003eRhemtulla, M., Brosseau-Liard, P. \u0026Eacute;., \u0026amp; Savalei, V. (2012). When can categorical variables be treated as continuous? A comparison of robust continuous and categorical SEM estimation methods under suboptimal conditions. Psychological methods, 17, 354-373.\u003c/li\u003e\n\u003cli\u003eRichter, T., Naumann, J., \u0026amp; Groeben, N. (2001). The computer literacy inventory (INCOBI): An instrument for the assessment of computer literacy and attitudes toward the computer in university students of the humanities and the social sciences. Psychologie in Erziehung und Unterricht, 48(1), 1\u0026ndash;13.\u003c/li\u003e\n\u003cli\u003eRodriguez, A., Reise, S. P., \u0026amp; Haviland, M. G. (2016). Evaluating bifactor models: Calculating and interpreting statistical indices. \u003cem\u003ePsychological Methods, 21\u003c/em\u003e(2), 137\u0026ndash;150. https://doi.org/10.1037/met0000045\u003c/li\u003e\n\u003cli\u003eRunge, I., Lazarides, R., Rubach, C., Richter, D., \u0026amp; Scheiter, K. (2023). Teacher-reported instructional quality in the context of technology-enhanced teaching: The role of teachers\u0026rsquo; digital competence- related beliefs in empowering learners. Computers \u0026amp; Education, 198, 104761.https://doi.org/10.1016/j.compedu.2023.104761.\u003c/li\u003e\n\u003cli\u003eS\u0026aacute;nchez-Caball\u0026eacute;, A., Gisbert-Cervera, M., \u0026amp; Esteve-Mon, F. (2020). The digital competence of university students: A systematic literature review. Revista de Psicologia, Ci\u0026egrave;ncies de l\u0026apos;Educaci\u0026oacute; i de l\u0026apos;Esport, 38(1), 63\u0026ndash;74. https://bit.ly/3lkkh3D\u003c/li\u003e\n\u003cli\u003eSatorra, A., \u0026amp; Bentler, P. M. (1994). Corrections to test statistics and standard errors in covariance structure analysis. In A. von Eye \u0026amp; C. C. Clogg (Eds.), \u003cem\u003eLatent variables analysis: Applications for developmental research\u003c/em\u003e (pp. 399\u0026ndash;419). Sage Publications\u003c/li\u003e\n\u003cli\u003eSatorra, A., \u0026amp; Bentler, P. M. (2010). Ensuring positiveness of the scaled difference chi-square test statistic. \u003cem\u003ePsychometrika, 75\u003c/em\u003e(2), 243\u0026ndash;248. https://doi.org/10.1007/s11336-009-9135y[1](https://publications.jrc.ec.europa.eu/repository/bitstream/JRC101254/jrc101254_digcomp%202.0%20the%20digital%20competence%20framework%20for%20citizens.%20update%20phase%201.pdf)\u003c/li\u003e\n\u003cli\u003eSavalei, V. (2021). Improving fit indices in structural equation modeling with categorical data. Multivariate Behavioral Research, 56(3), 390-407.\u003c/li\u003e\n\u003cli\u003eSchmitt, T.A., \u0026amp; Sass, D.A. (2011). Rotation criteria and hypothesis testing for exploratory factor analysis: implications for factor pattern loadings and interfactor correlations. Educational \u0026amp; Psychological Measurement, 71, 95-113.\u003c/li\u003e\n\u003cli\u003eScholtes, V. A. B., Terwee, C. B., \u0026amp; Poolman, R. W. (2011). What makes a measurement instrument valid and reliable? \u003cem\u003eInjury\u003c/em\u003e, \u003cem\u003e42\u003c/em\u003e(3), 236\u0026ndash;240. https://doi.org/10.1016/j.injury.2010.11.042\u003c/li\u003e\n\u003cli\u003eSelber, S. A. (2004). Reimagining the functional side of computer literacy. College Composition and Communication, 55(3), 470\u0026ndash;503. doi:10.2307/4140696.\u003c/li\u003e\n\u003cli\u003eShi, D., Maydeu-Olivares, A., \u0026amp; Rosseel, Y. (2020). Assessing fit in ordinal factor analysis models: SRMR vs. RMSEA. Structural Equation Modeling: A Multidisciplinary Journal, 27(1), 1-15.\u003c/li\u003e\n\u003cli\u003eSiddiq, F., Hatlevik, O. E., Olsen, R. V., Throndsen, I., \u0026amp; Scherer, R. (2016). Taking a future perspective by learning from the past: a systematic review of assessment instruments that aim to measure primary and secondary school students\u0026rsquo; ICT literacy. Educational Research Review, 19, 58\u0026ndash;84. https://doi.org/10.1016/j.edurev.2016.05.002\u003c/li\u003e\n\u003cli\u003eSillat, L. H., Tammets, K., \u0026amp; Laanpere, M. (2021). Digital competence assessment methods in higher education: A systematic literature review. Education in Science, 11, 402. https://doi.org/10.3390/educsci11080402\u003c/li\u003e\n\u003cli\u003eSolomon, L., \u0026amp; Rothblum, E. (1984). Academic procrastination: Frequency and cognitive-behavioral correlates. Journal of Counseling Psychology, 31, 503\u0026ndash;509. http://dx.doi.org/10.1037/0022-0167.31.4.503.\u003c/li\u003e\n\u003cli\u003eSoper, D. S. (2025). \u003cem\u003eA-priori sample size calculator for structural equation models\u003c/em\u003e [Software]. Available from https://www.danielsoper.com/statcalc\u003c/li\u003e\n\u003cli\u003eSotelo-N\u0026uacute;\u0026ntilde;ez, A. C., Herrera Rojas, J. J., Herrera Rojas, M. Z., and L\u0026oacute;pez-Regalado, O. (2024). Competencia digital en estudiantes universitarios: una revisi\u0026oacute;n sistemtica. Horizontes 8, 1781\u0026ndash;1800. doi: 10.33996/revistahorizontes.v8i34.833\u003c/li\u003e\n\u003cli\u003eSpante M, Hashemi SS, Lundin M, Algers A. Digital competence and digital literacy in higher education research: Systematic review of concept use. Cogent Education. 2018; 5(1):1\u0026ndash;21. https://doi.org/10.1080/2331186X.2018.1519143\u003c/li\u003e\n\u003cli\u003eUNESCO. (2017). Albania. Educational policy review; Issues and Recommendations. Paris, April 2017, p. 24. Retrieved from https://unesdoc.unesco.org/ark:/48223/pf0000247993\u003c/li\u003e\n\u003cli\u003eUNESCO. (2011). \u003cem\u003eICT competency framework for teachers\u003c/em\u003e. United Nations Educational, Scientific and Cultural Organization. https://unesdoc.unesco.org/ark:/48223/pf0000213475\u003c/li\u003e\n\u003cli\u003evan de Vijver, F., \u0026amp; Tanzer, N. K. (2004). Bias and equivalence in cross-cultural assessment: An overview. European Review of Applied Psychology, 54, 119-135.\u003c/li\u003e\n\u003cli\u003evan Deursen, A. J. A. M., \u0026amp; van Dijk, J. A. G. M. (2010). \u003cem\u003eInternet skills and the digital divide\u003c/em\u003e. \u003cem\u003eNew Media \u0026amp; Society\u003c/em\u003e, 12(8), 1239\u0026ndash;1259. https://doi.org/10.1177/1461444809355564\u003c/li\u003e\n\u003cli\u003evan Deursen, A. J. A. M., Helsper, E. J., \u0026amp; Eynon, R. (2016). Development and validation of the Internet Skills Scale (ISS). Information Communication \u0026amp; Society, 19(6), 804\u0026ndash;23. doi:10.1080/1369118x.2015.1078834.\u003c/li\u003e\n\u003cli\u003evan Laar, E., van Deursen, A. J. A. M., van Dijk, J. A. G. M., \u0026amp; de Haan, J. (2017). The relation between 21st-century skills and digital skills or literacy: A systematic literature review. Computers in Human Behavior, 72, 577\u0026ndash;588. doi:10.1016/j.chb.2017.03.010\u003c/li\u003e\n\u003cli\u003evan Zyl LE and ten Klooster PM (2022) Exploratory Structural Equation Modeling: Practical Guidelines and Tutorial With a Convenient Online Tool for Mplus. Front. Psychiatry 12:795672. doi: 10.3389/fpsyt.2021.795672\u003c/li\u003e\n\u003cli\u003eVoogt, J., \u0026amp; Roblin, N. P. (2012). A comparative analysis of international frameworks for 21st century competences: Implications for national curriculum policies. Journal of Curriculum Studies, 44(3), 299\u0026ndash;321. doi:10.1080/00220272.2012.668938\u003c/li\u003e\n\u003cli\u003eVuorikari, R., Punie, Y., Carretero, S., \u0026amp; Van den Brande, G. (2016). \u003cem\u003eDigComp 2.0: The Digital Competence Framework for Citizens. Update Phase 1: The Conceptual Reference Model\u003c/em\u003e. Luxembourg: Publications Office of the European Union. https://doi.org/10.2791/11517\u003c/li\u003e\n\u003cli\u003eVuorikari, R., Casta\u0026ntilde;o Mu\u0026ntilde;oz, J., Punie, Y., \u0026amp; Redecker, C. (2022). \u003cem\u003eDSI 2.0: Towards a more impactful Digital Skills Indicator\u003c/em\u003e. Publications Office of the European Union. https://data.europa.eu/doi/10.2760/47949[1](https://link.springer.com/book/10.1057/9781137437037)\u003c/li\u003e\n\u003cli\u003eVuorikari, R., Jerzak, N., Karpinski, Z., Pokropek, A., \u0026amp; Tudek, J. (2022). \u003cem\u003eMeasuring digital skills across the EU: Digital Skills Indicator 2.0\u003c/em\u003e (EUR 31193 EN). Publications Office of the European Union. https://doi.org/10.2760/897803\u003c/li\u003e\n\u003cli\u003eVuorikari, R., Kluzer, S., \u0026amp; Punie, Y. (2022). \u003cem\u003eDigComp 2.2: The Digital Competence Framework for Citizens \u0026ndash; With new examples of knowledge, skills and attitudes\u003c/em\u003e (EUR 31006 EN). Publications Office of the European Union. https://doi.org/10.2760/115376\u003c/li\u003e\n\u003cli\u003eVuorikari, R., Pokropek, A., \u0026amp; Casta\u0026ntilde;o Mu\u0026ntilde;oz, J. (2025). Enhancing digital skills assessment: Introducing compact tools for measuring digital competence. \u003cem\u003eTechnology, Knowledge and Learning\u003c/em\u003e. https://doi.org/10.1007/s10758-025-09825-x\u003c/li\u003e\n\u003cli\u003eWessel, D., Attig, C., and Franke, T. (2019). \u0026ldquo;ATI-S-An ultra-short scale for assessing affinity for technology interaction in user studies,\u0026rdquo; in Proceedings of the mensch und computer 2019 (MuC\u0026rsquo;19), eds F. Alt, A. Bulling, and T. D\u0026ouml;ring (New York, NY: Association for Computing Machinery), 147\u0026ndash;154. doi: 10.1145/3340764.3340766\u003c/li\u003e\n\u003cli\u003eWest, S. G., Finch, J. F., \u0026amp; Curran, P. J. (1995). Structural equation models with nonnormal variables: Problems and remedies. Thousand Oaks, CA: Sage.\u003c/li\u003e\n\u003cli\u003eWidowati, A., Siswanto, I., and Wakid, M. (2023). Factors affecting students\u0026rsquo; academic performance: self-efficacy, digital literacy, and academic engagement effects. Int. J. Instr. 16, 885\u0026ndash;898. doi: 10.29333/iji.2023.16449a\u003c/li\u003e\n\u003cli\u003eYockey, R. D. (2016). Validation of the Short Form of the academic procrastination scale. Psychological Reports, 118(1), 171-179.\u003c/li\u003e\n\u003cli\u003eYuan, X., Rehman, S., Altalbe, A., Rehman, E., \u0026amp; Shahiman, M. A. (2024). \u003cem\u003eDigital literacy as a catalyst for academic confidence: Exploring the interplay between academic self-efficacy and academic procrastination among medical students\u003c/em\u003e. BMC Medical Education, 24, Article 1317. https://doi.org/10.1186/s12909-024-06329-7\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Footnotes","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003e 2024 State of the Digital Decade package (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://digital-strategy.ec.europa.eu/en/policies/2024-state-digital-decade-package\u003c/span\u003e\u003cspan address=\"https://digital-strategy.ec.europa.eu/en/policies/2024-state-digital-decade-package\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e)\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://education.ec.europa.eu/focus-topics/digital-education/action-plan\u003c/span\u003e\u003cspan address=\"https://education.ec.europa.eu/focus-topics/digital-education/action-plan\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003e International Society for Technology in Education. (2016). \u003cem\u003eISTE standards for students\u003c/em\u003e. ISTE. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.iste.org/standards/iste-standards-for-students\u003c/span\u003e\u003cspan address=\"https://www.iste.org/standards/iste-standards-for-students\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003e UNESCO ICT Competency Framework for Teachers (ICT-CFT)\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Digital skills assessment, Educational policy, Exploratory Structural Equation Modeling (ESEM), Instrument development, Psychometric validation","lastPublishedDoi":"10.21203/rs.3.rs-7289357/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7289357/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThis study presents the development and validation of a 24-item instrument designed to assess students\u0026rsquo; digital skills, an essential competency in modern education. Grounded in a robust conceptual framework, the instrument captures key dimensions of digital literacy and was tested using cross-sectional data alongside advanced latent variable modeling techniques.\u003c/p\u003e\u003cp\u003eThe analytical methods applied included Confirmatory Factor Analysis (CFA), Exploratory Structural Equation Modeling (ESEM), bifactor-CFA, and bifactor-ESEM. Among these, ESEM yielded the best model fit, offering a nuanced representation of the multidimensional structure of students\u0026rsquo; digital skills.\u003c/p\u003e\u003cp\u003eThe model demonstrated strong psychometric properties, including high internal consistency, solid construct validity, and measurement invariance across gender. Predictive validity was also confirmed through significant associations with relevant educational outcomes.\u003c/p\u003e\u003cp\u003eThese findings support the instrument\u0026rsquo;s application in large-scale assessments and policy initiatives aimed at improving digital literacy. The validated framework provides a foundation for evidence-based decisions in curriculum design, teacher training, and educational planning, and is well-suited for integration into broader SEM framework-based research.\u003c/p\u003e","manuscriptTitle":"Validating a Psychometric Instrument for Assessing Students’ Digital Skills: A Latent Variable Approach for Policy-Oriented Educational Research","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-08-08 04:59:20","doi":"10.21203/rs.3.rs-7289357/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"06e4db47-25e3-42af-aaa3-27d5b726b9b3","owner":[],"postedDate":"August 8th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":52614090,"name":"Applied Statistics"},{"id":52614091,"name":"Educational Philosophy and Theory"}],"tags":[],"updatedAt":"2025-08-08T04:59:20+00:00","versionOfRecord":[],"versionCreatedAt":"2025-08-08 04:59:20","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7289357","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7289357","identity":"rs-7289357","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}