Access Profiles and Motivational Predictors of Online Reading Performance: A Study of Digital Equity in Taiwan

Abstract

The growth of the digital economy has reshaped educational opportunities worldwide, but it has also widened gaps in access, digital skills, and learning outcomes. This study investigates how digital device access profiles and psychological factors influence online reading performance among secondary school students in Taiwan. Drawing on data from a large-scale computer-based assessment, two complementary analyses were conducted. Objective 1 (n = 1,169) used latent class analysis to identify students’ digital device access profiles and examined their relationship to online reading outcomes across regions with differing levels of digital development. Findings show that students with “Basic” access (Profile 3: smartphone + computer) and those from digitally mature regions performed significantly better. Objective 2 (n = 2,246) focused on two key psychological factors: intrinsic motivation and self-efficacy, which were selected for their well-established roles in promoting reading engagement and self-regulated learning, particularly in digital environments. Multi-group structural equation modeling was then employed to explore how these factors predict the use of online reading strategies and, in turn, influence reading performance. Results showed that strategy use mediated the effects of motivation and self-efficacy in all regions, but intrinsic motivation had a direct impact on performance only in digitally mature areas. These findings reveal the complex interplay between access, learner psychology, and strategy use in shaping online reading outcomes. The study highlights the need for equity-focused investments in infrastructure, as well as pedagogical interventions that promote digital agency. Implications are offered for education policy, teacher practice, and digital governance aligned with SDGs 4 (quality education) and 10 (reduced inequalities).

Plain Language Summary

How Students’ Access to Technology and Motivation Affect Their Online Reading in Taiwan

In today's digital world, students need strong online reading skills to succeed in school and life. But not all students have equal access to digital devices, and not all feel confident or motivated when reading online. This study looked at junior high school students in Taiwan to understand how two types of factors, external (like access to digital tools and the level of digital development in their region) and internal (like students’ confidence and motivation), work together to influence online reading performance. First, we grouped students based on their patterns of device access, such as having laptops, tablets, or smartphones. Then, we compared these groups across different regions with higher or lower levels of digital infrastructure. Surprisingly, students with full access to all devices didn’t always perform the best; especially in more developed regions, where distractions or over-reliance on technology may play a role. We also found that students who were more motivated and confident in their reading abilities used better reading strategies, which in turn improved their reading outcomes. These effects were stronger in areas with better digital resources. Our results show that it’s not enough to simply give students devices. Schools and teachers also need to help students build motivation and confidence, and teach them strategies for reading and learning online. These findings can help policymakers and educators design more effective programs that close the digital divide and support all learners; especially those in underserved areas. By focusing on both access and internal factors like self-belief and interest, we can create more equal opportunities for success in the digital age.

Keywords

digital inclusion access profiles online reading performance intrinsic motivation self-efficacy

Introduction

The current digital age demands that citizens not only adapt to technological change, but also demonstrate proficiency in computer and information literacy (CIL); a critical competency for meaningful participation in the digital economy (Fraillon et al., 2018). At the core of CIL is reading proficiency, which enables individuals to critically engage with digital information and navigate online environments effectively (Taylor & Jaeger, 2022). As digital platforms increasingly become the primary sources of information, effective online reading skills are not only essential for academic success (Alexander et al., 2017; Jang et al., 2023), but also central to realizing the goals of sustainable and inclusive development in education. Given the increasing reliance on digital texts in learning and everyday life (García-Hernández et al., 2023; Napal et al., 2020), researchers have sought to understand how CIL and reading proficiency intersect (Yuan et al., 2021), both of which are foundational for equitable digital inclusion and lifelong learning (Tachie-Donkor & Ezema, 2023).

A growing body of research has explored the contextual factors that facilitate CIL acquisition, including access to computers and digital infrastructure (Alkan & Meinck, 2016; Aydin, 2022; Bundsgaard & Gerick, 2017). Scholars have increasingly recognized that online reading performance is influenced by multiple interrelated factors, including access to digital technologies, metacognitive reading strategies, and learners’ self-perceptions of competence (self-efficacy; Chang et al., 2023; Lim & Jung, 2019). Given that online reading is both a prerequisite and a reflection of broader information literacy skills, recent studies have begun to explore its link to CIL and other 21st-century competencies (Gubbels et al., 2020; Li & Chen, 2024; Zhang, 2021). As screen-based reading becomes more prevalent, students must contend with challenges such as evaluating digital content credibility, managing distractions, and mastering non-linear information pathways (Chen & Xiao, 2024). To address these challenges, Zheng et al. (2024) emphasize the importance of considering both social contexts and psychological processes when analyzing students’ digital reading behaviors. This aligns with the broader imperative of ensuring equitable access to digital learning and closing the digital divide, particularly in under-resourced communities.

Among the many psychological factors influencing digital reading, intrinsic motivation (IM) and self-efficacy (SE) have been consistently associated with greater strategy use, persistence, and reading comprehension in both traditional and online contexts (Hatlevik et al., 2018; Wigfield & Guthrie, 1997). However, these internal attributes do not operate in isolation. Their effects are often conditioned by external environments, particularly students’ access to technology and the maturity of regional digital infrastructure (Mohammed, 2022). These two constructs were selected because they are central to both theory and practice: IM is a key component of Self-Determination Theory (Deci & Ryan, 1985; Ryan & Deci, 2000), capturing students’ internal drive for learning, while SE is grounded in Social Cognitive Theory (Bandura, 1986), which reflects confidence in one’s ability to perform tasks. Both are critical for regulating effort and engagement, especially in autonomous digital learning contexts. This dual-layered phenomenon; wherein external disparities in digital device access and regional development, and internal differences in motivation and efficacy jointly affect online reading proficiency forms the foundation of the present study.

To investigate this, the study addresses two objectives: (a) to examine how environmental factors, specifically digital device access profiles and levels of regional digital development influence students’ online reading performance; and (b) to assess how psychological traits (IM and SE) affect students’ reading strategy use and performance, including whether these relationships differ across digital contexts. These objectives are grounded in Social Cognitive Theory (Bandura, 1986) and Self-Determination Theory (Deci & Ryan, 1985), which emphasize the importance of self-beliefs, motivation, and context in shaping academic behavior. By integrating these perspectives, the study contributes to a more comprehensive understanding of the digital divide, offering insight into how both access and learner agency determine educational equity in the digital age.

Literature Review

Digital Device Access and Regional Disparities in Online Reading

The growing centrality of digital reading in education has raised concerns about unequal access to technology and information. The digital divide; a gap in infrastructure, device ownership, and usage quality, remains a key determinant of educational inequality (Alkan & Meinck, 2016; Bundsgaard & Gerick, 2017). Beyond binary access (e.g., has or does not have a device), recent scholarship emphasizes a more layered understanding of access involving availability, frequency, purpose, and digital readiness (KC et al., 2025; Moore et al., 2018). In Taiwan and other high-performing Asian contexts, digital reading performance is significantly shaped by the home environment, school Information and Communication Technology (ICT) conditions, and regional disparities in infrastructure (Li & Chen, 2024). For example, Deng and Sun (2022), using Program for International Student Assessment (PISA) 2018, found that ICT development at the individual, school, and national levels all significantly predicted students’ digital reading literacy. Their findings also highlighted a “Matthew effect,” where well-resourced regions see accelerated gains, exacerbating short-term inequalities.

Moreover, Hu and Hu (2024) further demonstrated the long-term cognitive consequences of delayed access to digital devices, revealing that students who begin regular use after age 9 exhibit lower digital reading performance, partly due to reduced cognitive flexibility. These insights underscore that not only access, but also the timing and quality of access, are crucial. Similarly, within the Taiwanese context, students in urbanized or digitally mature areas have significantly better access to devices and reliable internet, while those in rural regions face ongoing constraints (García-Hernández et al., 2023). This access gap restricts opportunities to practice online reading skills and develop digital literacy.

Evidence from Thailand further confirms that simply providing devices may not improve outcomes unless accompanied by structured integration into the learning process (Dipendra et al., 2025). Policies focused purely on hardware provision fail to close equity gaps without parallel improvements in teacher preparation, curriculum design, and pedagogical use of ICT. These regional and international findings highlight the need for fine-grained profiling of digital device access. In global contexts such as the Western Balkans, digital transformation remains uneven, requiring not only regional cooperation, but also inclusive policies to ensure that digitization benefits all sectors of society (Mrdović, 2023). Overall, these cross-regional insights further reinforce the call for equity-focused approaches to digital education infrastructure and strategy. Latent Class Analysis (LCA), as used in this study, enables the identification of meaningful subgroups of students based on patterns of device usage across multiple platforms (Moore & Vitale, 2018). By analyzing these access profiles alongside regional development levels, this study contributes to a more varied understanding of digital inclusion and performance. Specifically, it extends existing research by integrating student-level digital access patterns with regional digital infrastructure maturity to examine their joint impact on online reading performance. Whereas prior research often treated infrastructure or device use in isolation, this study offers a more holistic perspective by examining how these elements interact. This approach directly addresses the first research objective: to examine how environmental factors, particularly digital device access profiles and regional digital development, predict students’ online reading achievement. This approach provides empirical grounding for policy and pedagogical decisions aimed at fostering equitable outcomes in digitally diverse settings.

Psychological Predictors of Digital Reading: Intrinsic Motivation and Self-Efficacy

Reading in online environments differs fundamentally from traditional, paper-based reading experiences (Singer & Alexander, 2017). Conventional reading relies on foundational skills such as word decoding and language understanding (Gough & Tunmer, 1986) and involves retrieving information, draw inferences, and synthesizing content across linear texts (Woolley, 2011). These are typically acquired through structured instruction (Pearson & Gallagher, 1983). In contrast, online reading requires additional competencies, such as navigating hypertext, evaluating sources, and integrating information across non-linear text structures; all of which challenge students’ cognitive and metacognitive abilities (Hahnel et al., 2016; Marchionini, 1989). Although print-based reading tends to result in better comprehension (Clinton, 2019; Kong et al., 2018), the dominance of screen-based technologies nowadays necessitates digital reading skills proficiency (Mangen & van der Weel, 2016).

While much of the early literature on IM and SE focused on traditional print reading (Wigfield & Guthrie, 1997), emerging studies have begun to examine how these constructs operate in digital environments, where students face heightened cognitive demands and greater autonomy (Mohammed, 2022; Unrau et al., 2018). SE is rooted in Bandura’s Social Cognitive Theory (Bandura, 1986, 2011), refers to belief in one’s ability to complete specific tasks. In digital contexts, higher SE correlates with better navigation, strategy use, and academic performance (Hatlevik et al., 2018; Mohammed, 2022; Rohatgi et al., 2016). Furthermore, challenges unique to digital reading, such as evaluating hypermedia and multitasking, heighten the importance of SE (Alghamdi & Sideridis, 2025; Artino, 2008). Importantly, Orakci (2023), using structural equation modeling (SEM), confirmed that SE significantly predicts creative problem-solving skills and reading achievement in technology-mediated learning environments.

In addition to Social Cognitive Theory, which emphasizes self-beliefs and efficacy (Bandura, 1986), Self-Determination Theory provides a complementary perspective. Self-Determination Theory posits that IM arises when individuals experience autonomy, competence, and relatedness; factors particularly salient in student-driven digital learning environments (Deci & Ryan, 1985; Ryan & Deci, 2000). However, despite the well-established role of IM in traditional literacy, fewer studies have directly investigated its role in digital reading contexts. This gap highlights the need for further research, such as the present study, to clarify how IM contributes to students’ success in online reading tasks. IM which is defined as the engagement driven by curiosity, enjoyment, or personal interest, has consistently been linked to reading success (Cambria & Guthrie, 2010; Troyer et al., 2019). Some studies noted that while IM may decline during adolescence, it remains predictive of reading comprehension across cultural contexts (Wang & Guthrie, 2004; Wigfield et al., 2016). Notably, Kanonire et al. (2022) and Zhu et al. (2024) found that IM directly or indirectly enhances reading performance and engagement, even after controlling for early literacy skills or ideal self-concept. Studies also confirm the importance of IM in digital settings. Forzani et al. (2021) developed a validated instrument; the Motivations for Online Reading Questionnaire (MORQ), which includes IM components such as curiosity and self-improvement, shown to predict online reading comprehension. In digital contexts, Mohammed (2022) found that students with higher IM and SE employed more sophisticated reading strategies, which in turn predicted better online reading performance.

In Taiwan, Liu and Ko (2014) distinguished between online reading comprehension and operational skills. They noted that mastery of both was rare among students, and recommended explicit instruction and collaborative activities. Subsequent studies (Chen, 2011; Hong et al., 2020; Liu et al., 2014) confirmed that pedagogical interventions enhanced IM and SE. Lastly, Huang and Yang (2015) showed that digital reading remediation programs effectively boosted students’ reading efficacy and motivation. Taken together, these studies highlight that both IM and SE are indispensable for online reading success. Their influence is amplified in digital settings, where learners face greater cognitive load and must independently deploy reading strategies. As such, targeted support for cultivating motivation and confidence remains a key priority for promoting digital literacy.

The Mediating Role of Reading Strategies in Digital Contexts

While psychological traits like IM and SE influence student engagement, their effects on reading outcomes often operate through an important intermediary, such as strategy use. Online reading demands the use of intentional, adaptive strategies to navigate hyperlinked texts, evaluate information credibility, and synthesize fragmented content. These strategies are more than technical skills; they reflect metacognitive regulation that supports deeper comprehension (Coiro & Dobler, 2007; Hahnel et al., 2016). Habók et al. (2024) confirmed the central role of reading strategies in mediating digital reading success. They found that among Hungarian students, problem-solving strategies exerted the strongest impact on online reading comprehension, outperforming global, and support strategies. Their results further demonstrate that successful readers in digital settings actively monitor comprehension, revise tactics when confusion arises, and apply integrative reasoning.

Within the digital environments, strategy use is not only more varied; it is also more necessary. Students must search, scan, and critically appraise sources in real-time, often without the support scaffolds available in print-based tasks (Marchionini, 1989). Comparative studies show that while students may have higher access to content online, this does not automatically translate into deeper learning without strategic engagement (Clinton, 2019; Singer & Alexander, 2017). Importantly, strategy use often mediates the relationship between internal psychological traits and reading performance. For instance, Zhu et al. (2024) demonstrated that IM influenced digital reading achievement largely through increased engagement; a proxy for strategic activity. Similarly, Mohammed (2022) and Forzani et al. (2021) found that students with stronger IM and SE were more likely to deploy higher-order reading strategies, suggesting an indirect pathway from beliefs to outcomes. Strategy use thus serves as a bridge between learner dispositions (IM, SE) and observable academic outcomes.

Likewise within the Taiwanese context, Liu and Ko (2014) emphasized the importance of teaching strategic online reading comprehension explicitly, given that only a small fraction of students independently applied such skills. Follow-up studies have also supported the effectiveness of guided instruction, particularly in collaborative or remedial settings (Hong et al., 2020; Huang & Yang, 2015). Overall, these findings reinforce the importance of strategy use as both a cognitive skillset and a mediating mechanism through which motivation and efficacy translate into performance. In this study, the mediating role of strategy use is tested within a multi-group SEM framework to determine whether it explains how IM and SE affect performance across digital development levels. While several studies have explored the mediating role of strategy use, few have formally examined how these psychological processes interact with broader contextual differences. In particular, there is limited research on whether the effects of IM and SE on strategy use and performance differ across regions with varying levels of digital development. By applying a multi-group SEM approach, this study not only tests the mediating effect of reading strategy use, but also explores moderation by digital environment. This dual analytic framework provides a more integrated understanding of how internal dispositions and external conditions jointly shape students’ digital reading outcomes.

Conceptual Framework and Hypotheses Development

As noted earlier, the current study draws on a multidimensional conceptual framework integrating elements from Self-Determination Theory, Social Cognitive Theory, and digital divide literature. It aims to understand how external environmental factors (digital device access profiles and regional digital development) and internal psychological factors (IM and SE) jointly influence students’ online reading performance. The proposed framework posits two pathways:

Access-performance pathway: Students’ digital device access profiles and regional digital development directly affect online reading outcomes.

Psychological pathway: IM and SE influence strategy use, which in turn mediates their effects on reading performance. Regional digital development moderates these relationships.

This framework informs the following hypotheses:

H1: Students with more advanced digital device access profiles and those from digitally mature regions will demonstrate higher online reading performance.

H2: IM (H2a) and SE (H2b) will positively predict online reading strategy use.

H3: Online reading strategy use will mediate the relationship between IM (H3a) /SE (H3b) and online reading performance.

H4: The mediating effect of online reading strategies between IM/SE and reading performance will vary depending on students’ regional digital development level.

Figure 1 illustrates the hypothesized relationships among contextual (H1) and psychological predictors (H2a–H2b), the mediating role of strategy use (H3a–H3b), and the moderating influence of regional digital development (H4). By integrating LCA and multi-group SEM, the study seeks to offer a comprehensive model of digital inclusion that bridges infrastructure, learner psychology, and reading performance.

Figure 1.

Theoretical framework and hypothesized model diagram.

As noted earlier that while several studies have examined how psychological traits such as IM and SE influence reading engagement (Mohammed, 2022; Wigfield et al., 2016), fewer have explored how these relationships might vary across differing environmental contexts. In particular, the moderating role of regional digital development in online reading has been largely understudied. Given well-documented disparities in infrastructure and digital access across regions in Taiwan (Deng & Sun, 2022; García-Hernández et al., 2023), it is reasonable to hypothesize that the pathways from psychological traits to performance may differ by context. Multi-group SEM was therefore used as a strategy to test for moderation by digital development level, a method commonly employed in educational research to examine cross-group differences in latent variable relationships (Cheung & Rensvold, 2002; Putnick & Bornstein, 2016).

Materials and Methods

Research Design and Context

This study employed a quantitative, cross-sectional research design using data from a large-scale, government-supported educational assessment conducted in Taiwan during the 2022 to 2023 academic year (Hsieh et al., 2024; Wu et al., 2023). The assessment aimed to evaluate students’ online reading performance and examine factors related to digital device access and psychological readiness. Data were collected through a standardized system coordinated by national testing authorities. The dataset comprises three main components:

Online Reading Assessment: A computer-based reading test administered to Grade 7 students to evaluate their comprehension and navigation of digital texts;

Student Questionnaire: A self-report survey completed by all participating students that included measures of IM, SE, and online reading strategy use; and

Household Digital Access Survey: A supplementary module completed by a subset of students whose parents or guardians consented to provide information about home internet connectivity, device availability, and usage conditions. Of the 2,246 students who completed the core modules, 1,169 provided complete data for this household component.

These three data sources were integrated into a single analytic framework. While the student questionnaire and reading performance data were universally collected, the inclusion of household-level digital access data enabled further subgroup comparisons based on home environment and regional development indicators. This design allowed for the modeling of relationships between environmental factors (digital device access profiles and regional digital development), psychological factors (IM and SE), and digital reading performance using LCA and multi-group SEM.

Sampling Design and Participants

Taiwan offers a robust digital learning ecosystem supported by widespread internet infrastructure (Kuo et al., 2023). However, disparities remain, particularly between urban centers and more remote or Indigenous communities (Chou & Pan, 2024). To address these gaps, the Taiwan National Development Council classifies regions into four digital development levels: mature, potential, beginning, and emerging areas; based on indicators such as ICT application, education and culture, and infrastructure (United Marketing Research, 2020). In parallel, the Executive Yuan (2002) designates 55 townships as Indigenous areas, where differentiated education policies are implemented to promote cultural inclusion.

To ensure broad and equitable representation, a stratified two-stage cluster sampling design was employed (S. E. Lee et al., 2016). Stratification considered: (a) digital development level, (b) school type (public or private), and (c) geographic designation (Indigenous or non-Indigenous area). A total of 16 theoretical strata were identified, from which 10 strata were sampled due to availability. Within each stratum, schools were selected using probability proportional to size (PPS) sampling (Rosén, 1997). In total, 179 schools and 345 classes participated in the study, yielding 9,009 valid student responses (51.2% boys, 48.8% girls). Due to the matrix sampling design of the assessment program (Gonzalez & Rutkowski, 2010), students received different combinations of test and questionnaire booklets. Two relevant subsamples were used:

For Objective 1, data from 1,169 students were analyzed. These students completed both the online reading test and a parent-linked module on household digital access.

For Objective 2, 2,246 students who completed the online reading test and the student questionnaire on psychological constructs (IM and SE) were included.

These samples included approximately equal proportions of male and female students, with an average age of 14.8 years (Standard Deviation; SD = 0.4). Informed parental consent and student assent were obtained prior to data collection, in accordance with ethical research protocols approved by the institutional review board (IRB) overseeing the study. These targeted subsamples allowed for focused analysis of both environmental and psychological predictors of online reading performance, while ensuring regional and institutional diversity aligned with national digital equity goals.

Instruments and Variables

Objective 1

Digital Development Levels

As noted earlier, Taiwan’s National Development Council classifies each township into one of four digital development tiers: mature, potential, beginning, and emerging, based on six dimensions of digital readiness.

Digital Device Access

Student access to digital devices was measured using five items assessing the frequency of use for computers, smartphones, tablets/e-readers, wearable devices, and game consoles. Responses were originally collected on a 7-point frequency scale ranging from “almost every day” (6) to “almost never” (1), with an additional “no access” category coded as 0. For analytic clarity and latent profile modeling, responses were dichotomized into 1 = frequent use (more than once or twice a week) and 0 = infrequent or no access (once a week or less, or no access).

Online Reading Assessment

The online reading test included 63 items delivered digitally using a matrix sampling approach. Items measured four core competencies adapted from Progress in International Reading Literacy Study (PIRLS; Mullis & Martin, 2019): (a) retrieving information, (b) making inferences, (c) integrating ideas, and (d) evaluating and reflecting on content. The test was scored using the Multidimensional Random Coefficients Multinomial Logit Model (MRCMLM; Adams et al., 1997) using the Test Analysis Module (TAM) package in R (Robitzsch et al., 2024). Model fit indices were within acceptable ranges (outlier-sensitive fit; outfit 0.65–1.60; information-weighted fit; infit 0.86–1.18); infit and outfit mean square values between 0.5 and 1.5 are generally considered acceptable, while values near 1.0 indicate good model fit (Wright & Linacre, 1994), and the reliability statistics were satisfactory (Expected A Posteriori Reliability; EAP = 0.97; Warm Likelihood Estimate Reliability; WLE = 0.74; Warm, 1989). Scale scores were standardized (Mean = 500, SD = 100; Mullis & Martin, 2019).

Data were collected during the second semester of the 2022 to 2023 academic year. Online reading tests and student questionnaires were administered in school computer labs under the supervision of trained facilitators. The household digital access survey was distributed electronically to parents/guardians through school communication platforms. Ethical approval was obtained from the IRB office and informed consent was secured from parents and assent from students prior to participation.

Objective 2

Psychological constructs are composed of three latent variables measured through student self-report:

Online Reading Strategies (ORS): 10 items, 4-point Likert (1932) type scale (e.g., “I check my understanding of online texts as I read”); Cronbach (1951)α = .94, signifying quite good internal consistency (Cohen et al., 2007).

Intrinsic Motivation (IM): 4 items (e.g., “I enjoy searching for new and interesting topics online”); α = .88.

Self-Efficacy (SE): 6 items (e.g., “Online reading is one of my strengths”); α = .84.

Construct validity was confirmed using Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA). The three-factor model explained 58.17% of the variance, and CFA indices indicated good model fit (Root Mean Square Error of Approximation; RMSEA = 0.063, cutoff value <0.08 for acceptable fit; comparative fit index; CFI = 0.95, cutoff value ≥0.95 for good fit), all of which are within the acceptable values suggesting the measurement model adequately fits the observed data (L.-T. Hu & Bentler, 1999; Schreiber et al., 2006). Chi-square to degrees of freedom ratio (χ²/df = 2.87) was within acceptable range (<3.00), further supporting model fit (Schreiber et al., 2006). In addition, descriptive statistics showed that all continuous variables exhibited acceptable levels of skewness and kurtosis (range: −1.12–1.35), indicating approximate univariate normality. Given the large sample size, normality assumptions for SEM were considered sufficiently met (Kline, 2011)

All survey instruments were administered in Mandarin Chinese; the students’ native language. Where original items were sourced from English-language scales, translations and cultural adaptations were made by a bilingual expert panel following standard forward–backward translation procedures. Furthermore, in this study, the proportion of missing values for most questionnaire items ranged from 2.4% to 4%, with only one item (SPQ_05_10) reaching a missing rate of 7.6%. Since all missing rates were below the commonly accepted threshold of 10%, mean imputation was used to preserve sample size and data completeness (Schafer & Graham, 2002). A paired-sample comparison of the original and imputed variables showed that the means, SDs, and standard errors were identical across all items. As the differences were zero, t-tests could not be computed. This indicates that the imputation had no impact on the central tendency or variability of the data and did not affect the validity of subsequent statistical analyses (Tabachnick & Fidell, 2019). Therefore, the analyses conducted in this study can be considered statistically stable and representative, unaffected by the imputation process.

Data Analysis

For Objective 1, a LCA was employed to classify students into distinct digital device access profiles based on their reported frequency of using various technology devices. LCA was conducted using Mplus version 8.0 software (Muthén & Muthén, 1998-2017). Model fit was evaluated using standard criteria, including entropy values (≥0.60), the Bayesian Information Criterion (BIC), and the statistical insignificance of G², the Lo-Mendell-Rubin adjusted likelihood ratio test (LMRp), and the bootstrap likelihood ratio test (BLRTp; Nylund et al., 2007). Based on the optimal model, conditional probabilities were examined and student groups were labeled accordingly. To examine performance differences, a two-way analysis of variance (ANOVA) was conducted to test the main and interaction effects of digital development area and access profile on online reading performance. Prior to analysis, the assumptions of ANOVA, including homogeneity of variance, were verified using Levene’s test (Andrade, 2024). Where appropriate, post hoc comparisons were conducted to identify statistically significant group differences (Hair et al., 2019).

For Objective 2, a multi-group SEM was conducted to examine the effects of IM and SE on ORS, and their subsequent influence on online reading performance. Multi-group SEM analysis was conducted in several stages. First, the model was estimated separately for students residing in mature digital environments and for those in the other three digital development areas combined. Given prior research showing significant disparities in digital infrastructure and learning conditions across Taiwanese regions (Deng & Sun, 2022; Kuo et al., 2023), students were grouped into two categories: (a) those in digitally mature areas and (b) those in potential, beginning, or emerging areas. This grouping enabled multi-group SEM to test for moderation effects of digital development context on psychological pathways. Overall model fit was assessed using established criteria, including a non-significant chi-square value (with the understanding that chi-square is sensitive to large sample sizes), an RMSEA below 0.08, and a CFI above 0.90 (Bagozzi & Yi, 1988; Kline, 2011). In addition, factor loadings and item error variances were examined to ensure that all items met the requisite fit criteria. Following the initial model estimation, measurement invariance across the two groups was tested using a series of constrained models in AMOS version 26 software (Byrne, 2001). The baseline model (m0) and subsequent models (m1 through m4), which introduced additional constraints on regression coefficients, intercepts, and other parameters, were evaluated. Chi-square difference tests were conducted at each step to assess whether imposing these constraints resulted in statistically significant differences between models, thereby informing our understanding of group differences in the measurement model (J. L. Lee & Stankov, 2023). Lastly, to assess the significance of indirect effects, nonparametric bootstrapping with 5,000 resamples was conducted. Bias-corrected confidence intervals (CI) were used to determine mediation significance, as this method does not assume normality of the indirect effect distribution (Preacher & Hayes, 2008).

Results

Objective 1: Digital Device Access Profile and Regional Development Effects on Online Reading

Table 1 shows the distribution of valid samples and the population living in the four areas. In each area, the percentages were similar to those obtained from the population data. Based on these results, it is evident that Study 1 participants comprised a representative sample of the population. Note that the information about population distribution were retrieved from United Marketing Research (United Marketing Research, 2020).

Table 1.

Sample and Population Distribution (n = 1,169).

Digital development area	Sample		Population
Digital development area	Number	Percentage	Percentage
Mature area	727	62.2	58.8
Potential area	268	22.9	24.8
Beginning area	109	9.3	10.8
Emerging area	65	5.6	5.6

For the construct “access to information technology equipment,” students were instructed to indicate the frequency with which they utilize digital technology devices. The construct was comprised of five items, with respondents indicating the frequency of their use on a rating scale ranging from 1 (almost every day) to 7 (almost never). The detailed structure of the scale are as follows: 1 (almost every day), 2 (three or four times a week), 3 (once or twice a week), 4 (three or four times a month), 5 (once or twice a month), 6 (a couple of times a year), and 7 (almost never). A rating of 0 indicates that there is no such equipment item. Due to the limited number of responses in some categories, the ratings were simplified and the options were reduced from eight to two. The study combined responses ranging from one to three, labeling this category as “more than once or twice a week,” and combined responses ranging from four to seven, as well as zero, labeling this category as “less than once a week or without this device.”Table 2 illustrates that smartphones were the most prevalent device, with a usage rate of 90%, followed by computers (including desktops and laptops) at 55%. Conversely, game consoles (e.g., XBOX, PS5, Switch) exhibited the lowest usage rate at 26% .

Table 2.

Descriptive Statistics of Access to Information Technology Equipment (n = 1,169).

Device	Usage frequency	%	n
Computer (including desktop or laptop)	More than once or twice a week	55	646
Computer (including desktop or laptop)	Less than once a week or without this device	45	523
Smartphone	More than once or twice a week	90	1,050
Smartphone	Less than once a week or without this device	10	119
Wearable devices (Apple watch, Xiaomi, …)	More than once or twice a week	40	470
Wearable devices (Apple watch, Xiaomi, …)	Less than once a week or without this device	60	699
Tablet or eBook reader (iPad, Kindle, …)	More than once or twice a week	39	454
Tablet or eBook reader (iPad, Kindle, …)	Less than once a week or without this device	61	715
Game consoles (XBOX, PS5, Switch, …)	More than once or twice a week	26	308
Game consoles (XBOX, PS5, Switch, …)	Less than once a week or without this device	74	861

LCA of Digital Device Access Profiles

To examine students’ access to digital technology, a LCA was performed on five device use variables. The model fit evaluation (Table 3) indicated that the three-class model provided the optimal solution, with adequate entropy (0.72), a low BIC value (6,676.83), and non-significant G², LMRp, and BLRTp statistics. The resulting three profiles were:

Minimal Access (Category 1): Limited to primarily smartphones, 3.51% of the sample.

Basic Access (Category 3): Regular use of smartphones and computers, 83.23%.

Full Access (Category 2): Regular use of all devices (smartphones, computers, tablets, wearables, game consoles), 13.26%.

Table 3.

Latent Class Analysis Model Fit.

Model	NPara	Entropy	BIC	G ²	df	LMRp	BLRTp
Null	5	1.00	6,897.63	328.39**	26	—	—
2-class	11	0.78	6,648.78	37.16*	20	<.001	<.001
3-class	17	0.72	6,676.83	22.82	14	.07	.07
4-class	23	0.52	6,708.31	11.93	8	.10	.11

Note. NPara = Number of parameters estimated in the model; Entropy = Values closer to 1 indicate better separation between classes; BIC = Bayesian Information Criterion; G² = Likelihood Ratio Chi-square statistic; df = degrees of freedom; LMRp = The p-value for the Lo-Mendell-Rubin; BLRTp = The p-value for the Bootstrap Likelihood Ratio Test.

p < .05, **p < .01.

Conditional probabilities across device types for each group are shown in Table 4. Smartphone use was high across all groups, but full access users showed notably higher engagement with tablets, wearables, and game consoles. The current level of access to technology has been classified in accordance with previous categorizations made by other scholars (Moore & Vitale, 2018; Moore et al., 2018; Ogundari, 2024).

Table 4.

Conditional Probabilities by Access Profile.

Device	Category 1	Category 2	Category 3
Computer (including desktop or laptop)	.24	1.00	.52
Smartphone	.67	.99	.91
Wearable devices (Apple watch, Xiaomi, …)	.02	.85	.38
Tablet or eBook reader (iPad, Kindle, …)	.08	1.00	.33
Game consoles (XBOX, PS5, Switch, …)	.00	.84	.20
Number of clusters (probability)	41 (3.51%)	155 (13.26%)	973 (83.23%)
Group label	Minimal access	Full access	Basic access

Two-Way ANOVA: Impact of Region and Access Profile

A two-way ANOVA was conducted to test the effect of digital development area and access profile on online reading performance. Levene’s test confirmed homogeneity of variance (F = 1.26, p = .24; Andrade, 2024; Hair et al., 2019). The interaction term (A × B) was not significant, but both main effects were statistically significant (see Table 5). Moreover, the post-hoc comparisons indicated that:

Students from mature digital development areas scored significantly higher than those in potential, beginning, and emerging areas.

Students in the Basic Access group (smartphone + computer) outperformed both Full and Minimal Access groups, indicating that moderate, but purposeful technology use may be more beneficial than widespread device saturation.

Table 5.

Two-Way ANOVA Results.

Source	Type III SS	df	Mean square	F value	η²
Digital development areas (A)	306,118.47	3	102,039.49	10.64**	.03
Access to IT equipment (B)	398,049.19	2	199,024.59	20.75**	.04
A*B	78,335.55	6	13,055.92	1.36	.01
Error	11,100,000.00	1,157	9,590.86
Total	307,400,000.00	1,169

Note. SS = Sum of squares; df = degrees of freedom; η² (Eta-squared) = Effect size.

p < .01.

This pattern aligns with previous studies suggesting that unstructured or excessive access to digital tools, such as wearables or gaming consoles, which may dilute the educational utility of technology access (Lim & Jung, 2019; Moore & Vitale, 2018). Overall, these findings shed light on how both regional infrastructure and patterns of digital access contribute to disparities in digital reading proficiency (as illustrated in Figure 2, wherein the figure shows that students in mature digital areas consistently exhibit higher online reading performance than those in less developed regions. Additionally, students with basic access; primarily to computers and smartphones, outperformed those with either minimal or full access, highlighting that quality and relevance of access, rather than sheer number of devices, plays a more critical role in supporting online reading proficiency).

Figure 2.

Students’ online reading performance across digital development areas and digital device access profiles.

In sum, LCA identified three distinct digital device access profiles, with the Basic Access (category 3) group achieving the highest online reading scores. Regional disparities persisted, with mature digital areas showing clear advantages. These findings support Hypothesis 1 (H1) and underscore the importance of not only equitable access to devices, but also the nature and quality of technology engagement.

Objective 2: Psychological and Contextual Predictors of Online Reading Performance

To examine the interplay between internal and external factors in students’ online reading performance, a multi-group SEM was used. The proposed model posited that students’ IM and SE toward online reading significantly influence their use of ORS, which in turn affect overall online reading performance. In addition, the study hypothesized that the model is moderated by the digital development areas in which students reside (see Figure 3, wherein the structural model illustrates the relationships between intrinsic motivation (IM) toward online reading (ξ1), self-efficacy (SE) toward online reading (ξ2), online reading strategy (ORS; η1), and online reading performance (η2) in predicting online reading ability (Vach). Observed variables (V1–V20) represent measurement indicators corresponding to each latent construct. Arrows indicate hypothesized causal paths between constructs, suggesting the mediating role of online reading strategy and performance in the development of online reading ability). Multi-group SEM analysis was conducted in several stages. First, the model was estimated separately for students residing in mature digital environments and for those in the other three digital development areas combined. Overall model fit was assessed using established criteria, including a non-significant chi-square value (with the understanding that chi-square is sensitive to large sample sizes), an RMSEA below 0.08, and a CFI above 0.90 (Bagozzi & Yi, 1988; Kline, 2011). In addition, factor loadings and item error variances were examined to ensure that all items met the requisite fit criteria. Following the initial model estimation, measurement invariance across the two groups was tested using a series of constrained models. The baseline model (m0) and subsequent models (m1 through m4), which introduced additional constraints on regression coefficients, intercepts, and other parameters, were evaluated. Chi-square difference tests were conducted at each step to assess whether imposing these constraints resulted in statistically significant differences between models, thereby informing our understanding of group differences in the measurement model.

Figure 3.

Influence of individual psychological factors on online reading performance.

Model Fit and Overall Evaluation

To verify the interplay between internal and external factors in students’ online reading performance, a multi-group SEM was implemented. The model proposed that students’ IM and SE toward online reading have a significant impact on their use of ORS, and thereby on their performance in online reading. Moreover, it was also proposed that this model is moderated by the digital development areas wherein the students reside. Multi-group SEM analysis was conducted using a series of procedures. The initial step was to ascertain model fit for students residing in the mature area and the other three combined areas. The evaluation criteria for overall fit included a non-significant Chi-square value, a RMSEA of less than 0.08, and a CFI greater than 0.90. The results demonstrated that the error variances of all items from the models were positive and statistically significant, with factor loadings ranging from 0.61 to 0.85. This satisfied the requisite criteria for item fit. Table 6 presents the results of the remaining model fit indices. The Chi-square goodness-of-fit value was statistically significant due to its sensitivity to sample size, particularly in the context of large sample sizes (Jöreskog & Sörbom, 1993). Moreover, the RMSEA values ranges from 0.063 to 0.065, all of which met the fit criteria. In general, these results indicated that the proposed model was acceptable.

Table 6.

SEM Model Fit by Digital Development Area.

Model	n	χ2	df	p	RMSEA	CFI
Mature area	1,403	1,203.53	184	<.001	0.063	0.94
Other areas	843	844.22	184	<.001	0.065	0.94

Note. n = sample size; χ2 = Chi-square statistics; df = degrees of freedom; p = significant value; RMSEA = Root Mean Square Error of Approximation; CFI = Comparative Fit Index.

Standardized Effect Sizes and Key Relationships

Table 7 presents the standardized effects of IM and SE on ORS and online reading performance. Key findings include:

Both IM and SE had significant positive effects on ORS across both groups (β ≈ .36–.38), indicating that motivated and confident students are more likely to use effective reading strategies.

IM and SE did not have significant direct effects on performance in either group; however, their indirect effects through ORS were significant (β = .06–.11), confirming the mediating role of strategy use.

ORS had a strong direct effect on performance, especially in less developed areas (β = .29), indicating its central role in bridging motivation and achievement.

Table 7.

Standardized Effects in SEM.

Outcome variables	Effect typed	Predictors	Mature (β [95% CI])	Others(β [95% CI])
Online reading strategy (ORS)	Total	Intrinsic Motivation (IM)	.36* [.25, .45]	.38** [.26, .51]
		Self-Efficacy (SE)	.37* [.28, .46]	.36* [.25, .51]
Online reading performance	Direct	IM	.08 [−.05, .23] (ns)	−.19 [−.19, .05] (ns)
		SE	−.06 [−.20, .10] (ns)	.10 [−.09, .32] (ns)
		ORS	.17** [.11, .25]	.29** [.20, .37]
	Indirect	IM	.06* [.02, .11]	.11* [.06, .15]
		SE	.06* [.02, .11]	.11* [.06, .15]
	Total	IM	.14* [.02, .24]	−.08 [−.19, .05] (ns)
		SE	.00 [−.14, .13] (ns)	.21* [.09, .32]
		ORS	.17** [.11, .25]	.29** [.20, .37]

Note. ORS = Online reading strategies; IM = Intrinsic motivation; SE = Self-efficacy; CI = 95% bias-corrected bootstrap confidence interval; ns = not significant (CI includes 0).

p < .05, **p < .01.

These effects varied in magnitude across digital contexts, providing empirical support for H4, which posits that the strength and direction of these relationships are moderated by the digital development area.

H2a and H2b: Psychological Predictors of Strategy Use

The SEM results revealed that both IM and SE had significant positive effects on students’ use of online reading strategies in both the mature digital development area and the other three areas combined. As shown in Table 7, wherein:

IM → ORS: β = .36 [.25, .45] in mature areas; β = .38 [.26, .51] in other areas

SE → ORS: β = .37 [.28, .46] in mature areas; β = .36 [.25, .51] in other areas

These findings support H2a and H2b, confirming that students who are more intrinsically motivated or have higher confidence are more likely to adopt effective reading strategies.

H3a and H3b: Strategy Use as a Mediator of Reading Performance

The indirect and total effects were examined. IM and SE did not have significant direct effects on online reading performance (β not significant), but their indirect effects through ORS were significant in both groups:

IM → ORS → Performance: β = .06* [.02, .11] in mature areas; β = .11* [.06, .15] in other areas

SE → ORS → Performance: β = .06* [.02, .11] in mature areas; β = .11* [.06, .15] in other areas

Moreover, ORS had a strong direct effect on performance:

ORS → Performance: β = .17** [.11, .25] in mature areas; β = .29** [.20, .37] in other areas

These results support H3a and H3b, emphasizing that strategy use is a key mediator between motivation/beliefs and actual reading performance.

Measurement Invariance Testing

In the multi-group SEM analysis, parameter estimation was conducted using AMOS, software in accordance with methodological guidelines established by Bollen (1989), Byrne (2001), and Kline (2011). To ensure that the measurement model was consistent across groups, measurement invariance was assessed using a series of constrained models. The baseline model (m0) in AMOS constrained 24 item error variance parameters and 17 item factor loading parameters to be invariant across the groups. The model in question comprised 136 parameters for the two groups combined. Following the imposition of constraints on 24 item error variance parameters and 17 item factor loading parameters, 95 parameters were able to be freely estimated. The degrees of freedom for the baseline model (m0) were calculated as 409, derived from the formula 184 × 2 + 41. Subsequently, further constraints were imposed on the parameters of the m0 model in each subsequent iteration. Model m1 introduced constraints on the regression coefficients β21, γ12, and γ22. Model m2 further constrained the intercepts, while model m3 imposed constraints on the γ11 and γ21 coefficients. Ultimately, model m4 imposed constraints on the factor score (θ). The results demonstrated that the imposition of constraints on the regression coefficients β21, γ12, and γ22 resulted in statistically significant discrepancies between models m0 and m1 as well as the imposition of constraints on the item intercept between models m1 and m2.

H4: Moderating Role of Digital Development Area

To test H4, multi-group SEM was conducted by comparing models with increasing parameter constraints. Table 8 presents the results of measurement invariance testing:

Significant model differences emerged when constraints were placed on regression paths (m0 vs. m1: Δχ2 = 10.21, p = .02) and intercepts (m1 vs. m2: Δχ2 = 92.67, p < .001)

Post-hoc tests indicated significant differences for γ12 (IM → Performance; Δχ2 = 8.98, p < .001) and intercept Vach (Δχ2 = 61.54, p < .001)

Table 8.

Measurement Invariance Testing.

Model	Model fit					Model comparison	χ² Difference test
Model	χ ²	df	p	RMSEA	CFI	Model comparison	Δχ²	△df	p
m0	2,120.24	409	<.001	0.043	0.94	m0–m1	10.21	3	.02
m1	2,130.45	412	<.001	0.043	0.94	m1–m2	92.67	21	<.001
m2	2,223.12	433	<.001	0.043	0.94	m2–m3	0.87	2	.65
m3	2,223.99	435	<.001	0.043	0.94	m3–m4	0.33	1	.57
m4	2,224.32	436	<.001	0.043	0.94

Note. m0 = baseline model (configural invariance); m1 = Model with constraints on regression coefficients β21, γ12, and γ22; m2 = Model with intercepts constrained; m3 = Model with additional constraints on regression coefficients γ11 and γ21; m4 = Model with the constraint on the factor score (θ); χ2 = Chi-square statistics; df = degrees of freedom; p = significant value; RMSEA = Root Mean Square Error of Approximation; CFI = Comparative Fit Index; Δχ2 = Chi-square change; Δdf = change in degrees of freedom.

These findings confirm H4, indicating that the influence of IM on performance is moderated by regional digital context. Students in mature areas benefit more from motivation in terms of performance outcomes.

Post Hoc Comparisons and Group Differences

The parameter estimates for each model are presented in Table 9, with constrained parameters clearly highlighted with gray background. Given the significant χ² differences between models m0 and m1, as well as between models m1 and m2, the study proceeded with post hoc comparisons by incrementally constraining the regression coefficients β₂₁, γ₁₂, γ₂₂ and the intercepts individually. The results revealed statistically significant differences between the maturity and other three areas combined for γ₁₂ (△χ² = 8.98, △df = 1, p < .001) and for intercept V_ach (△χ² = 61.54, △df = 1, p < .001). These findings suggest that IM exerts a differential impact on online reading performance across different digital development areas, with students in mature area demonstrating significantly better online reading performance compared to those in the other three areas combined.

Table 9.

Unstandardized Estimates from Multiple Groups SEM.

Parameters	m0		m1		m2		m3		m4
	Maturity area	Other areas	Maturity area	Other areas	Maturity area	Other areas	Maturity area	Other areas	Maturity area	Other areas
β₂₁	28.45**	48.37**	35.43**	35.43**	36.51**	36.51**	36.55**	35.55**	.36.53**	36.53**
γ₁₂	12.36	−31.36*	−4.17	−4.17	−4.93	−4.93	−5.00	−5.00	−4.80	−4.80
γ₂₂	−13.48	22.86	1.22	1.22	0.32	0.32	0.35	0.35	0.11	0.11
γ₁₁	0.35**	0.39**	0.36**	0.38**	0.36**	0.38**	0.37**	0.37**	0.37**	0.37**
γ₂₁	0.50**	0.50**	0.50**	0.51**	0.50**	0.51**	0.50**	0.50**	0.50**	0.50**
$\emptyset_{12}$	0.20**	0.20**	0.20**	0.20**	0.20**	0.20**	0.20**	0.20**	0.20**	0.20**
Intercept V₁	2.98**	3.03**	2.98**	3.03**	3.00**	3.00**	3.00**	3.00**	3.00**	3.00**
Intercept V₂	3.24**	3.22**	3.24**	3.22**	3.23**	3.23**	3.23**	3.23**	3.23**	3.23**
Intercept V₃	3.01**	3.04**	3.01**	3.04**	3.02**	3.02**	3.02**	3.02**	3.02**	3.02**
Intercept V₄	2.57**	2.65**	2.57**	2.65**	2.60**	2.60**	2.60**	2.60**	2.60**	2.60**
Intercept V₅	3.31**	3.30**	3.31**	3.30**	3.31**	3.31**	3.31**	3.31**	3.31**	3.31**
Intercept V₆	3.10**	3.12**	3.10**	3.12**	3.11**	3.11**	3.11**	3.11**	3.11**	3.11**
Intercept V₇	2.69**	2.72**	2.69**	2.72**	2.70**	2.70**	2.70**	2.70**	2.70**	2.70**
Intercept V₈	2.77**	2.82**	2.77**	2.82**	2.79**	2.79**	2.79**	2.79**	2.79**	2.79**
Intercept V₉	2.92**	2.95**	2.92**	2.95**	2.93**	2.93**	2.93**	2.93**	2.93**	2.93**
Intercept V₁₀	3.03**	3.03**	3.03**	3.03**	3.03**	3.03**	3.03**	3.03**	3.03**	3.03**
Intercept V₁₁	2.65**	2.71**	2.65**	2.71**	2.67**	2.67**	2.67**	2.67**	2.67**	2.67**
Intercept V₁₂	2.83**	2.85**	2.83**	2.85**	2.84**	2.84**	2.84**	2.84**	2.84**	2.84**
Intercept V₁₃	3.10**	3.09**	3.10**	3.09**	3.10**	3.10**	3.10**	3.10**	3.10**	3.10**
Intercept V₁₄	3.07**	3.03**	3.07**	3.03**	3.06**	3.06**	3.06**	3.06**	3.06**	3.06**
Intercept V₁₅	3.10**	3.04**	3.10**	3.04**	3.08**	3.08**	3.08**	3.08**	3.08**	3.08**
Intercept V₁₆	2.91**	2.93**	2.91**	2.93**	2.92**	2.92**	2.92**	2.92**	2.92**	2.92**
Intercept V₁₇	3.00**	2.99**	3.00**	2.99**	2.99**	2.99**	2.99**	2.99**	2.99**	2.99**
Intercept V₁₈	2.89**	2.92**	2.89**	2.92**	2.91**	2.91**	2.91**	2.91**	2.91**	2.91**
Intercept V₁₉	2.98**	2.98**	2.98**	2.98**	2.98**	2.98**	2.98**	2.98**	2.98**	2.98**
Intercept V₂₀	3.12**	3.10**	3.12**	3.10**	3.11**	3.11**	3.11**	3.11**	3.11**	3.11**
Intercept V_ach	512.81**	478.69**	512.81**	478.69**	500.00**	500.00**	500.00**	500.00**	500.00**	500.00**

Note. The gray background refers to parameters that are constrained to be equal across groups; m0 = baseline model (configural invariance); m1 = Model with constraints on regression coefficients β21, γ12, and γ22; m2 = Model with intercepts constrained; m3 = Model with additional constraints on regression coefficients γ11 and γ21; m4 = Model with the constraint on the factor score (θ).

p < .05, **p < .01.

Discussions

In an era increasingly shaped by the digital economy, proficiency in online reading is more than an academic skill; it is a core competency for full participation in knowledge-based societies. The findings of this study highlight the dual influence of external and internal factors on students’ online reading performance, offering significant insights into digital inclusion, equitable education, and sustainable development within the context of Taiwan’s digital transformation.

The first objective explored how environmental factors, particularly regional digital development and students’ digital device access shape online reading outcomes. LCA revealed three distinct digital device access profiles, and ANOVA results confirmed that students in digitally mature areas consistently outperformed peers in less developed regions. Notably, students with “Basic” access (frequent use of both smartphones and computers) achieved higher reading performance than those with either limited or full-spectrum device access. This challenges the common assumption that more devices automatically enhance digital learning. In fact, excessive access to non-academic devices such as gaming consoles and wearables may distract from academic engagement, as echoed in prior research (Beuermann et al., 2015; Carter et al., 2017; J. Hu & Hu, 2024). Furthermore, as Figure 2 illustrates this phenomenon: in mature digital regions, students with full-spectrum device access performed the lowest. One explanation is that excessive device availability may lead to multitasking, fragmented attention, or entertainment-driven screen time, undermining purposeful academic engagement. The superior performance of the “Basic access” group suggests that optimal access and not maximum access is more conducive to learning. This reinforces the importance of structured and purposeful technology integration into students’ learning environments.

Overall, these results highlight the importance of strategic, pedagogically guided access to digital tools rather than unchecked device proliferation. Furthermore, the strong performance of students with basic access in mature digital areas illustrates a synergistic effect between local infrastructure and digital engagement. Even modest access, when situated in robust digital environments, can translate into high digital literacy and academic success. This supports targeted regional investment to close opportunity gaps and promote sustainable digital education (Deng & Sun, 2022; Dipendra et al., 2025; OECD, 2019; UNESCO, 2021).

The second objective examined how IM and SE predict students’ use of online reading strategies, and how these in turn affect reading outcomes. SEM results confirmed that both psychological traits significantly influenced strategy use, which served as a robust mediator of reading performance. Hypotheses H2 and H3 were supported. These further demonstrates the intertwined relationship between IM and SE. The strong path coefficients between them reflect a “pull association,” whereby students who are intrinsically motivated are also likely to believe in their reading competence and vice versa. This synergy amplifies their strategy use and, in turn, performance. Furthermore, the multi-group SEM indicated contextual moderation (H4) in digitally mature areas, IM had a marginal and statistically non-significant direct effect on performance, while in less developed areas, its effect was fully mediated by strategy use. These findings align with Social Cognitive Theory (Bandura, 1986) and Self-Determination Theory (Deci & Ryan, 1985), which emphasize the role of both personal agency and contextual support in driving motivation and learning outcomes. Moreover, they resonate with findings by Mohammed (2022) and Forzani et al. (2021), who showed that both IM and SE predict strategy use, which in turn enhances online reading performance.

This differential pattern highlights that even highly motivated students in under-resourced settings may be constrained by infrastructural or instructional limitations. In contrast, students in mature digital regions are better positioned to convert motivation into achievement. These results mirror the observations of Wigfield et al. (2016) and Zhu et al. (2024), who found that IM influences digital reading performance largely through strategic engagement, particularly in less supported environments. This insight calls for systemic interventions that combine digital access, learner agency, and instructional alignment to foster inclusive and effective digital education.

Taken together, the findings affirm that digital inclusion is multidimensional, spanning infrastructure, access, psychological readiness, and strategy use. As Taiwan and other economies advance in their digital agendas, it is crucial to balance technology provision with investments in learner-centered pedagogy and supportive environments. These findings support the goals of SDG 4 (inclusive and equitable education) and SDG 10 (reduced inequalities), showing that effective digital transformation depends not just on access, but on fostering the conditions in which students can thrive.

Conclusion

This study examined how digital device access profiles, regional digital development, and psychological factors predict online reading performance among Taiwanese middle school students. It found that students with basic, purposeful access to core devices in digitally mature regions consistently outperformed their peers. Moreover, the study demonstrated that motivation and self-efficacy influence performance primarily through the mediating role of strategy use, especially in less developed areas where direct effects of motivation were minimal. These findings advance understanding of digital inclusion as an interplay of structural and psychological elements. They show that providing devices alone is insufficient. Instead, educational systems must ensure that students are supported by aligned pedagogy, robust infrastructure, and motivational scaffolds. The results reinforce the importance of context-sensitive digital strategies that center both access and student agency in efforts toward equity.

Based on the findings, the following recommendations are proposed:

Infrastructure Investment: Education ministries should prioritize investments in digital infrastructure in underserved regions. Providing stable internet and access to essential tools (e.g., smartphones, computers) is foundational for reducing digital learning gaps.

Strategy-Based Digital Literacy: Schools should embed instruction in online reading strategies, such as source evaluation, synthesis, and critical reading into the curriculum. These skills are crucial mediators between motivation and achievement.

Teacher Professional Development: Teachers should be trained not only in ICT integration, but also in supporting student motivation and SE. This is especially vital in less digitally mature areas (Ertmer & Ottenbreit-Leftwich, 2010; Tondeur et al., 2017).

Context-Aware Policy: Policymakers must avoid equating access with equity. Effective digital governance requires attention to the interaction of technological, psychological, and pedagogical factors.

Ecological and Instructional Research: Future studies should include school-level and community-level variables, such as digital leadership, instructional design, and parental involvement, to better understand and support digital equity.

By implementing these strategies, educators and policymakers can better support sustainable and inclusive digital learning environments that empower all students to succeed.

Importantly, several practical strategies can support students’ digital reading development in the classroom. First, teachers should explicitly teach online reading strategies, including how to evaluate the credibility of sources, navigate hyperlinked text, and synthesize information from multiple tabs or platforms. Embedding these skills into regular lessons can strengthen metacognitive regulation. Second, given the role of IM, educators are encouraged to design student-centered tasks that foster curiosity and autonomy, such as allowing students to choose reading topics or engage in inquiry-based online projects. Third, to enhance SE, teachers can incorporate scaffolded digital reading tasks with gradual release of responsibility, combined with formative feedback that reinforces students’ confidence in navigating online texts. Finally, for learners in less digitally mature regions or with limited home access, structured in-school access to devices and peer collaboration activities can help bridge gaps and provide meaningful practice in authentic digital contexts.

While this study provides meaningful insights into the dynamics of digital inclusion and online reading performance, several limitations must be acknowledged. First, the data were drawn exclusively from seventh-grade students in Taiwan. Although the national scope and sampling design support internal validity, generalizing findings to other educational systems or cultural contexts should be done cautiously. Cross-national comparative studies could strengthen the global applicability of these results. Second, the study employed a cross-sectional design, capturing student performance and perceptions at a single point in time. This limits the ability to draw causal inferences or observe developmental trajectories. Future research using longitudinal or intervention-based designs would provide greater insight into how digital access, motivation, and strategy use evolve over time—and how these variables respond to policy and pedagogical changes. Third, intrinsic motivation and self-efficacy were assessed through self-report measures, which are subject to potential biases such as social desirability and inaccurate self-perception. Although widely accepted in educational research, future studies may benefit from multi-method approaches, including behavioral data, teacher assessments, or digital trace analysis to triangulate findings. Fourth, while the study addresses technological access and psychological readiness, it does not fully account for the instructional and pedagogical factors that may mediate these relationships. For example, students with basic access may experience very different instructional quality depending on their teachers’ ICT integration practices. Investigating classroom-level variables, such as digital pedagogy, curriculum design, or school leadership, would help to contextualize these outcomes. Finally, other potentially influential factors, such as parental support, school-level policies, teacher digital competencies, and broader community resources, were not included in the current models. These ecological and organizational variables may interact with both access and motivation to shape reading outcomes and should be included in more comprehensive future frameworks. Future research should continue to bridge psychological, infrastructural, and pedagogical dimensions of digital education to inform inclusive, sustainable practices in diverse educational contexts.

Footnotes

Acknowledgements

During the preparation of this manuscript, the authors used Wordtune for the purpose of language check and readability of the text. The authors have reviewed and edited the output and take full responsibility for the content of this publication.

ORCID iDs

Pei-Ching Chao

Jin-Chang Hsieh

Gregory S. Ching

Ethical Considerations

The protocol of the study was reviewed and approved by the National Cheng Kung University Human Research Ethics Committee with Case Number 112-095 and Academia Sinica Institutional Review Board with Case Number AS-IRB-HS07-112054.

Consent to Participate

Informed consent was obtained from all subjects involved in the study.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research is funded by the Taiwan National Academy for Educational Research under project number NAER-2023-011-C-1-1-C3-01. Open-access publication fees were supported by Taiwan Academia Sinica.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

Data is available upon formal written request to the second author.

References

Adams

R. J.

Wilson

Wang

W.-C.

(1997). The multidimensional random coefficients multinomial logit model. Applied Psychological Measurement, 21(1), 1–23. https://doi.org/10.1177/0146621697211001

Alexander

Adams Becker

Cummins

Hall Giesinger

(2017). Digital literacy in higher education, Part II: An NMC horizon project strategic brief. The New Media Consortium.

Alghamdi

Sideridis

(2025). The dual role of digital self-efficacy in reading engagement from a nonlinear dynamics perspective. Children, 12(3), 292. https://doi.org/10.3390/children12030292

Alkan

Meinck

(2016). The relationship between students’ use of ICT for social communication and their computer and information literacy. Large-scale Assessments in Education, 4(15), 1–17. https://doi.org/10.1186/s40536-016-0029-z

Andrade

(2024). Understanding factorial designs, main effects, and interaction effects: Simply explained with a worked example. Indian Journal of Psychological Medicine, 46(2), 175–177. https://doi.org/10.1177/02537176241237066

Artino

A. R.

(2008). Motivational beliefs and perceptions of instructional quality: Predicting satisfaction with online training. Journal of Computer Assisted Learning, 24(3), 260–270. https://doi.org/10.1111/j.1365-2729.2007.00258.x

Aydin

(2022). A multilevel modeling approach to investigating factors impacting computer and information literacy: ICILS Korea and Finland sample. Education and Information Technologies, 27, 1675–1703. https://doi.org/10.1007/s10639-021-10690-1

Bagozzi

(1988). On the evaluation of structural equation models. Journal of the Academy of Marketing Sciences, 16, 74–94. https://doi.org/10.1007/BF02723327

Bandura

(1986). Social foundations of thought and action: A social cognitive theory. Prentice Hall.

10.

Bandura

(2011). A social cognitive perspective on positive psychology. International Journal of Social Psychology, 26(1), 7–20. https://doi.org/10.1174/021347411794078444

11.

Beuermann

D. W.

Cristia

Cueto

Malamud

Cruz-Aguayo

(2015). One laptop per child at home: Short-term impacts from a randomized experiment in Peru. American Economic Journal: Applied Economics, 7(2), 53–80. https://doi.org/10.1257/app.20130267

12.

Bollen

K. A.

(1989). Structural equations with latent variables. John Wiley & Sons.

13.

Bundsgaard

Gerick

(2017). Patterns of students’ computer use and relations to their computer and information literacy: Results of a latent class analysis and implications for teaching and learning. Large-scale Assessments in Education, 5(16), 1–15. https://doi.org/10.1186/s40536-017-0052-8

14.

Byrne

B. M.

(2001). Structural equation modeling with AMOS: Basic concepts, applications, and programming. Lawrence Erlbaum Associates Publishers.

15.

Cambria

Guthrie

J. T.

(2010). Motivating and engaging students in reading. The NERA Journal, 46(1), 16–29.

16.

Carter

S. P.

Greenberg

Walker

M. S.

(2017). The impact of computer usage on academic performance: Evidence from a randomized trial at the United States Military Academy. Economics of Education Review, 56, 118–132. https://doi.org/10.1016/j.econedurev.2016.12.005

17.

Chang

Wang

Liu

Feng

Zhang

(2023). Study on factors influencing college students’ digital academic reading behavior. Frontiers in Psychology, 13, 1007247. https://doi.org/10.3389/fpsyg.2022.1007247

18.

Chen

H. Y.

(2011). Online reading comprehension strategies among fifth- and sixth-grade general and special education students. Education Research and Perspectives, 37(2), 79–109.

19.

Chen

Xiao

(2024). Pathways to digital reading literacy among secondary school students: A multilevel analysis using data from 31 economies. Computers & Education, 218. https://doi.org/10.1016/j.compedu.2024.105090

20.

Cheung

G. W.

Rensvold

R. B.

(2002). Evaluating goodness-of-fit indexes for testing measurement invariance. Structural Equation Modeling: A Multidisciplinary Journal, 9(2), 233–255. https://doi.org/10.1207/S15328007SEM0902_5

21.

Chou

Y.-C.

Pan

H.-H.

(2024). Effect of digital gap on virtual care in the time of COVID-19 pandemic: Empirical evidence in Taiwan. Asia Pacific Journal of Social Work and Development, 34(2), 86–100. https://doi.org/10.1080/02185385.2023.2210112

22.

Clinton

(2019). Reading from paper compared to screens: A systematic review and meta-analysis. Journal of Research in Reading, 42(2), 288–325. https://doi.org/10.1111/1467-9817.12269

23.

Cohen

Manion

Morrison

(2007). Research method in education. Routledge.

24.

Coiro

Dobler

(2007). Exploring the online reading comprehension strategies used by sixth-grade skilled readers to search for and locate information on the Internet. Reading Research Quarterly, 42(2), 214–257. https://doi.org/10.1598/RRQ.42.2.2

25.

Cronbach

L. J.

(1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16, 197–334. https://doi.org/10.1007/BF02310555

26.

Deci

E. L.

Ryan

R. M.

(1985). Intrinsic motivation and self-determination in human behavior. Plenum Press. https://doi.org/10.1007/978-1-4899-2271-7

27.

Deng

Sun

(2022). How the ICT development level influences students’ digital reading literacy: A multi-level model comparison based on PISA 2018 data. Frontiers of Education in China, 17(2), 151–180. https://doi.org/10.3868/s110-007-022-0008-9

28.

Dipendra

K. C.

Pramod

K. C.

Rado

Vichit-Vadakan

(2025). Digital inequality and learning outcomes: Evidence from Thailand. Education and Information Technologies, [Advance online publication]. https://doi.org/10.1007/s10639-025-13570-0

29.

Ertmer

P. A.

Ottenbreit-Leftwich

A. T.

(2010). Teacher technology change: How knowledge, confidence, beliefs, and culture intersect. Journal of Research on Technology in Education, 42(3), 255–284. https://doi.org/10.1080/15391523.2010.10782551

30.

Executive Yuan. (2002). Xíng zhèng yuàn 91 nián 4 yuè 16 rì yuàn tái jiāng zì dì 0910017300 hào hán tóng yì hé dìng “yuán zhù mín dì qū” bāo kuò 55 gè xiāng (zhèn, shì) [Letter No. 0910017300 from the Taiwan Executive Yuan on April 16, 1991 - Approval of 55 townships and municipalities as areas of indigenous people]. Retrieved May 1, 2025, from https://www.cip.gov.tw/zh-tw/news/data-list/C6DFB32460630CF4/0C3331F0EBD318C213147655B37A59D5-info.html

31.

Forzani

Leu

D. J.

E. Y.

Rhoads

Guthrie

J. T.

McCoach

(2021). Characteristics and validity of an instrument for assessing motivations for online reading to learn. Reading Research Quarterly, 56(4), 761–780. https://doi.org/10.1002/rrq.337

32.

Fraillon

Ainley

Schulz

Duckworth

Friedman

(2018). IEA international computer and information literacy study 2018 assessment framework. Springer Cham. https://doi.org/10.1007/978-3-030-19389-8

33.

García-Hernández

Muñoz-Repiso

A. G.-V.

Casillas-Martín

Cabezas-González

(2023). Sustainability in digital education: A systematic review of innovative proposals. Education Sciences, 13(1), 33. https://doi.org/10.3390/educsci13010033

34.

Gonzalez

Rutkowski

(2010). Principles of multiple matrix booklet designs and parameter recovery in large-scale assessments. IEA-ETS Research Institute Monograph, 3, 125–156.

35.

Gough

P. B.

Tunmer

W. E.

(1986). Decoding, reading, and reading disability. Remedial and Special Education, 7(1), 6–10. https://doi.org/10.1177/074193258600700104

36.

Gubbels

Swart

N. M.

Groen

M. A.

(2020). Everything in moderation: ICT and reading performance of Dutch 15-year-olds. Large-scale Assessments in Education, 8(1), 1–17. https://doi.org/10.1186/s40536-020-0079-0

37.

Habók

T. Z.

Magyar

(2024). The effect of reading strategy use on online reading comprehension. Heliyon, 10(2), Article e24281. https://doi.org/10.1016/j.heliyon.2024.e24281

38.

Hahnel

Goldhammer

Naumann

Kröhne

(2016). Effects of linear reading, basic computer skills, evaluating online information, and navigation on reading digital text. Computers in Human Behavior, 55(A), 486–500. https://doi.org/10.1016/j.chb.2015.09.042

39.

Hair

J. F.

Babin

B. J.

Black

W. C.

Anderson

R. E.

(2019). Multivariate data analysis. Cenage.

40.

Hatlevik

O. E.

Throndsen

Loi

Gudmundsdottir

G. B.

(2018). Students’ ICT self-efficacy and computer and information literacy: Determinants and relationships. Computers and Education, 118, 107–119. https://doi.org/10.1016/j.compedu.2017.11.011

41.

Hong

H. Y.

Lin

P. Y.

Lee

K. Y. H.

(2020). Advancing third graders’ reading comprehension through collaborative knowledge building: A comparative study in Taiwan. Computers & Education, 157, Article 103962. https://doi.org/10.1016/j.compedu.2020.103962

42.

Hsieh

J.-C.

Wang

H.-L.

Chen

K.-Z.

F.-L.

Huang

Y.-R.

Chen

Y.-H.

Chao

P.-C.

Tsai

M.-H.

(2024). Guó jiā jiào yù yán jiū yuàn yán jiū chéng guÇ bào gào: Shù wèi shì dài xià tái wān xué shēng chéng jiù cháng qī zhuī zōng píng liáng (I) [Final report of the Taiwan Assessment of Student Achievement: Longitudinal study in the digital generation (I)] (NAER-2023-011-C-1-1-C3-01). National Academy for Educational Research.

43.

(2024). The influence of age at first regular digital device access on digital reading performance: The mediating effect of cognitive flexibility. Humanities and Social Sciences Communications, 11, 799. https://doi.org/10.1057/s41599-024-03292-y

44.

L.-T.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6(1), 1–55. https://doi.org/10.1080/10705519909540118

45.

Huang

C. T.

Yang

S. C.

(2015). Effects of online reciprocal teaching on reading strategies, comprehension, self-efficacy, and motivation. Journal of Educational Computing Research, 52(3), 381–407. https://doi.org/10.1177/0735633115571924

46.

Jang

Seo

Y. S.

Brutt-Griffler

(2023). Building academic resilience in literacy: Digital reading practices and motivational and cognitive engagement. Reading Research Quarterly, 58(1), 160–176. https://doi.org/10.1002/rrq.486

47.

Jöreskog

K. G.

Sörbom

(1993). LISREL 8: Structural equation modeling with the SIMPLIS command language. Scientific Software International.

48.

Kanonire

Lubenko

Kuzmina

(2022). The effects of intrinsic and extrinsic reading motivation on reading performance in elementary school. Journal of Research in Childhood Education, 36(1), 1–13. https://doi.org/10.1080/02568543.2020.1822961

49.

Kline

R. B.

(2011). Principles and practice of structural equation modeling. Guilford Press.

50.

Kong

Seo

Y. S.

Zhai

(2018). Comparison of reading performance on screen and on paper: A meta-analysis. Computers & Education, 123, 138–149. https://doi.org/10.1016/j.compedu.2018.05.005

51.

Kuo

B.-C.

Doody

Bardone

Chang

F. T.-Y.

Jang

H.-J.

Masiello

Wong

J. M. S.

Toennsen

K.-C.

K. C.

Cameron

Lee

L.-S.

Pedaste

Kim

N. T. P.

Busharian

Avidov-Ungar

Häkkinen

Bacsich

… Mohseni

Z. A.

(2023). Trends and issues of promoting digital learning in high-digital-competitiveness countries: Country reports and international comparison. Technological and Vocational Education Research Center (TVERC), National Taiwan Normal University, K-12 Education Administration (K12EA), Ministry of Education, Taiwan.

52.

Lee

J. L.

Stankov

(2023). Multilevel latent profile analysis of global competence in PISA 2018: Small country-level differences. Journal of Cross-Cultural Psychology, 54(6–7), 658–683. https://doi.org/10.1177/00220221231183150

53.

Lee

S. E.

Lee

P. R.

Shin

K.-I.

(2016). A composite estimator for stratified two stage cluster sampling. Communications for Statistical Applications & Methods, 23(1), 47–55. https://doi.org/10.5351/CSAM.2016.23.1.047

54.

Chen

(2024). Impacts of ICT-related factors on students’ digital reading literacy: Evidence from high-performing Asian countries and regions. Education and Information Technologies, 29, 16717–16747. https://doi.org/10.1007/s10639-024-12501-9

55.

Likert

(1932). A technique for the measurement of attitudes. Archives of Psychology, 22(140), 55.

56.

Lim

H. J.

Jung

(2019). Factors related to digital reading achievement: A multi-level analysis using international large scale data. Computers & Education, 133, 82–93. https://doi.org/10.1016/j.compedu.2019.01.007

57.

Liu

I.-F.

H.-W.

(2014). Guó xiăo xué shēng “xiàn shàng yuè dú sù yăng cè yàn” zhī biān zhì yŭ xiàn shàng yuè dú néng lì biăo xiàn zhī chū tàn [Developing online reading skill test to assess elementary students’ online reading performance]. Psychological Testing, 61(4), 509–532.

58.

Liu

I. F.

H. W.

S. Y.

(2014). Learning reading strategies with online discussion. Journal of Educational Computing Research, 50(2), 231–247. https://doi.org/10.2190/EC.50.2.e

59.

Mangen

van der Weel

(2016). The evolution of reading in the age of digitisation: An integrative framework for reading research. Literacy, 50(3), 116–124. https://doi.org/10.1111/lit.12086

60.

Marchionini

(1989). Making the transition from print to electronic encyclopaedias: Adaptation of mental models. International Journal of Man-Machine Studies, 30(6), 591–602. https://doi.org/10.1016/S0020-7373(89)80012-4

61.

Mohammed

G. M. S.

(2022). The impact of reading strategies and self-efficacy on reading comprehension: The case of Saudi EFL learners. Journal of Language & Linguistics Studies, 18(SI1), 627–639.

62.

Moore

Vitale

(2018). High school students’ access to and use of technology at home and in school. ACT Center for Equity in Learning.

63.

Moore

Vitale

Stawinoga

(2018). The digital divide and educational equity: A look at students with very limited access to electronic devices at home. ACT Center for Equity in Learning.

64.

Mrdović

(2023). The role of digitalisation in transforming Western Balkan societies. ÖGfE Policy Brief, 14'2023.

65.

Mullis

I. V. S.

Martin

M. O.

(2019). PIRLS 2021 assessment framework. TIMSS & PIRLS International Study Center, Lynch School of Education, Boston College and International Association for the Evaluation of Educational Achievement.

66.

Muthén

L. K.

Muthén

B. O.

(1998-2017). Mplus user’s guide (8th ed.). Muthén & Muthén.

67.

Napal

Mendióroz-Lacambra

A. M.

Peñalva

(2020). Sustainability teaching tools in the digital age. Sustainability, 12(8), Article 3366. https://doi.org/10.3390/su12083366

68.

Nylund

K. L.

Asparouhov

Muthén

B. O.

(2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling: A Multidisciplinary Journal, 14(4), 535–569. https://doi.org/10.1080/10705510701575396

69.

OECD. (2019). Trends shaping education 2019. OECD.

70.

Ogundari

(2024). Student access to technology at home and learning hours during COVID-19 in the US. Educational Research for Policy and Practice, 22, 443–460. https://doi.org/10.1007/s10671-023-09342-7

71.

Orakci

(2023). Structural relationship among academic motivation, academic self-efficacy, problem solving skills, creative thinking skills, and critical thinking skills. Psychology in the Schools, 60(7), 2173–2194. https://doi.org/10.1002/pits.22851

72.

Pearson

P. D.

Gallagher

M. C.

(1983). The instruction of reading comprehension. Contemporary Educational Psychology, 8(3), 317–344. https://doi.org/10.1016/0361-476X(83)90019-X

73.

Preacher

K. J.

Hayes

A. F.

(2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40, 879–891. https://doi.org/10.3758/BRM.40.3.879

74.

Putnick

D. L.

Bornstein

M. H.

(2016). Measurement invariance conventions and reporting: The state of the art and future directions for psychological research. Developmental Review, 41, 71–90. https://doi.org/10.1016/j.dr.2016.06.004

75.

Robitzsch

Kiefer

(2024). TAM: Test analysis modules. R package version 4.2-21. Retrieved May 1, 2025, from https://CRAN.R-project.org/package=TAM

76.

Rohatgi

Scherer

Hatlevik

O. E.

(2016). The role of ICT self-efficacy for students’ ICT use and their achievement in a computer and information literacy test. Computers and Education, 102, 103–116. https://doi.org/10.1016/j.compedu.2016.08.001

77.

Rosén

(1997). On sampling with probability proportional to size. Journal of Statistical Planning and Inference, 62(2), 159–191. https://doi.org/10.1016/S0378-3758(96)00186-3

78.

Ryan

R. M.

Deci

E. L.

(2000). Self-determination theory and the facilitation of intrinsic motivation, social development, and well-being. American Psychologist, 55(1), 68–78. https://doi.org/10.1037/0003-066X.55.1.68

79.

Schafer

J. L.

Graham

J. W.

(2002). Missing data: Our view of the state of the art. Psychological Methods, 7(2), 147–177. https://doi.org/10.1037/1082-989X.7.2.147

80.

Schreiber

J. B.

Nora

Stage

F. K.

Barlow

E. A.

King

(2006). Reporting structural equation modeling and confirmatory factor analysis results: A review. The Journal of Educational Research, 99(6), 323–338. https://doi.org/10.3200/JOER.99.6.323-338

81.

Singer

L. M.

Alexander

P. A.

(2017). Reading on paper and digitally: What the past decades of empirical research reveal. Review of Educational Research, 86(7), 1007–1041. https://doi.org/10.3102/0034654317722961

82.

Tabachnick

B. G.

Fidell

L. S.

(2019). Using multivariate statistics (7th ed.). Pearson.

83.

Tachie-Donkor

Ezema

I. J.

(2023). Effect of information literacy skills on university students’ information seeking behaviour and lifelong learning. Heliyon, 9(8), Article e18427. https://doi.org/10.1016/j.heliyon.2023.e18427

84.

Taylor

N. G.

Jaeger

P. T.

(2022). Foundations of information literacy. American Library Association.

85.

Tondeur

van Braak

Ertmer

P. A.

Ottenbreit-Leftwich

(2017). Understanding the relationship between teachers’ pedagogical beliefs and technology use in education: A systematic review of qualitative evidence. Educational Technology Research and Development, 65, 555–575. https://doi.org/10.1007/s11423-016-9481-2

86.

Troyer

Kim

J. S.

Hale

Wantchekon

K. A.

Armstrong

(2019). Relations among intrinsic and extrinsic reading motivation, reading amount, and comprehension: A conceptual replication. Reading and Writing, 32, 1197–1218. https://doi.org/10.1007/s11145-018-9907-9

87.

UNESCO. (2021). Reimagining our futures together: A new social contract for education. UNESCO. https://doi.org/10.54675/ASRB4722

88.

United Marketing Research. (2020). Xiāng zhèn shì qū péng wèi fā zhăn fēn lèi bào gào [Report on the classification of digital development in rural and urban Area]. Commissioned Research Report of the Taiwan National Development Council. Retrieved May 1, 2025, from https://www-api.moda.gov.tw/File/Get/moda/zh-tw/a8GFGMlOV35oeoY

89.

Unrau

N. J.

Rueda

Son

Polanin

J. R.

Lundeen

R. J.

Muraszewski

A. K.

(2018). Can reading self-efficacy be modified? A meta-analysis of the impact of interventions on reading self-efficacy. Review of Educational Research, 88(2), 167–204. https://doi.org/10.3102/0034654317743199

90.

Wang

J. H.-Y.

Guthrie

J. T.

(2004). Modeling the effects of intrinsic motivation, extrinsic motivation, amount of reading, and past reading achievement on text comprehension between U.S. and Chinese students. Reading Research Quarterly, 39(2), 162–186. https://doi.org/10.1598/RRQ.39.2.2

91.

Warm

T. A.

(1989). Weighted likelihood estimation of ability in item response theory. Psychometrika, 54(3), 427–450. https://doi.org/10.1007/BF02294627

92.

Wigfield

Gladstone

J. R.

Turci

(2016). Beyond cognition: Reading motivation and reading comprehension. Child Development Perspectives, 10(3), 190–195. https://doi.org/10.1111/cdep.12184

93.

Wigfield

Guthrie

J. T.

(1997). Relations of children’s motivation for reading to the amount and breadth or their reading. Journal of Educational Psychology, 89(3), 420–432.

94.

Woolley

(2011). Reading comprehension. In Woolley

(Ed.), Reading comprehension: Assisting children with learning difficulties (pp. 15–34). Springer. https://doi.org/10.1007/978-94-007-1174-7_2

95.

Wright

B. D.

Linacre

J. M.

(1994). Reasonable mean-square fit values. Rasch Measurement Transactions, 8(3), 370.

96.

C.-I.

Jou

Y.-H.

Lin

T.-S.

Huang

L.-W.

Chen

H.-J.

Chen

Y.-F.

C.-W.

Tsai

M.-H.

Chen

K.-Z.

Hsieh

J.-C.

Tsai

H.-F.

(2023). Taiwan i-Generation Panel Study (1/5). Commissioned Research Program of National Science and Technology Council.

97.

Yuan

Y.-H.

Liu

C.-H.

Kuang

S.-S.

(2021). An innovative and interactive teaching model for cultivating talent’s digital literacy in decision making, sustainability, and computational thinking. Sustainability, 13(9), Article 5117. https://doi.org/10.3390/su13095117

98.

Zhang

(2021). The influence of computer information activities of middle school students on reading literacy based on interactive behavior theory. Open Journal of Social Sciences, 9(2), Article 107384. https://doi.org/10.4236/jss.2021.92027

99.

Zheng

J.-Q.

Cheung

K.-C.

Sit

P.-S.

Lam

C.-C.

(2024). Unraveling the relationship between ESCS and digital reading performance: A multilevel mediation analysis of ICT-related psychological needs. Education and Information Technologies, [Advance online publication]. https://doi.org/10.1007/s10639-024-12458-9

100.

Zhu

Chan

S. D.

Yao

Zhu

(2024). The different effects of the ideal L2 self and intrinsic motivation on reading performance via engagement among young Chinese second-language learners. International Review of Applied Linguistics in Language Teaching, 62(4), 1539–1560. https://doi.org/10.1515/iral-2022-0189