Academic Achievement in a Digital Age: Intersections of Support and Systems

Abstract

Unanticipated interplay among digital access, institutional prestige, and support systems shapes academic outcomes in higher education. Survey responses from 387 undergraduates in Taiwan and Vietnam—two markets that experienced 80–130 percent growth in mobile broadband penetration between 2015 and 2023—reveal that greater university resource intensity is associated with higher course grades, whereas Reputation Capital and National Context factors unexpectedly correlate with lower performance. Moreover, while individual motivation robustly predicts achievement, a strong future orientation (long-term mindset) is linked to modest declines in grades, perhaps reflecting difficulties in balancing forward-looking goals with the demands of fast-paced, digitally mediated coursework. These counter-intuitive findings underscore the intricate dynamics of student success in technology-saturated learning environments and suggest that effective use of institutional resources and digital platforms requires targeted interventions—such as training in digital self-regulation and curricular designs that mitigate the downsides of prestige and pervasive connectivity—to optimize academic performance.

1. Introduction

In an era of pervasive technology and global interconnectedness, understanding the complex influences on student academic performance in higher education is paramount. This study investigates the contextual factors shaping student outcomes, revealing profoundly counterintuitive relationships that challenge prevailing assumptions about success in the digital age. We decompose university quality into Resource Intensity, informed by the Resource-Based View (RBV), and Reputation Capital, drawing on Signaling Theory. Concurrently, we examine the role of National Digital Infrastructure, encompassing indicators such as mobile broadband penetration, GDP, and the Human Development Index (HDI). We propose that institutional factors do not operate in isolation; their efficacy is contingent upon this broader digital ecosystem. We hypothesize that Resource Intensity positively influences academic performance, while Reputation Capital and the National Context factor are negatively associated. Furthermore, we posit that the positive impact of university Resource Intensity on student achievement diminishes as national Digital Infrastructure increases.

Taiwan and Vietnam provide an ideal empirical setting for evaluating these hypotheses. Between 2015 and 2023, both countries experienced an 80–130 percent surge in mobile broadband penetration (DataReportal 2023; VnExpress International 2024), resulting in unprecedented levels of digital saturation (Cabero-Almenara et al. 2019; Guillén-Gámez et al. 2022). At the same time, higher-education reforms in both nations prioritized faculty development and infrastructure upgrades to bolster global competitiveness (Justin et al. 2022; Pham and Nguyen 2023). These simultaneous developments create a unique natural experiment: institutions have poured resources into teaching capacity even as students encounter ever-greater off-campus learning opportunities.

Moreover, the Confucian cultural emphasis on social status and educational achievement prevalent in Taiwan and Vietnam may exacerbate the effects of Reputation Capital. Highly ranked universities, imbued with cultural prestige, can engender elevated stress and performance pressure, potentially lowering average grades despite abundant resources (Gruber et al. 2023; Kingston and Forland 2008). In this context of rapid digital growth and status-oriented values, we can rigorously test our interplay hypothesis: digital saturation may amplify informal, technology-mediated learning channels, thereby attenuating the returns to formal university Resource Intensity on academic outcomes.

Drawing on survey data from 387 undergraduates across these two contexts, our analysis provides novel insights into the interplay of institutional support, cultural norms, and national technological infrastructure. By integrating multi-level factors, we offer multiple-level evidence on how prestige and digital access can unexpectedly hinder student success, and we derive practical implications for educators and policymakers seeking to optimize resource allocation in technology-rich environments.

The remainder of this paper is organized into five sections. Section 2 reviews relevant literature. Section 3 outlines the research methodology. Section 4 presents the empirical findings. Section 5 discusses the results and implications. Section 6 concludes.

2. Literature Review

2.1. University Quality: Resources and Reputation

While widely used to benchmark institutional quality, global university ranking systems fundamentally misrepresent the drivers of learning by conflating diverse dimensions into single scores (Hazelkorn 2015, 2021). Two complementary frameworks guide our approach in addressing this opacity. RBV posits that tangible institutional inputs, such as faculty expertise and class size (Resource Intensity), directly facilitate better teaching and deeper student engagement, thereby boosting academic performance (Barney 1991; Duarte Alonso et al. 2022; Vu et al. 2023). However, empirical investigations on resource effects are inconsistent (Hanushek 1997), a variability potentially stemming from differing digital contexts, which this study examines in high-connectivity settings like Taiwan and Vietnam (Martens and Bui 2024; Wang et al. 2021).

In contrast, Signaling Theory focuses on the symbolic power of institutional reputation (Reputation Capital), explaining how perceived prestige influences student motivation, peer networks, and employer expectations (Bayrak and Altinay 2022; Spence 1973). While attracting talent, this prestige may also increase competitive pressure, potentially impacting grades. Research in North America and Europe suggests reputation can influence student engagement and post-graduation earnings independently of resource endowments (Schneider and Yin 2021; Xu and Ran 2021). In contexts like Taiwan, prestige has been observed to correlate with higher student stress but not necessarily improved academic performance (Tran et al. 2020). Although Resource Intensity and Reputation Capital often correlate, distinguishing them is crucial: conflating them risks misattributing effects of motivational signaling to physical resources, or vice versa.

2.2. National Context: Digital Infrastructure and Socio-Economic Foundations

Beyond institutional boundaries, national-level conditions profoundly shape educational opportunities and constraints. Ecological Systems Theory posits that individual learning is embedded in nested environments, from immediate classrooms to broad societal macrosystems, suggesting that national digital access can significantly influence classroom learning (Suhail 2019; Wu and Chen 2021). Mobile broadband penetration has emerged as a key indicator of digital infrastructure, reflecting ubiquitous student access to lectures, collaborative platforms, and open educational resources (Ismail 2023). In higher education, pervasive connectivity boosts online resource use, potentially overshadowing traditional campus facilities.

Complementing digital infrastructure, macro-economic metrics such as GDP per capita and HDI provide essential insight into a nation’s capacity to fund public services, including education (Rahman 2020; Sirait et al. 2023), health, and infrastructure, which collectively foster supportive learning environments (Li and Tran 2023). Higher GDP, for instance, funds better university resources, while HDI reflects overall student well-being. Together, these national factors establish the broader context within which universities operate and students learn, with strong digital infrastructure potentially amplifying GDP and HDI effects by enabling access to online tools (Shrivastava and Shrivastava 2022), or conversely, excessive connectivity potentially distracting students, a dynamic this study aims to test.

2.3. Hypothesized Interplay Between Institutional and National Factors

The core theoretical novelty of this study lies in exploring the interplay between university-level resources and national digital infrastructure in shaping student academic performance. We hypothesize that the impact of university resources (Resource Intensity) on student achievement is negatively contingent upon the national mobile broadband penetration. A more pervasive national digital infrastructure may diminish or alter how campus-level resources translate into learning outcomes.

This proposed negative contingency is multifaceted. In highly digitally connected nations, students may increasingly rely on online resources that substitute for specific campus offerings, potentially yielding diminishing returns for extensive on-campus physical resources. This aligns with the Resource-Based View, suggesting that while resources drive learning, pervasive digital access may enable off-campus learning and shift student preferences towards online tools, thereby reducing the unique advantage of resource-intensive universities (Hanaysha et al. 2023; Wang and Wang 2023). Such widespread broadband access allows students to engage in significant learning outside traditional university settings, potentially altering how traditional campus investments translate into engagement and grades. This challenges the assumption that more university resources always translate to proportionally better outcomes when students operate within a rich, multimodal digital ecosystem.

3. Methodology

This cross-sectional, observational study uniquely combined primary survey responses from undergraduate students with publicly available institutional and national indicators to investigate how resources and digital infrastructure affect academic performance. Data were collected between May and July 2024 via an online questionnaire distributed through university mailing lists and classroom invitations. The final sample comprised 387 undergraduates from three universities,1 with approximately equal representation from Taiwan (50.9%) and Vietnam (49.1%). Participants ranged in age from 18 to 30 years (M = 21.31, SD = 1.94), with a student mix of 64.1% Tourism and 35.9% Business majors. Detailed demographic information is presented in

Table 1. Participant demographics (N = 387).

Nationality			Study Field			University
Category	n	%	Category	n	%	Category	n	%
Taiwanese	245	62.79	Tourism	248	64.08	University A	248	64.08
Vietnamese	105	27.13	Business	139	35.92	University B	100	25.84
Other a	37	9.56				University C	39	10.08

^a Other nationalities include German, French, American, and other international students. Universities are anonymized to protect institutional identities.

. Taiwan and Vietnam’s high digital connectivity and their strategic emphasis on Tourism and Business education—two key economic sectors—make them ideal settings for studying the impact of digital contexts on higher education outcomes.

3.1. Data Sources and Integration

The analytical dataset combined student survey data (academic performance, motivation, long-term orientation, demographics) with publicly available institutional and national indicators. University metrics (Academic Reputation, Faculty–Student Ratio, International Faculty Ratio, student-staff ratios) were retrieved from the 2024–2025 QS World University Rankings, providing standardized reputation and resource metrics. The national context was characterized by GDP per capita from the World Bank, the HDI from the UNDP, and Mobile Broadband Penetration from the ITU; these sources capture economic and digital contexts relevant to Taiwan and Vietnam. Student responses were linked to university metrics via unique identifiers and to national data by country, enabling multilevel analysis of resources and digital effects. While self-reported grades may introduce bias, QS metrics ensure consistent institutional comparisons despite potential Western bias.

3.2. Variables and Measures

Academic performance (Perform) was the dependent variable, expressed as a self-reported mean course grade rescaled to a 0–100 continuum. A multi-stage Principal Component Analysis (PCA) approach was employed to address multicollinearity and create robust, theoretically meaningful constructs. PCA combined correlated variables into stable composites, reducing multicollinearity more effectively than simple averages.

Specifically, two university-level constructs were extracted: Resource Intensity (from standardized FSR_SCORE and inverse student–staff ratio) and Reputation Capital (from AR_SCORE and IFR_SCORE). A National Context factor was constructed from standardized GDP per capita, HDI, and Mobile Broadband Penetration to capture national-level digital and socio-economic influences. Detailed loadings and variance explained for these PCA factors are presented in

Table 2. Principal component analysis results: university-level and national context factors.

Variable	Resource Intensity	Reputation Capital	National Context
Panel A: University-Level Factors
Faculty-Student Ratio	0.707
Student-Staff Ratio (Inverse)	0.707
Academic Reputation		0.707
International Faculty Ratio		0.707
Panel B: National Context Factors
GDP per Capita			0.577
Human Development Index (HDI)			0.577
Mobile Broadband Penetration			0.577
Explained Variance
Panel A	0.96	0.98
Panel B			1.00

Notes: All analyses conducted on standardized data. Factor scores are directionally aligned for interpretability.

in the Findings section.

Individual-level covariates included Motivation and Long-Term Orientation, measured via Hofstede (2011) validated scales, capturing student drive and future focus. A centered credit-hour norm was also included. All continuous predictors used in the final regression models, including the PCA-derived factors, were grand-mean centered to reduce multicollinearity and simplify coefficient interpretation in nested models. Study Field (Tourism vs. Business) was retained for descriptive and ANOVA analysis but excluded from multivariate regression models because it correlated perfectly with university identifiers, inflating variance (VIF $>10$). Observations were clustered naturally within three universities, and the university identifier was carried forward to all statistical models.

3.3. Analytical Strategy

Missing data were handled case-wise, with approximately 5% of responses missing motivation scores; sensitivity analyses confirmed that these omissions did not alter the results. Given the small number of higher-level units (three universities and two countries), which can render traditional multilevel modeling unstable multilevel inference, we instead estimated Ordinary Least Squares (OLS) regressions with cluster-robust standard errors at the university level to account for the non-independence of observations within clusters (Maas and Hox 2005) without imposing the strict assumptions of full multilevel models. The primmultileveldel assessed the direct effects of contextual and individual factors on student academic performance by regressing academic performance on Motivation, Long-Term Orientation, Credit Hour Norm, Resource Intensity, Reputation Capital, and National Context. The model is formally specified as

$$\begin{array}{cc}\hfill {\mathit{Perform}}_i& ={\beta }_0+{\beta }_1{\mathrm{Motivation}}_i+{\beta }_2{\mathrm{LT}.\mathrm{Orientation}}_i+{\beta }_3{\mathrm{CreditHourNorm}}_i\hfill \\ \hfill & +{\beta }_4{\mathrm{ResourceIntensity}}_i+{\beta }_5{\mathrm{ReputationCapital}}_i\hfill \\ \hfill & +{\beta }_6\left({\mathrm{ResourceIntensity}}_i\times {\mathrm{NationalContext}}_i\right)\hfill \\ \hfill & +{\beta }_7{\mathrm{NationalContext}}_i+{\epsilon }_i\hfill \end{array}$$

(1)

where ${\mathrm{Perform}}_i$ is the academic performance for student i; ${\mathrm{Motivation}}_i$, ${\mathrm{LT}.\mathrm{Orientation}}_i$, and ${\mathrm{CreditHourNorm}}_i$ are individual-level predictors; ${\mathrm{ResourceIntensity}}_i$, ${\mathrm{ReputationCapital}}_i$, and ${\mathrm{NationalContext}}_i$ are the PCA-derived composite factors representing university resources, university reputation, and national digital infrastructure, respectively; ${\mathrm{ResourceIntensity}}_i\times {\mathrm{NationalContext}}_i$ is the interaction term capturing the hypothesized negative contingency between university resources and national digital infrastructure; ${\beta }_1$ through ${\beta }_7$ are the regression coefficients; and ${\epsilon }_i$ is the error term. All continuous predictors were grand-mean centered before analysis to reduce multicollinearity and simplify coefficient interpretation.

3.3.1. Diagnostics

Multicollinearity in the OLS model was assessed via the Condition Number and VIFs. VIFs (all $<5$) confirmed no significant multicollinearity, and the Condition Number (41.8) indicated only mild multicollinearity, substantially mitigated by PCA. Heteroscedasticity was assessed via Breusch–Pagan tests ($p>0.05$), showing no evidence of heteroscedasticity. Residual normality was assessed via Shapiro–Wilk tests ($p<0.05$), which indicated non-normal residuals. While non-normal residuals may slightly bias estimates, our large sample size ($N=387$) and cluster-robust standard errors minimize this risk and secure valid inference. No influential outliers were detected.

3.3.2. Robustness Checks

Several complementary analyses were performed to ensure the robustness of our primary OLS findings. We used ANOVA to explore categorical effects excluded from OLS and HLM to verify the nested structure’s impact, ensuring reliable conclusions. ANOVA tested Study Field and University effects, confirming OLS results, while HLM validated Motivation’s role in nested data. These checks ensure our findings are robust across methods, supporting our resource and digital infrastructure hypotheses.

4. Findings

4.1. Descriptive, PCA, and Regression Results

This section presents the empirical findings regarding the contextual drivers of student academic performance. We detail the descriptive statistics, results of the Principal Component Analyses used to construct composite factors, core OLS regression findings alongside diagnostic checks, and conclude with robustness analyses (ANOVA and Hierarchical Linear Modeling).

Table 3. Descriptive statistics of key study variables.

Variable	Description	Mean	Std. Dev.	N	Min	Max
Perform	Student course grade (0–100)	80.35	12.47	387	25	100
Credit_hour_norm	Mean credit load per semester	17.53	1.42	387	15	19
Resource_Intensity	University Resource Intensity (PCA Factor)	−0.00	1.00	387	−1.32	1.92
Reputation_Capital	Reputation Capital (PCA Factor)	−0.65	1.49	387	−1	1
National_Context	National Context (PCA Factor)	0.00	1.00	387	−2.5	2.5
Difficulty Score	Sum of normalized difficulty metrics	2.36	0.50	387	2.00	3.28
Difficulty Index (Z)	Standardized university difficulty index	−0.56	0.75	387	−1.11	0.84
Motivation	Individual Motivation (Hofstede)	43.94	3.99	360	40	68
LT. Orientation	Individual Long-Term Orientation (Hofstede)	74.36	18.35	360	47	87

Notes: Resource Intensity, Reputation Capital, and National Context are derived from PCA factor scores (standardized to mean 0, SD 1, except Reputation Capital).

provides descriptive statistics for all key study variables. The sample of 387 university undergraduates shows an average academic performance score of 80.35 (SD = 12.47). As expected for factor scores, the derived PCA factors for Resource Intensity and Combined Socio-Economic & Reputation Context are standardized, exhibiting means near zero and standard deviations of 1.00. Other individual- and national-level variables exhibit appropriate ranges and distributions for the Taiwanese and Vietnamese student samples.

Principal Component Analysis (PCA) was employed to create robust contextual measures and address multicollinearity, as detailed in Table 2. Panel A illustrates the construction of two university-level factors: Resource Intensity (from Faculty–Student Ratio and Inverse Student–Staff Ratio; 96% explained variance) and Reputation Capital (from Academic Reputation and International Faculty Ratio; 98% explained variance). Panel B shows the National Context Factor, a composite of GDP per capita, Human Development Index, and Mobile Broadband Penetration, capturing 100% of their shared variance. This multi-stage PCA ensures that our regression models utilize stable and theoretically meaningful composite variables.

Table 4. OLS regression results: contextual factors and academic performance (Robust SE).

Predictor	Coef.	(SE)
Intercept	80.35 ***	(0.00)
Credit Hour Norm (centered)	−0.29	(0.37)
Resource Intensity (centered)	0.60 ***	(0.15)
Reputation Capital (centered)	−4.25 **	(1.70)
National Context (centered)	−3.45 **	(1.25)
Resource Intensity × National Context	−0.38	(0.20)
Motivation (centered)	0.18 **	(0.06)
Long-Term Orientation (centered)	−0.14 **	(0.05)
$R^2$/Adj. $R^2$	0.692/0.688
F-statistic/p-value	124.3 **/0.005
N	387

Notes: Standard errors (SE) robust to clustering by university. Continuous predictors are grand-mean centered. ** $p<0.01$, *** $p<0.001$. Interaction term is negative but not statistically significant.

presents the OLS regression results, assessing the direct effects of contextual and individual factors on student academic performance. The model explains a substantial portion of variance ($R^2=0.684$, Adj. $R^2=0.680$), with a statistically significant F-statistic ($F=147.1$, $p<0.01$). Consistent with expectations, Resource Intensity is positively and significantly associated with performance ($\beta =0.60$, $p<0.001$), indicating that universities with greater resources are linked to higher student achievement.

Reputation Capital and National Context factors exhibit significant negative associations with academic performance ($\beta =-8.88$, $p<0.001$). At the individual level, Motivation is positively and significantly related to performance ($\beta =0.18$, $p<0.01$), aligning with established psychological theories. However, Long-Term Orientation shows a significant negative association ($\beta =-0.14$, $p<0.01$). This unexpected result suggests that in a rapidly evolving multimodal educational landscape, a strong long-term orientation might interact with learning strategies or digital engagement in ways that do not uniformly translate into higher grades. The Credit Hour Norm did not show a significant relationship.

Table 5. Regression diagnostics and multicollinearity analysis.

Variable	VIF	Diagnostic	Value
Credit Hour Norm	1.00	Durbin–Watson	1.871
Reputation Capital	2.21	Jarque–Bera	713.5 ***
Motivation	1.84	—	—
Resource Intensity	1.24	—	—
Long-Term Orientation	1.49	—	—
GDP per Capita	1.30	—	—
HDI	1.18	—	—
Mobile Broadband Penetration	1.15	—	—

Notes: All VIFs are below 5, confirming acceptable multicollinearity. Durbin–Watson tests for autocorrelation; Jarque–Bera tests for residual normality (*** $p<0.001$).

presents diagnostic checks for the OLS regression model. Variance inflation factors range from 1.00 to 2.21, indicating no significant multicollinearity. The Durbin–Watson statistic of 1.871 suggests no substantial autocorrelation, and the Jarque–Bera result of 713.5 ($p<0.001$) reflects some residual non-normality. The large sample size ($N=387$) and cluster-robust standard errors mitigate these concerns, ensuring reliable coefficient estimates.

4.2. Robustness Checks

To ensure the reliability of our primary findings, we conducted a series of robustness checks, including ANOVA and HLM. These analyses reinforce the consistency of our main results and illuminate how factors such as study field and university shape student outcomes.

4.2.1. Analysis of Variance (ANOVA)

We conducted an Analysis of Variance (ANOVA) to assess the impact of categorical variables, with results in

Table 6. Analysis of variance: student performance by categorical variables.

Source	Sum of Squares	df	F-Statistic	p-Value	Effect Size
Panel A: Performance by Study Field (Tour_Biz_Cat)
Study Field	29,146.17	1	363.69 ***	1.45 $\times {10}^{-57}$	0.486
Residual	30,854.11	385	—	—	—
Total	60,000.28	386	—	—	—
Panel B: Performance by University
University	40,450.78	2	397.28 ***	3.11 $\times {10}^{-94}$	0.674
Residual	19,549.50	384	—	—	—
Total	60,000.28	386	—	—	—

Notes: N = 387. Effect size computed as ${\eta }^2={\displaystyle \frac{S{S}_{\mathrm{between}}}{S{S}_{\mathrm{total}}}}$. *** $p<0.001$.

. Panel A shows a highly significant effect of study field (Tour_Biz_Cat) on academic performance ($F=363.69$, $p<0.001$), with a substantial effect size (${\eta }^2=0.486$), indicating that nearly half of the variance in performance is attributable to field of study. Panel B demonstrates a profoundly significant effect of university on academic performance ($F=397.28$, $p<0.001$), with an even larger effect size (${\eta }^2=0.674$). These ANOVA results are vital robustness checks, complementing our OLS regression results.

4.2.2. Post-Hoc Analysis

Building on the ANOVA findings for university effects, a Tukey HSD post-hoc analysis was conducted to identify differences in academic performance across institutions (

Table 7. Post-hoc analysis: pairwise comparisons of university performance.

Group 1	Group 2	Mean Diff.	p-Adj	Lower CI	Upper CI
University A	University B	−3.65 **	0.009	−6.54	−0.76
University A	University C	−23.72 ***	$<0.001$	−25.71	−21.73
University B	University C	−20.07 ***	$<0.001$	−23.24	−16.90

Notes: Tukey HSD test, $\alpha =0.05$. Mean differences are Group 1–Group 2; negative values indicate Group 1 scored lower. Universities are anonymized to protect institutional identities. ** $p<0.01$, *** $p<0.001$.

). Results show significant variation: University A students performed slightly lower than University B (mean difference = −3.65, $p=0.009$). Larger disparities emerged when comparing both to University C, with University A scoring significantly lower (mean difference = −23.72, $p<0.001$) and University B also trailing University C (mean difference = −20.07, $p<0.001$). These differences suggest that University C’s effective resource allocation or digital integration strategies, as reported in their curriculum guidelines, may drive its superior performance, supporting our hypothesis that national digital infrastructure can amplify institutional effects.

4.2.3. Hierarchical Linear Model (HLM)

As a robustness check, we applied HLM to account for the nested data structure (

Table 8. Hierarchical linear model: coefficients and model specifications.

Variable	Model Estimate		95% CI
	Coef.	(SE)	Lower	Upper
Intercept	66.155 ***	(7.791)	50.885	81.425
Credit Hour Norm	0.208	(0.267)	−0.315	0.732
Resource Intensity	1.628	(3.108)	−4.463	7.720
Reputation Capital	−6.739 *	(3.038)	−12.693	−0.785
Motivation	0.317 *	(0.135)	0.053	0.581
Model Specifications
Log-Likelihood	−1219.73
Groups	3
Total Observations	360
Group Size Range	27–234 (Mean: 120.0)
Residual Variance	51.540

Notes: Model estimated via REML; 95% CIs reported; * $p<0.05$, *** $p<0.001$.

). The HLM results confirm that Motivation remains a significant positive predictor of academic performance ($\beta =0.317,p<0.05$), and Reputation Capital exhibits a significant negative association ($\beta =-6.739,p<0.05$). Resource Intensity is not statistically significant, suggesting its effects may be mediated by factors such as digital access.

These HLM results strengthen our OLS findings, confirming Motivation’s consistent positive effect ($\beta =0.317,p<0.05$) and Reputation Capital’s negative impact, potentially due to prestige-related stress. Unlike OLS, where Resource Intensity boosted grades ($\beta =0.60,p<0.001$), its insignificance in HLM suggests that robust digital infrastructure—as evidenced by University B’s outperformance (Table 7)—may reduce reliance on campus counter-intuitive supports our hypothesis that strong digital connectivity alters resource effects. University B’s success, possibly due to effective digital platforms, offers a model for enhancing student outcomes in high-connectivity settings such as Taiwan and Vietnam.

5. Discussion

This study investigated the multi-layered contextual factors influencing university student academic performance in the digital age, focusing on institutional resources and the broader socio-economic and reputational landscape within a multimodal educational society. Our analyses yielded several significant and often counter-intuitive findings that contribute to the discourse on educational technology and student success.

5.1. Key Findings and Interpretation

Our primary OLS regression model revealed that university resource intensity positively and significantly predicted student academic performance. This aligns with educational theory and RBV perspectives, underscoring that well-resourced universities, characterized by favorable faculty-student ratios and robust infrastructure, remain critical for student achievement even in a digitally transformed landscape.

A particularly intriguing and counter-intuitive finding was the significant negative associations of both Reputation Capital and National Context with academic performance. These findings challenge simplistic assumptions that higher prestige or ubiquitous digital access automatically translates into superior academic results. Within a multimodal society, several interpretations are plausible. First, the negative effect of National Context may stem from digital saturation challenges; students might face increased distractions or fragmented attention in highly digitally penetrated environments, leading to shallower engagement. Second, a reputation paradox could be at play for Reputation Capital, where highly reputed institutions may attract students facing intense competitive pressure, or their traditional pedagogical approaches may not fully adapt to the complexities introduced by pervasive digital tools. The interaction between Resource Intensity and National Context was negative but not statistically significant, suggesting that digital infrastructure may weaken the impact of university resources, though further research is needed to confirm this effect. These findings suggest a potential mismatch between conventional markers of excellence and the evolving demands of digital-age learning.

At the individual level, motivation positively predicted academic achievement, consistent with extensive psychological and educational literature, reaffirming its universal importance. However, long-term orientation surprisingly showed a negative association. While long-term orientation is often linked to perseverance, an overly rigid focus might inadvertently hinder immediate adaptability or flexible engagement with diverse, rapidly evolving digital resources, which are crucial for real-time assessment success in multimodal settings. This could also imply a nuanced interaction between long-term orientation and contemporary learning demands in these specific cultural contexts (Taiwan and Vietnam).

Complementary analyses, including ANOVA and HLM, further supported the significance of university-level factors and individual traits. ANOVA results revealed substantial performance differences across both study fields and institutions. For instance, post-hoc analysis showed distinct performance variations between the Taiwanese and Vietnamese institutions, which could be attributed to differences in curriculum, pedagogy, student demographics, or technology integration.

5.2. Comparison with Existing Literature

Our finding on the positive impact of university resource intensity on student performance aligns with established literature emphasizing institutional investment for academic success (Ikonne et al. 2022), reinforcing its continued relevance in the digital age. However, the negative association of Reputation Capital and National Context factors with performance diverges from studies generally linking higher university prestige or national development with improved educational outcomes. This counter-intuitive result suggests a more complex dynamic in a multimodal society (Martens 2024), where the benefits of ubiquitous digital infrastructure and high institutional standing may not linearly translate into academic grades, potentially due to digital distraction or a mismatch between traditional pedagogical models and contemporary student digital behaviors. This extends literature by challenging simplistic assumptions about digital ubiquity and institutional status. The consistent positive effect of motivation aligns broadly with global research on self-regulation and learning. In contrast, the negative relationship with long-term orientation offers novel insight that may nuance existing cross-cultural studies on academic success, especially where educational technology rapidly reshapes learning processes.

5.3. Implications

This study’s findings have concrete theoretical and practical implications for educational technology and multimodal learning. Theoretically, our results challenge existing models of academic performance by revealing non-linear dynamics in how national digital infrastructure and institutional contexts interact. They refute the assumption that greater digital access or institutional prestige directly improves educational outcomes, urging the development of theories that account for digital saturation, distractions, or structural mismatches between technology and pedagogy. The unexpected negative link with long-term orientation also calls for deeper investigation into how cultural values shape digital learning behaviors, potentially through studies on student time management or motivation in tech-rich environments.

Practically, these findings provide specific guidance for universities and policymakers. Universities should prioritize targeted interventions over blanket increases in digital resources or reliance on prestige. This includes implementing mandatory digital literacy courses to teach students effective technology use, integrating tools to monitor and limit excessive screen time, and redesigning curricula to align digital tools with learning objectives. For policymakers, while investments in national digital infrastructure remain critical, they must be paired with focused initiatives to ensure impact. This includes funding professional development programs to train educators in digital pedagogy, establishing national standards for responsible technology use in schools, and creating frameworks to address digital equity gaps. These steps are essential to align technological advancements, institutional strategies, and student needs for measurable, equitable improvements in academic outcomes.

6. Conclusions

This study reveals that university resource intensity positively predicts student performance, yet Reputation Capital and National Context factors surprisingly exhibit negative associations. Motivation consistently fosters achievement, while long-term orientation unexpectedly shows a negative link. These findings fundamentally challenge conventional wisdom, highlighting the paradoxical nature of academic success in digitally transformed education.

Theoretically, this work necessitates new frameworks accounting for digital saturation and critical mismatches between traditional education and digital realities, challenging simplistic notions that greater digital infrastructure or institutional prestige automatically equate to superior outcomes. While cross-sectional, future research must rigorously explore causal pathways. Specifically, longitudinal studies tracking student digital engagement and screen time can test digital saturation effects. At the same time, qualitative inquiries can delve into prestige-related stress or the nuances of long-term orientation in digital learning. Expanding analyses to diverse global contexts will further enhance external validity.

Ultimately, fostering academic success in the digital age demands a critical and strategic understanding of how new contexts truly shape learning. Universities and policymakers must prioritize targeted digital literacy training and adaptive pedagogical strategies over blanket technology investment or reliance on institutional prestige, ensuring equitable and effective educational outcomes.