This study investigates student satisfaction with blended learning in Indian higher education, examining how demographic, academic and technological factors shape satisfaction and identifying which dimensions of the learners' satisfaction in blended learning scale (LSBLS) most strongly predict overall satisfaction. The study positions its findings as applicable to resource-constrained higher education contexts across Asia and the Global South.
A quantitative cross-sectional survey was conducted with 210 undergraduate and postgraduate students from three institutions in Uttar Dinajpur district, India. The self-developed LSBLS (26 items; five dimensions) was validated through exploratory factor analysis and demonstrated excellent reliability (α = 0.92). Four research questions guided the study, addressed through independent samples t-tests, one-way ANOVA and multiple linear regression supplemented with Cohen's d and η2 effect sizes.
Students reported moderate overall satisfaction. Access-layer variables – residence, Internet accessibility and device type – produced the most practically significant group differences, while demographic variables (gender, family type and course level) produced negligible effects. Regression analysis identified Interaction and Course Administration as the dominant satisfaction predictors, followed by Learning Outcomes, Technological Support and Assessment, supporting the proposed two-layer distinction between access-enabling and pedagogy-determining factors.
This study contributes the validated LSBLS instrument and a theoretically grounded two-layer satisfaction model extending TAM and the Community of Inquiry framework. The model proposes that access conditions enable participation while pedagogical quality – mediated by learner engagement – determines its ceiling, offering institutional decision-makers an empirically supported priority hierarchy applicable to open and distance learning contexts worldwide.
Introduction
The rapid integration of technology into higher education has established blended learning – the systematic combination of face-to-face instruction with online environments – as a central pedagogical model for enhancing flexibility, accessibility and learner engagement (Garrison and Vaughan, 2008; Graham, 2013). Student satisfaction, widely recognised as a key indicator of instructional quality and sustainability in technology-mediated settings, has become an essential lens for evaluating blended learning effectiveness (Kuo et al., 2014). For open and distance learning (ODL) institutions in particular, where learners are often geographically dispersed and technologically constrained, understanding what drives satisfaction is a matter of institutional viability as well as educational quality.
In India, blended learning adoption accelerated following the COVID-19 pandemic and has been institutionally reinforced by the National Education Policy 2020; Ministry of Education (2020). Yet what determines satisfaction in blended learning remains contested: studies consistently identify interaction quality and technological access as critical factors (Wang et al., 2023; Nyathi, 2024; Qamar et al., 2024), but the relative predictive weight of pedagogical versus infrastructural determinants has not been rigorously examined in district-level Indian settings. Crucially, satisfaction does not arise in isolation – it is shaped by learner engagement, which mediates the relationship between instructional quality and perceived outcomes (Fredricks et al., 2004; Hrastinski, 2019). The student profile of district-level settings – predominantly rural, smartphone-dependent, with intermittent Internet access – mirrors that of open university learners across South and Southeast Asia, sub-Saharan Africa and Latin America, making findings from this context broadly relevant to the international ODL community.
This study addresses these gaps by examining blended learning satisfaction among 210 higher education students in Uttar Dinajpur district, West Bengal, using the self-developed learners' satisfaction in blended learning scale (LSBLS). Integrating exploratory factor analysis (EFA), effect size estimation and multiple regression, the study produces both a validated measurement instrument and a theoretically grounded two-layer satisfaction model. Four research questions guide the study.
What is the overall level of postgraduate students' satisfaction with blended learning, as measured by the LSBLS and how is this satisfaction structured across its dimensions?
Do postgraduate students' satisfaction with blended learning differ meaningfully across access-related variables (residence, Internet accessibility and device type) and background variables (gender, family type, stream, course level and prior experience)?
Which LSBLS dimensions most strongly predict overall satisfaction, and do these results support distinguishing between access-enabling and pedagogy-determining factors?
What do these findings suggest for blended and open learning institutions in resource-constrained settings, particularly for learner engagement and educational equity?
Blended learning satisfaction, theoretical foundations and conceptual framework
Multidimensional nature of blended learning satisfaction
Student satisfaction in blended learning is consistently conceptualised as a multidimensional construct shaped by instructional design, interaction quality, technological adequacy and assessment practices (Diep et al., 2017; Al-Fraihat et al., 2020; Rahman et al., 2015; Sembiring, 2018). The central debate concerns the relative weight of these dimensions: some studies foreground technological access and system quality (Al-Fraihat et al., 2020; Khan et al., 2025), while others emphasise pedagogical interaction and course organisation (Wang et al., 2023; Nyathi, 2024). Wang et al. (2023) found teacher–student interaction to be the most robust predictor across multiple Chinese institutions; Nyathi (2024), studying sub-Saharan African universities, demonstrated that social presence and IT infrastructure jointly determined satisfaction; and Qamar et al. (2024) confirmed that blended and online learning indicators were significantly more predictive than face-to-face components in Indian post-COVID classrooms. Wong (2024) found that students in agile-blended environments most valued interactional quality over technological features, consistent with a broader programme of research on agile-blended pedagogy that has evaluated hybrid teaching practices from academics' perspectives (Li et al., 2023), identified persistent pedagogical challenges in blended course design (Li, 2013) and formalised these insights into a dedicated pedagogical framework (Li and Wong, 2025). Critically, satisfaction does not emerge directly from instructional conditions – it is mediated by learner engagement, defined as the cognitive, behavioural and emotional investment students make in their learning activities (Fredricks et al., 2004). When instruction is interactive and well-administered, students engage more deeply; deeper engagement produces higher perceived learning quality and, consequently, greater satisfaction. Despite this theoretical linkage, none of the studies above simultaneously employed validated instruments, effect size estimation across demographic subgroups and regression-based dimension ranking in a district-level Indian setting – the specific analytical gap addressed here.
Theoretical foundations: TAM, CoI, engagement and the two-layer model
Four complementary frameworks underpin the study's design and collectively motivate the two-layer satisfaction model. The Technology Acceptance Model (TAM; Davis, 1989; Venkatesh and Davis, 2000) posits that perceived usefulness and ease of use determine technology adoption and satisfaction – directly relevant to RQ2's investigation of Internet accessibility and device type. UTAUT (Venkatesh et al., 2003) extends TAM by including facilitating conditions, capturing institutional infrastructure as a satisfaction determinant. Kampa (2023), applying TAM to Indian ODL students, confirmed that technology optimism predicted ease of use, which in turn determined satisfaction. Khan et al. (2025), studying distance learners in Pakistan, further confirmed that system quality and service quality are key LMS predictors. These findings establish the empirical basis for the access layer.
The Community of Inquiry (CoI) framework (Garrison et al., 2000) conceptualises effective blended learning as the intersection of cognitive, social and teaching presence – directly operationalised in the LSBLS dimensions of Interaction (social presence) and Course Administration (teaching presence). Constructivist theory (Vygotsky, 1978) underscores that learning and satisfaction emerge from active, collaborative knowledge construction. Bridging these frameworks, engagement theory (Fredricks et al., 2004; Hrastinski, 2019) proposes that instructional quality shapes satisfaction through its effect on cognitive, behavioural and emotional engagement: when social and teaching presence are strong, students engage more deeply with content and peers and this engagement generates satisfaction. The two-layer model proposed in this study integrates these four frameworks: the access layer (TAM/UTAUT) determines whether students can participate meaningfully in blended learning; the pedagogy layer (CoI/constructivism), mediated by learner engagement, determines the ceiling of satisfaction once access is secured. This sequential architecture – access first, pedagogy second, engagement as mediator – is the model's distinguishing theoretical contribution and represents a testable extension of prior frameworks.
Research gap, international relevance and ODL context
Blended learning satisfaction studies in Indian higher education remain limited in analytical rigour, geographic diversity and instrument validation (Goyal and Tambe, 2015; Johry et al., 2019). District-level rural settings – where smartphone dependency and intermittent Internet access are the norm – remain virtually unstudied despite representing the majority of Indian higher education enrolments. This profile closely mirrors that of learners served by open universities across Asia: the Indira Gandhi National Open University (IGNOU) serves over three million students; Open University Malaysia and the Korea National Open University operate across similarly dispersed, device-constrained populations. The present study's findings are therefore positioned not as India-specific evidence but as a contextualised test of whether the access-pedagogy-engagement sequence holds in the resource-constrained settings that characterise much of Asian ODL.
Conceptual framework
Figure 1 presents the integrated conceptual framework. The two-layer architecture – the access layer (TAM/UTAUT) and the pedagogy layer (CoI/constructivism) – is the framework's distinguishing feature. Learner engagement is positioned as a mediating construct between the pedagogy layer and student satisfaction: strong teaching and social presence foster cognitive, behavioural and emotional engagement, which in turn generates satisfaction. This mediation pathway is theoretically grounded but requires direct empirical testing in future research. RQ2 tests the access layer; RQ3 tests whether the pedagogy layer dominates the regression, consistent with the engagement-mediated pathway.
The diagram presents a two-layer satisfaction model integrating TAM, UTAUT, CoI, and engagement theory. The context includes Indian higher education transformation, digital infrastructure gaps, and institutional readiness, applicable to ODL institutions across Asia, Africa, and Latin America. The access layer includes technological factors such as internet accessibility, device type, perceived ease of use, facilitating conditions, and institutional infrastructure. The pedagogy layer includes instructional factors such as interaction, course administration, assessment alignment, and learning outcomes. Learner engagement, positioned as a mediating construct, includes cognitive, behavioral, and emotional engagement. Student satisfaction is measured through five dimensions: course administration, interaction, technological support, assessment, and learning outcomes. The outcomes include blended learning quality, learner equity, and ODL institutional improvement.Two-layer satisfaction model integrating TAM, UTAUT, CoI and engagement theory
The diagram presents a two-layer satisfaction model integrating TAM, UTAUT, CoI, and engagement theory. The context includes Indian higher education transformation, digital infrastructure gaps, and institutional readiness, applicable to ODL institutions across Asia, Africa, and Latin America. The access layer includes technological factors such as internet accessibility, device type, perceived ease of use, facilitating conditions, and institutional infrastructure. The pedagogy layer includes instructional factors such as interaction, course administration, assessment alignment, and learning outcomes. Learner engagement, positioned as a mediating construct, includes cognitive, behavioral, and emotional engagement. Student satisfaction is measured through five dimensions: course administration, interaction, technological support, assessment, and learning outcomes. The outcomes include blended learning quality, learner equity, and ODL institutional improvement.Two-layer satisfaction model integrating TAM, UTAUT, CoI and engagement theory
Methodology
Research design and participants
A quantitative cross-sectional survey design was adopted to address RQ1–RQ4 (Creswell and Plano Clark, 2018). Data were collected in 2024–2025 from 210 undergraduate (n = 110) and postgraduate (n = 100) students across three purposively selected institutions in Uttar Dinajpur district, West Bengal: Dr Meghnad Saha College, Surendranath College and Raiganj University – all actively implementing blended learning. Of 220 questionnaires distributed by in-person convenience sampling, 210 were retained after excluding ten for incompleteness or systematic responding (Table 1).
Instrument
The LSBLS was developed in three stages, following established scale-development procedures (DeVellis, 2016), because no validated context-specific instrument was available. First, 42 candidate items across five theoretically grounded dimensions – Course Administration, Interaction, Technological Support, Assessment and Learning Outcomes – were generated from prior instruments (Kuo et al., 2014; Wang et al., 2023). Second, five subject-matter experts reviewed items for relevance and clarity, reducing the pool to 30. Third, pilot testing with 100 students excluded four items with corrected item-total correlations below 0.30, yielding a 26-item, five-point Likert scale (1 = Strongly Disagree to 5 = Strongly Agree; 9 items reverse-scored; maximum total = 130). Dimensions map explicitly to the framework: Interaction and Course Administration operationalise CoI social and teaching presence; Technological Support operationalises TAM/UTAUT facilitation conditions; Learning Outcomes reflects perceived usefulness.
Psychometric properties
The overall LSBLS yielded α = 0.92, exceeding conventional reliability thresholds (Nunnally and Bernstein, 1994); dimension alphas ranged from 0.79 (Assessment) to 0.85 (Interaction). EFA suitability was confirmed (KMO = 0.83; Kaiser, 1974; Bartlett's χ2 = 2,318.47, df = 325, p < 0.001). Principal Axis Factoring with Promax rotation extracted five factors consistent with the a priori structure (loadings: 0.68–0.81; no cross-loadings above 0.40), confirming construct validity (Fabrigar et al., 1999). Full results are in Tables 3 and 4.
Data collection and analysis
Questionnaires were administered offline, face-to-face, following institutional permission; participation was voluntary, anonymous and constituted implicit consent. All analyses used IBM SPSS 23. Distributional assumptions – normality (skewness/kurtosis ±1), homogeneity (Levene's test) and outliers (boxplots) – were confirmed prior to inferential testing. To address RQ2, independent samples t-tests with Cohen's d were conducted for binary group comparisons; one-way ANOVA with η2 examined prior experience. To address RQ3, simultaneous multiple regression identified the relative predictive contribution of each LSBLS dimension to overall satisfaction. Regression assumptions were confirmed: VIF = 1.69–1.93; Durbin–Watson ≈2.0; residual plots confirmed linearity and homoscedasticity.
Results
Participant characteristics
The sample was predominantly female (73.81%), rural (78.57%) and from nuclear families (65.24%). Undergraduate and postgraduate students were nearly equally represented (52.38 vs. 47.62%); Arts stream students constituted 57.14% of the sample. Most reported intermittent Internet connectivity (61.90%) and smartphones as their primary learning device (94.76%). Almost half (48.57%) had less than one year of prior blended learning experience (Table 2).
Demographic and academic characteristics (N = 210)
| Variable | Category | n (%) |
|---|---|---|
| Gender | Male | 55 (26.19%) |
| Female | 155 (73.81%) | |
| Family type | Nuclear | 137 (65.24%) |
| Joint | 73 (34.76%) | |
| Residence | Urban | 45 (21.43%) |
| Rural | 165 (78.57%) | |
| Course level | UG | 110 (52.38%) |
| PG | 100 (47.62%) | |
| Stream | Arts | 120 (57.14%) |
| Science | 90 (42.86%) | |
| Internet access | Always | 76 (36.19%) |
| Sometimes | 130 (61.90%) | |
| Rarely/Never | 4 (1.90%) | |
| Device used | Smartphone | 199 (94.76%) |
| Laptop/Desktop | 11 (5.24%) | |
| BL experience | 0–1 year | 102 (48.57%) |
| 1–2 years | 81 (38.57%) | |
| More than 2 years | 27 (12.86%) |
Note(s): UG = Undergraduate; PG = Postgraduate; BL = Blended Learning
RQ1: LSBLS validation and overall satisfaction profile
The LSBLS demonstrated excellent overall reliability (α = 0.92). Dimension-level alphas: Course Administration (0.83), Interaction (0.85), Technological Support (0.80), Assessment (0.79), Learning Outcomes (0.82). EFA confirmed a clear five-factor solution (Table 3); inter-factor correlations (Table 4) ranged from r = 0.28 to r = 0.50, confirming convergent and discriminant validity (Kline, 2016).
EFA pattern matrix: principal axis factoring, promax rotation (N = 210)
| Item | F1 (CA) | F2 (I) | F3 (TS) | F4 (Asst) | F5 (LO) |
|---|---|---|---|---|---|
| CA_1 | 0.79 | – | – | – | – |
| CA_2 | 0.77 | – | – | – | – |
| CA_3 | 0.74 | – | – | – | – |
| CA_4 | 0.72 | – | – | – | – |
| CA_5 | 0.70 | – | – | – | – |
| I_1 | – | 0.81 | – | – | – |
| I_2 | – | 0.79 | – | – | – |
| I_3 | – | 0.76 | – | – | – |
| I_4 | – | 0.74 | – | – | – |
| I_5 | – | 0.72 | – | – | – |
| TS_1 | – | – | 0.78 | – | – |
| TS_2 | – | – | 0.76 | – | – |
| TS_3 | – | – | 0.74 | – | – |
| TS_4 | – | – | 0.72 | – | – |
| TS_5 | – | – | 0.70 | – | – |
| TS_6 | – | – | 0.68 | – | – |
| Asst_1 | – | – | – | 0.77 | – |
| Asst_2 | – | – | – | 0.75 | – |
| Asst_3 | – | – | – | 0.73 | – |
| Asst_4 | – | – | – | 0.71 | – |
| Asst_5 | – | – | – | 0.69 | – |
| LO_1 | – | – | – | – | 0.80 |
| LO_2 | – | – | – | – | 0.78 |
| LO_3 | – | – | – | – | 0.76 |
| LO_4 | – | – | – | – | 0.73 |
| LO_5 | – | – | – | – | 0.71 |
Note(s): CA = Course Administration; I = Interaction; TS = Technological Support; Asst = Assessment; LO = Learning Outcomes. Loadings ≤0.40 suppressed. α: CA = 0.83; I = 0.85; TS = 0.80; Asst = 0.79; LO = 0.82; overall α = 0.92
Inter-factor correlation matrix (LSBLS)
| Factor | CA | I | TS | Asst | LO |
|---|---|---|---|---|---|
| Course administration (CA) | 1.00 | – | – | – | – |
| Interaction (I) | 0.48 | 1.00 | – | – | – |
| Technological support (TS) | 0.32 | 0.30 | 1.00 | – | – |
| Assessment (Asst) | 0.37 | 0.41 | 0.33 | 1.00 | – |
| Learning outcomes (LO) | 0.46 | 0.50 | 0.29 | 0.43 | 1.00 |
Note(s): r = 0.28–0.50 across all pairs. Convergent and discriminant validity confirmed (Kline, 2016)
The overall LSBLS mean (M = 74.15, SD = 8.81; per-item M = 2.85) falls within the moderate satisfaction band (2.34–3.67). Course Administration (M = 16.84, SD = 2.18) and Interaction (M = 16.71, SD = 2.25) scored highest; Assessment (M = 12.81, SD = 2.41) and Learning Outcomes (M = 12.71, SD = 2.46) lowest; Technological Support showed greatest variability (M = 15.08, SD = 2.87). This pattern – high pedagogical satisfaction, variable technological satisfaction – aligns with the two-layer model prediction and is consistent with comparable resource-constrained contexts (Nyathi, 2024; Qamar et al., 2024). Full statistics are in Table 5.
Overall and dimension-wise satisfaction scores (N = 210)
| Dimension | Items | M | SD | Mean item score |
|---|---|---|---|---|
| Course administration | 5 | 16.84 | 2.18 | 3.37 |
| Interaction | 5 | 16.71 | 2.25 | 3.34 |
| Technological support | 6 | 15.08 | 2.87 | 2.51 |
| Assessment | 5 | 12.81 | 2.41 | 2.56 |
| Learning outcomes | 5 | 12.71 | 2.46 | 2.54 |
| Overall LSBLS | 26 | 74.15 | 8.81 | 2.85 |
Note(s): Maximum score = 130. Per-item M = 2.85 (moderate band: 2.34–3.67)
RQ2: Group differences – access-layer and background variables
Table 6 presents t-test results with Cohen's d. Access-layer variables produced the largest effects: device type (d = 0.76) and Internet accessibility (d = 0.45) yielded the most practically significant differences, directly supporting the framework's prediction that connectivity and device type are primary access-layer determinants. Residence (d = 0.39) and academic stream (d = 0.31) produced small-to-medium effects. Background demographic variables – gender (d = 0.23), family type (d = 0.13) and course level (d = 0.12) – produced trivial to small, non-significant differences, indicating that demographic background does not meaningfully differentiate satisfaction once access conditions are met.
Independent samples t-test results with Cohen's d (Addressing RQ2)
| Variable | Group | n | M | SD | df | Mean difference | t | p | d |
|---|---|---|---|---|---|---|---|---|---|
| Gender | Male | 55 | 75.64 | 9.52 | 208 | 2.01 | 1.46 | >0.05 | 0.23 |
| Female | 155 | 73.62 | 8.52 | ||||||
| Residence | Rural | 165 | 74.88 | 8.75 | 208 | 3.44 | 2.34 | <0.05* | 0.39 |
| Urban | 45 | 71.44 | 8.60 | ||||||
| Family type | Joint | 73 | 74.89 | 9.71 | 208 | 1.13 | 0.89 | >0.05 | 0.13 |
| Nuclear | 137 | 73.75 | 8.31 | ||||||
| Course level | UG | 110 | 74.67 | 8.95 | 208 | 1.10 | 0.90 | >0.05 | 0.12 |
| PG | 100 | 73.57 | 8.66 | ||||||
| Stream | Arts | 120 | 75.30 | 8.92 | 208 | 2.68 | 2.20 | <0.05* | 0.31 |
| Science | 90 | 72.61 | 8.48 | ||||||
| Internet access | Always | 76 | 76.57 | 8.16 | 208 | 3.65 | 3.09 | <0.05* | 0.45 |
| Sometimes | 130 | 72.92 | 8.15 | ||||||
| Device used | Smartphone | 199 | 74.49 | 8.54 | 208 | 6.58 | 2.43 | <0.05* | 0.76 |
| Laptop/Desktop | 11 | 67.91 | 11.51 |
Note(s): d = Cohen's d (pooled SD). Benchmarks: trivial <0.20; small 0.20–0.49; medium 0.50–0.79 (Cohen, 1988). *p < 0.05. Access-layer variables (Internet, Device) produced the largest effects
For prior blended learning experience (ANOVA), satisfaction was nearly identical across groups (0–1 year: M = 73.92; 1–2 years: M = 74.23; >2 years: M = 74.74); the result was non-significant (F(2, 207) = 0.098, p = 0.907, η2 < 0.01). Prior exposure explains less than 1% of satisfaction variance – course quality and access conditions matter substantially more than accumulated experience (Table 7).
RQ3: Predicting overall satisfaction – pedagogy layer dominance
The regression model was significant (F(5, 204) = 44.17, p < 0.001; R2 = 0.52; Adjusted R2 = 0.51), explaining 52% of satisfaction variance (Table 8). The regression pattern directly addresses RQ3 and provides empirical support for the two-layer model's pedagogy-layer prediction: the three pedagogy-layer dimensions – Interaction (β = 0.32), Course Administration (β = 0.30) and Learning Outcomes (β = 0.29) – are the dominant predictors. Technological Support, classified as an access-adjacent dimension, contributed a smaller but significant effect (β = 0.18). Assessment was the weakest predictor (β = 0.15), signalling a persistent evaluation-alignment gap in blended learning design (Boelens et al., 2017). All VIF values were below 2.0, confirming no multicollinearity.
Multiple regression: LSBLS dimensions predicting overall satisfaction (RQ3)
| Predictor | B | SE B | β | t | p | VIF | Layer |
|---|---|---|---|---|---|---|---|
| Interaction | 0.79 | 0.10 | 0.32 | 7.90 | <0.001 | 1.93 | Pedagogy |
| Course administration | 0.74 | 0.09 | 0.30 | 8.22 | <0.001 | 1.87 | Pedagogy |
| Learning outcomes | 0.71 | 0.10 | 0.29 | 7.10 | <0.001 | 1.91 | Pedagogy |
| Technological support | 0.42 | 0.08 | 0.18 | 5.25 | <0.001 | 1.69 | Access-adjacent |
| Assessment | 0.36 | 0.08 | 0.15 | 4.50 | <0.001 | 1.74 | Pedagogy |
| Constant | 4.87 | 1.09 | – | 4.47 | < 0.001 | – | – |
Note(s): F(5, 204) = 44.17, p < 0.001; R2 = 0.52; Adjusted R2 = 0.51. Predictors ordered by β. The Layer column maps each dimension to the two-layer model. All VIF <2.00
Discussion
RQ3: The two-layer model – empirical support and theoretical extension
The two-layer satisfaction model rests on a straightforward but empirically testable claim: access conditions and pedagogical quality operate as sequential, not concurrent, determinants of blended learning satisfaction. The data support this claim. The access layer – device type (d = 0.76), Internet accessibility (d = 0.45) and residence (d = 0.39) – produced the largest group-level differences observed in this study, while the pedagogy layer – Interaction (β = 0.32), Course Administration (β = 0.30), Learning Outcomes (β = 0.29) – dominated the regression. These two patterns are not independent findings; they are theoretically linked by engagement. Access enables participation; participation creates the conditions for interaction and instructional engagement; engagement, in turn, generates satisfaction. That Interaction is the strongest regression predictor across this rural Indian sample, urban Chinese institutions (Wang et al., 2023) and Hong Kong's emergency online education (Wong, 2024) suggests the engagement-mediated pathway is not culturally contingent – it is a structural feature of blended learning wherever social presence is meaningfully delivered.
The two-layer model extends TAM and CoI in a specific and theoretically motivated way. TAM explains why access conditions determine initial adoption and baseline satisfaction; CoI explains why social and teaching presence – when access is sufficient – determine deeper satisfaction. Neither framework accounts for the sequential, threshold nature of this relationship. By formally introducing engagement as the mediating construct between the pedagogy layer and satisfaction, this study proposes a testable extension: once access conditions cross a minimum threshold, pedagogical quality shapes engagement quality, which in turn determines the satisfaction ceiling. This proposition awaits structural equation modelling in future research but is grounded in engagement theory (Fredricks et al., 2004) and consistent with Hrastinski's (2019) synthesis of participation and engagement in online learning.
RQ2: Access-layer effects and equity implications
The near-medium device-type effect (d = 0.76) is the most practically significant group-level finding. Smartphone users reported substantially higher satisfaction than laptop/desktop users (M = 74.49 vs. 67.91). This is not about hardware capability; it reflects socio-technological normalcy. Kampa (2023) found that technology optimism predicted mobile learning ease of use among Indian ODL students – device familiarity compensates for hardware limitations. For the 94.76% of students who learn primarily through smartphones, the mobile interface is not a compromise; it is the norm. Open universities designing content for comparable populations must treat mobile-first architecture as a prerequisite, not an enhancement. The Internet accessibility effect (d = 0.45) reinforces this: reliable connectivity is foundational. Khan et al. (2025) found system quality and service quality to be key LMS predictors for distance learners in Pakistan – a directly analogous ODL context. Without reliable access, even well-designed pedagogy cannot generate high engagement or satisfaction.
The small-to-medium residence effect (d = 0.39) – rural students reporting higher satisfaction than urban peers – reflects a genuine equity dividend. Blended learning disproportionately benefits learners with historically constrained access (UNESCO, 2021): for rural students in Uttar Dinajpur, digital instruction represents a qualitative expansion of educational opportunity unavailable through traditional attendance-only models. IGNOU and comparable ODL institutions should read this not as complacency but as confirmation that their rural target populations respond positively to flexible digital delivery when connectivity is reliable. The absence of significant gender differences (d = 0.23, p > 0.05) is equally important: blended learning does not appear to replicate gender-based educational disadvantages, consistent with Dziuban et al. (2018) and Masrom et al. (2019) across culturally diverse settings.
RQ4: Implications for ODL and blended learning institutions
The regression coefficient hierarchy produces a four-step institutional action sequence. First, address the access layer: mobile-compatible LMS platforms, reliable campus connectivity and data-cost support are prerequisites – not optional enhancements – for equitable blended participation (Khan et al., 2025), consistent with frameworks for technology-enabled institutional readiness in agile-blended contexts (Wong and Li, 2025). Without this, the pedagogy layer cannot operate. Second, invest in interaction design: synchronous check-ins, structured online discussions, rapid instructor feedback and collaborative peer tasks. Interaction is the single highest-return pedagogical investment identified in this study – and in Wang et al. (2023) and Wong (2024). Third, ensure course administration coherence: transparent objectives, integrated online-offline scheduling and consistent expectations reduce cognitive load and allow students to direct their attention to learning rather than navigation. Fourth, reform assessment: formative, criteria-aligned and blended-format-appropriate evaluation is the weakest-scoring and weakest-predicting dimension in this study – a high-impact, low-cost improvement target. Assessment redesign should be treated as the minimum condition for pedagogical coherence, not an afterthought. This four-step sequence is applicable to any institution where the two-layer model operates, including open universities across Asia, Africa and Latin America. National policy frameworks such as India's NEP 2020 should explicitly fund both access-level interventions (connectivity, devices) and pedagogy-level faculty development in parallel, rather than treating digital infrastructure as a precondition for educational investment rather than a component of it (Ministry of Education, 2020; UNESCO, 2021).
Conclusion
This study examined blended learning satisfaction among 210 higher education students in West Bengal through a validated instrument (LSBLS; α = 0.92), EFA, effect size estimation and multiple regression, proposing and empirically supporting a two-layer satisfaction model. The access layer – connectivity, device type and geographic context – determines whether students can participate meaningfully in blended learning. The pedagogy layer – interaction, course administration and learning outcomes, operating through learner engagement – determines the ceiling of satisfaction once access is secured. Interaction is the single most important pedagogical investment. Assessment remains a persistent design gap, and the access-before-pedagogy sequence is the primary institutional planning principle. That rural students reported higher satisfaction than urban peers confirms a genuine equity dividend: well-implemented blended learning expands access for historically marginalised learners, not merely for the already-advantaged. The LSBLS provides a replicable measurement tool and the two-layer model a transferable planning framework, both applicable across the resource-constrained ODL contexts that characterise much of Asian, African and Latin American higher education.
Future research should prioritise three directions. First and most important: formally incorporate learner engagement as a mediating construct between the pedagogy layer and satisfaction. The present study provides the theoretical architecture; an SEM study with a validated engagement measure (Fredricks et al., 2004) could empirically test the full access → engagement → satisfaction causal chain, transforming the two-layer model into a testable three-construct path model. Second, validate the LSBLS across other national ODL contexts through confirmatory factor analysis, enabling cross-country comparison of satisfaction profiles. Third, examine whether the access-pedagogy threshold varies systematically with institutional infrastructure maturity, using multi-level modelling across institutions with differing digital readiness levels.
