Blended Learning Data: How to Tell If Your Engineering Program Is Actually Working

Blended learning has become the default promise of modern engineering education, yet most institutions still evaluate it through the shallow lens of attendance logs and completion certificates. A newly published open dataset from Vietnamese engineering universities, containing 436 valid survey responses, cuts through that complacency by measuring what actually matters: teaching presence, student engagement, and perceived achievement.

This is not another theoretical white paper; it is raw, structured evidence that challenges how universities determine whether their hybrid programs genuinely work.

The dataset, hosted on ScienceDirect, represents a rare empirical window into the mechanics of blended learning effectiveness. It moves beyond the tired debate of online versus offline and instead interrogates the quality of interaction, the clarity of instructor presence, and the psychological connection students feel toward their own progress.

For administrators, faculty, and curriculum designers, this data offers a replicable framework to audit their own programs with precision rather than guesswork.

This analysis dissects the dataset's methodology, its key findings, and the practical metrics institutions should adopt immediately. We will explore how teaching presence functions as the backbone of engagement, why perceived achievement often diverges from actual grades, and how universities can build a continuous evaluation loop that turns raw responses into actionable curriculum reform.

TL;DR A new open dataset from Vietnamese engineering universities provides 436 validated responses measuring teaching presence, engagement, and perceived achievement in blended learning environments. The data reveals that teaching presence is the strongest predictor of both engagement and perceived achievement, while infrastructure quality plays a moderating role. Institutions should adopt a three-tier evaluation framework: process metrics (teaching presence), experience metrics (engagement), and outcome metrics (perceived achievement), rather than relying on attendance or completion rates alone.

The Anatomy of the Dataset: What 436 Responses Actually Reveal

This dataset is not a casual survey; it is a rigorously constructed instrument designed to capture the multidimensional reality of blended learning. The researchers targeted engineering-related universities across Vietnam, a country rapidly expanding its higher education infrastructure and digital capacity.

The 436 valid responses represent a filtered subset, meaning invalid or incomplete submissions were removed to preserve statistical integrity.

The survey instrument was built around three core constructs: teaching presence, engagement, and perceived achievement. Teaching presence refers to the instructor's ability to design, facilitate, and direct cognitive and social processes. Engagement captures the student's behavioral, emotional, and cognitive investment in the learning process. Perceived achievement measures the student's self-assessed learning gains, not just their final grades.

What makes this dataset particularly valuable is its openness. Any researcher, administrator, or curriculum designer can download the raw responses and run their own analyses. This transparency stands in stark contrast to the proprietary, closed-door evaluations most universities conduct internally, which rarely see the light of peer review.

Teaching Presence as the Dominant Predictor

The data consistently shows that teaching presence outperforms other variables in predicting both engagement and perceived achievement. Students who reported clear instructor guidance, timely feedback, and well-structured course design were significantly more likely to report high engagement levels.

This finding aligns with the Community of Inquiry framework, which has long argued that teaching presence is the glue holding cognitive and social presence together.

In practical terms, this means universities cannot simply upload lecture videos and call it blended learning. The instructor must remain an active, visible force throughout the digital components of the course.

Discussion forums need instructor moderation, assignments need personalized feedback, and course navigation must be intuitive enough that students never feel lost in the digital labyrinth.

The dataset quantifies this relationship with correlation coefficients that demand attention. Teaching presence showed a stronger correlation with perceived achievement than raw engagement did, suggesting that students attribute their learning gains directly to instructor quality. This is a humbling reminder that technology is merely a delivery mechanism; pedagogy remains the differentiator.

Institutions that ignore this finding risk investing heavily in learning management systems while neglecting the human element that makes those systems effective. A well-designed course with an absent instructor will underperform a modestly designed course with a highly present instructor, according to the patterns in this data.

Engagement: The Mediating Variable That Connects Teaching to Outcomes

Engagement does not exist in a vacuum; it is the bridge between teaching presence and perceived achievement. The dataset reveals a clear mediation pathway: teaching presence drives engagement, and engagement drives perceived achievement.

This means engagement is not an end in itself but a mechanism through which good teaching translates into student outcomes.

Behavioral engagement, such as attending synchronous sessions and completing assignments, was the easiest to measure but not necessarily the most impactful. Emotional engagement, the student's sense of belonging and interest in the subject, showed stronger correlations with perceived achievement.

Cognitive engagement, the willingness to grapple with complex problems, emerged as the most demanding yet most rewarding form.

Universities that track only behavioral metrics, such as login frequency or video watch time, are capturing the surface of engagement. The dataset suggests that emotional and cognitive engagement are where the real learning gains occur, and these require qualitative measurement methods such as surveys, reflective journals, and peer assessments.

The practical implication is that engagement surveys should be administered at multiple points throughout the semester, not just at the end. This allows instructors to identify disengagement early and intervene before it hardens into failure. The dataset provides a validated instrument that institutions can adapt for this purpose.

Perceived Achievement: The Subjective Metric That Predicts Retention

Perceived achievement, the student's self-assessment of their learning gains, is often dismissed as a soft metric compared to exam scores. However, the dataset shows that perceived achievement correlates strongly with long-term outcomes such as course persistence and program completion.

Students who believe they are learning are more likely to stay enrolled and recommend the program to peers.

This finding challenges the assumption that objective grades are the only meaningful measure of educational success. Grades measure performance at a single point in time, while perceived achievement captures the cumulative sense of growth that sustains motivation.

A student who earns a B but feels they learned nothing is at higher risk of dropping out than a student who earns a C but feels they mastered the material.

The dataset also reveals that perceived achievement is not uniformly distributed across demographic groups. Students with weaker prior preparation reported lower perceived achievement even when their grades were comparable, suggesting that confidence and self-efficacy play a significant role. Universities should therefore pair academic support with psychological support to address this gap.

For program evaluation, perceived achievement offers a leading indicator that predicts downstream outcomes before they manifest in retention statistics. Institutions that monitor this metric can identify struggling programs early and allocate resources before attrition becomes a crisis.

Infrastructure and Contextual Factors: The Hidden Moderators

While teaching presence and engagement dominate the narrative, the dataset also captures contextual factors that moderate their impact. Internet reliability, access to devices, and the quality of the learning management system all influenced how effectively teaching presence translated into engagement. Students with poor infrastructure reported lower engagement even when instructor quality was high.

This finding is particularly relevant for developing economies like Vietnam, where digital infrastructure is still catching up with educational ambition. Universities cannot assume that a well-designed blended course will work equally well across all student populations. Contextual audits must accompany pedagogical audits to ensure equity.

The dataset includes demographic variables such as year of study, major, and prior academic performance, allowing researchers to segment the analysis. This granularity reveals that first-year students are more sensitive to teaching presence than final-year students, who have developed independent learning strategies. Curriculum designers should therefore calibrate instructor involvement by student cohort.

Institutions in developed economies should not dismiss these findings as irrelevant to their context. Even in high-infrastructure environments, the quality of the digital learning environment varies significantly across departments and courses. The dataset provides a template for auditing these variations systematically.

Building a Blended Learning Evaluation Framework: From Data to Action

The dataset's true value lies not in its descriptive statistics but in its prescriptive potential. Universities can adopt the survey instrument and analytical approach to build their own continuous evaluation loops.

This section outlines a practical framework that transforms raw data into curriculum reform, moving beyond the anecdotal and the anecdotal into evidence-based decision-making.

The framework rests on three pillars: process metrics that measure teaching presence, experience metrics that measure engagement, and outcome metrics that measure perceived achievement. Each pillar requires distinct data collection methods and analytical techniques. Together, they provide a holistic view of program effectiveness that no single metric can offer.

Critically, this framework must be embedded in the institutional rhythm, not treated as a one-off research project. Semesterly administration, rapid analysis, and visible action are the ingredients that turn evaluation from a bureaucratic exercise into a genuine improvement engine.

Process Metrics: Auditing Teaching Presence in Digital Spaces

Teaching presence is not a vague quality; it is a set of observable behaviors that can be systematically audited. The dataset's survey items provide a validated checklist: instructor clarity in explaining course goals, timeliness of feedback, facilitation of discussions, and design of learning activities. Each item can be scored on a Likert scale and aggregated into a teaching presence index.

Universities should administer this audit both to students and to trained observers who review course materials and discussion transcripts. Triangulating student perception with expert observation reduces bias and provides a more complete picture. Discrepancies between the two perspectives often reveal blind spots in course design.

The frequency of measurement matters. A single end-of-semester survey captures only the final impression, which may be colored by recent events. Administering the audit at weeks four, eight, and twelve allows institutions to track how teaching presence evolves and to identify courses where instructor engagement declines over time.

Benchmarking against the Vietnamese dataset provides a useful starting point, but institutions should eventually develop their own norms. A teaching presence score that is high relative to a peer institution may still be inadequate for a specific student population. Contextual calibration is essential for meaningful interpretation.

Experience Metrics: Measuring Engagement Beyond the Clickstream

Learning management systems generate vast amounts of behavioral data, but the dataset demonstrates that this data captures only a fraction of engagement. Emotional and cognitive engagement require self-report instruments that probe students' interest, effort, and willingness to persist through difficulty. These instruments should be short enough to administer frequently without causing survey fatigue.

The dataset's engagement scale includes items on active participation, concentration during learning activities, and the perceived relevance of course content. These items can be adapted into a weekly pulse survey that takes less than two minutes to complete. The resulting time series reveals engagement patterns that a single snapshot would miss.

Qualitative methods complement the quantitative survey. Focus groups and reflective journals provide rich narratives that explain why engagement rises or falls. A quantitative decline in engagement scores becomes actionable when accompanied by student voices describing confusing assignments or disconnected course modules.

Institutions should also examine engagement disaggregated by student subgroup. The dataset shows that engagement patterns vary by year of study and prior performance, meaning aggregate scores can mask significant disparities. Equity-focused evaluation requires subgroup analysis to ensure that no student population is silently disengaging.

Outcome Metrics: Reframing Achievement as Perceived Growth

Perceived achievement should not replace objective grades but should complement them. The dataset's achievement scale measures students' self-assessed gains in knowledge, skills, and confidence. When perceived achievement diverges from actual grades, the gap itself becomes a diagnostic signal worth investigating.

A student who earns high grades but reports low perceived achievement may be gaming the assessment system without deep learning. Conversely, a student who earns low grades but reports high perceived achievement may be developing skills that the assessment does not capture. Both scenarios warrant instructor attention.

Longitudinal tracking of perceived achievement across a program reveals growth trajectories that single-course evaluations cannot. Institutions can identify courses where perceived achievement consistently lags, indicating a need for pedagogical redesign. They can also identify courses where perceived achievement exceeds expectations, providing models for replication.

The dataset's finding that perceived achievement predicts retention suggests that this metric deserves a prominent place in early warning systems. Students whose perceived achievement drops sharply in the first semester are at elevated risk of attrition. Proactive advising based on this signal can intervene before the student decides to leave.

Closing the Loop: From Evaluation to Curriculum Reform

Data collection without action is a waste of student time and institutional resources. The final step of the framework is the translation of findings into concrete curriculum changes. This requires a governance structure that reviews evaluation results, prioritizes interventions, and tracks their impact over time.

Faculty development programs should be directly informed by teaching presence scores. Instructors who score low on feedback timeliness should receive targeted training and coaching. The dataset provides evidence that such investments yield measurable returns in engagement and achievement, justifying the resource allocation.

Course redesign should follow a similar evidence-based path. If engagement scores drop in the middle of the semester, the course structure may need adjustment, such as breaking long modules into shorter segments or adding interactive elements.

The evaluation loop ensures that redesign decisions are driven by data, not by the loudest voice in the faculty meeting.

Finally, institutions should publish their evaluation results, both internally and externally. Internal publication creates accountability and spreads best practices. External publication, as demonstrated by the Vietnamese dataset, contributes to the global knowledge base and invites peer scrutiny that improves methodological rigor.

Comparative Analysis: How the Vietnamese Dataset Stacks Against Global Blended Learning Research

The Vietnamese dataset does not exist in isolation; it joins a growing body of international research on blended learning effectiveness. Comparing its findings with studies from Europe, North America, and other Asian countries reveals both universal patterns and context-specific variations. This comparative lens helps institutions judge which findings are transferable to their own settings.

One striking consistency across studies is the centrality of teaching presence. Research from the Netherlands, Australia, and the United States has repeatedly identified instructor involvement as the strongest predictor of student satisfaction and learning outcomes in blended environments. The Vietnamese dataset reinforces this consensus, suggesting that the finding transcends cultural and infrastructural boundaries.

However, the magnitude of the teaching presence effect varies across contexts. In high-infrastructure environments where students have reliable internet and devices, the effect may be smaller because the baseline digital experience is already adequate. In lower-infrastructure environments, teaching presence may compensate for technical shortcomings, making it even more critical.

Cross-Cultural Validity of the Community of Inquiry Framework

The Community of Inquiry framework, which underpins the dataset's theoretical foundation, was developed primarily in Western educational contexts. Its application in Vietnam tests whether the framework's constructs translate across cultures. The dataset's strong internal consistency suggests that teaching presence, social presence, and cognitive presence are meaningful constructs for Vietnamese engineering students.

Yet cultural nuances emerge in the relative weight of each presence. Vietnamese students, influenced by Confucian educational traditions that emphasize teacher authority, may place even greater importance on teaching presence than their Western counterparts. The dataset's findings are consistent with this hypothesis, though direct cross-cultural comparison would require a multi-country study.

Institutions in other Asian countries with similar cultural traditions should therefore treat the Vietnamese findings as highly relevant. The dataset provides a validated instrument that can be adapted with minimal modification, saving significant development time and enabling direct benchmarking.

Conversely, institutions in Western contexts should be cautious about overgeneralizing the specific correlation coefficients. The structural relationships are likely similar, but the magnitude may differ. Replication studies in diverse contexts are needed to establish the generalizability of the findings.

Methodological Innovations: What This Dataset Teaches About Survey Design

The dataset's methodology offers lessons beyond its substantive findings. The researchers clearly invested in instrument validation, reporting reliability coefficients and conducting confirmatory factor analysis. This rigor is unfortunately rare in educational research, where many studies rely on ad hoc survey items with unknown psychometric properties.

Institutions building their own evaluation instruments should follow this example. Validated scales, pilot testing, and psychometric analysis are not optional extras; they are essential for producing trustworthy data. A poorly constructed survey can yield misleading results that lead to counterproductive decisions.

The dataset also demonstrates the value of open data practices. By publishing the raw responses alongside the analysis, the researchers enable independent verification and secondary analysis. This transparency builds trust in the findings and accelerates scientific progress by allowing others to build on the work.

Universities should consider publishing their own evaluation datasets, at least in anonymized form. The competitive advantage lies not in hoarding data but in using it effectively. Publication invites collaboration and positions the institution as a thought leader in educational innovation.

Limitations and Future Research Directions

No dataset is perfect, and this one has clear limitations that must be acknowledged. The sample is drawn exclusively from engineering-related universities in Vietnam, limiting generalizability to other disciplines and countries. The data is cross-sectional, capturing a single point in time rather than tracking change over semesters.

The reliance on self-report measures introduces the possibility of common method bias. Students who report high teaching presence may also report high engagement simply because they have a positive overall disposition. Future research should incorporate objective measures, such as learning analytics and performance data, to triangulate the self-reports.

Longitudinal studies are urgently needed to establish causal direction. The dataset's mediation analysis suggests that teaching presence drives engagement, which drives perceived achievement, but cross-sectional data cannot definitively prove this sequence. Panel studies that track the same students over multiple semesters would provide stronger evidence.

Despite these limitations, the dataset represents a valuable contribution to the field. Its openness, rigor, and substantive findings provide a foundation for future research and a practical tool for institutional evaluation. The limitations are not reasons for inaction but rather signposts for careful interpretation.

Actionable Takeaways for University Administrators and Faculty

For administrators, the dataset's most urgent message is that blended learning evaluation must move beyond attendance and completion rates. These metrics are necessary but grossly insufficient. A program can have perfect attendance and high completion while failing to deliver meaningful learning, and the dataset provides the tools to detect this failure.

For faculty, the dataset offers a clear roadmap for improving their blended courses. Investing in teaching presence, through timely feedback, clear communication, and active facilitation, yields measurable returns in student engagement and perceived achievement. These behaviors are teachable and improvable with deliberate practice.

For curriculum designers, the dataset highlights the importance of cohort-specific calibration. First-year students need more instructor presence than final-year students. Courses in the middle of the curriculum may need engagement boosters to counteract mid-program fatigue. The evaluation framework enables these nuanced adjustments.

The ultimate takeaway is that blended learning effectiveness is not a mystery. It can be measured, analyzed, and improved with the right instruments and the right commitment. The Vietnamese dataset provides a proven starting point; the rest is institutional will.

Strategic Recommendations: Turning Data into Institutional Advantage

The Vietnamese dataset is not merely an academic exercise; it is a strategic asset for any university serious about blended learning quality. Institutions that adopt its framework gain a competitive edge in an increasingly crowded higher education market. Students and employers are becoming more discerning, and evidence of learning effectiveness is becoming a differentiator.

The recommendations that follow are designed for immediate implementation. They require no new technology investments, only a commitment to measuring what matters and acting on the results. The cost of inaction is far higher: continued investment in blended learning programs that may not be delivering their promised value.

Universities that embrace this data-driven approach will not only improve their programs but also build a culture of evidence-based decision-making that extends beyond blended learning to all aspects of teaching and learning.

Immediate Actions for Program Directors

Program directors should begin by adapting the dataset's survey instrument to their local context. This involves translating items, adjusting terminology, and piloting the instrument with a small student group. The validation process should include reliability analysis and factor structure confirmation to ensure the instrument performs as intended.

Once validated, the instrument should be administered at multiple points during the semester. The dataset's finding that teaching presence evolves over time means a single measurement is insufficient. A three-point administration schedule, at weeks four, eight, and twelve, provides a trajectory that reveals improvement or decline.

Directors should also establish a data review cadence. Monthly meetings to review engagement and achievement trends allow for timely intervention. Waiting until the end of the semester to analyze results means missed opportunities to correct course midstream.

Finally, directors should communicate results transparently to faculty and students. Sharing aggregate findings builds trust and demonstrates that evaluation is not punitive but developmental. Faculty who see their teaching presence scores improve over time are more likely to embrace the process.

Faculty Development Priorities Based on the Data

The dataset identifies specific teaching behaviors that drive engagement and achievement. Faculty development programs should focus on these behaviors: providing timely and specific feedback, designing clear and navigable course structures, and facilitating active discussion in both physical and digital spaces. These are teachable skills, not innate talents.

Workshops should include hands-on practice with feedback techniques, such as rubric-based commenting and audio feedback. Course design reviews should be conducted using a checklist derived from the teaching presence survey items. Peer observation should be structured around the same framework to ensure consistency.

Development should be ongoing, not a one-time workshop. The dataset shows that teaching presence can decline over the semester, suggesting that faculty need sustained support, not just initial training. Coaching and mentoring programs that provide continuous feedback are more effective than isolated training events.

Institutions should also recognize and reward faculty who demonstrate high teaching presence. The dataset provides evidence that these behaviors produce measurable student outcomes, justifying their inclusion in promotion and tenure criteria. What gets rewarded gets repeated.

Curriculum Redesign Informed by Engagement Patterns

The dataset's engagement findings suggest that curriculum structure significantly influences student investment. Courses that front-load difficult content without adequate support may see engagement drop sharply. Redesigning such courses to scaffold learning progressively can maintain engagement throughout the semester.

Mid-semester engagement dips, which the dataset reveals as common, can be addressed by inserting interactive elements such as case studies, group projects, or guest lectures. These interventions break the monotony of routine and re-energize students. The evaluation loop identifies when these interventions are needed.

Program-level curriculum mapping should identify courses where perceived achievement is consistently low. These courses become priorities for redesign. The dataset's finding that perceived achievement predicts retention means that fixing these courses directly impacts student persistence and graduation rates.

Curriculum redesign should also consider cohort-specific needs. First-year courses may need more instructor presence and structured support, while capstone courses may benefit from greater autonomy. The dataset provides the evidence to calibrate these differences.

Building a Sustainable Evaluation Culture

The ultimate goal is not a one-time evaluation but a sustainable culture of continuous improvement. This requires institutional structures that support data collection, analysis, and action. A dedicated office of educational effectiveness, or a similar unit, should own the evaluation process and report directly to academic leadership.

Data infrastructure must be designed for longitudinal tracking. Student responses should be linked across semesters to enable growth trajectory analysis. Privacy protections must be robust, with anonymization and secure storage as non-negotiable requirements.

Leadership commitment is essential. Deans and provosts must signal that evaluation is a priority, not a compliance exercise. This means allocating resources, celebrating successes, and holding units accountable for acting on findings. Without leadership support, evaluation efforts will wither.

Finally, institutions should contribute to the broader knowledge base by publishing their findings. The Vietnamese dataset demonstrates the value of open data. By sharing their own results, universities accelerate global progress in blended learning effectiveness and position themselves as thought leaders.