BWM analysis of online and offline learning effectiveness in Bangladesh

Abstract

The shift to online learning during the COVID-19 pandemic presented significant challenges and opportunities, particularly in developing nations like Bangladesh, where digital infrastructure is limited. This study compares the effectiveness of online and offline learning for Bangladeshi engineering students using the Best Worst Method (BWM), a robust decision-making tool that simplifies the analysis by focusing on the best and worst criteria, ensuring consistency and accuracy. Eight key factors, including cost, flexibility, learning effectiveness, and technological challenges, were evaluated to identify the distinct strengths and limitations of each learning modality. The findings highlight that online learning is favored for its affordability and adaptability, with cost (25.74%) and flexibility (19.38%) emerging as the highest-priority criteria. In contrast, offline learning is valued for its hands-on practicality and structured environments, where technological challenges (19.25%) and concentration (18.47%) ranked as the most critical factors. Sensitivity analysis confirms the robustness of these rankings, reinforcing the reliability of the results. This research uniquely applies BWM to a resource-constrained educational context, addressing gaps in the literature. Its findings have broader implications for educational policy and resource allocation, providing actionable insights for designing blended learning strategies. By integrating the flexibility of online platforms with the immersive, practical benefits of offline learning, this study proposes a scalable framework for improving learning outcomes in Bangladesh and similar developing regions.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-92831-x.

Introduction

Global educational systems have undergone significant change as a result of the COVID-19 pandemic, which has accelerated the transition from conventional in-person instruction to online learning environments¹. This change presented substantial challenges, particularly in underdeveloped nations such as Bangladesh, where access to reliable internet and digital resources is frequently constrained. These criteria have elicited apprehensions over the efficacy of online learning in comparison to conventional offline approaches². Consequently, comprehending the advantages and disadvantages of all learning modalities has become essential for guaranteeing that students obtain excellent education under diverse conditions. The sudden shift required the implementation of new teaching methods and technical solutions, which have achieved differing levels of success in maintaining educational continuity and effectiveness³.

In Bangladesh, the transition to online learning has amplified existing educational challenges, including inadequate digital infrastructure, high costs of internet services, and frequent power outages⁴. Many students, particularly in rural areas, face difficulties accessing stable internet connections and digital devices, further widening the gap in educational opportunities⁵. Additionally, engineering education, which requires hands-on practical training and laboratory work, suffers significantly in online settings where such experiences are limited or absent. Conversely, while offline learning offers structured environments conducive to practical education, it often poses financial and logistical constraints, particularly for students from low-income households. These unique challenges call for a comprehensive evaluation of the strengths and limitations of both learning modalities⁶.

Online education offers several advantages, including flexibility, cost-effectiveness, and convenience, since it allows students to learn at their own pace and from any location. This instructional strategy has been quite beneficial throughout the pandemic since it addresses diverse learning needs and provides teaching to a global audience. Nonetheless, online learning presents disadvantages, including the necessity for substantial self-discipline, limited interaction, and technological difficulties⁷. Conversely, conventional offline education offers a controlled and interactive setting that can enhance engagement and practical experiences, however it frequently entails greater expenses and reduced flexibility⁸.

It is essential to thoroughly assess the efficacy of both online and offline learning approaches in light of these divergent features. This study aims to address this need by employing a popular multi-criteria decision-making (MCDM) technique: the Best-Worst Method (BWM). These methods enable a comprehensive analysis of the key criteria affecting learning effectiveness, such as cost, flexibility, effectiveness of learning, interaction, concentration, pace of learning, technological criteria, and practicality. The Best-Worst Method (BWM), introduced by Jafar Rezaei in 2015, simplifies the decision-making process by focusing on identifying the best and worst criteria and making fewer comparisons, which reduces the potential for inconsistency⁹. Though it is new technique, however it has been deployed in several studies^10–13.

BWM emphasizes the identification of optimal and suboptimal criteria through minimal comparisons. In comparison to other MCDM approaches such as AHP, TOPSIS, and ELECTRE, the Best Worst Method (BWM) presents unique benefits. It necessitates fewer pairwise comparisons, hence diminishing cognitive strain and rendering it less time-intensive than AHP. BWM concentrates solely on the optimal and suboptimal criteria, hence improving consistency and precision in weight estimates, whereas methodologies such as AHP include more intricate pairwise comparisons that may lead to discrepancies. In contrast to TOPSIS, which needs normalization and distance computations, BWM is more straightforward to execute. BWM surpasses ELECTRE in computing efficiency by circumventing intricate outranking procedures. The systematic methodology of BWM facilitates the identification of discrepancies and preserves resilience despite low data availability. Moreover, BWM’s adaptability enables its successful use across many decision-making contexts.

Including sensitivity analysis strengthens this study by ensuring the robustness of findings, as it verifies that priority rankings hold steady even with changes in criterion weights. This approach enhances result reliability, confirming that identified factors are resilient and dependable for guiding improvements in online and offline learning.

This study’s primary objectives are listed below:

1. To evaluate and rank key criteria influencing the effectiveness of online and offline learning for engineering students in Bangladesh using the Best Worst Method (BWM).

2. To compare the priority of criteria between online and offline learning environments, highlighting the differences in student preferences and needs across these modalities.

3. To assess the robustness of the identified criteria through sensitivity analysis, ensuring reliable insights for developing effective, adaptable learning strategies in the Bangladeshi educational context.

This rest of the paper is organized as follows: the Literature Review summarizes relevant research on the effectiveness of online and offline learning, addressing key gaps this study aims to fill. The Methodology section details the Best Worst Method (BWM) used for evaluating criteria. The Results section presents findings on the prioritized criteria for each learning modality, followed by Sensitivity Analysis at the end of the results to confirm robustness. The Discussion interprets the findings and compares them with previous studies, and the Conclusion highlights key insights, limitations, and suggestions for future research.

Literature review on the effectiveness of online and offline learning

The efficiency of online and offline learning has been a hot topic of discussion, especially in the wake of the COVID-19 epidemic and the global transition to online platforms. Research consistently recognizes that online education provides flexibility and convenience, enabling students to engage with classes at their own speed and from any place (Dhawan, 2020)¹⁴. Research reveals that the absence of face-to-face interaction in online environments frequently constrains student engagement and emotional connection, both of which are essential for profound learning^15,16. The study by Amit et al. investigates COVID-19-related mental stress among 651 Bangladeshi adults, revealing over 83% experience issues like short temper, sleep disorders, and family chaos. Key stress factors include financial hardship, disrupted education, and food crises, highlighting the need for targeted strategies to improve mental health and resilience during pandemics¹⁷.

A study by Alserhan et al. explores how integrating a third-generation Learning Management System (LMS) into Personal Learning Environments (PLEs) enhances self-regulated learning (SRL) using Zimmerman’s SRL model¹⁸. Through pretest and posttest evaluations during a PLE course applied in the COVID-19 pandemic, the findings reveal a model of SR factors in PLEs using partial least squares structural equation modeling (SEM). Another study systematically reviews 104 studies on student behavior in online social networks (OSNs)¹⁹, focusing on trends, theories, and methods while utilizing the Stimulus-Organism-Response (SOR) model to classify influencing factors. Key findings reveal five research streams—academic purpose, cyber victimization, addiction, personality issues, and knowledge sharing—offering a framework for future research on OSNs’ impact on students. This study by Abuhassna et al. (2023) grounded in transactional distance theory (TDT) and collaborative learning, examines how collaborative learning influences students’ academic achievements, autonomy, and satisfaction with online platforms during the COVID-19 pandemic. It focuses on Malaysian higher education’s shift to online learning, prompted by campus closures and government directives, highlighting the critical role of collaboration in online education²⁰.

Technological obstacles, such inconsistent internet connectivity and restricted availability of digital devices, impede the effectiveness of online education, especially in poor nations²¹. These factors are exacerbated by the digital divide, which disproportionately impacts pupils in areas such as South Asia and sub-Saharan Africa. Although online education can save expenses for both institutions and students, it sometimes necessitates considerable upfront investments in infrastructure²². Conversely, offline learning offers an organized setting with prompt feedback, which several students like due to its in-person connection and collaborative prospects. Research indicates that traditional schooling is more effective for developing practical abilities, particularly in disciplines such as engineering and science, where experiential learning is essential²³.

Online learning effectiveness

Several studies have examined the effectiveness of online learning in Bangladesh, particularly during the COVID-19 pandemic. A study by Haque (2021) investigated the perspectives of 310 students from various universities, revealing that approximately 60% were satisfied with online education due to close interaction and teacher efforts. However, the remaining students reported dissatisfaction, citing health and mental issues from prolonged online classes²⁴. Similarly, Saarmean et al. (2021) conducted research among private university students, highlighting challenges such as lack of physical interaction, limitations in monitoring during online exams, and mental and physical health complications. These factors contributed to perceptions of reduced learning effectiveness in online settings²⁵.

Offline learning effectiveness

Traditional classroom settings offer structured environments and direct interactions, which are crucial for effective learning. A study by Shahariar (2022) assessed the effectiveness of online versus offline schooling methods among 56 students in Bangladesh. The findings indicated that while both methods have their advantages, offline schooling provided more effective academic direction, assessment, and interactivity. Students reported that online classes often suffered from poor internet connectivity, data limitations, and power outages, affecting the overall learning experience²⁶. Further, a study by Chowdhury and Behak (2022) explored students’ experiences with blended learning in a private university in Bangladesh²⁷. The research revealed that while blended learning offers flexibility, students still value face-to-face interactions for immediate feedback and personal engagement, which are inherent benefits of offline learning.

Blended approaches

Blended learning, integrating online and offline components, is frequently seen as the most successful approach, since it provides the flexibility of online education while preserving the interaction aspects of conventional classroom instruction²⁸. The efficacy of both modalities is predominantly contingent upon the caliber of instructional design and the capacity to meet the distinct demands of students across various situations²⁹. Recognizing the strengths and limitations of both online and offline learning, blended learning approaches have been proposed. The University Grants Commission (UGC) of Bangladesh has developed a policy on blended learning, aiming to combine online and offline methods to enhance educational outcomes. This approach seeks to leverage the flexibility of online learning while maintaining the interactive benefits of traditional classrooms³⁰. Additionally, Khan (2021) examined the challenges and prospects of deploying blended learning in the new normal pedagogy in Bangladesh³¹. The study identified barriers such as inadequate internet facilities and untrained teachers but also highlighted the potential of blended learning to provide a balanced educational system in the post-pandemic period.

While online learning has considerable benefits regarding flexibility and affordability, it encounters substantial challenges concerning engagement, retention, and technology accessibility, particularly in poor countries. In contrast, offline learning has superior engagement and practical application, however it is deficient in flexibility and cost-effectiveness compared to online platforms. Blended learning seems to be a viable method that integrates the advantages of both modes.

Research gap and significance of the study

The novelty of this study lies in its focus on comparing the effectiveness of online and offline learning specifically in Bangladesh, a developing country where technological and digital infrastructure criteria are more prominent. The current body of research highlights the effectiveness of online, offline, and blended learning, yet several critical gaps remain unaddressed. Existing studies often lack context-specific evaluations, particularly for disciplines like engineering that require hands-on practical training, making the unique needs of engineering students in resource-constrained settings like Bangladesh underexplored. While technological challenges, such as unstable internet and limited device accessibility, are frequently noted in online learning research, there is insufficient focus on how these issues impact the practical implementation of blended models. Furthermore, although policies on blended learning, such as those proposed by the University Grants Commission in Bangladesh, offer potential solutions, their effectiveness in addressing the shortcomings of online and offline learning modalities remains empirically unverified. Additionally, the psychosocial impacts of online learning, including mental and physical health complications, are acknowledged but not thoroughly examined in the context of blended learning adoption and its effectiveness. Finally, there is a lack of comparative assessments employing robust methods like the Best Worst Method (BWM) to systematically evaluate the relative strengths and weaknesses of online, offline, and blended learning approaches in developing countries. This study seeks to address these gaps by utilizing BWM to evaluate and compare the effectiveness of these learning modalities for engineering students in Bangladesh, providing actionable insights to tailor blended learning models that address educational, technological, and psychosocial challenges in resource-constrained settings.

Methodology

This study employed the Best Worst Method (BWM), a well-established Multi-Criteria Decision-Making (MCDM) method, to evaluate the effectiveness of online and offline learning based on key criteria. BWM was selected due to its robustness in handling both qualitative and quantitative data, making it highly suitable for complex educational evaluations where diverse criteria are involved⁹. Firstly, data about online and offline education system has been collected from students of engineering background. Then those data were analyzed by BWM and the ranking of criteria of each learning method has been determined.

Sample selection

The data were gathered from engineering students at both public and private universities in Bangladesh. The university selection aimed to encompass a varied array of viewpoints, reflecting different socio-economic origins and resource accessibility. Engineering students were selected due to their education, which encompasses both theoretical and practical elements, rendering them suitable for assessing the efficacy of online and offline learning settings. Nonetheless, the selection of colleges lacked randomness; they were instead picked for their convenience and accessibility. The admission criteria targeted students with expertise in both online and offline learning environments throughout the epidemic. A sample size of 21 students was considered adequate based on previous research utilizing AHP in analogous educational settings, where smaller, concentrated samples facilitated comprehensive pairwise comparisons as suggested by Vaidya et al.³². Studies employing the Best Worst Method (BWM) demonstrate its effectiveness with small, expert-focused sample sizes. For instance, Shabur et al. used 9 experts in a study on Industry 4.0 in the steel sector¹⁰ and 10 experts in a similar study for the fertilizer sector in Bangladesh¹². The study “Soil Quality Ranking Using AHP” applied AHP to rank soil quality with a sample size of 18 participants, demonstrating the method’s effectiveness in such contexts³³. Similarly, the “Application of AHP in Construction Management” review highlighted that AHP is particularly suitable for studies with small sample sizes, often around 15 experts, due to its high level of consistency and simplicity³⁴. In the educational field, research utilizing AHP found it effective even with 20 faculty members, supporting its applicability in similar contexts³⁵. Furthermore, a simulation study on “Sample Size Considerations in AHP-Based Surveys” explored appropriate sample sizes, ranging from 10 to 25 participants, indicating that smaller samples can be sufficient depending on the study’s objectives³⁶.These focused samples, comprising industry professionals and academics, provided reliable and valid results through targeted expertise. This justifies the sample size in our study, as BWM prioritizes depth of insight over sample size, ensuring robust outcomes even with limited participants.

Data collection process

The data gathering technique entailed the administration of a standardized questionnaire particularly formulated for the AHP method. The questionnaire was designed following a comprehensive literature analysis of learning effectiveness criteria, encompassing eight principal criteria: Cost and Time, Flexibility, Learning Effectiveness, Interaction, Concentration, Pace of Learning, Technological Criteria, and Practicality. Each criterion was explicitly delineated, and participants received guidance on doing pairwise comparisons of these criteria. The questionnaire was administered to students in person, assuring their comprehension of the procedure. The duration needed to finish the survey was around 90 min.

Development of structured questionnaire

For the BWM method, we developed the structured questionnaire through extensive literature review and expert consultation, selecting eight criteria based on their relevance in prior research on online and offline learning, tailored to the Bangladeshi context. Participants conducted pairwise comparisons, rating each criterion’s relative importance on a scale from 1 to 9, consistent with BWM principles. To ensure clarity, we piloted the questionnaire with a small student group before wider distribution, allowing for a structured, accurate representation of participants’ preferences in the final rankings.

Reliability was ensured by employing a structured questionnaire based on a comprehensive review of the literature, with clearly defined criteria relevant to the context of online and offline learning. A pilot study was conducted with a small group of students to refine the questionnaire for clarity and consistency. The Best Worst Method (BWM) was chosen due to its established robustness in handling multi-criteria decision-making, minimizing inconsistencies in pairwise comparisons.

For validity, the criteria were selected based on their prominence in prior studies (e.g., Rezaei, 2015; Dhawan, 2020)^9,14, ensuring relevance to the research objectives. Sensitivity analysis was conducted to confirm the stability of the results under varying weight scenarios, a process similarly employed in other BWM-based studies, such as Alserhan et al. (2021)¹⁸. These steps collectively enhance the reliability and validity of the findings, ensuring they are grounded in rigorous methodologies and reflective of the study’s context.

Sampling bias and mitigation strategies

Participants in this study were engineering students from both public and private universities in Bangladesh. This selection was made to capture a diverse range of experiences and perspectives related to online and offline learning. Engineering students were specifically chosen due to their unique blend of theoretical and practical coursework, making them well-suited to evaluate both learning modalities. While the study includes participants from diverse institutional backgrounds, the focus solely on engineering students may limit the generalizability of the findings to other disciplines. Additionally, non-random participant recruitment could introduce selection bias, as students from specific institutions may share similar experiences or preferences.

To address potential biases, the selection process included students from multiple public and private universities, ensuring representation from institutions with varying levels of resources and technological infrastructure. Efforts were also made to include students with substantial experience in both online and offline learning environments, particularly during the COVID-19 pandemic. Finally, results were cross-validated with findings from the existing literature to strengthen the reliability and applicability of the conclusions.

Best worst method (BWM)

The Best Worst Method (BWM) involves fewer pairwise comparisons than other MCDM tools, thus reducing the potential for inconsistency. The BWM process involves the following steps as shown in Fig. 1⁹:

Fig. 1.

Steps of BWM process.

BWM simplifies the decision-making process by reducing the number of comparisons required, making it a faster and often more consistent method compared to AHP.

Details of criteria

The criteria were chosen based on a thorough review of the literature and their relevance to assessing the effectiveness of online and offline learning modalities, particularly in a resource-constrained context like Bangladesh. Key factors such as cost, flexibility, and technological challenges were prioritized due to their significant impact on accessibility and feasibility, as highlighted in prior studies. Learning effectiveness, concentration, and interaction were included to capture cognitive and engagement-related aspects crucial for educational success. Practicality was selected to address the unique needs of engineering education, which relies heavily on hands-on training. These criteria comprehensively address logistical, cognitive, and contextual factors, making them well-suited for evaluating learning effectiveness in this specific context. These are following:

CR1: Cost and Time considers all expenses a student faces during their education, such as transportation, meals, internet, equipment, books, and other resources. This criterion also accounts for time spent commuting, attending classes, and completing assignments. It encompasses both tangible and intangible costs involved in online and offline learning environments, highlighting the full scope of investment required for each mode of education^28,37.

CR2: Flexibility represents students’ freedom in managing their study time and location, enabling them to attend classes at convenient times and access materials from anywhere. This criterion is especially crucial for online learning, which supports diverse schedules and learning needs, accommodating various lifestyles more effectively than traditional classroom settings^38,39.

CR3: Effectiveness of Learning assesses how well the educational process meets its intended outcomes. This criterion examines the extent to which students achieve the course goals, retain knowledge, and apply it effectively, while also considering the learning method’s role in developing essential skills. Ultimately, it evaluates how successfully an educational approach fulfills its purpose in fostering meaningful, skill-based learning^22,40.

CR4: Interaction evaluates the quality and depth of communication during the learning process, focusing on opportunities for students to engage with instructors for feedback and to participate in discussions with peers. This criterion emphasizes the level of student engagement driven by interaction, which is vital for fostering curiosity, deeper understanding, and enriching the learning experience^15,16.

CR5: Concentration is the power of students to sustain or keep focused during the learning session. It looks at their ability to avoid any disturbance and the time, which the student can hold or sustain attention on an educational activity. He sees that maintenance of focus was very important in efficient learning and that this criterion evaluated factors impacting concentration of students in the online as well as offline environment^41,42.

CR5: Concentration assesses students’ ability to maintain focus throughout a learning session, examining their capacity to avoid distractions and sustain attention on educational tasks. This criterion highlights the importance of focus for effective learning and evaluates factors that influence concentration in both online and offline environments^29,43.

CR7: Technological Problems refer to technical issues that disrupt the learning process, primarily in online environments. This criterion includes factors like audio and video quality, internet connectivity, and stability. For instance, a slow or unstable connection can significantly impact the learning experience, making it a critical parameter in evaluating the effectiveness of online learning^44,45.

CR8: Practicality assesses the extent to which learning includes hands-on or experiential components. It considers opportunities for students to engage in practical tasks, activities, and real-world applications, as well as the balance of theory and application in the curriculum. This criterion emphasizes the value of practical experiences in reinforcing theoretical concepts, contributing to comprehensive skill development^23,46.

The participants received explanations about the criteria and the ranking process, followed by filling out the questionnaires.

Results and discussions

Data analysis using BWM method

Sample Calculation: 01

Table 1 outlines the determination of the best criterion for online learning by comparing “Effectiveness of Learning” against other criteria. It lists preference scores ranging from 1 to 9, reflecting the degree of importance each criterion has relative to “Effectiveness of Learning.” The high score assigned to “Technological Criteria” (9) indicates significant perceived impact. Lower scores, like those for “Cost and Time,” suggest these criteria were seen as less critical. The table helps highlight which factors are prioritized by participants in an online learning context.

Table 1.

Determine the best criteria for online.

Best to others	CR1	CR2	CR3	CR4	CR5	CR6	CR7	CR8
Effective-ness of learning	7	6	1	3	2	9	5	4

Table 2 displays the identification of the best criterion for offline learning, with “Cost and Time” being compared against other factors. Preference scores indicate that “Flexibility” (score of 9) is seen as the most important after “Cost and Time,” whereas “Practicality” receives a lower preference score. The table demonstrates a distinct pattern where time and financial considerations dominate decision-making in offline settings. This reflects participants’ recognition of tangible costs associated with traditional education. The scores also suggest varying levels of importance for different factors in offline learning.

Table 2.

Determine the best criteria for offline.

Best to others	CR1	CR2	CR3	CR4	CR5	CR6	CR7	CR8
Cost and Time	1	9	3	4	5	6	8	2

Table 3 lists the preferences of other criteria in relation to the least important criterion, “Pace of Learning,” for online learning. High scores, such as those for “Effectiveness of Learning” (9), reveal the significant value placed on impactful learning outcomes. Lower scores for “Cost and Time” show it is considered less critical in the online setting.

Table 3.

Preferences for online.

Others to the worst	Pace of learning
Efficiency of learning	9
Cost and time	2
Flexibility	3
Interaction	6
Concentration	7
Pace of learning	1
Technological issue	4
Practicality	5

Table 4 shows the preferences of criteria relative to “Flexibility,” identified as the least important criterion in the offline context. “Cost and Time” receives the highest score (9), indicating that participants see it as the most crucial factor. The table reveals that while “Flexibility” might be less valued offline, other factors like “Interaction” and “Concentration” are still prioritized.

Table 4.

Preferences for offline.

Others to the worst	Flexibility
Efficiency of learning	6
Cost and time	9
Flexibility	1
Interaction	5
Concentration	4
Pace of learning	3
Technological issue	2
Practicality	7

Table 5 provides the calculated weights for each criterion in the online learning context, using BWM to determine relative importance. The highest weight is assigned to “Effectiveness of Learning” (0.338), showing its priority over other criteria. Weights for factors like “Pace of Learning” and “Technological Issues” are lower, suggesting less emphasis in the online environment. The table shows a balanced approach to weighing cognitive, practical, and logistical aspects of learning. These calculated weights are crucial for understanding which areas should be targeted for improvement in online education.

Table 5.

Criteria weight for online.

	CR1	CR2	CR3	CR4	CR5	CR6	CR7	CR8
Weights	0.05653	0.06595	0.33844	0.13191	0.19786	0.03124	0.07914	0.09893

Table 6: Lists the weights assigned to criteria for offline learning, indicating how participants prioritize different factors. “Cost and Time” has the highest weight (0.340), while “Flexibility” is the least weighted factor. The table reflects a greater emphasis on practical and logistical considerations in traditional learning. Weights for factors like “Practicality” suggest that hands-on experiences are valued offline, compared to lower weights for “Technological Issues.” This distribution provides a clearer picture of the key concerns influencing offline education.

Table 6.

Criteria weight for offline.

	CR1	CR2	CR3	CR4	CR5	CR6	CR7	CR8
Weights	0.3408	0.0315	0.1328	0.0996	0.0797	0.0664	0.0498	0.1993

Figure 2 is the graphical representation of the results for the weights of different criteria in online learning, determined using the Best Worst Method (BWM). The bar chart highlights “Effectiveness of Learning” as the most important factor, with the highest weight, while criteria such as “Pace of Learning” and “Technological Issues” have lower weights, indicating they are considered less critical by participants. This graphical representation helps visualize the varying levels of importance assigned to each criterion, identifying key strengths and areas for improvement in online education.

Fig. 2.

Weights of criteria for online.

Figure 3 provides a graphical representation of the weights of criteria for offline learning, based on BWM analysis. The bar chart shows that “Cost and Time” is the most heavily weighted factor, indicating its significance in traditional education, while “Flexibility” has the lowest weight. The graphical representation illustrates that practical considerations like “Practicality” and “Concentration” are more prioritized in offline settings compared to other criteria. This visual summary allows for a straightforward comparison with the results shown in Fig. 1, demonstrating the differing priorities between online and offline learning environments. The figure effectively conveys the participants’ evaluations, emphasizing the distinctions in how learning factors are valued across modalities.

Fig. 3.

Weights of criteria for offline.

Consistency check

Table 7 shows the consistency check results for online learning, confirming that the pairwise comparisons meet an acceptable consistency ratio. The input-based consistency ratio (CR) is 0.1645, indicating that participant evaluations are within a reasonable range. The associated threshold of 0.362 validates the reliability of the results. This table helps ensure that the comparisons made are not arbitrary but based on consistent judgment. Such consistency strengthens the credibility of the weight calculations for online learning criteria.

Table 7.

Consistency check for online.

Input-based CR	0.1645	The pairwise comparison consistency level is acceptable
Associated threshold	0.362

Table 8 displays the consistency check for offline learning, with an input-based CR of 0.152, meeting the same threshold (0.362). This demonstrates that offline learning criteria comparisons also exhibit an acceptable level of consistency. The results validate the reliability of the participants’ evaluations in an offline context. Consistency checks like this help identify any deviations that may affect decision-making quality. The table confirms the robustness of the results derived from the BWM process.

Table 8.

Consistency check for offline.

Input-based CR	0.152	The pairwise comparison consistency level is acceptable
Associated threshold	0.362

Sum of criteria weight using BWM

Table 9 summarizes the total weight sums for each criterion across all participants, specifically for the online learning context. “Cost and Time” and “Effectiveness of Learning” show substantial variation in total weight sums, reflecting different participant perspectives. The cumulative values help in understanding common priorities and disparities among the participants’ responses. The table provides a collective view of how various criteria are evaluated in online education. Such aggregation enables a comprehensive assessment of learning factors across different participants.

Table 9.

Total sum of every considered criterion for online.

Criteria	CR1	CR2	CR3	CR4	CR5	CR6	CR7	CR8
Participant-1	0.341	0.032	0.133	0.100	0.080	0.066	0.050	0.199
Participant-2	0.343	0.202	0.081	0.067	0.081	0.134	0.035	0.058
Participant-3	0.237	0.267	0.041	0.067	0.134	0.067	0.134	0.053
Participant-4	0.333	0.196	0.078	0.065	0.065	0.131	0.034	0.098
Participant-5	0.330	0.132	0.079	0.079	0.198	0.099	0.033	0.050
Participant-6	0.329	0.203	0.081	0.102	0.068	0.136	0.036	0.045
Participant-7	0.339	0.196	0.065	0.098	0.131	0.079	0.036	0.056
Participant-8	0.317	0.121	0.073	0.091	0.121	0.182	0.034	0.061
Participant-9	0.033	0.322	0.126	0.076	0.126	0.190	0.063	0.063
Participant-10	0.314	0.130	0.065	0.130	0.194	0.056	0.078	0.034
Participant-11	0.142	0.356	0.036	0.061	0.071	0.085	0.142	0.107
Participant-12	0.285	0.087	0.058	0.069	0.173	0.260	0.025	0.043
Participant-13	0.192	0.334	0.128	0.077	0.128	0.064	0.036	0.043
Participant-14	0.039	0.333	0.079	0.066	0.197	0.099	0.056	0.132
Participant-15	0.243	0.100	0.150	0.031	0.060	0.299	0.043	0.075
Participant-16	0.357	0.042	0.189	0.076	0.126	0.094	0.054	0.063
Participant-17	0.347	0.129	0.034	0.078	0.097	0.194	0.065	0.055
Participant-18	0.322	0.189	0.189	0.094	0.047	0.075	0.030	0.054
Participant-19	0.333	0.101	0.068	0.203	0.081	0.135	0.051	0.029
Participant-20	0.034	0.261	0.030	0.103	0.310	0.062	0.044	0.155
Participant-21	0.198	0.338	0.079	0.066	0.099	0.132	0.031	0.057
Sum	5.406	4.070	1.860	1.797	2.587	2.638	1.108	1.528

Table 10 presents the aggregated weights for offline learning, indicating the sum of weights for each criterion as judged by all participants. “Technological Issues” and “Concentration” have the highest weight sums, showing participants’ concerns with technical barriers and focus in offline learning. The table illustrates the extent to which each criterion impacts overall learning effectiveness in a traditional context. Differences in weight sums highlight varying levels of significance assigned to each factor. This consolidated data helps compare the offline learning priorities against online ones, guiding targeted improvements.

Table 10.

Total sum of every considered criterion for offline.

Criteria	CR1	CR2	CR3	CR4	CR5	CR6	CR7	CR8
Participant-1	0.071	0.035	0.059	0.071	0.051	0.118	0.298	0.298
Participant-2	0.110	0.031	0.073	0.063	0.147	0.088	0.342	0.147
Participant-3	0.052	0.035	0.346	0.138	0.104	0.104	0.083	0.138
Participant-4	0.035	0.047	0.066	0.270	0.332	0.054	0.083	0.111
Participant-5	0.035	0.055	0.077	0.335	0.194	0.077	0.129	0.097
Participant-6	0.061	0.027	0.246	0.101	0.152	0.038	0.299	0.076
Participant-7	0.050	0.032	0.174	0.087	0.300	0.116	0.069	0.174
Participant-8	0.120	0.040	0.032	0.305	0.060	0.072	0.249	0.120
Participant-9	0.027	0.051	0.103	0.068	0.082	0.137	0.205	0.326
Participant-10	0.034	0.060	0.119	0.060	0.392	0.080	0.096	0.159
Participant-11	0.033	0.056	0.098	0.130	0.195	0.065	0.326	0.098
Participant-12	0.049	0.030	0.114	0.172	0.172	0.086	0.069	0.308
Participant-13	0.056	0.033	0.131	0.079	0.197	0.079	0.328	0.098
Participant-14	0.055	0.034	0.347	0.065	0.194	0.097	0.129	0.078
Participant-15	0.076	0.029	0.063	0.047	0.126	0.076	0.306	0.277
Participant-16	0.029	0.054	0.145	0.109	0.062	0.154	0.348	0.145
Participant-17	0.064	0.040	0.077	0.313	0.193	0.055	0.129	0.129
Participant-18	0.031	0.039	0.248	0.062	0.310	0.155	0.078	0.078
Participant-19	0.036	0.049	0.099	0.132	0.079	0.066	0.341	0.198
Participant-20	0.057	0.066	0.338	0.132	0.198	0.031	0.079	0.099
Participant-21	0.066	0.031	0.079	0.198	0.338	0.099	0.057	0.132
Sum	1.093	0.873	3.035	2.936	3.877	1.854	4.042	3.284

Ranking of criteria using BWM

Table 11 ranks the criteria for the online context based on their average weight and percentage using the BWM method.

Table 11.

Criteria ranking for online using BWM.

In Table 11, the ranking of criteria for online learning reveals that Cost and Time is the most critical factor, representing 25.74% of the weight. This high priority indicates that affordability and time efficiency are essential for students when considering online learning. The emphasis on reducing expenses (such as transportation and material costs) and saving time from commuting and structured class hours highlights the value students place on the financial and logistical benefits of online education. The ability to minimize these commitments makes online learning particularly appealing, allowing students to balance their studies with other financial and time constraints.

Following closely, Flexibility, with a weight of 19.38%, ranks as the second most important criterion for online learning. Flexibility here encompasses the convenience of learning from any location and at any time, which enables students to fit education around other responsibilities. This adaptability is especially beneficial in contexts like Bangladesh, where technological access varies, and students often juggle personal, educational, and professional obligations. The ability to control their schedule without the rigidity of traditional classrooms is a significant advantage for online learners. Mid-ranking criteria include the Pace of Learning (12.56%) and Concentration (12.31%). The weight assigned to the pace of learning indicates that students value being able to adjust the speed of their learning. Many students find the self-paced nature of online learning appealing, as it allows them to tailor the speed of their studies to match their comprehension and schedules. Similarly, concentration, while also significant, suggests that students recognize the criteria of maintaining focus in an online environment, where distractions are more prevalent, and a structured environment may be lacking. These results indicate that an effective online setup should be mindful of students’ need to manage distractions while maintaining a pace conducive to engagement.

Finally, lower-ranking criteria such as Technological Issues (5.27%) and Practicality (7.27%) reflect lesser concern in the online setting, as students seem to prioritize cognitive and logistical benefits over hands-on learning elements. The low weight for technological issues suggests that students are less concerned with technical criteria or have found ways to manage them, while practicality, which involves experiential learning, is naturally less emphasized in online formats due to the limitations of virtual platforms.

In Table 12, which ranks criteria for offline learning, the results differ significantly, highlighting the distinct priorities in traditional settings. Technological Issues top the list with a weight of 19.25%, emphasizing the importance students place on reliable technology even in offline settings, likely due to the limited digital resources and support in these environments. This high ranking suggests that any technological barriers, even minimal, can heavily impact students’ learning experience in offline contexts.

Table 12.

Criteria ranking for offline using BWM.

Concentration is the second highest priority in offline learning (18.47%), underscoring the value of an immersive, distraction-free environment, which traditional classrooms are better equipped to provide. The emphasis here suggests that students view offline environments as better suited to sustained focus and engagement, an advantage that direct, face-to-face learning provides. Following these is Practicality (15.64%), which reflects the hands-on, experiential value offline learning offers, especially important in fields like engineering. The ability to engage in practical, tactile learning experiences strengthens students’ skills and comprehension, elements challenging to replicate online. Effectiveness of Learning (14.46%) and Interaction (13.98%) rank fourth and fifth, respectively, demonstrating that students value direct engagement with instructors and peers, which facilitates deeper understanding and immediate feedback.

In contrast, Cost and Time and Flexibility hold the lowest weights, at 5.21% and 4.16% respectively, indicating that logistical factors are less of a concern in offline education. This disparity between online and offline priorities provides essential insights: while online learning emphasizes flexibility and affordability, offline learning values immersive and interactive experiences. The findings suggest that both online and offline modalities have unique advantages, and future learning models could benefit from blending these priorities to enhance educational effectiveness across different contexts.

For online learning, the weights highlight “Effectiveness of Learning” as the most prioritized criterion, emphasizing its critical role in improving educational outcomes, while “Technological Issues” received a lower weight, reflecting lesser perceived challenges in this area. These outputs indicate that targeted efforts to enhance the effectiveness of online learning tools will yield significant benefits. In offline learning, “Cost and Time” emerged as the most heavily weighted factor, demonstrating its importance in resource-constrained environments like Bangladesh, where traditional classroom settings still dominate. “Flexibility” was the least weighted criterion, indicating a lower emphasis on adaptable schedules in offline contexts.

Additionally, graphical representations have been included to visualize these weight distributions for better clarity. Each weight and ranking have been interpreted in the context of the participants’ responses, linking them to practical implications for educational policy and planning. This approach ensures that even readers unfamiliar with BWM can understand the relevance and application of the findings.

Sensitivity analysis

Sensitivity analysis in the Best Worst Method (BWM) is essential for evaluating the robustness of judgments by analyzing the impact of fluctuations in criterion weights on the final outcome. It aids in pinpointing the paramount criteria, guiding decision-makers to focus on critical matters. The approach increases confidence in the stability and validity of results, despite variations in assessments. It also enhances risk management by clarifying the impacts of ambiguity. Sensitivity analysis enables the exploration of many possibilities through “what-if” scenarios. Ultimately, it ensures that the MCDM process is robust, reliable, and validated.

Sensitivity analysis for online learning

The weight of “C1-Cost and Time” has been adjusted in this study from 0.1 to 0.9, examining its impact on other identified criteria in online learning. Upon altering the weight of the “C1-Cost and Time” problem from 0.1 to 0.9, the proportional significance of each challenge is encapsulated in Table 13. Consequently, Table 14 illustrates the hierarchy of the criteria rated.

Table 13.

The weights of eight criteria of online learning during sensitivity analysis.

Selected criteria		Values denoting the optimal relative significance of the criteria of online learning
	Normal (0.2575)	0.1	0.2	0.3	0.4	0.5	0.6	0.7	0.8	0.9
CR1	0.2575	0.1000	0.2000	0.3000	0.4000	0.5000	0.6000	0.7000	0.8000	0.9000
CR2	0.1939	0.2350	0.2089	0.1828	0.1567	0.1305	0.1044	0.0783	0.0522	0.0261
CR3	0.0886	0.1074	0.0955	0.0835	0.0716	0.0597	0.0477	0.0358	0.0239	0.0119
CR4	0.0856	0.1038	0.0922	0.0807	0.0692	0.0576	0.0461	0.0346	0.0231	0.0115
CR5	0.1232	0.1494	0.1328	0.1162	0.0996	0.0830	0.0664	0.0498	0.0332	0.0166
CR6	0.1257	0.1523	0.1354	0.1185	0.1015	0.0846	0.0677	0.0508	0.0338	0.0169
CR7	0.0528	0.0640	0.0569	0.0498	0.0426	0.0355	0.0284	0.0213	0.0142	0.0071
CR8	0.0728	0.0882	0.0784	0.0686	0.0588	0.0490	0.0392	0.0294	0.0196	0.0098
Total	1.0000	1.0000	1.0000	1.0000	1.0000	1.0000	1.0000	1.0000	1.0000	1.0000

Table 14.

The ranking of selected online criteria utilizing sensitivity analysis.

Selected criteria		Ranking of criteria (online)
	Normal (0.2575)	0.1	0.2	0.3	0.4	0.5	0.6	0.7	0.8	0.9
CR1		1	6	2	1	1	1	1	1	1
CR2		2	1	1	2	2	4	2	4	4
CR3		5	4	5	5	5	5	5	5	5
CR4		6	5	6	6	6	7	6	7	7
CR5		4	3	4	4	4	3	4	3	3
CR6		3	2	3	3	3	2	3	2	2
CR7		8	8	8	8	8	9	8	9	9
CR8		7	7	7	7	7	6	7	6	6

Simultaneously, the scales of further criteria are modified. Table 14 displays a ranking of the identified criteria associated with the online learning, derived from sensitivity analysis. Table 14; Fig. 4, and Fig. 4 illustrate the findings of the sensitivity study, indicating that the factor designated as “C1-Cost and Time” consistently receives the highest rating. In contrast, the challenge labeled “C7-Technological Issues” routinely attains the lowest grade. Figures 4 and 5 illustrate the changes in weight and ranks that occurred throughout the sensitivity testing. Sensitivity study substantiates the claim that outcomes generated by the Best Worst Method (BWM) demonstrate consistency, absence of bias, and enhanced trustworthiness.

Fig. 4.

Graphical representation of weights of the criteria of online education during sensitivity analysis.

Fig. 5.

Spider chart of ranking of the online learning criteria after sensitivity analysis.

Stability in sensitivity analysis ensures that criteria like cost, internet accessibility, and flexibility maintain their priority rankings even when input data or assumptions slightly change. For policymakers, this stability validates investments in online infrastructure, such as improving internet access or subsidizing digital devices, ensuring these initiatives have a consistent impact. Education planners can confidently design online learning strategies knowing these factors will reliably enhance accessibility and student engagement in Bangladesh.

Sensitivity analysis for offline learning

The similar procedures of sensitivity analysis have been carried out for offline learning system. Here the weight of “C7-Technological Issues” has been modified from 0.1 to 0.9, analyzing its influence on other specified criteria in online learning. By adjusting the weight of the “C7-Technological Issues” issue from 0.1 to 0.9, the relative importance of each criterion is summarized in Table 15. Thus, Table 16 delineates the hierarchy of the assessed criteria.

Table 15.

The weights of eight criteria of offline learning during sensitivity analysis.

Selected criteria		Values denoting the optimal relative significance of the criteria of offline learning
	Normal (0.1925)	0.1	0.2	0.3	0.4	0.5	0.6	0.7	0.8	0.9
CR1	0.0521	0.0580	0.0516	0.0451	0.0387	0.0322	0.0258	0.0193	0.0129	0.0064
CR2	0.0416	0.0463	0.0412	0.0360	0.0309	0.0257	0.0206	0.0154	0.0103	0.0051
CR3	0.1446	0.1611	0.1432	0.1253	0.1074	0.0895	0.0716	0.0537	0.0358	0.0179
CR4	0.1398	0.1559	0.1386	0.1212	0.1039	0.0866	0.0693	0.0520	0.0346	0.0173
CR5	0.1847	0.2058	0.1830	0.1601	0.1372	0.1144	0.0915	0.0686	0.0457	0.0229
CR6	0.0883	0.0984	0.0875	0.0766	0.0656	0.0547	0.0437	0.0328	0.0219	0.0109
CR7	0.1925	0.1000	0.2000	0.3000	0.4000	0.5000	0.6000	0.7000	0.8000	0.9000
CR8	0.1564	0.1744	0.1550	0.1356	0.1162	0.0969	0.0775	0.0581	0.0387	0.0194
Total	1.0000	1.0000	1.0000	1.0000	1.0000	1.0000	1.0000	1.0000	1.0000	1.0000

Table 16.

The ranking of selected offline criteria utilizing sensitivity analysis.

Selected criteria		Ranking of criteria (offline)
	Normal (0.2575)	0.1	0.2	0.3	0.4	0.5	0.6	0.7	0.8	0.9
CR1		1	6	2	1	1	1	1	1	1
CR2		2	1	1	2	2	4	2	4	4
CR3		5	4	5	5	5	5	5	5	5
CR4		6	5	6	6	6	7	6	7	7
CR5		4	3	4	4	4	3	4	3	3
CR6		3	2	3	3	3	2	3	2	2
CR7		8	8	8	8	8	9	8	9	9
CR8		7	7	7	7	7	6	7	6	6

The criterion scales are concurrently adjusted. Table 16 presents a rating of the found factors related to online learning, obtained via sensitivity analysis. Table 16; Figs. 6, and 7 present the results of the sensitivity analysis, demonstrating that the variable labeled “C7-Technological Issues” regularly attains the highest rating. Conversely, the challenge designated “C2-Flexibility” consistently receives the lowest rating. Figures 6 and 7 depict the variations in weight and rankings observed during the sensitivity testing. The sensitivity research confirms that the results produced by the Best Worst Method (BWM) for offline learning also exhibit consistency, lack bias, and possess increased reliability.

Fig. 6.

Graphical representation of weights of the criteria of offline education during sensitivity analysis.

Fig. 7.

Spider chart of ranking of the offline criteria after sensitivity analysis.

For offline learning, stability in sensitivity analysis confirms the importance of criteria like practical engagement, structured environments, and student-teacher interaction, regardless of minor variations in assumptions. This reliability supports policymakers in prioritizing investments in laboratory facilities, hands-on training, and teacher development. Education planners can trust that focusing on these aspects will consistently improve the effectiveness of offline education for engineering students in Bangladesh.

Discussion

The findings of this study provide valuable insights into the prioritized criteria for online and offline learning among students in Bangladesh, highlighting both logistical and cognitive factors that influence learning experiences. “Cost and Time” emerged as a highly influential criterion in online learning, drawing connections to socio-economic factors in Bangladesh. During the pandemic, many families faced financial hardships, making affordability a critical factor for accessing education. Online learning provided a cost-effective alternative by eliminating travel and accommodation expenses, which are significant for students from rural areas. Time constraints also played a role, as online platforms allowed for greater scheduling flexibility, enabling students to balance education with household responsibilities or part-time jobs. These socio-economic realities highlight the importance of prioritizing cost and time efficiency in designing future online learning strategies for Bangladesh.

The findings provide valuable insights for developing blended learning models and shaping educational policies in Bangladesh and similar developing countries. By integrating the flexibility and affordability of online platforms with the interactive and practical advantages of offline settings, blended learning can address diverse student needs. However, potential barriers to its adoption must be considered, including resistance from educators who may prefer traditional methods and lack the necessary training in hybrid models. Additionally, insufficient digital infrastructure, such as unreliable internet and the high cost of devices, poses significant challenges. To overcome these barriers, policymakers should prioritize targeted training programs for educators to build their capacity for delivering hybrid instruction while simultaneously investing in digital and offline infrastructure. These strategies will ensure the effective implementation and sustainability of blended learning, enhancing educational resilience in resource-constrained environments.

Through the Best Worst Method (BWM), Cost and Time emerged as the top priority in online learning, closely followed by Flexibility, confirming previous studies that emphasize online education’s appeal due to its affordability and time efficiency. This aligns with Dhawan’s¹⁴ findings, which noted that the ease of learning at one’s own pace and reduced commuting costs make online learning particularly attractive in resource-limited settings. Similarly, Adnan and Anwar’s¹ study found that online learning offers flexibility and convenience, supporting our findings that flexibility is highly valued in online formats. This study by Abuhassna et al. (2020) complements our research by highlighting factors like collaboration, autonomy, and interaction, which positively influence satisfaction and academic achievement in online learning platforms. These insights align with our findings, emphasizing the need to address such factors when designing blended learning models for engineering students in Bangladesh.

However, the ranking of Technological Issues as the least important factor in online learning diverges from Adedoyin and Soykan’s²¹ results, which highlighted significant barriers due to limited internet access in developing countries. This discrepancy may be due to adaptive strategies students in this study have adopted to manage technological challenges. In offline learning, Technological Issues emerged as the highest priority, consistent with Gonzalez et al.⁸ and Muangmee et al.³, who identified technology as crucial for ensuring efficient offline education and overcoming disruptions due to the pandemic.

Additionally, Concentration and Practicality ranked higher in offline settings than in online settings, a trend observed in Kolb’s²³ work on experiential learning, which emphasizes the value of hands-on practice, particularly for engineering students. This supports the relevance of structured, interactive environments in traditional classrooms, as noted by Dewey⁴⁶. Lastly, Flexibility was ranked as the least significant criterion in offline learning, echoing Schramm²⁹, who highlighted the rigid nature of traditional classrooms as less adaptable to individual schedules. These findings underscore the nuanced differences in learning needs across online and offline modalities, providing a framework for educational improvements tailored to Bangladesh’s unique educational landscape.

Section: qualitative insights into student experiences

Participant feedback provides valuable context to the findings, illustrating the lived experiences of students navigating online and offline learning. One student shared, “Online classes helped me save on transportation costs and time, which was crucial during the pandemic. However, the cost of mobile data and frequent power outages made it challenging to stay consistent with my studies.” This highlights the duality of online learning as both a cost-effective alternative and a source of frustration due to technological barriers. Another student remarked, “The flexibility of online learning allowed me to manage both household responsibilities and my studies, but I missed the structured environment of in-person classes.” Such sentiments emphasize the trade-off between flexibility and the need for structured guidance in academic settings.

Another students underscored the importance of practical engagement, with one stating, “As an engineering student, practical lab sessions are irreplaceable. In offline classes, I could directly interact with equipment and instructors, which is impossible to replicate online.” Another participant voiced concerns about online disruptions, stating, “I often struggled with unstable internet connections during online classes, which disrupted my focus and left me feeling disconnected from the course.” Despite these challenges, many participants expressed optimism about hybrid models, with one summarizing, “A combination of online and offline learning would be ideal. Online platforms provide flexibility, but offline sessions are essential for building practical skills and fostering deeper interactions with teachers and peers.”

These qualitative insights enrich the study by humanizing the statistical outputs and providing actionable implications for policymakers to consider in future educational strategies.

Conclusion

Key findings

This study provides a detailed evaluation of the effectiveness of online and offline learning modalities for engineering students in Bangladesh, utilizing the Best Worst Method (BWM) to rank key criteria. The novelty of applying BWM lies in its ability to prioritize and rank multiple criteria systematically, ensuring consistent and reliable decision-making, even with small expert-focused samples, making it uniquely suited for evaluating complex issues like online and offline learning effectiveness in resource-constrained contexts. The findings highlight that “Cost and Time” and “Flexibility” are the most influential factors in online learning, showcasing its significant advantages in affordability and adaptability, particularly for students in resource-constrained settings. However, challenges such as unstable internet connections and limited device accessibility remain critical barriers. For offline learning, “Technological Issues” and “Concentration” emerged as top priorities, emphasizing the importance of practical, hands-on experiences and structured environments essential for engineering education. These insights highlight the complementary strengths of both modalities and provide a basis for developing blended learning strategies that address the unique needs of Bangladeshi students.

Policy implications

This study offers actionable recommendations to guide policymakers in improving education systems in Bangladesh and similar contexts. To address affordability challenges in online learning, specific interventions include providing subsidized internet access and offering low-cost digital devices to students from low-income households. For offline learning, investments in laboratory facilities and other infrastructure are essential to support hands-on training, particularly for engineering disciplines. Additionally, developing blended learning modules that integrate the flexibility of online platforms with the interactive and practical benefits of offline learning can create a more effective and balanced educational framework. Policymakers should also focus on training educators in hybrid teaching methods and equipping them with modern digital tools to ensure the successful implementation of blended approaches. By implementing these strategies, policymakers can create a sustainable and inclusive education system, bridging the digital divide and enhancing the quality of learning in resource-constrained environments.

Limitations and future research directions

This study provides valuable insights into the effectiveness of online and offline learning for engineering students in Bangladesh but has certain limitations. The exclusive focus on engineering students limits the generalizability of the findings to other disciplines; future research should explore diverse academic fields for broader applicability. The Best Worst Method (BWM) is effective for prioritizing factors in structured contexts but has limitations, such as its reliance on expert judgment and inability to capture qualitative nuances or dynamic interactions between criteria. While suitable for focused studies like this one, its applicability to diverse educational contexts may require adaptations. Combining BWM with qualitative methods could address these limitations and enhance its generalizability. While the Best Worst Method (BWM) offers robust results, the reliance on a small, expert-focused sample introduces potential selection bias. Expanding the sample to include students, educators, and policymakers could enhance representativeness.

The study also relies primarily on quantitative methods, limiting the depth of participant perspectives. Incorporating mixed methods, such as interviews or focus groups, could provide richer insights into the experiences of learners and educators. Additionally, while the findings are contextually relevant to Bangladesh, cross-country comparisons could explore how local socio-economic factors influence learning effectiveness. Finally, this study focuses on learning during the pandemic; future research should examine the long-term impacts of blended learning on academic performance and satisfaction in a post-pandemic setting. Addressing these limitations will strengthen the understanding and implementation of effective learning strategies in diverse contexts.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Syeda Sharaban Tahura: Conceptualization, Methodology, Data collection Software, Investigation, Writing- Original draft preparation, Md. Abdus Shabur: Conceptualization, Methodology, Supervision, Validation, Writing- Original draft preparation, Review and editing Md. Tasnuva Jahan Nuva: Project administration, Visualization, Review and, editing. All authors read and approved the final manuscript.

Data availability

The datasets generated are available from the corresponding author upon reasonable request.

Declarations

Ethical approval and consent to participate

This study was approved by the Department of Mechatronics and Industrial Engineering, Chittagong University of Engineering and Technology, Chattogram-4349, Bangladesh. The ethical approval number is- CUET/MIE/2023/1709028 which certify that the study was performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. Informed consent was obtained from all subjects and/or their legal guardian(s).

Competing interests

The authors declare no competing interests.