The Educational Digital Divide for Vulnerable Students in the Pandemic: Towards the New Agenda 2030

Abstract

The COVID-19 pandemic has caused school closures worldwide and has disrupted nearly 1.6 billion students across the globe. This has widened existing digital gaps and has caused vulnerable students to be further digitally displaced. In efforts to mitigate this issue, various strategies have been used to cater for the educational digital divide of vulnerable students. However, there is a lack of studies investigating the relationship between access and connectivity of learning and use and exploitation of technology, particularly with regards to iPads during the pandemic. Thus, the present study investigates this scenario by examining the digital educational divide for vulnerable students in the pandemic, in terms of access and connectivity and use and exploitation. A survey was distributed to 518 vulnerable students in schools between the ages of 10 and 15 years old, and results were analyzed using partial least squares–structural equation modeling (PLS-SEM). The findings indicate that asynchronous learning is a stronger construct than synchronous learning, while creativity skills was stronger than productivity skills with regard to the use and exploitation of technology for pandemic learning of vulnerable students. This study’s findings could assist future developers and educators in the development of effective emergency teaching and learning strategies and design.

1. Introduction

1.1. Global and Local Educational Scenario due to the COVID-19 Pandemic

The COVID-19 pandemic has caused school closure worldwide and has disrupted nearly 1.6 billion students across the globe. The closure of schools has impacted educational systems with regard to short- and immediate long-term effects, in which short-term effects include learning loss and school dropout, whereas long-term ones include threats to gender equality, health, well-being, and digital equity and inclusion [1]. While it might be too soon to grasp the full extent of the impact of school closure, this study describes an effort by the national government to mitigate harm and safeguard process to digital equity and inclusion in and through remote learning via its 1:1 iPad initiative.

Ref. [1] reported that by August 2020, school closures had occurred in at least 194 countries, which affected nearly 1.6 billion learners, equivalent to over 90 percent of the global population of learners. It was also reported that by mid-September 2021, schools worldwide had been fully closed for an average of 18 weeks since the outbreak of COVID-19. This had an immense effect on face-to-face learning, as partial school closures have caused nearly the equivalent of two-thirds of a typical school year of learning to be lost due to the pandemic. As nations devised mitigation plans for school closure globally, new challenges and issues arose as nations were not fully equipped to implement remote learning on a large scale.

It has been reported that nearly 500 million learners from pre-primary to upper secondary school had no access to any remote learning modality, among whom 75% lived in the poorest households. Learning modalities varied from high-technological learning modalities (online platforms that required digital access) to low-technological learning modalities (television and radio broadcasts) to non-technological learning modality (textbooks). Early in the pandemic, in April 2020, it was discovered that 110 of 118 countries had developed policies related to the implementation of at least one learning modality for students in lower education. By July 2020, it was reported that 134 of 149 countries had implemented high-technological and low-technological learning modalities. As for non-technological ones, it was reported that 106 nations used radio for learning, while 127 countries provided take-home materials for learning. After a year of school closure, it was discovered that nations implemented multi-modality learning approaches, in which 143 nations integrated more than three learning modalities that included learning via online platforms, mobile phones, television, radio, take-home packages, and other distance-learning platforms [1].

The effects of the impact of school closure have been discussed. The first effect is related to gender expectations, where male and female school students spend their time at home during the COVID-19 pandemic. Gender roles result in the fact that female students might be expected to take a higher percentage of domestic labor and spend more time taking on roles such as caretakers when not at school. In addition, female students might also be expected to take on additional home duties such as cooking and childcare at home. The situation leads to less time participating in live and scheduled learning that incorporates remote learning as well as television and radio broadcasts. Furthermore, highly remote learning strategies also have an impact on inequities with regard to the gender digital divide. It has been reported that female school students are less likely to own or have access to smartphones and internet-enabled devices, and they are more likely to have limited skills and confidence using technology, thus causing them to be less likely to be involved and benefit from high-technological learning modalities implemented during school closures [1].

However, male school students were also deprived of learning in certain contexts, as poverty and economic pressure might have caused them to be involved in their households’ need for additional income via pandemic-related jobs, thus reducing their time spent learning [1]. Other effects were related to access to school dropouts. It has been reported that 23.8% of school students from pre-lower to higher education may have dropped out or have no access to school due to the pandemic situation [2]. The World Bank has concluded that among female and male school students aged 12 to 17, female students in low-, lower-, and middle-income countries have a higher risk of not returning to school [3]. In a related study among poor households in three Indian states, the findings revealed that approximately 80 percent of male and female students who did not experience food or monetary shortages during COVID-19 lockdowns reported that they would return to school after schools reopened. However, contradicting findings were discovered in the study, as 50 percent of the students who faced food or monetary shortages during lockdowns stated that they were uncertain about returning to school after the reopening of schools [4]. In another study in Kenya, Ref. [5] discovered that 16 percent of female students and 8 percent of male students aged 15 to 19 did not enroll during the two months after school reopening. In contrast, in Mexico, male students were at a higher risk of becoming school dropouts as compared to female students. Out of the one-quarter of students aged 14 to 17 in Chiapas, Mexico, male students obtained the highest dropout rates at 37.5 percent [5].

1.2. Malaysia Educational Scenario and Mitigation Initiatives

Similar to the global situation, Malaysian students have also suffered from school closures and the suspension of formal classroom learning. Since March 2020, school closures have impacted 4.9 million students nationwide. The national government has responded with several mitigation strategies, including multiple learning modalities such as high-technological, low-technological, and non-technological learning modalities. High-technological were implemented by the government via the Digital Learning Initiative Malaysia (DELIMa) initiated by the Education Ministry in collaboration with Apple, Microsoft, and Google, where Apple provided pedagogical and technical training to teachers nationwide, Microsoft to students, and Google provided a national learning management system [6].

However, access to online education has proven to be a challenge, particularly for low-income households in Malaysia, and low device ownership is a major issue, which makes the implementation of online learning quite challenging. Ref. [7] reported that the COVID-19 pandemic negatively changed the mean monthly household gross income from MYR 7901 (USD 1865) to MYR 7089 (USD 1673) in 2020. In terms of household distribution by decile, a majority of households experienced a decline in income, where many households from a higher decile group shifted to a lower income group. In 2020, it was reported that there was an additional 12.5 percent of households with income less than RM2500 (USD 590). Statistics also reported that 20 percent of 580,000 households in Malaysia shifted from middle-income groups to lower-income groups during the pandemic. The scenario revealed the socio-economic disparities of lower-income communities and a lack of learning-device ownership in these communities.

Thus, in the wake of the challenges faced by the low-income school students in Malaysia, the Ministry of Finance, Ministry of Education, and CERDIK, a government-linked company (GLC) and government-linked investment company (GLIC) under its corporate responsibility pilot initiative, provided digital access, which included iPads, laptops, and data connectivity, to primary and secondary school students for 150,000 students in lower-income families in Malaysia. The initiative involved contributions worth RM 150 million (USD 35.4 million) from 31 GLCs and GLICs, which aimed to provide underprivileged students with devices, access to data, and learning connectivity and bridge the gap in terms of low learning-device ownership and internet connectivity for remote learning. Since 29 October 2021, 150,000 devices were distributed nationwide in 1264 schools in 144 educational districts in Malaysia. This study only focuses on the iPad initiative, where 30,000 iPads were distributed to 30,000 students in 16 districts, involving 1000 administrators and 2500 teachers.

As such, the purpose of the research was to investigate the digital divide in education among vulnerable students in the COVID-19 pandemic, in terms of the access and connectivity and the use and exploitation of information and communication technology (ICT), particularly iPads with mobile and internet connections. The research contributes to the body of knowledge in terms of a new research model for the field analyzed using partial least squares–structural equation modeling (PLS-SEM), with regard to aspects such as access and connectivity and use and exploitation and elements such as synchronous learning, asynchronous learning, productivity skills, and creativity skills, among vulnerable students during the pandemic. The paper is structured as follows. The paper starts with a literature review on the digital divide and digital inclusion. This is followed by a methodology section involving the research model applied, participant selection, data collection procedures, online learning environment used, and data analysis procedures. The results and discussion are then presented to highlight the contribution of the study to the field, as well as the assessment of the research hypotheses. The paper ends with conclusions, implications, and future directions in the field.

2. Literature Review: Digital Divide and Digital Inclusion

The digital divide is defined as the gap between those who have and do not have access to information and communication technology (ICT) that includes computers, smartphones, and the internet [8], as well as the essential skills to benefit fully from information society [9]. The digital divide generates an inequality of opportunities that are important for improving standards of living [8], which can be categorized according to criteria that explain the difference in participation such as age, education, gender, geographic location, income, and socioeconomic status [9]. In 2016, the United Nations declared the significance of employing a human-rights-based approach in providing and increasing internet access, demanding that all countries make efforts to bridge the digital divide [10]. The United Nations announced access to the internet as a human right, and certain countries have already made it a government policy [10]. After all, the act does not mean that the service is provided for free but considers internet access as a public utility similar to electricity and water [10]. Research by [8] identified two dimensions of the digital divide: access and connectivity and use and exploitation. Access and connectivity aspect is related to how vulnerable students are required to have access to ICT for learning, while use and exploitation aspect refers to the competency of ICT usage and being able to benefit from the technologies and hence to exploit the advantages for learning [8]. The COVID-19 pandemic has highlighted the digital education divide among the vulnerable and has shown severe implications of that vulnerability, in which the internet and ICT have become a necessity not only for education, but also for health, well-being, and survival. In a situation where restrictions are imposed for social and physical distancing, access and connectivity to ICT is crucial for learning, as is use and exploitation for learning purposes. This is important in ensuring vulnerable students are prevented from becoming further marginalized and socially excluded due to the digital divide caused by the pandemic [8,11,12].

Similarly, the term “digital inclusion” refers to government policies and strategies aiming at a country’s digital transformation by ensuring all individuals and communities, including the underprivileged and vulnerable ones, have meaningful access to and use of ICT [13]. Ref. [13] also discussed multidimensional aspects of meaningful connectivity, including internet usage, device appropriateness, data sufficiency, and fast connection. Thus, bridging the digital divide should be geared towards various aspects, ranging from the technological accessibility to the ability and skills to use those technologies.

In general, technology accessibility, skills, and usage are contributing factors related to the concept of the digital divide. In a report by Cambridge Assessment, Ref. [14] explained that there are four levels of the digital divide in COVID-19: (i) the digital divide in relation to access; (ii) the digital divide related to digital skills and types of digital media; (iii) the digital divide related to outcomes of technology access, skills, and usage; (iv) the digital divide related to a variety of other social and environmental factors. At the first level, the digital divide was originally conceptualized to include physical access to devices and the internet and has now been expanded to include types of digital technologies and the quality of access that they offer. For example, in remote digital education, the quality of education could be affected by aspects such as screen size, up-to-date software, and audiovisual capabilities that differ across devices such as laptops and mobile phones. Meanwhile, the second level of the digital divide concerns digital skills and types of digital media. Ref. [14] emphasized that it is often assumed that children who are “digital natives” have strong digital skills, yet this may not be true in some cases. Additionally, at this level of the digital divide, factors such as device usage frequency and usage for learning or entertainment aims are deemed to be related to digital inequalities. On the third level, the digital divide has been linked to the outcome of technology access, skills, and usage, as well as the effects of technology access, skills, and use on educational performance. At the fourth level, the digital divide is related to children’s surroundings such as parental support, home learning infrastructure, home learning environment (e.g., size of learning space, proximity of the other sibling who is also learning at home), and teachers’ access, skills, and usage of digital technologies.

With regard to the global digital divide scenario during the pandemic, in the United Kingdom (the UK), digital exclusion was reported to be more severe for secondary school children as compared to primary [14]. Reports have suggested that remote education provisions for the UK were consistent during the early phases of the COVID-19 pandemic, during which a variety of digital remote education techniques and physical resources were applied. Later, in January 2021, lockdowns accelerated the level of synchronous digital remote learning strategies such as the usage of video conferencing for live classroom lessons, in which private schools, but not public schools were reported to be hosting live online lessons. Additionally, some schools also resorted to physical resources for vulnerable learners (e.g., sending physical learning packs), to cater to students who had difficulties in learning such as limited technological access and connectivity [14,15,16,17,18]. In Nigeria, Ref. [19] reported that access to remote learning was quite different, where challenges were related to the affordability of internet data and phone credit as well as electricity and access to devices. The study also reported that the internet penetration in Nigeria is still relatively low, such that only 42% of the population is online. As is the case in the UK, this study discovered that there were differences in access to digital tools in public and private schools, in which public schools had difficulties in access as compared to private schools. This further resulted in academic engagement levels, where students in private schools had a higher engagement level as compared to public ones.

In the United States (the US), Ref. [20] reported that variations in internet access in rural areas had limited distance education, causing children to be left behind. It was reported that 51.6% of rural US residents had had 250/25 megabits per second (Mbps) internet access in 2018 compared to 94% of urban residents. On top of that, there are also additional factors that can affect connectivity speed, including multiple devices within a household, internet usage in peak time, connectivity hardware, technological infrastructure, and distance to servers. Furthermore, although schools and public places offer a higher speed of internet connectivity as compared to students’ homes, rural children face further challenges due to distance and travel mechanisms to reach these access points [20,21]. In a related study, Ref. [22] reported that many rural residents only had access to one service provider for home internet. Other than that, it was reported that fifteen states in the US have median download speeds of less than 25 Mbps. These data were collected from 7.2 million individual households from 30 December 2019 through 30 June 2020. This situation has a major impact on access to learning, in which speeds less than 25 Mbps are only capable of supporting one or two devices simultaneously, while high-quality learning content on multiple devices would require over 50 Mbps [20].

With regard to the digital divide in the local context, since 2015, mobile broadband subscription rates in Malaysia showed substantial growth, which can be attributed to the migration from 2G/3G to 4G that improved the quality of service offered through mobile broadband [23]. Additionally, smartphones are more widely affordable, as the selling prices tend to drop when the models mature on the market. Meanwhile, telecommunications companies actively promote mobile data packages that include an internet plan and mobile devices through close collaboration with various partners [23]. However, in reality, most vulnerable communities are less digitally included and need better connectivity to compensate for the disadvantages associated with being remote. A video of Veveonah Mosibin, a university student in Sabah, who spent 24 h in a tree for internet connectivity to take her online examination made international headlines in June 2020. Sabah is one of 13 states of Malaysia, located on the island of Borneo, in East Malaysia, and its geography is naturally mountainous. This incident was an example of a digital territorial divide that occurred between rural and urban areas, as well as between low-population or unpopulated territories. The Malaysian Communications and Multimedia Commission then released a statement that they were going to build a new telecommunications tower in the area, upgrading the 3G coverage to 4G [23]. Regardless of the demographic and socioeconomic status of an individual or group, internet access and quality of service will be made available depending on the residential area [8]. In other words, the digital divide in connectivity and use negatively impacts rural communities, creating pockets of marginalization and exclusion that put those vulnerable populations at risk of being left behind in education. Additionally, if the connectivity and technology use in rural areas are not improved, not only education but also other public services such as health, social services and other opportunities will be lost.

According to Article 26 of the Universal Declaration of Human Rights [24], everyone is entitled to education as a human right. Additionally, SDG-Education 2030 agenda states that all vulnerable groups should have the right to education [25]. Vulnerable students refer to learners who are living in marginalized or vulnerable situations including gender inequality, physical or mental disability, and low socio-economic status, as well as in rural, urban, post-conflict, or post-disaster situations [25]. However, in an emergency context such as the COVID-19 pandemic, ensuring and protecting the right to education for marginalized vulnerable groups can be difficult. During emergencies, the value of education is not considered as lifesaving, yet it can bring emotional stability to parents and students. To some extent, education can help people affected by the COVID-19 pandemic in reintegrating back into society after the emergency is over and preventing similar occurrences in the future [25]. In light of this event, UNESCO ensures that human rights law is applied across all contexts by the international community to minimize the harmful effects of the pandemic. As such, our study focuses on vulnerable students who are living in marginalized situations, specifically those with low socioeconomic status and living in difficult or disaster situations, in this case, the COVID-19 pandemic.

The COVID-19 pandemic has exposed the digital divide and digital inclusion in both developed and developing countries. Governments and private sectors have been giving attention to policies and projects that could help improve the digital inequalities and inadequacies in our education systems [3,4,5]. According to [3], mitigation effectiveness for high-income countries to compensate for school closures and learning losses was high, ranging from 15% to 60%, in terms of technology and internet access. For upper-middle and lower-middle income countries, the government’s ability to mitigate this crisis was moderate, ranging from 7% to 40%, reflecting household access to mobile phones, computers, and the internet. In low-income countries, technology and internet access were around 7% and 6%, respectively, and this low effectiveness limits the ability of governments to mitigate this crisis. COVID-19 provided an opportunity to reduce learning poverty, address inequality, and reimagine education. Schools and educational institutions need to change to prepare students for the future and ensure that all students are learning. A successful strategy for online and remote learning relies on many delivery approaches [3,4,5]. The efforts made for online and remote learning, such as digital content and assessments, as well as online training and support for teachers, are more personalized, providing a more resilient education [3].

3. Method

3.1. Research Model Used for Data Collection and Analysis

The study investigates the educational digital divide of vulnerable students in the pandemic and suggests implications of the findings for moving towards the new Agenda 2030. The modified research model used for data collection and analysis is based on the study of [8] and Apple’s Everyone Can Create curriculum [26], illustrated in Figure 1. The constructs of the model were based on the study by [8], where the digital divide was categorized under two constructs: (i) access and connectivity and (ii) use and exploitation. For the sub-constructs and indicators, they were further classified under asynchronous learning and synchronous learning with regard to internet browsing and text-based messaging for asynchronous learning and learning management system (LMS) learning, video and audio chatting, and video conferencing for synchronous learning. As for the use and exploitation construct, two sub-constructs were categorized according to Apple’s Everyone Can Create curriculum, where the indicators of digital note-taking, mathematical calculation, and presentation slide creation were categorized under productivity skills, whereas augmented reality (AR) creation, animation creation, and digital book creation were categorized under creativity skills. All constructs, sub-constructs, and indicators of the research model are summarized in

Table 1. The constructs, sub-constructs, and respective indicators of the research model.

Construct	Sub-Construct	Indicator
Access and connectivity	Asynchronous learning	Internet browsing LMS learning Text-based messaging
	Synchronous learning	Video and audio chatting Video conferencing
Use and exploitation	Productivity skills	Digital note-taking Mathematical calculation Presentation slide creation
	Creativity skills	Augmented reality (AR) creation Animation creation Digital book creation Digital music composition Video production

. The research hypotheses are as follows:

H1. 

Access and connectivity is positively related to asynchronous learning.

H2. 

Access and connectivity is positively related to synchronous learning.

H3. 

Use and exploitation is positively related to access and connectivity.

H4. 

Use and exploitation is positively related to creativity skills.

H5. 

Use and exploitation is positively related to productivity skills.

3.2. Participants

The participants consisted of 518 vulnerable students in schools between the ages of 10 and 15 years old. The participants were from vulnerable communities and low-income households with a maximum household income of RM4850 (1050 EUR) and no access to connectivity and technology at their homes. All the participants did not have access to iPads with internet connectivity, and this was the first time that they used iPads for learning purposes. The iPads were provided for all 518 students from the CERDIK program, in which the participants were from the north, south, east, and west regions of peninsular Malaysia and the states of Sabah and Sarawak.

3.3. Data Collection Procedure

An online questionnaire was developed using Google Forms, consisting of two sections and 13 questions. The questionnaire was designed to assess the demographical data of students and the frequency of iPad usage. The demographical section was designed to investigate aspects including gender, age, household income, school location, access to iPads with internet connectivity, and usage of iPads for learning purposes. Meanwhile, for the second section, questions were designed in terms of access and connectivity as well as use and exploitation for four categories: asynchronous learning, synchronous learning, productivity skills, and creativity skills. For asynchronous learning, the questions covered iPad usage frequency, which involved three elements: internet browsing using Safari (internet browsing), LMS learning using the national learning management platform (LMS learning), and text-based messaging using Messages. Meanwhile, synchronous learning included questions covering iPad usage frequency with regard to video and audio chatting using Facetime (video audio chatting) and video conferencing (video conferencing). As for the productivity skills category, the three questions covered iPad usage frequency in terms of digital note-taking using Notes (digital note-taking), mathematical calculations using Numbers (mathematical calculation), and presentation slides creation using Keynote (presentation slide creation). With regard to creativity skills, five questions covered iPad usage frequency in terms of augmented reality content using Reality Composer (AR creation), animation creation using Clips (animation creation), digital book creation using Notes (digital book creation), digital music composition using GarageBand (digital music composition), and video production using iMovie (video production). The questionnaire was developed by a panel of experts consisting of subject matter experts, educational technologists, professors and lecturers, instructional designers, technical experts, and language experts.

The data were collected via online surveys distributed to 518 vulnerable students at lower education levels, from schools located in the north, south, east, and west regions of peninsular Malaysia (i.e., 12 states), and the states of Sabah and Sarawak. The questionnaires were distributed in November 2021 to investigate iPad usage for online learning after a two-month period of receiving their devices during the COVID-19 pandemic. All received responses were based on purposive sampling, and the respondents were from throughout Malaysia. The online survey method was chosen due to geographical restrictions imposed during the pandemic. It is worth noting that the response rate of the survey was 86 percent, in which the answering of the survey was assisted by teachers and parents as needed. Online workshops were conducted by educational technology consultants for students according to regions during the two-month period. The questionnaire consisted of questions related to constructs, sub-constructs, and items of the research model related to iPad access and connectivity and iPad use and exploitation, as well as sub-constructs that include synchronous and synchronous learning, and productivity and creativity skills, as summarized in Table 1. The questionnaire was on a 10-point scale ranging 1 to 10 (very infrequent to very frequent). The questionnaire was reviewed by a group of experts comprising school administrators, schoolteachers, educational technologists, instructional designers, technical experts, and language experts. Consent was received from the participants for their participation in the survey.

3.4. Learning Management System Used as an Online Learning Environment

The national learning management system (i.e., Digital Educational Learning Initiative Malaysia or DELIMa) was used as an online learning environment for the study. The system was developed by the Ministry of Education Malaysia in collaboration with Apple, Google, and Microsoft. The LMS was launched on 16 June 2020 [6]. Resources related to iPad usage in teaching and learning can be accessed under the student menu “Apple Resources,” as shown in Figure 2. Helpdesk support with regards to technical and operational support was also provided according to regions asynchronously via the LMS, and synchronously via social media using instant messaging.

3.5. Data Analysis Procedure

The data collected were analyzed using partial least square–structural equation modeling (PLS-SEM). This allowed for exploratory investigation of the research model for the digital educational divide among vulnerable students, as well as the relationship between the constructs of access and connectivity and use and exploitation, and relationships among sub-constructs of synchronous learning, asynchronous learning, productivity skills, and creativity skills.

As the study was exploratory in nature, PLS-SEM was chosen as the data analysis procedure. This procedure allows predictions and explanations of target constructs to be conducted rather than confirmatory analysis with the capability of small sample sizes and complex models. PLS-SEM analysis does not make any assumptions about underlying data; thus, 518 participants are sufficient to carry out the analysis [27,28]. Thus, this study focuses only on describing the structural model analysis results (via PLS-SEM model diagrams) in terms of loadings of each construct and does not report on measurement model analysis results. The results of loadings would assist in understanding the most and least important constructs and sub-constructs of the educational digital divide among vulnerable students. The software used to run PLS-SEM analysis is SmartPLS version 3.3.9 by SmartPLS GmbH, Hamburg, Germany. The research model used in the analysis is illustrated in Figure 1.

4. Results and Discussion

4.1. Demographical Findings

More than half (56%, 290/518) of the participants were female, while the remaining participants were male (44%, 228/518). The age range included students aged 10 to 15 years old. All the participants were students from low-income families, with a maximum household income of RM4850 (EUR 1050). Of the participants, 37.1% (192/518) were from the north region of peninsular Malaysia, while 252 participants were from the east region of peninsular Malaysia. The remaining participants were from the west and south regions of peninsular Malaysia, as well as from the states of Sabah and Sarawak. All the participants did not have access to iPads with internet connectivity, and this was the first time that had they used iPads for learning purposes.

4.2. Results and Discussion of Partial-Least Squares Equation Modeling Results for the Digital Education Divide among Vulnerable Students in the Pandemic

4.2.1. Results and Discussion of the Overall Measurement Model and Structural Model Analysis

The findings of the measurement model analysis show that the indicators of the constructs achieved internal consistency, reliability, convergent validity, and divergent validity. The results are summarized in

Table 2. Internal consistency reliability results.

Construct, Sub-Construct, and Indicator	Average Variance Extracted (AVE)	Composite Reliability	R2	Cronbach’s Alpha
Access and connectivity	0.622	0.891	0.850	0.847
Asynchronous learning	0.631	0.837	0.711	0.708
Synchronous learning	0.738	0.850	0.646	0.646
Use and exploitation	0.658	0.939	0.927	0.925
Productivity skills	0.673	0.861	0.758	0.757
Creativity skills	0.760	0.940	0.922	0.920

and

Table 3. Convergent validity results.

Construct, Sub-Construct, and Indicator	Loading	Average Variance Extracted (AVE)	Composite Reliability
Asynchronous learning		0.631	0.837
Internet browsing	0.804
Text-based messaging	0.761
LMS learning	0.817
Synchronous learning		0.738	0.850
Video audio chatting	0.856
Video conferencing	0.863
Productivity skills		0.672	0.860
Digital note-taking	0.823
Mathematical calculation	0.817
Presentation slide creation	0.820
Creativity Skills		0.760	0.940
AR creation	0.900
Animation creation	0.912
Digital book creation	0.841
Digital music composition	0.803
Video production	0.896

. The findings of the structural model analysis (Figure 3) revealed that there are two constructs: access and connectivity, and use and exploitation. There are also two sub-constructs related to the former construct (access and connectivity), namely asynchronous learning (with the loading of 0.969) and synchronous learning (with loading of 0.949), while productivity skills (with the loading of 0.909) and creativity skills (with the loading of 0.974) are related to the latter construct (use and exploitation). This indicates that all the indicators (e.g., internet browsing) are related to their respective sub-constructs (e.g., asynchronous learning). These results corroborate the work of [8], which revealed that the rural digital divide due to the COVID-19 pandemic in Europe is related to access and connectivity, as well as use and exploitation. The results also corroborate Apple’s Everyone Can Create curriculum with regard to the sub-constructs of productivity skills and creativity skills related to the use and exploitation construct [26].

This study can be linked to [29], which investigated how online learning was supported in rural areas of China during the COVID-19 pandemic. The study involved 1409 students, and the researchers assessed the Community of Inquiry model with the perceptions of technology use. The findings indicated that facilitating conditions had an influence on students’ online learning quality via enhanced technology self-efficacy and perceived usefulness. In the study, the authors discovered an imbalance in access to digital equipment, resources, and competencies in rural and underdeveloped areas of China as compared to urban areas and stressed the importance of providing these facilitating conditions to ensure learning and digital equity are achieved. The study also emphasized that as access and connectivity are available for learning, students’ confidence levels are increased, thus contributing to students’ online learning success. Linking back to this study, the findings signify that the use and exploitation construct is related to access and connectivity—in other words, when students are provided with access and connectivity for learning, this can facilitate conditions for the use and exploitation of technology for learning. On the contrary, when students experience technological problems with regard to access and connectivity, this would cause them to lose confidence, thus disrupting their learning process.

The study’s findings also indicate that the loadings obtained for the access and connectivity sub-constructs were 0.969 for asynchronous learning and 0.901 for synchronous learning, which signifies that the strongest sub-construct is asynchronous learning followed by synchronous learning. This is an especially interesting finding in relation to pandemic learning for the vulnerable, which contradicts the common perception that synchronous learning is perceived to be more effective than asynchronous learning. A related point that Ref. [28] reported is that many teachers attempt to sustain students’ positive affect and maintain social relationships in online instruction during the pandemic, as they were concerned with overcoming perceived effects of pandemic restrictions that included isolation and anxiety. Although these results may be true for the vulnerable (in terms of synchronous and asynchronous learning), they must be dealt with cautiously, as these students may be introverted and hence have a higher tendency to prefer asynchronous learning rather than synchronous ones [20].

As for the use and exploitation construct, the creativity skills construct obtained a higher loading (0.948) as compared to productivity skills (0.909). This signifies that creativity skills are stronger than productivity skills in terms of the use and exploitation of technology in pandemic learning for vulnerable students. This can be explained by the fact that students in the pandemic reverted to digital play for learning, as they moved to onsite learning in schools to online learning in LMS and online environments. In a study by [30], they discovered that children in the pandemic reverted to activities that involved creativity skills such as video production and animation creation, as they grappled with issues of the pandemic lockdown. This also resonated with the findings of [31,32], in which they found out that students in the pandemic reverted to activities involving creativity skills, which included learning with AR and learning via online music performances and auditions.

4.2.2. Results and Discussion for Sub-Constructs

The results showed that the highest loading received for the asynchronous learning sub-construct was LMS learning (0.817), followed by internet browsing (0.804), and text-based messaging (0.761), as shown in Table 3. The results signify that the strongest asynchronous learning technique is via LMS learning, followed by internet browsing. These results can be linked with the study by [33]. Ref. [33] studied the US’s response to emergency online learning in the state of Mississippi during the pandemic. The study reported that LMS learning was deployed at the start of the 2021 school year, during which school districts integrated additional instructional models that included both synchronous and synchronous learning that was coupled with pre-printed learning packets. Almost all the districts offered one mode, namely an online learning mode, with 80% of the districts applying blended learning, in which online learning was blended with onsite instruction. However, as the school year continued, it was reported that the use of asynchronous learning via LMS and the use of pre-printed packets dropped dramatically, as schools generally viewed the approaches as less effective. However, the use of blended learning steadily decreased as the year went on. It worth noting that, contrary to the schools’ perception of LMS use for learning, superintendents rated the fully online instructional models to be more successful than the blended learning mode. With regard to text-based messaging for asynchronous learning, Ref. [33] discovered that teachers experimented with many different communication methods for emergency learning. Commonly used techniques were social media, automated voice calls, automated text messaging services, mass emails, postal mail, video calls, newspaper and radio announcements, announcements through the LMS, printed newsletters, individual visits from social workers, and home visits. The most effective technique reported for primary and middle grades were text messages, emails, and voice calls—in that order—while for high schools, the most effective ones were via LMS and automated text messages. This signifies that although the highest loading received for the asynchronous learning sub-construct was LMS learning, followed by internet browsing and text-based messaging, this should be investigated in detail for asynchronous learning for vulnerable students, as different types of school levels have been shown to have a prevalent or preferred medium of instruction.

As for synchronous learning, the highest loading obtained was video conferencing (0.863), while the lowest loading gained was for video and audio chatting (0.856). In Table 2, the findings also show that the strongest learning technique for synchronous learning was video conferencing, while the weakest technique was via video and audio chatting. These results can be linked to a study by [34], which investigated virtual education activities among children, focusing on remote book reading over video chat, and compared them to reading over pre-recorded video (i.e., asynchronous book reading) and traditional book reading. They found out that children were more responsive to live and video chat conditions as compared to pre-recorded videos, since there were high levels of interactions between children and book readers. They further suggested that children are more responsive in video chat learning, as they receive feedback that is contingent on their actions and interactions with book readers. Linking back to this study, although video conferencing received a higher loading as compared to video and audio chatting as both methods allow for live interactions and feedback between learners and teachers, this indicates that both methods acted as viable alternative approaches for online synchronous learning to consider the digital divide in learning during the pandemic. Ref. [35] carried out an investigation on the role of technology in children due to the COVID-19 pandemic quarantine, in which they carried out a cross-sectional survey on 1860 children in Italy. Their study discovered that during the pandemic, social media platforms became crucial for maintaining and enhancing socialization among adolescents, as well as maintaining friendships and emotional connections. The survey also revealed that due to a lack of face-to-face contact during the pandemic, children resorted to video chats and text messages for virtual communication and socialization.

Meanwhile, in other related studies, videoconferencing for learning was reported to cause negative effects on learning instead of positive ones. Ref. [36] reported that students often felt a high degree of isolation, anxiety, and depression as compared to onsite physical classes. They also revealed that students appeared to be nonresponsive when attending classes via videoconferencing, and this further had a negative impact of nonverbal dynamics of student–instructor interaction. Furthermore, communication issues were also raised regarding the lack of facial expression, body appearance, and movement among peers and teachers, as well as internet connectivity for videoconferencing. In addition, Ref. [37] reported that videoconferencing technologies for synchronous learning were found to cause learners to be mentally exhausted. This was more evident in developed countries, where poor internet connectivity and constantly seeing themselves during videoconferencing resulted in high levels of stress among students. Thus, although videoconferencing and video and audio chatting were found to be appropriate methods of online synchronous learning, the mental and psychological issues it caused should be looked at carefully for vulnerable students, so as to ensure that methods enhance learning instead of disrupting it.

The results show that the indicators that received the highest loading for the productivity skills sub-construct were digital note-taking (0.828) and mathematical calculation (0.828). Meanwhile, the weakest loading gained was for the presentation slide creation indicator, which received a loading of 0.805. These results can be linked to a study by [38] on the use of touchscreens and stylus pen tablets for remote learning. The study revealed that the use of such devices increased attention spans during mathematics learning. The study also emphasized that touchscreens allowed for a stronger association with students’ hand gestures and on-screen results as opposed to using a mouse or a physical keyboard. Furthermore, the study stressed that pen-based tablets enhanced mathematical learning, as they allow students to write and draw mathematical equations and diagrams. To connect this with our study, we can note that the use and exploitation of technology with regard to digital note-taking and mathematical calculations could be useful for vulnerable students in emergency remote learning.

The results also indicated that the highest loading received for the creativity skills sub-construct was animation creation (0.910), followed by augmented reality (AR) creation (0.897) and video production (0.897), as shown in Table 3. The lowest loading obtained was digital music composition. These findings can be related to the studies of [31,32,39,40,41,42,43,44]. In the first study, Ref. [45] studied distance learning among school students in distance learning environments during the COVID-19 pandemic in the United Arab Emirates (UAE). The study found that distance learning environments were perceived by students as having a positive impact on creativity and collaborative skills while learning about chemistry, despite the issue of carrying out laboratory-related demonstrations and hands-on learning activities due to pandemic restrictions. The study also reported that the online mode of learning chemistry presents benefits with regard to flexibility and autonomy in learning, as well as an increase in the sense of community-belonging and an increase in levels of safety as students conduct virtual experiments rather than physical ones, which tend to have a certain degree of risk attached to them. In another related study on AR, Ref. [31] studied the effects of augmented reality technology in marine education in Taiwan during the pandemic. The study discovered that AR picture books increased the level of learners’ engagement while learning about oysters without having them make a physical and onsite field trip to an oyster farm. This was due to the fact that AR offered the opportunity to perform close-up examinations, which was less likely to be possible in onsite visits.

Meanwhile, Ref. [30] studied digital play for children in a 15-month project during the pandemic in the United Kingdom. They discovered that digital play contributed to an enhancement of children’s experiences during the pandemic, in which children engaged in playful activities such as animation creation and video production. The study shared multiple global case studies in Australia, the UK, and Chile with regard to animation creation and video production during the pandemic. In Australia, it was reported that children had increased motivation levels while producing movies by combining a series of photos edited together into a short film, as well as 3D model creations for props in 3D movies. In the UK, teenagers created stop-motion movies using iMovie (a video-editing app on iPad OS) to share “positive energy” during the pandemic lockdown, whereas children also created stop-motion videos using iPads using loose parts and LEGO minifigures in reference to cartoon series and comics. In Chile, video production was carried out by using wooden animals as protagonists, in which stories were created related to their feelings and emotions of being trapped in the middle of lockdowns.

With regard to music composition, Ref. [46] investigated the use of a music apps for learning Shubailan, a form of music folk-talk-singing with 122 secondary school students in Hong Kong. The study applied a flipped learning strategy in which students were given preassigned recorded videos and collaborative activities during online synchronous lessons. They found out that by providing prerecorded videos before synchronous online classes, teachers had the opportunity to engage students in higher-order thinking activities, including music performance, music composition, and problem-solving, during the classes. This allowed teachers to design activities that were more task-based and involved a high level of student interactivity. These four studies highlight creativity skill activities that were conducted during the pandemic with regard to animation creation, AR creation, video production, and music composition and show students involved in media creation involving audiovisual and musical forms. To connect this with our study, we can note that, although animation creation received the highest loading as compared to the other remaining items, learning activities involving AR creation, video production, and music composition can act as alterative activities for enhancing creative skills in pandemic learning.

4.2.3. Results and Discussion for Hypothesis Testing and Bootstrapping

To investigate the hypotheses developed in this study, a bootstrap with 5000 resamplings was operated by [47,48] to generate the beta values, standard errors, t-values, and decisions.

Table 4. Structural model results.

Hypothesis	Relationship	Std Beta	Std Error	t-Value	p-Value	Decision
H1	Access and connectivity → Asynchronous learning	0.969	0.003	311.378	0.000	Support
H2	Access and connectivity → Synchronous learning	0.949	0.004	208.490	0.000	Support
H3	Use and exploitation → Access and connectivity	0.696	0.025	26.977	0.000	Support
H4	Use and exploitation → Creativity skills	0.974	0.003	369.344	0.000	Support
H5	Use and exploitation → Productivity skills	0.909	0.009	101.297	0.000	Support

shows that access and connectivity (β = 0.969, p < 0.00) is positively related to asynchronous learning, access and connectivity is positively related to synchronous learning (β = 0.949, p < 0.00), use and exploitation is positively related to access and connectivity (β = 0.696, p < 0.00), use and exploitation is positively related to creativity skills (β = 0.974, p < 0.00), and use and exploitation is positively related to productivity skills (β = 0.909, p < 0.00). Therefore, all hypotheses (H1, H2, H3, H4, and H5) were supported.

5. Conclusions, Implications, and Future Directions

The paper discovered two constructs related to the educational digital divide for vulnerable students in the pandemic, namely access and connectivity and use and exploitation, as well as four related sub-constructs, namely synchronous learning, asynchronous learning, productivity skills, and creativity skills. The study also revealed that asynchronous learning is a stronger construct for access and connectivity as compared to synchronous learning, while creativity skills was stronger than productivity skills with regard to the use and exploitation factor. In addition, the study discovered 13 indicators that were related to their respective sub-constructs, which included LMS learning, internet browsing, and text-based messaging for asynchronous learning; video conferencing and video audio chatting for synchronous learning; digital note-taking, mathematical calculations, and presentation slide creation for productivity skills; and augmented reality creation, animation creation, digital book creation, digital music composition, and video production for creativity skills. This study’s findings with regard to the digital education divide for vulnerable students in the pandemic could assist future developers and educators in the development of effective emergency teaching and learning strategies and design. This study could also help educational managers foresee and implement technological integration, in terms of enabling and empowering vulnerable students for synchronous and asynchronous learning, including software that allows for video audio chatting such as FaceTime, as well as production and creativity skills including software that allows for augmented reality creation such as Reality Composer. However, these technologies should be carefully looked at, as they may disrupt learning if not designed properly with appropriate pedagogical approaches.

Future directions in this area are as follows. First, the constructs and sub-constructs were only tested in specific vulnerable conditions with lower education in one geographical region. It would be interesting to investigate whether the findings are corroborated or contradicted in other educational settings, in terms of the level and type of student vulnerability, education levels, and different geographical locations. Second, the 13 indicators of the model (e.g., augmented reality creation) were treated in the study only as indicators. It would be beneficial to further investigate the indicators as sub-constructs to further re-confirm their relation to their respective sub-constructs and constructs. Third, the study used PLS-SEM for statistical analysis. Applying a different type of analysis (e.g., covariance-based structural equation modeling) could have yielded different results. Fourth, as the study did not focus on a particular instructional design strategy, it would be interesting to investigate how different instructional designs influence results. Fifth, the focus of the study was to investigate the relationships of the constructs and sub-constructs of the research model with regard to measurement and structural model analysis using PLS-analysis. It would be interesting to measure students’ performance with regard to constructs and sub-constructs in the study in order to understand how vulnerable students perform under emergency remote learning conditions. Finally, it would also be interesting to see whether access and connectivity to technology during the pandemic had a negative effect on vulnerable students.