Family income does not necessarily lead to a holistic e-learning experience during COVID-19: a study in Sri Lanka - Nature

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Humanities and Social Sciences Communications volume 11 , Article number: 1625 ( 2024 ) Cite this article

The COVID-19 outbreak caused a disruption of the conventional education system worldwide. This resulted in a shift towards online mode of delivery. This study analyzed the pandemic’s effect on the education of students from diverse socioeconomic groups in Sri Lanka. The required data for this study was collected through a nation-wide field survey, covering 3020 households. The factor analysis (FA) performed identified seven significant factors. Thereafter, K-means and spectral clustering were used to cluster the population in order to identify the socioeconomic groups that were impacted in a similar manner. This resulted in the identification of three clusters whose demographics and relationship to the extracted factors were extensively analyzed and interpreted. The study concluded that students in high-income households were more emotionally affected, whereas students in low-income households were mostly affected by a lack of resources. The study also concluded that students from mid-income households were the least affected emotionally.

The COVID-19 pandemic caused by the novel coronavirus SARS-CoV-2 has had an unprecedented global impact since its emergence in December 2019 (Novel Coronavirus – China n.d. ). The global tally, as of 16th of August 2023, shows that there have been over 769 million reported cases and an excess of 6.9 million recorded deaths worldwide (WHO n.d. ). The pandemic rapidly spread causing widespread disruptions in various sectors, including healthcare, economy (Benneker et al. 2023 ; Betthäuser et al. 2023 ; Hecht et al. 2023 ; Jay et al. 2020 ; Josephson et al. 2021 ; UNSDG n.d. ), and has profoundly impacted the education systems of countries worldwide (Almossa and Alzahrani 2022 ; Mpungose 2020 ; Hoofman and Secord 2021 ; Kuhfeld et al. 2020 ; Lichand et al. 2022 ; Parolin and Lee 2021 ; Salleh et al. 2023 ; Tanemura and Chiba 2022 ; Van Lancker and Parolin 2020 ; Zamir and Wang 2023 ). The widespread closure of schools affected approximately 95% of the global student population (UNSDG n.d. ).

The situation in South Asia was similar to other countries (Matta 2020 ; Campillo-Ferrer and Miralles-Martínez 2021 ; Chandasiri 2020 ; Haththotuwa and Rupasinghe 2021 ; Mizani et al. 2022 ; Rasul et al. 2021 ; Sandamali 2016 ) and upon identifying the first case of COVID-19 in Sri Lanka in March 2020, the government promptly resorted to drastic measures such as closing schools, travel restrictions, and strict social distancing protocols, intended at curbing the spread of the virus. Similar to any other country, this approach was essentially an evolving process until the end of the process and Sri Lanka faced three different waves of the pandemic. The first wave was from 27th March 2020 to 3rd October 2020, the second wave was from 4th October 2020 to 14th April 2021, and the third wave was from 15th April 2021 to 31st December 2022 (Epidemiology Unit n.d. ). With a diverse population and a spectrum of populations with varying family incomes, Sri Lanka is a good candidate to study the impact of COVID-19 on important aspects, such as education, as modulated by socioeconomic factors. In low-income countries, which includes Sri Lanka, the impact of the pandemic on education has been particularly severe (Betthäuser et al. 2023 ; Chandasiri 2020 ; Da Costa Cunha et al. 2020 ; Dutta and Smita 2020 ; Haththotuwa and Rupasinghe 2021 ; Ilankoon et al. 2020 ; Rameez et al. 2020 ; Rasul et al. 2021 ).

The closure of schools and universities for extended periods disrupted the learning process, leaving students with limited or no access to in-person education worldwide (Garbe et al. 2020 ; Fairlie and Loyalka 2020 ; Hecht et al. 2023 ; Lan et al. 2022 ; Liyanagunawardena and Williams 2021 ; Suheera and Wazeema 2022 ; Takács et al. 2023 ; von Stumm et al. 2022 ) and remote learning became the alternative, relying on online platforms, digital resources, and remote teaching methods (Campillo-Ferrer and Miralles-Martínez 2021 ; Coskun and Kara 2022 ; Angrist 2022 ; Hayashi et al. 2020 ; Ilangarathna et al. 2022 ; Ilankoon et al. 2020 ; Sucena et al. 2022 ; Tarkar 2020 ). This study provides the opportunity to test and analyze the hypothesis/argument presented by many social scientists (Goudeau et al. 2021 ; Masonbrink and Hurley 2020 ; Schulz and Robinson 2022 ) that the digital divide deepened existing educational inequalities, as students from disadvantaged backgrounds faced greater difficulties in accessing and participating in remote learning. Especially as all these factors are true/applicable in the context of Sri Lanka.

Schools and universities often serve as essential support systems, providing meals, healthcare, and psychosocial support to students (Chandasiri 2020 ; Da Costa Cunha et al. 2020 ; Fairlie and Loyalka 2020 ; Hecht et al. 2023 ; Jackman et al. 2021 ; Li et al. 2021 ; Törő and Csikós 2022 ; von Stumm et al. 2022 ). The prolonged closure of these institutions disrupted these crucial services, leaving vulnerable students without vital resources and support networks (Afiqah et al. 2020 ; Betthäuser et al. 2023 ; Demetriou et al. 2021 ; Mahmudah et al. 2021 ; Tadesse et al. 2020 ; Mohammed et al. 2020 ). Furthermore, schools and other educational institutes serve as places where children and adolescents learn social interactions and behaviors required to become successful in life. Thus, the closure of schools and educational institutes may have many more broad ramifications in addition to the factors mentioned above (Demetriou et al. 2021 ; Gayatri et al. 2020 ; Tadesse et al. 2020 ; Mohammed et al. 2020 ). Limited resources and inadequate funding for educational infrastructure have also hindered the implementation of effective remote learning strategies.

Due to the distinct culture of South Asian countries (Betthäuser et al. 2023 ; Haththotuwa and Rupasinghe 2021 ; Ilankoon et al. 2020 ; Rameez et al. 2020 ; Rasul et al. 2021 ) like Sri Lanka, the impacts of COVID-19 on socioeconomic and behavioral changes among the Sri Lankan population may vary compared to other regions worldwide. The unique cultural context of Sri Lanka can result in different responses and outcomes in the face of the pandemic, setting it apart from other global regions. Subsequently, socioeconomic conditions in Sri Lanka during and after the pandemic presented a certain level of uniqueness. As an example, during the first wave of the pandemic, there were hardly any covid deaths in Sri Lanka when the covid death toll was quite significant in other countries. Furthermore, the average income classes of the population were also consistent and this observation which is also shown by Fig. 1 is explained in a broader manner in the results section. Hence it is important to appreciate that the results would have a mixture of both these unique as well as generic factors when interpreting the data (Iwuoha and Jude-Iwuoha 2020 ) and this study will present an opportunity to draw conclusions about online education, the impact of the pandemic on education for other countries with similar socioeconomic conditions to Sri Lanka.

Average Household Income in Sri Lankan Rupees Variation During the Pre-pandemic (PP), First Wave (FW), Second Wave (SW), and Third Wave (TW) of the pandemic in Sri Lanka.

In Sri Lanka as well as most low-income countries, there is a lack of studies focused on analyzing the pandemic’s impact on both university and school students. This research study aims to address this gap by investigating the effects of COVID-19 on the education of students in all nine provinces of the country. This comprehensive study sets itself apart from other research endeavors as it encompasses both university and school students, providing unique insights into the pandemic’s educational ramifications in Sri Lanka. The study was a face-to-face field survey which provided the essential level of interaction needed for an extensive impact analysis (Ilangarathna et al. 2024 ; Howard 2009 ; Zhang et al. 2017 ). This was also beneficial compared to an online survey in reaching out to some of the more vulnerable groups of society that were most affected by lack of such resources, computer literacy, and device familiarity as an online survey might have not been able to extract the required information (Hlatshwako et al. 2021 ; Wardropper et al. 2021 ). On the other hand, certain groups might not have participated in an online survey due to privacy concerns because of the unfamiliarity with the entity that conducts the survey. Furthermore, there could have been the possibility of the parents/ guardians filling out the survey questions for the children, if it was conducted as an online interview. On the other hand, the emotional impact of the children from government or self-imposed social distancing (Akpınar 2021 ) might not have been captured through the prepared set of questions if they were presented in an online survey as some of the results indicate in the latter sections. Therefore, through a face-to-face survey, it was possible to interact with them in a close fashion and involve such groups to the level expected.

Our main objective is to evaluate the direct impact of the COVID-19 pandemic on the education of both school and university students in Sri Lanka. While our statistical analysis does not establish causal identification, we employ survey questions to directly capture the experiences of respondents regarding how the pandemic and related restrictions have affected them. The findings from this research can provide valuable information to governments and international aid organizations, enabling them to make well-informed decisions on effectively mitigating the long-lasting consequences of the COVID-19 pandemic in low and lower-middle-income countries like Sri Lanka and this research will provide valuable insights for future researchers in this field.

The dataset was created from the data extracted by the main survey conducted in the period starting from 6th November 2021 to 10th December 2021 (Ilangarathna et al. 2024 ). The main purpose of implementing the survey was to capture changes made in the Socioeconomic and Behavioral categories due to COVID-19.

The validity and reliability of a dataset are the most significant factors that ensure the consistency and stability of the dataset and that the intended measures are carried out as planned. Since these measures govern the interpretations/outcomes precisely, conducting a validity and reliability analysis before using the dataset for factor analysis was considered vital. Kaiser–Meyer–Olkin Measure (KMO) is used as a content validity measure (Hou 2023 ; Watson and Thompson 2006 ; Wetzel 2012 ). A measure of the factorability of a matrix is given by KMO by indicating the correlation between the questionnaire and the variables. Cronbach alpha and McDonald’s omega were used as the statistical measures used to ensure reliability. Table 1 shows the results of the validity and reliability analysis for the educational questionnaire.

Factor extraction methods are imperative to identify the latent factors in a multivariate dataset. There is a range of commonly used factor extraction methods: Principal Component Analysis (PCA), common factor analysis, principal factor analysis, and maximum likelihood (ML) factor analysis. In this study, SPSS statistical analysis software was utilized to extract the factors from both datasets. Under the module of factor analysis in SPSS, PCA was used to identify the underlying factors.

Through FA, a set of variables in a multivariate dataset was converted to a low-dimensional set of variables in such a way that they summarize the impact of the original variables. Therefore, deciding precisely the number of factors present in the multivariate dataset having a well-defined theory to guide is a critical step. One popular theory to determine the number of factors in FA is the Kaiser Criterion, which is based on Eigenvalues. The magnitude of an Eigenvalue in PCA is directly proportional to the data’s explainable variance. Any variables with associated eigenvalues greater than 1 are considered factors under this criterion (Acar Güvendir and Özer Özkan 2022 ; Hou 2023 ; Muilenburg and Berge 2005 ; Watson and Thompson 2006 ; Zaman et al. 2020 ). Next, the Varimax rotation was applied to the factor-loading matrix obtained through PCA by applying the cutoff factor loading as 0.4 (Muilenburg and Berge 2005 ). Reducing the intricacy of the factor loadings to make an easy arrangement to interpret is the basic intuition behind the factor-loading rotation. The cutoff value has been selected to minimize the presence of cross-loading items in the factor-loading matrix. Cross-loading items are referred to questions or variables exhibiting factor loadings on more than one factor, with loading differences less than 0.1 (Acar Güvendir and Özer Özkan 2022 ). The aforementioned phenomenon makes it difficult to achieve a simple factor load arrangement and creates confusion for researchers when interpreting their results. Therefore, the variables with cross-loading are eliminated from the analysis, and the entire procedure starting by applying PCA until getting the rotation matrix, was repeated without the cross-loaded variables. This entire process is further explained in the Results section.

In natural disasters such as epidemics, pandemics, tsunamis, and earthquakes people in society are often affected disproportionately based on demographics. Hence examination of how the underlying factors affected different demographic groups is important to understanding the overall impact of such a phenomenon.

Therefore, for the education impact analysis conducted in this study to be complete, the population had to be grouped based on how the factors influencing education (identified through the previously mentioned factor analysis) affected the sample population. This clustering enables us to identify the different socioeconomic groups that were similarly affected by factors influencing education during the COVID-19 pandemic. This would enable us to derive important conclusions through the linkage of those groups, their demographics, and socioeconomic status with how they affected the child’s education. Therefore, the examination of factors derived from various socioeconomic classes of participants can be considered of great importance to estimate the overall impact of COVID-19 on a country as well.

After the identification of the factors from the multivariate dataset, the dataset can be viewed as a set of answers given to sub-questions that fall under a primary question, where each factor can be considered as a primary question. Therefore, the original higher dimensional data structure is not indicative of the identified key factors. As the process has already identified that most of the information resides on these specific dimensions pertaining to the identified factors, it is only logical to remove the influence of the noisy insignificant dimensions and the survey data be refined by projecting the entire dataset to the subspace identified through PCA that is entirely formed by the identified key factors as the basis. This is in essence a noise removal of the survey data.

The dataset created from the responses of the survey participants is not labeled to reflect their socioeconomic background. Therefore, it is essential to find a reliable mechanism to find the groups of people that have answered similarly. Utilizing machine learning techniques for clustering is an effective gateway for identifying similar groups because clustering algorithms can group similar data points based on shared characteristics. The two primary types supervised and unsupervised clustering are used in general, and labels associated with data points are utilized by supervised clustering, whereas unsupervised clustering is employed when labels are not available. In this study, spectral clustering, an unsupervised clustering method, was used to identify social groups that were affected similarly by the pandemic from the refined dataset. By employing this approach, it was aimed to group survey participants based on similarities in their responses and discern patterns among these groups.

Initially, the refined dataset derived by projecting the original dataset in the principal directions can be visualized as a big single cluster lying in the high-dimensional feature space, which is built by combining several sub-clusters. However, when zooming in on the feature/cluster space, sub-clusters start to form within the feature space. The mechanism of zooming in to detect the number of small clusters submerged inside the initial supercluster is called finding the “Modes” of clustering (Qin et al. 2016 ; Rupasinghe et al. 2016 ; van Dam et al. 2021 ). Furthermore, the number of clusters in a mode serves as the distinctive identifier for that mode. Hence different modes are discovered by varying the free parameter “σ” which embodies the zooming effect in the Standard Spectral Clustering algorithm, this process is known as “Sigma Sweep”. Sigma Sweep’s main steps are as follows:

Affinity matrix A, for the refined dataset was generated using the transformation.

Degree matrix D is a diagonal matrix where ( i , i )th element of D is the row sum of As’ i th row.

Calculate the Laplacian matrix L, from the relationship,

Obtain eigenvalues of L and arrange them into descending order. Then eigengaps were computed by getting the difference between two successive eigenvalues.

Plot the graph of the variation of the eigengap with log σ by repeating steps 1–4 for different σ values to determine the number of sub-clusters K.

To determine the optimum number of clusters K, two methods based on the properties of the sigma-sweep curves are used. One is to determine the value of K by selecting the mode number that corresponds to the highest eigengap. This is the curve that yields the highest valued peak among all modes during the sigma-sweep and the resulting curve is shown by Fig. 2 .

Variation of Eigengap of Six Modes with logarithmic sigma for Education data.

After identifying the number of socioeconomic categories/classes/clusters in the refined dataset, the K-means algorithm was used to group respondents into clusters based on similar experiences/responses (Ikotun et al. 2023 ). Income level of the population was identified as the primary factor distinguishing these clusters, considering the average income variation from pre-pandemic to the third wave of the pandemic (see Fig. 1 ), names were given to these identified social groups. This procedure is further explained in the Results section. For instance, the group of people who had the highest income-level throughout the period was named “High Income” and vice versa. This procedure was followed to make it simple to identify them for further analysis and results interpretations.

Next, the factor loadings belonging to different social groups separated via K-means were factor-wise averaged to find the average impact of different factors on different identified social groups. Then the average factor loadings variations of each factor through demographic variations were analyzed for interpretation. The Python scikit-learn module was utilized for the implementation of K-means algorithms and obtaining the region separation.

A door-to-door survey (Ilangarathna et al. 2024 ) was conducted to collect data by covering 3020 households as data points that are scattered over 20 districts out of the total of 25 districts, considering the rate of infection data provided by the Department of Census and Statistics, Sri Lanka. Households were selected utilizing the multistage clustering technique which was the hierarchical structure of the power distribution in administrative divisions in Sri Lanka (Ilangarathna et al. 2024 ). The regions for the selection were based on extensive analysis on factors such as the population, economic status, high-risk areas depending on the dependency ratio of the given region. To make sure the data has a good representation, areas consisting of urban areas, rural areas, agricultural zones, industrial zones, fisheries, estate sectors were included (Ilangarathna et al. 2024 ). The households which were most affected by the pandemic in more than one way such as the entire household being infected or facing severe financial difficulties due to the pandemic was chosen with the help of the respective Village Officer (Ilangarathna et al. 2024 ). This selection process is further elaborated in Fig. 3 and the more details are available on the dataset & the publication available online (Ilangarathna et al. 2024 ). The survey was around 40 min of face-to-face interviews with a questionnaire covering the socioeconomic and behavioral impact on a household influenced by the pandemic. Initially, the questions related to the basic demographic information of the family were answered by the main respondent of the survey. The questions related to a particular impact area were answered by the most relevant member of each household. The gender-based distribution and ethnicity-based distribution of the surveyed population are compared with the Sri Lankan population data (`Mid-year population estimates by district & sex' n.d. ) in Tables 2 and 3 . The spread of our data pool compared to the Sri Lankan population in each district is shown by Fig. 4 . Some districts, which also happen to be the most densely populated districts in Sri Lanka, have accounted for a greater percentage of the survey respondents compared to its percentage in the Sri Lankan population and this was primarily due to more groups being present in those districts that were severely affected from the pandemic compared to the rest of the country. Even though the surveyed population is not a perfect representation of the national population the survey was designed to make the data distribution as representative as possible. The responses to the questions were recorded using two types of Likert scales designed to assess the perceived impact. These Likert scales vary from 1 to 5, where a rating of 1 represents “strongly agree” and a rating of 5 represents “strongly disagree” for positively worded questions. For negatively worded questions, a flipped scale was used (see the Data Availability section).

A summary of the household selection process for the survey.

Percentage distribution comparison of the surveyed vs. Sri Lankan population in each district.

In this study, there are three predominant outcomes. First, identifying the factors that influenced the overall education experience of the students during the pandemic. Second, clustering the surveyed population in an unsupervised manner and identifying the resulting clusters of groups that were similarly affected by the underlying factors affecting education. Furthermore, upon analyzing the above clusters/groups using demographics of the population such as income & education, it was concluded that the unsupervised clustering has resulted in a division through income level, and it is explained further in the subsequent results section. Finally, by combining the above two outcomes, this work studies the variations of the identified factors affecting education, over the income levels of the identified classes.

Initially, responses to education were selectively extracted from the original dataset, yielding 1643 out of the 3020 households (54.4%). There were 22 questions in the education impact analysis component of the survey and the answers to these 22 questions were considered in the factor analysis. These 22 questions were grouped into the previously mentioned seven factors as the result from the factor analysis and those 22 questions and their respective factors are shown in Table 4 . There were no questions where none of the loadings were more than 0.4. The scree plot for all the factors can be seen in Fig. 5 and the 8th factor does not fulfill the criteria but it is not dramatically lower than 1. Since we followed the criteria of considering factors with magnitude of eigenvalues greater than one, clearly distinguishable seven factors were identified and for those seven, the question selection criteria was imposed with the 0.4 threshold. The rotated component matrix of the factor analysis shown in Table 5 . This was obtained from the first iteration of factor analysis itself, therefore, there were no other factors or questions that we needed to eliminate from the analysis.

The data selection process for the education impact analysis from the survey dataset can be summarized in two steps. Members that fall under some of the studied impact areas were not present in every household since the survey was conducted to analyze the impact of numerous areas such as mobility patterns, psychological impact, health & nutrition and not just education. Therefore, not every household had a student to answer the questions covering the education impact but still were massively affected in other impact areas which were studied by the survey. Therefore, the first step involved removing the household entries that had no members involved in ongoing studies. Furthermore, the household data entries which had errors in collection or entering of the questionnaire responses, were removed since they were labeled as missing data (Ilangarathna et al. 2024 ). Therefore, the second step involved dropping such household data points from the education impact analysis which eventually resulted in 1643 households for the study. It was validated that all of these 1643 households had complete and accurate information about educational impact. The study employed the KMO, Cronbach alpha coefficient, and McDonald’s omega coefficient as mentioned (Hou 2023 ; Watson & Thompson 2006 ) in the Method section, to assess the validity and reliability of the dataset (Table 1 ).

PCA with varimax rotation (Watson & Thompson 2006 ) was employed using the SPSS version 27 to extract the underlying factors of the dataset. Further, the questions associated with each factor were chosen according to the findings of the literature review (Goudeau et al. 2021 ; Gunawardana 2020 ; staff et al. 2020 ) and based an initial hypothesis that education would have been impacted due to reasons such as lack of resources, network connection issues, loneliness & other psychological reasons, lack of motivation, lack of peer presence. To further elaborate, these questions covered whether the students had a well-functioning device for online education, how many people had to share the devices, state of the power supply to the residence, computer literacy, internet connection status and quality, availability of peer support, and support received from family members.

These questions were selected using a large spanning criteria to have as many questions as possible to cover the entirety of the education impact analysis. Only the questions with a factor loading above 0.4 were deemed to be representative of the underlying factor and the 0.4-factor-loading threshold was selected consistent with literature (Bavel et al. 2020 ; Hou 2023 ; Watson & Thompson 2006 ). Moreover, the study utilized the Kaiser criterion as the theoretical framework for factor extraction in the education dataset as mentioned in the Methods section. Specifically, factors with eigenvalues greater than 1 were selected, resulting in the identification of seven distinct factors within the education dataset. The obtained factors are shown in Table 5 . The cumulative explainable variances of the seven factors are 60.7% of the total observed variance for the education dataset and since the resulting rotated component matrix was obtained from the first iteration, the cumulative explainable variance (60.7%), was also obtained from the first round of factor analysis iteration itself.

Once the questionnaire has been factored in, the study of how these changes have influenced the different socioeconomic groups is vital. The spectral clustering described in the method section was used to identify the dominant mode which corresponds to the number of social groups. These identified social groups in the dataset formed clusters from the spectral clustering algorithm described in the method section. As detailed in spectral clustering, the dominant mode was found to be three, as the maximum eigengap was between three to four, which means that the dataset primarily can be grouped into three distinguishable classes (Rupasinghe et al. 2016 ).

After finding the number of underlying clusters, individuals belonging to these clusters were separated from the pool of people by employing the K-means algorithm (Ikotun et al. 2023 ) which was implemented using Python scikit-learn module. That resulted in three regions in the clustering diagram as shown in Fig. 6 . Upon analyzing the average income in these three regions, they corresponded to three income levels which could be appropriately designated as “low, mid and high”. For instance, in the pre-pandemic stage, one end had households with more than LKR 80,000 average income (i.e., the “high” income cluster); the other end had households with an average income lower than LKR 50000 (i.e., the “low” income cluster); and the middle segment had an income in between LKR 50,000–LKR 80,000 (i.e., the “mid” income cluster). However, as illustrated in Fig. 1 , average income level corresponding to these three clusters changed over the three pandemic stages.

Regions resulted from K-means clustering for the education impact data.

It should be noted that the average income data of each household was collected in the survey under the Economic Impact section (see the Data Availability section). The respondents have been asked to choose an income level according to five predefined categories instead of responding with a numerical value for their average income. Average income “Less than LKR 10,000”, average income in between “LKR 10,000–LKR 25,000”, average income in between “LKR 25,000–LKR 50,000”, average income in between “LKR 50,000–LKR 80,000” and average income “More than LKR 80000” were the 1st, 2nd, 3rd, 4th & 5th income groups respectively. Further as explained above, the analysis distilled the five classes into three contextual income categories (high, mid, low), which were then identified to be corresponding with the regions separated by the unsupervised clustering.

The main intuition of labeling the people based on their pre-pandemic income level can be considered more reasonable and unbiased as the people’s income may have deviated from a small amount to a large amount during the pandemic period owing to the economic downfall experienced by Sri Lanka and the heightened job insecurity resulting from multiple lockdowns. Based on that criterion, people who had a high average pre-pandemic income were considered a high-income class. Following that, others are labeled as mentioned above.

Subsequently, a comprehensive analysis was conducted to explore the effects of the identified factors on individuals’ lives and lifestyles, aiming to gain a deeper understanding of their impact and how they manifested across diverse groups. As mentioned previously, the identified factors from the initial part of the study (factor analysis) were clustered using unsupervised clustering, and that resulted in three groups. It was observed that the resulting three groups were equivalent to the high, middle, and low-income categorized subjects. Thereafter, outcomes of factor identification and unsupervised clustering were combined to analyze the variational impact of factors through the demographic groups for further impact assessment. This involved delving into the intricacies of their demographics, and meticulously examining the reasons behind these consequences. Subsequently, focusing on the above social groups, the authors examined the average impact of the identified factors on the identified different categories of income, revealing valuable insights. Figure 7 represents how the seven extracted factors influenced the identified data-driver clusters primarily distinguishable via income levels. In the second phase of this study, the authors made significant efforts to consider the demographics of the primary respondents, who represent households and play a crucial role in shaping their children’s education. The intention was to capture a comprehensive understanding of the household’s background and how it affected the factors impacting education. Therefore, among the diverse demographic details available in the dataset, particular emphasis was placed on investigating the impact of marital status, education level, and employment sector. The results of Fig. 7 are further explained later in this section and in the discussion.

Distribution of the impact of extracted factors on different socioeconomic groups.

The marital status of the parents is graphically illustrated in Fig. 8a , highlighting five marital statuses: Married, Divorced, Widowed, Separated, and Unmarried. A concise summary of the different marital statuses that are considered is shown in Fig. 8a and it can be observed that the number of divorced households is comparatively large in high-income families. When considering Widowed families, low-earning families have outnumbered all the other socioeconomic groups. Additionally, the mid-income groups have shown a higher percentage of possession in the remaining marital states than the other two groups.

( a ) Distribution of the Marital Status across the socioeconomic groups with number of households ( b ) Distribution of the Educational Level across the socioeconomic groups with number of households.

Next, Fig. 8b yields valuable insights into the education level of the parents of the surveyed households. According to this plot, it is distinguished that most low-income respondents have not completed their secondary or higher education. Compared to the other two income groups, the main respondent of poor households has never had the opportunity to attend any form of formal education and it is 6% as a percentage. The uneducated percentage in middle-income households was less than one-fifth that of low-income families surveyed. Moreover, it is consistently observed that a greater number of impoverished families attain education up to Ordinary Levels (O/L) compared to the other two socioeconomic categories (see Fig. 8b ). Additionally, there is a clear trend indicating higher rates of advanced-level education (A/L) among middle-income families. In contrast, individuals residing in high- and middle-income households exhibit the highest levels of educational achievements, including diplomas, undergraduate degrees, master’s degrees, and even Ph.Ds.

From Fig. 7 , it can be observed that factor impact variations can be primarily classified into distinct categories based on their relationship with financial strength. The way in which the quantified impact of each factor has fluctuated across the three data points “High, Mid, Low” income households, was observed to be of linear, nonlinear & independent behavior. In other words, as shown in Fig. 7 , the factors which showed a proportional growth or a reduction with income level was classified as linear, factors which showed a more complex relationship where an up down motion was present across the three income levels was classified as nonlinear, factors which hardly showed any perturbation or sensitivity on the factor weights through income were classified as independent. Therefore, we can conclude that “Network quality and connectivity”, “Resource affordability for online learning”, and “Household environment”, showed a linear relationship with the income levels and “Device accessibility and know-how” showed a slightly linear relationship with income level. Person inherited qualities like “Self-motivation” showed an independent variation from income and “Sociocultural and emotional”, and “Teacher availability and preparedness” have varied nonlinearly.

The identification of the key factors that have influenced students’ education during the COVID-19 pandemic was considered as the primary objective of this study and seven distinct factors that have had an impact on remote education during this unprecedented time were identified through factor analysis. Various aspects including students’ perception of online education compared to traditional classroom learning, the influence of limited resources, data plans available for remote learning, the role of household support and conducive learning environments, students’ personal interest and motivation in engaging with online education, the significance of a stable and strong internet connection, the challenges faced due to the unavailability or limited access to necessary equipment for online classes, and the extent of institutional support provided to ensure students’ continued engagement and success in their studies were encompassed in these factors. However, an equal impact of the above seven factors on the entire Sri Lankan community was not evident. As a result, this study was expanded to investigate how it impacts various socioeconomic groups. The demographical information of the Sri Lankan households is explained below and according to the findings of this study, a more expanded description about how each of the seven identified factors affected the Sri Lankan population is explained in the subsequent sections.

Considering the demographics of the three income groups, most high-income class families tend to reside close to metropolitan areas, where they have access to all the amenities while maintaining sparse relationships with others (CEPA n.d. ). A majority of the high-income class population is well-educated and has fewer issues of maintaining social distancing even after the government-imposed restrictions were lifted due to resource availability. In comparison to the high-income class, most of the middle-income class resides near urban areas or cities. These groups needed to mix with the other groups after the restrictions were removed due to the economic situation as many of them relied on social interactions to keep a steady income. Low-earning groups in Sri Lanka tend to choose dwellings in locations where the lack of infrastructure facilities aligns with their income. In contrast to the high & middle-income population, the low-income class generally resides near densely populated urban areas in flats or slums, and when they move far away from cities, they settle in sparsely populated rural villages. Low-income families are almost entirely dependent on interactions which made them mix with the general population as soon as the restrictions were lifted for survival. During the pandemic, it was observed that hygiene practices and adherence to health guidelines were very low among urban poor communities (Corburn et al. 2020 ), as their earnings were only sufficient to cover necessities. Consequently, these places were identified as COVID-19 spreading hotspots. Therefore, to curb the spread of the pandemic, the Sri Lankan government had to enforce selective area lockdowns.

Our research findings can be categorized according to how the seven identified factors affected each of the three income clusters. Considering the demographics of each income class, we can draw conclusions about how the results generated from the study came to be. The factors that have behaved nonlinearly, linearly, and independently from income have been discussed in this section below.

The factors which have shown a nonlinear behavior and how the context of the Sri Lankan society resulted in a such nonlinear pattern are discussed in the first part of this section. “Sociocultural and emotional” factor showed a nonlinear concave pattern in terms of the income levels of the families by exhibiting notably elevated levels among students hailing from high and low-income families. In the context of the Sri Lankan community, as mentioned above, most high-income families prefer to live close to big cities where they have access to all essential needs conveniently while having limited social interactions. Much like the rest of the world, the high-income class in this study which has a significant portion of the educated middle class also practiced social distancing well into the later stages of the pandemic even when the government lowered restrictions, simply as they could afford to. This is also suggested by the factors explained later. This self-imposed distancing had an impact on the child’s overall education experience. Consequently, children growing up in such social environments find their circle of interaction narrowed down to a select group of individuals, including teachers and peers, with whom they engage daily. As the answers to the survey suggest, with the introduction of online education facilitated by platforms like Zoom and WhatsApp, this circle of interaction was further diminished as it became confined to their own homes or rooms. Furthermore, it is evident that they missed their classmates and their contemporaries’ engagement in extracurricular activities from the answers to the questions related to the above-mentioned sociocultural and emotional factors. As online learning failed to fully replicate the traditional classroom experience, a significant number of students expressed dissatisfaction with e-learning. In contrast to the other income classes, lower-middle-income and lower-income families have been inclined to mix with the population as more government-imposed restrictions relaxed. A majority of these people’s income depends on the quality of their interactions with others and the frequency with which they go out to work. This has helped the children from such backgrounds indirectly in terms of social interactions and extracurricular activities. These differences in the approaches have resulted in the nonlinear behavior of the above factor against the income variations. The door-to-door survey has helped in identifying the true nature of the emotional factor that might not have been able to be captured from an online survey.

The “Teachers’ availability and preparedness” factor has also nonlinearly varied with household income, following a convex trend across income levels. However, as the deduced results from the answers to the survey questions suggest, the high-income and upper middle-income level households have only been moderately impacted by the teacher’s preparedness and availability, as they attended better private, international, or competitive government schools. Additionally, private tutoring is popular among students in this income class to survive the competition among peers in such schools. Therefore, since school was not the only option to fulfill their need for teachers’ support, this might have led them to feel that they were not greatly affected by this matter (see Table 6 ). Learners from middle-earning families experienced the fewest issues stemming from the lack of sociocultural and emotional factors and were affected by tutor availability and preparedness. Compared to those groups, other lower-income groups also had a separate set of priorities with the dire economic situation in addition to bringing solutions to the teacher preparedness issues as explained later under the other factors. As many students in these communities were not very enthusiastic about their studies and attended schools with low facilities, they felt more anxious about schoolwork. The parents/ guardian’s apathy for education due to their lack of educational qualifications (see Fig. 8b ), and the usual lack of inclination for studies from students had an average impact on the teachers’ preparedness and availability according to their sentiments.

The device and network-related factors had affected similarly in a linear fashion on education and these three factors; “Device accessibility and know-how,” “Network quality and connectivity,” and “Resource affordability for online learning” were examined collectively. The internet plays a pivotal role in e-learning. To access the Internet, individuals must meet three primary requirements. Firstly, they need a device, such as a mobile phone or a computer, along with the knowledge to operate it. Secondly, a reliable and high-quality network connection is essential. Lastly, individuals must be able to afford a sufficient data package. Overall, the low-earning community has faced considerably more challenges regarding network quality and resource affordability, with a slight increase in issues related to device availability compared to the other two groups. In general, these factors were mostly linearly correlated with income levels. According to the published reports, Sri Lanka can be identified as a country with a high presence of mobile phone penetration (Digital Outlook Sri Lanka 2021 n.d. ). However, in Sri Lanka, the type and demand for mobile phones and their ownership depend on several socioeconomic characteristics (Haththotuwa & Rupasinghe 2021 ; Hayashi et al. 2020 ; Ilankoon et al. 2020 ). Having to share the resources for learning and the limitations of data plans has had a proportional impact on households based on their income levels. Affluent families opt for high-speed and high-data bundle plans, while impoverished communities are constrained to using mobile prepaid low-data bundle plans due to their unaffordability. As a result, the severity of issues concerning peak/off-peak timings, speed, and connection largely depends on the data bundle plans that different income groups can afford. Our study reveals another significant finding that the impact of COVID-19 on educational issues within the household environment varies among students from different economic backgrounds.

One notable finding from our survey indicates that there are no significant differences in “Self-motivation” among students across various income levels. Consequently, we can conclude that the impact of COVID-19 on the self-motivation of students across all economic levels is moderate. This implies that the pandemic has influenced the motivation levels of students to a certain extent, regardless of their financial backgrounds. The underlying cause for this phenomenon can be attributed to the influence of internal factors on self-motivation, namely personal goals, aspirations, and intrinsic motivation. These internal drivers hold greater significance in shaping a student’s motivation levels compared to their economic background as can be concluded from the results. This draws a hugely important conclusion that the motivation for education is independent of the income levels of the students’ households and that is why education can be considered a universal human right. This provides proof to the argument that universal access to education must be allowed regardless of social class.

In conclusion, six of the seven factors were sensitive to income variation linearly or nonlinearly. Only the self-motivation factor was independent of the income variation and the middle-income group was disproportionately affected in all the factors except self-motivation which is completely independent as mentioned above. During the COVID-19 pandemic, the household environment of students from low-income families experienced the most significant impact. Conversely, the household environment of middle-income families was moderately affected, while students from high-income families encountered the least disruption in their household environments. The impact of COVID-19 on the household environment in low-income families can be attributed to multiple determinants. Firstly, children in low-income families in Sri Lankan culture often bear greater household responsibilities, including looking after their siblings, cooking, washing clothes, and assisting their parents with their jobs. In contrast, high-income families tend to have fewer household responsibilities for their children, while middle-income families fall somewhere in between. Additionally, the parents in low-income families in Sri Lanka often have lower levels of education compared to high and middle-income families (see Fig. 8b ). As a result, parents in low-income families may have limited knowledge and resources to assist their children with their studies, potentially impacting their academic progress. Furthermore, Low-income households often have limited living space, leading to overcrowding and a louder household environment. This can make it difficult for children to effectively participate in online classes and concentrate on their studies compared to students from more spacious and quieter environments in higher-income households.

The dataset used for the research is accessible in Harvard Dataverse under the title “Socioeconomic Impact of COVID-19 during three pandemic waves” ( https://doi.org/10.7910/DVN/CXMJSM ).

Acar Güvendir M, Özer Özkan Y (2022) Item removal strategies conducted in exploratory factor analysis: a comparative study. Int J Assess Tools Educ 9(1):165–180. https://doi.org/10.21449/ijate.827950

Afiqah N, Abdullah A, Rahmat H, Zawawi Z, Adib M, Khamsah N, Humaira A, Akademi A, Bahasa P, Anuarsham AH (2020) Coping with post covid-19: can work from home be a new norm? Eur J Soc Sci Stud. https://doi.org/10.46827/EJSSS.V5I6.933

Akpınar E (2021) The effect of online learning on tertiary level students’ mental health during the Covid19 lockdown. Eur J Soc Behav Sci 30(1):52–62. https://doi.org/10.15405/EJSBS.288

Almossa SY, Alzahrani SM(2022) Assessment practices in Saudi higher education during the COVID-19 pandemic. Humanit Soc Sci Commun 9(1):1–9. https://doi.org/10.1057/s41599-021-01025-z

Angrist N, Bergman P, Matsheng M (2022) Experimental evidence on learning using low-tech when school is out. Nat Hum Behav 6:941–950. https://doi.org/10.1038/s41562-022-01381-z

Bavel JJVan, Baicker K, Boggio PS, Capraro V, Cichocka A, Cikara M, Crockett MJ, Crum AJ, Douglas KM, Druckman JN, Drury J, Dube O, Ellemers N, Finkel EJ, Fowler JH, Gelfand M, Han S, Haslam SA, Jetten J, Willer R (2020) Using social and behavioural science to support COVID-19 pandemic response. Nat Hum Behav 4(5):460–471. https://doi.org/10.1038/s41562-020-0884-z

Benneker IMB, Lee NC, van Atteveldt N(2023) Mindset and perceived parental support of autonomy safeguard adolescents’ autonomous motivation during COVID-19 home-based learning. Npj Sci Learn 8(1):1–9. https://doi.org/10.1038/s41539-023-00153-2

Betthäuser BA, Bach-Mortensen AM, Engzell P (2023) A systematic review and meta-analysis of the evidence on learning during the COVID-19 pandemic. Nat Hum Behav 7(3):375–385. https://doi.org/10.1038/s41562-022-01506-4

Campillo-Ferrer JM, Miralles-Martínez P(2021) Effectiveness of the flipped classroom model on students’ self-reported motivation and learning during the COVID-19 pandemic. Humanit Soc Sci Commun 8(1):1–9. https://doi.org/10.1057/s41599-021-00860-4

CEPA , Provincial inequalities in income and its distribution in Sri Lanka (n.d.). https://www.cepa.lk/blog/provincial-inequalities-in-income-and-its-distribution-in-sri-lanka/ . Accessed 4 Sep 2023

Chandasiri O (2020) The COVID-19: impact on education. J Asian Afr Soc Sci Humanit 6(2):37–42. https://www.aarcentre.com/ojs3/index.php/jaash/article/view/207

Corburn J, Vlahov D, Mberu B, Riley L, Caiaffa WT, Rashid SF, Ko A, Patel S, Jukur S, Martínez-Herrera E, Jayasinghe S, Agarwal S, Nguendo-Yongsi B, Weru J, Ouma S, Edmundo K, Oni T, Ayad H (2020) Slum health: arresting COVID-19 and improving well-being in urban informal settlements. J Urban Health 97(3):348–357. https://doi.org/10.1007/s11524-020-00438-6

Article PubMed Central PubMed Google Scholar

Coskun K, Kara C (2022) Impact of school closure due to COVID-19 on phonemic awareness of first-grade primary school children. Humanit Soc Sci Commun 9:196. https://doi.org/10.1057/s41599-022-01218-0

Da Costa Cunha K, Mavrić B, Buheji M, da Costa Cunha K, Beka G, Leandro do Carmo de Souza Y, Souza da Costa Silva S, Hanafi M, Chetia Yein T (2020) The extent of COVID-19 pandemic socio-economic impact on global poverty. a global integrative multidisciplinary review. Researchgate.Net. https://doi.org/10.5923/j.economics.20201004.02

Demetriou L, Keramioti L, Hadjicharalambous D (2021) Examining the relationship between distance learning processes and university students’ anxiety in times of COVID19. Eur J Soc Sci Studi. https://doi.org/10.46827/EJSSS.V6I2.1012

Digital Outlook Sri Lanka 2021 Annual Market Insight Report (n.d.) https://apidm.lk/apidm/Digital%20Outlook%202021.pdf

Dutta S, Smita MK (2020) The impact of COVID-19 pandemic on tertiary education in Bangladesh: students’ perspectives. Open J Soc Sci 8(9):53–68. https://doi.org/10.4236/JSS.2020.89004

Epidemiology Unit (n.d.). https://www.epid.gov.lk/epidemiological-summary . Accessed 4 Sep 2023

Fairlie R, Loyalka P(2020) Schooling and Covid-19: lessons from recent research on EdTech. Npj Sci Learn 5:13. https://doi.org/10.1038/s41539-020-00072-6

Garbe A, Ogurlu U, Logan N, Cook P (2020) COVID-19 and remote learning: experiences of parents with children during the pandemic. Am J Qual Res 4(3):45–65. https://doi.org/10.29333/AJQR/8471

Gayatri M, Kependudukan B, Keluarga D, Nasional B(2020) The implementation of early childhood education in the time of COVID-19 pandemic: a systematic review. Humanit Soc Sci Commun 8(6):46–54. https://doi.org/10.18510/hssr.2020.866

Goudeau S, Sanrey C, Stanczak A, Manstead A, Darnon C (2021) Why lockdown and distance learning during the COVID-19 pandemic are likely to increase the social class achievement gap. Nat Hum Behav 5(10):1273–1281. https://doi.org/10.1038/s41562-021-01212-7

Gunawardana DP (2020) The impact of COVID 19 on the MSME sector in Sri Lanka. https://sustainabledevelopment.un.org/content/documents/26277Report_The_Impact_of_COVID19_to_MSME_sector_in_Sri_Lanka.pdf

Haththotuwa PMPS, Rupasinghe RAHM (2021) Adapting to online learning in higher education system during the Covid-19 pandemic: a case study of universities in Sri Lanka. Sri Lanka J Soc Sci Humanit 1(2):147–160. https://doi.org/10.4038/sljssh.v1i2.46

Hayash R, Garcia M, Maddawin A, Hewagamage KP (2020) Online Learning in Sri Lanka’s Higher Education Institutions during the CO VID-19 Pandemic. ADB Briefs. 151(151). https://doi.org/10.22617/brf200260-2

Hecht CA, Murphy MC, Dweck CS, Bryan CJ, Trzesniewski KH, Medrano FN, Giani M, Mhatre P, Yeager DS (2023) Shifting the mindset culture to address global educational disparities. Npj Sci Learn 8(1):29. https://doi.org/10.1038/s41539-023-00181-y

Hlatshwako TG, Shah SJ, Kosana P, Adebayo E, Hendriks J, Larsson EC, Hensel DJ, Erausquin JT, Marks M, Michielsen K, Saltis H, Francis JM, Wouters E, Tucker JD (2021) Online health survey research during COVID-19. Lancet Digit Health 3(2):e76–e77. https://doi.org/10.1016/S2589-7500(21)00002-9

Article CAS PubMed Central PubMed Google Scholar

Hoofman J, Secord E (2021) The effect of COVID-19 on education. Pediatr Clin North Am 68(5):1071–1079. https://doi.org/10.1016/j.pcl.2021.05.009

Hou D (2023) Constructing and realizing a model of college students’ learning engagement strategy based on factor analysis. In: Proceedings of the 2nd international conference on internet, education and information technology (IEIT 2022). Atlantis Press International BV, pp 390–397. https://doi.org/10.2991/978-94-6463-058-9_64

Howard SB (2009). The benefits of face-to-face interaction in the online freshman composition course. MERLOT J Online Learn Teach. http://jolt.merlot.org/vol5no4/howard_1209.pdf

Ikotun AM, Ezugwu AE, Abualigah L, Abuhaija B, Heming J (2023) K-means clustering algorithms: a comprehensive review, variants analysis, and advances in the era of big data. Inf Sci 622:178–210. https://doi.org/10.1016/j.ins.2022.11.139

Ilangarathna GA, Ranasinghe Y, Weligampola H, Attygalla E, Ekanayake J, Yatigammana S, Pinnawala M, Godaliyadda R, Herath V, Ekanayake P, Thilakaratne G, Dharmarathne S (2022) A comprehensive overview of education during three COVID-19 pandemic periods: impact on engineering students in Sri Lanka. Educ Sci 12(3):197. https://doi.org/10.3390/educsci12030197

Ilangarathna GA, Ramanayake L, Senarath N, Ranasinghe Y, Weligampola H, Dedunupitiya W, Thilakasiri I, Godaliyadda R, Ekanayake P, Herath V, Ekanayake J, Yatigammana S, Rathnayake A, Pinnawala M, Maheswaran M, Thilakaratne G, Dharmarathne S (2024) A dataset on the socioeconomic and behavioural impacts in Sri Lanka through multiple waves of COVID-19. Data Brief 53:110063. https://doi.org/10.1016/j.dib.2024.110063

Ilankoon IMPS, Kisokanth G, Warnakulasuriya SSP (2020) COVID-19: impact on undergraduate nursing education in Sri Lanka. J Public Health Res 9(s1):1–3. https://doi.org/10.4081/JPHR.2020.1916

Iwuoha JC, Jude-Iwuoha AU (2020) COVID-19: Challenge to SDG and globalization. Electron Res J Social Sci Hum, Vol 2. Issue III, Preprint at https://ssrn.com/abstract=3670330

Jackman JA, Gentile DA, Cho NJ, Park Y(2021) Addressing the digital skills gap for future education. Nat Hum Behav 5:542–545. https://doi.org/10.1038/s41562-021-01074-z

Jay J, Bor J, Nsoesie EO, Lipson SK, Jones DK, Galea S, Raifman J (2020) Neighbourhood income and physical distancing during the COVID-19 pandemic in the United States. Nat Hum Behav 4(12):1294–1302. https://doi.org/10.1038/s41562-020-00998-2

Josephson A, Kilic T, Michler JD (2021) Socioeconomic impacts of COVID-19 in low-income countries. Nat Hum Behav 5:557–565. https://doi.org/10.1038/s41562-021-01096-7

Kuhfeld M, Soland J, Tarasawa B, Johnson A, Ruzek E, Liu J (2020) Projecting the potential impact of COVID-19 school closures on academic achievement. Educ Res 49(8):549–565. https://doi.org/10.3102/0013189X20965918

Lan M, Pan Q, Tan CY, Law NWY (2022) Understanding protective and risk factors affecting adolescents’ well-being during the COVID-19 pandemic. Npj Sci Learn 7(1):32. https://doi.org/10.1038/s41539-022-00149-4

Li M, Luo L, Sikdar S, Nizam NI, Gao S, Shan H, Kruger M, Kruger U, Mohamed H, Xia L, Wang G (2021) Optimized collusion prevention for online exams during social distancing. Npj Sci Learn 6(1):5. https://doi.org/10.1038/s41539-020-00083-3

Lichand G, Doria CA, Leal-Neto O, Fernandes JPC (2022) The impacts of remote learning in secondary education during the pandemic in Brazil. Nat Hum Behav 6(8):1079–1086. https://doi.org/10.1038/s41562-022-01350-6

Liyanagunawardena T, Williams SA (2021) Emergency remote education: experience from Sri Lanka during Covid-19. Asian J Distance Educ 16(1):207–229. https://www.asianjde.com/ojs/index.php/AsianJDE/article/view/554

Mahmudah FN, Putra Eka CS, Wardana BH (2021) The impacts of Covid-19 pandemic: external shock of disruption education and financial stress cohesion. FWU J Soc Sci 15(2):42–64. https://doi.org/10.51709/19951272/Summer-2/3

Masonbrink AR, Hurley E (2020) Advocating for children during the COVID-19 school closures. Pediatrics 146(3):20201440. https://doi.org/10.1542/PEDS.2020-1440/36730

Matta G (2020) Science communication as a preventative tool in the COVID19 pandemic. Humanit Soc Sci Commun 7(1):159. https://doi.org/10.1057/s41599-020-00645-1

Mid-year population estimates by district & sex. Department Of Census And Statistics (n.d.) http://www.statistics.gov.lk/Population/StaticalInformation/VitalStatistics/ByDistrictandSex2024

Mizani H, Cahyadi A, Hendryadi H, Salamah S, Retno Sari S (2022) Loneliness, student engagement, and academic achievement during emergency remote teaching during COVID-19: the role of the God locus of control. Humanit Soc Sci Commun 9(1):305. https://doi.org/10.1057/s41599-022-01328-9

Mohammed KS, Rashid CA, Salih HA, Budur T (2020) The role of online teaching tools on the perception of the students during the lockdown of Covid-19. Int J Soc Sci Educ Stud. https://doi.org/10.23918/IJSSES.V7I3P178

Mpungose CB (2020) Emergent transition from face-to-face to online learning in a South African University in the context of the Coronavirus pandemic Humanit Soc Sci Commun 7(1):1–9. https://doi.org/10.1057/s41599-020-00603-x

Muilenburg LY, Berge ZL (2005) Student barriers to online learning: a factor analytic study. Distance Educ 26(1):29–48. https://doi.org/10.1080/01587910500081269

Novel coronavirus – China. (n.d.). https://www.who.int/emergencies/disease-outbreak-news/item/2020-DON233 . Accessed 4 Sep 2023

Parolin Z, Lee EK (2021) Large socio-economic, geographic and demographic disparities exist in exposure to school closures. Nat Hum Behav 5(4):522–528. https://doi.org/10.1038/s41562-021-01087-8

Qin X, Dai W, Jiao P, Wang W, Yuan N (2016) A multi-similarity spectral clustering method for community detection in dynamic networks. Sci Rep. 6(1):31454. https://doi.org/10.1038/srep31454

Article ADS CAS PubMed Central PubMed Google Scholar

Rameez A, Fowsar MAM, Lumna N (2020) Impact of Covid-19 on higher education sectors in Sri Lanka: a study based on South Eastern University of Sri Lanka. J Educ Soc Res 10(6):341. https://doi.org/10.36941/jesr-2020-0132

Rasul G, Nepal AK, Hussain A, Maharjan A, Joshi S, Lama A, Gurung P, Ahmad F, Mishra A, Sharma E (2021) Socio-economic implications of COVID-19 pandemic in South Asia: emerging risks and growing challenges. Front Sociol 6:629693. https://doi.org/10.3389/FSOC.2021.629693/BIBTEX

Rupasinghe RAA, Senanayake SGMP, Padmasiri DA, Ekanayake MPB, Godaliyadda GMRI, Wijayakulasooriya JV (2016) Modes of clustering for motion pattern analysis in video surveillance. In: 2016 IEEE International Conference on Information and Automation for Sustainability (ICIAfS). pp 1–6. https://doi.org/10.1109/ICIAFS.2016.7946550

Salleh MIM, Alias NA, Ariffin S, Ibrahim Z, Ramli AR, Aliman S (2023) The sudden transition to remote learning in response to COVID-19: lessons from Malaysia. Humanit Soc Sci Commun 10(1):254. https://doi.org/10.1057/s41599-023-01751-6

Sandamali MDJS (2016) Influence of family background on children’s education: a small pilot sample from Sri Lanka. Eur Cent Sci Technol 01:1–5

Schulz J, Robinson L (2022) Distance learning, digital inequality, and COVID-19: visualizing learning channels among California public school students. First Monday. https://doi.org/10.5210/fm.v27i4.12585

staff, R., Policy Centre for Inclusive Growth, I., & Sri Lanka, U (2020) The International Policy Centre for Social protection responses to COVID-19 in Sri Lanka 1

Sucena A, Silva AF, Marques C (2022) Reading skills intervention during the Covid-19 pandemic. Humanit Soc Sci Commun 9:45. https://doi.org/10.1057/s41599-022-01059-x

Suheera MYM, Wazeema TMF (2022) The dark side of online home-schooling after Covid-19 in Sri Lanka. Simulacra 5(1):69–82. https://doi.org/10.21107/sml.v5i1.14102

Tadesse S, Muluye W, Tadesse S, Muluye W (2020) The impact of COVID-19 pandemic on education system in developing countries: a review. Open J Soc Sci 8(10):159–170. https://doi.org/10.4236/JSS.2020.810011

Takács R, Takács S, Kárász JT, Oláh A, Horváth Z (2023) The impact of the first wave of COVID-19 on students’ attainment, analysed by IRT modelling method. Humanit Soc Sci Commun 10(1):127. https://doi.org/10.1057/s41599-023-01613-1

Tanemura N, Chiba T (2022) The usefulness of a checklist approach-based confirmation scheme in identifying unreliable COVID-19-related health information: a case study in Japan. Humanit Soc Sci Commun 9(1):270. https://doi.org/10.1057/s41599-022-01293-3

Tarkar P (2020) Impact of Covid-19 pandemic on education system. Int J Adv Sci Technol 29(9s):3812–3814. http://sersc.org/journals/index.php/IJAST/article/view/16620

Törő K, Csikós G (2022) Families face pandemic – a review of psychological suggestions and empirical findings regarding the Covid-19 pandemic. Eur J Soc Sci 5(1):127–138. https://doi.org/10.26417/958LQT68G

UNSDG, Policy brief: education during COVID-19 and beyond. (n.d.). https://unsdg.un.org/resources/policy-brief-education-during-covid-19-and-beyond . Accessed 4 Sep 2023

UNSDG, Policy brief: the impact of COVID-19 on children. (n.d.). https://unsdg.un.org/resources/policy-brief-impact-covid-19-children . Accessed 4 Sep 2023

van Dam A, Dekker M, Morales-Castilla I, Rodríguez MÁ, Wichmann D, Baudena M (2021) Correspondence analysis, spectral clustering and graph embedding: applications to ecology and economic complexity. Sci Rep 11(1):8926. https://doi.org/10.1038/s41598-021-87971-9

Van Lancker W, Parolin Z (2020) COVID-19, school closures, and child poverty: a social crisis in the making. Lancet Public Health 5(5):e243–e244. https://doi.org/10.1016/S2468-2667(20)30084-0

von Stumm S, Cave SN, Wakeling P (2022) Persistent association between family socioeconomic status and primary school performance in Britain over 95 years. Npj Sci Learn 7(1):4. https://doi.org/10.1038/s41539-022-00120-3

Wardropper CB, Dayer AA, Goebel MS, Martin VY (2021) Conducting conservation social science surveys online. Conserv Biol 35(5):1650–1658. https://doi.org/10.1111/cobi.13747

Watson R, Thompson DR (2006) Use of factor analysis in Journal of Advanced Nursing: literature review. J Adv Nurs 55(3):330–341. https://doi.org/10.1111/j.1365-2648.2006.03915.x

Wetzel AP (2012) Factor analysis methods and validity evidence: a review of instrument development across the medical education continuum. Acad Med 87(8):1060–1069. https://doi.org/10.1097/ACM.0b013e31825d305d

WHO, Coronavirus (COVID-19) Dashboard | WHO coronavirus (COVID-19) dashboard with vaccination data. (n.d.). https://covid19.who.int/ . Accessed 4 Sep 2023

Zaman NU, Bibi Z, Sheikh SUR, Raziq A (2020) Manualizing factor analysis of Likert scale data. J Manag Sci 7(2):56–67. https://ideas.repec.org/a/gei/journl/v7y2020i2p56-67.html

Zamir S, Wang Z (2023) Uncovering Covid-19, distance learning, and educational inequality in rural areas of Pakistan and China: a situational analysis method. Humanit Soc Sci Commun 10:679. https://doi.org/10.1057/s41599-023-02025-x

Zhang XC, Kuchinke L, Woud ML, Velten J, Margraf J (2017) Survey method matters: online/offline questionnaires and face-to-face or telephone interviews differ. Comput Hum Behav 71:172–180. https://doi.org/10.1016/J.CHB.2017.02.006

Faculty of Engineering, University of Peradeniya, Peradeniya, Sri Lanka

Neranjan Senarath, Wathsala Dedunupitiya, Gayanthi Ilangarathna, Isuru Thilakasiri, Bhagya Samarakoon, Lakshitha Ramanayake, Roshan Godaliyadda, Parakrama Ekanayake, Vijitha Herath & Janaka Ekanayake

Department of Geography, Virginia Polytechnic Institute and State University, Blacksburg, VA, USA

Faculty of Arts, University of Peradeniya, Peradeniya, Sri Lanka

Mallika Pinnawala & Sakunthala Yatigammana

Faculty of Medicine, University of Peradeniya, Peradeniya, Sri Lanka

Institute of Policy Studies, Colombo, Sri Lanka

Search author on: PubMed Google Scholar

Neranjan Senarath (NS) - Conceptualization, Methodology, Visualization, Validation, Writing (original draft). Wathsala Dedunupitiya (WD) - Conceptualization, Methodology, Validation, Writing (original draft). Gayanthi Ilangarathna (GI) - Conceptualization, Methodology, Visualization. Isuru Thilakasiri (IT) - Methodology, Investigation, Visualization, Validation, Writing (original draft), Writing (review & editing). Bhagya Samarakoon (BS) - Conceptualization, Methodology. Lakshitha Ramanayake (LR) - Conceptualization, Methodology. Roshan Godaliyadda (RG) - Conceptualization, Methodology, Data curation, Supervision, Writing (original draft), Writing (review & editing), Funding acquisition. Parakrama Ekanayake (PE) - Conceptualization, Methodology, Data curation, Supervision, Writing (original draft), Writing (review & editing), Funding acquisition. Vijitha Herath (VH) - Conceptualization, Methodology, Data curation, Supervision, Writing (original draft), Writing (review & editing), Funding acquisition. Mallika Pinnawala (MP) - Conceptualization, Methodology, Data curation, Supervision. Sakunthala Yatigammana (SY) - Conceptualization, Methodology, Data curation, Supervision. Janaka Ekanayake (JE) - Conceptualization, Methodology, Data curation, Supervision, Funding acquisition, Resources. Samath Dharmarathne (SD) - Conceptualization, Data curation, Supervision. Ganga Tilakaratne (GT) - Conceptualization, Data curation, Supervision.

Correspondence to Isuru Thilakasiri .

The authors declare no competing interests.

The ethical clearance for this study was obtained from the Ethical Review Committee, Faculty of Arts, University of Peradeniya, Sri Lanka, (ARTS/ERC/2021/01 and 18 September 2021) with the support of the Department of Sociology. Administrative clearance was obtained by the Ministry of Home Affairs of Sri Lanka, relevant Divisional Secretariate (DS) offices, and relevant Grama Niladari (GN) offices.

Informed verbal and written consent were obtained from all the participants. Voluntary participation was ensured. Privacy and confidentiality of data were maintained during each step of data collection, adhering to standard practices and protocols.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Senarath, N., Dedunupitiya, W., Ilangarathna, G. et al. Family income does not necessarily lead to a holistic e-learning experience during COVID-19: a study in Sri Lanka. Humanit Soc Sci Commun 11 , 1625 (2024). https://doi.org/10.1057/s41599-024-04099-7

Version of record : 29 November 2024

DOI : https://doi.org/10.1057/s41599-024-04099-7

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

Humanities and Social Sciences Communications ( Humanit Soc Sci Commun )

© 2026 Springer Nature Limited