Demand for internet services before and during the Covid-19 pandemic: what lessons are we learning in South Africa?

The primary aim of this study was to assess the demand for internet services before and during the Covid-19 pandemic, considering the challenges and opportunities brought about by the global health crisis. While the pandemic has had numerous negative impacts on people's lives, it has also facilitated advancements in technology, particularly the adoption of the 4th industrial revolution. To explore the positive impacts of these technological advancements, the study focused on analysing changes in household internet usage using the 2019 and 2021 General Household Survey data obtained from STATS SA. The study examined the shifts in internet usage between the two data sets and found a modest increase in internet usage over time. To further investigate the determinants of household internet usage, the study employed descriptive analysis, cross-tabulations, and a binary logistic regression model. Income, age, household size, and gender were used as independent variables, while internet usage served as the dependent variable. The results revealed that all the independent variables were statistically significant factors influencing the probability of internet usage. Income and household size demonstrated a positive relationship with internet usage, indicating that higher levels of income and larger household sizes were associated with increased demand for internet services. Conversely, the age of the household head showed a negative effect on internet usage, suggesting that as individuals grew older, their likelihood of using the internet decreased. Additionally, the study found that male-headed households exhibited higher levels of internet usage compared to their female counterparts. To ensure that digital inclusion is prioritized, it is crucial for authorities to ensure that internet access is accessible to low-income households. Addressing the disparity in internet usage between higher and lower-income households is essential. Government regulators can encourage broadband providers to expand affordable internet access, while reducing administrative burdens to facilitate network deployment, thereby supporting the current levels of internet usage, and promoting further growth. By considering these findings, policymakers and stakeholders can develop strategies to bridge the digital divide and ensure equal access to internet services for all segments of society. This will contribute to a more inclusive and equitable digital landscape, fostering social and economic development in the medium to long term. © 2023 by the authors. Licensee SSBFNET, Istanbul, Turkey. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).


Introduction
Globally, the COVID-19 pandemic had an impact on every element of human life, including research, sports, entertainment, worship, transportation, social gatherings and interactions, commercial operations, and politics.The reality of the situation was difficult to bear because of the threat of COVID-19, and the education sector is still one of those most severely affected by the outbreak (Onyema et al., 2020).Due to disruptions in education and issues with global health that were exceedingly challenging for international health systems to handle, the Coronavirus pandemic outbreak widened the world's educational gaps.Everyone in the world was affected by the pandemic, as COVID-19's rapid spread and catastrophic effects wiped out entire nations or races.The coronavirus outbreak had a worldwide impact that spread quickly and quickly.The disease radically altered global lifestyles just a few months after it first 627 emerged, forcing billions of people to "remain at home," "observe self-isolations," and work and study from home (Mhlanga and Moloi 2020).
The pandemic restricted people's freedom of movement, trade, and association, and COVID-19 not only put many nations on complete lockdown but also killed thousands of people, mostly women and old people.Knowing that studies from other continents, including America, Africa, Asia, and Europe, suggested a surge in COVID-19-related new cases and mortality daily was more concerning (Onyema et al., 2020).According to experts, any disease epidemic must be contained as soon as possible, but the coronavirus caught everyone off guard, and many nations including the superpowers were originally unprepared for the pandemic (Benfer et al., 2019, Clipper 2020, Mbunge et al., 2022).The World Health Organization (W.H.O.) released recommendations and updates on how to stop the spread of the pandemic soon after COVID-19 first appeared, and several nations subsequently added additional measures to the W.H.O.recommendations to stop the disease's spread (Shahbaz et al., 2020, Cohen et al., 2021).To impose coronavirus limitations and lower the number of covid cases, there were lockdowns in most of the world's cities, people were urged to work from home, and some nations even sent out their armed forces.Some chosen COVID-19 comments from throughout the world, especially concerning the closing of schools because of the coronavirus.Numerous schools were closed in South Africa, and exams and tests that were scheduled were also postponed.
Numerous public and private institutions turned to technology as part of their pandemic response plans.Numerous studies were done in a variety of sectors in reaction to the epidemic to attempt and comprehend how it affected humanity.For instance, Onyema et al. (2020) examined the effects of the coronavirus pandemic on schooling in the Education Literature.According to Onyema et al. (2020), the coronavirus pandemic caused severe worries for the world's educational systems and forced the unscheduled shutdown of schools in more than 100 nations.Onyema et al. (2020) also discovered that COVID-19 has negative consequences on education, such as disruptions to study, limited access to facilities for education and research, job losses, and higher student debt.Furthermore, Onyema et al. (2020) discovered that while many instructors and learners relied on technologies to deliver that learning continued online during the Coronavirus global epidemic, online education was hampered by subpar infrastructures such as network, power, lack of access and lack of availability issues, and low digital literacy.In line with the new worldwide trends and realities in education, Onyema et al. (2020) emphasize the negative consequences of COVID-19 on the education sector and the necessity for all academic institutions, educators, and students to use technology and enhance their digital abilities.Mukuna and Aloka (2020) investigated the perceived difficulties rural educators faced with online learning in the wake of the COVID-19 outbreak at a particular rural school in South Africa.According to Mukuna and Aloka (2020), there are many obstacles to online learning, including a lack of learning tools, inadequate personal protective equipment, poor network connectivity, and a lack of parental involvement in children's homework assignments.Mukuna and Aloka (2020) recommended that stakeholders like the Department of Education, Department of Basic Education, and School Governing Bodies, among others, make sure that parents are aware of the necessity for students to manage the resources at their disposal and that it is crucial to provide adequate resources like internet connectivity and learning devices for information and communication technologies.In a different study, Mhlanga and Moloi (2020) evaluated the role of the COVID-19 epidemic in spurring digital change in South Africa's education system.The research by Mhlanga and Moloi (2020) assumed that learning ceased because of the lockdown caused by COVID-19 in South Africa and most of the world and that education was moved online.In South Africa, Mhlanga and Moloi (2020) discovered that a variety of 4IR tools were released during the lockdown from early schools to higher and postsecondary education where learning programs shifted to remote (online) learning.These observations represent that South Africa essentially does have some pockets of excellence to push the education system into the 4th industrial revolution, which can increase access.Mhlanga and Moloi (2020) conclude that despite the immense human suffering this epidemic has caused around the world, it has also provided a chance to evaluate the achievements and shortcomings of deployed technologies, their costs, and how these technologies might be scaled to increase access.Mbunge et al. (2022), also conducted a study to determine the digital health technologies and virtual healthcare services used in South Africa during the coronavirus and the difficulties connected with their use, which placed fourth in the field of health literature.The study found that South Africa used digital technologies like artificial intelligence, chatbots, robotics, and SMS-based solutions to deliver healthcare services during the COVID-19 pandemic.It also found that South Africa adopted telemedicine and telehealth, mobile health applications, telemedicine, and WhatsApp-based systems.In addition to "screening infectious and non-infectious diseases, disease surveillance and monitoring, medication and treatment compliance, creating awareness and communication", these cutting-edge technologies have also been employed for other uses.The study also found that during COVID-19 in South Africa, "virtual healthcare services were provided employing digital health technologies in the areas of teleconsultation and e-prescription, telelaboratory and telepharmacy, teleeducation and teletraining, teledermatology, teleradiology, telecardiology, teleophthalmology, teleneurology, telerehabilitation, Tele oncology, and telepsychiatry".These smart digital health technologies, according to Mbunge et al. (2022), also confront several obstacles, including infrastructural and technological ones, organizational and financial ones, governmental and regulatory ones, as well as cultural ones.Considering this, the current study aims to examine the demand for internet services before and during the COVID-19 pandemic and to draw conclusions for South Africa that can assist policymakers on what should be done to take advantage of the benefits derived from mobile broadband use.

Literature Review
Several studies were undertaken especially in South Africa to understand the impact of the pandemic on various sectors of the economy, for instance, Mhlanga and Moloi (2020), Mbunge et al., (2022), Mukuna and Aloka (2020), Onyema et al., (2020) among others.The current study is the first of its kind in South Africa to examine how the pandemic affected the demand for the internet before and after the pandemic.As a result, it is equally important to have a review of the literature that will provide a clear picture of the situation that was taking place in various sectors.Having a thorough definition of all the different parts of this study will be the first step.

Definition of Key Terms in the Study
The terms that will be described in this study are internet broadband and coronavirus disease.Definitions of these terms are necessary to paint a clear picture of the study's overall direction.

The Coronavirus Disease
A contagious illness called the coronavirus sickness first appeared in Wuhan, China, in 2019.The World Health Organization later designated it as "COVID-19," which stands for Coronavirus Disease 2019.(WHO, 2020, Mhlanga and Moloi 2020, Onyema et al., 2020).One of the most serious global pandemics in recent memory is the coronavirus outbreak.The high fatality rate and distressing ease of transmission were both problems.According to research, those who were older or have underlying medical conditions including cancer, diabetes, chronic respiratory disease, or cardiovascular disease were more prone to experience serious coronavirusrelated illnesses.The symptoms of a coronavirus include a sore throat, runny nose, persistent coughing and sneezing, difficulty breathing, and exhaustion (Onyema et al., 2020).

Internet Broad Band
According to many definitions, the internet is a vast network of computers that functions similarly to the postal service but only at frequencies of less than one second (Mhlanga and Beneke 2021).According to the BBC-WebWise (2012), because users may transmit envelopes containing messages and packages of digital data through the internet, it functions similarly to the postal service.Internet protocol (TCP/IP) is the common language that the internet uses to function, and the IP address facilitates internet connectivity in this manner.The US military department's advanced research project agency's creation of the ARPAnet in the 1960s is recognized as being the forerunner of the internet (BBC-WebWise, 2012).Then, in many nations, private commercial companies developed a wide variety of networks.The inability of these networks to link was the only issue.In 1974, Vint Cerf and Bob Kahn created the TCP/IP, which produced a system for connecting packet networks.As a result, despite the IP's dominance, the internet evolved into a network of networks (BBC-WebWise, 2012).
Broadband Internet is one popular form of access to the internet.Due to its fast access speeds, broadband Internet service which is available in four different configurations is the most popular way to access the Internet.These configurations include fibre-optic, cable, satellite, and Digital Subscriber Line or DSL.Only the outdated dial-up connection is a non-broadband option, and although being less expensive, most Web users are switching to the speedier broadband connectivity.For instance, Internet service through a digital subscriber line connects by using unused phone lines, which don't interfere with your phone service.The speed of a DSL connection depends on how far away you are from the switch point.When trying to decide between a DSL line and a cable connection, the speed will be slower the farther you are from the switchboard and faster the closer you are.

Empirical Literature Review
Globally, the COVID-19 pandemic had an impact on every element of human life, including research, sports, entertainment, worship, transportation, social gatherings and interactions, commercial operations, and politics.Several studies were conducted in response to better understand its effects on the economy and society at large.The impact of the COVID-19 pandemic on technology adoption in the healthcare industry was examined by Clipper (2020).According to Clipper (2020), it is unlikely that we will completely reverse course and go back to where we started in the areas of telehealth and virtual care, artificial intelligence, and the use of robots in healthcare in February 2020, even though some technology adoption leaps seem extraordinary.Even though it has only been a few months, Clipper (2020) concluded that it is difficult to remember life before the COVID-19 pandemic.This is true both in our personal lives and in how we use technology in healthcare.Again, according to Clipper (2020), while the COVID-19 epidemic has had terrible effects on people, it has also increased the need for and capacity for incorporating technological solutions at a rate that has never been witnessed before.Al-Maroof et al. (2020) also investigated how the fear emotion affected teachers' and students' adoption of technology during the COVID-19 epidemic, utilizing Google Meet as a social learning platform at for-profit universities.According to Al-Maroof et al. (2020), the most frequent threats that kids and teachers/educators may encounter during the Coronavirus pandemic include fear related to a family lockdown situation, fear of educational failure, and worry of losing social contacts.Ahin and Ahin (2022) looked at the factors that influenced pre-service teachers' enthusiasm to accept new technologies during the COVID-19 pandemic as well as their psychological needs and feelings.According to Ahin and Ahin (2022), the importance of motivation becomes even more clear when considering how learning and teaching environments are being digitally transformed, particularly considering the pandemic's effects.Ahin and Ahin (2022) concluded that innovativeness, which is connected to technological use and motivation, had a variety of moderating impacts on the links between intrinsically and extrinsically encouragement and technology adoption, particularly during the pandemic.
In South Africa, Aruleba et al. (2022) looked at disadvantaged colleges' readiness to adopt new technology, during the COVID-19 epidemic.According to Aruleba et al. (2022), the Covid-19 pandemic has killed over four million people and affected hundreds of millions more, and governments and policymakers have recognized the necessity for immediate action to stop the virus' spread.According to Aruleba et al., (2022), to contain the virus, governments and policymakers around the world implemented a variety of protection measures and interventions to alter the behaviour of their citizens, primarily through social exclusion, interprovince lockdown, stay-at-home tactics, and quarantines.Aruleba et al. (2022), however, stated that the various lockdown tactics had produced unusual and difficult circumstances that had no known analogue in the education sector.Many higher education institutions around the world were not prepared to adapt to online teaching and learning when the governments announced the sudden lockdown, according to Aruleba et al., (2022).'s analysis.Once more, Aruleba et al. (2022) found that the respondents' eagerness to adopt technology was correlated with their average optimism and innovativeness.On the other hand, higher average levels of unease and discomfort reflect the respondent's difficulties embracing new technologies.Mhlanga and Beneke (2021) looked at the elements that affect South African households' access to the internet in a different study conducted in that country.
According to Mhlanga and Beneke (2021), access to broadband facilities by individuals, households, and consumers is one of the essential elements of a country's economic growth and prosperity given the rapid technological advancements and changes that are being observed everywhere in the Fourth Industrial Revolution.According to Mhlanga and Beneke (2021), a community or country's access to broadband technologies contributes to increased productivity, which is a major driver of economic growth and influences poverty levels.Mhlanga and Beneke (2021) used logistic regression to determine that the significant variables influencing the demand for internet access by households in emerging economies were race, access to a landline, access to a mobile phone, access to electricity, ownership of a home, gender, the age of the household head, net household income per month, and household expenditure.The availability of energy and access to a cellphone, per Mhlanga and Beneke (2021), were the elements that had the most impact on people's capacity to use the internet.
Additionally, Fu and Mishra (2022) investigated the use of fintech during the COVID-19 era.Fu and Mishra (2022) found that the proliferation of COVID-19 and associated government lockdowns caused a sizable increase in the rate of finance app downloads by using mobile application data from a globally representative sample.Again, Fu and Mishra (2022) discovered that while established incumbents initially had the greatest growth in their digital products, "BigTech" firms and more recent fintech providers ultimately beat them in the long run.In the health literature, Mhlanga (2022) also discovered that the COVID-19 pandemic caused disturbances in every facet of human existence and had a significant impact on all global economic sectors.Mhlanga (2022) also discovered that the pandemic, particularly in developing and underdeveloped countries, halted and reversed progress in health and consequently shortened life expectancy.However, machine learning and artificial intelligence significantly aided in the global response to the pandemic.
According to Mhlanga (2022), among other significant accomplishments, artificial intelligence and machine learning played a significant role in the response to the COVID-19 pandemic challenges by scaling customer communications, providing a platform for understanding how COVID-19 spreads and speeding up research and treatment of COVID-19.Mhlanga (2022) concluded that governments must increase their trust in artificial intelligence and machine learning to address future health issues and ensure that the sustainable development goals related to good health and wellbeing are met, despite the disruptions and rise in the number of unintended consequences brought on by technology in the fourth industrial revolution.By adopting and using the iterative independent review approach from Braun and Clarke's framework, Lalani et al. (2021) developed a thematic assessment and crucial evaluation of organizational behaviour patterns to provide a critical comprehension of organizational behaviour patterns that influence the adoption and implementation of educational technology in the higher education sector.Authentic leadership, servant leadership, leader-member interaction, contextual management, shared leadership, and unproductive forms of leadership are among the leadership styles that emerged from the thematic analysis, according to Lalani et al. (2021).Lalani et al. (2021) discovered further that the intersectionality of these styles with elements of the leadership environment, such as financial management and social networks, provides a greater understanding of the kinds of leadership behaviours that, if established in higher academic leaders, might assist larger readiness for prospective times of uncertainty and change.Bokolo (2021) investigated the use of virtual software and telemedicine for the treatment of outpatients during and after the COVID-19 epidemic.Bokolo (2021) found that software-based solutions, such as medical software applications, could offer helpful advice on health-related information to doctors to improve quality of life, particularly for outpatients like the elderly, immunosuppressed, and pregnant women.The utilization of virtual software and telemedicine, according to Bokolo (2021), shows encouraging possibilities in the struggle against COVID-19.Bokolo (2021) concluded that by virtual treatment of patients during and after the COVID-19 pandemic, telemedicine and virtual software can reduce emergency department visits, safeguard healthcare resources, and minimize the spread of COVID-19.
In a related study, Ren, and Kwan (2009) focused specifically on gender variations in Internet management and leisure activities while examining the intricate relationships between various types of Internet use and physical activity.According to Ren and Kwan's (2009) findings, there are gender-specific differences in how Internet use affects people's behaviour, such as travel patterns.Overall, Ren and Kwan (2009) discovered that whereas men's physical activities and travel are more significantly impacted by Internet use for leisure, women's physical activities and travel are more significantly impacted by Internet use for maintenance purposes.The findings of Ren and Kwan (2009) show that categorizing Internet activities into separate groups shows some underlying trends that would not have been noticed if these various forms of Internet activity were grouped as a single category.Golin (2021) examines the causal relationship between adult mental health and broadband Internet access.According to Golin's results from 2021, broadband Internet use is associated with lower mental health in women, especially those between the ages of 17 and 30, but not in men, expanding the gender gap in mental diseases.Broadband connection, according to Golin (2021), worsens socializing behaviour and one's capacity for coping with emotional issues, which are related to sub-aspects of mental health.The findings are concentrated among younger female cohorts, which suggests that high Internet activity intensity magnifies the detrimental effects of internet access on mental health, according to Golin's (2021) analysis.

Research & Methodology
The main objective of this study is to investigate the enduring impact of the COVID-19 pandemic on changes in access to internet usage in South Africa, considering the post-pandemic landscape.By utilizing a data-driven approach, the study aims to comprehensively examine the evolving patterns of internet usage following the pandemic and identify the factors influencing these changes.The General Household survey conducted by Stats SA (2021) will continue to serve as the primary dataset, enabling a comparative analysis of internet usage data collected before and after the outbreak.In addition to assessing the overall impact, this study will specifically focus on understanding the challenges faced by socio-economically disadvantaged households in attaining internet usage.By analyzing data from different income groups, the study aims to identify any disparities and variations in internet access and utilization.Specifically, it will investigate whether there are differences in internet usage between affluent and economically disadvantaged households and explore the underlying factors contributing to such disparities.Through the enhanced methodology, incorporating descriptive analysis, cross-tabulation, and regression analysis techniques, this study will provide insights into the lasting impact of COVID-19 on internet usage patterns, as well as the specific challenges faced by economically disadvantaged households in accessing and utilizing the internet.The findings will contribute to a deeper understanding of the digital divide in South Africa and inform policies and interventions aimed at reducing inequalities in internet access and promoting digital inclusivity in the post-COVID era.

Regression and model specification
This study employed a binary logistic regression analysis to delve into the determinants of demand for internet usage, drawing on the general household data from General household data from South Africa in 2021.The main objective was to examine the significant household characteristics that contribute to the likelihood of whether a household utilized the internet or not.The regression model was formulated as follows, with internet usage serving as the dependent variable.Internet usage was measured as a categorical variable with two distinct categories: households that reported using the internet and households that reported not using the internet.To explore the relationships between internet usage and various household characteristics, the study incorporated several independent variables.These variables included income, age, household size, gender, and race.The inclusion of income aimed to gauge the influence of households' financial resources on their internet usage patterns.Age was considered to assess potential variations in internet usage across different age groups within households.Household size was considered to explore whether the number of individuals within a household affected their likelihood of internet usage.Gender was considered as a potential factor influencing internet usage, examining whether there was gender-based differences in access and utilization.Lastly, race was included to explore any disparities in internet usage patterns among different racial groups within households.
To examine the relationships between the dependent variable (internet usage) and the independent variables, a binary logistic regression model was specified.This statistical approach allowed for the estimation of the probability of a household using the internet based on the specified independent variables.By analyzing the coefficients and significance levels of the independent variables, the study aimed to identify which household characteristics were most influential in determining the likelihood of internet usage.Overall, the utilization of a binary logistic regression model in this study provided a rigorous and quantitative approach to explore the determinants of internet usage demand among households in South Africa.By considering multiple household characteristics, the study aimed to provide valuable insights into the factors shaping internet usage patterns and highlight potential areas for intervention and policy development to promote broader access and utilization of internet services.
The regression model was specified as follows: In equation ( 1), IUt represents the estimated value of the dependent variable (often denoted as Y) for a particular observation or case.The terms β₀, β₁, β₂, ..., βₙ represent the regression coefficients or parameters associated with the independent variables X₁, X₂, ..., Xₙ respectively.These coefficients quantify the impact or effect that each independent variable has on the dependent variable.The term εᵢ represents the error term or residual, which captures the unexplained variability or randomness in the relationship between the dependent and independent variables.
To construct the regression model, all the variables of interest are included in equation (1).The independent variables X₁, X₂, ..., Xₙ represent the factors or variables that are hypothesized to influence or explain the variation in the dependent variable IUt.These variables can be quantitative (such as age, income, or household size) or qualitative (such as gender or race).The coefficients β₁, β₂, ..., βₙ represent the estimated effects of these independent variables on the dependent variable, controlling for other variables included in the model.
By estimating the values of the regression coefficients (β₀, β₁, β₂, ..., βₙ), the model allows for the quantification of the relationships between the independent variables and the dependent variable.The coefficients indicate the direction (positive or negative) and magnitude of the impact of each independent variable on the dependent variable.The error term εᵢ accounts for the unobserved or omitted factors that influence the dependent variable but are not included in the model.
Overall, the regression model (equation 1) provides a systematic framework for analyzing and understanding the relationship between the dependent variable (IUt) and the independent variables (X₁, X₂, ..., Xₙ), enabling the estimation of the dependent variable's value based on the specified set of independent variables.
The regression therefore will have all the variables of interest included as follows.
=  0 +  1 () +  2 () +  3 (ℎℎ ) +  4 () …    1 _ 4 are the coefficients for the corresponding variables, In equation ( 2), The dependent variable, IUt, is defined as a binary variable where 1 represents the individual has used the internet and 0 represents the individual has not used the internet.The regression model includes the variables of interest, namely Income, age, household size, and gender.Each independent variable is associated with a coefficient (β₁, β₂, β₃, β₄) that measures the impact or effect of that variable on the probability of using the internet.For example, β₁ represents the change in the probability of using the internet for a unit increase in Income, while β₄ represents the difference in the probability of using the internet between males and females.
It is important to note that in this specific model, gender is treated as a categorical variable, with males assigned a value of 1 and females assigned a value of 0.
The constant term, β₀, represents the intercept of the regression model, capturing the baseline probability of using the internet when all independent variables are zero or absent.The error term, εᵢ, accounts for unexplained variability or random factors that influence the dependent variable but are not included in the model.

Findings and Discussion
This section presents the results of the study, as indicated in the methodology the paper compared the usage of the internet before Covid (2019) and after Covid (2021).The first part presents demographic results followed by cross-tabulation and lastly regression results.The results are presented as follows: The data reveals that the majority of individuals accessed the internet through mobile devices, with 51% of the population indicating that they had internet access via mobile devices.In contrast, 49% of the population reported not having access to the internet through mobile devices.There was a significant disparity in internet access from the workplace, with only 16% of respondents reporting that they were able to access the internet in this setting, while the remaining 84% indicated otherwise.The high prevalence of internet access through mobile devices suggests a notable shift in technology adoption.This trend aligns with previous literature, which discusses the increasing affordability of mobile devices, potentially contributing to the greater accessibility of internet usage.It is worth noting that the descriptive analysis provides an overview of internet usage patterns in 2019.These findings lay the groundwork for further analysis and exploration of the factors influencing internet access and usage in subsequent years.) increase is evident between the two years, the increase is mainly seen amongst those who indicated to have been able to use internet access through the mobile device between 2019 and 2021.The increase has led to a decrease in the percentage of the population who indicated not have been able to use the internet in the same period.The results indicate a decrease in usage of the internet from the workplace this could be because most people are now working from home due to restrictions of Covid-19.This was consistent with the findings of Comini ( 2020), who concluded that there had been an increase in internet usage because of the COVID-19 epidemic and the ensuing lockdown measures, which increased internet traffic globally.Comini (2020) went further to argue that African nations exhibit this global trend, with data traffic increasing in the months when "stay at home" orders were issued.
According to Comini (2020), the COVID-19 issue offered the possible impetus required to bring about ubiquitous connectivity.The COVID-19 and the associated shutdown steps to stop the virus transmission, according to Comini (2020), have underlined how crucial it is for everyone to have access to dependable and resilient digital infrastructure.In a similar vein, Feldmann et al. ( 2021) discovered a rise in internet traffic of roughly 15-20% within a couple of weeks of the lockdown, and the increase was frequently spread over numerous months under normal operation.

Demographic results of usage of internet usage in 2022.
This section presents demographic descriptive results of usage of the internet in 2021.At this point, the study highlights more in 2021 to uncover the effects of Covid-19 in terms of internet usage and further highlight some of the determinants of the increase of the usage of data in this period post-Covid-19.Table 3 presents descriptive results of all continuous variables in the selected data set.It indicates that according to age the minimum age of the household head was 15 years and the maximum was 116 with an average of 51 years.On household size, the minimum number per household was 1 and maximum 24 and a mean of 3 people per household.On income, the lowest income was 100 rand and the highest 66000 with a mean of 9329.02rand per month.

Figure 1: Household Expenditures
Figure 1 displays the findings on household expenditures based on the monthly household income.The results reveal distinct patterns in expenditure distribution among households in South Africa.Notably, a significant proportion of households (25%) allocate a monthly budget ranging from 2500 to 5000 rand.Conversely, approximately 12% of households have higher spending capacity, with monthly expenditures surpassing 10,000 rand.Conversely, a negligible percentage of households allocate less than 200 rand or report no expenditure at all.These findings underscore the increasing cost of living in South Africa.The data indicates that a considerable portion of households must allocate a substantial portion of their monthly income to meet various expenses.As a result, the financial burden on households is evident, emphasizing the growing challenges of maintaining a satisfactory standard of living.

Figure 2. Gender of household head
Figure 2 provides an overview of the gender dynamics within households, specifically focusing on the distribution of household heads.The findings reveal that 47% of households in South Africa are headed by males, while 53% are headed by females.The results highlight the significant presence of female-headed households in South Africa.This suggests a prevailing trend where women assume the role of household heads, taking on responsibilities related to decision-making, financial management, and overall household management.The data reflects the importance of recognizing the central role played by women in South African households and the implications this has for gender dynamics and gender equality in the country.These findings contribute to a deeper understanding of the gendered nature of household leadership and shed light on the distribution of decision-making power within households.Recognizing the predominance of female-headed households is crucial for policy-makers and researchers aiming to address gender disparities and promote inclusive socio-economic development.The results indicate that the majority of household heads in South Africa belong to the black African population group, accounting for 86% of the total.The second largest population group among household heads is coloured, representing 6.6% of the sample.The white population group follows closely, comprising 5.5% of household heads.Lastly, the Indian population group constitutes the smallest percentage, with 1.6% of household heads.These findings highlight the demographic diversity within South African households and provide insights into the representation of different population groups in household leadership roles.The overrepresentation of black African household heads suggests the significance of this population group in shaping and influencing household dynamics, decision-making processes, and socio-economic outcomes.It also reflects the country's historical context and the ongoing efforts towards achieving social cohesion and addressing historical inequalities.Understanding the distribution of household heads across population groups is vital for policymakers, researchers, and organizations working towards promoting inclusivity, addressing socio-economic disparities, and ensuring equitable opportunities for all population groups.These findings contribute to a more comprehensive understanding of the demographic landscape and inform targeted interventions and policies aimed at fostering social cohesion and equality in South Africa.

Cross tabulation results
This section presents cross-tabulation results comparing the usage of the internet in 2021 with the selected demographics as follows: Table 4 provides cross-tabulation results examining the relationship between the demand for internet usage and household monthly income.The findings reveal a clear association between income levels and the demand for internet access.Among those who reported using the internet, a significant majority (55.7%) belonged to the income bracket of 0 to 5000 rand per month.As income levels increased, the percentage of individuals indicating internet usage also increased, with the trend continuing to rise.Notably, among those with an income of 100,000 rand and above, over 80% reported using the internet during this period.Conversely, among those who reported not using the internet, the proportion of individuals in the lower income brackets was higher.Specifically, 44% of this group fell within the income range of 0-5000 rand per month.As income levels rose, the percentage of individuals indicating no internet usage decreased.For instance, among those with an income of over 100,000 rand per month, the figure dropped to below 20%.
These results highlight a clear relationship between income and the demand for internet data.It is reasonable to conclude that individuals with higher incomes are more likely to afford and purchase data, allowing for greater internet usage.On the other hand, 86.30% 6.60% 1.60%

5.50%
African/Black Coloured Indian/Asian White population group individuals with lower incomes may face financial constraints that limit their ability to access and utilize the internet.Understanding the link between income and the demand for data is essential for policymakers, service providers, and organizations aiming to bridge the digital divide and promote equitable access to internet services.By recognizing the income-related disparities in internet usage, targeted interventions can be designed to ensure affordability and accessibility for individuals across different income brackets.These findings emphasize the importance of addressing economic inequalities to achieve a more inclusive and digitally connected society.Table 5 presents the cross-tabulation results examining the relationship between the demand for internet usage and the population group of households.The findings reveal disparities in internet access across different population groups.Among the African/Black and Coloured households, 64% indicated that they were able to use the internet.This suggests a relatively lower level of internet access compared to the other population groups.In contrast, a larger percentage of Indian and White households, with over 80%, indicated that they had access to the internet.Furthermore, when considering those who indicated no usage of the internet, a higher percentage belonged to the Black and Coloured population groups, with 35% representing this category.On the other hand, a smaller percentage of Indians and White people, at 18% and 23% respectively, indicated no usage of the internet.These findings suggest that there are disparities in internet access and usage based on population group.The higher percentage of African/Black and Coloured households without internet access indicates potential challenges in affordability or availability of data services.The lower internet access among these population groups may be attributed to various factors, such as financial constraints or limited infrastructure in certain areas.Addressing these disparities in internet access and usage across population groups is crucial for promoting digital inclusion and bridging the digital divide.It highlights the need for targeted initiatives and policies that aim to improve affordability, infrastructure, and digital literacy among marginalized communities.By addressing the data problems faced by the African/Black and Coloured population groups, efforts can be made to ensure equitable access to the internet, fostering equal opportunities for all individuals to participate in the digital age.The findings suggest that, overall, a higher percentage of males indicated that they used the internet compared to females.However, the difference in usage between the two genders was relatively small, with 66% of males and 64% of females reporting internet usage.The results raise the question of why males showed slightly higher internet usage compared to females.One possible explanation is that male-headed households tend to be more advantaged in various aspects, including access to resources and opportunities.This advantage may extend to digital resources, such as internet access and devices, which could contribute to the higher percentage of males using the internet.It is important to note that the difference in internet usage between genders, although statistically significant, is not substantial.This suggests that both males and females have relatively high levels of internet usage, indicating a growing trend of digital connectivity among households regardless of gender.
However, it is essential to explore the underlying factors that contribute to the gender disparity in internet usage.Potential factors could include differences in digital literacy levels, access to devices and infrastructure, and socio-cultural norms and expectations regarding internet usage.To address the gender gap in internet usage, it is crucial to promote digital literacy and provide equal access to resources and opportunities for both males and females.By fostering gender equality in digital connectivity, societies can ensure that everyone, regardless of their gender, can fully participate in the digital age and benefit from the opportunities it offers.

The binary logistic Regression result
This section focuses on the binary logistic regression results aimed at examining the factors influencing internet usage in the year 2021.The dependent variable used in the regression analysis was "internet usage," which was categorized into two groups: "yes" for households that reported using the internet and "no" for households that did not use the internet.Given the categorical nature of the dependent variable, a binary logistic regression model was deemed appropriate for the analysis.In this model, the "used internet" category was treated as the success category and assigned a code of 1, while the "not used internet" category was assigned a code of 0. This coding scheme allowed for a meaningful interpretation of the regression coefficients.
Table 7 presents the frequency distribution of the two categories, providing an overview of the distribution of households based on their internet usage status.The table displays the number of households falling into each category, allowing for a visual understanding of the distribution patterns.The binary logistic regression analysis conducted in this study aimed to identify the determinants or factors associated with internet usage among households in 2021.By examining the relationship between the dependent variable (internet usage) and a set of independent variables (such as demographic characteristics, income, and household size), the regression model sought to provide insights into the factors that influence the likelihood of households using the internet.The results of the binary logistic regression analysis will be presented and discussed in subsequent sections, providing a comprehensive understanding of the determinants of internet usage and shedding light on the factors that contribute to the digital divide among households in the study population.As discussed in the methodology the regression of four independent variables was employed, and the results show that all independent variables were statically significant predictors of the model all with a p-value of 0.00.The first independent variable examined in the binary logistic regression analysis was income, which was transformed into log income and treated as a continuous variable.The results indicated that income was a statistically significant predictor of internet usage, with a p-value of .000.This suggests that income significantly influenced the probability of a household using the internet.The coefficient for income was positive at 0.524, indicating that as income increased, the likelihood of using the internet also increased.The odds ratio of 0.592 further supports this relationship, implying that households with higher incomes had greater odds of using the internet compared to those with lower incomes.These findings align with common sense, as households with higher incomes generally have more disposable income to allocate towards purchasing internet bundles and accessing online services.However, addressing the disparities in internet usage between the poor and the rich is crucial for promoting digital inclusion and reducing social inequities.By closing the gap in internet usage between different socioeconomic groups, we can foster greater equality of opportunity, empower marginalized communities, and enhance social and economic development.Access to the internet and digital resources is increasingly vital for individuals to fully participate in the modern world, and ensuring equitable access is an essential step toward building a more inclusive and fair society.This income-related digital divide is not unique to the study population and can be observed at a global level, where developed countries typically provide more free internet access points, making it easier for low-income individuals to access the internet.
The second independent variable examined in the regression analysis was gender, which was treated as a categorical variable with males coded as 1 and females coded as 0. The variable was found to be statistically significant, with a p-value of 0.000, indicating that gender played a role in predicting internet usage during the Covid-19 pandemic.The coefficient for gender was 0.167, and the odds ratio was 1.181.Since males were coded as 1 and females as 0, the positive coefficient suggests that male-headed households were more likely to use the internet compared to their female counterparts.The odds ratio of 1.181 can be interpreted as males having approximately 1.181 times the odds of using the internet compared to females.These findings are consistent with the patterns observed in the cross-tabulation analysis between gender and internet usage.Gender disparities in internet access and usage have been well-documented, and this study's results align with previous research.It is noteworthy that Ren and Kwan (2009) found that the impact of internet use on individuals' behaviors, such as travel habits, varied significantly based on gender.Their study highlighted that men's physical activities and travel were more influenced by internet use for leisure purposes, while women's physical activities and travel were more influenced by internet use for maintenance purposes.Overall, the gender disparity in internet usage does raise concerns about women potentially falling behind in technological advancement.Access to and effective use of technology, including the internet, are crucial for participating in the digital economy and benefiting from the opportunities it offers.When women have limited access to digital resources and are less engaged in the online world, they may face challenges in acquiring digital skills, accessing educational and employment opportunities, and leveraging technology for personal and professional growth.Efforts to bridge the gender digital divide should focus on promoting digital literacy and skills training for women, providing affordable and accessible internet connectivity, and fostering inclusive digital environments.By empowering women with the necessary tools, knowledge, and support, they can actively engage in technological advancements, contribute their unique perspectives and talents, and help shape the future of technology.

Age and household size
Another variable included in the binary logistic regression analysis was the age of the household head.This variable was considered as a continuous variable and was found to be a statistically significant predictor of internet usage with a p-value of 0.000.The coefficient for age was -0.027, indicating a negative relationship between age and the probability of using the internet.The odds ratio of 0.974 further supports this relationship, suggesting that as the age of the household head increased, the likelihood of internet usage decreased.These results can be attributed to the differences in internet usage patterns between different age groups.It is commonly observed that younger individuals, who are more technologically savvy and have grown up in the digital age, are more active internet users compared to older generations.The younger generation often spends more time on social media, online games, and various other online platforms, while older individuals may not have the same level of interest or familiarity with internet activities.This finding aligns with the research conducted by Mhlanga and Beneke, who also concluded that the age of the head of the family influences the demand for internet services, with older age having a negative impact on internet demand.The reduced interest and lower income streams of older individuals may contribute to their decreased desire for internet access.
Regarding household size, the results were found to be statistically significant at a 1% level, indicating that it is a good predictor in the regression model.The variable showed a positive coefficient of 0.117 and an odds ratio of 1.124.This implies that as the number of people residing in a household increase, the probability of internet usage in that household also increases.This finding can be explained by the fact that larger households tend to have more individuals who require internet access for various purposes.With more people in the household, there is a greater demand for data and the need for multiple devices, leading to increased internet usage.This trend is particularly evident in households with older children and even younger children, as the internet has become a necessity for online activities, especially following the COVID-19 pandemic, where many aspects of life have shifted to online platforms.
In summary, the binary logistic regression analysis provided insights into the impact of age and household size on internet usage.The results indicated that older age was associated with a decreased probability of internet usage, likely due to differences in technological literacy and interests across age groups.On the other hand, larger household size was positively correlated with internet usage, reflecting the increased demand for internet access within households with more residents.These findings contribute to our understanding of the factors influencing internet usage and highlight the importance of considering demographic variables when examining patterns of digital inclusion and access.

Conclusions
The study's main objective was to access and further analyze the determinants of the usage of the internet before Covid 19 pandemic and during the pandemic.As discussed in the literature Covid-19 pandemic has impacted our daily livelihood in so many ways both positively and negatively.From the positive side, we have seen the progression of advancement in technology where now most countries who were still behind in terms of technological advancement have been forced to adopt some of the 4 th industrial revolution changes.Such changes have caused many people to adapt to the usage of the internet as a way of conducting business, schooling and so many economic activities in their daily lives.This study, therefore, used the general household survey for 2019 which was before the pandemic and that of 2021 which is after the pandemic to achieve the main objective of the study.The study first analyzed the changes in the usage of the internet over the years provided the results showed that there was a significant change in the usage of the internet, the most prominent usage was that of using the mobile phone as a device.The study expanded its analysis by utilizing the more recent 2021 data to delve deeper into the topic.The initial focus was on examining the demographics of the households and their relationship to internet usage.Specifically, cross-tabulation was conducted to explore how internet usage correlated with other demographic factors.One significant finding from the analysis was the association between income and internet usage.The results revealed that individuals with higher incomes tended to use the internet more frequently compared to those with lower incomes.This suggests that there is a positive relationship between income level and internet usage, indicating that individuals with greater financial resources are more likely to have access to and utilize the internet.The observed pattern aligns with the notion that economic factors play a significant role in determining internet usage.Higher-income individuals may have greater access to reliable internet services, possess the means to afford internet subscriptions or data packages, and have access to devices such as smartphones, laptops, or computers that facilitate internet usage.In contrast, individuals with lower incomes may face financial constraints that limit their ability to afford internet services or acquire necessary devices.The findings highlight the potential for digital inequality based on income disparities.The fact that individuals with higher incomes are more likely to use the internet implies that those with limited financial resources may face barriers in accessing the opportunities and benefits associated with internet usage.This underscores the importance of addressing the digital divide and ensuring that internet access becomes more equitable and affordable for all segments of the population.Finally, the study looked at the determinants of usage of the internet at the household level.To achieve this a binary logistic regression model was used having usage of the internet as a dependent variable and income, age, gender, and household size as independent variables.
The study findings on gender reveals that males had a slightly higher rate of internet usage compared to females.This highlights the existence of gender disparities in accessing and utilizing digital technologies.These disparities may arise from various factors, including socio-cultural norms, educational opportunities, and access to resources.To promote digital equity, it is essential to address the underlying gender inequalities and provide equal opportunities for women and girls to access and benefit from the internet.The findings revealing differences in internet usage between the poor and the rich have significant implications for digital equity and social inequality.The study's results suggest that individuals with higher income levels have a greater probability of using the internet compared to those with lower incomes.This disparity in internet usage based on socioeconomic status can further deepen existing social inequalities.Limited access to the internet among economically disadvantaged individuals can create a "digital divide" that perpetuates disparities in education, employment opportunities, and access to information and resources.The internet has become an essential tool for communication, education, job searching, accessing government services, and participating in the digital economy.Those who lack internet access or have limited usage are at a disadvantage in terms of accessing these opportunities and resources, which can hinder social and economic mobility.
On household age, the study highlights a negative relationship between age and internet usage, indicating that as individuals grow older, their likelihood of using the internet decreases.This finding aligns with the notion that younger generations tend to be more digitally engaged and have higher rates of internet adoption.To address this age-based disparity, efforts should focus on promoting digital literacy and providing targeted support for older adults to enhance their digital skills and confidence.On Population Group, the study reveals disparities in internet usage across different population groups.Black African households had the highest rate of internet usage, followed by coloured, white, and Indian households.This suggests that certain population groups may face specific barriers to internet access and usage.Efforts should be made to understand and address these barriers, taking into account the socioeconomic, cultural, and historical contexts of each group, to ensure equitable digital inclusion for all.
Overall, the implications of these findings highlight the need for comprehensive and inclusive strategies to bridge the digital divide.The study recommends that policymakers, governments, and relevant stakeholders should work together to address the underlying factors contributing to these disparities, such as socio-economic inequalities, gender biases, age-related challenges, and historical disadvantages.By promoting digital literacy, improving affordability, and expanding access to digital infrastructure, we can create a more inclusive digital society that benefits individuals of all genders, income levels, ages, and population groups.Such efforts are crucial for fostering social cohesion, economic development, and ensuring that no one is left behind in the digital era.

Figure 3
Figure 3:Population group of household head Figure3illustrates the distribution of household heads across different population groups.The results indicate that the majority of household heads in South Africa belong to the black African population group, accounting for 86% of the total.The second largest population group among household heads is coloured, representing 6.6% of the sample.The white population group follows closely, comprising 5.5% of household heads.Lastly, the Indian population group constitutes the smallest percentage, with 1.6% of household heads.These findings highlight the demographic diversity within South African households and provide insights into the representation of different population groups in household leadership roles.The overrepresentation of black African household heads suggests the significance of this population group in shaping and influencing household dynamics, decision-making processes, and socio-economic outcomes.It also reflects the country's historical context and the ongoing efforts towards achieving social cohesion and addressing historical inequalities.Understanding the distribution of household heads across population groups is vital for policymakers, researchers, and organizations working towards promoting inclusivity, addressing socio-economic disparities, and ensuring equitable opportunities for all population groups.These findings contribute to a more comprehensive understanding of the demographic landscape and inform targeted interventions and policies aimed at fostering social cohesion and equality in South Africa.

Table 1 :
Usage of the internet in 2019

Table 1
presents the descriptive findings on internet usage across different platforms in 2019.

Table 2 :
Access to the internet in 2021Table2indicates the results of Internet access and demand in South Africa using different platforms this time using data from 2021.As indicated, the highest internet access was through mobile devices, 66% of the population indicated to have been able to access the internet using mobile devices and only 34% indicated not to have been able to access the internet through this platform.Comparing results from table 1 of internet access through mobile devices collected before the Covid-19 pandemic with table 2 of the same data collected in 2021 after the Covid-19 pandemic, the results indicate a significant increase between the data sets.A 15% (from 51% in 2019 to 66% in 2021

Table 3 :
Descriptive results of continuous variables on the usage of the internet in 2021

Table 4 :
Cross-tabulation of internet usage and Household monthly total income

Table 5 :
Cross-tabulation results between demand for internet and group household head

Table 6 :
Cross tabulation between gender and usage of the internet

Table 6
displays the cross-tabulation results examining the relationship between the gender of the household head and internet usage.

Table 7 :
frequency of distribution of internet usage Table8presents regression results of the regression model accessing the determinants of usage of the internet in South Africa data collected in 2020 which was after the Covid-19 pandemic.

Table 8 :
regression results Sex of household head, age of household head, household size.