EXAMINING THE RELATIONS BETWEEN DIGITAL DIVIDES AND INFORMATION ABOUT COVID-19 AMONG SOUTH KOREAN FARMERS AND FISHERS By Angie Yeonjoo Nam A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Media and Information – Master of Arts 2023 ABSTRACT The COVID-19 pandemic sharpened our awareness of digital divides. Information related to COVID-19 was crucial for maintaining daily lives during the pandemic. South Korea is a country with a unique technological context of nearly ubiquitous Internet and digital device access. However, marginalized groups with limited digital skills and uses also exist in South Korea. Farmers and fishers are one of these groups, but they have not received the same level of attention from researchers as other marginalized subgroups. This thesis aims to overview and examine indicators of the digital usage divide regarding COVID-19-related information, specifically on stimulus check requesting services and information services among South Korean farmers and fishers. It is an explanatory study to develop a first level of understanding of the digital usage divide in a highly connected society. The study uses data from the 2021 Digital Divide Report released by the Korean National Information Society Agency (NIA). Resources and appropriation theory was used as a theoretical framework as it considers a wide range of factors influencing digital divides. Given the explanatory nature of the study, empirical work used multivariate regression analyses to discern patterns in the data. The work examined the associations between sociodemographic factors, such as income level and education level, attitude toward digital technology, and digital skills and the usage of COVID-19-related services. Notably, there was no association between digital material access and the use of COVID-19- related services. This study contributes the understanding digital inequalities in the unique environment of a nation that has achieved high digital connectivity. It also provides initial insights that can help in the design of more effective policies to reduce digital divides among South Korean farmers and fishers. Copyright by ANGIE YEONJOO NAM 2023 TABLE OF CONTENTS 1. INTRODUCTION ...................................................................................................................... 1 2. LITERATURE REVIEW ........................................................................................................... 4 3. RESEARCH QUESTIONS AND METHODS ........................................................................ 16 4. RESULTS ................................................................................................................................. 23 5. DISCUSSION ........................................................................................................................... 58 6. CONCLUSION ......................................................................................................................... 66 BIBLIOGRAPHY ......................................................................................................................... 68 APPENDIX A: CODE BOOK OF VARIABLES THAT USED IN THIS STUDY ................... 76 APPENDIX B: EACH RESULT OF MULTUPLE SIMPLE REGRESSION ANALYSES FOR USAGE COVID-19 RELATED STIMULUS CHECK REQUESTING SERVICES ................. 82 APPENDIX C: EACH RESULT OF MULTUPLE SIMPLE REGRESSION ANALYSES FOR USAGE COVID-19 RELATED INFORMATION SERVICES .................................................. 87 iv 1. INTRODUCTION In high-income countries, most people use the Internet and digital devices daily. From simple information searching to streaming live orchestra concerts, Internet access and digital devices, such as smartphones, laptops, and tablets, are embedded in life. They are used to interact socially, for entertainment, for education, and to obtain information about political, cultural, and economic matters. Statista (2023) estimates that 64.4% of the world’s population used the Internet in January 2023. For users with adequate digital skills, using the Internet allows finding and generating helpful information and sharing it with others (DiMaggio et al., 2001; Pearce & Rice, 2013; Van Deursen & Van Dijk, 2015). However, because of the existence of digital divides, not everyone can benefit from the Internet equally. Digital divides are gaps in access, uses, and skills needed to use the Internet and obtain proper information on the Internet. These discrepancies are related to factors such as spatial, demographic, and socioeconomic attributes. They are typically also related to social capital (DiMaggio et al., 2001; Selwyn, 2004; Van Dijk, 2005). The existence of digital divides was first noticed in the 1990s. During the first stages of growing awareness, most researchers and policymakers focused on the access divide. A widely held assumption was that such access divides would be resolved as adoption rates of the Internet approximated saturation levels (Van Deurson & Van Dijk, 2019). However, researchers noticed that other forms of divides (or inequalities) arose, including discrepancies in digital usage and outcomes. These divides have appeared even though the access gap has been narrowed and the adoption of digital services and devices has increased (Hargittai, 2002; Hargittai & Hinnant, 2008). Moreover, researchers theorized that these second and third level divides could potentially widen within or between groups (Norris, 2001; Van Deursen et al., 2017; Van Deursen & Van Dijk, 2019; Van Dijk, 2013). The COVID pandemic brought the severity of digital divides to the attention of individuals and policymakers. People realized that digital communication and networks had become necessities, essential goods and services required to participate in society. Social distancing policies required tremendous changes in work practices. Workers were asked to perform their work remotely, and students were to take their classes virtually. Scheduling of government and health appointments through websites and apps replaced walk-ins. People increasingly order their food in restaurants and cafes from a kiosk, not in person. Most daily 1 activities now require internet connection, digital devices, and digital skills. However, in many countries a noticeable share of the population does not own digital devices, such as smartphones, laptops, and home wireless networks, and more do not know how to use them effectively (Van Dijk, 2020). Consequently, people who cannot access and use digital devices and services properly were unable to keep up with their daily lives during the pandemic. Marginalized groups with limited digital skills and uses also exist in South Korea, a country known for its high national level of broadband Internet access. According to the Korea Communications Commission (2022), 99.7% of Korean households have internet access at home, and 93.4% of Korean people have smartphones. In other words, most Korean people have access to the Internet at their home and have technological means that suggest digital access is at or near saturation level. Nonetheless, digital gaps exist between most users and marginalized communities, such as older people, disabled, low-income, farmers and fishers, marriage immigrants, and North Korean defectors. The discrepancies experienced by these groups are the subject of the annual digital divide report conducted by Korea National Information Society Agency (NIA), for Internet and mobile use and digital skills. Thus, South Korea allows conducting research on digital usage and outcome divides within a highly connected society. Among these marginalized groups, fewer studies document farmers and fishers compared to older people and disabled people. The Korean government has made efforts to narrow the digital divide affecting farmers and fishers since the 1990s (Kim & Sung, 2020). Park & Nam (2015) insisted that the access gap has been narrowed significantly since the government conducted an initiative project named the ‘Information Network Village Program,’ distributing personal computers and proving broadband internet access, in 2001 (Park, 2015). Nevertheless, the digitalization level developed and measured by NIA of farmers and fishers is 78.1 compared to the entire population of 100, according to the 2021 Digital Divide Report (NIA, 2021). Several studies regarding the digital divide for farmers and fishers have been conducted in South Korea (Lee & Kim, 1997; Kim & Sung, 2020), the United States (Strover, 2001), and Britain (Townsend et al., 2013), but most prior studies are mainly focused on the gap between rural and cities, the location farmers and fishers live, not farmers and fishers themselves (Kim & Sung, 2020). Kim & Sung (2020) studied the factors that affect the digital divide among farmers and fishers associated with attitudes toward technology, physical access, and digital skills, but the study only focused on mobile internet use. 2 After the COVID-19 outbreak, digital skills and uses became crucial in the context of obtaining information. COVID-19-related information is a critical resource for people to continue their lives, especially if it is related to health care and financial support. Therefore, this study aims to explore the connection between usage and digital skills among farmers and fishers and the obtaining of COVID-19-related information. Specifically, the study will generate empirical evidence for South Korean farmers and fishers, a minority facing serious digital divide challenges that is currently overlooked by digital divide policies in South Korea. This will allow insights into which measures might exist to help mitigate the observed forms of social exclusion due to the digital divide. This study is conceptualized as a first explanatory examination of the data, not a confirmatory analysis. It seeks to discern first patterns of the nature and dimensions of digital divides in a highly connected country, South Korea.The thesis is organized as follows. The next section provides an overview of the research on digital divides. Section three demonstrates research questions and explains the data set and variables used in this study. In section four, the results of the analyses are articulated. Lastly, main insights from the findings and suggestions for future studies are presented in the discussion part. 3 2. LITERATURE REVIEW 2.1. Access Divides (First Level Digital Divides) Awareness of the importance and pervasive presence of digital divides has grown since the mid-1990s. Scholars assert that the origin of the term ‘digital divide’ is uncertain (Gunkel, 2003). However, most researchers believe its use spread after the U.S. National Telecommunications and Information Administration (NTIA) used it in the ‘Falling Through the Net: Defining the Digital Divide.’ report (NTIA, 1999). This report defined the digital divide as the discrepancy between those with access to the latest information and communication technologies and those without. The definition focuses on access (“first level” digital divide). During the initial stages of internet diffusion, in the 1980s and 1990s, communication scholars considered the possession of hardware as the primary determinant of the digital divide (e.g., Van Dijk, 1999). In his book, Van Dijk distinguished four kinds of barriers to access, namely, lack of elementary digital experience (mental access), no possession of computers and network connections (material access), lack of digital skills (skill access), and lack of significant usage opportunities (usage access). At the initial stage of diffusion of the Internet and digital technology, focusing specifically on material access among the above four barriers, which is at the center of public opinion and public policy, many decision-makers expect the problem of the digital divide and information inequality can be solved when everyone has access to a personal computer and connection to the Internet (Van Dijk & Hacker, 2003). Thus, many countries continue to focus their digital divide and information inequality policy on building an infrastructure for universal access (Dewan & Riggins, 2005) to improve access to digital technologies (Ahn, 2006; Van Dijk & Hacker, 2003). However, awareness of the importance of skills and uses is now widespread. Early research indicated that individual factors affect the level of digital access, which is based on so-called methodological individualism suggested by Wellman and Berkowitz (1988) (Van Dijk, 2013). Sociodemographic factors, such as income level, gender, age, education, and race, are the main subjects of digital access studies. In addition, researchers identified individual characteristics like family, friends, colleagues, and social networks as crucial factors affecting digital access (Ahn, 2006). Male users, highly educated and earning a high income, are more likely to have digital access than other users based on multiple digital divide studies (Van Dijk, 2020). Sociodemographic factors identified in earlier studies are used as a predictor in studies of 4 digital access divides. They are also considered in the design of policies intended to improve access of digitally marginalized groups. These studies reveal that the digital divide is a multidimensional phenomenon (Ahn, 2006; Ferro et al., 2011; Gunkel, 2003; Newhagen & Bucy, 2004; Selwyn, 2004). Only increasing access to digital devices and technologies will be insufficient to overcome digital inequalities. Scholars discuss the more advanced of recent digital devices as one of the factors resulting in digital inequalities because higher-level skills are necessary to use them. Digital devices, such as tablet PCs and smartphones, are devices that combine portable telephone functions and the Internet. Users, therefore, can use their devices in diverse ways depending on the types of applications they install and their digital usage skills level (Keum & Cho, 2010). Consequently, higher-level digital divides could arise due to a mismatch of between the power of digital devices and an individuals’ medium-related and usage skills. For instance, in 2020, 94.4% of Korean people could access digital technologies and the internet with their own devices, but only 63.8% of them knew how to utilize the devices and technologies properly (National Information Society Agency of Korea, 2021). In other words, about one-third of the people who owned digital devices could not utilize their full benefits. This empirical case shows that the digital divide does not result only from the lack of devices or internet access. Socioeconomic inequality interacts with other factors to generate diverse digital divide in access, usage, and outcomes. This often exacerbates prevailing social inequalities (Ahn, 2006; Lee & Youk, 2014; Van Dijk, 2005). 2.2. Second and Third Level Digital Divides The notion of digital usage divides, also referred to as the second level of the digital divide, is inspired by the knowledge gap hypothesis, which was suggested in 1970 by Tichenor, Donohue, and Olien (1970) (Van Dijk, 2020). This hypothesis was formulated in the context of mass media, such as television and newspapers. It states that the access to such media varies, by demographic factors and socioeconomic status. These variations have repercussions for uses and outcomes. This concept was applied to digital divides (Van Dijk, 2020). In that context, it means that the level of digital usage will likely be different depending on the user’s demographic characteristics, socioeconomic status, social networks, employment status, or digital skills. Digital use is not limited to computers and information behavior but includes various digital devices, such as smartphones, smartwatches, and tablets that are used in daily lives. Therefore, the effects of digital usage gaps are a multi-phenomenon as well (Van Dijk, 2020). 5 Blank & Groselj (2014) suggested categorizing digital usage gaps into three categories, each measured by an index: amount, variety, and type. According to the authors, the amount is the actual time and frequency of Internet use, variety is the number of activities that users can do on the Internet, and types mean how users actively or passively use the Internet. These indicators are often employed in digital usage-related studies (Van Dijk, 2020). The NIA also uses one of them, a variety of digital use, as items for measuring digital usage scores in South Korea in its annual empirical survey. Studies have examined the usage gap of digital technologies (Ahn, 2006; Dewan & Riggins, 2005; Hargittai, 2002; Hargittai & Hinnant, 2008; Hargittai & Walejko, 2008; Lee & Youk, 2014; Van Dijk & Hacker, 2003). Prior studies have identified factors, including digital access level, which affect the digital usage divide. For instance, the cost of digital devices and the price of monthly Internet service can be a burden to low-income people. Consequently, they may not be able to use the Internet and individual digital devices consistently even though they have digital access (Ahn, 2006). Or some people do not want to be connected because they have negative thoughts regarding being connected and assume using the internet and digital devices might result in potential harm. These kinds of people choose not to use digital technologies even though they are aware of the convenience of using them. And their choice of non-use also could lead to differentiation in digital skills and usage among people regardless of their digital access status. Also, the internet does not appeal to low-income and low-educated people (Katz & Rice, 2022). Based on this phenomenon, Van Dijk insisted that the motivations to get access also must be considered when researchers address the digital divide (2013). This shows why in developed countries that have already reached a high access level to digital technologies, the emphasis has shifted to a second level of digital divides, known as the usage gap. As some empirical studies partially demonstrated above, studies of the second-level digital divide also show that age, gender, income, and educational level, are associated with the usage gap in digital technologies as well (Ahn, 2006; Dewan & Riggins, 2005; Hargittai, 2002; Hargittai & Hinnant, 2008; Hargittai & Walejko, 2008; Lee & Youk, 2014; Van Dijk & Hacker, 2003). These results support the assumption that existing social inequalities result in diverse types of digital divides and eventually deepen structural inequalities (Ahn, 2006; Lee & Youk, 2014; Van Dijk, 2005). Moreover, social network resources, for example family, friends, and colleagues, can help 6 overcome shortcomings of digital skills. Social support and relationships of users could impact the amount of Internet use (Selwyn, 2003). Recent research has again shifted emphasis. In addition to the second-level digital divide, a third-level digital divide was recognized. The second-level digital divide refers to the disparity of digital skills and utilization levels among people with access to digital devices and technologies (Lee & Youk, 2014). The notion of a third-level digital divide revealed the outcomes, the effects of using digital technologies on users, are not identical (Lee & Youk, 2014; Wei et al., 2011). For instance, researchers have detected educational outcome differences (Hampton et al., 2021; Wei et al., 2011) and life satisfaction differences (Hwang, 2019; Kim & Kwon, 2022; Koh, 2017) based on the digital skills that users have. Considering that the digital access divide and usage divide affect social inequality, it seems plausible that the third-level digital divide, the digital outcomes divide, might intensify existing inequalities as well. Nevertheless, studies regarding this third-level digital divide are limited compared to the usage gap studies, and the interactions between these multiple levels of digital discrepancies are poorly understood. 2.3. Resources and Appropriation Theory Resources and appropriation theory is one of the theories that can help explain the multiple levels of digital divides in an integrative framework. Van Dijk suggested resources and appropriation theory in 2005. He asserted that the theories commonly used in digital divide studies could not fully cover the digital divide issue since the digital divide is a multidimensional problem. The digital divide could be approached from various perspectives as a theoretical framework: technology acceptance (attitude), motivation, capital for using technology, and sociocultural perspective (Van Dijk, 2020). For instance, the Technology acceptance model (TAM) (Davis, 1989) is usually applied to explain the factors of Internet diffusion, the attitude towards information technology, and from a digital access perspective (Han & Nam, 2021; Lu, 2001; Zhang, 2013). However, TAM only examines the process of technology acceptance and attitudes when people encounter innovative technologies and which factors improve the level of digital access. Thus, TAM has less to say about why and how the existing social inequalities affect the digital divide. Van Dijk included complementary perspectives based on the relational view of equality in the resources and appropriation theory (Van Dijk, 2013). The theory articulates four 7 successive kinds of access in the appropriation of digital technology: motivation, physical and material access, digital skills, and usage (Van Dijk, 2013). To accept innovative technology, people must have some motivation to use, acquire or consume technology, develop skills, and learn how to use technology (Van Dijk, 2013; 2020). In theory, the motivation not to get access considers both have nots and want nots, unlike only focusing on have nots at the earlier stage of digital divide studies. After each kind of factor meets, develops, or reaches a sufficient level, then the following factors could make and lead to an appropriation of digital technology. This theory has been supported empirically in the Netherlands and Germany to explain the digital divide (Van Deursen, & Van Dijk, 2019; Van Dijk, 2013). Figure 1a. Four successive kinds of access in the appropriation of digital technology (Van Dijk, 2013, p.34) The resources and appropriation framework also explains the factors shaping the digital divide and its consequences. It demonstrates how categorical inequalities in society produce an unequal distribution of resources and eventually lead to unequal access to digital technology and social exclusion, such as labor market, education, politics, cultures, and health provisions (Sala et al., 2022). The theory argues that social and structural inequalities cause digital access and use disparities, which, in turn, exacerbate socioeconomic inequalities (Fang et al., 2019; Sala et al.,2022). Therefore, the framework can reflect multiple dimensions of the digital divide and explain more than other commonly used theories or frameworks in digital divide studies. 8 Figure 1b. A causal model of resources and appropriation theory (Van Dijk, 2013, p.34) A systematic literature review of 55 digital divide-related studies based on resources and appropriation theory (Fang et al., 2019) demonstrates its power to help explores a wide range of factors influencing digital divides: sociodemographic determinants of ICT adoption and use (context), ICT resources/motivations/skills (mechanisms), ICT disparities across social interaction (outcomes), and organizing principles that try to mitigate ICT access and use challenges. In addition, resources, motivation, and skills are identified as three primary mechanisms of the digital divide. Because of its generality, the resources and appropriation theory has become the preeminent theoretical framework for examining the importance of recognizing and responding to the multiple layers of digital access and use inequalities (Fang et al., 2019; Van Dijk, 2020). This study builds on the resources and appropriation framework to examine the association between sociodemographic factors as well as the use and ICT resources/skills, and the digital divide among South Korean farmers and fishers. Furthermore, resources and appropriation theory pays only limited attention to changes in external factors, such as changes in the environment of technology use. The theory primarily focuses on factors that individual users have, as mentioned above. That is, there is scarce empirical evidence regarding how external changing factors, in this case, the outbreak of COVID-19, would affect digital divides within the resources and appropriation framework. 2.4. Digital Divides and Health Specifically, the digital divide in health-related uses has become critical after the COVID-19 outbreak, even though the repercussions in health-related uses appeared years ago. The Internet is one of the potentially valuable tools for getting health information, care, and services (Rains, 2008). Health information is exceedingly popular among the wealth of 9 information sources that are accessible on the Internet, used by a large share of the Internet population. According to Wyatt et al.’s (2005) study, more people go online for health information than consult a doctor. In other words, people nowadays quickly and easily get health-related information from the Internet. The potential consequences of inequalities in access to broadband Internet or digital devices may therefore be substantial (Rains, 2008). Researchers examined the relationship between the digital divide and health information, communication, and literacy (Mackert et al., 2016; Neter & Brainin, 2012; Wyatt, 2005). These studies primarily focus on older people (Hong et al., 2017; Mackert et al., 2016; Mitchell et al., 2019; Levy et al., 2015; Rains, 2008) since obtaining and utilizing health-related information is more important to older people. Moreover, older adults are often not as familiar with the Internet and digital devices as younger people. Prior studies show that health disparities also relate to the level of digital access. The Internet provides opportunities for improving access to health information and thus is a tool that can contribute to improved health (Mackert et al., 2009; 2016; Mitchell et al., 2019). A prerequisite of obtaining health information on the Internet is having access to network services and devices. Most of the previous research (Hong et al., 2017; Mackert et al., 2016; Mitchell et al., 2019; Neter & Brainin, 2012; Levy et al., 2015; Rains, 2008; Wyatt, 2005) focused on digital access divides. They identified demographics, socioeconomic status, education level, race, ethnicity, and age as associated with health disparities (López et al., 2011; Rains, 2008; Wyatt, 2005). Researchers also recognized the importance of digital and health literacy for reducing health disparities (Hong et al., 2017; Mackert et al., 2016; Neter & Brainin, 2012). Unlike the general digital divide, the health-related digital divide has a unique aspect, health literacy. According to Berkman et al. (2010), health literacy refers to how people obtain, understand, use, and communicate health information to make informed decisions. Access to the Internet and knowledge of how to find necessary health information needs to be complemented by health literacy. Someone with low health literacy may not understand and probably not make best use of the obtained information. A vicious cycle of low health literacy, lower Internet use to search for health information, and worse health practices may exacerbate the digital divide (Mackert et al., 2016). Thus, health literacy is one of the crucial variables that must be considered when examining the digital divide regarding health. 10 As technology develops, there are other ways to get health-related information not on the Internet, such as fitness applications, nutrition applications, and activity trackers. Moreover, some countries that achieved a certain level of digital access, such as South Korea, Japan, and the United States, provide public health information through their official governmental department social network channel to improve health (Lee, 2014). According to a study by Lee (2014), other factors, such as perceived usefulness, perceived ease of use, and digital literacy play a critical role and interact with conventional determinants, such as socioeconomic status, for the digital divide. 2.5. Digital Divides in South Korea South Korea is known for having a high Internet adoption rate and one of the fastest Internet networks around the world (Jobst, 2022). According to a Korea Communication Commission (2022) report, in 2021 99.7% of the Korean population had Internet access at home, and 93.4% of the population had a smartphone. That is, most Korean people are connected to the Internet and have technological means. Not only has the penetration rate of 4G networks almost reached 100 percent, but 5G wireless services have accumulated more than 20 million subscribers at the end of 2021 (Jobst, 2022). These numbers suggest that the adoption and expansion of innovative digital technology in South Korea is faster than in any other country. GDP per capita of South Korea in 2021 was 1.81 trillion US dollars. According to the World Bank (2023), income per capita increased has steadily since 2009. The country also boasts a high university enrollment rate. In 2022, 73.8% of eligible students at South Korean universities enrolled universities (Statista, 2023). These socioeconomic metrics in addition to high connectivity data suggest that South Koreans can afford adoption and utilize the Internet and other information and communication technology. At the same time, South Korean society is aging. People older than 65 years old accounted for 17.5% of the population in 2022 (Statista, 2022). Those older than 60 years accounted for 26.4% in 2022, according to data by the Korean Ministry of the Interior and Safety (2023). That is, the digital divide is likely to widen by an aging population (Han et al., 2022). One of the drivers of the high quality and fast expansion of the Internet in South Korea was government effort. The Korean government took a vital role in the development of the expansion of the Internet, and its influence has not been declining since 1987 (Jobst, 2022; Lee, 2017). To overcome the digital divide, the Korean government established digital divide-related 11 law in 2001 and the Korea National Information Society Agency (NIA) in 2003 as an agency dedicated to help overcome the digital divide (Ko et al., 2021). After the establishment of NIA, the Korean government tried to narrow the divide by establishing broadband internet service in rural areas and providing informatization education (Park, 2015; Ko et al., 2021). Even though digital access in South Korea is nearly ubiquitous, digital divides and inequalities in usage and outcomes persist. Since 2002, the Korea National Information Society Agency (NIA) has conducted annual research to examine the status of the digital divide and to assess the impact of the government’s digital divide policies, especially for marginalized groups. According to the report on the digital divide in 2021 (NIA, 2022), 94.4% of the whole Korean population had digital access. Only 63.8% of those with access had digital skills, such as sending digital files or making digital content on PC or mobile, and 77.6% had experience of digital usage, including activities of social participation, networking, sharing, and creating digital content. Compared to the full population, the level of digital access, skills, and usage was much lower within marginalized communities, such as older people, disabled people, low-income, and farmers and fishers. The index developed, measured, and published by NIA consisted of three weighted components-20% of digital access level, 40% of digital skills, and 40% of digital usage-has been consistently increasing since 2002, when the first annual report was published. Since South Korea has already reached nearly universal physical and material access, usage and outcome divides may be more relevant than in other developed countries. In that sense, South Korea is a good place to examine the existence of digital outcome divides and their determinants. South Korea has also recognized the necessity for studying the digital usage divide and outcome divide (Lee, 2013). Several studies examine the digital usage divide. Lee & Youk (2014) conducted a study to find factors that impact the digital usage and outcome divide in South Korea. The authors defined and offered general variables for exploring the second-level digital divide: information use skills, content production skills, and social association skills. Many studies (Baek et al., 2022; Kim & Lee, 2018; Kim & Lee, 2021; Yim et al., 2021) used these variables when examining the digital usage divide. Furthermore, the annual NIA digital divide report has stimulated digital divide studies in South Korea. 2.5.1. Digital Divides among South Korean Farmers and Fishers Farmers and fishers in South Korea are digitally marginalized groups and were one of the targeted subjects of the digital divide policy in South Korea (NIA, 2021). According to NIA 12 (2021), farmers and fishers in South Korea refer to people who are working in the agricultural or fishing business and are over 15 years old. Farmers and fishers are categorized as one of the targeted subject groups for several reasons. First, most farmers and fishers live in rural areas. Despite governmental efforts to offer Internet networks an initiative named the ‘Information Network Village Program’ (Park, 2015), differences in network infrastructure between rural and cities exist. Second, farmers and fishers are experiencing aging and a population decline. 68.1% of farmers and fishers are older than 60 in the NIA (2021) sample data. This implies that fewer young people have digital skills or give help to people with a low level of digital skills than in other areas. Lastly, farmers and fishers are a small group compared to the entire population. Only 2.2 % of the full population in South Korea are farmers and fishers, according to the Statistics Korea census 2021 data. Studies of the digital divide among farmers and fishers are scarce in South Korea compared to other subgroups and studies between rural areas and cities (Ko et al., 2021). Factors influencing digital divides have been studied in other contexts. However, usage and outcome divides among South Korean farmers and fishers remain to be studied. Moreover, farmers and fishers are the subgroups that have the second lowest digital use and skills among the subjects of NIA’s digital divide report data. Digital informatization education provided by the Korean government, which aims to improve digital literacy and digital skills, is mainly focused on older and disabled people. South Korea’s budget for digital information education has decreased since 2016 (Kim, 2020) after the adoption (digital access level) reached a certain level. The remaining digital divide policies mainly focus on other social groups, currently bypassing farmers and fishers. Also, farmers and fishers with low digital skills and usage might experience social exclusion and even violate human rights as much information, participation, and access to vital public services can be achieved by digital technology (Han & Nam, 2021; Muchran & Ahmar, 2019; Schmidt-Hertha & Strobel-Dümer, 2014). Kim & Sung (2020) studied aspects of the digital usage divide among farmers and fishers focusing on mobile use and social media. The study identified that attitude, skills, physical access, age, and area of residence are associated with mobile use and social media. These factors could be associated with digital usage divides among farmers and fishers, but an empirical examination is missing. Therefore, studying the digital divide among Korean farmers and fishers offers a case study to deepen our understanding of the repercussions of digital divides for a 13 specific, marginalized population. It will also contribute to a better understanding of the role of policies in reducing digital divides among such populations. 2.5.2. Digital Divide and COVID-19 Related Information The unprecedented COVID-19 pandemic revealed digital vulnerabilities of people with low digital skills and usage since a contactless society has become the new normal. As resources and appropriation theory asserted, digital inequalities might affect existing social inequalities more severely under the current pandemic. Thus, NIA added COVID-19-related questions to the Digital Divide Report from 2020. NIA is particularly interested in changes regarding digital divide influenced by COVID-19 outbreak. The NIA measured subjective change that users self- reported regarding digital use in general and specific digital services after the COVID-19 outbreak: the amount of time for using digital devices and services, awareness of COVID-19- related services, the experience of COVID-19-related services, and reason for not using COVID- 19-related services even if they are aware of it. NIA collected information on several COVID-19-related services. Among the activities surveyed, services to request emergency funding assistance and information services about COVID-19 directly influence people's life. As other studies have shown, obtaining dependable COVID-19-related information has repercussions for people’s health care and financial support (Abdulai et al, 2021; Goel & Nelson, 2021; Park, 2022). For example, people must take a COVID-19 test and show a negative result before visiting a hospital at an early stage of the COVID-19 outbreak in South Korea. Information regarding testing site locations was provided online. If a person has no access to the Internet or digital devices or has low digital skills, they may not be able to find the testing site information. Consequently, he/she could visit the hospital late or not at all. In addition, in the initial stages of COVID-19, the Korean government published information about places that COVID-19-positive patients had visited to help other citizens monitor potential exposure and adopt preventive action every day. Again, people with low digital skills, who cannot access the information online, have fewer options to respond. Thus, the lack of health-related information due to the lack of digital access and skills could negatively affect people’s well-being. The Korean government also provided emergency funding assistance to Korean citizens, like stimulus checks in the United States due to the COVID pandemic. Funding could be requested in two ways: by visiting an online bank website or by setting up an in-person bank 14 appointment. If someone had not requested this funding assistance within three months, the money was automatically donated to the government. Besides the Korean version of the stimulus check, there were several other forms of emergency funding assistance for people with unstable employment status and marginalized groups during COVID-19 pandemic. Even though the Korean government offers an in-person, offline way of requesting funding, most of the requested services are processed through the Internet and digital devices including personal computers and smartphones. In other words, people who have less experience using digital technology might have had trouble accessing these kinds of services and could not get governmental financial support. And this could lead to another shape of exacerbating existing social inequalities. According to NIA data, the second-ranked reason marginalized people did not use these kinds of services was that they did not know how to utilize digital services (NIA, 2021). These are survey responses by marginalized people, but no research has investigated these matters. An empirical study examining the association between digital usage and the acquisition of COVID- 19-related information could help develop a better understanding of the role of digital divides in an overall highly connected country. Thus, this study aims to explore which factors are associated with usage of COVID-19 related services, limitedly on stimulus check request services and information services. 15 3. RESEARCH QUESTIONS AND METHODS As explained in the introduction, this study aims to conduct a first explanatory analysis of multidimensional digital divides in a unique context, an exceptionally highly connected society. The study explores which indicators are related to the use of COVID-19-related services and, specifically, the digital usage divide among South Korean farmers and fishers. The research questions in this study are as follows. 3.1. Research Questions RQ1: Among South Korean farmers and fishers, are sociodemographic factors, digital access, skills, and activities associated with the use of digital services to obtain COVID-19- related information on stimulus check requesting services and information services? RQ1-1: Are sociodemographic factors (age, education level, income level, attitudes towards digital technology) of South Korean farmers and fishers associated with obtaining COVID-19-related information on stimulus check requesting services and information services? RQ1-2: Is the level of digital material access (internet, devices) of South Korean farmers and fishers associated with obtaining COVID-19-related information on stimulus check requesting services and information services? RQ1-3: Is the level of digital skills (medium-related, usage) of South Korean farmers and fishers associated with obtaining COVID-19-related information on stimulus check requesting services and information services? RQ1-4: Is the extent to which South Korean farmers and fishers use digital technologies and service associated with obtaining COVID-19-related information on stimulus check requesting services and information services? 3.2. Population and Sample The population of this study is individuals working in the agricultural or fishing industry aged 15 and over in South Korea. The sample is 2,200 Korean farmers and fishers and the survey was conducted from September to December 2021. There are 1,114 male (50.6%) and 1,086 female (49.4%) respondents in the sample. The average age of respondents in the sample was 62.98. 42.1 % of respondents answered that high school is their highest educational achievement, and 23.2 % answered that it was middle school. Regarding monthly income, 60.2% of respondents said their monthly income was less than 2.99 million won (₩, KRW) (approximately $1,523.54, USD). 16 Table 1 Demographic information of the sample Category Label Frequency Percentage (%) Male 1,114 50.6 Gender Female 1,086 49.4 10~19 4 0.2 20~29 23 1.0 30~39 68 3.1 Age 40~49 211 9.6 50~59 541 24.6 60~69 683 31.0 Above 70 670 30.5 Elementary school or 509 23.1 below Middle school 511 23.2 Education level High school 927 42.1 University/College or 253 11.5 above Less than 1 million 282 12.8 1~1.99million 473 21.5 2~2.99million 570 25.9 3~3.99million 456 20.7 Income level 4~4.99million 203 9.2 (Monthly, Korean 5~5.99million 104 4.7 won, ₩) 6~6.99million 71 3.2 7~7.99million 21 1.0 8~8.99million 8 0.4 9~9.99million 6 0.3 More than 10million 6 0.3 N 2200 Note: ₩ 1 million (KRW) = $ 763.53 (USD) (April 5 , 2023) th 3.3. Data The 2021 Digital Divide Report in South Korea was used as data for this study. It is a cross-sectional dataset conducted by the Korean National Information Society Agency (NIA) and the ICT department from September to December 2021. The Digital Divide Report is the annual survey conducted by NIA, using face-to-face interviews with structured survey questions. These surveys have been conducted every September to December since 2002. The survey is designed to measure the status of the digital divide in South Korea. 17 The survey subjects are Korean citizens over 7 years old, including older people, disabled people, low-income people, farmers and fishers, marriage immigrants, and North Korean defectors. The sample is stratified, with an overall size of 15,000. Of that total, 7,000 respondents are drawn from the population of Korean citizens at large. In specific, 2,300 respondents over 55 years old as the older people group, and 2,200 respondents for each group of disabled people, low-income, and farmers and fishers, respectively, were asked for the survey. And 700 respondents for marriage immigrants and North Korean defectors of each group were drawn for the survey. Respondents had to answer between 75 and 79 questions, depending on the groups they belonged to. The questionnaires cover various digital divide-related questions, including the level of digital access, use, skills, social support, social capital, attitude towards digital technology, self-efficacy for using digital devices, demand for digital literacy education, previous experience with governmental digital informatization education, and obtaining COVID- 19-related information. Demographic data, such as age, gender, race, occupation, education level, income, and area of residence of the respondents, were also collected. The raw data of the 2021 Digital Divide Report in South Korea for each marginalized group, older people, disabled people, low-income people, farmers and fishers, marriage immigrants, and North Korean defectors, was released at the end of September 2022 on the Open Data Portal (https://www.data.go.kr/) in South Korea. 3.4. Variables and Measurement NIA developed its digital divide measures that are most suitable to meet the specific goals of the South Korean government. The agency developed a measurement scale of digital accessibility, usage, and skills level for South Korea in 2016. Because some questions contained more than one variable, the raw dataset comprises 221 indicators in total. Purpose of the study, I selected and analyzed indicators that are relevant to the research questions. Also, several indicators were aggregated to form new variables that work compatible with the conceptual framework. Cronbach's alpha was calculated for each variable to check the internal reliability of the variables before conducting data analyses in this study. The new variables were mutually exclusive and can be considered in specific strata of measurement. Detailed variable measurements are presented in the codebook in Appendix A. 18 Usage of COVID-19-related services (dependent variable): My dependent variables are the use of information services and stimulus check requesting services. Each of the two types is measured. In the survey, responses to COVID-19 are measured using four groups of indicators. 1) Changes in the amount of time of use after the COVID-19 outbreak: device (a personal computer and a mobile device), digital activities (search and email activity, social media activity, everyday life information activity, information seeking and sharing activity, networking activity, social engagement activity, and consumption activity), 2) Awareness of COVID-19-related services: stimulus check requesting services, information services, groceries/food delivery services, and content subscription services 3) Experience of COVID-19-related services, and 4) Reason for not using of each COVID-19-related service even if they are aware of it. In this study, the awareness and experience of each COVID-19-related service were selected and analyzed as dependent variables among four categories. Among four COVID-19-related services, I chose stimulus check requesting services and information services because they are closely related to the respondents’ daily lives, specifically their health and financial resources. The survey only measured the experience of COVID-19-related services if the respondents were aware of them. To simplify the analyses the two measures of awareness and usage experience were combined into one new variable with three values: unaware, aware_not use, and aware_use. Figure 2a. Transformation of survey data into dependent variable Digital Access: Digital Access was divided into two categories in the survey: Possession of a wired/wireless digital device and Availability of a stable connection to the Internet. The respondents were asked whether they have a desktop, laptop, smartphone or feature phone, tablets, and other smart devices such as a smartwatch or AI speaker or not for digital device 19 possession (Yes/No). Also, they were asked whether a stable broadband internet connection was installed in their home. There were five different devices and internet access availability in digital access category. The value of each device of 0 referred to ‘Do not have’ and a value of 1 referred to ‘Have’. For the analysis in this study, two variables of possession of tablets and other smart devices were aggregated into one variable, possession of smart devices, to align with the conceptual framework (∝=0.542, M=0.49, SD=0.17). Digital Skills: Digital Skills was divided into two categories in the survey: Medium- related skills in personal computers (desktops and laptops) and smart devices (mobile and tablet) use, and Usage skills. Seven statements were used to measure medium-related skills related to personal computers and mobile devices, respectively. Respondents were asked to check their digital skill level on a 4-point Likert scale, where a value of 1 referred to ‘Not at all,’ and a value of 4 referred to ‘Very likely.’ The statements for personal computer skills included: “I could install/update/delete software I need on my computer.”, “I could connect wired/wireless internet connection to my computer.”, “I could utilize web browser settings (blocking pop-ups, text size, security settings).”, “I could connect and use various external devices to my computer.”, “I could send a digital file in my computer to other people.”, “I could run and resolve computer virus or spyware problems,” and “I could write text files on my computer.” (∝=0.974, M=1.94, SD=0.98) The statements for mobile device skills included: “I could change display/sounds/security/alarms settings on my devices.”, “I could connect wireless internet.”, “I could transmit digital files on smart devices to personal computers.”, “I could send digital files and photos to other people.”, “I could install/update/delete applications on my smart devices.”, “I could run and resolve virus problems on smart devices.”, and “I could write text files on smart devices.” (∝=0.953, M=2.46, SD=0.98). For analyses, new variables were generated by averaging the Likert scale score for the seven statements related to the personal computer and mobile device, respectively.Twenty statements were used to measure usage skills, and the respondents were asked to check their digital usage level on a 5-point Likert scale, where a value of 1 referred to ‘Not at all,’ and a value of 5 referred to ‘Very likely.’ The statements for usage skills included: “I could host and attend virtual meetings.”, “I can distinguish reliable information among results from information- seeking activity in search engines by comparing them to other resources.”, “I can purchase items by using online payment.”, “I can find a route by using navigation or online map services.” 20 (∝=0.98, M=2.18, SD=1.05). For each respondent, new variable was generated by averaging the Likert scale score for the twenty statements related to the usage skills. Digital Usage: Digital Usage was divided into two categories in the survey: Frequency of digital use and Variance in digital services use. Regarding frequency, the respondents were asked to write how many days they used a personal computer, smartphone, and tablet pc within one month from the survey. To measure variance in digital usage, the respondents were asked to check the frequency in the last year of each of seven types of digital activity in both computer and mobile on a 4-point Likert scale, where a value of 1 referred to ‘Not used at all,’ and a value of 4 referred to ‘Frequently used.’ The digital activities include: (1) Search, email, and content activity (search information/news, e-mail, media content use (films, music, electronic books), and education content use) (∝=0.886, M=1.96, SD=0.69), (2) Social media activity (social network services, instant messenger, personal blog, online community, and cloud service) (∝=0.918, M=1.75, SD=0.65), (3) Everyday life information activity (weather, news, maps, and public transportation information, e-commerce, online banking, and public services) (∝=0.906, M=2.04, SD=0.73), (4) Information seeking and sharing activity (sharing digital content, uploading user-created content, and sharing hyperlinks) (∝=0.868, M=1.72, SD=0.77), (5) Networking activity (maintain personal relationships and meet and communicate with people) (∝=0.823, M=1.98, SD=0.74), (6) Social engagement activity (express opinion, provide policy proposal or political comments, donating or volunteering activities, and participate in online voting or opinion poll) (∝=0.933, M=1.42, SD=0.63), (7) Financial activity (help job seeking or promotion, marketing activities, obtain financial information, and cost reduction activities) (∝=0.923, M=1.62, SD=0.71). For each respondent, new variables were generated by averaging the Likert scale score for the four statements related to each seven above digital activities. Attitude towards technology: There were four statements in the survey for measuring attitudes towards digital technology in the survey. A 4-point Likert scale was used, where a value of 1 referred to ‘Strongly disagree,’ and a value of 4 referred to ‘Strongly agree.’ The statements for attitude towards digital technology included: “Digital technology is useful,” “Digital technology makes my life easier,” “Digital technology is good for me,” and “I would like to use digital technology more than now.” (∝=0.901, M=2.75, SD=0.70). For the regression analyses, new variables were generated by averaging the Likert scale scores for the four statements. 21 Demographic information: The survey collected information on Age, Education Level, and Monthly Income of the respondents. 3.5. Methods To get an overview of the data, basic descriptive statistics were generated, including cross-tabulations of selected variables. I also examined the pattern of correlations between variables and ran a few selected simple regression analyses to get a first understanding of the factors that might influence the dependent variables (obtaining COVID-19-related information, stimulus check requesting services, and information services). To discern patterns and explore the research questions, multivariate regression analyses were conducted. I examined the relationships between attitudes, socioeconomic status, digital skills, and digital usage levels of farmers and fishers in the contexts of obtaining COVID-19- related information on stimulus check requesting services and information services. Multiple specifications of the multivariate regression analyses were conducted to find the model with the best explanatory power. The theoretical framework in this study suggests that multiple factors are in play. Moreover, South Korea has ubiquitous Internet and device access with a unique technological context compared to other developed countries. In other words, it is not clear what is the best explanatory model in this unique context, which is why exploring factors is necessary for the empirical study. First, all independent variables run simultaneously as a full model, and then each group of factors respectively to distinguish which factors are associated with the outcome variables. Also, multivariate regression analyses with three split age groups were conducted. And the results were compared since the age variable had been identified not only as one of the factors that strongly affect COVID-19-related information in cross-tabulation and simple regression analyses but also related to multicollinearity issues after empirical trials. IBM SPSS Statistics (Version 25) was used as a data analysis tool for this study. 22 4. RESULTS 4.1. Descriptive statistics Descriptive statistics and cross-tabulations were used to understand the distribution and basic patterns of the variables in the dataset. This provided initial, general insights regarding the digital divide among South Korean farmers and fishers. 4.1.1. Responses to the COVID-19 outbreak After the COVID-19 outbreak, the Korean government provided several COVID-19- related services to help citizens. To support financial challenges, the government issued a stimulus check for all populations with South Korean nationality. In addition, Korea Disease Control and Prevention Agency (KDCA) announced the number of COVID-19-positive patients and the location they visited before confirmed COVID-19 every 9 AM for preventive action against potential exposure to the virus after the outbreak of COVID-19. Obtaining this information was critical to maintaining the daily lives of people since it is closely related to their financial resources and health issues. The survey revealed variations among respondents regarding the awareness of COVID-19-related services, including stimulus check requesting services and information services, and the experience of using them. Regarding stimulus check requesting services, 18.5% of respondents answered that they were not aware of the existence of it. In comparison, 24.7% reported being unaware of the information service. Among respondents who were aware of stimulus check requesting services, 65.5% of them used the service. On the other hand, 72.7% of respondents who were aware of information services used it. Among those aware of the services, the most prevalent reason for not using them is ‘no need’ for both stimulus check requesting and information services among respondents who had not used the services. However, the second most important reason was the ‘difficulty in use.’ This suggests that digital skills directly or indirectly affect the usage of COVID-19-related services. And these responses yielded similar outcomes from earlier studies that adequate digital skills help generate information and use services (DiMaggio et al., 2001 Pearce & Rice, 2013; Van Deursen & Van Dijk, 2015). 23 Table 2 Awareness and usage of COVID-19 related information in respondents Aware_Not Type of services Unaware Aware_Use N use Stimulus check Frequency 408 618 1174 requesting 2200 Percentage services 18.5 28.1 53.4 of total (%) Frequency 543 452 1205 Information 2200 services Percentage 24.7 20.5 54.8 of total (%) Various cross-tabulations between awareness and usage of COVID-19-related services and digital access level, socioeconomic status, attitude towards digital technology, digital skill level, and digital usage level were conducted. Only selected interesting findings are reported here. One result, related to the age of respondents, showed a high association between age and the awareness and usage of COVID-19-related services. Figures 3a ~ 3d show the association of age and digital usage gaps for COVID-19-related services. As respondents get older, the number of unawareness of both types of services significantly increases. In addition, the number of respondents who had not used the services significantly increased despite the awareness of both types of services as the age of respondents increased. Respondents above 70 years old were significantly less aware of both types of services than any other age group. Moreover, the frequency of using the services with awareness decreased sharply among respondents over 60 in both types of services. 24 Figure 3a. Awareness and usage of COVID-19 related stimulus check requesting services by age group (frequency) Stimulus check requesting services (Frequency) Above 70 60~69 50~59 40~49 30~39 20~29 10~19 0 50 100 150 200 250 300 350 400 450 10~19 20~29 30~39 40~49 50~59 60~69 Above 70 Aware Use 3 23 65 196 406 324 157 Aware_Not use 1 0 1 13 104 262 237 Unaware 0 0 2 2 31 97 276 Aware Use Aware_Not use Unaware Figure 3b. Awareness and usage of COVID-19 related stimulus check requesting services by age group (percentages of each age group) Stimulus check requesting services (Percentage) Above 70 60~69 50~59 40~49 30~39 20~29 10~19 0% 20% 40% 60% 80% 100% 120% 10~19 20~29 30~39 40~49 50~59 60~69 Above 70 Aware_Use 75% 100% 95.60% 92.90% 75.00% 47.40% 23.40% Aware_Not use 25% 0% 1.50% 6.20% 19.20% 38.40% 35.40% Unaware 0% 0% 2.90% 0.90% 6% 14.20% 41.20% Aware_Use Aware_Not use Unaware 25 Figure 3c. Awareness and usage of COVID-19 related information services by age group (frequency) Information services (Frequency) Above 70 60~69 50~59 40~49 30~39 20~29 10~19 0 50 100 150 200 250 300 350 400 450 10~19 20~29 30~39 40~49 50~59 60~69 Above 70 Aware Use 3 22 61 189 398 353 179 Aware_Not use 1 1 4 17 99 175 155 Unaware 0 0 3 5 44 155 336 Aware Use Aware_Not use Unaware Figure 3d. Awareness and usage of COVID-19 related information services by age group (in percentages of age groups) Information services (Percentage) Above 70 60~69 50~59 40~49 30~39 20~29 10~19 0% 20% 40% 60% 80% 100% 120% 10~19 20~29 30~39 40~49 50~59 60~69 Above 70 Aware_Use 75% 95.70% 89.70% 89.60% 73.60% 51.70% 26.70% Aware_Not use 25% 4.30% 5.90% 8.10% 18.30% 25.60% 23.10% Unaware 0% 0% 4.40% 2.40% 8.10% 22.70% 50.10% Aware_Use Aware_Not use Unaware 26 4.1.2. Physical and material access Table 3 Physical and material access Percentage Access category Values Frequency N (%) Do not have 1072 48.7 Desktop 2200 Have 1128 51.3 Do not have 1639 74.5 Laptop 2200 Have 561 25.5 Have nothing 42 1.9 Have mobile Mobile 2200 (smartphone or 2158 98.1 feature phone) Device Have nothing 1884 85.6 Have 1 type (tablet, smartwatch, AI 241 11.0 speaker, smart Smart devices 2200 peripheral devices) Have all kinds (tablet and other 75 3.4 smart device) Internet Unavailable 20 0.9 Internet connection at 2200 home Available 2180 99.1 Table 3 shows that the level of access to internet connections and digital devices among South Korean farmers and fishers was high, particularly regarding internet access and mobile devices. Almost every respondent reported having their own mobile device (98.1%), either a smartphone or a feature phone, and internet connection at their home (99.1%). Ownership of personal computers, either desktop or laptop, and other smart devices, such as tablet, smartwatch, AI speaker, or other smart peripheral devices, was much lower than mobile devices. The number of respondents who owned a desktop (51.3%) was more than double of those who had a laptop (25.5%). Prevalence of ownership of a desktop showed the preference regarding digital devices, excluding mobile phones, to older respondents. This prevalence also showed in the low possession of smart devices, including tablets, smartwatches, AI speakers, and smart peripheral devices. Only 316 respondents (14.4%) answered that they own at least one of the several types of smart devices. Even though digital and device access cannot automatically guarantee usage, 27 the level of digital physical and material access is very high. Thus, it is less of a constraint and may even facilitate the development of appropriate skills. 4.1.3. Digital skills According to the 2021 NIA report, the index of digital skills of South Korean farmers and fishers is 69.6 out of 100, the index of the Korean population. The index consists of three weighted components: 20% of digital access level, 40% of digital skills, and 40% of digital usage. Digital skills are divided into two categories in the survey and dataset: medium-related and content-related (usage) skills. The data measured medium-related skills of personal computers, including desktops and laptops, and mobile devices. The medium-related digital skills (M=2.20, SD=0.93) were slightly more developed than usage skills (M=2.17, SD=1.05). However, digital skills in personal computers and mobile devices differed. Digital medium- related skills in mobile devices (M=2.46, SD=0.98) were much higher than in personal computers (M=1.94, SD=0.98). Since most respondents have mobile phones, users may develop medium-related skills related to mobile devices more quickly than skills related to personal computers. 4.1.4. Digital usage Among respondents, 1,803 farmers and fishers (81.9%) answered that they had used the internet within one month of the survey. However, 340 respondents (15.5%) said that they had never used the internet, and 57 respondents (2.6%) said that they had not used the internet for over one month. These numbers show that there are still parts of the population who do not use the internet, despite high internet use among South Korean farmers and fishers. Since these 397 respondents had answered that they had no use of the internet at all or within the last month, measuring their digital usage level was impossible. Thus, the following analysis related to the digital usage level considered the data of 397 respondents as missing values. 28 Figure 4a. Internet use in the last month Number of Respondents 2000 1803 1800 1600 1400 1200 1000 800 600 400 340 200 57 0 Use within one month Not use over one month Have never used the internet 4.2. A First Step Toward Examining the Data: Correlation Matrix and Simple Regressions 4.2.1. Correlations among variables To gain insights into the relations among the variables and for diagnostic purposes, a correlation matrix was calculated (see Tables 4 ~ 7). The Pearson correlation coefficients examine the strengths of the linear relations among variables. This information also helped in refining the multivariate analyses. Among the independent variables, only the respondent’s age shows a negative correlation with the usage of both stimulus check requesting and information services. Within digital access and connectivity, the mobile phone, either a smartphone or a feature phone, had the strongest positive correlation (r = .687, p<.01) with internet access than any other devices among respondents. This shows that mobile devices might be the dominant technical means to connect to the Internet to respondents. Also, medium-related skill in a personal computer had a slightly stronger positive correlation with usage skill (r = .829, p<.01) than that in a mobile (r = .812, p<.01) even though the percentage of possession of mobile phone (98.1%) is much higher than a personal computer (58.8%). Lastly, medium-related skill in a mobile phone was the strongest 29 correlated with the usage of both COVID-19-related stimulus check requesting (r = .678, p<.01) and information services (r =.652, p<.01) among independent variables. These correlation results provide a basic understanding of the relationship between variables. They also provide first indications for the presence of multicollinearity. 30 Table 4 Pearson Correlation matrix for COVID-19 related stimulus check requesting and information services (1/4) Mobile Smart Educa- PC Mobile Desktop Laptop Internet Attitude Age Income phone devices tion skills skills Desktop 1 .226** .097** .215** .089** .375** -.351** .375** .349** .492** .458** Laptop 1 .066** .397** .056** .320** -.227** .197** .233** .433** .381** Mobile 1 .046* .687** .154** -.082** .153** .094** .118** .173** phone Smart 1 .037 .263** -.168** .142** .195** .387** .322** devices Internet 1 .118** -.052* .101** .071** .082** .118** Attitude 1 -.421** .449** .356** .563** .654** Age 1 -.673** -.534** -.492** -.547** Educa- 1 .617** .469** .557** tion Income 1 .377** .436** PC skills 1 .784** Mobile 1 skills Usage skills 31 Table 5 Pearson Correlation matrix for COVID-19 related stimulus check requesting and information services (2/4) Use_ Use_ Usage Use_ Use_ Use_Info Use_Net Social Use_ Stimulus Info- Social N skills Search Everyday - seeking -working engage- Financial check services media ment Desktop .449** .407** .314** .423** .292** .293** .235** .320** .411** .415** 2200 Laptop .419** .427** .371** .430** .306** .325** .262** .354** .306** .282** 2200 Mobile .151** - - - - - - - .179** .149** 2200 phone Smart .388** .438** .423** .402** .370** .344** .332** .366** .236** .204** 2200 devices Internet .105** - - - - - - - .117** .103** 2200 Attitude .617** .500** .436** .494** .397** .407** .262** .361** .561** .581** 2200 Age -.512** -.329** -.265** -.394** -.237** -.270** -.200** -.261** -.489** -.454** 2200 Educati .489** .322** .228** .368** .193** .229** .168** .204** .577** .489** 2200 on Income .405** .285** .239** .351** .190** .239** .189** .232** .463** .419** 2200 PC .829** .766** .639** .741** .556** .559** .434** .576** .585** .558** 2200 skills Mobile .812** .665** .530** .666** .483** .506** .342** .471** .670** .652** 2200 skills Usage 1 .743** .709** .752** .643** .653** .553** .678** .636** .622** 2200 skills 32 Table 6 Pearson Correlation matrix for COVID-19 related stimulus check requesting and information services (3/4) Mobile Smart Educati PC Mobile Desktop Laptop Internet Attitude Age Income phone devices on skills skills Use_ search Use_ social media Use_Every day Use_Info seeking & sharing Use_Net- working Use_Social engage- ment Use_ Financial Stimulus check Information services 33 Table 7 Pearson Correlation matrix for COVID-19 related stimulus check requesting and information services (4/4) Use_ Use_ Usage Use_ Use_ Use_Info Use_Net Social Use_ Stimulus Info- Social N skills Search Everyday - seeking -working engage- Financial check services media ment Use_ 1 .818** .858** .639** .645** .528** .649** .495** .449** 1803 search Use_ social 1 .782** .697** .684** .584** .676** .399** .383** 1803 media Every-day 1 .651** .659** .547** .687** .558** .513** 1803 life info Info seeking & 1 .776** .625** .689** .318** .330** 1803 sharing Networking 1 .603** .715** .338** .372** 1803 Social 1 .724** .269** .227** 1803 engagement Financial 1 .374** .353** 1803 Stimulus 1 .756** 2200 check Information 1 2200 services Note: **. Correlation is significant at the 0.01 level (2-tailed), *. Correlation is significant at the 0.05 level (2-tailed) in Tables 4~7. 34 4.2.2. Simple Regression Analyses 4.2.2.1. Usage of COVID-19-related information on stimulus check requesting services Before conducting multivariate regression analyses, multiple single regressions were conducted to get a first understanding of the associations of independent variables with the dependent variables. This approach would only hold under the strict methodological assumption that all other factors are equal (or irrelevant). Because this is likely not the case, we use insights only to inform and guide subsequent analyses and interpretations. Simple regressions were conducted for digital material access level (availability of internet connection and possession of devices), socioeconomic status, attitude toward technology, digital skills, and digital usage in seven different activities. Given the results of the correlation analysis and the large sample, it was not surprising that all independent variables were statistically significant (p<.001). The following results illustrate selected associations. For information purposes, all results of the simple regressions for COVID-19 related stimulus check requesting services are in Appendix B. Table 8 Simple regression analysis between age and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Age -.326 .012 -.489 -26.28 690.86 .239 2200 Table 9 Simple regression analysis between attitude toward technology and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Attitude .618 .019 .561 31.80 1011.31 .315 2200 Table 10 Simple regression analysis between medium-related skill in a mobile device and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Mobile skill .527 .012 .670 42.31 1789.79 .449 2200 35 Table 11 Simple regression analysis between digital usage skills and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Usage skill .470 .012 .636 38.64 1493.33 .405 2200 Table 12 Simple regression analysis between medium-related skill in a personal computer and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β PC skill .460 .014 .585 33.80 1142.64 .342 2200 Table 13 Simple regression analysis between everyday life information activity usage and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Everyday life .497 .017 .558 28.51 812.60 .311 1803 information Note: All values shown in table 5a ~ 5f are significant of p<.001. Notably, age was negatively associated with the usage of stimulus check requesting service (Table 8) (β = -.326, p <.001). All the other independent variables were positively associated with the usage of stimulus check requesting service. The strength of the association for usage stimulus check requesting service varied in each independent variable and every single simple regression. Among socioeconomic status and attitude towards technology, the single strength of the association with attitude toward technology (β =.618, p <.001) was the highest (Table 9), which means that as attitude towards technology is more favorable, more likely to use stimulus check requesting services. Regarding digital skills (Tables 10 ~ 12), unlike the results of correlation, the single strength of association with medium-related skills in a mobile device (β =.527, p <.001) was higher than that in a personal computer (β =.460, p <.001). Also, digital usage skills showed positive association (β =.470, p <.001) and were higher than PC skills. The respondents are more 36 likely to use the stimulus check requesting services if they have higher medium-related specifically in mobile and usage skills, regardless of devices. At the digital usage level, usage in everyday life information activity (Table 13) was strongly associated (β =.497, p <.001) with usage stimulus check requesting service. 4.2.2.2. Usage of COVID-19 related information services Multiple simple regressions were also conducted for the usage of COVID-19-related information services. All independent variables were associated with the dependent variables in a statistically significant way (p<.001). Patterns of association were like the simple regressions of usage stimulus check requesting service. There was a slight variance in the degree of association in each independent variable between the two types of COVID-19-related services. Only a few selected results from the simple regression analyses are presented below. A complete discussion of the results can be found in Appendix C. Table 14 Simple regression analysis between age and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Age -.328 .014 -.454 -23.88 570.11 .206 2200 Table 15 Simple regression analysis between attitude toward technology and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Attitude .694 .021 .581 33.47 1119.91 .338 2200 Table 16 Simple regression analysis between medium-related skill in a mobile device and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Mobile skill .556 .014 .652 40.32 1625.35 .425 2200 37 Table 17 Simple regression analysis between digital usage skills and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Usage skill .498 .013 .622 37.22 1385.52 .387 2200 Table 18 Simple regression analysis between medium-related skill in a personal computer and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β PC skill .476 .015 .558 31.55 995.42 .312 2200 Note: All values shown in table 6a ~ 6e are significant of p<.001. Like the results for usage of stimulus check requesting services, the age variable was negatively associated with the usage of COVID-19 related information services (Table 14) (β = - .328, p <.001). The other independent variables were all positively associated with the usage of COVID-19 related information services. Among socioeconomic status and attitude towards technology, the strength of the association in attitude towards technology (Table 15) showed higher (β =.694, p <.001) compared to the usage of stimulus check requesting service (Table 9). Regarding digital skills (Tables 16 ~ 18), the order of strength of the association was slightly different from the degree of correlation between digital skills and the usage of COVID- 19 related information services. The strength of the association showed in the order of medium- related skills in a mobile device (β =.556, p <.001), usage skills (β =.498, p <.001), and medium- related skills in personal computers (β =.476, p <.001), respectively. 4.3. Multivariate Regression Analyses Given the ordinal dependent variable, methods such as logistical regression would have to be used in confirmatory analysis. However, given the aim to conduct preliminary, explanatory analysis, multivariable regression analysis method was employed as a first approximation in this study. The multivariate regression analyses were conducted in two steps for each independent variable. Initial model runs with independent variables suggested by our theoretical framing showed the presence of strong multicollinearity among variables. Several reasons may cause that 38 problem, including the high correlation among independent variables noticed in the correlation matrix (Tables 4 ~ 7). For example, in an earlier study, sociodemographic factors were identified as one of the influential factors in material access, and attitudes were also specified as one factor that could cause inequalities in material access (Pearce & Rice, 2013). Since socioeconomic status factors and digital access levels could move in parallel, multicollinearity issues would occur. Moreover, other variables aware highly correlated, such as internet connectivity and mobile device possession or medium-related skill in a personal computer and usage skill. To mitigate the multicollinearity problem, multiple models were run using a stepwise approach. In the first set of analyses, the entire data set was used, and independent variables were added and dropped until a model with an acceptable condition index was found. The threshold for entering and removing independent variables was set at a probability αE to enter the variable of ≤.05 and a probability αR to remove of ≥.10. To compare the results of the best-fitting model with potential, simpler explanations, multivariate regression analyses were conducted in two ways: analyses of all independent variables together and analyses of sub-groups of factors. In these models, age was one of the causes of multicollinearity and hence had to be eliminated. However, descriptive inspection of the data suggested a strong association of age with outcomes. To examine this issue while avoiding the multicollinearity problem, a split sample analysis was conducted for three separate age groups. This approach allowed assessing variations among the factors associated with the usage of COVID-19 related information. The sample was divided into three groups: 15~39, 40~59, and above 60. The age thresholds were chosen based on suggestions in the research literature. The age 25~40 years are considered peak users of digital access (Van Dijk, 2013). Individuals sixty years and above are considered older adults. 4.3.1. Usage of COVID-19-related information on stimulus check requesting services 4.3.1.1. All respondents Multivariate regression analysis of the full sample using the stepwise method identified incrementally better models. Table 19 shows selected models for COVID-19 related stimulus check requesting services. Only Models 6-9 shown in Table 19. Models 1-5 are not presented because they had a low value of r². 39 Model 9 had the highest value of r² but presented multicollinearity problems. There are multiple ways to determine whether multicollinearity is present: checking collinearity tolerance, statistics VIF, and the condition index. If at least one of the following criteria is met, then it is sufficient to say that multicollinearity is present: collinearity tolerance ≤ 0.1, VIF ≥ 10, or condition index ≥15. The collinearity tolerance and statistics VIF were calculated and did not indicate multicollinearity. However, the condition index, widely regarded as the most reliable statistic to examine multicollinearity, indicated the presence of multicollinearity among variables in several models. The most widely accepted and desirable condition index is below 15, although this threshold is not rigid. Condition indices between 15 and 20 may be acceptable but will be associated with wider confidence intervals of the estimates. If this trade-off is acceptable, higher conditions indices can be tolerated. Given the number of observations, a higher threshold for the condition index was acceptable, if other test statistics, such as r², improved and parameter estimates remained statistically significant. Table 19 shows results of models with better goodness of fit. Models 1-5 are not shown, as they had worse goodness of fit. In Model 9, the attitude towards technology, social media activity usage, and networking activity usage was entered as independent explanatory variables. These variables were highly correlated with the independent explanatory variables used in Model 6. The correlation matrix (Table 4) shows that, the attitude variable is highly correlated with medium-related skills in a mobile device (r = .654, p<.01), and usage of digital activities arehighly correlated with each other. These high correlations cause the multicollinearity problem in Model 9. Thus, of the options, Model 6 was selected from this first set of models based on an analysis of all independent variables since it had an acceptable condition index (r² = .411). 40 Table 19 Selected multiple result models of multivariate regression analysis in all respondents for COVID-19 related stimulus check requesting services Variables Model 6 Model 7 Model 8 Model 9 Possession of desktop Possession of laptop Possession of Access mobile phone Possession of smart devices Internet connection Attitude Attitude .068** .069** .070** Age Socioeconomic Income level .084*** .082*** .082*** .084*** status Education level .162*** .159*** .155*** .154*** H/W skills_PC H/W Digital skills skills_Mobile .173*** .157*** .149*** .150*** devices Usage skills .165*** .150*** .169*** .177*** Search, email, and content service Social media -.078* -.069* Everyday life .306*** .299*** .339*** .345*** information Digital usage Information seeking and -.118*** -.122*** -.101*** -.068* sharing Networking -.064* Social engagement Financial activities Adjusted r² .411 .414 .415 .416 Condition index 17.183 22.427 24.053 25.421 N 2200 * p<0.1, ** p<0.01, *** p<0.001 The values shown are standardized beta coefficients. 41 To examine whether more parsimonious models with goodness of fit could be generated with sub-groups of factors, multivariate regression analyses within each category of independent variables were run. Table 20 summarizes the results in each category of multivariate regression analyses for the usage of COVID-19 related stimulus check requesting services. Model 6, thus far the best fit, is shown for the purposes of comparison. Table 20 Summary of multivariate regression analyses in all respondents for COVID-19 related stimulus check requesting services Model 11: Model 10: Model 12: Model 13: Model 14: Socioecon Variables Model 6 Access Digital Socioecon Digital omic level skills & skills usage status Possession .338*** of desktop Possession .188*** of laptop Possession of mobile .130*** Access phone Possession of smart .083*** devices Internet connection Attitude Attitude .366*** .143*** Age Socioecon Income .084*** .126*** .084*** omic level status Education .162*** .335*** .222*** level H/W skills_PC H/W Digital skills_Mob .173*** .451*** .256*** skills ile devices Usage .165*** .270*** .198*** skills Search, email, and .148** Digital content usage service Social -.121** media 42 Table 20 (cont’d) Everyday life .306*** .564*** informatio n Informatio n seeking -.118*** -.059* and Digital sharing usage Networkin g Social engagemen t Financial activities Adjusted r² .411 .238 .457 .473 .543 .319 Condition index 17.183 18.498 11.245 9.939 15.671 17.881 N 2200 2200 2200 2200 2200 1803 * p<0.1, ** p<0.01, *** p<0.001 The values shown are standardized beta coefficients. The following paragraphs discuss highlights from the results presented in Table 20. The results indicate that this strategy allows improvements in the overall goodness of fit. Eventually, Model 13 was selected as the best-fitting model to explain the usage COVID-19 related stimulus check requesting services. The model was selected because it explains a higher share of the variance than other specifications. Its condition index is higher than that of Models 11 and 12 but is in an acceptable range. This finding suggests that socioeconomic factors and skills are the best predictors of the dependent variable. In the comparison model (Table 20, Model 6), the included independent explanatory variables were statistically significant. Everyday life information activity usage showed the strongest positive association (β = .306, p <.001) with the usage of stimulus check requesting services among all identified independent variables. Explanatory variables related to digital skills also showed strong associations. Medium-related skills in a mobile device (β = .173, p <.001) and digital usage skills (β = .165, p <.001) showed a high positive association with the usage of stimulus check requesting services. Socioeconomic status factors showed a positive association as well. Education level (β = .162, p <.001) showed higher strength of the association than income level (β = .084, p <.001). On the other hand, no variables from the digital access subgroup, attitude towards technology, age, medium-related skills in a personal computer, and 43 digital activities usage made a sufficient incremental contribution to the explanatory power to be included in the model. Information seeking and sharing activity usage showed a negative association (β = -.118, p <.001). A separate analysis of the digital access subgroup of variables (Table 20, Model 10), showed that the availability of the internet connection at home was not associated with usage of stimulus check requesting services. Possession of a desktop (β = .338, p <.001) showed the strongest positive association. Access to devices was also associated with the usage of stimulus check requesting services. Regarding sociodemographic and attitudinal factors (Table 20, Model 11), attitude towards technology showed the strongest positive association (β = .366, p <.001), followed by the education level (β = .335, p <.001). In the digital skills subgroup analysis (Table 20, Model 12), medium-related skills in a personal computer were not associated as presented in the full model (Model 6). Strength of the association with medium-related skills in a mobile device (β = .451, p <.001) showed much higher than digital usage skills (β = .270, p <.001). Among digital activities usage (Table 20, Model 14), as presented in the full model (Model 6), digital activities usage level of everyday life information, such as news, weather, and public transportation information, show the highest association (β = .564, p <.001). Other activities usage, such as search, email, and content services activities (β = .148, p <.001), social media activity (β = -.121, p <.01), and information seeking and sharing (β = -.059, p <.1), had weaker association with the usage of COVID-19 related stimulus check requesting services. Usage of social media and information seeking and sharing activities showed a negative association. The model including variables related to socioeconomic status and digital skills (Table, 20, Model 13) emerged as the best-fitting model to explain the usage of COVID-19 related stimulus check request services. The value of the condition index for Model 13 was 15.671, which is still acceptable. Also, the value of r² was the highest (r² = .543) among all tested models. That is, Model 13 had the highest explanatory power compared to other result models, such as the digital skills model (Model 12, r² = .473). In Model 13, the strength of association declined from medium-related skills in a mobile device (β = .256, p <.001), education level (β = .222, p 44 <.001), digital usage skills (β = .198, p <.001), attitude towards technology (β = .143, p <.001), and income level (β = .084, p <.001). 4.3.1.2. Age group comparison Table 21 summarizes the results for the usage of COVID-19 related stimulus check requesting services differentiated for three age groups. This modeling strategy was used to examine whether age is working as a moderator, as suggested by initial descriptive analyses. The following paragraphs discuss selected highlights for each model and which specifications offered the best overall fit. The results revealed that different models explain the association between independent and dependent variables for different age groups. Best models were selected by considering acceptable levels of the condition index and the value of r². Model 3 (Table 21) which included only socioeconomic indicators had the highest explanatory power for the age groups under-40 (r² =.372, condition index = 17.48) and 40-59 years old (r² = .252, condition index = 17.46). On the other hand, for the above 60 years old group, the model including digital skills (Table 21, Model 4) was the best fitting (r² =.426, condition index = 8.75). For the sake of clarity, I will briefly comment on other model specifications. All other models have weaker explanatory power. For example, in the full model with age group comparison (Table 21, Model 1), as the age group of respondents got older, more associated variables appeared. In the under-40 age group, only the education level (β = .377, p <.001) and medium-related skills in a personal computer (β = .337, p <.001) showed an association with using stimulus check requesting services. In the middle age group (40-59 years old), everyday life information activity usage (β = .237, p <.001), the education level (β = .206, p <.001), and the attitude towards technology (β = .181, p <.001) showed a positive association. In the above 60 years old group, everyday life information activity usage showed the strongest positive association (β = .442, p <.001). Digital usage skill was also highly associated (β = .244, p <.001). However, two factors were negatively associated with the dependent variable: social media (β = - .157, p <.01) and information seeking and sharing activities usage (β = -.118, p <.01). Only the education level contributed to explaining the use of stimulus check requesting services in all age groups. A degree of variance in the strength of the associations appeared in Models 2 and 3 (Table 21) regarding digital material access and socioeconomic status. In the material access 45 subgroup analysis, only possession of a laptop was statistically significant and positively associated with the dependent variable in the under-40 years old group (β = .228, p <.1). With the other two groups, possession of a desktop, laptop, mobile phone, and other smart devices were positively associated with the dependent variable. The degree of the association was higher in the above 60 years old group than the 40-59 group. However, the availability of an internet connection was not associated with the dependent variable in all age groups. In the model focusing on digital skills (Table 21, Model 4) only medium-related skills in a personal computer showed an association in the under-40 years old group (β = .560, p <.001). On the other hand, all types of digital skills were associated with the 40-59 years old group. Also, medium-related skills in a mobile device and digital usage skills were positively associated with the rest of the two age groups. Still, the order of the strength of the association was different. For the 40-59 age group, digital usage skills had a higher association (β = .260, p <.001) than medium-related skills in a mobile device (β = .172, p <.001), but for above 60 years old group, medium-related skills in a mobile device (β = .410, p <.001) had higher association than digital usage skills (β = .277, p <.001). Among the digital activities usage group of variables (Table 21, Model 5), usage of everyday life information activities showed a strong association in all age groups. 46 Table 21 Summary of multivariate regression analyses with age groups for COVID-19 related stimulus check requesting services Model 1: Model 2: Model 3: Model 4: Model 5: Variables All Access level Socioeconomic status Digital skills Digital usage 40≤G 40≤G 40≤G 40≤G 40≤G Age group G<40 60≤G G<40 60≤G G<40 60≤G G<40 60≤G G<40 60≤G <60 <60 <60 <60 <60 Possession of .167* .294* desktop ** ** Possession of .135* .179* .228* laptop ** ** Possession of .119* .142* Access mobile phone * ** Possession of .089* .084* smart devices * Internet connection .181* .290* .396* Attitude Attitude .204* ** ** ** .108* .108* .133* Socioec Income level ** * ** o-nomic Education .377* .206* .120* .537* .270* .250* status level ** ** ** ** ** ** H/W .337* .560* .113* skills_PC ** ** H/W Digital .155* .172* .410* skills_Mobile skills ** ** ** devices .244* .213* .277* Usage skills ** ** ** Search, email, and .241* content ** service Digital - - usage Social media .157* .270* * ** Everyday life .237* .442* .307* .438* .562* information ** ** * ** ** 47 Table 21 (cont’d) Information - - seeking and .118* .087* sharing * Digital Networking usage Social engagement Financial activities Adjusted r² .322 .221 .401 .042 .087 .201 .372 .252 .354 .306 .194 .426 .085 .157 .320 Condition index 14.70 19.90 19.03 2.67 25.73 15.50 17.48 17.46 9.80 10.33 15.58 8.75 9.29 9.20 15.78 N 95 752 1353 95 752 1353 95 752 1353 95 752 1353 94 727 982 * p<0.1, ** p<0.01, *** p<0.001 The values shown are standardized beta coefficients. 48 4.3.2. Usage of COVID-19 related information services 4.3.2.1. All respondents Multivariate regression analyses of the entire sample for usage of COVID-19 related information services using the stepwise method also identified incrementally better models. Table 22 shows models for COVID-19 related information services. Only Model 4-Model 8 areshown in Table 22 since Model 1-Model 3 had a low value of r². In determining the best- fitting specification, the same considerations as in the previous analyses were adopted for the condition index. Model 8 (Table 22) had the highest value of r² but presented multicollinearity problems. In Model 8, social engagement activity usage, possession of smart devices, and possession of desktop were added as independent explanatory variables compared to Model 5. Model 4 and Model 5 had advantages and disadvantages regarding the condition index and explanatory power. Model 4 had a lower condition index (17.355) but lower explanatory power (r² = .329) as well, compared to Model 5 (r² = .334). On the other hand, Model 5 had a slightly higher condition index (19.420) and a value of r². The condition index of Model 5 was still below 20, which is the threshold I set. Also, Model 5 could explain more than Model 4 for the usage of COVID-19 related information services. Therefore, Model 5 was selected among the options since it had an acceptable condition index and explanatory power (r² = .334). 49 Table 22 Selected multiple result models of multivariate regression analysis in all respondents for COVID-19 related information services Variables Model 4 Model 5 Model 6 Model 7 Model 8 Possession of .061** desktop Possession of laptop Possession of Access mobile phone Possession of -.070** -.071** smart devices Internet connection Attitude Attitude .175*** .160*** .154*** .156*** .153*** Age Socioeconomic Income level .118*** .118*** .119*** .120*** .113*** status Education level H/W skills_PC H/W Digital skills skills_Mobile .163*** .110*** .084** .084** .079* devices Usage skills .132*** .185*** .190*** .189*** Search, email, and content service Social media Everyday life .276*** .219*** .260*** .277*** .260*** information Digital usage Information seeking and sharing Networking Social -.109*** -.099*** -.099*** engagement Financial activities Adjusted r² .329 .334 .341 .345 .347 Condition index 17.355 19.420 21.371 21.856 22.974 N 2200 * p<0.1, ** p<0.01, *** p<0.001 The values shown are standardized beta coefficients. 50 To examine whether more parsimonious models could be generated with sub-groups of factors, multivariate regression analyses within each category of independent variables were conducted, as well. Table 23 summarizes the results in each category of multivariate regression analyses for the usage of COVID-19 related information services. Table 23 Summary of multivariate regression analyses in all respondents for COVID-19 related information services Model 10: Model Model Model 9: Model 11: Socioecon 12: 13: Variables Model 5 Access Digital omic Socioecon Digital level skills status & skills usage Possession .354*** of desktop Possession .174*** of laptop Possession of mobile .101*** Access phone Possession of smart .054** devices Internet connection Attitude Attitude .160*** .440*** .216*** Age Socioecon Income .118*** .134*** .091*** omic level status Education .209*** .095*** level H/W skills_PC H/W Digital skills_Mo .110*** .432*** .262*** skills bile devices Usage .132*** .271*** .193*** skills Search, email, and content service Digital Social usage media Everyday life .219*** .504*** informatio n 51 Table 23 (cont’d) Informatio n seeking and sharing Networkin Digital .109*** g usage Social engageme -.115*** nt Financial activities Adjusted r² .334 .221 .413 .450 .499 .272 Condition index 19.420 18.498 11.245 9.939 15.671 9.776 N 2200 2200 2200 2200 2200 1803 * p<0.1, ** p<0.01, *** p<0.001 The values shown are standardized beta coefficients. The following paragraphs discuss highlights from the results in Table 23, focusing on selected variables in each specification that were associated with the usage of COVID-19 related information services. Eventually, Model 12 in Table 23 was selected as the best-fitting specification to explain the usage of COVID-19 related information services among respondents. The patterns and strengths of the association for the usage COVID-19 related information services were different from that for stimulus check requesting services. In the full model (Table 23, Model 5), the included independent explanatory variables were all statistically significant. Everyday life information activity usage showed the strongest positive association (β = .219, p <.001) with the usage of information services among all independent variables, which was identical to the findings for stimulus check requesting services (Table 20, Model 6). Digital skills were also shown as independent explanatory variables. However, the order of the degree of association was different from the analysis regarding stimulus check requesting services. For information services, digital usage skills (β = .132, p <.001) had a higher association than medium-related skills in a mobile device (β = .110, p <.001). Socioeconomic status factors showed a positive association as well. Attitude towards technology (β = .160, p <.001) showed higher strength of the association than income level (β = .118, p <.001), which was the second strongest association following the everyday life information activity usage. On the other hand, variables in the digital access subgroup, age, education level, medium-related skills in a personal computer, and digital activities usage were not associated with the usage COVID-19 related information services. 52 In the digital access subgroup analysis (Table 23, Model 9), the strength was identical to the results of analyses in stimulus check requesting services. The availability of an internet connection at home was not associated, but possession of a desktop (β = .354, p <.001) showed the strongest positive association with usage of COVID-19 related information services. Possession of a laptop (β = .174, p <.001), mobile phone (β = .101, p <.001), and other smart devices (β = .054, p <.001) were positively associated in the material access level subgroup. In the sociodemographic factors and attitude group (Table 23, Model 10), attitude showed the strongest positive association (β = .440, p <.001), followed by education level (β = .209, p <.001) and income level (β = .134, p <.001). In the digital skills subgroup (Table 23, Model 11), medium-related skills on a personal computer were not associated. The single value of association with medium-related skills in a mobile device (β = .432, p <.001) showed a much higher association than digital usage skills (β = .271, p <.001). In the digital activities usage subgroup model (Table 23, Model 13), the associated types of digital activities use varied from the result of stimulus check requesting services. Still, the usage of everyday life information activities had the strongest association, but information seeking and sharing activity usage (β = .109, p <.001) and social engagement activity (β = -.115, p <.01) also showed an association. Social engagement activities usage was negatively associated with the dependent variable. Like the findings for stimulus check requesting services, the model including socioeconomic status and digital skills (Table 23, Model 12) generated the best-fit to explain the usage of COVID-19 related information services. The value of the condition index for Model 12 was 15.671, which is acceptable. The value of r² was the highest (r² = .499) among the models with an acceptable level of condition index, below 20. Thus, Model 12 was considered as the specification with the highest explanatory power compared to other models, such as the digital skills model (Model 11, r² = .450), even though it had a slightly higher condition index. In Model 12, medium-related skills in a mobile device (β = .262, p <.001), attitude towards technology (β = .216, p <.001), digital usage skills (β = .193, p <.001), education level (β = .095, p <.001), and income level (β = .091, p <.001) were all positively and significantly associated with the dependent variable. 53 4.3.2.2. Age group comparison Overall, the explanatory power of the analyses regarding COVID-19 related information services was much lower than that with stimulus check requesting services. Table 24 summarizes the results for the usage of COVID-19 related information services differentiated for three age groups. As mentioned in the previous analyses, this modeling comparison was used to examine whether age is working as a moderator for information services use. The following paragraphs discuss selected highlights for each model and which specifications offered the best overall fit. Like the age group comparison model for the usage of stimulus check requesting services, the results showed that different models explain the association between independent and dependent variables for different age groups. Model 4 (Table 24), which focused only on digital skills indicators, had the highest explanatory for the age groups under-40 (r² =.268, condition index = 10.33) and above 60 years old (r² = .426, condition index = 8.75). On the other hand, for the 40-59 years old group, the model only including socioeconomic indicators (Table 24, Model 3) was the best fitting (r² =.199, condition index = 17.46). For clarity, the following paragraphs will comment on the other model specifications. For instance, in the full model with age group comparison (Table 24, Model 1), as the age group of respondents got older, more variables were associated with the dependent variable. In the under- 40 years old groups, only medium-related skills in a personal computer (β = 431, p <.001) was statistically significant and had a positive association. On the other hand, in the 40-59 years age group, attitude towards technology (β = 212, p <.001), everyday life information activity usage (β = .136, p <.01), and digital usage skill (β = .130, p <.01), were positively associated with the dependent variable. In the above 60 years old group, everyday life information activity usage showed the strongest positive association (β = .438, p <.001). Also, digital usage skill was highly associated (β = .272, p <.001) among digital skills in the above 60 years old group. Attitude towards technology (β = .157, p <.001) and income level (β = .127, p <.001) were associated with usage of COVID-19 related information services among socioeconomic status factors. However, several negative associations were detected for the above 60 years old group. These were social media (β = -.124, p <.1), social engagement activities usage (β = -.115, p <.01), and possession of smart devices (β = -.069, p <.1). 54 There was no associated variable with the digital material access subgroup result model under-40 years old respondents (Table 24, Model 2). In material access (Model 2), possession of desktop (β = .162, p <.001) and laptop (β = .126, p <.001) were statistically significant and positively associated with in 40-59 years old group. In the above 60 years old group, all types of devices were associated with the usage of COVID-19 related information services. The strength of the association was higher in the above 60 years old group than the 40-59 group. However, the availability of internet connection was not associated with the dependent variable in all age groups. In the subgroup analysis for socioeconomic factors (Table 24, Model 3), most were associated with the usage of COVID-19 related information services in all age groups, except for income level in the 40-59 age group. The most substantial associated independent variable varied among age groups. Education level (β = .297, p <.001) had the highest strength in the under 40 years old group, while attitude towards technology showed the strongest associations in 40-59 (β = .331, p <.001) and above 60 years old (β = .470, p <.001) group. In the digital skill subgroup (Table 24, Model 4), only medium-related skills in a personal computer showed an association in the under 40 years old group (β = .525, p <.001). On the other hand, medium-related skills in a mobile device and digital usage skills were positively associated with the rest of the two age groups. Still, the order of the strength of the association was different. For the 40-59 age group, digital usage skills had a higher association (β = .260, p <.001) than medium-related skills in a mobile device (β = .165, p <.001), but for the above 60 years old group, medium-related skills in a mobile device (β = .409, p <.001) had higher association than digital usage skills (β = .277, p <.001). Among digital activities usage (Table 24, Model 5), usage of everyday life information activities showed a strong association in all age groups. And networking activity usage had a positive association, and social engagement activity showed a negative association in both the 40-59 and above 60 years old group. 55 Table 24 Summary of multivariate regression analyses with age groups for COVID-19 related information services Model 1: Model 2: Model 3: Model 4: Model 5: Variables All Access level Socioeconomic status Digital skills Digital usage 40≤G 40≤G 40≤G 40≤G 40≤G Age group G<40 60≤G G<40 60≤G G<40 60≤G G<40 60≤G G<40 60≤G <60 <60 <60 <60 <60 Possession of .162* .322* desktop ** ** Possession of .174* laptop ** Possession of .126* .099* Access mobile phone ** ** Possession of - .079* smart devices .069* * Internet connection .212* .157* .268* .331* .470* Attitude Attitude ** ** * ** ** .127* .110* .133* Socioec Income level ** * ** o-nomic .297* .144* .129* status Education level * ** ** .431* .525* H/W skills_PC ** ** Digital H/W skills_Mobile .165* .409* skills devices ** ** .130* .272* .260* .277* Usage skills * ** ** ** Search, email, and content service Digital - usage Social media .124* Everyday life .136* .438* .295* .298* .524* information * ** * ** ** 56 Table 24 (cont’d) Information seeking and sharing .105* Digital Networking .118* * usage - - - Social engagement .115* .135* .103* * * * Financial activities Adjusted r² .177 .144 .362 .060 .201 .183 .199 .336 .268 .149 .426 .077 .103 .292 Condition index 11.11 19.13 19.73 24.64 15.50 17.48 17.46 9.80 10.33 12.79 8.75 9.29 10.90 9.33 N 95 752 1353 95 752 1353 95 752 1353 95 752 1353 94 727 982 * p<0.1, ** p<0.01, *** p<0.001 The values shown are standardized beta coefficients. 57 5. DISCUSSION 5.1. Findings Findings from this study are explanatory and need to be interpreted with caution. The data suggest that, in the unique context of a highly connected society, digital access does not have a statistically significant association with the two outcomes in which I am interested. The finding in this study is compatible with previous research that asserted that the digital divide might shift. Data examined in this empirical study suggests that this also seems to hold for digitally marginalized groups. The literature review showed that current research considers the digital divide as a multifaceted phenomenon, with multiple factors shaping it. Findings of this study support this view. Usage of both types of COVID-19 related services, stimulus check requesting services and information services, among South Korean farmers and fishers, was associated with multiple variables across sociodemographic, access, skill, and usage levels. Thus, any policy seeking to reduce digital divides must consider all elements at each level and stage. Moreover, the evidence suggests that addressing digital divides will require multiple perspective actions at the same time, as a previous study had asserted (Epstein et al., 2011; Van Deursen & Van Dijk, 2015). 1) Sociodemographic Factors, Attitudes, and Digital Skills Research question1-1 and Research question1-3 asked whether sociodemographic factors (age, education level, income level, and attitudes toward digital technology) and digital skills (medium-related and usage) are associated with obtaining COVID-19-related information on stimulus check requesting services and information services among South Korean farmers and fishers. The result of this study showed that the above indicators were associated with the digital usage divide regarding COVID-19-related services. As many digital divide studies have demonstrated (Ahn, 2006; Dewan & Riggins, 2005; Hargittai, 2002; Hargittai & Hinnant, 2008; Hargittai & Walejko, 2008; Lee & Youk, 2014; Van Dijk, 2020; Van Dijk & Hacker, 2003), sociodemographic factors are associated with digital usage gaps. This study is compatible with these findings. It is consistent with the argument that existing structural inequalities reinforce digital inequalities, as in Van Dijk’s resources and appropriation theory (2005) and previous studies (Ahn, 2006; Lee & Youk, 2014; Van Dijk, 2005). Moreover, attitude towards technology appeared to be one of the significant factors for the usage of both types of COVID-19-related services among respondents. 58 Besides socioeconomic status and attitude towards technology, digital skills were also identified as indicators for the usage of COVID-19 related services among South Korean farmers and fishers. Based on a broad range of alternative specifications, the socioeconomic status and digital skills model was identified as the best explanatory model for the usage of both types of COVID-19 related services. In addition, comparisons among three age groups showed that the best fitting model for each age group in both types of services was either the socioeconomic status subgroup model or digital skills subgroup, despite the variance among age groups. These results show that digital skills have become one of the critical indicators to shape digital usage divides, known as the second-level digital divide, as Hargittai (2002) already asserted. As mentioned above, the level of digital access is close to a saturation rate, even for the digitally marginalized group of farmers and fishers. In contrast, digital skills are strongly associated with COVID-19 related services. Preliminary analyses for this study also demonstrated the demographic differences are associated with skills. This is in line with findings by Hargittai & Micheli (2019) and Scheerder et al. (2017). Thus, sociodemographic factors and digital skills are closely related to each other and to the digital usage divide as well. Among sociodemographic factors, attitude towards technology and digital skills, medium-related mobile device skills were associated with both types of COVID-19-related services among South Korean farmers and fishers. This study’s theoretical framework (Van Dijk, 2013) and the previous study (Pearce & Rice, 2013) might explain why medium-related skills are more decisive than usage skills. Digital skills are developed sequentially. Medium-related skills need to develop first. Usage skills develop after a certain threshold level of medium-related skills (Pearce & Rice, 2013; Van Dijk, 2013). Respondents’ digital medium-related skills have not sufficiently developed to acquire high level of digital usage skills enough to have stronger association for COVID-19-related services use. Findings of this study suggest that digital divide policy makers in South Korea might consider designing learning programs of medium-related skills as an introductory class and usage skills as an advanced course for their farmers and fishers. Interestingly, the medium-related skills of the mobile device were more associated with other sociodemographic factors. This can be studied in further research. Among sociodemographic factors, age showed an interesting association in this study. As the correlation matrix showed, age is highly correlated with several of the other explanatory variables. Other analysis methods, such as path analysis or structural equation modeling, which 59 is beyond the scope of this thesis, might help disentangle these associations. In this thesis, a different, less demanding, approach was chosen. To examine the association of the age with dependent variables, I performed split sample analyses. The various model specifications did not identify a significant association between age and dependent variables when all respondents were included. However, I found an effect of age in the differentiated analyses of age groups. This is plausible given that age is one of the identified sociodemographic factors to the digital divides from previous studies (Ahn, 2006; Dewan & Riggins, 2005; Hargittai, 2002; Hargittai & Hinnant, 2008; Hargittai & Walejko, 2008; Lee & Youk, 2014; Van Dijk & Hacker, 2003). It was also identified as one of the influential factors in my own cross-tabulation of the usage of COVID-19- related services in this study. More research could help discern additional reasons for these variations. In addition, attitudes toward technology showed significant associations in this study. The relationship between attitude and material access, internet skills, and uses was also identified in several studies (Van Deursen & Van Dijk, 2015; 2019). The strength of association increased as the age of respondents got older with both types of COVID-19-related services used in age group comparison result models. That is, the attitude towards digital technology is a more crucial factor in older than younger farmers and fishers in South Korea. There might be less variance in attitude towards digital technology as respondents are young, and most of them are favorable to technology since the younger generation is more familiar with it as they encountered it when they were children. On the other hand, the older generation is not as friendly as the younger, so the level of attitude might vary, leading to the strength of association. Furthermore, a favorable attitude towards digital technology might give respondents a motivation to use the services. Only having internet access and a device to access the internet cannot guarantee internet use (Van Dijk, 2013). Under the digital material access saturation level in this study, as we saw in a previous discussion, the attitude toward technology might play a vital role in leading to the actual usage of COVID-19-related services. These differential associations and gaps might give insights to policymakers in South Korea when they design a digital divide policy for farmers and fishers. 2) Digital Access Research question1-2 asked whether the level of digital material access (internet and device) is associated with obtaining COVID-19-related information on stimulus check requesting 60 services and information services among South Korean farmers and fishers. This study found no association between digital material access and the usage of both COVID-19 related stimulus check requesting and information services (except when the analysis was restricted to the digital access variables subgroup). The availability of an internet connection, either narrow or broadband, and possession of devices as a means of accessing the internet is a necessary condition for internet and digital technology use. Initial digital divide studies focused on the digital access level (Ahn, 2006; Dewan & Riggins, 2005; Van Dijk, 1999; Van Dijk & Hacker, 2003), and digital access was considered an influential primary factor of the digital divide (Van Dijk, 1999). Van Deursen and Van Dijk (2015) demonstrated that the broader property of material access is critical to shaping the digital divide. This conclusion not only holds in developing countries where the variance of internet connection rate and possession of devices is high but often also in societies and places where the average adoption rates are high. Material access, specifically device access, is still considered a significant indicator for the digital usage divide, according to previous studies (Pearce & Rice, 2013; Van Deursen & Van Dijk, 2015). In contrast, this empirical study did not find a statistically significant association of connectivity and device access, even for one of the digitally marginalized groups in South Korea, farmers and fishers. Once other factors were introduced, digital material access showed was not associated with the usage of COVID-19 related services. South Korean farmers and fishers have already reached a saturation level on internet connection (99.1%) and possession of the mobile phone (98.1%). Given ubiquitous availability, connectivity and devices access do not serve as potential predictors for my dependent variables. This result shows two things First, after a certain saturation rate of internet connection and possession of at least one device to access the internet, digital material access might have no association with activities that are less demanding, such as accessing the COVID-19-related services analyzed in this study. Requesting stimulus check services is slightly more demanding than information services since users must find and visit a specific website and then supply personal information to apply. However, this is less sophisticated work compared to taking part in or hosting a virtual meeting. Whether the existing level of connectivity is sufficient may vary depending on the types of digital activity and the country’s social, economic, technological, and cultural circumstances. 61 Second, the South Korean government’s efforts and investment allowed building internet network infrastructure in rural areas. This lowered access barriers for user groups. From 2001, an initiative named the ‘Information Network Village Program’ (Park, 2015) has distributed personal computers and set up broadband internet networks to mitigate digital access divide at rural areas. In addition, local governments in South Korea continue to implement public Wi-Fi in public spaces, including buses, bus stops, public health centers, and libraries (Park, 2020). This enables people to use the Internet or digital technology despite their economic constraints. Consequently, the level of digital material access does not contribute to explaining variations in digital uses of COVID-19-related services, even for South Korean farmers and fishers. Some associations between possession of the device and usage of COVID-19-related services were identified, unlike the internet connectivity, even though most respondents answered that they could connect both the internet at home and have their mobile phone. With a saturation rate of both internet connection and at least one technical means to access the internet, the variance in association levels of the device might be because different devices offer different attributes and affordances (Pearce & Rice, 2013). Digital divide studies suggest that personal computers, either desktop or laptop, are superior to mobile devices for a range of digital activities (Chigona et al., 2009; Donner, 2015; Napoli & Obar, 2014; Pearce & Rice, 2013; Rice et al., 2022; Tsetsi & Rains, 2017). The presumption comes from limited storage capacity and the slow and less reliable web-browsing speed of mobile devices (Napoli & Obar, 2014; Tsetsi Rains, 2017). Thus, from this perspective, only having access to the Internet and mobile devices might not be sufficient to use specific services. Moreover, according to the data, slightly more than half of respondents (51.3%) only have a desktop compared to 98.1% who have a mobile phone. The different rates of possession of desktop and mobile phones among respondents might lead to a variance of association level in devices when respondents use COVID-19-related services because the capability of a desktop is better for processing multiple digital activities than a mobile phone. However, recent studies argue that the inferiority of mobile phones compared to personal computers has decreased, and mobile phones can be superior in some limited activities due to the changing environment and technological improvement (Al Ghamdi et al., 2016; Kortum & Sorber, 2015; Quaglione et al., 2020; Rice et al., 2022). Based on these previous studies, examining how each device’s type, characteristics, affordances, and capability are related to the 62 usage of COVID-19-related services might be examined in a future study, improved from this empirical study. 3) Digital Usage Research question1-4 asked whether the extent to the use of digital technologies and services is associated with obtaining COVID-19-related information on stimulus check requesting services and information services among South Korean farmers and fishers. The association of the level of digital usage was less prominent than other subgroup factors. However, certain types of activities showed a strong association with the usage of COVID-19-related services among South Korean farmers and fishers. The most substantial activity was everyday life information activity, which included the usage of weather, news articles, maps, public transportation information, public service, and online banking. Everyday life information activity contains typical and the most frequently used services to users, closely related to our daily lives compared to other types of digital activities measured in this study. Thus, the strong association between everyday life information activity and COVID-19-related services use might suggest that heavy digital technology users are more likely to use COVID-19-related services. This interpretation is aligned with several previous studies that demonstrated internet usage level is related to digital usage gap (Scheerder et al., 2017; Wei & Hindman, 2011). Specifically, regarding the usage of COVID-19 related stimulus check requesting services, the negative association with information seeking and sharing activities were recognized in the full model. This association might be related to the low-income level of respondents in this study. According to the Korean Ministry of Economy and Finance, the average monthly income of the Korean population in the fourth quarter of 2021 was 4.642 million won (₩, KRW) (approximately $3,551.64, USD). Compared to our respondents, 60.2% earned monthly less than 2.99 million won (₩, KRW) (approximately $1,523.54, USD), with the average of the population, farmers and fishers having lower income. Since low-income users are likely to use less in information seeking and sharing activities (Van Deursen & Van Dijk, 2015), the negative association can be understood from the relationship between socioeconomic status and the digital divide. This also shows that the digital usage gap is a multidimensional phenomenon. 63 5.2. Limitations Since the digital divide is a multidimensional phenomenon and is influenced by several factors simultaneously, other model structures should be explored in further studies. Based on prior research, each level of digital access, sociodemographic factors, digital skills, and digital activities influence the usage of internet activities (Donner, Gitau, & Marsden, 2011; Pearce & Rice, 2013; Van Deursen & Van Dijk, 2019). Only conducting multivariate regression analyses cannot identify all effects of independent factors, including an interaction between independent variables, moderate effect, and mediate effect, on the dependent variables, obtaining COVID-19- related information among South Korean farmers and fishers. More advanced data analysis methods, such as structural equation model (SEM) or path analysis, might be helpful to examine additional indirect effects and associations with the usage of COVID-19-related information. Moreover, several multicollinearity issues in the results models were present, and the apparent reason for multicollinearity cannot be explained. In some instances, such as medium- related skills in a mobile device and usage of digital activities, an examination of the correlation matrix provides hints to its causes. In other instances, multicollinearity is based on more complex, direct and indirect patterns of association. The correlation matrix is not sufficient to fully discern them, and additional, supplementary analyses will be necessary to disentangle the causes. In addition, this study examined the association of selected digital activities with dependent variables. Additional dimensions could be analyzed. For example, digital activity usage can be measured in three types (Blank & Groselj, 2014), which are amount (actual time of usage), variety (number of activities in digital usage), and type (actively or passively use). The only variety of digital activity usage was measured and analyzed limitedly in this study, and the amount and type of digital usage were not considered. If these wide ranges of digital usage were considered, a more extended understanding of the determinants of the digital usage divide can be achieved. Furthermore, digital activity usage was only measured by the frequency of each seven types of activities across the device types in this empirical study. According to previous studies, different types of devices may lead to differentiation in activity and use (Pearce & Rice, 2013; Tsetsi & Rains, 2017; Van Deursen & Van Dijk, 2019). Research has shown a relationship between device type and the level of engagement in a variety of online activities (Rice et al., 2022). For example, entertainment activities like gaming are more pervasive on mobile phones 64 (Adepu & Adler, 2016), and personal computers are more task- or work-based activities (Murphy et al., 2016; Pearce & Rice, 2013; Van Deursen & Van Dijk, 2019; Zillien & Hargittai, 2009). Because of data limitations, this study could not examine these effects. If the variance in prevalent use depending on device types was considered, the aspects of devices and activities on the digital usage divide might be recognized. There are some limitations regarding the data as well. The study was not conducted separately on farmers and fishers in South Korea. Also, the study had not considered how rural the respondents’ residences were. Since the dataset only measured the province, like states in the United States, the variations in development in the region where respondents live could not be considered, although living in rural areas might be one of the factors of the digital divide. The data used in this study was cross-sectional, despite the variety and volume of the dataset. The annual report from the NIA regarding the digital divide in South Korea has collected and measured multiple various digital divide-related variables over 20 years from digitally marginalized groups. However, the respondents for each year change, thus NIA generates a series of cross-sectional data. Thus, comparisons across years are limited. Collecting, measuring, and analyzing longitudinal data might generate additional insights regarding the digital usage divide in highly connected countries in further studies. 65 6. CONCLUSION This study aimed to understand how South Korean farmers and fishers use digital technologies and services to obtain COVID-19-related information. The main purpose of this study is to improve our understanding of digital inequalities in the unique environment of a nation that has achieved high digital connectivity. The findings lend credence to the hypothesis that the digital divide transitions to a higher level of the system. The study found no significant impact of digital access to specific digital use after a certain saturation level of connectivity and device access. Digital access has been specified as one of the influential primary factors shaping the digital divide since the digital divide was defined. However, no association between digital access and the usage of COVID-19- related services was found in this study. This is remarkable, as farmers and fishers are considered one of the digitally marginalized groups in South Korea. Gap in the usage of COVID-19 related services among South Korean farmers and fishers did exist, though but they were rooted in other factors. South Korea is an example that the initial model of digital divides that emphasized access needs to be modified. In future study, it would be interesting to examine the conditions or thresholds when digital access fades to serve as an indicator of the digital divide. It is also possible that future technologies and services require higher quality access so that access conditions might again become relevant. While access did not have a statistically significant association with outcomes in the models with the highest explanatory power, the association of sociodemographic factors and digital skills on the digital usage divide was corroborated. Specifically, digital skills associated with the use of information and services offered by government (Van Deursen & Van Dijk, 2009). Thus, as seems plausible, mitigating the digital access barrier loses its effectiveness in narrowing digital divides. As research has argued, digital divides are more multifaceted and complicated. As conjectured in the conceptual framework, existing structural inequalities worsen digital divides. Therefore, measures to weaken or break these structural inequalities will be important. Moreover, it is likely that there is continuous feedback from digital inequalities into sociodemographic inequalities. Breaking these continuous cycles will be an important path for digital divide policy. In addition to this general insight, this study supplies insights that can help in designing more effective digital divide policies to reduce the usage gap for South Korean farmers and 66 fishers. As mentioned, the current digital divide policy in South Korea focuses on older and disabled people. No customized digital divide policy exists for other digitally marginalized groups, including farmers and fishers. Among farmers and fishers, the strength of factors associated with the usage of COVID-19-related services by age groups varied. For instance, the strength of attitudes towards technology varied with COVID-19-related services among age groups. Based on this result, for above 60 years old farmers and fishers, making digital services attractive and providing user-friendly applications (Van Deursen & Van Dijk, 2015) to improve favorable attitudes towards technology might be one way to mitigate the digital usage gap. Thus, policymakers must understand the characteristics of each digitally marginalized group and consider which policy actions and programs might be the most helpful to them. Providing adequate literacy programs might further reduce the gap in obtaining financial resources and health information among farmers and fishers. 67 BIBLIOGRAPHY Abdulai, A. F., Tiffere, A. H., Adam, F., & Kabanunye, M. M. (2021). COVID-19 information- related digital literacy among online health consumers in a low-income country. International journal of medical informatics, 145, 104322. Adepu, S., & Adler, R. F. (2016, October). A comparison of performance and preference on mobile devices vs. desktop computers. In 2016 IEEE 7th Annual Ubiquitous Computing, Electronics & Mobile Communication Conference (UEMCON) (pp. 1-7). IEEE. Ahn, J. (2006). Digital Divide and Digital Literacy on the Perspective of Audience Welfare. Korean Journal of Communication & Information, 36, 78-108. Al Ghamdi, E., Yunus, F., Da’Ar, O., El-Metwally, A., Khalifa, M., Aldossari, B., & Househ, M. (2016). The effect of screen size on mobile phone user comprehension of health information and application structure: an experimental approach. Journal of medical systems, 40, 1-8. Baek, J. Y., Jang, E. G., & Lee, J. M. (2022). The study of consumer types according to the level of digital divide. Journal of Digital Convergence, 20(4), 193-202. Berkman, N. D., Davis, T. C., & McCormack, L. (2010). Health literacy: what is it? Journal of health communication, 15(S2), 9-19. Blank, G., & Groselj, D. (2014). Dimensions of Internet use: amount, variety, and types. Information, Communication & Society, 17(4), 417-35. Chigona, W., Beukes, D., Vally, J., & Tanner, M. (2009). Can mobile internet help alleviate social exclusion in developing countries? The electronic journal of information systems in developing countries, 36(1), 1-16. Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS quarterly, 319-340. Dewan, S., & Riggins, F. J. (2005). The digital divide：Current and future research directions. Journal of the Association for information systems, 6(12), 298-337. DiMaggio, P. & Hargittai, E. (2001). “From the ‘digital divide’ to ‘digital inequalities’: Studying Internet use as penetration increases.” Princeton: Center for Arts and Cultural Policy Studies, Woodrow Wilson school, Princeton University, 4(1), 4-2 Donner, J. (2015). After access: Inclusion, development, and a more mobile Internet. MIT press. Donner, J., Gitau, S., & Marsden, G. (2011). Exploring mobile-only Internet use: Results of a training study in urban South Africa. International Journal of Communication, 5, 574- 597. Epstein, D., Nisbet, E. C., & Gillespie, T. (2011). Who's responsible for the digital divide? Public perceptions and policy implications. The Information Society, 27(2), 92-104. 68 Eysenbach, G. (2001). What is e-health? Journal of medical Internet research, 3(2), e833. Fang, M. L., Canham, S. L., Battersby, L., Sixsmith, J., Wada, M., & Sixsmith, A. (2019). Exploring privilege in the digital divide: implications for theory, policy, and practice. The Gerontologist, 59(1), e1-e15. Ferro, E., Helbig, N. C., & Gil-Garcia, J. R. (2011). The role of IT literacy in defining digital divide policy needs. Government Information Quarterly, 28(1), 3-10. Goel, R. K., & Nelson, M. A. (2021). COVID-19 internet vaccination information and vaccine administration: Evidence from the United States. Journal of Economics and Finance, 45(4), 716-734. Gunkel, D. J. (2003). Second thoughts: toward a critique of the digital divide. New media & society, 5(4), 499-522. Hampton, K. N., Robertson, C. T., Fernandez, L., Shin, I., & Bauer, J. M. (2021). How variation in internet access, digital skills, and media use are related to rural student outcomes: GPA, SAT, and educational aspirations. Telematics and Informatics, 63, 101666. Han, S. M., Han, S. E., Lim, S. H., & Lee, S. H. (2022, July 20). Widening digital divide isolates the elderly. Korea JoongAng Daily. https://koreajoongangdaily.joins.com/2022/07/20/englishStudy/journalism%20internship/ KIS2021-Widening-digital-divide-isolates-the-elderly/20220720154137997.html Han, S., & Nam, S. I. (2021). Creating supportive environments and enhancing personal perception to bridge the digital divide among older adults. Educational Gerontology, 47(8), 339-352. Hargittai, E. (2002). Second-level Digital Divide: Differences in People’s Online Skills, First Monday, 7(4) Hargittai, E., & Hinnant, A. (2008). Digital Inequality: Differences in Young Adults’ Use of the Internet. Communication Research, 35, 602-621 Hargittai, E., & Micheli, M. (2019). Internet skills and why they matter. In M. Graham & W. H. Dutton (Eds.), Society and the Internet: How networks of information and communication are changing our lives (pp. 109-124). Oxford University Press. Hargittai, E., & Walejko, G. (2008). The Participation Divide: Content Creation and Sharing in the Digital Age. Information, Communication, & Society, 11(2), 239-256 Hong, Y. A., Zhou, Z., Fang, Y., & Shi, L. (2017). The digital divide and health disparities in China: evidence from a national survey and policy implications. Journal of medical Internet research, 19(9), e7786. Hwang, J. H. (2019). The Effect of Digital Divide on Life Satisfaction of Disabled People. Journal of Informatization Policy, 26(3), 53-68. 69 Jang, S. J. (2016). Examining Factors Affecting the Digital Divide of the Elderly. Media Perform. Arts, 11, 146-170. Jobst, N. (April 29, 2022). Internet usage in South Korea - Statistics & Facts, Statista. https://www.statista.com/topics/2230/internet-usage-in-south-korea/#dossierKeyfigures Katz, J. E. & Rice, R. E. (2002). Social consequences of Internet use, access, involvement, and interaction. Cambridge, MA, US: MIT Press. Keum, H. J., & Cho, J. H. (2010). Smartphone, Communication Gap, and Political Participation: The Moderating Role of Smartphone Use in the Effects of Social Media on Participation. Korean Journal of Journalism & Communication Studies, 54(5), 348-371. Kim, B. Y., & Kwon, J. S. (2022). The Mediating Effect of the Outcome of Digital Device Use between Digital Informatization & Life Satisfaction. Journal of Informatization Policy, 29(2), 91-105. Kim, H. J., & Lee, J. M. (2018). Determinants of Mobile digital information usage among senior consumers: Focusing on secondary digital divide. Family and Environment Research, 56(6), 493-506. Kim, N. J. (2020 December 4). Resolution for Digital Divide in Contactless Society. Issue and Point, 1774, National Assembly Research Service, ISSN 2005-744X. Kim, S. H., & Sung, W. J. (2020). A study on the mobile internet use and digital divide of farmers and fishers. Informatization Policy, 27(3), 19-38. Kim, W. S., & Lee, S. B. (2021). The Influence of Media Use Characteristics on Social Relationships and School Adjustment through Adolescents' Self-Efficacy and Commitment. International journal of advanced smart convergence, 10(1), 38-55. Ko, J., Kang, W., & Lee, J. (2021). Research Trend Analysis of Digital Divide in South Korea. Journal of Korean Library and Information Science Society, 52(4), 179-203. Koh, H. (2017). A Study on the Digital Divide and Life Satisfaction: Focusing on Generation, SES, and an Urban-Rural Comparison. Journal of the Korea Contents Association, 17(5), 633-641. Koo, J.(2007). A Study on Digital Divide of Aged Women. Information Society & Media, (11), 1-24. Korea Communications Commission. (2022, February 8). Media Usage Behavior Research in Korea. ISSN 2005-498X https://kcc.go.kr/user.do?mode=view&page=A02060100&dc=&boardld=1027&boardSe q=52581 Korea Ministry of Interior and Safety. (n.d.). Korean population age distribution, Retrieved March 20, 2023, from https://jumin.mois.go.kr/ageStatMonth.do 70 Korea National Information Society Agency (NIA) & Korea Ministry of Science and ICT. (2022, March 24). 2021 Report on the Digital Divide. (2021 디지털정보격차 실태조사) https://www.nia.or.kr/site/nia_kor/ex/bbs/View.do;jsessionid=0F2859B4AA588CAFFF0 9D668BBE4F31B.c6f38c20c4b206361143?cbIdx=81623&bcIdx=24287&parentSeq=24 287 Kortum, P., & Sorber, M. (2015). Measuring the usability of mobile applications for phones and tablets. International Journal of Human-Computer Interaction, 31(8), 518-529. Lee, D. & Kim, J. (1997). Digital divide between regions and informatization of rural areas. Journal of Rural Development/Nongchon-Gyeongje, 20(2), 1-26 Lee, M. J. (2014). A Study on Determinants Factors of SNS Public Health Information: Based on the Digital Divide. [Master’s thesis, Ewha Womans University] Major in Mass Communication, The Graduate School of Policy Science, Ewha Womans University, Seoul. Lee, J. (July 19, 2017). Why does South Korea have faster Internet for a cheap price tag? Public Knowledge, https://publicknowledge.org/why-does-south-korea-have-faster-internet-for- a-cheaper-price-tag/ Lee, S. J., & Youk, E. H. (2014). Digital Capability Divide and Digital Outcome Divide: Gaps in the Digital Capability and its Effects on Informational Support. Korean Journal of Journalism & Communication Studies, 58(5), 206-232. Lee, W. (2013). Conceptualizations and Policy Directions of Participation Divide between Smart Mobile Media Users. 21st Century Political Science Review, 23(2), 371-395. Levy, H., Janke, A. T., & Langa, K. M. (2015). Health literacy and the digital divide among older Americans. Journal of general internal medicine, 30(3), 284-289. López, L., Green, A. R., Tan-McGrory, A., King, R. S., & Betancourt, J. R. (2011). Bridging the digital divide in health care: the role of health information technology in addressing racial and ethnic disparities. The Joint Commission Journal on Quality and Patient Safety, 37(10), 437-445. Lu, M. T. (2001). Digital divide in developing countries. Journal of Global Information Technology Management, 4(3), 1-4. Mackert, M., Love, B., & Whitten, P. (2009). Patient education on mobile devices: an e-health intervention for low health literate audiences. Journal of Information Science, 35(1), 82- 93. Mackert, M., Mabry-Flynn, A., Champlin, S., Donovan, E. E., & Pounders, K. (2016). Health literacy and health information technology adoption: the potential for a new digital divide. Journal of medical Internet research, 18(10), e6349. Mitchell, U. A., Chebli, P. G., Ruggiero, L., & Muramatsu, N. (2019). The digital divide in health-related technology use: The significance of race/ethnicity. The Gerontologist, 59(1), 6-14. 71 Muchran, M., & Ahmar, A. S. (2019). Application of TAM model to the use of information technology. arXiv preprint arXiv:1901.11358. Murphy, H. C., Chen, M. M., & Cossutta, M. (2016). An investigation of multiple devices and information sources used in the hotel booking process. Tourism management, 52, 44-51. Napoli, P. M., & Obar, J. A. (2014). The emerging mobile Internet underclass: A critique of mobile Internet access. The Information Society, 30(5), 323-334. National Information Society Agency (NIA). (2021). 2021 Report on the Digital Divide. Seoul, Republic of Korea: Ministry of Science and ICT & Ministry of the Interior and Safety, available at https://www.nia.or.kr/site/nia_kor/ex/bbs/View.do?cbIdx=81623&bcIdx=24287&parentS eq=24287 National Telecommunications and Information Administration (NTIA). (1999). Falling through the Net: Defining the Digital Divide. Washington, DC: US Department of Commerce, available at http://www.ntia.doc.gov/ntiahome/fttn99/contents.html. Neter, E., & Brainin, E. (2012). eHealth literacy: extending the digital divide to the realm of health information. Journal of medical Internet research, 14(1), e1619. Newhagen, E. J., & Bucy, E. P. (2004). Routes to media access. In E. P. Bucy & J. E. Newhagen (Eds.). Media access: Social and psychological dimensions of new technology use (pp.3- 23). London: Lawrence Erlbaum Associates. Norris, P. (2001). Digital Divide: Civic engagement, information poverty, and the Internet worldwide. Cambridge: Cambridge University Press. Park, E. Y. (2022). Effect of COVID-19 on Internet Usage of People with Disabilities: A Secondary Data Analysis. International Journal of Environmental Research and Public Health, 19(13), 7813. Park, S. J. (2020, December 14). S. Korea establishes free nationwide WiFi network on public buses for first time. Aju Business Daily. https://www.ajudaily.com/view/20201214144523898 Park, S. Y., & Kim, J. W. (2019). A Panel Analysis (2010–2018) of the Digital Divide among Different Age Groups among Older Adults. J. Korean Data Anal. Soc, 21, 2407-2418. Park, Y. (2015). The Correlation between the Role of Local Government to the Operational Performance of Information Network Village Project (INVIL). Journal of Korean Association for Regional Information Society, 18(3), 95-122. Park, Y. & Nam, K. (2016). Strategies of Local Government for Solving Rural Smart Divide: Focusing on Information Network Village (INVIL) Project. Journal of Korean Association for Regional Information Society, 19(3), 1-26. Pearce, K. E., & Rice, R. E. (2013). Digital divides from access to activities: Comparing mobile and personal computer Internet users. Journal of communication, 63(4), 721-744. 72 Quaglione, D., Matteucci, N., Furia, D., Marra, A., & Pozzi, C. (2020). Are mobile and fixed broadband substitutes or complements? New empirical evidence from Italy and implications for the digital divide policies. Socio-Economic Planning Sciences, 71, 100823. Rains, S. A. (2008). Health at high speed: Broadband Internet access, health communication, and the digital divide. Communication Research, 35(3), 283-297. Rice, R. E., Pearce, K. E., & Calderwood, K. J. (2022). W (h) ither the device divide? Changing relationships between personal computer or mobile device with online activities. Mobile Media & Communication, 20501579221142134. Sala, E., Gaia, A., & Cerati, G. (2022). The gray digital divide in social networking site use in Europe: Results from a quantitative study. Social Science Computer Review, 40(2), 328- 345. Scheerder, A., Van Deursen, A., & Van Dijk, J. (2017). Determinants of Internet skills, uses and outcomes. A systematic review of the second-and third-level digital divide. Telematics and informatics, 34(8), 1607-1624. Schmidt-Hertha, B., & Strobel-Dümer, C. (2014). Computer literacy among the generations: how can older adults participate in digital society? In Challenging the 'European Area of Lifelong Learning' (pp. 31-40). Springer, Dordrecht. Selwyn, N. (2003). Apart from technology: understanding people’s non-use of information and communication technologies in everyday life, Technology in Society, 25(1), 99-116. Selwyn, N. (2004). Reconsidering Political and Popular Understanding of the Digital Divide. New Media & Society, 6(3), 341-362. Seol, J., Yu, D., & Lim, G. G. (2018). The Effect of Personal Characteristics on Use Outcomes of Smart Device: Moderating Effects of Mobile Divide. Journal of Information Technology Services, 17(2), 17-33. Statista. (2022, October 24). Share of elderly population aged 65 years and above in South Korea from 2010 to 2022. https://www.statista.com/statistics/995650/south-korea-elderly- population-share/ Statista. (2023, January 6). Enrollment rate in university in South Korea from 2013 to 2022. https://www.statista.com/statistics/629032/south-korea-university-enrollment- rate/#:~:text=At%20around%2073.8%20percent%20as,over%20the%20past%20few%20 years. Statista (2023, February 24). Worldwide digital population 2023. https://www.statista.com/statistics/617136/digital-population-worldwide/ Strover, S. (2001). Rural internet connectivity, Telecommunications policy, 25(5), 331-347. Tichenor, P. J., G. A. Donohue, and C. N. Olien (1970). Mass Media Flow and Differential Growth in Knowledge. Public Opinion Quarterly, 34, 159-170. 73 Townsend, L., Sathiaseelan, A., Fairhurst, G., & Wallace, C. (2013). Enhanced broadband access as a solution to the social and economic problems of the rural digital divide. Local Economy, 28(6), 580-596. Tsetsi, E., & Rains, S. A. (2017). Smartphone Internet access and use：Extending the digital divide and usage gap. Mobile Media & Communication, 5(3), 239-255. Van Deursen, A. J., Helsper, E., Eynon, R., & Van Dijk, J. A. (2017). The compoundness and sequentiality of digital inequality. International Journal of Communication, 11, 452-473. Van Deursen, A. J., & Van Dijk, J. A. (2009). Improving digital skills for the use of online public information and services. Government information quarterly, 26(2), 333-340. Van Deursen, A. J., & Van Dijk, J. A. (2015). Toward a multifaceted model of internet access for understanding digital divides: An empirical investigation. The Information Society, 31(5), 379-391. Van Deursen, A. J., & Van Dijk, J. A. (2019). The first-level digital divide shifts from inequalities in physical access to inequalities in material access. New media & society, 21(2), 354-375. Van Dijk, J. (1999). The network society, Social aspects of new media. Thousand Oaks, CA: Sage. Van Dijk, J. (2005). The deepening divide: Inequality in the information society. Thousand Oaks, CA: Sage. Van Dijk, J. (2013). Chapter 2 A theory of the digital divide. In The digital divide (pp. 49-72). Routledge. Van Dijk, J. (2020). The digital divide. John Wiley & Sons. Van Dijk, J., & Hacker, K. (2003). The digital divide as a complex and dynamic phenomenon. The information society, 19(4), 315-326. Wellman, B. and Berkowitz, S.D. (Eds.). (1988). Social structures, A network approach. London: JAI Press. World Bank. (n.d.). GDP (current US$) - Korea, Republic. Retrieved March 20, 2023, from https://data.worldbank.org/indicator/NY.GDP.MKTP.CD?locations=KR Wyatt, S., Henwood, F., Hart, A., & Smith, J. (2005). The digital divide, health information and everyday life. New Media & Society, 7(2), 199-218. Wei, L., & Hindman, D. B. (2011). Does the digital divide matter more? Comparing the effects of new media and old media use on the education-based knowledge gap. Mass Communication and Society, 14(2), 216-235. Wei, K. K., Teo, H. H., Chan, H. C., & Tan, B. C. (2011). Conceptualizing and Testing a Social Cognitive Model of the Digital Divide. Information Systems Research, 22(1), 170-187. 74 Yim, J. H., Lee, H. J., Lee, J. H., & Lee, J. H. (2021). The Influence of Intelligence Information Society Perception Expectation Anxiety on Life Satisfaction among the Elderly: Moderating effect of On/Offline Social Capital. Journal of Information Technology Services, 20(1), 117-130. Youn, S. M., Song, C. H., Kim, Y. K., & Kim, J. H. (2004). Mobile Phone Divide. Korean Journal of Journalism & Communication Studies, 48(3), 354-378. Zhang, X. (2013). Income disparity and digital divide: The Internet Consumption Model and cross-country empirical research. Telecommunications Policy, 37(6-7), 515-529. Zillien, N., & Hargittai, E. (2009). Digital distinction: Status‐specific types of internet usage. Social Science Quarterly, 90(2), 274-291. 75 APPENDIX A: CODE BOOK OF VARIABLES THAT USED IN THIS STUDY Table A1 Independent variable definitions and measurement Variable Category Questionnaire/Label Measurement name Desktop MA1B1 Possession of desktop 0.Don’t Have Laptop MA1B2 Possession of laptop 1. Have 0.Have nothing Mobile 1. Have mobile Digital MA2B Possession of mobile phone phone (smartphone or Access_Device featurephone) 0. Have nothing Smart Possession of smart devices (tablet, smart watch, AI speaker, smart MA3B 1. Has 1 type devices peripheral device) 2. Have all kinds Digital Access_Internet 0. Unavailable IA Availability of using the internet at home connection 1. Available I can install, update, and delete the necessary software on my MRS1P1 computer (desktop or laptop). I can set up wired or wireless internet network on my computer Please indicate MRS1P2 (desktop or laptop). your level of I can set preferences such as blocking pop-ups, text size, and security digital skills in Digital Skills_Medium MRS1P3 of web browsers (Chrome/Safari/Microsoft Edge) on my computer each activity related_Personal (desktop or laptop). 1. Not at all computer I can connect and use various external devices such as digital camera, 2. Not so much MRS1P4 printer, scanner, and USB drive on my computer (desktop or laptop). likely I can send a digital file on my computer (desktop or laptop) to other 3. Somewhat likely MRS1P5 people by using the Internet. 4. Very likely I can scan and repair the malware such as viruses and spyware on my MRS1P6 computer (desktop or laptop) 76 Table A1 (cont’d) Digital Skills_Medium I can write documents and create digital materials such as excel and related_Personal MRS1P7 PowerPoint file on my computer (desktop or laptop). computer I can set preferences such as display, sound, security, alarm, and input MRS2M1 Please indicate method on a smart device (smartphone or tablet). your level of I can set up wired or wireless internet network on a smart device MRS2M2 digital skills in (smartphone or tablet) each activity I can transfer digital files from a smart device (smartphone or tablet) to MRS2M3 1. Not at all a personal computer (desktop or laptop). 2. Not so much Digital Skills_Medium I can send a digital file or photo on a smart device (smartphone or MRS2M4 likely related_Mobile device tablet) to other people. 3. Somewhat I can install, update, and delete applications on a smart device MRS2M5 likely (smartphone or tablet). 4. Very likely I can scan and repair the malware such as viruses and spyware on a MRS2M6 smart device (smartphone or tablet) I can write documents and create digital materials such as notes or MRS2M7 words on a smart device (smartphone or tablet). I can install/remove/upgrade software programs or DS1 copy/delete/transfer/edit files and folders on a computer I can install/remove/upgrade applications on a smart device Please indicate DS2 (smartphone or tablet) your level of I can use applications such as a calculator, calendar, and contacts on a digital skills in DS3 smartphone each activity I can write and share a file by using smart office programs such as 1. Not at all DS4 Digital Skills_Usage Evernote, Google docs, Naver office, MS Office 365, etc 2. Not so much I can host/attend virtual meetings using Google Meet and Zoom likely DS5 applications 3. Average I can utilize smart devices that are linked to a smartphone, such as a 4. Somewhat DS6 likely smart watch, smart fridge, or IoT devices I can distinguish reliable information among results from information- 5. Very likely DS7 seeking activity in search engines by comparing them to other resources 77 Table A1 (cont’d) DS8 I can use references or websites to find out which one is fake news DS9 I can set up settings for filtering harmful content I can modify content file types from original movies, animation, and DS10 music videos for other kinds of videos, such as GIF files I can work with other people by using online applications such as DS11 Google docs I can purchase items by using online payment such as PayPal, Naver DS12 Please indicate Pay, and Kakao Pay I can find a route by using navigation or online map services your level of DS13 digital skills in I can find and participate in a community that I am interested in on a DS14 each activity computer or smart device 1. Not at all Digital Skills_Usage I can discuss or make a petition regarding political/social issues on a DS15 2. Not so much computer or smart device likely DS16 I can set up security settings on a computer or smart device 3. Average I can delete cookies or history in a web browser on a computer or 4. Somewhat likely DS17 smart device 5. Very likely I can set up the extent to which people disclose information when I DS18 upload a post to my social media account or board on a computer or smart device I know how to take temporary measures when I find out some posts DS19 that insult or defame me I know how to report when someone violates my rights (defamation, DS20 copyright infringement) on a web portal or social media U1A1 Searching information or news articles Frequency of use in the last year Digital Use_ Search, U1A2 Email 1. Not used at all email, and content U1A3 Media Contents (Movie, Music, eBooks, and webtoons) 2. Not so much services used U1A4 Educational Contents (Online courses for degrees or hobby) 3. Somewhat used 4. Frequently used 78 Table A1 (cont’d) U2A1 Social Network Service (Twitter, Facebook, Instagram, etc) Frequency of use in the last year Instant Messenger (Kakao Talk, Facebook messenger, Line, Instagram U2A2 1. Not used at all Digital Use_Social DM, Telegram, etc) 2. Not so much media Private Blog (Daum blog, Nate Blog, Tstory, Cyworld, Blogger.com, U2A3 used etc) 3. Somewhat used U2A4 Community (online communities, club, group, etc) 4. Frequently used Everyday life Information (Weather, News, Public transportation, Frequency of use U3A1 maps, find routes, etc) in the last year U3A2 E-commerce (Online shopping, Reservation, Booking tickets, etc) 1. Not used at all Digital Use_Everyday U3A3 Online Banking 2. Not so much life information used U3A4 Public Service 3. Somewhat used 4. Frequently used I have uploaded user-created content such as information, knowledge, Frequency of use U4A1 Digital Use_Information news, video content, or photos made by myself or other users in the last year seeking and sharing I have shared hyperlinks to digital content that I consume on the 1. Not used at all U4A2 Internet 2. Not so much U5A1 I have used the Internet to maintain personal relationships used Digital Use_Networking I have used the Internet to meet and communicate with people who I 3. Somewhat used U5A2 do not know before 4. Frequently used I have used the Internet to express my opinion regarding public issues U6A1 Frequency of use by writing comments, posting blogs, or online discussion in the last year I have used the Internet to provide policy proposals, political U6A2 1. Not used at all Digital Use_Social comments, or civil complaints to the government 2. Not so much engagement I have experience in donating or volunteering activities through the U6A3 used Internet 3. Somewhat used I have experience participating in online voting, opinion poll, or U6A4 4. Frequently used signature campaign through the Internet 79 Table A1 (cont’d) I have experience in activities that help my job-seeking or promotion U7A1 Frequency of use through the Internet in the last year I have experience in marketing activities for my businesses through the U7A2 1. Not used at all Digital Use_Financial Internet 2. Not so much activites I have experience in obtaining financial information that helps the U7A3 used increase of my income through the Internet 3. Somewhat used I have experience in cost reduction activities such as price comparison U7A4 4. Frequently used or group buying through the Internet A1 Digital technology is useful Please indicate A2 Digital technology makes my life more convenient how much you A3 Digital technology is good for me agree with each statement. Attitude towards digital 1. Strongly technology disagree A4 I would like to use digital technology more than now 2. Disagree 3. Agree 4. Strongly agree 80 Table A2 Dependent variable definitions and measurement Variable Category Questionnaire/Label Measurement name Do you know of and have Request service for stimulus check due experience UCI1A to the COVID-19 outbreak to the using this government through the Internet service after Use_COVID19 the COVID-19 related information outbreak (Jan service Information service related to COVID- 2021)? 19, such as the location of testing sites, 0. Unaware UCI2A places that COVID-19-positive patients 1. Aware_Not had visited, etc. use 2. Aware_Use 1st reason for no experience in Please choose requesting service for stimulus checks one among RCIA1 due to the COVID-19 outbreak the following government through the Internet items 2nd reason for no experience in 1. No need requesting service for stimulus checks 2. Difficulty in RCIA2 due to the COVID-19 outbreak the use government through the Internet 3. Cost burden 1st reason for no experience in using 4. Decrease in information service related to COVID- income RCIB1 19, such as the location of testing sites, 5. No device Reason for not places that COVID-19-positive patients 6. No internet use_COVID19 had visited, etc. access related information 7. Physical service restrictions 8. Negative perspective 2nd reason for no experience in using regarding information services related to COVID- Internet use RCIB2 19, such as the location of testing sites, 9. Concerns of places that COVID-19-positive patients harmful had visited, etc. results 10. Others (write it in detail) 81 APPENDIX B: EACH RESULT OF MULTUPLE SIMPLE REGRESSION ANALYSES FOR USAGE COVID-19 RELATED STIMULUS CHECK REQUESTING SERVICES Note: All values shown in table B1 ~ B19 are significant of p<.001. Table B1 Simple regression analysis between possession of desktop and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Possession of .635 .030 .411 21.12 446.12 .169 2200 desktop Table B2 Simple regression analysis between possession of laptop and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Possession of .542 .036 .306 15.05 226.62 .093 2200 laptop Table B3 Simple regression analysis between possession of mobile phone (smartphone or feature phone) and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Possession of 1.010 .119 .179 8.52 72.60 .032 2200 mobile phone Table B4 Simple regression analysis between possession of smart devices and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Possession of .394 .035 .236 11.39 129.65 .056 2200 smart devices 82 Table B5 Simple regression analysis between availability of internet connection at home and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Internet .957 .173 .117 5.55 30.75 .014 2200 connection Table B6 Simple regression analysis between attitude towards technology and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Attitude towards .618 .019 .561 31.80 1011.31 .315 2200 technology Table B7 Simple regression analysis between income level and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Income level .213 .009 .463 24.52 601.22 .215 2200 Table B8 Simple regression analysis between education level and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Education .461 .014 .577 33.15 1098.82 .333 2200 level Table B9 Simple regression analysis between age and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Age -.326 .012 -.489 -26.28 690.86 .239 2200 83 Table B10 Simple regression analysis between medium-related skills in a personal computer and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β H/W .460 .014 .585 33.80 1142.64 .342 2200 skills_PC Table B11 Simple regression analysis between medium-related skills in a mobile device and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β H/W skills_Mobile .527 .012 .670 42.31 1789.79 .449 2200 device Table B12 Simple regression analysis between digital usage skills and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Usage skills .470 .012 .636 38.64 1493.33 .405 2200 Table B13 Simple regression analysis between search, email, and content service activity usage and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Search, email, and content .468 .019 .495 24.18 584.61 .245 1803 service activity 84 Table B14 Simple regression analysis between social media activity usage and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Social media .403 .022 .399 18.49 341.94 .160 1803 activity Table B15 Simple regression analysis between everyday life information activity usage and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Everyday life information .497 .017 .558 28.51 812.60 .311 1803 activity Table B16 Simple regression analysis between information seeking and sharing activity usage and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Information seeking and .271 .019 .318 14.23 202.36 .101 1803 sharing activity Table B17 Simple regression analysis between networking activity usage and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Networking .297 .019 .338 15.24 232.34 .114 1803 activity 85 Table B18 Simple regression analysis between social engagement activity usage and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Social engagement .278 .023 .269 11.84 140.27 .072 1803 activity Table B19 Simple regression analysis between financial activity usage and COVID-19 related stimulus check requesting services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Financial .343 .020 .374 17.09 292.07 .140 1803 activity 86 APPENDIX C: EACH RESULT OF MULTUPLE SIMPLE REGRESSION ANALYSES FOR USAGE COVID-19 RELATED INFORMATION SERVICES Note: All values shown in table C1 ~ C19 are significant of p<.001. Table C1 Simple regression analysis between possession of desktop and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Possession of .696 .033 .415 21.36 456.37 .172 2200 desktop Table C2 Simple regression analysis between possession of laptop and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Possession of .544 .039 .282 13.80 190.45 .080 2200 laptop Table C3 Simple regression analysis between possession of mobile phone (smartphone or featurephone) and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Possession of .914 .129 .149 7.06 49.91 .022 2200 mobile phone Table C4 Simple regression analysis between possession of smart devices and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Possession of .370 .038 .204 9.77 95.46 .042 2200 smart devices 87 Table C5 Simple regression analysis between availability of internet connection at home and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Internet .909 .188 .103 4.85 23.50 .011 2200 connection Table C6 Simple regression analysis between attitude towards technology and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Attitude towards .694 .021 .581 33.47 1119.91 .338 2200 technology Table C7 Simple regression analysis between income level and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Income level .209 .010 .419 21.65 468.49 .176 2200 Table C8 Simple regression analysis between education level and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Education .424 .016 .489 26.26 689.77 .239 2200 level Table C9 Simple regression analysis between age and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Age -.328 .014 -.454 -23.88 570.11 .206 2200 88 Table C10 Simple regression analysis between medium-related skills in a personal computer and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β H/W .476 .015 .558 31.55 995.42 .312 2200 skills_PC Table C11 Simple regression analysis between medium-related skills in a mobile device and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β H/W skills_Mobile .556 .014 .652 40.32 1625.35 .425 2200 devices Table C12 Simple regression analysis between digital usage skills and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Usage skills .498 .013 .622 37.22 1385.52 .387 2200 Table C13 Simple regression analysis between search, email, and content service activity usage and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Search, email, and content .469 .022 .449 21.34 455.45 .202 1803 service activity 89 Table C14 Simple regression analysis between social media activity usage and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Social media .427 .024 .383 17.58 309.18 .147 1803 activity Table C15 Simple regression analysis between everyday life information activity usage and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Everyday life information .504 .020 .513 25.35 642.43 .263 1803 activity Table C16 Simple regression analysis between information seeking and sharing activity usage and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Information seeking and .310 .021 .330 14.85 220.41 .109 1803 sharing activity Table C17 Simple regression analysis between networking activity usage and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Networking .361 .021 .372 17.01 298.21 .138 1803 activity 90 Table C18 Simple regression analysis between social engagement activity usage and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Social engagement .259 .026 .227 9.88 97.61 .051 1803 activity Table C19 Simple regression analysis between financial activity usage and COVID-19 information services Standardized Unstandardized B Coefficients Variable t(p) F(p) R² N Beta B SE β Financial .358 .022 .353 16.03 256.93 .125 1803 activity 91