Data and Participants
The data for this study were obtained from a National Survey of Older Koreans in 2020. The survey has been conducted every three years since 2008, and questions about digital access and usage have been included since 2020. The survey is a nationwide, self-report survey conducted by the Ministry of Health and Welfare of Korea and the Korea Institute for Health and Social Affairs. The survey participants were adults 65 years and above and were stratified by the type of housing and administrative district units. We excluded participants who were unable to live independently through activities of daily living (ADL) and instrumental activities of daily living (iADL). The survey was conducted between September 14, 2020, and November 20, 2020. A trained researcher visited the participants and conducted a 1:1 tablet PC-assisted personal interview. For quality control, a post-survey telephone interview was conducted with 10% of the participants. Institutional review board approval was waived because the national survey of older Korean data are publicly available for research.
Dependent Variable
Since the early stage of ICT distribution, there have been many measurement tools for the digital divide [16]. According to a recent study, the diversity of Internet availability is divided into four factors: motivation, materials, skills, and usage [17]. Specifically, we focused on Internet usage as a major factor. The gap in Internet usage, the so-called “second digital divide,” was evaluated by a range of factors, such as the frequency and type of online activities. In this study, there were a total of ten types of online activities. These included receiving messages (through text messages or message applications), sending messages, searching for information (news, weather), taking pictures or videos, listening to music, playing games, watching videos (movie, television, platform), using social networking services, engaging in online shopping or making reservations, using online banking, and installing applications. However, there is no authorized and comprehensive evaluation tool for the digital divide among older Korean adults. Therefore, in this study, older adults who were able to perform less than three out of ten types of online activities were classified as having “poor digital literacy,” and this cutoff value was defined by a receiver operating characteristic curve. Additionally, we defined “poor digital literacy” not using digital devices, such as personal computers, notebooks, tablets, and smartphones.
Independent Variables
Social participation was evaluated by regular offline participation in activities in the previous year, such as senior community activity, learning activity, religious activity, volunteer work, leisure activity, political group activity, and club activity. Occupation-related or money earning activities were excluded. Based on the number of participating activities, social participation was divided into four groups: no participation, participation in only one activity, participation in two kinds of activities, and participation in more than three kinds of activities. Senior community activities included regular visits to senior community centers or senior halls to communicate or engage in hobbies and entertainment activities with friends of their same age. Learning activities included learning foreign languages, health information, computer classes, and employment education. Religious activities included to every faith-based activity, regardless of religion. Volunteer work included all unpaid social work-based activities, such as disability or elderly service, safety guards, recycling campaigns, and teaching students. Leisure activities included watching movies, performances, sports, playing musical instruments, and traveling. Club activities referred to social gatherings, such as an alumni association or a reading club. Total number of offline social participations, the frequency of offline social participations and the types of offline social participations were investigated.
Covariates
We selected several socioeconomic and health-related characteristics that could affect both independent and dependent variable as covariates. Specifically, gender, age, area of residence, family structure, economic activity, level of education, economic status, alcohol and smoking status, physical activity, depression, self-reported health status, and comorbidity were assessed. Gender was divided into two subgroups: Men and Women. Age was divided into four subgroups: 65–69, 70–74, 75 − 59, and above 80. Area of residence was divided into two subgroups: Metropolitan and province (rural area). Family structure was divided into three subgroups: With their offspring(s), Single elderly household, and Elderly in couple household. Economic activity was classified into two groups, and paid workers were classified as having a job. Level of education was divided into three subgroups: Below elementary school, Middle and High School, and Above university. Economic status was divided into three subgroups: The bottom 25% of household incomes were classified as the Low group, the top 25% as the High group, and the remainder were categorized as the Middle group. Drinking/smoking status were divided into two categories: Non- or ex-drinker/smoker and Current-drinker/smoker. Physical activity was assessed by the question “Do you usually exercise for at least 10 minutes a day?” Depression was evaluated using the Korean version of the Short Form Geriatric Depression Scale [18]. The comorbidity variable was calculated using the number of chronic diseases (hypertension, hyperlipidemia, type II diabetes mellitus, asthma, and cancer). All variables were assessed through self-reporting.
Statistical Analysis
Chi-square tests were used to analyze participant general characteristics. Multivariable logistic regression analysis was conducted to determine the association between the digital divide and number of social participations. Moreover, we performed subgroup analyses considering age and the presence of a depressed mood. We examined the association between the frequency of social participant and digital literacy. We also considered different types of social engagement participants. Adjusted odds ratios (aOR) with 95% confidence intervals (CI) were used for data presentation. There was no evidence of multicollinearity among variables based on variance inflation factor (VIF). Statistical analysis was performed using SAS software (version 9.4; SAS Institute, Cary, North Carolina, USA). A p-value of < 0.05 was determined to be statistically significant.