Social Network Typologies and Digital Literacy Differences among Korean Older Adults: A Latent Class Analysis

Social Network Typologies and Differences in Digital Literacy

Article information

Ann Geriatr Med Res. 2024;28(2):134-143

Publication date (electronic) : 2024 February 22

doi : https://doi.org/10.4235/agmr.23.0174

Hamin Lee ¹, Mi Kyeong Kim ¹, Hae Yean Park ², Kang-Hyun Park^,³

¹Department of Occupational Therapy, Graduate School, Yonsei University, Wonju, Korea

²Department of Occupational Therapy, College of Software and Digital Healthcare Convergence, Yonsei University, Wonju, Korea

³Department of Occupational Therapy, Baekseok University, Cheonan, Korea

Corresponding Author: Kang-Hyun Park, PhD, OTR Department of Occupational Therapy, Baekseok University, 1, Baekseokdaehak-ro, Dongnam-gu, Cheonan 31065, Korea E-mail: kanghyun@bu.ac.kr

This study was presented at the Alzheimer’s Association International Conference 2023 held in Amsterdam, the Netherlands.

Received 2023 October 15; Revised 2024 February 7; Accepted 2024 February 16.

Abstract

Background

This study categorized older Korean adults’ social networks and analyzed their characteristics and digital literacy differences based on type.

Methods

We analyzed data from 9,377 Korean older adult participants of the 2020 National Survey of Older Koreans, and performed latent class analysis (LCA) chi-square and Welch’s F analyses to understand the characteristics of each social network type. The Games–Howell post-hoc test was applied to determine the significance of differences between groups.

Results

The three social network types derived using LCA were “child-centered,” “child-friend,” and “friend-centered.” The digital literacy levels differed significantly according to social network type.

Conclusion

The results of this study can be used to propose intervention programs and services associated with older adults’ social networks by examining their social network types and the corresponding differences in digital literacy.

Keywords: Computer literacy; Korean; Aged; Social network

INTRODUCTION

The rapid growth of digital technology in Korea leads to offline public and private service substitution with digital devices such as smartphones and computers.1) These changes have been accelerated by the coronavirus disease 2019 (COVID-19) impact, which started in 2020, and the Fourth Industrial Revolution. Digital literacy combines literacy, the ability to read and write, and proficiency in digital technology. Glister’s definition encompasses the ability to understand computer-mediated information, evaluate and integrate diverse information, and accurately use it.2)

Although digitalizing animate services brings convenience, it aggravates alienation among individuals vulnerable to such digitalized information, facing difficulty accessing digital information and tools.3,4) The National Information Society Agency provides reports on South Korea’s digital divide. According to their 2021 report, older adults are more vulnerable than low-income earners, disabled individuals, farmers, and fishermen.5) Despite the high demand for online services such as shopping and insurance claims, older adults may face discrimination due to difficulties in digital access.6)

Continuously excluding older people from the benefits of information society could degrade the quality of life of individuals with extended life expectancies.6) With Korea projected to become a super-aged nation by 2025,7) the number of socially isolated older adults is expected to increase owing to reduced social networks with advancing age.8) Thus, preventing social isolation among seniors and improving their quality of life requires fostering digital literacy.

Researchers’ definitions of social networks vary. Ell9) defined a social network as one encompassing an individual’s interpersonal contacts, spanning formal and informal support from family, friends, neighbors, colleagues, and caregivers. This multidimensional concept involves network size, components, frequency of contact with family and friends, aid level, acceptance, and relationship satisfaction.10,11) Social network size decreases with age due to retirement, physical illness, or the death of friends.8)

Maintaining good social networks enhances multiple aspects of older adults’ lives positively. Social networks are linked to cognitive, emotional, and mental health status for older adults.12,13) Moreover, strengthening social networks can help prevent suicide attempts among older adults.14) According to health indicators from the Organisation for Economic Cooperation and Development, Korea has the highest suicide rate among all countries.15) Korea’s suicide rate among those aged 65 years and older is twice the average, underscoring the growing need for research on social networks in older adults.16) Furthermore, older adults’ social networks affect health-promoting behaviors such as reducing smoking and alcohol consumption17) and life satisfaction.18)

Previous research on social networks in older Korean adults has focused on assessing the links between variables such as spouses, children, and friends and outcomes such as depression or health. However, recent research trends have aimed to elucidate the complex social network patterns in older adults. However, few studies have analyzed digital literacy discrepancies among older adults across various social network types. Therefore, this study aimed to categorize the social networks of older Korean adults and examine the characteristics and differences in digital literacy linked to these network types. The goal was to identify the most susceptible network type for digital devices and facilitate targeted interventions.

MATERIALS AND METHODS

Study Data

This study used data from the 2020 National Survey on Older Koreans conducted by the Ministry of Health and Welfare of Korea. The survey was designed to understand the living conditions and needs of older individuals in South Korea, making it suitable for analyzing the characteristics of social networks and digital literacy disparities. This national survey employed a stratified two-stage cluster sampling method for community-dwelling older adults aged ≥65 years. Data were collected through in-person interviews conducted between September and November 2020.

Participants

This study initially included 10,097 Korean adults aged ≥65 years. After excluding 185 individuals with missing digital literacy or social network values, the final analysis included a total of 9,377 individuals.

Study Variables

Social network typology indicator

Latent class analysis was used to identify network types based on earlier social network typology studies. This study included seven social network variables from the 2020 National Survey on Older Koreans, based on the typology described by Kang et al.19) These seven variables were spousal presence, child encounter frequency, child contact, friend encounter frequency, friend contact, club participation, and social activity engagement. For each variable, encounters refer to in-person encounters with a child or friend, whereas contact refers to remote contact with a child or friend, such as phone calls or texts. All variables were coded dichotomously, as follows.

Spousal presence was coded as 0 (no spouse) or 1 (spouse present). The frequency of child encounters was coded as 0 (less than once every three months) or 1 (more than once every three months). The frequencies of child contact, friend encounters, and friend contact were coded as 0 (less than once a week) or 1 (more than once a week). Club and social activity participation was coded as 0 (did not participate in the past year) or 1 (participated in the past year). Club participation refers to club involvement, while social activities involve engagement in social groups such as alumni associations and social gatherings.

Digital literacy

Older adults’ digital literacy was measured using five questions regarding challenges with information service devices. The question “Is online or internet-based information and services usage challenging?” was coded as 0 (yes) or 1 (no). The remaining four questions used a five-point Likert scale reflecting discomfort levels (1=not uncomfortable at all, 2=not uncomfortable, 3=normal, 4=uncomfortable, and 5=very uncomfortable). However, for this analysis, we dichotomously coded the four questions (1 of 1–3 points, 0 of 4–5 points) to standardize the scales across the five questions. The four questions were related to (1) difficulty with online train/bus reservations, (2) difficulty with restaurant orders through kiosks (digital machines), (3) inconvenience caused by ATM usage at banks or fewer offline bank locations, and (4) difficulty caused by an increase in credit card-only establishments. We summed the responses to the five questions, with higher scores indicating higher digital literacy.

Demographic and health status

We used the respondents’ demographic characteristics and health-related variables to analyze disparities in attributes among older Korean adults based on their social network type. The demographic and sociological characteristics included age, sex (male=0, female=1), educational level (illiterate=1, elementary school=2, middle school=3, and high school or higher=4), residential area (large city=1, small and medium-sized city=2, and rural area=3), and economic level (national basic livelihood security or medical benefit recipient=1, if not=0). We also measured subjective age, which refers to one’s self-perceived status as an older adult or otherwise.20,21) We determined the subjective age by subtracting perceived age from actual age22) based on responses to the question, “At what age do you consider someone an older adult?” A negative subjective age indicated self-perceived older status, whereas a positive value indicated the opposite.

Health status evaluations included functional independence, cognitive health, and depression indicators. First, the independence level of instrumental activities of daily living (IADL) was used to gauge daily life difficulties through 10 questions: grooming, housework, meal preparation, laundry, taking medicine on time, financial management, short-distance outings, making decisions on payment and changes, making phone calls, and using transportation. Questions #1–7 were rated on a three-point scale, whereas questions #8–10 were rated on a four-point scale. The coding was as follows: Questions #1–7 (complete self-reliance=3, partial help=2, and complete help=1) and Questions #8–10 (complete self-reliance=4, some help=3, considerable help=2, and complete help=1). We used the scores for 10 IADL items in the analysis, with a higher score indicating greater functional independence. The Korean version of the Mini-Mental State Examination for Dementia Screening (MMSE-DS) was used to evaluate cognitive function. The MMSE-DS includes 19 questions and a 0–30-point scale, wherein a lower score indicates a lower cognitive function. The 15-item short form of the Geriatric Depression Scale (SGDS) measured depression over the past week using “yes” or “no” responses. The existing score scale was reverse-coded for this analysis, in which the higher the SGDS score, the lower the depression level.

Instruments for Validation of the Data Analyses

For the latent class analysis, we used Mplus version 8.0 (https://www.statmodel.com/) to categorize older adults into groups based on their social relationships involving family, friends, and social activities. Latent class analysis employs the maximum likelihood estimation method for continuous variables while simultaneously assessing an individual’s likelihood of belonging to a group and the overall group model.23,24) This approach classifies groups similarly to conventional cluster analysis.25) This employs an objective approach to determine the optimal number of groups using a model-based stochastic analysis method.26)

To determine the optimal number of groups, goodness-of-fit indices were used, including the Akaike information criterion (AIC),27) Bayesian information criterion (BIC),28) and sample-size adjusted Bayesian information criterion (SABIC).29) Lower values indicated a better-fit model. Additionally, we assessed the classification quality of the latent groups using the entropy index,30) which ranged in values from 0 to 1. An entropy index closer to 1 indicates a more accurate profile, and values of ≥0.8 are considered favorable.31,32) We used the Lo–Mendell–Rubin likelihood ratio test (LMR-LRT) to assess statistical significance by comparing the k and k-1 latent profile group models.33)

We analyzed the digital literacy differences within detailed social network types using IBM SPSS Statistics for Windows, version 26.0 (IBM Corp., Armonk, NY, USA). The frequency and descriptive statistics provided an overview of the respondents’ characteristics. Chi-square and Welch’s F analyses were performed to confirm each group’s characteristics in the case of no equality of variances based on Levene’s test, followed by the Games–Howell post-hoc test to determine the significance of differences between groups. Statistical significance was set at p<0.05. We further analyzed whether the differences in digital literacy varied according to social network type by stratifying confounding factors such as education level, sex, and age.

Ethical Considerations

Participants in the National Survey on Older Koreans provided written informed consent before enrolling in the survey. This national survey was approved by the Institutional Review Board of the Korea Institute for Health and Social Affairs (IRB No. 2020-36). This study obtained a review exemption from the Institutional Review Board of Yonsei University Mirae Campus (No. 1041849-202210-SB-180-01).

RESULTS

The participants’ demographic and sociological characteristics are summarized in Table 1.

Table 1.

Demographic and sociological characteristics (n=9,377)

Social Network Typologies of Older Adults: Latent Class Analysis

Determination of latent classes of social network

We performed latent class analysis using the presence of a spouse, child encounters (at least once a week), child contact (at least once a week), friend encounters (at least once a week), club participation, and engagement in social activities as dependent variables to confirm the social network types of the older adults. The analysis compared the information criterion indices and likelihood ratio verification statistics as the number of latent classes increased from two to four (Table 2). Information reference values, including AIC, BIC, and SABIC, decreased as the fit of the four-class model improved. The statistical significance of the LMR-LRT signified that the current number of k-groups was more suitable than that of k-1 groups.34) In this estimation, all four classes were statistically significant (p=0.001). The second and third classes exhibited excellent entropy values of ≥0.8. The number of latent classes for older adults’ social networks was determined based on statistical information and a comprehensive consideration of class interpretability and case counts. Consequently, three latent classes were determined to be optimal.

Table 2.

A comparison of the fit statistics of different latent class models

Social network types in older adults

We classified the latent classes based on the response patterns to each question within seven categories (spouse, child, child, friend, friend, club, and social activity). The latent class demonstrating substantially high encounter and contact frequency with children was labeled the “child-centered type.” Similarly, the class with elevated encounter and contact frequencies with friends and children was designated the “child-friend type.” Finally, the latent class showing a high frequency of encounters and contact with friends was termed the “friend-centered type.” Fig. 1 shows the conditional response probabilities for each latent class using average values. The specific values of the response probabilities for each class are presented in Supplemental Table S1. Figs. 2–4 show the latent classes.

Fig. 1.

Profile of latent classes. Conditional response probabilities for each latent class are shown using the averages.

Fig. 2.

Characteristics of the child-centered type.

Fig. 3.

Characteristics of the child-friend type.

Fig. 4.

Characteristics of the friend-centered type.

Characteristics of the social network types

We analyzed the differences in demographics, health status, and digital literacy factors among the social network types of older adults (Table 3).

Table 3.

Characteristics of social network types

First, our examination of demographic factors, including age, sex, educational level, residential area, economic status, and degree of older adult perception, according to the social network type, revealed significant differences in all items. Regarding age, the friend-centered and child-friend types showed the highest and lowest average ages, respectively. The proportions of women were higher than those of men for all types.

Regarding education level, the child-centered type exhibited the highest proportion of high school graduates or those with higher education, whereas the child-friend and friend-centered types showed high ratios of elementary school graduates. Regarding residential areas, the child-centered and child-friend types had the highest percentages of individuals living in large cities and the lowest in towns. However, the friend-centered type was characterized by a low percentage of residents in large cities and a high percentage in rural areas. Regarding economic conditions, the rate of non-recipients of government support was high for all three types. The friend-centered type comprised the highest proportion of individuals who perceived themselves as older adults.

We analyzed functional independence, cognitive health, and depression as health status factors with significant differences in all items. Regarding functional independence, the child-centered type experienced the most difficulties, whereas the child-friend type showed the fewest challenges. Regarding cognitive health, the child-friend and friend-centered groups scored the highest and lowest, respectively. Regarding depression, older adults were less depressed in the order of child-friend, child-centered, and friend-centered relationships.

Finally, the digital literacy factors differed significantly according to the social network type. The friend-centered type was the most vulnerable to digital literacy, whereas the child-centered type was the most competent. We performed additional analysis to determine whether there were differences in digital literacy according to social network type when stratifying potential confounding factors such as education level, sex, and age (Table 4). When stratified by sex, the friend-centered and child-centered types were the most vulnerable in both men and women. This finding was similar to the results obtained without stratification. In terms of education level, the difference in digital literacy according to social network type was significant at all levels except for illiteracy. At the other levels, the results were comparable to the findings without stratification. Moreover, the differences in digital literacy were significant after classifying the participants into three age groups. The youngest old and middle old groups demonstrated results similar to those without stratification. However, for the oldest old, friend-centered behavior remained the most vulnerable, whereas the child-friend type was the most competent.

Table 4.

Digital literacy difference of social network types

DISCUSSION

This study aimed to understand social network types based on the frequency and size of social networks among older Korean adults, analyze each type’s characteristics, and examine digital literacy disparities among these types. The summary and implications of the results are as follows.

First, the social network types were classified according to the frequency and size of the social network of older Korean adults into three latent classes: “child-centered type,” “child-friend type,” and “friend-centered type.” The child-friend type accounted for the most participants, followed by the child-centered and friend-centered types. All three types exhibited lower engagement in club participation and social activities among the seven areas used to determine the social network types. This deviated from the patterns reported in other countries, where more diverse networks involve connections among relatives, friends, and communities35) and reflects the family-centered culture in Korea. These findings are partially consistent with those of Im et al.,36) who highlighted the prominence of social networks centered on cohabiting partners and non-cohabiting children among the oldest Korean adults. In contrast, in their 2014 national survey data analysis, Park et al.37) reported higher club or social group participation among older Koreans. This result reflects the unique situation posed by the COVID-19 pandemic, which reduced social networks and activities during the survey period.38)

Second, the results of this study revealed significant differences in the descriptive characteristics of each social network type. A comparative analysis revealed a relatively high average age for the friend-focused type among social network types. In contrast, the child-focused group was younger. While the terminology may differ, these findings align with those of Park et al.,39) demonstrating higher average age among restricted and friend-focused types. Kang et al.19) added that with aging, the proportions of the friend-focused types tend to shrink due to factors such as death or health deterioration, in contrast to the expansion of child-centered types due to family care. Moreover, institutional support is required to form and maintain diverse social networks. Sex differences were also apparent in the present study, with the child-friend type showing a significantly higher proportion of women. This is consistent with earlier research demonstrating women’s broader and more diverse networks and heightened social support exchanges.40) In addition, distinctions in residential areas according to social network type revealed higher proportions of older adults of child-centered and child-friend types living in large cities, whereas the friend-centered type was marked by higher rural residence. This might stem from younger generations migrating to urban areas, leading to depopulation and aging in the rural areas of Korea.41) Economic status varied across social networks in the present study, with the child-friend type showing fewer national basic livelihood security support beneficiaries than the friend-centered and child-centered types. This implies that older adults with better economic standing are more likely to engage in diverse social networks, consistent with the findings of previous studies.37,39) These results suggest the need to reinforce various social networks, particularly for older adults in rural areas and beneficiaries of the national basic livelihood security support. Expanding social networks through public support or developing visiting programs to facilitate interaction with local neighbors is crucial.

Third, the social network type was associated with health outcomes. Regarding the health sub-areas, the child-friend type was healthier in terms of functional independence, cognition, and depression. This finding was consistent with those previous studies showing that broader social networks are correlated with more positive health outcomes.42) Differentiating between child-centered and friend-centered health statuses proved complex due to inconsistencies across health sub-areas. This mirrors earlier studies that reported varying findings on this topic. Regarding functional independence, previous studies suggested that the friend-centered type is healthier than the child-centered type. However, the disparities in depression between friend-centered and child-centered types in the present study were not statistically significant. Research on older adults from Western cultures has revealed a better depressive status for the friend-centered type. In contrast, studies from Eastern cultures such as Hong Kong43) and Japan44) have reported no significant differences in depression. This distinction might stem from the cultural emphasis on family bonds in Eastern societies, compared to the value placed on independence in Western cultures.45) Based on the finding that diverse social network types affect health status, fostering various networks is vital. This could involve creating opportunities for older adults to participate in community gatherings and social activities rather than relying solely on traditional family support.

Fourth, the results of the present study demonstrated that the child-centered type was the most competent, whereas the friend-centered type was the most vulnerable in digital literacy. This indicates that children adopt digital devices, and family composition is a learning environment for digital device usage.46,47) Since older adults in Korea are an information-vulnerable group and the digital divide rate is 68.6%, digital literacy among older Korean adults is generally low. Consequently, interventions and institutional alternatives should be developed to enable older adults to use digital devices without relying on their families. The results of this study demonstrated lower digital literacy among illiterate individuals, regardless of the social network type. This may occur due to the combined difficulty of learning new digital skills and understanding or using text. Therefore, strategies for digital device utilization among seniors must be tailored to the diverse situations of older adults.

During the pandemic, the importance of social isolation and digital literacy increased. The results of the present study are valuable for identifying which types of older adults are most at risk during this unprecedented time by collecting data during a period when older adults were experiencing both social and digital isolation. Particularly, the friend-centered type appeared to be the most vulnerable in terms of health and digital literacy. During the pandemic, when social distancing was required, friend-centered individuals with less interaction with their families may have been at a greater risk of isolation.

This study has several limitations. First, due to the use of secondary data, the variables were limited in selection and validation when identifying social network types and digital literacy in older adults. The study’s understanding of digital literacy levels in older adults was also hindered by the limited scope of items addressing digital use and the difficulties faced by this population. Therefore, follow-up research requires analysis and verification, including more comprehensive and diverse digital items. Second, analyzing social network types in older adults relied on cross-sectional data from 2020 onwards. Therefore, it is difficult to rule out the effects of COVID-19 on older adults’ social activities and networks. Furthermore, owing to aging, the distinct social network patterns among the youngest, middle-aged, and oldest participants necessitate further investigation through longitudinal data analysis. Finally, measuring social network types in older adults was primarily quantitative, focusing on network presence, frequency, and participation. Future studies should consider incorporating qualitative aspects of social networks, such as relationship satisfaction, for a more comprehensive understanding of social network types. Given the current analysis of the distinct characteristics of each network type, follow-up studies applying regression models to investigate the relationships between social network types and digital literacy are needed.

In conclusion, this study categorized the social networks of older adults in Korea into three distinct types and explored their general characteristics and differences in digital literacy. Among these, the child-friend type emerged as the most healthy, underscoring the significance of a wider social network in health preservation. Conversely, the friend-centered type was the most vulnerable to digital literacy. These findings suggest the need to prioritize digital literacy intervention programs and services for friend-centered older adults in South Korea.

Notes

CONFLICT OF INTEREST

The researchers claim no conflicts of interest.

AUTHOR CONTRIBUTION

Conceptualization, HL, MKK; Data curation, HL, MKK, KHP; Investigation, HL; Methodology, MKK; Project administration, HL, HYP; Supervision, HYP, KHP; Writing–original draft, HL, MKK; Writing–review & editing, HL.

FUNDING

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (NRF-2020R1C1C1011374).

SUPPLEMENTARY MATERIALS

Supplementary materials can be found via https://doi.org/10.4235/agmr.23.0174.

Table S1.

Class membership and item response probabilities of three latent classes

agmr-23-0174-Supplemental-Table-S1.pdf

References

1. Oh JH. Categorizing the level of digital literacy and the effect of social relationships: toward basic competence in use of digital media and services. J Cybercommun Acad Soc 2017;34:39–81.

2. Pool CR. A new digital literacy a conversation with Paul Gilster. Educ Leadersh 1997;55:6–11.

3. Hasan H, Linger H. Enhancing the wellbeing of the elderly: social use of digital technologies in aged care. Educ Gerontol 2016;42:749–57.

4. Kim HS, Shim JS. Digital literacy of elderly people and social activities. Korean J Policy Anal Eval 2020;30:153–80.

5. National Information Society Agency. The report on the digital divide, 2022 [Internet]. Sejong, Korea: Ministry of Science and ICT, National Information Society Agency; 2022. [cited 2024 Mar 1]. Available from: https://www.nia.or.kr/site/nia_kor/ex/bbs/View.do?cbIdx=81623&bcIdx=25353&parentSeq=25353.

6. Lee S. The effect of digital literacy in the elderly on quality of life and mediating effect of social network. J Humanit Soc Sci 2020;12:307–22.

7. Kim KW, Kim OS. Super aging in South Korea unstoppable but mitigatable: a sub-national scale population projection for best policy planning. Spat Demogr 2020;8:155–73.

8. Park H, Hong JH, Choi M, Kwon YD, Kim J, Noh JW. Social network changes of pre-and post-retirement. J Korea Contents Assoc 2014;14:753–63.

9. Ell K. Social networks, social support, and health status: a review. Soc Serv Rev 1984;58:133–49.

10. Krause N. Assessing change in social support during late life. Res Aging 1999;21:539–69.

11. Lubben JE. Assessing social networks among elderly populations. Fam Community Health 1988;11:42–52.

12. Kelly ME, Duff H, Kelly S, McHugh Power JE, Brennan S, Lawlor BA, et al. The impact of social activities, social networks, social support and social relationships on the cognitive functioning of healthy older adults: a systematic review. Syst Rev 2017;6:259.

13. Domenech-Abella J, Lara E, Rubio-Valera M, Olaya B, Moneta MV, Rico-Uribe LA, et al. Loneliness and depression in the elderly: the role of social network. Soc Psychiatry Psychiatr Epidemiol 2017;52:381–90.

14. Bradvik L. Suicide risk and mental disorders. Int J Environ Res Public Health 2018;15:2028.

15. Organisation for Economic Co-operation and Development. Health at a glance 2021: OECD indicators Paris, France: OECD Publishing; 2021.

16. National Statistical Office. Elderly statistics, 2014 Daejeon, Korea: Korea National Statistical Office; 2014.

17. Hong M, De Gagne JC, Shin H. Social networks, health promoting-behavior, and health-related quality of life in older Korean adults. Nurs Health Sci 2018;20:79–88.

18. Yim E, Koo B, Lim S. A study on the effect of social networking on life satisfaction on the rural elderly people: focusing on mediator effect of self-esteem. J Soc Sci 2014;25:567–90.

19. Kang EN, Kim HJ, Chung BO. The social network types and depression among old-old elderly. Korean J Soc Welf Stud 2015;46:229–55.

20. Choi EY, Cho SE, Oh YS, Chang HS, Kim YS. The association between subjective age and healthy aging in later life: an age-dependent analysis. Health Soc Welf Rev 2017;37:181–215.

21. Wurm S, Westerhof GJ. Longitudinal research on subjective aging, health, and longevity: current evidence and new directions for research. Annu Rev Gerontol Geriatr 2015;35:145–65.

22. Kaufman G, Elder GH Jr. Revisiting age identity: a research note. J Aging Stud 2002;16:169–76.

23. Hill AL, Degnan KA, Calkins SD, Keane SP. Profiles of externalizing behavior problems for boys and girls across preschool: the roles of emotion regulation and inattention. Dev Psychol 2006;42:913–28.

24. Little RJ, Rubin DB. Statistical analysis with missing data 3rd edth ed. Hoboken, NJ: John Wiley & Sons Inc.; 2020.

25. Morin AJ, Madore I, Morizot J, Boudrias JS, Tremblay M. Multiple targets of workplace affective commitment: factor structure and measurement invariance of the Workplace Affective Commitment Multidimensional Questionnaire. In : Colombus AM, ed. Advances in psychology research Hauppauge, NY: Nova Science Publishers; 2009. p. 45–75.

26. Shin H, Yun S. Middle school students’ academic and psychological adjustment among achievement goal orientation profiles. Korean J Sch Psychol 2015;12:271–93.

27. Akaike H. A new look at the statistical model identification. IEEE Trans Automat Control 1974;19:716–23.

28. Schwarz G. Estimating the dimension of a model. Ann Stat 1978;6:461–4.

29. Sclove SL. Application of model-selection criteria to some problems in multivariate analysis. Psychometrika 1987;52:333–43.

30. Morin AJ, Wang JK. A gentle introduction to mixture modelling using physical fitness performance data. In : Ntoumanis N, Myers N, eds. An introduction to intermediate and advanced statistical analyses for sport and exercise scientists London, UK: John Wiley & Sons; 2016. p. 183–209.

31. Hong W. A latent profile analysis of adolescents’ emotional/behavioral problems: the relationship between parenting attitudes and life satisfaction. Stud Korean Youth 2020;31:291–320.

32. Ko DH, Jung HS. A latent profile analysis of perceptions about mathematics teachers in school lessons. Math Educ 2018;57:75–92.

33. Kang SA, Park CS, Chang JY. An exploratory study on profiles of recovery experiences among Korean employees: application of latent profile analysis. Korean J Ind Organ Psychol 2018;31:643–67.

34. Pastor DA, Barron KE, Miller BJ, Davis SL. A latent profile analysis of college students’ achievement goal orientation. Contemp Educ Psychol 2007;32:8–47.

35. Litwin H, Shiovitz-Ezra S. Social network type and subjective well-being in a national sample of older Americans. Gerontologist 2011;51:379–88.

36. Im S, Kang M, Cho S. Social network and health status among Korean elders. J Welfare Aged 2013;59:281–308.

37. Park NS, Jang Y, Lee BS, Chiriboga DA, Chang S, Kim SY. Associations of a social network typology with physical and mental health risks among older adults in South Korea. Aging Ment Health 2018;22:631–8.

38. Sepulveda-Loyola W, Rodriguez-Sanchez I, Perez-Rodriguez P, Ganz F, Torralba R, Oliveira DV, et al. Impact of social isolation due to COVID-19 on health in older people: mental and physical effects and recommendations. J Nutr Health Aging 2020;24:938–47.

39. Park S, Smith J, Dunkle RE. Social network types and well-being among South Korean older adults. Aging Ment Health 2014;18:72–80.

40. Choi JM, Hwang SY, Kim SE. Social network and comprehensive aging problem: interaction effect analysis between gender and social network for political application. Korean J Policy Anal Eval 2015;25:83–114.

41. Garcilazo E, Springare LS, Sasaki M, Oropeza HD. OECD Regional Development Working Papers: Rural study of Korea Paris, France: Organisation for Economic Co-operation and Development; 2019.

42. Ali T, Nilsson CJ, Weuve J, Rajan KB, Mendes de Leon CF. Effects of social network diversity on mortality, cognition and physical function in the elderly: a longitudinal analysis of the Chicago Health and Aging Project (CHAP). J Epidemiol Community Health 2018;72:990–6.

43. Cheng ST, Lee CK, Chan AC, Leung EM, Lee JJ. Social network types and subjective well-being in Chinese older adults. J Gerontol B Psychol Sci Soc Sci 2009;64:713–22.

44. Fiori KL, Antonucci TC, Akiyama H. Profiles of social relations among older adults: a cross-cultural approach. Ageing Soc 2008;28:203–31.

45. Li T, Zhang Y. Social network types and the health of older adults: exploring reciprocal associations. Soc Sci Med 2015;130:59–68.

46. Hwang HJ, Hwang YS. Gaps and reasons of digital divide within the elderly in Korea: focusing on household composition. Res Inst Soc Sci 2017;24:359–86.

47. Wu YH, Damnee S, Kerherve H, Ware C, Rigaud AS. Bridging the digital divide in older adults: a study from an initiative to inform older adults about new technologies. Clin Interv Aging 2015;10:193–200.

Article information Continued

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (https://creativecommons.org/licenses/by-nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.

Variable	Value
Age (y)	73.51±6.52
Sex
Male	3,817 (40.1)
Female	5,621 (59.9)
Educational attainment
Illiteracy	1,072 (11.4)
Elementary school	3,126 (33.3)
Middle school	2,192 (23.4)
High school and above	2,987 (31.9)
Region
Large city	3,841 (41.0)
Small and medium-sized city	2,821 (30.1)
Rural area	2,715 (29.0)
Economic level	291 (51.14)
National basic livelihood security or medical benefit recipients	617 (6.6)
Not applicable	8,760 (93.4)
Elderly perception degree	3.27±7.93

Class	AIC	BIC	SABIC	Entropy	LMR-LRT	Proportions for the class (%)
Class	AIC	BIC	SABIC	Entropy	LMR-LRT	1	2	3	4
2	68628.216	68735.649	68687.981	1.000	4266.448^***	72.68	27.32	-	-
3	67150.092	67314.822	67241.732	0.849	1474.014^***	23.14	62.70	14.16	-
4	66589.534	66811.563	66713.049	0.717	568.797^***	21.82	21.82	17.38	38.98

	Child-centered type (a)	Child-friend type (b)	Friend-centered type (c)	F (Games-Howell)
Age (y)	73.53±6.91	73.38±6.43	74.24±6.25	8.659^*** (c>a, b)
Sex				67.586^***
Male	1,009 (47.3)	2,319 (37.4)	428 (40.3)
Female	1,111 (52.7)	3,872 (62.6)	638 (59.7)
Educational attainment				30.935^***
Illiteracy	272 (13.3)	655 (10.7)	145 (13.7)
Elementary school	664 (31.7)	2,090 (33.7)	372 (35.1)
Middle school	476 (22.3)	1,446 (23.4)	270 (25.2)
Over high school	708 (32.7)	2,000 (32.1)	279 (26.0)
Region				167.864^***
Large city	965 (45.4)	2,583 (41.6)	293 (27.9)
Small and medium-sized city	713 (33.7)	1,756 (28.5)	352 (32.8)
Rural area	442 (21.0)	1,852 (29.8)	421 (39.3)
Economic level				51.969^***
National basic livelihood security or medical benefit recipients	173 (8.1)	331 (5.3)	113 (10.5)
Not applicable	1,947 (91.9)	5,860 (94.7)	953 (89.4)
Subjective age	3.00±8.66	3.22±7.71	4.02±7.66	6.251^** (c>a, b)
Functional independence	3.19±0.36	3.26±0.20	3.24±0.27	61.148^*** (b, c>a)
Cognition	24.26±7.88	24.86±6.56	23.41±6.12	21.082^*** (b>a>c)
Depression	11.16±3.61	11.92±3.15	11.09±4.00	50.604^*** (b>a, c)
Digital literacy	1.94±1.66	1.79±1.62	1.49±1.47	30.363^*** (a>b>c)

	Digital literacy
	Child-centered type (a)	Child-friend type (b)	Friend-centered type (c)	F (Games-Howell)
Education
Illiteracy	0.68±1.02	0.74±1.13	0.74±1.08	0.319
Elementary school	1.52±1.50	1.24±1.37	1.23±1.39	9.290^*** (a>b, c)
Middle school	1.92±1.63	1.81±1.51	1.49±1.50	7.160^*** (a, b>c)
Over high school	2.83±1.55	2.70±1.60	2.23±1.44	17.023^*** (a, b>c)
Sex
Male	2.19±1.65	2.14±1.67	1.71±1.48	17.082^*** (a, b> c)
Female	1.71±1.64	1.58±1.56	1.35±1.44	11.801^*** (a, b> c)
Age group
Youngest old (65–74 y)	2.42±1.62	2.26±1.62	1.94±1.49	19.342^*** (a>b>c)
Middle old (75–84 y)	1.33±1.48	1.12±1.34	1.05±1.28	6.917^* (a>b, c)
Oldest old (≥85 y)	0.58±1.05	0.84±1.27	0.53±0.94	4.029^* (b>a, c)