A Typology of Aging Internet Users: Exploring Digital Gradations in Internet Skills and Uses

Abstract

Research suggests that older internet users are not a homogeneous group of users, as their level of digital inclusion varies widely, depending not only on their age, but also on their socio-demographic background, internet access characteristics, and the availability of social resources associated with “age-graded” life events (e.g., retirement and widowhood). This study presents a typology of internet users in late middle and late adulthood according to their levels of skills and uses and examines how socio-demographic characteristics, social networks and resources, and internet access shape differences between groups. Based on representative survey data, cluster analysis was used to identify four groups of users aged 50+: Apprehensive, Level-headed, Savvy, and Reluctant users. Three main conclusions emerge from their comparison. First, both internet skills and uses need to be considered, as more skills do not always lead to more varied use and the relationship is affected by access characteristics and proxy internet use. Second, socio-demographics remain critical in explaining gradations in digital inclusion, but their effects must be contextualized. Not all younger older adults were highly digitally engaged, so their broader life contexts need to be considered. Third, social networks and resources had little impact on aging internet users’ digital engagement.

Keywords

cluster analysis digital divide internet use older adults social support typology

Introduction

The gap in internet access between older adults and other age groups has been slowly closing over the last decade (Blank et al., 2020; Ofcom, 2021). While almost all recent growth in internet access and use in the general population has been associated with growth in users in older age groups (Blank et al., 2020), the narrowing of the digital divide is also being driven by the transition of long-time internet users into late middle and late adulthood (Matthews et al., 2019; Wang et al., 2018).

However, there remain significant differences in internet skills and use between younger and older internet users (Blank et al., 2020). This is also true for the aging population—younger older adults have been shown to be systematically more digitally engaged than older adults (Hargittai & Dobransky, 2017). In fact, in their study of internet users aged 50+, Matthews et al. (2019, p. 1) found that “rates of internet use increase faster among younger cohorts yet, despite initially increasing, begin to decline among older cohorts.” Likewise, Friemel (2016) argues that the cohort effect might apply to younger seniors, but not to older ones, for whom health problems and physical limitations seem to gain importance. Considering not only the different experiences older adults may have with internet technology, but also the fact that aging is characterized by normative and non-normative “age-graded” life events (e.g., retirement and family life cycle) and associated changes in the availability of various social resources (Baltes et al., 1980), the importance of understanding the nuances of digital inclusion among older adults becomes clear. In addition, as more and more everyday transactions––in the market, with the state and within social networks––are predicated upon internet use, research on age-related digital divides is crucial for forming and implementing adequate social policy measures to support the social inclusion of older adults with different needs and backgrounds as well as online capacities.

Older adults are indeed very heterogeneous in terms of their internet skills and use levels when comparing the degree of their digital inclusion with other age groups (Friemel, 2016; Hargittai & Dobransky, 2017). Differences in their levels of internet skills and use are related not only to their age but also to other socio-demographic characteristics (e.g., gender, education, and income), the availability of social resources and support (e.g., having children, association memberships, diversity of personal networks, opportunity for access to and mobilization of social support), characteristics of their internet access (Hargittai et al., 2019; Hargittai & Dobransky, 2017; Olsson et al., 2019), and previous internet experience, such as years online and occupational experience working with computers (Friemel, 2016; Hänninen et al., 2021; Hargittai & Dobransky, 2017; König et al., 2018; van Deursen et al., 2014; van Deursen & Helsper, 2015). Van Dijk’s resources and appropriation theory (van Dijk, 2005) posits that the personal (e.g., age and gender) and positional inequalities (e.g., education) described above determine the quantity and quality of resources available to an individual, which are the prerequisite of technology use. In fact, a recent qualitative study showed that these factors can help us better understand the social and personal contexts of different user groups among older adults, characterized by specific patterns of internet skills and uses (Quan-Haase et al., 2018).

To extend our understanding of such nuances, we draw on the second-level digital divide framework (Hargittai & Hsieh, 2012) to develop a typology of aging internet users based on differences in the level of their internet skills and uses. We study the aging population of internet users, defined as users aged 50+, for two reasons: first, to take into account the potential age-graded influences in the transition from late middle to late adulthood that have been suggested to affect levels of digital inclusion (Choi et al., 2020), and second, because a lower age threshold “might better capture the differentiation in findings across older adults, and better pinpoint disparity” (Hunsaker & Hargittai, 2018, p. 3948). Aside from identifying four distinct user groups, the original value of this study, based on multivariate analyses of survey data representative of the Slovenian population, is a comprehensive examination of the background characteristics of the four user groups using a range of indicators of internet access, demographics and social networks and resources, which have been indicated to shape gradations of digital inclusion in middle and late adulthood (Choi et al., 2020).

Literature Review

Typologies of Aging Internet Users

While early segmentations of internet user studies often excluded older adults or grouped them into one segment treating them as one homogenous group (e.g., Brandtzaeg et al., 2010), later research showed that there are differences among aging internet users that go beyond the distinction between internet use and non-use. For example, Vuori and Holmlund-Rytkönen (2005) identified two groups of Finnish older adults with specific internet-related needs. The group of healthier older adults, who were more likely to go online, were referred to as Healthy, indulging 55+ internet users (more educated users with higher incomes, still working), in contrast to Ailing outgoers, who used the internet but were not regular users. Health status was also found to be a relevant factor in influencing heterogeneity in internet use in the cluster analysis of van Boekel et al. (2017). Their results indicated four groups of older internet users. Minimizers were the oldest group with the lowest levels of psychological well-being and frequency of internet use. Conversely, Maximizers were the youngest and most digitally engaged cluster and, in turn, spent the most time on the internet. Their psychological well-being and overall health were better than those of the Minimizers and Social users, but similar to Practical users (who spent less time on the internet, which they used for functional and financially related activities).

Further, Nimrod (2013) derived a cross-cultural typology of internet users in later life based on context-specific internet activities related to participation in online communities. Three clusters were identified, with Aging-oriented standing out as the cluster with the lowest number of posts, whereas Information swappers had the strongest motivation to communicate and exchange information online. For the Socializers, online communities offered more than just practical information and social support, and they therefore reported the highest level of gratification. However, in a later study, Nimrod (2017) showed that a large majority of older adults still relied on traditional mass media rather than internet-based media when seeking information. Among the four identified clusters, digitally enhanced practices were typical only of Heavy asynchronous media users (17.7%) whose openness to innovation was associated with a wide range of internet uses and pronounced preference for asynchronous online communication. The largest group in the sample (43.9%) were Lighter users who predominantly used traditional over new media, while the two remaining groups, Heavy synchronous one-to-many users and Heavy one-to-many users, were both characterized by high overall media use, but not in terms of digitally enhanced practices (Nimrod, 2017). More recently, Vulpe and Crăciun (2020) identified three distinct clusters, further suggesting that different groups of older adults show diverse use of internet-enabled devices, frequencies of use, and communication preferences. The most diverse practices were identified among Digitally immersed communicators, who used the internet most frequently from various locations, in contrast to Asynchronous communicators, who were more likely to use a “traditional” means of communication, and Phone enjoyers, who accepted the internet and devices related to it, but were selective in their communication and internet activities.

Differences in the levels of digital skills and number of activities undertaken online were used to qualitatively identify five clusters of older internet users aged 65+ in Canada (Quan-Haase et al., 2018). Although Savvy users showed the highest levels of digital skills and uses, in-depth interviews revealed that older internet users with higher levels of online engagement do not always have pronounced digital skills. In particular, Go-getters and Apprehensive users engaged in multiple online activities despite having low- to mid-level digital skills. This was partly explained by their engagement in indirect internet access (e.g., asking other users to go online on their behalf) and by the fact that users’ motivation to improve internet skills in later life varied. In fact, participants often experienced internet anxiety and felt insecure, which were the main reasons for their lack of motivation to expand their digital skills (Quan-Haase et al., 2018). This applies to the largest group of Basic users as well as to Reluctant users, who also received limited social support. Members of other clusters showed interest in engaging in new online activities by improving their digital skills with the help of others.

Determinants of the Second-Level Digital Divide

The reviewed studies converge on the conclusion that diversity in internet use and/or skill levels between groups of older internet users can be explained by their different personal characteristics and by the social and technological context in which they live. This is in line with van Dijk’s (2005) resources and appropriation theory, which suggests that personal (e.g., gender and age) and positional inequalities (e.g., education and labor) are related to unequal distribution of resources (social and material, including internet access) which leads to differences in skills and uses. For example, user typology studies (Quan-Haase et al., 2018; van Boekel et al., 2017) confirm the findings of previous research (e.g., Friemel, 2016; Hargittai et al., 2019) that younger and more educated older adults are more likely to be savvy and more engaged internet users. Moreover, existing typologies and the related research suggest that socio-economic status (Cotten et al., 2016; Francis et al., 2019; Friemel, 2016; Nimrod, 2013, 2017; Olsson & Viscovi, 2020; Vulpe & Crăciun, 2020; Vuori & Holmlund-Rytkönen, 2005) and social capital (van Boekel et al., 2017) have both been associated with patterns of internet uses and skills in later life. However, online activities may not only enhance older adults’ social capital by expanding social contacts and access to information sources (e.g., Quan-Haase et al., 2018) but online activities may also be shaped by the social capital of older adults (Choi et al., 2020; Correa et al., 2017; Friemel, 2016). While van Boekel et al. (2017) did not find significant differences among the three clusters in terms of social and emotional loneliness, Vuori and Holmlund-Rytkönen (2005) noted that Ailing outgoers were more likely to spend their free time with family and friends. Relatedly, Vulpe and Crăciun (2020) found Digitally immersed communicators who used various devices and platforms for online communication to be more involved in political matters and to develop cultural beliefs influenced by social media.

Studies on user typologies have also reported different forms of internet access and types of devices as important causes of differential online engagement among older adults (Quan-Haase et al., 2018; Vulpe & Crăciun, 2020). Quan-Haase et al. (2018) argued that direct forms of internet access (e.g., types and number of devices, access location) need to be considered along with older adults’ engagement in indirect forms of access, such as proxy internet use (Dolničar et al., 2018; Grošelj et al., 2022). A number of older internet users mitigate their limited internet access or low internet skills by relying on the help of younger users in their social networks (Grošelj et al., 2019; Quan-Haase et al., 2018). However, Hunsaker et al. (2019) showed that inter-generational ties are not the only informal source of proxy internet use among older adults. In urgent cases or for minor technical problems, older users prefer peer support (e.g., spouse, friends, and neighbors). These findings suggest that internet users in middle and late adulthood not only receive support from others as users-by-proxy but also can help others in the role of proxy internet users (Hunsaker et al., 2019; Marler & Hargittai, 2022).

Given previous evidence of associations between digital engagement and transitions from late middle to late adulthood that are related to changes in the availability of various social resources, we expect varying levels of internet skills and uses to be associated not only with users’ socio-demographic characteristics and forms of direct and indirect internet access, but also with their social networks and resources. The aim of this study was to identify the clusters of aging adults based on their internet skills and uses, and to describe cluster characteristics based on socio-demographic background, social networks and resources, and (in)direct internet access. Four specific research questions (RQs) were asked:

RQ1: What groups of aging internet users can be identified based on their internet skills and breadth of internet use?

RQ2: What are the distinguishing characteristics of these groups according to their socio-demographic backgrounds?

RQ3: How do these groups differ in terms of their social networks and resources?

RQ4: What differences in internet access and proxy internet use exist among these groups?

Methods

Procedures and Data

Data were collected in the 2018 wave of the Slovenian Public Opinion Survey conducted as part of the International Social Survey Programme (ISSP). Computer-assisted face-to-face interviews were conducted from March to June 2018. Participants aged 18 years or older were selected from the Central Register of Population (CRP) using a two-stage random sampling with stratification by type of settlement and the statistical region of residence. The survey was completed by 1,047 participants, a response rate (RR1) of 57% after excluding non-eligible units (AAPOR, 2016). No post-stratification techniques were applied since the socio-demographic characteristics of the sample were shown to closely mirror the characteristics of the general population when compared with data retrieved from the CRP.

The analyses are based on the subsample of 322 participants aged 50+ who were internet users (hereafter referred to as respondents). They represented 30.8% of all participants in the sample and 39.6% of internet users. The socio-demographic characteristics of the total sample and the subsamples of internet users aged 50+ are presented in Table 1.

Table 1.

Socio-Demographic Characteristics of the Total Sample and the Subsample of Internet Users Aged 50+.

Variable	Category	Total sample^a		Internet users 50+^a
Variable	Category	N	%	N	%
Gender	Male	509	48.6	158	49.1
Gender	Female	538	51.4	164	50.9
Age	29 or less	151	14.4
	30–39	183	17.5
	40–49	165	15.8
	50–59	183	17.5	156	48.4
	60–69	180	17.2	116	36.0
	70 or more	185	17.7	50	15.5
Education	Primary basic education or lower	177	17.0	28	8.8
	Secondary education	546	52.4	174	54.4
	Higher education	318	30.5	118	36.9
Occupation	Second skill level or less	575	62.6	166	53.2
	Third skill level	124	13.5	61	19.6
	Fourth skill level	220	23.9	85	27.2
Employment status	Working	578	55.3	166	51.7
Employment status	Not working	467	44.7	155	48.3
Type of settlement	Urban	214	20.5	83	25.9
	Semi-Urban	267	25.6	74	23.1
	Rural	562	53.9	163	50.9
Marital status	Married, civil partnership	723	69.9	266	83.1
Marital status	Divorced, widowed, single	311	30.1	54	16.9
Individual monthly income^b	≤ 797.2 €	493	47.1	127	39.4
Individual monthly income^b	> 797.2 €	554	52.9	195	60.6
Household size	One person	133	12.7	39	12.1
	Two persons	312	29.8	142	44.1
	Three persons	217	20.7	70	21.7
	Four or more persons	385	36.8	71	22.0
Children in household	Yes	341	32.6	44	13.7
Children in household	No	706	67.4	278	86.3
Internet use	Yes	814	77.9
Internet use	No	231	22.1

Notes. ^aThe sample size might vary due to item nonresponse, which was most often caused by explicit refusal or “don’t know” answers. Response percentages may not add up to 100% due to rounding.

^bThe average individual monthly net income in Slovenia in 2018 was 797.2 € (SORS, 2020).

Measures

Variables used as the basis for clustering (i.e., clustering variables) included measures of internet skills and internet use. Internet skills were assessed with a short form of the Internet Skill Scale (van Deursen et al., 2016). The results of exploratory and confirmatory factor analyses showed high to adequate convergent and divergent validity of the scale (Grošelj et al., 2021). The four scale scores corresponding to operational, information navigation, social, and creative skills were calculated by averaging across corresponding items, with items measuring information navigational skills being reversed before summing so that higher values of all four skills scores indicated a higher level of proficiency. Breadth of internet use was assessed with an adapted 17-item inventory from Blank and Groselj (2014) measuring frequency of internet uses across cultural, personal, economic, and social domains. Four items each assessed cultural, social, and personal uses, while five items measured economic uses. For each of the four types, a composite score of breadth of use was created by summing the number of items that a user performs more often than “Never” and dividing it by the number of items in the corresponding type (see Reisdorf et al., 2021). We obtained four composite scores with higher values indicating a higher percentage of activities in which internet users engaged across the four domains.

In addition to describing the identified groups with their levels of internet skills and breadth of uses, we used three groups of variables as cluster descriptors. First, the standard ISSP socio-demographic module (ISSP Research Group, 2019) was used to assess participants’ gender, age, education level, occupation (based on International Standard Classification of Occupations––ISCO), employment, marital status, individual monthly net income, type of settlement, household size, and presence of children in household. Since 24.1% of respondents did not report their monthly net income, the missing values were replaced using regression imputation (Little & Rubin, 1987). Second, respondents’ social networks and resources were measured with the ISSP core social networks and social resources module (Sapin et al., 2020). The module included three dimensions: (1) a measure of diversity and hierarchy in personal social networks, assessed by a position generator; (2) a measure of social resources and support, assessed by a resource generator that captured personal and organizational forms of practical, informational, and emotional support; and (3) a measure of the structure of opportunity for access and mobilization of social resources based on seven items tapping sociability in kinship and friendship networks (Sapin et al., 2020). While the obtained factor solutions and scores with principal component analysis (PCA) and non-linear PCA¹ closely mirrored the results of Sapin et al. (2020) for the first two dimensions, we had to eliminate a total of three items from the inventory to retain a two-factor solution for the third dimension (see Table A4 in Appendix). Third, respondents’ internet access characteristics and proxy internet use were assessed. Type of internet access (PC-only or PC and other devices) and number of devices were assessed by five items measuring frequency of use of different devices (e.g., PC or laptop, smartphone, tablet). To determine how many different devices respondents used to go online and whether they were PC-only users the ordinal scale was collapsed into a dichotomous variable. If respondents used a selected device more often than “Never,” they were classified as device users (Grošelj & Matjašič, 2019). Among the respondents, 35.4% relied only on one device when accessing the internet and 28.0% used only a PC to go online (see Table 2). As a measure of indirect internet access, we used proxy internet use measures. To identify proxy internet users, respondents were asked if they have done any of the eight activities (e.g., searched for information about products or services, sent or received email, ordered or bought products or services) on the internet for someone else for private purposes in the last 12 months. Respondents who replied “Yes” to at least one of the listed activities were coded as proxy users (see Reisdorf et al., 2021). To identify users-by-proxy an adapted question from Grošelj et al. (2019) was used with Yes/No response options: “In the past 12 months have you turned to someone else to do something for you on the internet?” Respondents answering “Yes” were classified as users-by-proxy. The number of users-by-proxy (45.8%) was higher than the number of proxy internet users (31.4%). In terms of overall engagement in proxy internet use, 30.8% of respondents acted only as users-by-proxy as compared to 16.5% who were only proxy internet users. While 15% of respondents reported being both a user-by-proxy and a proxy internet user, 37.7% were not engaged in any kind of proxy internet use.

Table 2.

Internet Access and Proxy Internet Use among Internet Users Aged 50+.

Variable	Category	N ^a	%
Number of devices	One	114	35.4
Number of devices	Two or more	208	64.6
Type of access	PC-only	90	28.0
Type of access	PC and other devices	232	72.0
User-by-proxy	Yes	147	45.8
User-by-proxy	No	174	54.2
Proxy internet user	Yes	101	31.4
Proxy internet user	No	221	68.6
Engagement in proxy internet use	Only user-by-proxy	99	30.8
	Only proxy internet user	53	16.5
	User-by-proxy and proxy internet user	48	15.0
	Neither user-by-proxy nor proxy internet user	121	37.7

Notes. ^aThe sample size might vary due to item nonresponse, which was most often caused by explicit refusal or “don’t know” answers.

Analysis

The statistical analysis was carried out in two stages. First, a combination of hierarchical and k-means clustering with Euclidian distance as a similarity or closeness measure was used to identify clusters of aging internet users based on internet skills and breadth of use. This approach was selected because clustering makes no assumptions concerning the number of groups or group structure and groups members together based on their natural similarity (Johnson & Wichern, 2007, pp. 671–673). The Ward method and the cluster dendogram (Figure 1) were applied in hierarchical clustering to determine the number of clusters (Antonenko et al., 2012). K-means partitioning was used to determine the cluster membership of each unit. Three methods were used and compared to validate the optimal number of clusters obtained with hierarchical clustering (Figure 2; Rousseeuw, 1987; Tibshirani et al., 2001). Second, once the clusters were identified, the differences between them were explored using Pearson’s chi-square test for the categorical variables, and a one-way analysis of variance (ANOVA) with a post-hoc Tukey HSD test was used for normally distributed continuous variables. Meanwhile, a Kruskal–Wallis non-parametric test followed by a Dunn multiple comparison post-hoc was run in case of violation of normal distributions. All analyses were performed in R (R Core Team, 2019).

Figure 1.

Hierarchical cluster dendrogram.

Figure 2.

Results of the non-hierarchical k-means cluster analysis.

Results

With reference to RQ1, the results of hierarchical (Figure 1) and k-means clustering revealed four distinct groups of internet users, which we labeled as Apprehensive users, Level-headed users, Savvy users, and Reluctant users, based on distributions of internet skills and uses variables.

Table 3 presents the results of the one-way ANOVA and post-hoc tests, which showed that the effect of cluster membership was significant across all eight variables. Apprehensive users represented 19.1% of the sample. In general, they report, on average, a similar level of skills as Reluctant users, but lower levels when compared with Level-headed and Savvy users. However, their limited skills do not impede their engagement in different online activities. For example, they are on average engaged in relatively more internet uses (except for economic uses) than Level-headed users. Level-headers, who are the largest group in the sample (38.5%), resemble Savvy users in reporting the highest levels of operational, information and navigation, and social skills, while they are similar to Apprehensive users in terms of creative skills. Surprisingly, the reported internet skills of Level-headed users are not fully reflected in their online activities, as they show significantly lower engagement in all internet uses compared to Savvy and Apprehensive users––the latter are only significantly less engaged in economic uses. Savvy users are the second largest group in the sample (28.8%). On average, they significantly outperform Apprehensive and Reluctant users across all four internet skills, and report a significantly higher level of creative skills than Level-headers. They also report the highest average levels of engagement among the four groups. In fact, the only non-significant means difference between Savvy users and the other groups was found when comparing them to Apprehensive users in terms of cultural uses. Finally, Reluctant users are the smallest (13.6%) and the least engaged group. Their low levels of activity could be explained by an overall lack of internet skills. However, compared to Apprehensive users, they report significantly lower levels only in operational skills. The characteristics of the four groups according to their internet skills and uses can be found in Figure 3.

Table 3.

Summary of Means (SD) for Internet Skills and Uses Variables Across the Four Clusters.

	Apprehensive users (N = 45)	Level-headed users (N = 91)	Savvy users (N = 68)	Reluctant users (N = 32)	Total	F
Operational skills	3.4 (0.93)	4.4_a (0.66)	4.7_a (0.56)	1.6 (0.83)	3.9 (1.24)	130.27***
Information–navigation skills	3.0_a (0.80)	4.3_b (0.62)	4.3_b (0.70)	3.1_a (1.04)	3.9 (0.95)	49.637***
Social skills	3.7_a (0.88)	4.7_b (0.44)	4.7_b (0.42)	3.9_a (1.21)	4.4 (0.80)	20.155***
Creative skills	1.5_ab (0.57)	1.9_a (1.03)	2.9 (0.99)	1.1_b (0.28)	2.0 (1.60)	68.390***
Cultural uses	0.66_a (0.22)	0.46 (0.14)	0.74_a (0.19)	0.33 (0.21)	0.56 (0.24)	48.174***
Economic uses	0.36 (0.25)	0.56 (0.29)	0.82 (0.22)	0.14 (0.15)	0.54 (0.34)	64.474***
Personal uses	0.60 (0.22)	0.45 (0.23)	0.74 (0.22)	0.25 (0.21)	0.53 (0.27)	41.100***
Social uses	0.68 (0.30)	0.42 (0.25)	0.85 (0.19)	0.11 (0.17)	0.55 (0.34)	139.87***

Notes. N = 236. Group means with different subscripts within rows are not significantly different at the p < .05 based on Tukey HSD post-hoc pairwise comparisons. ***p ≤ .001.

Figure 3.

Graphical illustration of the four clusters according to internet skills and uses.

To answer RQ2, the demographic profiles of clusters were compared (Table 4). The results indicated significant differences by all analyzed variables, except gender, marital status, and children in the household. The age composition of Apprehensive users was very similar to Reluctant users, with more than 20% of respondents aged 70+. One out of five (22.2%) had primary education or lower, almost two out of three (64.4%) had an occupation that required the lowest skill level, and 66.7% were not working. They were also similar to Reluctant users in terms of income as 55.6% of Apprehensive users reported to have individual net monthly income below the nation's per capita average (i.e., 797.2 €). More than half (53.3%) were living in rural areas. Apprehensive users included the largest percentage (15.6%) of respondents living alone. The small number of 70+ users (8.8%) made Level-headed users very similar to Savvy users. An equal share (50%) of Level-headers obtained secondary or higher education. The majority (57.3%) had an occupation that required a third or fourth ISCO skill level, and almost two thirds (63.7%) were working. About half (52.7%) were living in rural areas, and only 6.6% were living alone. Savvy users were the youngest of the four groups, with a very similar educational attainment and workforce status as Level-headers. Yet, they were different from Level-headers in terms of occupation skills, as 45.3% (vs 32.6% of Level-headers) were working in occupations that required the highest skill level. Also, Savvy users differed from Level-headers in that only one in three (32.4%) were living in rural areas, and 20.6% lived in households with four or more persons. Although both of these groups were significantly more well off than Apprehensive and Reluctant users, a larger percentage of Savvy users (76.5%) had a monthly net income above the nation’s per capita average in comparison with Level-headers (65.9%). Reluctant users closely resembled Apprehensive users in age, education, employment status, and living area. Conversely, a significantly larger percentage (81.3%) of Reluctant users had an occupation that required the lowest skill level compared to Apprehensive users. Interestingly, of the four groups, Reluctant users had the largest share (59.4%) of respondents who were living in two-person households and the smallest percentage (6.2%) of respondents in three-person households.

Table 4.

Differences in Demographic Profiles of Clusters.^a

Variable	Category	Apprehensive users (%)	Level-headed users (%)	Savvy users (%)	Reluctant users (%)	Total (%)	χ ² (df) ^b
Gender	Male	40.0	49.5	51.5	53.1	48.7	1.844 (3)
Gender	Female	60.0	50.5	48.5	46.9	51.3	1.844 (3)
Age	50–59	35.6	52.7	57.4	37.5	48.7	13.189 (6)*
	60–69	42.2	38.5	36.8	40.6	39.0
	70+	22.2	8.8	5.9	21.9	12.3
Education	Primary education or lower	22.2	0.0	1.5	15.6	6.8	45.413 (9)**
	Secondary education	57.8	50.0	48.5	68.8	53.6
	Higher education	20.0	50.0	50.0	15.6	39.6
Occupation	Second skill level or lower	64.4	42.7	32.8	81.3	49.6	28.991 (6)**
	Third skill level	20.0	24.7	21.9	3.1	20.0
	Fourth skill level	15.6	32.6	45.3	15.6	30.4
Employment status	Working	33.3	63.7	61.8	37.5	46.2	16.355 (3)**
Employment status	Not working	66.7	36.3	38.2	62.5	53.8	16.355 (3)**
Type of settlement	Urban	28.9	28.6	27.9	31.2	28.8	15.404 (6)*
	Semi-Urban	17.8	18.7	39.7	12.5	23.7
	Rural	53.3	52.7	32.4	56.2	47.5
Marital status	Married, civil partnership	82.2	85.7	83.8	84.4	84.3	0.296 (3)
Marital status	Divorced, widowed, single	17.8	14.3	16.2	15.6	15.7	0.296 (3)
Individual monthly income^c	≤ 797.2 €	55.6	34.1	23.5	56.3	38.1	17.027 (3)**
Individual monthly income^c	> 797.2 €	44.4	65.9	76.5	43.8	61.9	17.027 (3)**
People living in a household	One	15.6	6.6	11.8	6.2	9.7	18.485 (9)*
	Two	42.2	44.0	30.9	59.4	41.9
	Three	22.2	20.9	36.8	6.2	23.7
	Four or more	20.0	28.6	20.6	28.1	24.6
Children in the household	Yes	15.6	12.1	20.6	6.2	14.4	4.279 (3)
Children in the household	No	84.4	87.9	79.4	93.8	85.6	4.279 (3)

Notes. N = 236. *.01 < p ≤ .05, **p ≤ .01.

^aResponse percentages may not add up to 100% due to rounding.

^bIn the case that cells had expected counts of less than 5, the Monte Carlo simulation for computing p-values was run on 10,000 replicates. Response percentages may not add up to 100% due to rounding.

^cThe average individual monthly net income in Slovenia in 2018 was 797.2 € (SORS, 2020).

To address RQ3, non-parametric tests were performed to see if there were differences across the four clusters according to the five investigated dimensions of social networks and resources (see Table 5). Only the position generator significantly distinguished the groups. Savvy users had more social capital than Reluctant users, while Apprehensive and Level-headed users did not differ from any other group. Moreover, while not significant, the observed differences between Savvy and Reluctant users suggested that the latter, on average, relied more on institutionalized sources of support and reported more frequent contact within the family. On average, Savvy users had less access to personal network support resources than Apprehensive users, even though they (like Reluctant users) reported having more frequent contact outside family than Level-headed and Apprehensive users.

Table 5.

Differences in Mean Factor Scores for Social Networks and Resources Variables Across Four Clusters.

Variable	Apprehensive users	Level-headed users	Savvy users	Reluctant users	Total	χ ² (df)
Social capital as measured by position generator^a	−0.01 (1.03)	0.11 (0.95)	0.28_a (0.92)	-0.27_a (1.02)	0.08 (0.98)	9.486 (3)*
Personal network support as measured by resource generator^b	0.10 (0.62)	−0.03 (1.23)	0.03 (0.80)	0.14 (0.62)	0.03 (0.94)	1.221 (3)
Organizational network support as measured by resource generator^c	0.06 (0.82)	0.09 (0.95)	−0.18 (1.28)	0.23 (0.59)	0.02 (1.01)	1.980 (3)
Sociability outside family^d	0.01 (0.88)	0.00(0.82)	0.21 (0.64)	0.16 (0.78)	0.08 (0.78)	1.972 (3)
Sociability within family^d	−0.06 (0.95)	0.06 (0.65)	-0.01 (0.85)	0.11 (0.86)	0.02 (0.80)	1.138 (3)

Notes. Group means in a row with a common superscript letter are significantly different at the p < .05 based on Dunn’s test post-hoc pairwise comparisons. *p ≤ .05.

^aN = 236.

^bN = 234.

^cN = 202.

^dN = 184.

RQ4 aimed to disentangle differences in internet access and proxy internet use. In general, the results showed that the internet access infrastructure of the four groups closely mirrored their internet skills and uses (see Table 6). Notably, while almost all Savvy users (97.1%) used at least two devices to access the internet, this percentage was significantly smaller among Reluctant users (31.2%). In fact, even more Savvy users (98.5%) relied on PCs and other devices to access the internet, whereas 56.2% of Reluctant users went online only on PCs. Apprehensive and Level-headed users did not show significant differences in internet access, even though there were more Apprehensive users than Level-headers (73.3% vs 64.8%) who used at least two internet-enabled devices. Thus, more Level-headers (28.6%) also used only PCs to go online. With reference to proxy internet use, more Apprehensive (68.9%) than Reluctant users (56.2%) compensated for their lack of internet skills with help from other internet users. Level-headed and Savvy users included a significantly smaller percentage of users-by-proxy (41.1% and 41.2%), but the latter (63.2%) were much more likely than the former (29.7%) to do something online on behalf of others. Probably due to their low internet skills, the proportion of proxy internet users among Apprehensive (20.0%) and Reluctant users (3.1%) was significantly smaller. The respondents in these two groups were also significantly less likely to be engaged in both proxy internet use roles. In fact, they were most likely only users-by-proxy (53.3% and 56.3%)––in contrast with 27.9% of Savvy users who get actively involved in both proxy roles. Interestingly, Level-headed users include the largest percentage of respondents (43.3%) who are neither user-by-proxy nor proxy users. Despite having the same level of internet skills as Savvy users, in withdrawing from proxy internet use, they resemble less-skilled Reluctant users.

Table 6.

Differences in Internet Access and Proxy Internet Use Between Clusters.^a

Variable	Category	Apprehensive users (%)	Level-headed users (%)	Savvy users (%)	Reluctant users (%)	Total (%)	χ ² (df) ^b
Devices used to access internet	One	26.7	35.2	2.9	68.8	28.8	48.965 (3)**
Devices used to access internet	Two or more	73.3	64.8	97.1	31.2	71.2
PC internet access	PC-only	22.2	28.6	1.5	56.2	23.3	39.011 (3)**
PC internet access	PC and other devices	77.8	71.4	98.5	43.8	76.7
User-by-proxy	Yes	68.9	41.1	41.2	56.2	48.5	11.686 (3)**
User-by-proxy	No	31.1	58.9	58.8	43.8	51.5	11.686 (3)**
Proxy internet user	Yes	20.0	29.7	63.2	3.1	33.9	44.248 (3)**
Proxy internet user	No	80.0	70.3	36.8	96.9	66.1	44.248 (3)**
Engagement in proxy internet use	Only user-by-proxy	53.3	26.7	13.2	56.3	31.9	57.821 (9)**
	Only proxy internet user	4.4	15.6	35.3	3.1	17.4
	User-by-proxy and proxy internet user	15.6	14.4	27.9	0	16.6
	Neither user-by-proxy nor proxy internet user	26.7	43.3	23.5	40.6	34.0

Notes. N = 236. *.01 < p ≤ .05, **p ≤ .01.

^aResponse percentages may not add up to 100% due to rounding.

^bIn the case that cells had expected counts of less than 5, the Monto Carlo simulation for computing p-values was run on 10,000 replicates. Response percentages may not add up to 100% due to rounding.

Discussion

Following from the framework of the second-level digital divide (Hargittai & Hsieh, 2012), we examined how internet users in late middle and late adulthood form different clusters according to their levels of skills and patterns of internet use. By modeling four types of skills and four types of uses, and by using a range of indicators to describe different clusters, the study provides a comprehensive depiction of gradations of digital inclusion among aging internet users. Our results confirm the conceptual difference between the notions of internet skills and uses (e.g., van Deursen et al., 2017; Helsper & Eynon, 2013) by showing that the relationship between internet skills and uses is not necessarily positive—higher levels of skills do not always lead to more varied use. Gradations in digital inclusion are much more complex, as we discuss further below.

We identified four distinct groups of users, which correspond to the groups identified by Quan-Haase et al. (2018) remarkably well, considering the different methodological approaches of the two studies and differences in the cultural contexts and ages of respondents. Thus, we adopted three group names from their typology: Reluctant, Apprehensive, and Savvy users. The fourth identified group was distinct, and we named it Level-headed users. In both studies, two groups that follow the expected positive association between skills and uses were identified: Savvy and Reluctant users. Savvy users are a typical highly digitally included group: they are younger older adults, are highly educated, have high social capital and income, have diverse internet access, and also engage in proxy internet use. By contrast, Reluctant users, who are the most digitally disadvantaged, especially in terms of economic and social uses, are older and less educated adults with lower incomes. They also have low social capital and less diverse internet access. Their lack of engagement in social uses may be explained by their relatively high reported levels of sociability outside and within the family. Quan-Haase et al. (2018) report that these users prefer to engage in task-oriented internet activities, and we found their self-reported operational skills considerably low. This may indicate that Reluctant users’ low engagement is in fact a reflection of their low skills levels.

Two identified groups are singular in exhibiting a negative association between skills and uses. Apprehensive users have relatively low internet skills but engage in a wide variety of internet uses, except for economic uses, which involve activities requiring financial transactions. Quan-Haase et al. (2018), in fact, noted that Apprehensive users are often nervous about doing something wrong. Their apprehensiveness in economic uses also echoes Helsper and Eynon’s (2013) observation that not all skills are equivalent and that confidence in one area does not necessarily translate to engagement in another. For social, cultural, and personal uses, where they were the second most engaged group, Apprehensive users seem to follow a “learning-by-doing” logic—while portraying lower skills, they present high levels of use. By contrast, Level-headed users tended to have high internet skills (except for creative skills) but reported only moderate use. This result is especially interesting, considering that most Level-headed users were 50–59 years old, whereas the majority of Apprehensive users were above 60 years of age. Further, Level-headed users were financially better off, had better education, and worked in higher-skilled jobs than Apprehensive users. Previous research suggests that education is usually positively related to internet use (Hargittai & Dobransky, 2017; Olsson & Viscovi, 2020; van Deursen & Helsper, 2015). We speculate that Level-headers are less digitally engaged because internet use does not fit their broader life contexts. Because a majority were working in high-skill occupations their relatively high internet skills could have been acquired at work (König et al., 2018). These opposing groups of older users suggest that digital inclusion may be shaped by age-graded life events and lifestyles.

This argument is further warranted when comparing Level-headed and Savvy users, because these clusters have a very similar age composition (most users were 50–59 years old) and very similar levels of skills (except for creative skills) but very different levels of use. In general, previous research is consistent in showing that younger older adults tend to engage online more extensively (Hargittai & Dobransky, 2017; van Deursen & Helsper, 2015). Likewise, previous typologies of older internet users characterized younger older adults as more digitally engaged (van Boekel et al., 2017; Vulpe & Crăciun, 2020). While most socio-demographic variables explored were significantly indicative of group membership, other examined variables may help better explain the differences between groups.

Specifically, access characteristics and proxy internet use also shaped group membership. Although Apprehensive users reported similar levels of social capital as Level-headed users, they were more likely users-by-proxy. We assume that engagement in indirect internet use may result in Apprehensive users’ more varied internet use, regardless of their relatively low levels of skills. Differences in internet access were also pronounced between Level-headed and Savvy users. Level-headers reported significantly poorer internet access, with about a third of them reporting going online only on one device and using PC-only access. Because their skill levels were relatively high, this finding suggests that poor quality internet access may in fact impede conversion of internet skills to uses. Further, while similar proportions of users in both groups engaged in use-by-proxy, Savvy users were much more likely to also act as proxy internet users. This finding suggests that among aging internet users, offering help to others as proxy users is driven by concrete experience with internet use, which is reflected in Savvy’s more varied internet access and higher levels of internet use (Quan-Haase et al., 2018).

Of the five dimensions of social networks and resources examined, only the social capital position generator significantly differentiated the groups. Savvy users had greater social capital than Reluctant users, while Apprehensive and Level-headed users did not differ from the other groups. This finding contrasts with previous studies that found social capital facilitating internet access and use (Chen, 2013; Hlebec et al., 2006). However, aging affects the size, composition, and interactions in individuals’ personal networks (c.f. Petrovčič et al., 2015), and these dynamics seem to not have an effect on older adults’ patterns of internet skills and uses (e.g., high sociability outside and within the family were characteristics of both Reluctant and Savvy users). Following the resources and appropriation theory (van Dijk, 2005), we may also speculate that among aging individuals, availability of resources in social networks plays a more important role at the stage of moving from non-use to (some) internet use (Friemel, 2016; Li & Chen, 2021) and not as much in shaping the second-level digital divide. For example, in a study in Slovenia, Dolničar et al. (2018) found that internet non-users who have (grand-)children in their socializing support network are more likely to engage in proxy internet use, thus gaining at least indirect internet access and bridging the first-level digital divide. This was also confirmed by Li and Chen (2021) in a study in the United States, who found that core tech support ties can act as important brokers to connect “digital laggards” to the internet.

Thus, it would be of interest in future to study the role of social resources in cultural contexts with a different level of internet uptake among the aging population. While Slovenia is an average-performing EU country in terms of internet access of the general population, it has a significantly smaller percentage of internet users among the 55–75 age group compared to the EU-27 average (2019: 47% vs 57%) (Eurostat, 2021). Relatedly, it would be interesting to replicate this study on a sample that would be more weighted towards retirement age (i.e., 65+) rather than working ages (i.e., 50–64) to make it more comparable with prior work on older internet users. It would also be worth considering the role of motivational aspects of digital engagement of aging internet users that were not included in this study with a comparative perspective to social resources. Having such information could also help us choose potentially more informative and less value-laden group labels. Finally, the use of survey data is accompanied by the limitations of the selected measures and biases in analytical procedures. For instance, using different social capital measures (Appel et al., 2014) could give a more detailed view of how social network resources shape the second-level digital divide among aging internet users.

Conclusion

In summary, three main conclusions can be drawn from this study. First, levels of both skills and uses need to be considered when depicting the second-level digital divide. The relationship between skills and uses is complex, and high skills do not always translate to varied use. Second, while socio-demographics remain important predictors of second-level digital inclusion, age-graded life events and different lifestyles of aging individuals may be equally important, which is also reflected in their differences in internet access and proxy internet use. Third, social networks and resources seem to have little impact on the second-level digital inclusion of aging users. We assume that social support is more important in overcoming the age-related first-level digital divide.

Footnotes

Acknowledgments

We are grateful for the assistance Špela Plemelj and Jerneja Laznik provided in data analysis. We would also like to thank the two anonymous reviewers for their insightful feedback on an earlier version of this manuscript.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Javna Agencija za Raziskovalno Dejavnost RS (BI-US/22-24-055, J5-2558, P5-0399, Z5-8234).

ORCID iDs

Andraž Petrovčič

Katja Prevodnik

Note

Appendix

Table A1.

Factor Loadings and Communalities Based on a Principal Components Analysis for Social Capital as Measured by Position Generator.

Variable	Factor loading
Maximum International Socio-Economic Index (ISEI)^a	.78
Range ISEI	.96
Mean ISEI	.91
Number of occupations mentioned	.93

Notes. N = 322. λ = 3.232, variance explained = .81, Kaiser-Meyer-Olkin measure = .63, Barlett’s test of sphericity: χ² = 1745.508, df = 6, p < .01, Cronbach’s α = 0.94.

^aDetails about ISEI score computation are presented in Sapin et al. (2020).

Table A2.

Factor Loadings Based on a Non-Linear Principal Components Analysis for Personal Network Support as Measured by Resource Generator.

Variable	Factor loading
Help you with a household or a garden job that you cannot do yourself?	.73
Help you around your home if you were sick and had to stay in bed for a few days?	.76
Be there for you if you felt a bit down or depressed and wanted to talk about it?	.67
Give you advice about family problems?	.59
Enjoy a pleasant social occasion with?	.61

Notes. N = 314. λ = 2.27, variance explained = .45, Kaiser-Meyer-Olkin measure = .61, Barlett’s test of sphericity: χ² = 282.3, df = 10, p < .01, Cronbach’s α = 0.63.

Table A3.

Factor Loadings Based on a Non-Linear Principal Components Analysis for Organizational Network Support as Measured by Resource Generator.

Variable	Factor loading
Help you if you needed to borrow a large sum of money?	.62
Help you if you needed to find a job?	.70
Help you with administrative problems or official paperwork?	.52
Help you if you needed to find a place to live?	.78
Look after you if you were seriously ill?	.72

Notes. N = 267. λ = 2.271, variance explained = .45, Kaiser-Meyer-Olkin measure = .67, Barlett’s test of sphericity: χ² = 171.8862, df = 10, p < .01, Cronbach’s α = .64.

Table A4.

Factor Loadings Based on a Principal Components Analysis with Varimax Rotation for Sociability Within and Outside Family.

Variable	Factor loading
Variable	Sociability outside family^a	Sociability within family^b
Go out with friends/acquaintances	.77	.05
Contact with parents or children	.05	.76
Contact with siblings	.17	.43
Contact with close friends	.55	.21

Notes. N = 243.

^aλ = 0.985, variance explained = .23.

^bλ = 0.288, variance explained = .20.

Kaiser-Meyer-Olkin measure = .56, Barlett's test of sphericity: χ² = 92.656, df = 6, p < .01. Cronbach’s α = .54. Variables measuring how often respondents make new friends (Q18), with how many people do they interact on a typical day (Q19) and how often they are in contact with other family members (Q24), were eliminated because they did not contribute to a simple factor structure and failed to meet minimum criteria of having a primary factor loading ≥ .4, and no cross-loading ≥ .3.

References

AAPOR . (2016). Standard definitions final dispositions of case codes and outcome rates for surveys (9th ed.). http://www.aapor.org/AAPOR_Main/media/publications/Standard-Definitions20169theditionfinal.pdf

Antonenko

P. D.

Toy

Niederhauser

D. S.

(2012). Using cluster analysis for data mining in educational technology research. Educational Technology Research and Development, 60(3), 383–398. https://doi.org/10.1007/s11423-012-9235-8

Appel

Dadlani

Dwyer

Hampton

Kitzie

Matni

Z. A.

Moore

Teodoro

(2014). Testing the validity of social capital measures in the study of information and communication technologies. Information Communication and Society, 17(4), 398–416. https://doi.org/10.1080/1369118X.2014.884612

Baltes

P. B.

Reese

H. W.

Lipsitt

L. P.

(1980). Life-span developmental psychology. Annual Review of Psychology, 31(1), 65-110. https://doi.org/10.1146/annurev.ps.31.020180.000433.

Blank

Dutton

W. H.

Lefkowitz

(2020). OxIS 2019: Digital Divides in Britain are Narrowing but Deepening. Oxford Internet Institute. University of Oxford.

Blank

Groselj

(2014). Dimensions of internet use: Amount, variety, and types. Information, Communication & Society, 17(4), 417–435. https://doi.org/10.1080/1369118X.2014.889189

Brandtzaeg

P. B.

Heim

Karahasanović

(2010). Understanding the new digital divide–A typology of Internet users in Europe. Journal of Human Computer Studies, 69(3), 123-138. https://doi.org/10.1016/j.ijhcs.2010.11.004.

Chen

(2013). The implications of social capital for the digital divides in America. Information Society, 29(1), 13–25. https://doi.org/10.1080/01972243.2012.739265

Choi

E. Y.

Kim

Chipalo

Lee

H. Y.

Meeks

(2020). Does perceived ageism widen the digital divide? And does it vary by gender? Gerontologist, 60(7), 1213–1223. https://doi.org/10.1093/geront/gnaa066

10.

Cotten

S.R.

Francis

Kadylak

Rikard

R. V.

Huang

Ball

DeCook

(2016). A Tale of Two Divides: Technology Experiences Among Racially and Socioeconomically Diverse Older Adults. In: Zhou, J., Salvendy, G. (Eds.) Human Aspects of IT for the Aged Population. Design for Aging. ITAP 2016. Lecture Notes in Computer Science, vol 9754, pp 167–177. Springer, Cham. https://doi.org/10.1007/978-3-319-39943-0_16

11.

Correa

Pavez

Contreras

(2017). Beyond access: A relational and resource-based model of household internet adoption in isolated communities. Telecommunications Policy, 41(9), 757–768. https://doi.org/10.1016/j.telpol.2017.03.008

12.

Dolničar

Grošelj

Filipovič Hrast

Vehovar

Petrovčič

(2018). The role of social support networks in proxy Internet use from the intergenerational solidarity perspective. Telematics and Informatics, 35(2), 305–317. https://doi.org/10.1016/j.tele.2017.12.005

13.

Eurostat . (2021). Statistics | eurostat. Eurostat. https://ec.europa.eu/eurostat/databrowser/view/ISOC_CI_IFP_IU/default/table?lang=en

14.

Francis

Ball

Kadylak

Cotton

S. R.

(2019). Aging in the digital age: Conceptualizing technology adoption and digital inequalities. In Neves

B. B.

Vetere

(Eds.), Ageing and digital technology (pp 35–49). Springer.

15.

Friemel

T. N.

(2016). The digital divide has grown old: Determinants of a digital divide among seniors. New Media and Society, 18(2), 313–331. https://doi.org/10.1177/1461444814538648

16.

Grošelj

Matjašič

(2019). Oblike in dejavniki dostopa do interneta v Sloveniji [Forms and factors of internet access in Slovenia]. Teorija in Praksa, 56(2), 601–622.

17.

Grošelj

Reisdorf

B. C.

Dolničar

Petrovčič

(2022). A decade of proxy internet use: The changing role of socio-demographics and family support in nonusers’ indirect internet access to online services. The Information Society, 1, 1–17. https://doi.org/10.1080/01972243.2022.2080782

18.

Grošelj

Reisdorf

B. C.

Petrovčič

(2019). Obtaining indirect internet access: An examination how reasons for internet non-use relate to proxy internet use. Telecommunications Policy, 43(3), 213–224. https://doi.org/10.1016/j.telpol.2018.07.004

19.

Grošelj

van Deursen

A. J. A. M.

Dolničar

Burnik

Petrovčič

(2021). Measuring internet skills in a general population: A large-scale validation of the short Internet Skills Scale in Slovenia. The Information Society, 37(2), 63-81. https://doi.org/10.1080/01972243.2020.1862377.

20.

Hänninen

Taipale

Luostari

(2021). Exploring heterogeneous ICT use among older adults:The warm experts’ perspective. New Media and Society, 23(6), 1548-1601. https://doi.org/10.1177/1461444820917353.

21.

Hargittai

Dobransky

(2017). Old dogs, new clicks: Digital inequality. Canadian Journal of Communication, 42(2), 195–212. https://doi.org/10.22230/cjc2017v42n2a3176

22.

Hargittai

Hsieh

Y. P.

(2012). Succinct survey measures of web-use skills. Social Science Computer Review, 30(1), 95–107. https://doi.org/10.1177/0894439310397146

23.

Hargittai

Piper

A. M.

Morris

M. R.

(2019). From internet access to internet skills: Digital inequality among older adults. Universal Access in the Information Society, 18(4), 881–890. https://doi.org/10.1007/s10209-018-0617-5

24.

Helsper

E. J.

Eynon

(2013). Distinct skill pathways to digital engagement. European Journal of Communication, 28(6), 696–713. https://doi.org/10.1177/0267323113499113

25.

Hlebec

Lozar Manfreda

Vehovar

(2006). The social support networks of internet users. New Media & Society, 8(1), 9-32. https://doi.org/10.1177/1461444806058166.

26.

Hunsaker

Hargittai

(2018). A review of Internet use among older adults. New Media and Society, 20(10), 3937–3954. https://doi.org/10.1177/1461444818787348

27.

Hunsaker

Nguyen

M. H.

Fuchs

Djukaric

Hugentobler

Hargittai

(2019). “He explained it to me and I also did it myself”: How older adults get support with their technology uses. Socius: Sociological Research for a Dynamic World, 5(January, 2019), Article 237802311988786. https://doi.org/10.1177/2378023119887866

28.

ISSP Research Group . (2019). International social survey Programme: Social networks and social resources - ISSP 2017. GESIS Data Archive. https://doi.org/10.4232/1.13322

29.

Johnson

R. A.

Wichern

D. W.

(2007). Applied multivariate statistical analysis (6th Ed.). Pearson Prentice Hall.

30.

König

Seifert

Doh

(2018). Internet use among older Europeans: An analysis based on SHARE data. Universal Access in the Information Society, 17(3), 621–633. https://doi.org/10.1007/s10209-018-0609-5

31.

Chen

(2021). Core tech support networks and digital inequalities in American disadvantaged urban communities. Journal of Computer-Mediated Communication, 26(2), 91–107. https://doi.org/10.1093/jcmc/zmaa019

32.

Little

R. J. A.

Rubin

D. B.

(1987). Statistical analysis with missing data. Wiley.

33.

Marler

Hargittai

(2022). Division of digital labor: Partner support for technology use among older adults. New Media & Society. https://doi.org/10.1177/14614448211068437

34.

Matthews

Nazroo

Marshall

(2019). Digital inclusion in later life: Cohort changes in internet use over a ten-year period in England. Ageing and Society, 39(9), 1914–1932. https://doi.org/10.1017/S0144686X18000326

35.

Nimrod

(2013). Probing the audience of seniors’ online communities. Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 68(5), 773–782. https://doi.org/10.1093/geronb/gbt059

36.

Nimrod

(2017). Older audiences in the digital media environment. Information Communication and Society, 20(2), 233–249. https://doi.org/10.1080/1369118X.2016.1164740

37.

Ofcom . (2021). Adults’ media use and attitudes report 2020/21. https://www.ofcom.org.uk/__data/assets/pdf_file/0025/217834/adults-media-use-and-attitudes-report-2020-21.pdf

38.

Olsson

Samuelsson

Viscovi

(2019). At risk of exclusion? Degrees of ICT access and literacy among senior citizens. Information Communication and Society, 22(1), 55–72. https://doi.org/10.1080/1369118X.2017.1355007

39.

Olsson

Viscovi

(2020). Who actually becomes a Silver Surfer? Prerequisites for digital inclusion. Javnost - The Public, 27(3), 230-246. https://doi.org/10.1080/13183222.2020.1794403.

40.

Petrovčič

Fortunati

Vehovar

Kavčič

Dolničar

(2015). Mobile phone communication in social support networks of older adults in Slovenia. Telematics and Informatics, 32(4), 642–655. https://doi.org/10.1016/j.tele.2015.02.005

41.

Quan-Haase

Williams

Kicevski

Elueze

Wellman

(2018). Dividing the grey divide: Deconstructing myths about older adults’ online activities, skills, and attitudes. American Behavioral Scientist, 62(9), 1207–1228. https://doi.org/10.1177/0002764218777572

42.

R Core Team . (2019). R: A language and environment for statistical computing. https://www.r-project.org/

43.

Reisdorf

B. C.

Petrovčič

Grošelj

(2021). Going online on behalf of someone else: Characteristics of Internet users who act as proxy users. New Media & Society. https://doi.org/10.1177/1461444820928051

44.

Rousseeuw

P. J.

(1987). Silhouettes: A graphical aid to the interpretation and validation of cluster analysis. Journal of Computational and Applied Mathematics, 20(November 1987), 53-65. https://doi.org/10.1016/0377-0427(87)90125-7.

45.

Sapin

Joye

Wolf

(2020). The ISSP 2017 social networks and social resources module. International Journal of Sociology, 50(1), 1–25. https://doi.org/10.1080/00207659.2020.1712157

46.

Statistical Office of the Republic of Slovenia . (2020). Income of the population, Slovenia, 2014–2018. https://www.stat.si/StatWeb/en/news/Index/9178

47.

Tibshirani

Walther

Hastie

(2001). Estimating the number of clusters in a data set via the gap statistic. Journal of the Royal Statistical Society: Series B (Statistical Methodology) 63(2), 411–423. https://doi.org/10.1111/1467-9868.00293

48.

van Boekel

L. C.

Peek

S. T.

Luijkx

K. G.

(2017). Diversity in older adults’ use of the internet: Identifying subgroups through latent class analysis. Journal of Medical Internet Research, 19(5), e180. https://doi.org/10.2196/jmir.6853.

49.

van Deursen

A. J. A. M.

Courtois

van Dijk

J. A. G. M.

(2014). Internet skills, sources of support, and benefiting from internet use. International Journal of Human-Computer Interaction, 30(4), 278-290. https://doi.org/10.1080/10447318.2013.858458.

50.

van Deursen

A. J. A. M.

Helsper

E. J.

(2015). A nuanced understanding of Internet use and non-use among the elderly. European Journal of Communication, 30(2), 171–187. https://doi.org/10.1177/0267323115578059

51.

van Deursen

A. J. A. M.

Helsper

E. J.

Eynon

(2016). Development and validation of the Internet Skills Scale (ISS). Information Communication and Society, 19(6), 804-823. https://doi.org/10.1080/1369118X.2015.1078834.

52.

van Deursen

A. J. A. M.

Helsper

E. J.

Eynon

van Dijk

J. A. G. M.

(2017). The Compoudness and Sequentiality of Digital Inequality. International Journal of Communication, 11, 452-473. https://ijoc.org/index.php/ijoc/article/view/5739.

53.

van Dijk

J. A. G. M.

(2005). The Deepening Divide: Inequality in the Information Society. Sage. https://doi.org/10.4135/9781452229812.

54.

Vulpe

Crăciun

(2020). Silver surfers from a European perspective: Technology communication usage among European seniors. European Journal of Ageing, 17(1), 125–134. https://doi.org/10.1007/s10433-019-00520-2

55.

Vuori

Holmlund-Rytkönen

(2005). 55 + people as internet users. Marketing Intelligence and Planning, 23(1), 58–76. https://doi.org/10.1108/02634500510577474

56.

Wang

Zhang

Wellman

(2018). Are older adults networked individuals? Insights from East Yorkers’ network structure, relational autonomy, and digital media use. Information Communication and Society, 21(5), 681–696. https://doi.org/10.1080/1369118X.2018.1428659