Article

Internet Usage among Senior Citizens: Self-Efficacy and Social

Influence Are More Important than Social Support
Mirjana Pejić Bach , Lucija Ivančić , Vesna Bosilj Vukšić, Ana-Marija Stjepić * and Ljubica Milanović Glavan

Faculty of Economics & Business, University of Zagreb, 10000 Zagreb, Croatia; mpejic@efzg.hr (M.P.B.);
livancic@efzg.hr (L.I.); vbosilj@efzg.hr (V.B.V.); lmilanovi@efzg.hr (L.M.G.)
* Correspondence: astjepic@efzg.hr

Abstract: For more than two decades, developed countries have been confronted with two trends
that have implications for the emergence of engaging senior citizens in the digital environment.
On the one hand, there is an increasing proportion of senior citizens in the total population. On
the other hand, the application of ICT in all areas of life and business is accelerating. This paper
investigates the relationship between self-efficacy, social support, and social influence on Internet
usage among senior citizens in Croatia. Survey research was conducted on a sample of Croatian
senior citizens, and a structural equation mode was developed for testing the research hypothesis.
Self-efficacy influenced both the Intensity and obstacles of Internet usage in a positive and negative
manner, respectively. Social influence directly decreased the obstacles to Internet usage, while the
relationship with the Intensity of the Internet was indirect through self-efficacy. Social support had
only an indirect association with Intensity of Internet usage. Results have relevant implications for
programmes aiming to enhance Internet usage among senior citizens, which should focus on the
educational programmes fostering perceived self-efficacy of Internet usage among senior citizens.

Keywords: senior citizens; Internet usage intensity; Internet usage obstacle; self-efficacy; social
influence; social support; older people
Citation: Pejić Bach, M.; Ivančić, L.;
Bosilj Vukšić, V.; Stjepić, A.-M.;
Milanović Glavan, L. Internet Usage
among Senior Citizens: Self-Efficacy 1. Introduction
and Social Influence Are More
Usage of the Internet and information and communication technology (ICT) has
Important than Social Support. J.
Theor. Appl. Electron. Commer. Res.
significantly disrupted both workplaces [1] and everyday life [2]. It is especially important
2023, 18, 1463–1483. https://
for senior citizens (65 years and older), whose Internet usage significantly increased from
doi.org/10.3390/jtaer18030074 the beginning of the Web 2.0 technology era in 2010 [3]. However, senior citizens have
problems accessing and using modern digital devices and thus face challenges in the
Academic Editor: Luis Felipe
qualitative use of the Internet [4]. Various authors have diverse focuses when researching
Luna-Reyes
the use of the Internet in the older population, investigating the availability of Internet
Received: 9 May 2023 services [5], the level of trust and readiness when using the Internet and Internet-based
Revised: 14 August 2023 applications [6,7], and awareness of the risks of using the Internet [8].
Accepted: 28 August 2023 As reported by Croatian Bureau of Statistics data, the average age of the Croatian
Published: 31 August 2023 population increased from 37 years in 1991 to 44 years in 2021 [9], following the trend in
most of the European countries [10]. According to [11], there is a noticeable increase in the
proportion of people over the age of 65, and their share in the total population is expected
to increase. Thus, in 2001, 15.7% of people living in the Republic of Croatia were over the
Copyright: © 2023 by the authors.
age of 65; in 2013, this share was 17.7%; and the most recent census revealed that nearly
Licensee MDPI, Basel, Switzerland.
25% of the Republic of Croatia’s population belonged to that segment [9,11].
This article is an open access article
A brief review of the literature reveals that a few studies conducted in Croatia address
distributed under the terms and
the issue of elderly digital inclusion. According to the results of a study conducted by [12],
conditions of the Creative Commons
due to the relatively low level of digital literacy among Croatian senior citizens, knowledge
Attribution (CC BY) license (https://
transfer “from the young to the elderly” needs to be made to achieve adequate inclusion
creativecommons.org/licenses/by/
4.0/).
of senior citizens in an information-based society. The findings of the study conducted

J. Theor. Appl. Electron. Commer. Res. 2023, 18, 1463–1483. https://doi.org/10.3390/jtaer18030074 https://www.mdpi.com/journal/jtaer
J. Theor. Appl. Electron. Commer. Res. 2023, 18 1464

by [13] show that the barriers related to seniors’ adoption of digital technologies and their
use of digital health and social care services provided by Croatian institutions are very
similar to those in many other countries. This paper is developing a structural equation
model that tests the association with self-efficacy, social impact, and social support on the
Intensity of Internet usage and observed barriers to Internet usage for senior citizens. This
research is conducted as part of the project “SENIOR 2030—a thematic network for active
ageing policy in Croatia”, developing a proposal for an active ageing strategy based on the
Silver Economy.
The remainder of the article is organised as follows. After the introduction, a literature
review and hypothesis development are presented. The methodology section describes the
data collection, the development of the research instrument, and the statistical analysis pro-
cess. Descriptive statistics, confirmatory factor analysis, and structural equation modelling
are used to analyse and test the data and hypotheses. The results are further analysed and
discussed. The limitations of the study are highlighted. Finally, the paper concludes with
a brief overview of the research findings. Limitations and recommendations for future
research are pointed out.

2. Literature Review and Hypothesis Development

In general, the term social influence refers to an alteration in the individual’s behaviour,
thinking, and feelings resulting from his presence and interaction with other people or
groups of people [14]. Likewise, in their work, [15] explains how the social influence
process certainly shapes the individual’s further behaviour and thinking. This influence
process lays the foundations for the later identity development of the individual, his later
decisions, and future socialisation [15]. Similarly, various authors accentuate how social
influence strongly impacts each person’s behaviour progress [16,17].
Ref. [16] developed the “attitude–social influence–efficacy model” in their work. Self-
efficacy can be explained as a tendency in which people are motivated to behave in a certain
way if they have sufficient self-confidence that they will be able to perform such behaviour
and that such behaviour will achieve the desired goal [18]. According to [17], there are three
types of self-efficacy which include (1) reactions of the individual that are motivated by self-
satisfaction and reactions that are motivated by dissatisfaction with oneself; (2) perception
of self-efficacy in achieving one’s goals; and (3) adjustment of planned objectives based on
one’s advancement. Accordingly, it can be concluded that self-efficacy and social influence
shape an emotional state, opinions, and motivations [19].
In addition to social influence and self-efficacy, many authors also emphasise the
concept of social support as one of the factors that influence the formation of an indi-
vidual in the context of his further behaviour [20]. Social support can be defined as the
individual’s perception or experience of care, appreciation, respect, acceptance, and social
inclusion, which can be provided by the society with which the individual interacts [21].
Therefore, social support for an individual can be provided by family, friends, colleagues,
acquaintances, and all other participants in the individual’s social life.
Social support, social influence, and self-efficiency interact, which is crucial in using
ICT among senior citizens [22]. According to all of the above and to examine the mutual
relations between the terms mentioned above: social influence, social support, and self-
efficacy, the three following hypotheses are defined and investigated in this work:
H1. Social influence is positively related to self-efficacy.
H2. Social support is positively related to self-efficacy.
H3. Social influence is positively related to social support.
When observing the use of information and communication technologies, including
the Internet, among older people, the authors of papers often emphasise self-efficacy as
one of the main determinants of the relationship [22–24]. According to [23], self-efficacy
can be defined as an individual’s positive or negative opinion of successfully using the
J. Theor. Appl. Electron. Commer. Res. 2023, 18 1465

Internet by themselves. In their work, ref. [25] describe the term Internet self-efficacy as the
ability of an individual to perform certain Internet activities independently to realise the
desired opportunities and services provided by the Internet. Consequently, the authors of
this work defined the following hypothesis:
H4. Self-efficacy is positively related to the Internet usage intensity.
Furthermore, many authors in their works point out that social influence encourages
older people to use ICT [22,23,26]. However, it is important to know that social influence in
such a context refers to recommendations and opinions about the use of the Internet and
digital technology that come from their immediate environment; that is, resources that are
given to them by their peers or family [22,26,27]. According to de Veer et al. (2015), social
influence as the perception of older users about the importance of using the new system,
in this case, the Internet, comes from the well-known theoretical model called the Unified
Theory of Acceptance and Use of Technology (UTAUT), which examines the plan of each
person in which he defines his behaviour in one way or another. Generally, within the
UTAUT model, the individual’s intention to use ICT is influenced by four determinants:
(1) the individual’s belief that the accepted technology will help him to improve the
performance of business tasks; (2) the individual’s attitude about the complexity of using
the accepted technology; (3) respect for the opinion of the environment about whether
or not one should use the planned technology in order to be accepted; and (4) personal
attitude about the technological and organisational readiness of the system in which it
operates [28]. Therefore, the following hypothesis was examined in this paper:
H5. Social influence is positively related to the Internet usage intensity.
Moreover, various authors emphasise social support as an important factor for suc-
cessfully using ICT and the Internet, such as [22,24,29,30]. According to [31], social support
is vital for older people to accept ICTs, such as using tablets for communication. Similarly,
ref. [32] points out the importance of social support in adopting and using information
and communication technologies and the Internet. Moreover, in their study, ref. [33,34]
accentuates the significance of social support as support from acquaintances in the context
of proper support in learning how to manage new technology or in the context of emo-
tional support in using information and communication technology. Accordingly, the next
hypothesis is imposed and investigated in this work:
H6. Social support is positively related to the Internet usage intensity.
As previously mentioned, self-efficacy is an individual’s awareness that they can
perform certain activities with the skills they possess [17,19,25]. Also, a person with a higher
level of self-efficacy will put more effort and willingness to overcome possible obstacles
that can stand in the way of preferred behaviour [17,19,25]. In this way, it is assumed
that an individual who believes that he has sufficient skills to use information technology,
hence, to search the Internet and use the provided Internet services, will definitely and
more easily overcome any obstacles in Internet usage [25]. Likewise, ref. [33,34] in his work
emphasises how it is necessary to improve self-efficacy due to the challenges of using the
Internet, specifically in online transactions. Accordingly, the authors of this paper want
to examine the negative influence of self-efficacy on Internet use obstacles by forming the
following hypothesis:
H7. Self-efficacy is negatively related to Internet usage obstacles.
Various authors are investigating how social influence can reduce potential obstacles
in new information and communication technologies [35–37]. For instance, the results
of [35] revealed how social influence positively reduces potential resistance to innovation
in banking services provided by the Internet. Similarly, ref. [38] emphasises the importance
of social influence, especially the influence of family members, in overcoming any insecu-
rities that could cause refusal of Internet usage. In their work, ref. [37] accentuates how
it is necessary to familiarise the individual’s social environment with the application of
J. Theor. Appl. Electron. Commer. Res. 2023, 18 ffi 1466

e-learning so that he receives sufficient information, knowledge, and skills from his environ-
ment to successfully overcome any challenges in learning how to use e-learning platforms.
Therefore, the following hypothesis is established for the examination in this paper:
ffi ff
H8. Social influence is negatively related to Internet usage obstacles.
In addition to self-efficacy and social influence, authors from different areas of research
note that social support also successfully contributes to eliminatingff potential obstacles in
ff
applying information communication technology [31,39]. Generally, social support can
help people more easily deal with the negative side effects of stress caused by different
ffi require them to change their existing behaviour or thinking [40,41].
life situations which
According to [39], social support enhances the motivation of older people to overcome pos-
sible difficulties thatfficome with learning new digital skills and generally using information
and communication technologies. Likewise, ref. [31] reveals that older people encounter
certain difficulties in learning how to work with information communication technology or
challenges in its active use; the help and encouragement of their family and friends will be
crucial in continuing to use the new technology and its functionalities. For that reason, the
final hypothesis is recognised and investigated in this work:
H9. Social support is negatively related to the Internet usage obstacles.
After defining the hypotheses following the results of the literature review of the
researched area, a conceptual research model was developed (Figure 1).

Figure 1. Research model. Source: Authors’ work.

3. Methodology
3.1. Data
The data analysed in this article were collected as part of a broader study on the
financial situation, well-being, and social inclusion of older people coordinated by the
Croatian Association of Pensioners as part of the project “Senior 2030—Thematic Network
for Active Ageing Policy in Croatia” funded by the European Social Fund. Accordingly, the
eligible population for this research was senior citizens from the territory of the Republic
of Croatia.
This research used a quota sample based on representative proportions according to
the 2011 census. Representativeness was considered for the region (NUTS 2 level) and
gender. Households were randomly selected, as was the selection of respondents in the
household if the household contained more than one person of older age. The term “senior
citizen” refers to persons 65 years of age and older, according to the OECD definition [42]
accepted in most developed countries. According to the United Nations report [43], the
number of senior citizens will increase to over 2 billion by 2050.
Therefore, data for this study were collected through a quantitative survey of the
sample of senior citizens (65+) in Croatia. The structured questionnaire was developed in
English, based on previously validated scales from the relevant literature [22,24,44]. The
questionnaire was then translated into Croatian for data collection. The responses to the
questionnaire were collected in January 2022 using the CATI (Computer Assisted Telephone
Interviewing) method.
J. Theor. Appl. Electron. Commer. Res. 2023, 18 1467

3.2. Research Instrument

The following constructs were used for the data collection: self-efficacy of Internet
usage, social influence for Internet usage, social support for Internet usage, obstacles in
Internet usage, and intensity of Internet usage. For the first three constructs, we used the
scales validated in previous research [22,24,30,44], while the constructs obstacles in Internet
usage and intensity of Internet usage were our operationalisations. These constructs are
presented in more detail in Table 1.
The self-efficacy measure assesses how easily respondents can use the Internet and
how comfortable they feel using it. Social influence measures include perceived incentives
to use the Internet from close social surroundings. The social support measure includes
activities from the social environment that aim to increase respondents’ Internet usage.
The above-mentioned variables had a seven-point Likert scale ranging from 1 (strongly
disagree) to 7 (strongly agree). Variable C11_10 was excluded from the further analysis
based on the results of exploratory factor analysis.
The obstacles in Internet usage dimension was operationalised with six items listing
common Internet use barriers among older people. Respondents could select none, one,
or more barriers that applied to their case. The total number of obstacles was used in the
model as an ordinal variable with values ranging from 0 (no obstacles) to 3 (three and
more obstacles).

Table 1. Research instrument.

Construct Code Research Item Item Measurement

Self-efficacy of C11_1 You feel comfortable using the Internet on your own.
Internet Usage C11_2 You can easily operate the Internet on your own. Likert scale
(Guan et al., 2017 [22];
(1—do not agree, 7—fully
Taylor and Todd, You feel comfortable using the Internet even if no one is
C11_3 agree)
1995 [44]; Hsieh, Rai and around you to tell you how to use it.
Keil, 2011 [24])
Social Influence on C11_4 Your family thinks that you should use the Internet.
Internet Usage C11_5 Your relatives think that you should use the Internet. Likert scale
(Guan et al., 2017 [22];
(1—do not agree, 7—fully
Taylor and Todd,
C11_6 Your friends think that you should use the Internet. agree)
1995 [44]; Hsieh, Rai and
Keil, 2011 [24])
You have someone to help solve Internet-related
C11_7
Social Support for problems.
Internet Usage You have friends or family to provide the necessary help
C11_8 Likert scale
(Guan et al., 2017 [22]; to use the Internet.
(1—do not agree, 7—fully
Wu and Rudkin, You have friends and family to help with solving
C11_9 agree)
2000 [30]; Hsieh, Rai and Internet-related problems.
Keil, 2011 [24]) You are supported by those around you when you have
C11_10
difficulty using the Internet.
The number of obstacles in Internet Usage that the
respondent faced in the last 3 months such as:
Lack of knowledge about using the device; Not having
anyone to help to install and use the device; The
0—no obstacles
appearance of the applications is complicated for the
Obstacles to Internet 1—one obstacle
C1_sum respondent and is not suitable for a user of the third age;
Usage 2—two obstacles
Respondent does not understand certain functions
3—three and more obstacles
because they are in a foreign language; Too much
distracting content (advertisements, etc.); Poorly
adapted for the vision, hearing and motor skills of
older people
J. Theor. Appl. Electron. Commer. Res. 2023, 18 1468

Table 1. Cont.

Construct Code Research Item Item Measurement

The number of purposes that the Internet was used for
by the respondent in the last 3 months such as:
Communication with family and friends via video calls
(ZOOM, Skype, Teams); Communication with family and
friends via e-mail and messaging applications (Viber,
WhatsApp, Messenger, etc.); Social networks (Facebook, etc.);
0—no form of use
News about everyday events (portals, magazines, etc.); For
1—one form of use
Intensity of Internet paying bills and other financial transactions; Ordering
C7_sum 2—two forms of use
Usage medical examinations in health institutions; For ordering
3—three and more forms
medicines and referrals, exchanging information with the
of use
family doctor; Internet shopping; To perform work (paid or
volunteering); For writing and other forms of creative
expression; For editing files (video, audio, photo); For
watching and listening to movies, music and photos; For
learning (independent or e-learning); Using the eCitizen
platform service; Using the service of the health care platform
Source: Authors’ work based on [22,24,30,44].

The intensity of the Internet usage dimension was operationalised with fifteen items
assessing the use of 15 Internet services. The items were measured as binary dummy
variables with a value of 0 if the respondent did not use the service and 1 if the respondent
used the service regularly. Respondents could select none, one, or more usage types that
applied to their case. The total number of forms of Internet usage was used in the model
for measuring the intensity of Internet usage as an ordinal variable (C7_sum), ranging from
0 (no form of use) to 3 (three and more forms of use).

3.3. Statistical Analysis

We applied several statistical analyses to ensure research or construct validity and to
test the hypothesis, as described in the continuation.
According to [45], a detailed assessment of the construct validity needs to be performed
to establish the accuracy of the research. Construct validity is the extent to which measured
items accurately reflect the theoretically designed latent constructs and should include four
components: face, nomological, convergent, and discriminant. While these four validity
assessments are appropriate and sufficient for establishing the construct validity when the
confirmatory factor analysis method is used [46], some research suggests combining the
confirmatory factor analysis with exploratory factor analysis [47,48], as was carried out in
this study.
To begin with, as we used previously developed scales from the literature for the
research instrument in this study and refined them to be suitable for our research context,
face validity was ensured [47].
Second, we calculated descriptive statistics of construct items presented in Table 2,
followed by a correlation analysis to check the measured variables’ consistency and ensure
that the instrument’s nomological validity was given [48].

Table 2. Sample characteristics.

Characteristics N %
Gender *
Male 273 38.9
Female 428 61.1
J. Theor. Appl. Electron. Commer. Res. 2023, 18 1469

Table 2. Cont.

Characteristics N %
Age (years) *
65–69 248 35.4
70–74 197 28.1
75–79 142 20.3
80–84 85 12.1
≥85 29 4.1
Size of the settlement *
Urban settlement 399 56.9
Suburban settlement 113 16.1
Rural settlement 182 26.0
A house outside the settlement 7 1.0
Education *
No education or less than eight grades of primary school 47 6.7
Elementary school (eight-year) 81 11.6
High school (three-year or four-year) 349 49.8
Higher education 224 32.0
* Total sample: N = 701. Source: Authors’ work, based on a survey, January 2022.

Third, we conducted an exploratory factor analysis to further inspect the instrument’s

validity. Even when scales that have already been validated in previous research are used,
as was the case with this study, an exploratory inspection of the measurement instrument
should be conducted, following the a priori assumption that any item can be associated
with any factor [47]. In exploratory factor analysis, we observed how each item loads on
each factor to preliminary examine the instrument scales. One of the items in the instrument
demonstrated cross-loading. Thus, it was excluded from further analysis, as it did not
represent uniquely one factor. All other items loaded strongly on one, as well as their
predicted factor from Table 1, and were used for creating the measurement model in the
fourth step.
Fourth, we evaluated a measurement model by using confirmatory factor analysis.
The purpose of evaluating a measurement model in SEM is to confirm how well the
items in the measurement instrument represent the hypothesised constructs for the data
obtained in the study, and it is a prerequisite for further assessment of the structural
model. We used a confirmatory factor analysis procedure suggested by [49] to inspect the
psychometric properties of the instrument. In this process, it is suggested first to assess
a set of goodness-of-fit indicators, including both absolute and incremental fit indices, to
test the overall validity and afterwards to test the convergent and discriminant validity
of the measurement model. Therefore, a set of goodness-of-fit indicators were used to
test the model’s overall validity, i.e., to check the representativeness of the model. We
checked absolute fit indices, namely the Chi-square index (χ2 ), Chi-square statistic ratio
or normed Chi-square (χ2 /df), Root Mean Square Error of Approximation (RMSEA),
Standardised Root Mean Residual (SRMR), and Goodness-of-fit index (GFI), as well as
incremental fit indices, namely Comparative Fit Index (CFI) and Tucker Lewis Index
(TLI). We then examined factor loadings, composite reliability (CR), and average variance
extracted (AVE) to determine each construct’s shared variance and test convergent validity.
Finally, we assessed discriminant validity by applying the Fornell–Larcker criterion [46] to
observe how uniquely the manifest variables represent each construct and other constructs.
This approach allowed us to obtain a complete picture of how the manifest (measured)
variables represent the latent variables (constructs), i.e., to achieve construct validity of the
measurement model, as suggested by [45].
Finally, after we evaluated the measurement model, we evaluated the structural
model. Multiple dependency relationships among constructs in the path model were tested
using structural equation modelling as proposed in (Hair, Babin and Krey, 2017 [46]). The
J. Theor. Appl. Electron. Commer. Res. 2023, 18 1470

overall fit of the model was tested using several indicators of goodness-of-fit, namely the
Chi-square index (χ2 ), Chi-square ratio or normed Chi-square (χ2 /df), Comparative Fit
Index (CFI), Tucker Lewis Index (TLI), Normed Fit Index (NFI), Non-Normed Fit Index
(NNFI), Root Mean Square Error of Approximation (RMSEA), and Standardised Root Mean
Residual (SRMR) (Hoe, 2008; Hair et al., 2014; Hair, Babin and Krey, 2017 [45,46,50]). Then,
the path model analysis was performed to examine the statistical significance of the path
coefficients and test the proposed hypothesis.
The statistical software JASP, version 0.16.3, was used to perform all the above-
mentioned statistical analysis procedures.

4. Results
4.1. Sample Characteristics
The sociodemographic characteristics of the sample are presented in Table 2. A
total of 701 responses were collected. A percentage of 61.1% of respondents identified
themselves as female, while 38.9% identified themselves as male. A percentage of 35.4% of
the respondents are between 65 and 69 years old; 28.1% are in the 70–74 age group; and
20.3% are in the 75–79 age group. Finally, 12.1% of respondents are between 80 and 84 years
of age, while 4.1% are 85 or older.
In the sample, the majority of respondents, 56.9%, live in urban settlements, and 16.1%
of respondents live in suburban settlements. In contrast, 26.0% of respondents live in rural
settlements. A percentage of 1% of the respondents from the sample say that they live in
a house isolated from a settlement. Most older people from our sample have a three- or
four-year high school diploma (49.8%), followed by 32.0% of respondents with a university
education. A percentage of 11.6% of the respondents have completed only elementary
school, and 6.7% reported having no formal education or completing only a few years of
elementary school.

4.2. Descriptive Statistics

The descriptive statistics for the three main constructs are given in Tables 3–5, and the
correlation matrix for the first three constructs is presented in Table 6.

Table 3. Descriptive statistics for the Internet usage.

N Minimum Maximum Mean Std. Dev.

Self-efficacy of Internet Usage
C11_1 701 1 7 4.11 2.355
C11_2 701 1 7 4.03 2.359
C11_3 701 1 7 3.98 2.348
Social Influence on Internet Usage
C11_4 701 1 7 4.12 2.186
C11_5 701 1 7 3.93 2.174
C11_6 701 1 7 3.77 2.134
Social Support for Internet Usage
C11_7 701 1 7 4.60 2.284
C11_8 701 1 7 4.62 2.315
C11_9 701 1 7 4.58 2.335
Source: Authors’ work, based on a survey, January 2022.

The highest mean value for the construct Internet usage (4.62) is obtained for the
variable C11_8 (Table 3). Variable C11_8 measures help from friends and family in Internet
services, suggesting that the strongest incentive for older people to use the Internet comes
from concrete informatics support from their closest social relationships. The smallest
mean value (3.77) was obtained for variable C11_6, which measures the perceived social
influence of friends as an incentive for Internet use. The smallest mean value of this Internet
J. Theor. Appl. Electron. Commer. Res. 2023, 18 1471

usage factor indicates that the social influence of friends to use the Internet is relatively low
compared to other Internet usage factors in the older population.

Table 4. Internet usage obstacles and Internet usage intensity.

Internet Usage Obstacles Internet Usage Intensity

# of Obstacles N % # of Usages N %
0—no obstacles 258 36.8% 0—no usage 295 42.1%
1—one obstacle 279 39.8% 1—one form of use 42 6.0%
2—two obstacles 81 11.6% 2—two forms of use 55 7.8%
3—three and more 83 11.8% 3—three and more 309 44.1%
Total 701 100% Total 701 100%
Source: Authors’ work, based on a survey, January 2022.

Table 5. Spearman’s rho correlation analysis of C11_1–C11_9 variables.

Variable C11_1 C11_2 C11_3 C11_4 C11_5 C11_6 C11_7 C11_8 C11_9
C11_1 1.000
C11_2 0.877 * 1.000
C11_3 0.875 * 0.874 * 1.000
C11_4 0.545 * 0.549 * 0.548 * 1.000
C11_5 0.519 * 0.514 * 0.517 * 0.771 * 1.000
C11_6 0.568 * 0.560 * 0.579 * 0.720 * 0.739 * 1.000
C11_7 0.524 * 0.521 * 0.509 * 0.581 * 0.512 * 0.507 * 1.000
C11_8 0.500 * 0.507 * 0.489 * 0.563 * 0.525 * 0.484 * 0.832 * 1.000
C11_9 0.521 * 0.524 * 0.500 * 0.570 * 0.526 * 0.504 * 0.826 * 0.830 * 1.000
Note: * statistically significant at 1%. Source: Authors’ work, based on a survey, January 2022.

Table 6. Rotated factor matrix for three factors.

Item Loadings
Item
PC1 PC2 PC3
C11_1 0.881
C11_2 0.878
C11_3 0.880
C11_4 0.795
C11_5 0.846
C11_6 0.794
C11_7 0.854
C11_8 0.868
C11_9 0.856
C11_10 0.593 0.509
Source: Authors’ work, based on a survey, January 2022.

The intensity of obstacles to Internet usage is summarised in Table 4. As shown in

Table 4, the majority of respondents, with a percentage of 39.8%, say they encounter only
one obstacle when using Internet services. This percentage is followed by a relatively
high percentage of 36.8% of respondents who indicated that they do not encounter any
obstacles when using the Internet. Finally, 11.6% of older people in our sample identi-
fied two obstacles, and 11.8% said they encountered three or more obstacles when using
Internet services.
The intensity of Internet services usage is summarised as follows: 44% of respondents
state they have three or more reasons for using the Internet, i.e., three or more types of
services that they regularly use, which represents the majority of the sample. A percentage
of 7.8% say they regularly use two types of Internet services, and 6.0% say they regularly
use one Internet service. However, 42.1% of respondents also do not regularly use a single
Internet service.
J. Theor. Appl. Electron. Commer. Res. 2023, 18 1472

A non-parametric correlation analysis was conducted for the items of the construct
Internet usage to assess the consistency of the measurement instrument. Items of the same
construct are expected to correlate, indicating that they are similar enough to measure the
same variable in the instrument.
Spearman correlation coefficients are presented in Table 5, and they show predomi-
nantly high correlations (>0.5) between construct variables, with several values below but
close to the 0.5 value. Since no negative or low correlations were found, and the correlation
coefficients are positive and strong, the correlation analysis confirms the consistency of the
measurement instrument [47].
In addition, the coefficients are very high (>0.7) between the items of each dimension
of Internet usage, i.e., between the items of self-efficacy (C11_1–C11_3), social influence
(C11_4–C11_6), and social support (C11_7–C11_9). Since the coefficients for the items in
a particular dimension are higher than those between the items belonging to different
dimensions, the correlation analysis first gave insight into the uniqueness of the latent
variables, which will be further investigated in the CFA analysis and confirms the nomo-
logical validity.

4.3. Exploratory Factor Analysis

We used an exploratory factor analysis (EFA) to examine the underlying structure
of the measurement instrument, following the a priori assumption that any item may be
associated with any factor [50]. Since the main statistical analysis method used in the
study was confirmatory factor analysis, EFA was used as a supplement for the preliminary
exploration of instrument properties.
Three latent factors were extracted using the principal component analysis with
varimax rotation, as presented in Table 6. Common thresholds for moderate and acceptable
loadings to interpret the factors range from 0.4 to 0.7, while values of 0.8 or greater indicate
a strong accuracy of the items [51]. One of our items, C11_10, demonstrated cross-loading,
as it loaded on two factors (PC1 and PC2). According to [51], the item should be discarded
if loaded adequately to strong, i.e., 0.5 or above on several factors. Therefore, since item
C11_10 had two loaders above the cut-off value of 0.5, it was excluded from further analysis,
as it did not uniquely represent one factor. All other items (C11_1–C11_9) had loaders
above 0.7 and 0.8, i.e., loaded strongly on one factor.
Since the convergent validity was examined using the EFA, a variable was deleted; as
a result, we continued with the validation of factor analysis in the following step, using
nine items (C11_1–C11_9) in a three-factor model.

4.4. Confirmatory Factor Analysis

We used a confirmatory factor analysis (CFA) procedure to inspect the properties of
the measurement model. As a first step in evaluating our measurement model, we observed
overall model fit using common goodness-of-fit indicators to check the representativeness
of the model. After that, we tested convergent validity by examining factor loadings,
composite reliability (CR), and average variance extracted (AVE) for each construct. Finally,
discriminant validity was assessed by applying the Fornell–Larcker criterion.
CFA analysis was performed for the entire sample. A three-factor model was estab-
lished, with the factors being namely self-efficacy, social influence, and social support. The
graph of items with their established factors is presented in Figure 2. Manifest variables
C11_1–C11_3 were loaded on the self-efficacy factor; the variables C11_4–C11_6 were
loaded on the social influence factor; and the variables C11_7–C11_9 were loaded on the
social support factor. All latent factors were assumed to be correlated with each other, as
the variance of the factors was set to 1.0. A covariance matrix with a maximum likelihood
(ML) method was used to estimate the parameters.
 ffi

J. Theor. Appl. Electron. Commer. Res. 2023, 18 1473

Figure 2. Model plot with standardised parameters; Source: Authors’ work, survey, January 2022.

Goodness-of-fit indicators are shown in Table 7. The Chi-square of 43.438, with

24 degrees of freedom and p = 0.009, is statistically significant at a 95% confidence level
and therefore does not indicate an adequate fit of the model. However, in such a case, other
tests should be performed to reject or accept the proposed measurement model, such as
the Chi-square statistic ratio (χ2 /df) and other goodness-of-fit indicators [45]. A (χ2 /df)
of 1.81 is considered very good, as it is below the recommended ratio of 3 or less (Hoe,
χ χ
2008) and even below the more conservative ratio of 2 or less [46]. Further, the value for
RMSEA is 0.034, well below the threshold of 0.08. The same is true for the SRMR of 0.017,
below the conservative threshold of 0.05 (Hair et al., 2014). CFI is 0.997, which exceeds the
recommended higher value of 0.94 for CFI [46]. Other goodness-of-fit measures also have
satisfactory values (TLI = 0.995 > 0.95; GFI = 0.987 > 0.95), indicating that the measurement
model fits the data well [45].

Table 7. The goodness-of-fit indicators for the measurement model evaluation.

ML Estimate Acceptable Value Source

Chi-square (χ2 ) 701 - -
Degrees of freedom (df ) 43.438 - -
χ
p-value 24 - -
Chi-square (χ2 ) 0.009 - -
CFI 0.997 >0.94 [46]
χ TLI 0.995 >0.95 [45]
GFI 0.987 >0.95 [45]
RMSEA 0.034 <0.08 [45]
SRMR 0.017 <0.05 [45]
Source: Authors’ work, based on a survey, January 2022.

Since the fit of the overall model was satisfactory, we proceeded with the estimates of
the factor loadings presented in Table 8. Factor loadings are correlation coefficients between
the manifest variable and the latent factor/construct, thus indicating how well the variable
represents the construct. The higher the loading, the more representative the variable is
of the factor. All unstandardised loading estimates are statistically significant, which is
required to establish convergent validity. In addition, all standardised factor loadings are
higher than the ideal or preferred value of 0.7 or greater [45]. Therefore, the results confirm
that the variables represent their latent factor well.
J. Theor. Appl. Electron. Commer. Res. 2023, 18 1474

Table 8. Unstandardised factor loadings, standardised factor loadings, average variance extracted,
and composite reliability.

Factor Indicator Symbol Est. Std. Est. Std. Error z-Value R-Squared AVE CR
C11_1 λ11 2.233 0.949 0.066 33.600 * 0.900 0.891 0.976
Self-efficacy C11_2 λ12 2.226 0.944 0.067 33.315 * 0.892
C11_3 λ13 2.201 0.938 0.067 32.920 * 0.880
C11_4 λ21 1.942 0.889 0.066 29.305 * 0.790 0.760 0.972
Social
C11_5 λ22 1.913 0.880 0.066 28.847 * 0.775
influence
C11_6 λ23 1.803 0.846 0.067 27.091 * 0.715
C11_7 λ31 2.103 0.921 0.066 31.626 * 0.849 0.841 0.974
Social
C11_8 λ32 2.121 0.917 0.068 31.363 * 0.840
support
C11_9 λ33 2.131 0.913 0.068 31.162 * 0.834
Note: * statistically significant at 1%. Source: Authors’ work, based on a survey, January 2022.

Additionally, we calculated the AVE and the CR to investigate convergent validity

further. The coefficient of construct reliability (CR) is analogous to the coefficient alpha and
is commonly used to estimate reliability in the CFA–SEM method (Hair et al., 2014). CR is
above the threshold of 0.7 for all latent variables, indicating adequate internal consistency
of the item scales. Finally, we estimated the average variance extracted (AVE) because
it is a more stringent criterion for internal consistency. AVE estimates for each construct
exceed the recommended minimum value of 0.5 [45]. Since all estimates of factor loadings,
construct reliability (CR), and average variance extracted (AVE) exceed the recommended
minimums of 0.7, 0.7, and 0.5, respectively, convergent validity was confirmed.
We apply a widely used approach to assess discriminant validity recommended by [46],
also known as the Fornell–Larcker criterion. According to this criterion (Table 9), the AVE
estimate for a given factor should be higher than the squared inter-construct correlations
related to that factor. Since all diagonal elements (AVE) are higher than the corresponding
(underlying) off-diagonal elements (squared inter-construct correlations), we can conclude
that the discriminant validity of our measurement model has also been demonstrated.

Table 9. Factor squared inter-construct correlations and average variance extracted estimate.

Self-Efficacy Social Influence Social Support

Self-efficacy 0.891
Social influence 0.465 0.760
Social support 0.380 0.520 0.841
Note: AVE estimate for a given factor (in bold). Source: Authors’ work, based on a survey, January 2022.

4.5. Structural Equation Modelling

While the measurement model assessment indicates how well the manifest (measured)
variables represent the latent variables (constructs), the structural model assessment in SEM
is used to examine the relationships between the constructs, i.e., the structure of the model.
Because the measurement model for our data showed strong convergent and discriminant
validity, we proceeded to evaluate the proposed structural model of the research.
First, we assessed the fit of the model using goodness-of-fit indicators. Then, we moved
to path analysis to test the proposed hypothesis. This step applied the SEM procedure to
the entire sample, while parameter estimates were obtained using the maximum likelihood
method (ML) of covariance matrices. A one-factor loading estimate was set to 1 for each
construct to determine the latent factor scale.
Table 10 provides estimates of the commonly used indices for evaluating the struc-
tural model.
J. Theor. Appl. Electron. Commer. Res. 2023, 18 1475

Table 10. Goodness-of-fit indicators for the structural model evaluation.

Indicator Estimated Value Recommended Value Source

N 701 - -
Chi-square (χ2 ) 58.820 - -
Degrees of freedom (df ) 36 - -
p-value 0.010 - -
Chi-square statistic ratio (χ2 /df ) 1.634 <2 [45]
CFI 0.997 >0.94 [46]
TLI 0.995 >0.95 [45]
NNFI 0.995 >0.9 [50]
NFI 0.992 >0.9 [50]
RMSEA 0.030 <0.07 [45]
SRMR 0.014 <0.08 [45]
Source: Authors’ work, based on a survey, January 2022.

As can be seen, all indicators met the requirements for the acceptable values for a good
model fit. The ratio of the Chi-square statistic (χ2 /df) is 1.634, well below the benchmark
ratio of 3 to 1 [50] and even below the more conservative maximum ratio of 2 [45]. Other
absolute fit indices, RMSEA and SRMR, are also well below the maximum thresholds of
0.07 and 0.08, respectively. The incremental fit indices, namely CFI, TLI, NFI, and NNFI,
are also in the acceptable range of 0.95 or more [45]. Moreover, all incremental indices have
values close to 1, indicating the model’s strong structural validity.
Since the goodness-of-fit indicators pointed out a sound specification of our model,
we were able to proceed with the path analysis to test whether the hypothesised theoretical
relationships among the constructs applied to our research context. The path analysis is
shown in Table 11.

Table 11. Standardised path coefficients and squared multiple correlation coefficient.

Estimate
Predictor Outcome z-Value p R-Squared Hypothesis
(Std. Error)
0.522
Social influence Self-efficacy 10.230 <0.001 *** 0.500 H1
(0.051)
0.277
Social support Self-efficacy 5.823 <0.001 *** H2
(0.047)
0.694
Social support Social influence 20.830 <0.001 *** 0.517 H3
(0.033)
0.912
Self-efficacy Intensity of Internet usage 26.046 <0.001 *** 0.720 H4
(0.035)
−0.052
Social influence Intensity of Internet usage −1.209 0.227 H5
(0.043)
0.025
Social support Intensity of Internet usage 0.656 0.512 H6
(0.037)
−0.279
Self-efficacy Obstacles to Internet usage −4.723 <0.001 *** 0.039 H7
(0.059)
0.177
Social influence Obstacles to Internet usage 2.312 0.021 ** H8
(0.077)
−0.008
Social support Obstacles to Internet usage −0.120 0.905 H9
(0.066)
Note: *** statistically significant at 1%: ** 5%. Source: Authors’ work, based on a survey, January 2022.

As expected, significant and positive relationships were found between social influence
and self-efficacy (H1); social support and self-efficacy (H2); and social influence and social
support (H3).
The path coefficient is positive and significant at 1% for the relationship between
the intensity of Internet usage and self-efficacy (H4), with a path coefficient of 0.912 and
p < 0.01. In contrast, the significance of the path coefficients between the intensity of
J. Theor. Appl. Electron. Commer. Res. 2023, 18 1476

Internet usage on the one side and social influence and social support on the other side was
not demonstrated in the model of our research (H5 and H6, respectively).
The path coefficient is negative and significant at 1% for the relationship between
obstacles in Internet usage and self-efficacy, with a path coefficient of −0.279, with p < 0.01,
indicating a strong negative correlation between these two variables (H7). The path between
obstacles in Internet usage and social influence has an estimated value of 0.177, with p < 0.05
indicating a significant positive relationship between the two constructs at 5%. However,
H8 was not confirmed because the relationship direction was different than expected
(positive instead of negative). In addition, the relationship was not significant between
obstacles in Internet usage and social support (H9).

5. Discussion
Table 12 summarises research results in the context of hypothesis testing. The path
analysis confirmed the hypothesised theoretical relationships in the structural model for
six hypotheses, namely H1, H2, H3, H4, H7, and H9, while hypotheses H5, H6, and H7
were not supported.

Table 12. Summary of research hypothesis conclusions.

Hypothesis Predictor Outcome Relationship Conclusion

H1 Social influence Self-efficacy Positive at 1% H1—Confirmed
H2 Social support Self-efficacy Positive at 1% H2—Confirmed
H3 Social influence Social support Positive at 1% H3—Confirmed
H4 Self-efficacy Intensity of Internet usage Positive at 1% H4—Confirmed
H5 Social influence Intensity of Internet usage Not significant H5—Not confirmed
H6 Social support Intensity of Internet usage Not significant H6—Not confirmed
H7 Self-efficacy Obstacles to Internet usage Negative at 1% H7—Confirmed
H8 Social influence Obstacles to Internet usage Positive at 5% H8—Not confirmed
H9 Social support Obstacles to Internet usage Not significant H9—Not confirmed
Source: Authors’ work, based on a survey, January 2022.

Concerning the determinants of the model of Internet usage among older people, the
results imply that the opinions of family, relatives, and friends about using the Internet
(measured by social influence) are associated with the personal attitude of each elderly
user about the pleasure and simplicity of using the Internet (measured by self-efficacy)
as well as the support of their family, friends, and acquaintances in using the Internet,
which confirms hypotheses H1 and H3. Generally, from the definition of social influence, it
can be explained how social influence is related to forming individual thoughts, opinions,
feelings, and behaviour about using s-commerce (e-commerce using social media) due to
the individual’s interactions with his environment [20,52]. According to [53], the perception
and expectations of family, friends, relatives, media, and community in general in Internet
usage strongly affect forming individuals’ attitudes, interests, and opinions toward Internet
usefulness and quality; this is in line with the confirmed hypothesis H1. Also, this finding
aligns with [22], whose results show how social influence impacts computer self-efficacy
among seniors. According to different authors [20,54,55], social influence encourages
people to acquire new knowledge, solve problems to meet other people’s expectations, be
socially involved, and improve self-perception, which will consequently impact the need
for a higher level of social support and more intensive social interactions. Such conclusions
can support the examined and confirmed hypothesis H3.
Similarly, hypothesis H2 is confirmed by showing how social support provided by
the social environment in which the seniors interact also positively influences the seniors’
self-efficacy in using the Internet. The given result from the hypothesis H2 is in line with the
previous study [31], in which they emphasise how to continue realising one’s own positive
opinion of information and communication technology: the encouragement and support of
the environment are extremely important. Moreover, the finding obtained from hypothesis
J. Theor. Appl. Electron. Commer. Res. 2023, 18 1477

H2 can be confirmed by [22], who also emphasises the influence of social support on
computer self-efficacy among elderly users of information and communication technology.
Concerning the determinants of intensity of Internet usage, the results suggest that
the consumption of Internet content and the use of electronic services for older people
(measured by the intensity of Internet usage) can be explained by their abilities and per-
ceived comfort in navigating the Internet (measured by self-efficacy), thus confirming H4.
At the same time, it is not influenced by incentives or support coming from their close
peers (measured by social influence and social support), thus not confirming H5 and H6.
The obtained result from the H4 hypothesis is also confirmed in the existing literature. For
instance, Ref. [56] also proved the importance of self-efficacy on the intensity of Internet
usage. Likewise, Ref. [57] also confirmed how higher self-efficacy in general Internet and
communication Internet usage positively influences informational Internet activities among
non-expert users.
Moreover, in their study, Ref. [23] stresses how self-efficacy is directly linked to higher
intention in e-Health application usage. The result from hypothesis H5 is in line with other
studies confirming that accepting and using new technology has not been influenced by the
individual perception and knowledge of how others from their environment use the same
technology [58]. According to [59], social support can be classified into three categories:
(i) providing emotional support, (ii) providing support in the context of information and
advice, and (iii) providing support in the context of help with daily tasks or finances.
Given that the Internet has become a part of everyday life, it can be considered that
providing support around the Internet is instrumental; it has been shown in research
to cause a higher level of depression among the elderly population [59–61]. Moreover,
receiving received, and not just perceived, social support in some studies resulted in higher
levels of emotional stress among respondents due to potential negative social interactions,
conflicts, and feelings of dependence on others that can accompany support from other
people [59–61]. Therefore, such results of previously conducted studies can support the
findings for hypothesis H6.
Concerning the determinants of the obstacles in Internet usage, the results indicate that
the obstacles for older people, similar to the intensity of Internet usage, can be explained by
the level of personal ability and comfort of using the Internet (measured by self-efficacy).
The obtained result from hypothesis H7 is also proved by [62], in which the authors argue
how individuals with a higher level of self-efficacy perceive difficulties as less challenging
than individuals with a lower level of self-efficacy; this consequently enables them to
overcome obstacles more easily. The social influence of the immediate environment has a
counterintuitive positive effect on obstacles in Internet usage, indicating that higher social
support increases anxiety in Internet usage, resulting in a higher perception of obstacles
in Internet usage. The association of social support with obstacles in Internet usage was
not confirmed. These findings about hypothesis H8 suggest that, as [63] points out, social
influence has a strong relationship with causing anxiety when using Internet services, for
instance, social media. The findings based on the result of hypothesis H9 can be related to
the result of hypothesis H6 and, therefore, explained by the fact that potential social support
can cause a feeling of dependence on others by older people who would potentially like to
independently solve the challenges they encounter by using the Internet. Moreover, such
an explanation can also be supported by the confirmation of hypothesis 4, which showed
that for seniors, their personal opinions about the comfort and ease of use of the Internet are
significant for deciding to use it more intensively. Also, such findings can suggest that the
sampled seniors are computer-literate enough to use the Internet independently. Figure 3
presents the summary of the testing of research model.
 ffi

J. Theor. Appl. Electron. Commer. Res. 2023, 18 1478

Figure 3. Summary of the testing research model; Source: Authors’ work, survey, January 2022.

6. Conclusions
6.1. Summary of the Research
In recent decades, developed nations have observed two significant trends: a growing
number of senior citizens and a rapid integration of ICT. This study delves into how self-
efficacy, social support, and influence affect Internet usage among Croatian senior citizens.
ffi
The research aimed to show the extent to whichff social influence, self-efficacy, and social
support are associated with the intensity of Internet use and reducing barriersffito Internet
use among older users. Through survey research and structural equation modelling, it was
found that self-efficacy positively impacts the intensity of Internet usage while also reducing
related obstacles. Social influence directly lessens these obstacles and indirectly affects
Internet usage intensity via self-efficacy. Meanwhile, social support indirectly influences
the intensity of Internet usage. These findings underscore the importance of educational
programs that boost seniors’ perceived self-efficacy in Internet usage.

6.2. Theoretical Implications

The work presented in this study has significant theoretical implications as it con-
tributes to the current body of literature investigating characteristics that facilitate the
use of information and communication technology (ICT) among older individuals. It
also enhances our understanding of the impact of social influence, social support, and
self-efficacy within this context. The theoretical contributions of this study offer a thor-
ough comprehension of the various elements that impact the utilisation of the Internet by
older individuals. These contributions highlight the significance of social influence, social
support, and self-efficacy in shaping Internet usage patterns among this demographic.
The study’s findings establish a correlation between the viewpoints of family mem-
bers, relatives, and friends towards Internet usage and the personal attitudes of senior
individuals towards the enjoyment and ease of using the Internet. This finding aligns
with other scholarly investigations that emphasise the role of social influence in shaping
individuals’ attitudes and behaviours regarding s-commerce and Internet use.
The study additionally validates that social support derived from the surrounding
context has a favourable impact on the self-efficacy of older individuals in utilising the
Internet. This finding is consistent with other research that underscores the significance of
contextual factors in fostering favourable attitudes towards information and communica-
tion technologies.
The prevalence of Internet usage among older adults can be attributed to their self-
efficacy, while social influence and social support do not appear to have a significant impact.
This discovery aligns with other studies that emphasise the significance of self-efficacy in
determining the level of Internet usage and the acceptance of e-Health applications.
The challenges encountered by elderly individuals in utilising the Internet can be
ascribed to their level of self-efficacy. Individuals with elevated levels of self-efficacy tend
J. Theor. Appl. Electron. Commer. Res. 2023, 18 1479

to regard difficulties as less formidable and are more adept at surmounting them. It is
noteworthy that evidence suggests that increased levels of social support can potentially
contribute to heightened anxiety in Internet usage, resulting in an amplified impression
of barriers. This indicates that using Internet services may lead to anxiety due to social
influence. At the same time, the senior population may experience feelings of dependence
due to potential social assistance.
The results indicate that the seniors who participated in the research have sufficient
computer literacy that enables them to utilise the Internet autonomously. The individual’s
subjective viewpoints regarding the convenience and user-friendliness of Internet usage
substantially influence their inclination to engage with it more extensively.
In conclusion, this study’s findings indicate a notable positive correlation between
social impact, social support, and older adults’ self-efficacy. Additionally, the findings indi-
cated a strong correlation between the social influence experienced by older Internet users
and the level of support received from their family, friends, and relatives. In contrast, of the
three criteria examined concerning the intensity of Internet usage among older individuals,
only self-efficacy has demonstrated a statistically significant positive correlation. Moreover,
based on the findings of the conducted research, it is evident that self-efficacy is the sole
factor that exhibits a substantial negative impact on the difficulties encountered during
Internet usage. Furthermore, it was shown that social influence exhibited a significant
correlation with the obstacles encountered by older individuals in utilising the Internet.

6.3. Practical Implications

The practical consequences of this study are the potential adaptation of Internet service
providers’ functionality to cater to the needs of older customers, hence facilitating their
inclusion in the dynamic realm of digitalisation.
The importance of self-efficacy in influencing the extent and challenges associated with
Internet usage among older adults necessitates the development of customised educational
initiatives. The primary objective of these programmes should be to improve seniors’
self-assurance and proficiency in utilising the Internet, fostering a sense of competence and
reducing apprehension about digital obstacles.
Given that the attitudes of older individuals regarding Internet usage are influenced by
the perspectives of their family members, relatives, and friends, it is imperative to engage
these social groups in any digital literacy initiative. Organising workshops or instructional
sessions can facilitate the participation of senior individuals and their close companions,
fostering a conducive environment for elderly individuals to acquire knowledge and
adjust accordingly.
The findings of this study may serve as a significant foundation for mental health in-
terventions, as they suggest that while social support can have positive effects, an excessive
amount of assistance or reliance on others can result in increased anxiety and a heightened
perception of barriers. Considering the potential anxiety often linked to Internet services,
particularly social media, it is advisable for practical programmes to integrate modules that
specifically target these concerns. The curriculum may encompass instructional modules
about various aspects of digital safety, including but not limited to online security measures,
privacy management mechanisms, and fostering constructive online engagements.
Future projects and initiatives may be directed towards advancing user-friendly digital
platforms. The prioritisation of user-friendly platforms by tech companies and service
providers is crucial in facilitating the uptake of the Internet among seniors, as personal
perceptions regarding comfort and ease of use significantly influence their decision-making
process. Simplified interfaces, increased text size, intuitive designs, and comprehensive
training can potentially enhance digital accessibility for older individuals. These platforms
have the potential to incorporate an automatic feedback mechanism, which would assist
educators and programme organisers in comprehending the distinct requirements, obsta-
cles, and preferences of senior participants. This measure can guarantee the continued
relevance and efficacy of digital literacy programmes.
J. Theor. Appl. Electron. Commer. Res. 2023, 18 1480

One potential strategy for disseminating information to the general public regarding
the potential beneficial impact they can have on the digital journeys of elderly individuals
is implementing awareness campaigns. By comprehending the significant impact of social
influence, individuals can adopt a proactive approach to promoting and directing the
elderly within their social networks.
Ultimately, establishing support groups and communities for seniors to share their
experiences, concerns, and solutions pertaining to Internet usage can prove advantageous.
These platforms have the potential to provide chances for peer-to-peer learning, thereby mit-
igating feelings of loneliness and fostering a sense of community among older individuals
who use the Internet.
By integrating these practical implications into strategies and programmes, there is
a notable potential to enhance the digital experience for older adults. This enhancement
would result in increased Internet usage and instil a sense of confidence and ease while
navigating online platforms.

6.4. Limitations and Future Research Directions

First, the primary limitation of this study is its geographic scope. The sample was
drawn exclusively from one country, which may not represent senior citizens’ experiences
and attitudes in other cultural or socio–economic contexts. Second, the sample predom-
inantly consisted of urban and highly educated elderly individuals. This demographic
bias may have influenced the results, as urban and educated seniors might have different
levels of exposure, comfort, and proficiency with the Internet compared to their rural or
less-educated counterparts. The third limitation stems from the cross-sectional design.
The study’s design captures a snapshot in time without tracking changes or evolutions
in attitudes and behaviours, which limits the ability to infer causality or observe how
perceptions and usage patterns might evolve.
Directions for future research include the following: First, to enhance the generalisabil-
ity of the findings, future research should consider a multi-country approach, considering
significant differences between European countries according to the level of digital trans-
formation [64]. Comparing results across countries can provide insights into cultural,
economic, or infrastructural factors influencing seniors’ Internet usage. Second, given
the urban and educated bias in the current study, future research should deliberately
include participants from rural areas and diverse educational backgrounds, offering a
more comprehensive understanding of the challenges and motivations across different
demographic groups. Third, implementing a longitudinal design can help track changes
in seniors’ Internet usage and attitudes over time, which would be particularly valuable
in understanding the long-term effects of interventions or the impact of rapidly evolving
digital landscapes on the elderly. Fourth, as recommended, in-depth interviews and case
studies can provide richer, more nuanced insights into seniors’ experiences and challenges
with the Internet. Qualitative research can uncover deeper motivations, fears, and barriers
that might not be evident in survey data. Finally, future research directions could cover
the diverse digital platforms and support systems. Future studies could explore seniors’
interactions with various digital platforms, not just the Internet, including mobile apps,
smart home devices, and other emerging technologies, to understand their adaptability
and challenges in a broader digital context. Given the importance of social influence and
support in the current study, future research could delve deeper into the role of different
support systems, like community centres, tech workshops, or peer-led groups, in enhancing
digital literacy among seniors.

Author Contributions: Conceptualization, M.P.B. and V.B.V.; methodology, M.P.B.; software, M.P.B.;
validation, M.P.B., V.B.V. and L.M.G.; formal analysis, L.I.; investigation, M.P.B. and V.B.V.; resources,
V.B.V.; data curation, V.B.V.; writing—original draft preparation, L.I. and A.-M.S.; writing—review
and editing, V.B.V. and M.P.B.; visualisation, M.P.B.; supervision, V.B.V. All authors have read and
agreed to the published version of the manuscript.
J. Theor. Appl. Electron. Commer. Res. 2023, 18 1481

Funding: This research has been funded within the project “Senior 2030—Thematic Network for
Active Ageing Policy in Croatia”, funded by the European Social Fund.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Data are available from the authors upon request.
Acknowledgments: This research has been conducted within the project “Senior 2030—Thematic
Network for Active Ageing Policy in Croatia”, funded by the European Social Fund.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Barišić, A.F.; Pejić Bach, M.; Miloloža, I. Human Resources Information Systems: Transactional and Strategic Paradigm.
ENTRENOVA—ENTerprise REsearch InNOVAtion 2018, 4, 224–230.
2. Barišić, A.F.; Rybacka Barišić, J.; Miloloža, I. Digital Transformation: Challenges for Human Resources Management.
ENTRENOVA—ENTerprise REsearch InNOVAtion 2022, 7, 357–366. [CrossRef]
3. Park, J.R.; Feng, Y. Trajectory tracking of changes digital divide prediction factors in the elderly through machine learning. PLoS
ONE 2023, 18, e0281291. [CrossRef]
4. Jiao, R. Is There a Barrier between Seniors and Smartphone Use in The Internet Age? A Study of Digital Disconnection among
Older Adults. SHS Web Conf. 2023, 155, 03010. [CrossRef]
5. Sum, S.; Mathews, R.M.; Hughes, I.; Campbell, A. Internet use and loneliness in older adults. CyberPsychol. Behav. 2008, 11,
208–211. [CrossRef]
6. van Houwelingen, C.T.; Ettema, R.G.; Antonietti, M.G.; Kort, H.S. Understanding older people’s readiness for receiving telehealth:
Mixed-method study. J. Med. Internet Res. 2008, 20, e8407. [CrossRef]
7. Xie, B. Older Chinese the Internet and well-being. Care Manag. J. 2007, 8, 33–38. [CrossRef]
8. Maaß, W. The elderly and the Internet: How senior citizens deal with online privacy. In Privacy Online: Perspectives on Privacy and
Self-Disclosure in the Social Web; Springer: Berlin, Germany, 2011; pp. 235–249.
9. Croatian Bureau of Statistics. Population Estimate of Republic of Croatia. 2022. Available online: https://podaci.dzs.hr/en/
statistics/population/ (accessed on 1 July 2023).
10. Drobne, S.; Bogataj, M. Migration flows through the lens of human resource ageing. Bus. Syst. Res. Int. J. Soc. Adv. Innov. Res.
Econ. 2022, 13, 47–62. [CrossRef]
11. Nejašmić, I.; Toskić, A. Starenje stanovništva u Hrvatskoj–sadašnje stanje i perspektive. Hrvat. Geogr. Glas. 2013, 75, 89–110.
[CrossRef]
12. Garbin Praničević, D.; Peterlin, J.; Bućan, M.J. Do older people benefit from digital services? DIEM Dubrov. Int. Econ. Meet. 2017,
3, 145–160.
13. Bosilj Vukšić, V.; Milanović Glavan, L.; Ivančić, L. Digital Technology in the Health Care and Social Care of Older Adults. In
Proceedings of the 4th International Conference on the Economics of Decoupling (ICED), Zagreb, Croatia, 30 November–1
December 2022; pp. 189–202.
14. Rashotte, L. Social Influence. In The Blackwell Encyclopedia of Sociology; John Wiley & Sons, Ltd.: Chichester, UK, 2007.
15. Friedkin, N.E.; Johnsen, E.C. Social influence and opinions. J. Math. Sociol. 1990, 15, 193–206. [CrossRef]
16. Vries, H.D.; Backbier, E.; Kok, G.; Dijkstra, M. The Impact of Social Influences in the Context of Attitude Self-Efficacy Intention
and Previous Behavior as Predictors of Smoking Onset 1. J. Appl. Soc. Psychol. 1995, 25, 237–257. [CrossRef]
17. Bandura, A. Social Foundations of Thought and Action: A Social Cognitive Theory; Prentice-Hall: New York, NY, USA, 1986.
18. Maddux, J.E.; Gosselin, J.T. Self-efficacy. In Handbook of Self and Identity; Leary, M.R., Tangney, J.P., Eds.; The Guilford Press: Ney
Work City, NY, USA, 2012; pp. 198–224.
19. Bandura, A.; Wessels, S. Self-Efficacy; W.H. Freeman & Company: New York, NY, USA, 1997.
20. Hu, X.; Chen, X.; Davison, R.M. Social support, source credibility, social influence, and impulsive purchase behavior in social
commerce. Int. J. Electron. Commer. 2019, 23, 297–327. [CrossRef]
21. Taylor, S.E. Social support: A review. In The Oxford Handbook of Health Psychology; Friedman, H.S., Ed.; Oxford University Press:
Oxford, UK, 2011; pp. 189–214.
22. Guan, C.; Wei, J.; Chan, C.M.; Chia, A.B. Senior citizens’ self-efficacy for ICT use: The influence of gender, social influence and
social support. In Proceedings of the International Conference on Electronic Business (ICEB), Dubai, United Arab Emirates,
4–8 December 2017; pp. 234–240.
23. de Veer, A.J.; Peeters, J.M.; Brabers, A.E.; Schellevis, F.G.; Rademakers, J.J.; Francke, A.L. Determinants of the intention to use
e-Health by community dwelling older people. BMC Health Serv. Res. 2015, 15, 103. [CrossRef]
24. Hsieh, J.J.P.-A.; Rai, A.; Keil, M. Addressing digital inequality for the socioeconomically disadvantaged through government
initiatives: Forms of capital that affect ICT utilisation. Inf. Syst. Res. 2011, 22, 233–253. [CrossRef]
J. Theor. Appl. Electron. Commer. Res. 2023, 18 1482

25. Eastin, M.S.; LaRose, R. Internet self-efficacy and the psychology of the digital divide. J. Comput.-Mediat. Commun. 2000, 6,
JCMC611. [CrossRef]
26. Ehrari, H.; Ulrich, F.; Andersen, H.B. Concerns and trade-offs in information technology acceptance: The balance between the
requirement for privacy and the desire for safety. Commun. Assoc. Inf. Syst. 2020, 47, 46. [CrossRef]
27. Hamari, J.; Koivisto, J. Working out for likes: An empirical study on social influence in exercise gamification. Comput. Hum. Behav.
2015, 50, 333–347. [CrossRef]
28. Venkatesh, V.; Morris, M.G.; Davis, G.B.; Davis, F.D. User acceptance of information technology: Toward a unified view. MIS Q.
2003, 27, 425–478. [CrossRef]
29. Heo, J.; Chun, S.; Lee, S.; Lee, K.H.; Kim, J. Internet use and well-being in older adults. Cyberpsychol. Behav. Soc. Netw. 2015, 18,
268–272. [CrossRef]
30. Wu, H.; Rudkin, L. Social contact socio-economic status and the health status of older Malaysians. Gerontologist 2000, 40, 228–234.
[CrossRef] [PubMed]
31. Tsai, H.S.; Shillair, R.; Cotten, S.R. Social support and “playing around”: An examination of how older adults acquire digital
literacy with tablet computers. J. Appl. Gerontol. 2017, 36, 29–55. [CrossRef] [PubMed]
32. Barnard, Y.; Bradley, M.D.; Hodgson, F.; Lloyd, A.D. Learning to use new technologies by older adults: Perceived difficulties,
experimentation behaviour and usability. Comput. Hum. Behav. 2013, 29, 1715–1724. [CrossRef]
33. Bandura, A. Self-efficacy mechanisms in human agency. Am. Psychol. 1982, 37, 122–147. [CrossRef]
34. Akhter, H.S. Privacy concern and online transactions: The impact of internet self-efficacy and internet involvement. J. Consum.
Mark. 2014, 31, 118–125. [CrossRef]
35. Matsuo, M.; Minami, C.; Matsuyama, T. Social influence on innovation resistance in internet banking services. J. Retail. Consum.
Serv. 2018, 45, 42–51. [CrossRef]
36. Martinez-Pecino, R.; Matos, A.D.; Silva, P. Portuguese older people and the Internet: Interaction, uses, motivations, and obstacles.
Communications 2013, 38, 331–346. [CrossRef]
37. Rabiee, A.; Nazarian, Z.; Gharibshaeyan, R. An explanation for internet use obstacles concerning e-learning in Iran. Int. Rev. Res.
Open Distrib. Learn. 2013, 14, 361–376. [CrossRef]
38. Van Biljon, J.; Renaud, K. A qualitative study of the applicability of technology acceptance models to senior mobile phone users.
In ER Workshops 2008 LNCS 5232; Song, I.-Y., Piattini, M., Phoebe Chen, Y.-P., Hartman, S., Grandi, F., Trujillo, J., Opdahl, A.L.,
Ferri, F., Grifoni, P., Caschera, M.C., et al., Eds.; Springer: Berlin, Germany, 2008; pp. 228–237.
39. Francis, J.; Kadylak, T.; Makki, T.W.; Rikard, R.V.; Cotton, S.R. Catalyst to connection: When technical difficulties lead to social
support for older adults. Am. Behav. Sci. 2018, 62, 1167–1185. [CrossRef]
40. Cohen, S.; Mermelstein, R.; Kamarck, T.; Hoberman, H.M. Measuring the functional components of social support. In Social
Support: Theory, Research and Applications; Springer: Dordrecht, The Netherlands, 1985; pp. 73–94.
41. Szkody, E.; Stearns, M.; Stanhope, L.; McKinney, C. Stress-buffering role of social support during COVID-19. Fam. Process 2021,
60, 1002–1015. [CrossRef]
42. OECD Data. Available online: https://data.oecd.org/pop/elderly-population.htm (accessed on 1 July 2023).
43. United Nations, Department of Economic and Social Affairs, Population Division World Population Ageing 2015
(ST/ESA/SER.A/390). 2015. Available online: https://www.un.org/en/development/desa/population/publications/
pdf/ageing/WPA2015_Report.pdf (accessed on 1 July 2023).
44. Taylor, S.; Todd, P.A. Understanding Information Technology Usage: A Test of Competing Models. Inf. Syst. Res. 1995, 6, 144–176.
[CrossRef]
45. Hair, J.F.; Black, W.C.; Babin, B.J.; Anderson, R.E. Multivariate Data Analysis, 7th ed.; Pearson: London, UK, 2014.
46. Hair, J.F., Jr.; Babin, B.J.; Krey, N. Covariance-Based Structural Equation Modeling in the Journal of Advertising: Review and
Recommendations. J. Advert. 2017, 46, 163–177. [CrossRef]
47. Pejić Bach, M.; Starešinić, B.; Omazić, M.A.; Aleksić, A.; Seljan, S. M-Banking quality and bank reputation. Sustainability 2020,
12, 4315. [CrossRef]
48. Hu, Z.; Li, J. The Integration of EFA and CFA: One Method of Evaluating the Construct Validity. Glob. J. Hum. Soc. Sci. 2015,
15, 15–19.
49. Jackson, D.L.; Gillaspy, J.A.; Purc-Stephenson, R. Reporting practices in confirmatory factor analysis: An overview and some
recommendations. Psychol. Methods 2009, 14, 6–23. [CrossRef]
50. Hoe, S.L. Issues and procedures in adopting structural equation modelling technique. J. Quant. Methods 2008, 3, 76–83.
51. Costello, A.B.; Osborne, J.W. Best practices in exploratory factor analysis: Four recommendations for getting the most from your
analysis. Pract. Assess. Res. Eval. 2005, 10, 7.
52. Amblee, N.; Bui, T.H. The influence of social proof in online shopping: The effect of electronic word of mouth on sales of digital
microproducts. Int. J. Electron. Commer. 2011, 16, 91–114. [CrossRef]
53. Klobas, J.E.; Clyde, L.A. Adults learning to use the Internet: A longitudinal study of attitudes and other factors associated with
intended Internet use. Libr. Inf. Sci. Res. 2000, 22, 5–34. [CrossRef]
54. Bearden, W.O.; Netemeyer, R.G.; Teel, J.E. Measurement of consumer susceptibility to interpersonal influence. J. Consum. Res.
1989, 15, 473–481. [CrossRef]
55. Fang, X.; Hu, P.J.-H. Top persuader prediction for social networks. MIS Q. 2018, 42, 82–83. [CrossRef]
J. Theor. Appl. Electron. Commer. Res. 2023, 18 1483

56. Lam, J.C.-Y.; Lee, M.K.O. Digital inclusiveness—Longitudinal study of Internet adoption by older adults. J. Manag. Inf. Syst. 2006,
22, 177–206. [CrossRef]
57. Jokisch, M.R.; Schmidt, L.I.; Doh, M.; Marquard, M.; Wahl, H.W. The role of internet self-efficacy, innovativeness and technology
avoidance in breadth of internet use: Comparing older technology experts and non-experts. Comput. Hum. Behav. 2020,
111, 106408. [CrossRef]
58. Braun, M.T. Obstacles to social networking website use among older adults. Comput. Hum. Behav. 2013, 29, 673–680. [CrossRef]
59. Kamin, S.T.; Beyer, A.; Lang, F.R. Social support is associated with technology use in old age. Inf. Commun. Technol. 2020, 18,
369–384. [CrossRef]
60. Reinhardt, J.P.; Boerner, K.; Horowitz, A. Good to have but not to use. Differential impact of perceived and received support on
well-being. J. Soc. Pers. Relat. 2006, 23, 117–129. [CrossRef]
61. Kaul, M.; Lakey, B. Where is the support in perceived support? The role of generic relationship satisfaction and enacted support
in perceived support’s relation to low distress. J. Soc. Clin. Psychol. 2003, 22, 59–78. [CrossRef]
62. Lucas, T.; Alexander, S.; Firestone, I.J.; Baltes, B.B. Self-efficacy and independence from social influence: Discovery of an
efficacy–difficulty effect. Soc. Influ. 2006, 1, 58–80. [CrossRef]
63. Abu-Taieh, E.M.; AlHadid, I.; Masa’deh, R.E.; Alkhawaldeh, R.S.; Khwaldeh, S.; Alrowwad, A.A. Factors Affecting the Use of
Social Networks and Its Effect on Anxiety and Depression among Parents and Their Children: Predictors Using ML, SEM and
Extended TAM. Int. J. Environ. Res. Public Health 2022, 19, 13764. [CrossRef]
64. Hunady, J.; Pisár, P.; Vugec, D.S.; Pejić Bach, M. Digital Transformation in European Union: North is leading, and South is lagging
behind. Int. J. Inf. Syst. Proj. Manag. 2022, 10, 58–81. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.