Anda di halaman 1dari 20

Telematics and Informatics 33 (2016) 452–471

Contents lists available at ScienceDirect

Telematics and Informatics


journal homepage: www.elsevier.com/locate/tele

Social networking time use scale (SONTUS): A new instrument


for measuring the time spent on the social networking sites
Yunusa Olufadi
Statistics and Mathematical Sciences, Kwara State University, PMB 1530, Ilorin, Nigeria

a r t i c l e i n f o a b s t r a c t

Article history: One of the key issues plaguing the existing studies on the use of the social networking sites
Received 12 June 2015 (SNSs) is the lack of a uniform index for measuring the time spent on the sites. The present
Received in revised form 31 July 2015 investigation tries to fill this gap by conceptualizing, developing, and validating a new con-
Accepted 3 November 2015
struct, which we referred to as social networking time use scale (SONTUS). To achieve this,
Available online 4 November 2015
two separate studies were conducted. The data for the first study was collected from 2049
individuals through quota sampling approach. Meanwhile, in this first study, we used
Keywords:
exploratory factor analysis to identify the dimension of the SONTUS construct. Findings
Social networking sites
Scale development
from this study reveal that SONTUS has five factors with good measurement properties.
Confirmatory factor analysis The main aim of the second study (where we utilized data from 1808 people) is to carry
Time of use out a confirmatory factor analysis (CFA) and tests three hypothesized models. In addition,
Validity it aims to investigate the construct validity of SONTUS; and to achieve this, we used 10 per-
sonality and well-being measures, and two theoretically related constructs to SONTUS. The
CFA results showed that SONTUS has five factor solution consisting of 29 items and that the
model with 5 first-order factors with 1 second-order factor is the most suitable model for
the study population. Additionally, the second study provides preliminary evidence for the
convergent, predictive, and incremental validity of SONTUS. Overall, the findings from our
exploratory (study 1) and confirmatory (study 2) studies shows that SONTUS can be used
as a standardized instrument for measuring time spent on sites.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction

For more than a decade, the social networking sites (SNSs) have witnessed a sporadic increase in number and popularity.
In fact, as the year passes-by, so is the popularity and number of SNS increases. This is evident in the report of Duggan et al.
(2015) and studies conducted by (Ellison et al., 2007; Kuss and Griffiths, 2011; Ryan and Xenos, 2011; Panek et al., 2013;
Olufadi, 2015). The SNSs has occupied a central role in the way people communicate and connect with each other; and is
been used by people for several reasons (e.g., communication, entertainment, learning, social, emotional etc.). Meanwhile,
many authors have described the use of SNSs as beneficial and harmful. However, most of the authors that describes its
use as harmful relies on the excessive use (i.e., time committed to the use of the SNSs), which might potentially affect
the individual’s work (e.g., performance at work) or health (e.g., addiction); see for example, studies by Shaffer et al.
(2004, Griffiths, 2005, Echeburúa and de Corral (2010. By this way, several authors (e.g., Ross et al., 2009; Ellison et al.,
2011; Junco, 2012a,b) have tried to estimate the amount of time people spend on the SNSs using various methods.

E-mail address: yolufadi@gmail.com

http://dx.doi.org/10.1016/j.tele.2015.11.002
0736-5853/Ó 2015 Elsevier Ltd. All rights reserved.
Y. Olufadi / Telematics and Informatics 33 (2016) 452–471 453

As for the time people spent on SNSs, evidence from the previous studies reveal varying results. For instance, 79% of the
participants in a study conducted by Ross et al. (2009) reported spending between 10 and 60 min on Facebook daily. They
obtained this result through categorical measure of time. In another study published in 2009, Pempek et al. (2009) asked
students to log their daily time spent on Facebook in a diary for 1 week and found that students reported spending an aver-
age of 27.93 min per day on weekdays and 28.44 min per day on weekends. Additionally, recent studies by Ellison et al.
(2011), Kalpidou et al. (2011), and Junco (2012a,b) show that students spent around 100 min on the SNSs per day. In a more
recent study carried out by Olufadi (2015), the participants in the study spent a substantial amount of time on the sites
(M = 175.4, SD = 117.26) per day.
While the previous studies reported above have made great progress in measuring the time spent on the SNSs, their use of
different measurement methods is problematic. Indeed, this could be a possible explanation for the variations in the esti-
mates of time spent on the SNSs (as reported earlier). We present in what follows a detailed account of some of the com-
monly used indices for measuring the time spent on the SNSs and their limitations.

(a) Categorical measures of time: Ellison et al. (2007), Pierce (2009), Ross et al. (2009), Memdouh and Taswir (2013), Wang
et al. (2014) and Lien and Cao (2014) are some of the authors that have used this measure. A very important limitation
of this approach is that people sometimes give answers they feel will reflect well on them; in addition, because people
like to think of themselves as normal or average, the range of answer choices provided when asking for a quantity or a
frequency can affect the results. For instance, the survey question developed by Ellison et al. (2007) and used by sev-
eral authors as a measure of time use on the sites asked the participants to respond to the following question: ‘‘on a
typical day, about how much time do you spend on Facebook?” The options given are (a) no time at all, (b) less than
10 min, (c) 10–30 min, (d) more than 30 min, up to 1 h. (e) More than 1 h., up to 2 h. (f) More than 2 h, up to 3 h. (g)
More than 3 h. It is highly likely we get fewer people picking 3 h or more, than if the choices offered are (a) 1 h or less,
(b) 2 h, (c) 3 h, (d) 4 h, (e) 5–6 h, (f) 7 h or more. In fact, only 1.5% of the participants in a study conducted by
Valenzuela et al. (2009) reported using the sites for more than 3 h. A possible explanation for this is that the first list
of choices makes 3 h sound extreme, while the second list of choices makes it seem typical. Moreover, the alternatives
listed may influence the opinion of the respondents as demonstrated above. In other words, the use of categorical
choices makes it difficult to include the respondents’ correct choice and may force them into an answer that would
not necessarily be a first choice. Additionally, Junco (2012a) reported that the use of categorical choices might reflect
an a priori bias on the part of the researcher regarding how much time she believe people spent on the SNSs per day.
Lastly, since categorical choices restricts respondents to select from a closed-ended options (which may not reflect and
captures respondents’ perceived time of use); this may lead to more introspection about how much actual time is
spent on the sites (Junco, 2012a).
(b) Time spent (in minutes) per day: By this method, participants are asked ‘‘how many minutes (per day) do you spend on
the sites?” A number of authors (Ellison et al., 2007, 2011; Junco, 2012a,b; Kalpidou et al., 2011; Kujath, 2011; Pempek
et al., 2009; Ross et al., 2009; Lubis et al., 2012) have employed this method. Unfortunately, this approach may be
problematic in the sense that it is difficult to account for the total amount of time spent on the sites. It is also possible
that people are unable to estimate the amount of the time they spent on the sites for the day. To be specific, there is
variation in the daily time spent on the SNSs; sometimes people have a lot of time, other times they hardly access their
SNSs account(s). Thus, there is a need to account for this variation. Moreover, if participants are for example, returning
the completed questionnaire (say, in the afternoon or evening), how do we account for the time on sites for the rest of
the day (e.g., at night).
(c) Use of daily/weekly diary: Many authors have employed self-reported daily and or weekly diary in order to measure
participants’ time use on SNSs (HERI, 2007; Pempek et al., 2009; Rideout et al., 2010; Jacobsen and Forste, 2011;
Junco, 2012a; Olufadi, 2015). One drawback of this approach is how to ensure people are filling the diary at the
end of each day (or any time they are required to fill it) and not that they just fill it on the last day and returned.
Another obvious limitation of this measurement technique is that respondents may under or overestimate their time
use on SNSs; admittedly, this limitation is not peculiar to this approach but any survey that is self-report in nature. We
refer readers to Junco (2013) for the details of the limitations of using self-report as a measure of time use on the SNSs.
(d) Time spent yesterday on the sites: For this measure, the real limitation lies in the possibility of cognitive impairment
that may affect respondents’ ability to recall the time spent on the SNSs the previous day. We could not rule out this
possibility. An example of authors that employ this measure in their studies is (Junco, 2012a,b).

The efforts of these authors are helpful as they provide an insight into the time use by the people on the sites, however, it
is not enough to capture the dimensionality of this complex construct (i.e., time use on the sites). Several other authors have
improved on the limitations highlighted above and have thus presented another view of SNS usage time. Some of these
authors have focused on the use of various functionalities of these sites while others have committed their time to the
use of these sites for a particular set of activities or specific applications area. For instance, to measure Facebook usage,
Joinson (2008) used a list of 28 activities, Pempek et al. (2009) used a list of 25 functionalities, Junco (2012a) used a list
of 14 activities, Mazman and Usluel (2010) used a list of 11 educational activities, Xu et al. (2012) used a list of 5 activities,
and Valenzuela et al. (2009) used a list of 4 activities. These new approach no doubt represents a major contribution to the
operational definitions of SNSs usage, this is because the users task was taken into consideration. In fact, the present study
454 Y. Olufadi / Telematics and Informatics 33 (2016) 452–471

benefitted immensely from their efforts. While their efforts are considered a positive leap in measuring SNSs use time, they
can be improved upon. Indeed, this is the goal of the present study.
Furthermore, evidence from the studies reported above showed that using different measures results in different esti-
mates of the time spent on the sites by the people. One major problem however, with the use of such a plethora of different
measures of the time spent on the SNSs is the difficulty in comparing the results between different studies. There is therefore
a need to have a universal index for measuring the time spent on the SNSs by the people (as suggested in Olufadi (2015)), if
we were to (1) compare the results between different studies and (2) study its relationship with any outcome of interest like
students’ learning and engagement, the big-five personality inventory, sleeping, anxiety, depression etc.
Another possible explanation for the variations in the estimates of the time people spent on the sites is that most of the
previous studies only focused on one of the SNSs. For example, Facebook (Joinson, 2008; Valenzuela et al., 2009; Mazman
and Usluel 2010; Xu et al., 2012; Junco 2012a,b, 2013), WeChat (Lien and Cao, 2014), KakaoTalk (Jo, 2013; Ha et al.,
2015), Twitter (Chen, 2011; Hughes et al., 2012; McKinney et al., 2012; Panek et al., 2013), Renren (Li and Chen, 2014). How-
ever, recent studies have shown that people subscribed to more than one SNSs platform (Helou et al., 2012; Memdouh and
Taswir, 2013; Duggan et al., 2015). For instance, Duggan et al. (2015) reported that more than half of the internet users (52%)
use two or more of the following SNSs (Facebook, Twitter, Instagram, Pinterest, and LinkedIn) compared with 42% who did so
in 2013. To this end, focusing on only one SNS platform is problematic.
Based on this finding, we have tried to correct the limitation of possible underestimation in the time data collected by
using a generic term (SNSs) rather than being specific to one SNSs platform. This is particularly important given that the pop-
ularity of SNSs differ by regions. For example, Qzone and Renren are the most popular among the Chinese (CNNIC, 2011,
2013; ResonanceChina, 2013; Wang et al. 2014), whereas, the SNSs platform that are widely used in Korea include KakaoTalk
and mobile Facebook (KAA, 2012). In one study by Jo (2013), almost 90% of smartphones users in Korea used KakaoTalk. In
addition, findings from the data collected by Wang et al (2014) revealed that Facebook and Twitter were the most popular
SNSs in the US; however, recent report from the PEW researchers reveals Facebook, Pinterest, and Instagram as the most
popular (Duggan et al., 2015).
Additionally, different SNSs are designed for different purposes and might offer different benefit to their users, thus, SNS-
users might be using two or more accounts (as evidenced in the studies by Helou et al. 2012; Memdouh and Taswir, 2013;
Duggan et al., 2015) for different purposes. Therefore, being specific to a particular platform would not give us the opportu-
nity of measuring time spent on the other sites. This approach therefore has the advantage of correcting for the underesti-
mation of the time spent on the sites. Moreover, focusing on one SNSs platform cannot allow us to generalize our findings to
other SNSs without proper validation, using this generic term (SNSs) corrects for this shortcoming.
To this end, we believe that the use of a generic term (SNSs) rather than being specific to a particular SNS platform (e.g.,
Twitter) is much more broad and have the advantage of allowing future researchers interested in studying the actual rela-
tionship between the time used on the sites (e.g., Facebook and Instagram) and any outcome of interest like wellbeing. Over-
all, since no attempts has been made in the past to develop a psychometrically tested scale capable of measuring time spent
by people on the SNSs, this study is designed to fill this gap.

1.1. What this study is not about

Firstly, the current study is not concerned about what the people are doing on the site(s) or the benefits and downsides of
the time spent on the SNSs as they have been addressed in some of the previous studies. See for example, studies conducted
by Valkenburg et al (2006), Valenzuela et al. (2009), Ross et al. (2009), Kirschner and Karpinski (2010), Ryan and Xenos
(2011) and Wang et al. (2014). Additionally, and as established in the literature, there are different types of SNSs with regard
to their primary functions to the users (see Ji et al., 2010). Thus, the relationship between SNSs usage and some outcomes of
interest (e.g., wellbeing) may depend on the type of usage. However, our concern is not about studying this relationship but
how to measure the time spent on the SNSs by the people. To this end, we have developed a psychometrically tested scale
which we termed social networking time use scale (SONTUS) for measuring the time spent by the people on the SNSs. To
wrap up this section, we present the debate between the ‘‘quality” and ‘‘quantity” of the time use on the sites – which is
more important?

2. Conceptual and theoretical framework

2.1. Using UGT to explain the quality and quantity of time use on SNSs

A theoretical study that is relevant to the conceptual discourse between quality and quantity of time use on SNSs by the
people is the uses and gratification theory (UGT) proposed by Katz (1959). Indeed, the UGT could easily explain the use and
the reasons people may decide to use a specific platform in a specific place or situations. The knowledge of such will no doubt
provide information about the quality or quantity of time use on the sites. For instance, the usefulness of the sites and the
ease with which people can connect or communicate with one another plays an important role on the amount of time (i.e.,
quantity) people use on the sites. For example, the benefits (e.g., meeting new people, entertainment, emotional, cognitive,
Y. Olufadi / Telematics and Informatics 33 (2016) 452–471 455

social, etc.) derived from the use of the sites could spur an individual to always appear on the sites and this could increase the
amount of time use on the sites.
In addition, previous studies (e.g., Davis et al., 1992; Kim et al., 2007; Lin and Bhattacherjee, 2008; Lu and Su, 2009; Moon
and Kim, 2001; van der Heijden, 2004) have reported some factors (e.g., usefulness, user-friendliness, enjoyment, flexibility
of interaction, appearance of the sites etc.) as being the driving force (gratification or motivation) behind the peoples’ use of
information technology including the SNSs. For example, Kim et al. (2007) pointed out that perceived benefit affects the indi-
vidual’s use of information technology, consisting of cognitive and affective benefit. As another example, studies by Kang and
Lee (2010), Kwon and Wen (2010) and Sledgianowski and Kulviwat (2009) revealed that users’ perceived usefulness of SNSs
affects their positive intention (i.e., quality) to use the sites. Additionally, SNSs users’ cares about whether the sites would
allow them to effectively build and maintain relationships among the mechanisms that allow strangers to become
acquainted and keep in touch (Li and Bernoff, 2008; Pfeil et al., 2009), this therefore allows the users to form profiles and
enable people to reach out toward one another.
Meanwhile, the UGT proposes that human have needs (e.g., emotional, cognitive, social, etc.) and are (human) driven to
gratify and/or reduce those needs through the use of social media including the SNSs. Notably, evidences from the previous
studies suggest that some of these needs are satisfied when people connect to the sites. For example, the cognitive related
needs of people (e.g., students) may be achieved through posting of some difficult homework questions on the sites. Simi-
larly, the social needs are met through connecting with others like friends and family. As an example, Ellison et al. (2007),
found that Facebook use was motivated primarily by social gratifications, which include maintaining existing social ties and
being able to reconnect with friends from the past. In an earlier study by these researchers, they found that Facebook was
used primarily to help university students keep in touch with high school friends, who are often geographically distant, and
to learn more about new people they have met offline (Lampe et al., 2006).
To conclude this section, we would like to make some remarks regarding why it is difficult to emphasize quality over
quantity when it comes to the use of the SNSs by people. Firstly, the definition of quality or quantity of time use on the sites
is relative and subjective in nature since there is no universally agreed definition for quality or quantity especially with
regard to peoples’ use of SNSs. In this regard, what one will term quality (or ‘‘overuse”) of the time on the sites, might
not be for others.
Importantly, the definition of ‘‘quality” has multiple definitions. For instance, Juran et al. (1974) defined quality as fitness
for use, Gilmore (1974) and Levitt (1972) defined it as conformance to specifications while Crosby (1979) defined it as con-
formance to requirements. Several other authors like Grönross (1983) and Parasuraman et al. (1985) have defined quality
based on customers’ perception as the ‘‘degree and direction of discrepancy between customers’ service perception and
expectations”. We also defined quality as fitness for purpose or meeting the needs, requirements, and expectations of the
customers (i.e., the users). Looking at these definitions, we may ask the following questions. What is the standard for mea-
suring the quality of time use on the SNSs? Are quality and quantity of the time use on the sites the same? In other words, are
they comparable? If quality is defined as fitness for use, how can we measure the purpose of SNSs use by the people? On the
other hand, how do we know whether the sites meet the needs, expectations, and requirements of its users? These and many
more questions are important when it comes to emphasizing quality over quantity of time use on the sites.
Furthermore, a plethora of evidences in the existing studies have shown that the SNSs is used for many different pur-
poses: pleasure and entertainment value (Lin and Lu, 2011; Cheung et al., 2011), learning and education (Greenhow and
Robelia, 2009; Junco et al., 2011; Polsgrove and Frimming, 2013; GreGory et al., 2014), social enhancement, maintaining
interpersonal connectivity, self-discovery, social presence (Cheung et al., 2011), need to belong and self-presentation
(Nadkarni and Hofmann, 2012), etc. Thus, being able to match the purpose for which people use the SNSs might help us tells
whether the quality of use has been achieved or not. The lack of criteria for judging the quality of use (time) of SNSs over the
quantity of time on the sites is another major reason it could be difficult to say one is better or deserves more attention than
the other.
Essentially, we would find that it is really a question of the definition of the quality and quantity of the time use on the
sites. Nevertheless, we believe that both are important, and the overlap between them is important. Overall, rather than
thinking of quality versus quantity of time use on the sites; a better approach might be to look at how we can use the amount
of time people use on the sites to create better quality of its use. However, this is not the purpose for which the present
investigation is being conducted.

2.2. Other theoretical frameworks

One of the primary functions of using the SNSs is connectivity (see Ji et al, 2010). Indeed, all other functions like commu-
nication, seeking information, and content sharing etc. are directly or indirectly linked to connection. For instance, when an
individual communicates or share something (post some homework assignment, personal photos etc.) with another person
(say, family, friends, or colleagues in the place of work) on the SNSs, there is an element of connection inherent in the com-
munication or content shared because the person is trying to reach out (i.e., connect) to someone. To this end, the present
study could benefit from the connectivism theory due to Siemens (2004). In recent times, people connect with each other
anytime and in different places on the SNSs more than ever before. This may be partly explained by the advent of the mobile
technology cum the low cost of accessing the sites and partly due to the popularity and proliferations of the sites. The theory
(connectivism theory) is thus relevant to this study because it has important principles for investigating the way and manner
456 Y. Olufadi / Telematics and Informatics 33 (2016) 452–471

in which people connect with each other on the sites. This therefore informed our decision to develop the items in the SON-
TUS based on the places and situations people have reported or found to be using the SNSs.
Another theoretical work that is relevant to the conceptual framework of the current study is the uses and gratification
theory (UGT) by Katz (1959). The UGT could easily explain the use, choice of SNSs to use and the reasons people may decide
to use a specific platform in a specific place or situations. For instance, UGT proposes that human have needs (e.g., emotional,
cognitive, social, etc.) and are (human) driven to gratify and reduce those needs through the use of social media including the
SNSs. There are evidences that some of these needs are satisfied when people connect to the SNSs. For example, the cognitive
related needs of people (e.g., students) may be achieved through posting of some difficult homework questions on the sites.
Similarly, the social needs are met through connecting with others like friends and family. As an example, Ellison et al.
(2007), found that Facebook use was motivated primarily by social gratifications, which include maintaining existing social
ties and being able to reconnect with friends from the past. In an earlier study by these researchers, they found that Facebook
was used primarily to help university students keep in touch with high school friends, who are often geographically distant,
and to learn more about new people they have met offline (Lampe et al., 2006). These likely reasons (e.g., social, stress, emo-
tional etc.) were kept in mind when generating items for use in the development of our instrument. For instance, one of the
reasons people may decide to use the SNSs is stress-related (e.g., emotional stress). Our pilot study suggest that some people
usually use the sites to relieve themselves from the stress of their daily work or when they are going through emotional
trauma.
Meanwhile, a very good model that could provide a framework for examining the time use on the SNSs by people is the
technology acceptance model (TAM) proposed by Davis (1989). The two established constructs: perceived ease of use (PEU)
and perceived usefulness (PU) is believed to be relevant to the conceptual framework of the present study.
For instance, the usefulness of the sites and the ease with which people can connect or communicate with one another
plays an important role on the amount of time people use on the sites. For example, the benefits (e.g., meeting new people,
entertainment, emotional, cognitive, social, etc.) derived from the use of the sites could spur an individual to always appear
on the sites and this could increase the amount of time use on the sites. In fact the perceived usefulness of the sites could
increase the amount of time people use on the sites if their friends are always on the sites and respond to them each time
they tried to communicate or share information with them. However, if the friends are not always on the sites or do not
always appear on the sites, then, this could lead to decrease in the time people might spend on the sites. Additionally,
the PU could decrease the time spent on the sites if people feels threatened by cyber-crime such as spam, fraud (e.g. bank
fraud, identity theft), offensive content, harassment (e.g. sexual, cyber-stalking, cyber-bullying), security threats, privacy
issues, ‘‘sexting”, lots of unsolicited friends request etc. (Tynes, 2007; Benson et al., 2014).
As for the PEU, defined by Davis (1989) as ‘‘the degree to which a person believes that using a particular system would be
free of effort” (p. 320); the advent of mobile technologies, having internet-connected smartphones with free access and/or
very low daily/weekly charges could make people to always connect or reach out to their friends through their account(s).
For instance, in Nigeria, almost all the mobile network service providers have packages that would enable the subscribers to
access some of the most popular SNSs on weekly basis at a very low cost. This may explain the reasons people have reported
using the SNSs platforms at different times and places and often connect to the sites. Thus, these two constructs (PEU and
PU) could both potentially affect the frequency (low or high) of using the site and are therefore very relevant to our concep-
tual framework because it allows the potential respondents to respond to the items in the questionnaire according to the
frequency with which he/she engages in the use of the sites.

3. Study 1: exploratory ANALYSIS

This first study aims at designing an exploratory construct validity and reliability analysis of our instrument (SONTUS).
This was achieved through the procedures highlighted in the sections that follow. Meanwhile, the desired information in line
with the objective of the study was elicited from participants through the design of self-rated questionnaire; and in accor-
dance with Cronbach’s (1971) recommendation on the development of a new scale, we strived to draw representative items
from a universal pool in order to ensure content validity. More so, because there are many situations and places people can
use the SNSs, the lists of items used for the development of SONTUS were generated through two methods in line with the
recommendations in Carmines and Zeller (1979), Netemeyer et al. (2003), Schultz and Whitney (2005). (1) A review of the
relevant empirical studies (e.g., Ellison et al., 2007; Joinson, 2008; Valenzuela et al., 2009; Mazman and Usluel 2010; Xu et al.,
2012) published on peoples’ use of SNSs and (2) nonprofessionals’ opinion. For the second approach, the author contacted 40
people1 and told them that he needed their help for his next research project. These 40 individuals responded to this question:
kindly list as many as possible places and situations in which you have used or found others (e.g., friends, family, parent, spouse,
colleagues, children etc.) used the SNSs, or the reasons you are likely to use the SNSs. Indeed, the present study has benefitted
hugely from these two approaches as they offered it valuable help in generating the items used in the design of the question-
naire. Using these approaches, we accumulated a list of 63 items covering the different places, situations, and reasons people are

1
Convenience sampling was employed to select the 40 participants in this pilot study. We selected 20 students from the 3 universities in Ilorin (the Kwara
State capital in Nigeria) and the other 20 participants from the people that are no longer in school or with no university education (e.g., civil servants, bankers,
businesspersons etc.).
Y. Olufadi / Telematics and Informatics 33 (2016) 452–471 457

highly likely to use the SNSs. It is pertinent to remark here that during the items generation process, we strived to screen the list
for repetition, and this initial quality control process proved to be helpful in increasing the face validity of the scale. Meanwhile,
to validate the items in the questionnaire, we employed the following methods of validation: translational validity (i.e., content
and face validity), factor analysis and reliability test of internal consistency.

(a) Content validity: The first validation method we employed when developing SONTUS is content validity, and according
to DeVon et al. (2007), content validity ensures the items in the questionnaire reflect a complete range of the attri-
butes under investigation. To apply this method, we contacted eight reviewers to review the potential items. Although,
the criteria for choosing the reviewers of the items were subjective, each of these reviewers were faculty member from
five institutions and researchers in information, communication, system research, and scale development with sub-
stantial experience in the areas of SNSs use by the people. It is pertinent to note here that a clear description of the
conceptual framework for this study (in line with the recommendation from the previous studies e.g., Nunnaly and
Bernstein, 1994; Walsh, 1995) was presented to the reviewers of the items in the draft questionnaire. These reviewers
were then asked to review the draft 63-item2 in the questionnaire to ensure it was consistent with the conceptual
framework. During the review, the reviewers independently rated the relevance of each item to the conceptual frame-
work using a 4-point Likert scale from 1 (not relevant) to 4 (very relevant and succinct) (Lynn, 1986). The Content Validity
Index (CVI)3 was used to estimate the validity of the items. Using this criterion, we dropped 11 of the items because they
have a CVI score below .83. The other 52 items were retained as they have a CVI score ranging from .88 to 1.00. We
remarked that the reviewers’ comments especially with regard to the need to reword some of the items and possible sim-
ilarities in some of the items were taken into account during pre-testing and before the actual survey.
(b) Face validity: Although, face validity has been described as the weakest form of validity because of its subjective assess-
ment nature (Trochim, 2001); it however provide insight into how the potential participants might interpret and
respond to the items (DeVon et al. 2007). For this reason, before the distribution of the questionnaires after implement-
ing the suggestions received from the content reviewers, we decided to conduct a pilot study (i.e., pre-testing) to exam-
ine the instrument and get feedback on the length, clarity, grammar, organization, appropriateness, formatting,
readability, layout, and overall design of the questionnaire from the participants. The sample in this preliminary study
consisted of 20 individuals who own an account in at least one of the SNSs and five colleagues in the author’s institution
who are skilled in questionnaire design and scale development. These individuals (who are exempted from the actual
survey) were asked to fill the draft questionnaire and complete the face validity evaluation form using a Likert scale of 1
(poor) to 5 (excellent) with a center-point at 3 (good). The evaluation form includes clarity, grammar, organization,
appropriateness, formatting, readability, and layout. In addition, the reviewers were provided space in the evaluation
form to comment on the overall design of the questionnaire and if there were any information, they may usefully add to
improve the questionnaire. All the respondents rated each item at four or five, with approximately 93% indicated they
understood the items in the questionnaire and found them easy to answer, while 98% indicated the clarity, readability,
appearance, and layout would be acceptable to the intended target audience. With these favorable findings regarding
the structure and content of the questionnaire, the questionnaire was ready to be tested in an expanded population.
(c) Factor analysis and internal consistency reliability: These two methods of validation are common and standard statistical
techniques when developing a scale; we therefore refer interested readers to the following authors Cronbach (1951),
Cortina (1993), Nunnaly and Bernstein (1994), Schmitt (1996), DeVon et al. (2007), and Tabachnick and Fidell (2007)
for details.

3.1. Participants, design, and selection

We conducted a self-reported cross-sectional survey within Ilorin metropolis (the capital of Kwara State, Nigeria) during
the first week of May 2015. The eligibility criteria to participate in the survey are, (1) the participants must have at least one
account in any of the SNSs and (2) he or she must be 18 years or above as at the time of conducting this survey. The potential
participants were asked two relevant questions to the above criteria in order to ensure the eligibility criteria is satisfied
before they are given the questionnaire.
As for the selection of participants, we used quota sampling technique. By this method and using the local government
areas as the stratification variable, we divided Ilorin metropolis into three regions. We have used the quota sampling tech-
nique for the selection of the participants because it is simple and convenient to apply, and more importantly because it
mimic the stratified sampling technique. We therefore took advantage of this stratification approach (which ensure repre-
sentativeness) to obtain the desired information from the different regions within Ilorin metropolis.
Using the procedure described above, we distributed 2500 questionnaires and informed the participants that the partic-
ipation was voluntary. These participants were further informed that any information provided would be treated confiden-

2
A non-overlapping list of 63 items were collated and compiled from the response of the 40 individuals and the ones selected through empirical literature
review. In the author’s opinion, these 63 items encapsulate the major categories of situations, places or reasons people are likely to use the SNSs.
3
According to the CVI (Lynn, 1986), a rating of three or four indicates the content is valid and consistent with the conceptual framework (Lynn 1986). For
example, if six of the eight reviewers rate an item as relevant (3 or 4) the CVI would be 6/8 = .75. This score does not meet the .83 level of endorsement required to
establish content validity using a panel of eight experts at the .05 level of significance, and indicates the item should be dropped (Lynn, 1986; DeVon et al. 2007).
458 Y. Olufadi / Telematics and Informatics 33 (2016) 452–471

tially and used only for the purpose for which it was collected. Thereafter, we present a brief description of the survey and
detailed instructions on how to complete the questionnaire to those who agreed to participate in the study. The interested
participants were later asked to sign the consent form for the study after which each of the participants were given a ques-
tionnaire designed to evaluate their time use on the SNSs.
The questionnaire consists of two sections (demographic and SNSs use). The demographic information obtained from the
respondents include age, gender, marital status, number of SNSs account, religion, monthly income, working status
(employed, unemployed, self-employed), highest educational qualifications and ethnicity. In the second section, participants
were asked using a scale of 1 (not applicable to me during the past week) to 11 (I used it more than 3 times during the past
week but spent more than 30 min each time) to indicate how often they use the SNSs (e.g., Facebook) during the past week.
Obviously, we deliberately avoid the use of the common Likert scales like ‘‘almost never”, ‘‘frequently”, ‘‘almost always”,
‘‘sometimes” ‘‘often” etc., because it is subjective. For instance, a person’s opinion of ‘sometimes’ may in reality be more
or less frequent than another person’s opinion. Additionally, we have included the ‘not applicable’ option as part of the scale
because some items (e.g., when you are cooking, when you are at the clinic/hospital receiving treatment, when you wake up
in the midnight and could not sleep again etc.) might not be applicable to all the participants or during the particular week in
question. We present in the Appendix the 52 items used for the factor analysis and the scoring details of the resulting scale
i.e., SONTUS, after taking into consideration the results from the confirmatory factor analysis.

3.2. Descriptive statistics of the participants

The total number of participants we used during this exploratory study consisted of 2049 people who are on various SNSs
platform, and are currently domicile in Ilorin, Nigeria. Among this sample, 54.5%, 24.5%, 8.34% and 5.28% are married, single,
widowed, and divorced respectively. Meanwhile, nearly 8% of the participants failed to report their marital status. This par-
ticipants have age range from 20 to 58 years (M = 32.43, SD = 7.73) and 57% (n = 1168) were female. Most participants were
predominantly Yoruba (69.56%), followed by Hausa (12.33%) and while the rest are from other ethnic groups (e.g., Igbo, Nupe
or Ebira). Additionally, the two dominants religions in the country (Islam and Christianity) were adequately represented in the
survey, however and as expected, the majority of the participants were Muslims (67.35) while 4.2% of the participants did not
specify their religion. More so, 13.17%, 27.35%, 19.37%, 34.33% and 5.78% had the primary, secondary, Diploma, BSc, and post-
graduate degree qualifications respectively with about 76.17% of the sample working (either self-employed or employed).
Furthermore, the participants belonged to a variety of SNSs platforms like Facebook (88.52%), WhatsApp (77.73%), Twitter
(52.75%) and so many others like Instagram (54%), Pinterest (35.4%), 2go (49.3%), Badoo (23.12%), LinkedIn (43.21%) etc., and
on average, they reported having active accounts on between two and four platforms (M = 2.57, SD = .32). Meanwhile, the
total amount reported by the participants as their monthly income ranges between $35 USD and $1,200 USD (M = 158,
SD = 56.33). We should note that the amount quoted was based on the exchange rate as at the time of writing this article.
Meanwhile, about 12.21% of the participants decline to provide their monthly income. Finally, the participants reported they
spend on average between 43 min and 480 min per day on these SNSs (M = 179.67, SD = 36.42).

4. Factor analyses

4.1. Justification for the use of factor analyses

Factor analysis using a principal component analysis (PCA) as extraction method, with varimax rotation, was performed
on the 52 items in the draft questionnaire. We present in this section the reasons behind the suitability of the present data
for factor analyses. First, the correlation matrix reveals that all variables inter-correlate with at least one other variable at .30,
this suggest reasonable factorability. Secondly, the determinant of the correlation matrix for this data was .0031; therefore,
multicollinearity is not a problem for this data. In other words, all the 52 items in the draft questionnaire correlate fairly well
and none of the correlation coefficients are particularly large; thus, there is no need to eliminate any item at this stage. This
further confirms the suitability of the data for factor analysis. Furthermore, an examination of the Kaiser–Meyer–Olkin
(KMO) measure of sampling adequacy indicated that we should be confident that factor analysis is appropriate for this data.
The KMO = .91, a value that is above the recommended value of .5 (Kaiser, 1974). More so, the Bartlett’s Test of Sphericity
was also significant (v2 (406) = 15035.45, p < .00001) indicating that factor analysis is an appropriate method. Given these
overall indicators, factor analysis was deemed suitable with all the 52 items. Meanwhile, a number of criteria were used
to determine the full and subscales of SONTUS, these are: (a) a minimum eigenvalue of one, (b) examination of the scree plot
(i.e., factors above the point of inflexion in the curve are retained), (c) exclusion of items with factor loadings less than .55
and (d) the conceptual coherence of the factors.

4.2. Factor analyses results

The 52 item questions relating to the reasons or places/situations where people have reported using the SNSs were factor
analyzed using PCA extraction method and varimax (orthogonal) rotation. However, an iterative process was followed to
reach a meaningful factor structure because the scree plot bended between three and seven factors, we therefore sought
Y. Olufadi / Telematics and Informatics 33 (2016) 452–471 459

Table 1
Factor loadings and communalities based on a principal components analysis.

Item Factors
1 2 3 4 5
Q2 When you are at home sitting idly .82
Q22 When you are watching TV, news, football, films, sports, etc. .81
Q7 When you are listening to music, radio, religious lectures etc. .78
Q12 When you are in bed about to sleep .78
Q21 When you are in the company of friends/family/colleagues having fun .77
Q24 When you are a passenger in a car/bus/train for at least 2 min .75
Q6 When you are waiting for someone (e.g., friends) either in their house or at a pre-arranged place .71
Q14 When you are at a place to repair your car, house appliances, etc. .68
Q26 When you are waiting for your boss in her office for at least 2 min when she is not attending to you .67
Q28 When you are online doing school or job-related works e.g., project, homework .86
Q1 When you are at a seminar/workshop or training program .82
Q5 When you are doing school or job-related assignment at home .81
Q29 Watching academic-related video lectures or those related to your job .78
Q10 When you are in the class receiving lecture .73
Q13 When you are reading in the library for academic purpose e.g., recommended text for class .69
Q4 When you go to the stadium to watch football, basketball etc. .84
Q17 When you are at a social gathering like wedding ceremony, birthday party, reception etc. .80
Q23 When you go to the cinema house to watch movie(s) .78
Q9 When you are in a meeting .75
Q19 When you are sitting in a religious place (e.g., church, mosque) and activities like sermon or prayer is .73
yet to start
Q3 When you need to reduce your mental stress .77
Q16 When you want to reduce the pressure of your daily routines .74
Q8 When you have gone through a lot of stress .71
Q27 When you are trying to forget your financial challenges .71
Q15 When you need to reduce your emotional stress .70
Q25 When you need to find people you haven’t seen for a while .82
Q20 When you need to find out more about people you met offline .78
Q18 When you need to communicate with your families and friends .75
Q11 When you need to maintain contact with existing friends .67
Eigen value 22.45 18.12 15.07 11.43 8.87
% variance extracted 21.96 16.51 11.22 9.13 6.74
Cronbach’s alpha .91 .89 .85 .86 .83

Factor loadings <.55 are suppressed.

solutions that are more efficient and conceptually meaningful. To this end, three, four, five, six, and seven factor-solutions
were examined and items with loadings less than the cut-off point or loaded on multiple factors were discarded and the
remaining pool of items was reanalyzed. The final outcome of this process yielded 29 items divided into five factors (see
Table 1), and together explain 65.45% of the variance in the measured construct. In other words, among the factor solutions
we examined, the five-factor solution proved the most conceptually meaningful construct. Meanwhile, PCA with a varimax
rotation was used in preference to other methods of extraction (e.g., maximum likelihood estimator) because (1) we are
interested in uncovering the underlying dimension of the SONTUS construct, (2) it yielded clearer dimensions with no
cross-loadings, and (3) because our main goal was to identify and compute composite scores for the factors underlying SON-
TUS. In addition, a review of the initial factor loadings suggests that the proper solution was attainable through PCA with
varimax rotation, as it converges in six iterations. Indeed, support for the convergence validity of the scales is evident
because each item converged significantly on the same factor tapping these items; additionally, since there were no multiple
loadings among the items, this provide evidence for the discriminant validity of the scale. This finding therefore
provides initial support for the validity of the instrument (i.e., SONTUS). Finally, we labeled each subscale of the SONTUS
as follows.
The first factor consists of nine items. It is clear from Table 1 that these nine items relate to situations in which people
might be less busy, relaxing, or have free time and seize the opportunity to use the SNSs, to this end, we labeled the first
factor as ‘‘Relaxation and Free Periods”. These nine items have loadings that range from .67 to .82, and explain together
21.96% of the variance. Meanwhile, we labeled the second factor as ‘‘Academic-related Periods,” because it contains items
relating to the use of the SNSs during academic-related periods or activities. This second factor consists of six items, has load-
ings that range from .69 to .86, and explain together 16.51% of the variance. The third factor is labeled ‘‘Public-places-related
use”, it consists of items that loads between .73 and .84, and explains 11.22% of the variance. The next factor is labeled
‘‘stress-related Periods”. This factor explains 9.13% of the variance and loads between .70 and .77. The last factor is labeled
‘‘Motives for use,” and comprises four items all of which talking about the reasons people are likely to use the SNSs. These
items load .67–.82 on their factor and explain 6.74% of the variance.
460 Y. Olufadi / Telematics and Informatics 33 (2016) 452–471

4.3. Internal consistency reliability

The internal consistency reliability of SONTUS and its five subscales were examined using Cronbach’s Alpha (Cronbach,
1951). The analysis reveals that alpha was .92 for the full scale while its five subscales have alpha values ranging from
.83 to .91 (see Table 1). We note that no substantial increase in alpha for SONTUS and its five subscales could have been
achieved by eliminating one or more items. We present in what follows, some possible explanation for the ‘‘high” values
of alpha we reported in this study which suggest sound reliability. First, the width of the scale employed in SONTUS provides
support for the ‘‘high” alpha values reported in the present study. This is in line with the suggestion given in Voss et al.
(2000) that instruments with more than 4-points Likert scales in width lead to greater variance and consequently increases
the value of alpha. For the SONTUS construct, we employed 11-point Likert scales in accordance with Voss’s et al. (2000)
recommendation, this may likely be a reason we have a ‘‘high” alpha values. In addition, although there is a debate about
what constitute an ‘‘appropriate sample size” when calculating the reliability of a scale, Helms et al. (2006) noted that
increasing the number of participants responding to a scale can increase the value of alpha. As we reported earlier, our sam-
ple size is 2049, which is considerably large and therefore, might be a pointer to the ‘‘high” alpha values we reported in this
study. Furthermore, it has been shown that alpha increases as the number of items in the scale increase (Cronbach, 1951;
Voss et al., 2000). In the present study, the number of items included in the full scale and subscales starts from four; this
may therefore explain the ‘‘high” value of alpha and may indicate that there is no issue with the construct or certain items.
Based on the above reasons, we submit that our instrument (i.e., SONTUS) is very reliable.

5. Study 2: confirmatory analysis

The data-gathering approach of study 2 were similar to those used in the exploratory study (i.e., study 1). However, in this
second study, we distributed 2200 questionnaire, however, only 1808 were useable. The questionnaire we used for this
confirmatory study consists of the 29 items we found from the factor analysis in study 1 together with several other
standardized measures of personality and wellbeing. The essence of these measures is to carry out further convergent, incre-
mental, and discriminant validity of our instrument (i.e., SONTUS). We described briefly information about the various
measures.

(a) Positive relations with others: Ryff and Keyes (1995) developed this scale in. It consist of nine-item and was used to
assess the quality of relations that participants tend to form with others. Participants rated each item on a 6-point
scale ranging from 1 (strongly disagree) to 6 (strongly agree). Higher scores indicate positive relations with others.
(b) Hospital anxiety and depression scale. The hospital anxiety and depression scale developed by Zigmond and Snaith
(1983) is a 14-item self-report scale that consists of a depression and an anxiety scale, each with 7 items. The scale
was designed to screen for mood disorders in general (non-psychiatric) medical outpatients. It focuses on subjective
disturbances of mood rather than physical signs, and aims at distinguishing depression from anxiety. Responses to
each item in the scale is rated on a four point Likert scale ranging between ‘‘No, not at all (0)” and ‘‘Yes definitely
(3)”. Sample item for anxiety and depression subscales are, ‘‘I wake early and then sleep badly for the rest of the night”
and ‘‘I feel life is not worth living” respectively.
(c) Shyness: Shyness was measured using the Revised Cheek and Buss Shyness scale (RCBS: Cheek, 1983). The RCBS is a
13-item measure of shyness that measure the extent to which the respondent feels that each item is characteristic
of them. The participants rated each item on a 5-point Likert scale ranging from 1 (very uncharacteristic) to 5 (very
characteristic). Its sound psychometric properties have been reported in various studies (e.g., Hopko et al., 2005).
The Cronbach’s alpha for scores on the overall scale was .87 in the current sample.
(d) Facebook addiction: The revised version of The Bergen Facebook Addiction Scale due to Andreassen et al. (2012) was
employed in the present study to measure participants’ level of addiction to Facebook. This revised scale consist of
six items which the participants rated on a 5-point scale ranging from 1 (very rarely) to 5 (very often). A sample item
from this scale is ‘‘How often during the last year, have you felt an urge to use Facebook more and more?” For this
study, Cronbach’s alpha is .78.
(e) Personality: The 10-item Big-Five Personality Inventory (i.e., Extraversion, Agreeableness, Emotional Stability, consci-
entiousness and Openness) developed by Gosling et al. (2003) was used in the present study to measure the person-
ality of the participants. This scale was used because it is very brief, takes little time to complete, and has been
reported as having good psychometric properties. The alpha values for each of the five scales falls in the range .72–
.81 in the present sample.
(f) Internet addiction: The internet addiction level of the participants was measured using Young’s Internet Addiction Test
(Young, 1998). The scale consists of 20 items rated using the following Likert scale: 0 = does not apply; 1 = rarely;
2 = occasionally; 3 = frequently; 4 = often; 5 = always. A sample item from this scale is ‘‘How often do you find that
you stay on-line longer than you intended?” The psychometric properties of this scale were investigated in 86 subjects
in a study conducted by Widyanto and McMurren (2004). They found that it showed good concurrent validity and
internal consistency. In the present study, the alpha value was found to be .84.
Y. Olufadi / Telematics and Informatics 33 (2016) 452–471 461

(g) Loneliness: Loneliness was measured using The Social and Emotional Loneliness Scale for Adults – Short version
(SELSA-S) as developed by DiTommaso et al. (2004). The SELSA-S consists of 15 items that measures loneliness in
an adult population. The participants in the present study rate their level of agreement with each item using a 7-
point Likert scale (1 = ‘‘Strongly disagree” to 7 = ‘‘Strongly agree”). SELSA-S has demonstrated good psychometric
properties and has been recommended by Cramer and Barry (1999) as a very good measure of loneliness. The Cron-
bach’s alpha value for this scale in the present study was found to be .88.

5.1. Confirmatory sample

Like in the first study, most of the participants in this confirmatory study were females 63.67% (n = 1152), and the average
age in the sample was 33.46 years (SD = 3.86). Close to 76% of the participants reported having average Internet experience,
and the remaining have either low or high Internet experience. Additionally, majority of the participants (76.2%) in this study
have been using the SNSs for more than 3 years, while others reported using it for less than 3 years. Furthermore, the par-
ticipants belonged to a variety of SNSs platforms like Facebook (78.12%), WhatsApp (67.73%), Twitter (32.45%), WeChat
(23.78%), Instagram (43.44%), LinkedIn (41.4%) etc., and on average, they reported having active accounts on between two
and five platforms (M = 3.24, SD = .46). More so, 18.57%, 22.45%, 13.31%, 38.11% and 7.56% had the primary, secondary,
Diploma, BSc, and postgraduate degree qualifications respectively with about 69.72% of them working (either self-
employed or employed).

5.2. Hypothesized models

 Model 1: This model hypothesized one first-order factor (SONTUS), accounting for all the common variance among the 29
items. By this model, we are assuming that time spent on the SNSs is a single first order construct. This assumption is
motivated by the typical practice of scaling new construct by adding the individual items in the construct to obtain
the total score.
 Model 2: By this model, we hypothesized that the 29 items divided into five uncorrelated or orthogonal first-order factors
(i.e., relaxation, academics, public, stress, daily routine) based on the results from the factor analysis in study 1. We should
recall that the use of varimax (orthogonal) rotation resulted in five uncorrelated factors; thus, Model 2 is considered a
plausible model of the underlying data structure.
 Model 3: This last model hypothesizes five first-order factors and one second-order factor (SONTUS). The model was
included because of the possible correlation that might exist between the factors. In fact, if the first-order factors are cor-
related, it is possible that the correlations between first order factors is statistically ‘‘caused” by a single second-order
factor.

5.3. Hypothesized model results

A number of indices was used in this study to check the adequacy of the hypothesized models in the present study. This is
because there is no universally acceptable statistic as an index of checking model adequacy. To this end, relative indices like
normed fit indices (NFI), target coefficient, and ratio of chi-square to its degree of freedom were used to compare models.
Additionally, absolute indices of goodness-of-fit such as chi-square, Goodness of Fit Index (GFI), Adjusted Goodness of Fit
Index (AGFI) are used to evaluate the individual model. More so, the Standardized Root Mean Squared Residual (SRMR)
and Root Mean Square Error of Approximation (RMSEA; Browne and Cudeck, 1993; Steiger and Lind, 1980) was used to esti-
mate the lack of fit in the model compared with a perfect model. The larger the value the greater the misspecification; a
RMSEA value less than 0.08 is considered indicative of an adequate fitting model and smaller values of the SRMR are asso-
ciated with better fitting models with scores below .05 considered as evidence of good fit (Byrne, 1989). Lastly, two incre-
mental fit index measures, namely the Tucker–Lewis Index (TLI: Tucker and Lewis, 1973) and Comparative fit Index (CFI:
Bentler, 1990) were also employed. These indices compare the fit between the specified model and an independent (null)
model. For these two indices, a value of .90 is generally considered to indicate an acceptable model and a value greater than
.95 is considered an excellent model (Bentler and Bonett, 1980; Hu and Bentler, 1999).
The goodness-of-fit indices for the hypothesized models and the null model are as presented in Table 2. We should note
that the null model was included in order to establish the zero-point for the NFI. Meanwhile, previous studies have made
effort to specify a threshold for measuring model fit. For instance, Marsh and Hocevar (1985) recommends using the ratio
of chi-square value to its degree of freedom when its value is as low as 2 or as high as 5, as evidence of a reasonable fit.
As for GFI and AGFI, researchers usually interpret GFI or AGFI scores in the .80–.89 range as representing reasonable fit;
scores of .90 or higher are considered evidence of good fit. Following this criteria, it is clear that the null model provides
a poor fit to the data (see Table 2).
Meanwhile, looking at all the indices of goodness-of-fit considered in this second study, Model 1 provides a substantially
better fit relative to the null model, however, by normal standards, Model 1 is not close to being considered a good fit with
the sample data. Moreover, Model 2 substantially improves all indices of goodness-of-fit relative to Model 1. Model 2 there-
fore shows good model-data fit, as indicated by the various indices of model fit considered in the present study (see Table 2).
Thus, from the empirical standpoint, this model provides a more than satisfactory solution. The last model (Model 3) shows
462 Y. Olufadi / Telematics and Informatics 33 (2016) 452–471

Table 2
Goodness-of-fit indices for the hypothesized models.

Fit measures Good fit Acceptable fit Null Model Value 1 first-order 5 first-order factors 5 first-order factors,
factor model (uncorrelated) 1 second-order
value model value factor model value
v2 (df) – – 10778.11 (406) 4259.554 (377) 1492.150 (367) 1659.768 (372)
v2 =d v2 =d < 3 3 < v2 =d < 5 26.55 11.3 2.27 2.46
NFI .95 6 NFI 6 1 .90 < NFI < .95 – .61 .92 .90
GFI .95 6 GFI 6 1 .90 < GFI < .95 0.52 .82 .95 .94
AGFI .9 6 AGFI 6 1 .85 < AGFI < .90 0.71 .79 .94 .93
SRMR 0 < SRMR < .05 0 < SRMR < .10 0.65 .073 .036 .044
RMSEA 0 < RMSEA < .05 .05 < RMSEA < .08 0.31 .075 .041 .044
TLI .95 6 TLI 6 1 .90 < TLI < .95 0.58 .60 .95 .94
CFI .95 6 CFI 6 1 .90 < CFI < .95 0.61 .63 .97 .95

NFI = normed fit indices; GFI = Goodness of Fit Index; AGFI = Adjusted Goodness of Fit Index; SRMR = Standardized Root Mean Squared Residual;
RMSEA = Root Mean Square Error of Approximation; TLI = Tucker–Lewis Index; CFI = Comparative fit Index.

excellent model-data fit, as indicated by the absolute indices (GFI, AGFI), incremental fit indices values (TLI and CFI). Mean-
while, as expected for a second-order model, Model 3’s GFI and AGFI scores are slightly lower than its first-order counterpart
(Model 2), however, it provides substantial improvement over Model 1, as evidenced by the changes in the NFI index (from
.61 to .90) and the ratio of chi-square to degrees of freedom (from 11.3 to 2.46). Thus, the results suggest that Models 2 and 3
are satisfactory and competing representations of the underlying structure of our construct.
Furthermore, the target coefficient was used to test for the existence of a higher-order user time construct. Using Model 2
as the target model, the target coefficient is the ratio of the chi-square of Model 2 to the chi-square of Model 3. In this case, a
target coefficient of .92 provides reasonable evidence of a second-order construct inherent in the time use on the sites. This
result indicates that Model 3 explains 92% of the variation in the five first-order factors in Model 2. Meanwhile, when com-
paring first-order and second-order models, it is important to realize that the higher order factors are merely trying to
explain the covariation among the first-order factors in a more parsimonious way (i.e., one that requires fewer degrees of
freedom). Consequently, even when the higher-order model is able to explain effectively the factor covariations, the
goodness-of-fit of the higher-order model can never be better than the corresponding first-order model (see Table 2). Over-
all, there is reasonable evidence to suggest a second-order construct.
It is possible to examine the validity and reliability of the individual items using either Models 2 or 3. In fact, estimates of
item validity and reliability are not sensitive to the addition of a second-order factor; thus, conclusions concerning the valid-
ity and reliability of the 29 items would be the same regardless of which model was selected. Meanwhile, because Model 3
has the additional advantage of providing estimates of the validity and reliability of the latent factors, we would recommend
Model 3. We therefore proceed with the analysis of the validity and reliability of factors and items assuming this second-
order model.
The estimates of Model 3’s standardized parameter estimates are presented in Table 3 for both latent variables and
observed variables. For the observed variables, Table 3 shows factor loadings, their corresponding t-values, and R-square val-
ues. According to Bollen (1989) and Mueller (1994), the larger the factor loadings or the corresponding t-values – the stron-
ger the evidence that the measured variables or factors represent the underlying constructs. With t-values above 2.0 being
considered significant, factor loadings can be interpreted as indicators of validity for the 29 items. All items have large
(greater than .74) and significant loadings on their corresponding factors, indicating evidence of good construct validity.
The proportion of the variances, or R-square, in the observed variables that is accounted for by its corresponding latent vari-
able is used as an indicator of each item’s common factor reliability, this has values that ranges from .53 to .86, indicating
acceptable reliability for all items. The total coefficient of determination for the first-order model is .93, indicating excellent
overall reliability of the 29 items combined.
For the latent factors in Table 3, the standard structural coefficients can be interpreted as indicators of validity of the
latent factors as components of the SONTUS construct; and as with the observed variables, the t-values are above 2.0 and
significant, all factors have large (greater than .82) and significant structural coefficients, these together indicates good con-
struct validity. The R-square values for each of the five latent factors range from .67 to .86, indicating acceptable reliability
for all factors. The total coefficient of determination for the structural equations of the second-order SONTUS factor is .93,
indicating excellent overall reliability of the five first-order factors combined.

6. Validity analyses

In this section, we provide evidence of convergent validity, predictive validity, and incremental validity of SONTUS. For
example, when testing its convergent validity, we hypothesized that SONTUS and its five subscales would exhibit convergent
validity by showing a modest positive and significant correlations with theoretically related constructs considered in the
present study. Furthermore, the predictive validity of SONTUS and its five subscales was assessed by correlating it with some
Y. Olufadi / Telematics and Informatics 33 (2016) 452–471 463

Table 3
Standardized parameter estimates and t-values for model 3.

Observed variables Latent variables


Item Factor loading R-square (reliability) Factor Std. structure coefficient R-square (reliability)
Q2 .83* .69 Relaxation and free periods .87 (14.65) .76
Q22 .76 (13.593) .58
Q7 .74 (13.685) .55
Q12 .75 (13.701) .56
Q21 .87 (12.763) .76
Q24 .76 (13.368) .58
Q6 .79 (13.666) .62
Q14 .85 (12.757) .72
Q26 .88 (13.160) .77
Q28 .79* .62 Academic-related periods .89 (11.744) .79
Q1 .86 (20.522) .74
Q5 .78 (20.661) .61
Q29 .85 (21.399) .72
Q10 .73 (21.985) .53
Q13 .83 (20.744) .69
Q4 .86* .74 Public-places-related use .89 (11.218) .79
Q17 .77 (16.737) .59
Q23 .82 (16.647) .67
Q9 .78 (17.737) .61
Q19 .77 (17.418) .59
Q3 .75* .56 Stress related periods .93 (10.852) .86
Q16 .83 (15.756) .69
Q8 .75 (14.583) .56
Q27 .85 (13.788) .72
Q15 .78 (12.667) .61
Q25 .82* .67 Motives for use .82 (5.476) .67
Q20 .93 (13.976) .86
Q18 .74 (13.552) .55
Q11 .86 (12.851) .74

t-Values for item factor loadings and factor structural coefficients are indicated in parentheses.
*
Indicates a parameter fixed at 1.0 in the original solution.

standardized measures of personality and wellbeing (i.e., Big-Five Personality Inventory, Positive Relations with Others,
Hospital Anxiety and Depression Scale, Cheek and Buss Shyness Scale, Social and Emotional Loneliness Scale for Adults –
Short version). Finally, we also examined the evidence of incremental validity of the SONTUS in the present study
(i.e., we examined the degree to which the five subscales of SONTUS predict personality and wellbeing beyond the effects
of demographic variables and other theoretically related constructs.

6.1. Convergent validity: correlations with theoretically related constructs

As expected, a moderately positive and significant association was obtained between the SONTUS, its five subscales and
two related constructs (Internet Addiction Test and Facebook Addiction Scale). In particular, greater levels of Internet Addic-
tion Test and Facebook Addiction Scale were related significantly to higher scores on SONTUS and its subscales with values
ranging between .26 and .43, except the Academic-related activities subscale (r = .22); on which greater levels of addiction
to internet and Facebook were related significantly to lower scores on this subscale.

6.2. Predictive validity: correlations with the personality and wellbeing measures

The correlations between SONTUS, its five subscales, and the personality and wellbeing measures are provided in Table 4.
Obviously, it is easy to observe from Table 4 that SONTUS subscales were associated with various personality and wellbeing
measures considered in this study. These results therefore support the predictive validity of the SONTUS.

6.3. Incremental validity: hierarchical regression analyses

In this section, we present the results of the Hierarchical Linear Multiple Regression (HLMR) analysis performed in order
to investigate the incremental validity of SONTUS, after controlling for participants’ demographic variables and the two the-
oretically related constructs to SONTUS. In the first step of the HLMR, six demographic variables and theoretically similar
constructs to SONTUS were entered; while the five factor analytically derived subscales of SONTUS were entered as one block
in the second stage. This process was repeated for each of the outcome measure (i.e., personality and wellbeing measure)
464 Y. Olufadi / Telematics and Informatics 33 (2016) 452–471

Table 4
Correlations between the SONTUS, its subscales, and the outcome measures.

Relaxation and free Academic-related Public-place-related Stress-related Motives for


periods activities use periods use
Loneliness .74** .19* .69* .33** .18**
Shyness .53* .21** .49* .41** .32*
Anxiety .44* .23* .37* .66* .11**
Depression .36** .13** .27** .37** .08**
Positive relations with others .51** .43* .43 .21 .21**
Extroversion .37* .31 .33* .15* .32*
Agreeableness .21** .12 .12* .34** .11*
Openness .12* .19* .23 .17* .21**
Conscientiousness .18* .43** .26 .21 .19
Neuroticism .28** .23* .21 .22* .22*
*
p < 05.
**
p < 01.

Table 5
Regression estimates of outcome measures on demographic variables, two related constructs to SONTUS and the factor analytically derived subscales of
SONTUS.

Variable Loneliness Shyness Anxiety Depression Extraversion Agree- Openness Positive Neuroticism Conscien-
ableness relation tiousness
Gender .18* .18* .34 .30 .27** .26 .31 .33* .21* .21**
Age .17 .14* .03 .17* .06 .05 .11* .11 .12 .10
Religion .08 .11* .14 .32 .27** .26 .01 .13 .21* .21**
Work status .18 .18* .24 .15 .27** .26 .31 .08 .14* .21**
Income .07 .05 .02 .03 .07 .09 .06 .03 .04 .01
No of SNSs account .05 .06 .41* .23 .09 .11 .13* .13 .03 .05
Facebook addiction .18** .18* .34* .30* .27** .26 .21* .13* .21* .21**
Internet addiction .18* .18* .34* .30* .27** .26 .11* .33* .21* .21**
Relaxation .28** .22** .28** .28** .28** 0.52*** .28** .34* .28** .59***
Academics .37* .33*** .37* .37* .37* 0.5*** .37* .23* .37* .60***
Public .43*** .42*** .43*** .43*** .43*** 0.29*** .43*** .04* .43*** .22**
Stress .34* .24* .34 .34* .34* .21* .34 .21* .34* .21*
Motives .21* .23 .21* .21* .21* .23 .21* .23 .21** .23*
R2 .72 .42 .30 .52 .25 .45 .57 .37 .43 .45
DR2 .13* .29*** .23** .49*** .10** .36*** .17* .20** .16*** .29***
F for DR2 14.95 10.34 7.12 12.94 23.07 30.07 11.35 24.48 15.37 24.24

Note: The coefficients listed are standardized regression weights obtained at the end of the analyses.
*
p < .05.
**
p < .01.
***
p < .0001.

considered in the present study. It is pertinent to note that since the outcomes of interest were many and were not identical,
it was not possible to provide straightforward description of the HLMR results. Meanwhile, after controlling for the demo-
graphic variables and related constructs to SONTUS, the five factor analytically derived subscales combined accounted for
unique variance in all of the outcome measures (see Table 5). They explained between 25% (Extraversion) and 72% (Loneli-
ness) of the variance.

7. Discussion

The main objective of the present investigation was to develop a simple, short, and psychometrically sound scale capable
of measuring the time spent by the people on the SNSs. To this end, we conceptualized and measured the time use by the
people on the SNSs as a multi-dimensional construct consisting of five factors: relaxation and free periods, academic-related
periods, public-place-related use, stress-related periods, and motives for use. The validation process we employed using both
exploratory and confirmatory factor analysis techniques produced a 29-item instrument, which exhibits excellent psycho-
metric properties. In addition, the investigation of the construct validity of the SONTUS using 10 personality and well-
being measures, and 2 theoretically related constructs to SONTUS provide preliminary evidence for the convergent, predic-
tive, and incremental validity of SONTUS.
Indeed, one of the main contribution of the present investigation is that for the first time, we are able to conceptualize the
time people committed to using the SNSs by developing a psychometrically tested scale. This new instrument benefitted
Y. Olufadi / Telematics and Informatics 33 (2016) 452–471 465

from three theories during its conceptual framework development, and is rooted in self-perceived experiences (e.g., social,
emotional, cognitive etc.) associated with the use of the SNSs. Another important contribution of the proposed construct is
its unifying feature. To our best knowledge, we now have a single measure with good psychometric properties that could be
used by interested researchers to measure the time use by the people on the sites. In addition, having a single instrument like
SONTUS would allow for easy comparison of the results from different studies and open up a new area of research on which
future researchers can validate the existing study using different population and many different outcomes of interest. Addi-
tionally, given the wide variability in measuring the time spent on site(s) in the existing studies, data collected from this
study may inform future data collection methods; and provide researchers a uniform and consistent measure of the time
use by the people on the sites.
A plethora of methods (following a superficial approach) have been employed when trying to measure the time use by the
people on the SNSs by the previous authors. For instance, majority of the previous studies have measured the time spent on
the site through a single-item scale that usually focused on frequency of use, duration of use, number of logins, history of use,
number of friends etc., on a daily or weekly basis (Ellison et al., 2007; Joinson, 2008; Pempek et al., 2009; Valenzuela et al.,
2009). These approaches, as discussed in Section 1 have led to underestimation of the time use and results in varying results.
Although, efforts by these previous scholars are helpful, useful, and provide insight into the operational definitions of the
time use on the sites, yet their effort are not enough to capture all the dimensions of this complex construct i.e., time use
on sites. However, the present study offered a promising measurement instrument through deep conceptualization of SNSs
usage time by taking into account the reasons for use, and places or situations where it is use. Additionally, the multidimen-
sional approach suggested in the present investigation will hopefully add insights to our understanding of SNSs use time and
encourage interested researchers to further develop and test the models reported here.
Furthermore, attempt by the previous authors were too fragmented and lack synthesis; with different authors defining
the time use on the sites in a different manner. To this end, we have improved on this by developing a scale that can be used
to measure time spent on any SNSs. In fact, having a uniform index for measuring time use on the sites does not only have
the capability of facilitating a scientific communication and/or interaction among interested researchers, but also allows for
easy comparability of results from different studies.

8. Study limitations

To our best knowledge, this is the first attempt at developing a scale (with good measurement properties) for measuring
the time spent on the SNSs. Although, the findings are quite promising, this study like many other scientific investigations
has some limitations that must be considered when interpreting its results. Firstly, the results of the confirmatory factor
analysis reported here should be interpreted with caution. This is because the criteria for judging the goodness-of-fit or com-
paring models are relative rather than absolute. In addition, there is no universally agreed threshold or standard cutoff values
for evaluating model-data fit or the existence of higher-order constructs. Furthermore, the fact that a model fits one sample
(like the one reported in the present investigation) does not imply that it is the ultimate solution. To this end, future
researchers can re-specified our model using a confirmatory factor analysis to revise and improve the model.
In addition, although the samples in the present investigation is representative of the users of the SNSs as it cuts across
different age-group with diverse socio-economic and demographic characteristics; the use of a non-probability sampling
approach may not provide an accurate representation of the population of SNSs users in the study area. By this way, cautions
should be taken when generalizing the findings despite the reliability and validity of the instrument. Also, the sample con-
sists of individuals residing in one of the cities the North central zone of Nigeria, excluding other people from the different
part of the country. However, this may not posed a problem as the samples in the present investigation represent the dif-
ferent age group and ethno-religious group in the country, nevertheless, future research must examine the statistical prop-
erties of this scale in other regions and national contexts. Moreover, the items in the scale is based entirely on literature
written/published in English and subjective viewpoints of 40 people. It is possible that other studies on SNSs time use have
been published in different languages not considered in this study.

9. Conclusions

This study reports an exploratory–confirmatory study through a rigorous validation of the proposed measurement instru-
ment. Despite the limitations highlighted in Section 8, the evidence presented here indicates that the scale is a promising
instrument for the measurement of time spent on the SNSs, through which researchers can measure and can make important
contributions to understanding and predicting how the times spent on the SNSs affect for example, individuals’ wellbeing.
Thus, SONTUS can be used as a standardized measure of time use on the sites. Overall, we hope the results presented in our
present investigation sparks the interests of researchers to refine this new measure possibly by analyzing the correlation
between the error terms of items or respecifying the model. In addition, we hope that this scale serves as a valuable resource,
reference, and guide for researchers interested in this fascinating field of inquiry.
466 Y. Olufadi / Telematics and Informatics 33 (2016) 452–471

Appendix

Kindly use the scale below to indicate how often you always use the social networking sites like Facebook, Instagram,
WhatsApp, Twitter, Myspace, Pinterest etc., during the past week in the following situations and places:

1 = Not applicable to me during the past week.


2 = I never used it during the past week.
3 = I used it once during the past week but spend less than 10 min.
4 = I used it once during the past week but spend between 10 and 30 min.
5 = I used it once during the past week but spent more than 30 min.
6 = I used it between 2 and 3 times during the past week but spend less than 10 min each time.
7 = I used it between 2 and 3 times during the past week but spend between 10 and 30 min each time.
8 = I used it between 2 and 3 times during the past week but spent more than 30 min each time.
9 = I used it more than 3 times during the past week but spend less than 10 min each time.
10 = I used it more than 3 times during the past week but spend between 10 and 30 min each time.
11 = I used it more than 3 times during the past week but spent more than 30 min each time.

Table A1. The 52 items used for the factor analysis.

Items 1 2 3 4 5 6 7 8 9 10 11
1. When you are at home sitting idly
2. When you are watching TV, news, football, films, sports, etc.
3. When you are in the office but idle
4. When you are waiting for someone (e.g., friends) either in their house or at a
pre-arranged place
5. When you are waiting for a bus/train at the bus/train station
6. When you are in the company of friends/family/colleagues having fun
7. When you are at a place to repair your car, house appliances, etc.
8. As a driver when trapped in heavy traffic for at least 2 min
9. When you are eating or drinking outside your home e.g., cafeteria
10. When you are relaxing
11. When you are in bed about to sleep
12. When you are at the clinic/hospital waiting to be attended to by the doctor
13. When you are listening to music, radio, religious lectures etc.
14. When you are a passenger in a car/bus/train for at least 2 min
15. When you are waiting for your boss in her office for at least 2 min when
she is not attending to you
16. When you are at the market, shopping mall etc.
17. When you are sitting in a religious place (e.g., church/mosque) and
religious activities like prayer or sermon is in progress
18. When you are at the place of work with a lot of work to do
19. When you are at the clinic/hospital receiving treatment
20. When you are sitting in a religious place (e.g., church, mosque) and
activities like sermon or prayer is yet to start
21. When you are walking on the street, road-side, class corridors etc.
22. When you go to the stadium to watch football, basketball etc.
23. When you are reading in the library for non-academic purpose e.g., reading
the newspaper
24. When you are at a social gathering like ⁄(wedding ceremony, birthday
party, reception etc)
25. hen you go to the cinema house to watch movie(s)
26. When you are in the class receiving lecture
27. When you are reading/studying for academic purpose outside the library
e.g., at home
28. When you are reading in the library for academic purpose e.g.,
recommended text for class
29. When you are reading or studying something related to your work/job
30. When you are online doing school or job-related works e.g., project,
Y. Olufadi / Telematics and Informatics 33 (2016) 452–471 467

Appendix (continued)

Items 1 2 3 4 5 6 7 8 9 10 11
homework
31. When you are at a seminar/workshop or training program
32. When you are doing school or job-related assignment at home
33. Watching academic-related video lectures or those related to your job
34. When you are sitting in a religious place (e.g., church, mosque) to learn
about your religion
35. When you need to reduce your mental stress
36. When you want to reduce the pressure of your daily routines
37. When you have emotional worries
38. When you have gone through a lot of stress
39. When you are trying to forget your financial challenges
40. When you need to reduce your emotional stress
41. When you need to reduce your physical stress
42. When you need to find people you haven’t seen for a while
43. When you need to find out more about people you met offline
44. When you need to communicate with your families and friends
45. When you need to maintain contact with existing friends
46. When you are cooking
47. When you are eating or drinking at home
48. When you are dressing up for class or office
49. When you wake up in the morning
50. When you wake up in the midnight and couldn’t sleep again
51. When you are on a queue for at least 2 min
52. When you are in a meeting

Table A2. Social networking time use scale (SONTUS).

Item 1 2 3 4 5 6 7 8 9 10 11
1 When you are at a seminar/workshop or training program
2 When you are at home sitting idly
3 When you need to reduce your mental stress
4 When you go to the stadium to watch football, basketball etc.
5 When you are doing school or job-related assignment at home
6 When you are waiting for someone (e.g., friends) either in their house or
at a pre-arranged place
7 When you are listening to music, radio, religious lectures etc.
8 When you have gone through a lot of stress
9 When you are in a meeting
10 When you are in the class receiving lecture
11 When you need to maintain contact with existing friends
12 When you are in bed about to sleep
13 When you are reading in the library for academic purpose e.g.,
recommended text for class
14 When you are at a place to repair your car, house appliances, etc.
15 When you need to reduce your emotional stress
16 When you want to reduce the pressure of your daily routines
17 When you are at a social gathering like wedding ceremony, birthday
party, reception etc.
18 When you need to communicate with your families and friends
19 When you are sitting in a religious place (e.g., church, mosque) and
activities like sermon or prayer is yet to start
20 When you need to find out more about people you met offline
21 When you are in the company of friends/family/colleagues having fun

(continued on next page)


468 Y. Olufadi / Telematics and Informatics 33 (2016) 452–471

Appendix (continued)

Item 1 2 3 4 5 6 7 8 9 10 11
22 When you are watching TV, news, football, films, sports, etc.
23 When you go to the cinema house to watch movie(s)
24 When you are a passenger in a car/bus/train for at least 2 min
25 When you need to find people you haven’t seen for a while
26 When you are waiting for your boss in her office for at least 2 min when
she is not attending to you
27 When you are trying to forget your financial challenges
28 When you are online doing school or job-related works e.g., project,
homework
29 Watching academic-related video lectures or those related to your job

Appendix Scoring. of the SONTUS

In scoring the SONTUS, five component scores are derived. The components scores are summed to produce a global score
that ranges from 5 to 23. This approach is in line with the results of our confirmatory factor analysis, which reveals a 5 first-
order factors with a 1 second-order factor as the best model for the SNOTUS construct.
Coding Instruction: each and every items in SONTUS is coded as follows:

1 = if a respondent select the Likert scale 1–3.


2 = if a respondent select the Likert scale 4– 6.
3 = if a respondent select the Likert scale 7–9.
4 = if a respondent select the Likert scale 10 or 11.

Component 1: relaxation and free periods.

Sum of items 2, 6, 7, 12, 14, 21, 22, 24 and 26 scores Component 1 score
9–12 1
13–16 2
17–20 3
21–24 4
25–28 5
29–32 6
>32 7

Component 2: academic-related periods

Sum of items 1, 5, 10, 13, 28, and 29 scores Component 2 score


6–9 1
10–13 2
14–17 3
18–21 4
>32 5

Component 3: public-places-related use.

Sum of items 4, 9, 17, 19, and 23 scores Component 3 score


5–8 1
9–12 2
13–16 3
17–20 4
Y. Olufadi / Telematics and Informatics 33 (2016) 452–471 469

Component 4: stress-related Periods.

Sum of items 3, 8, 15, 16, and 27 scores Component 4 score


5–8 1
9–12 2
13–16 3
17–20 4

Component 5: motives for use.

Sum of items 11, 18, 20, and 25 scores Component 5 score


4–7 1
8–11 2
>11 3

Global SONTUS score: sum of the five component scores: ________


Interpretation:

 An individual with a global score that ranges from 5 to 9 is regarded as low user of SNSs.
 An individual with a global score that ranges from 10 to 14 is regarded as average user of SNSs.
 An individual with a global score that ranges from 15 to 19 is regarded as high user of SNSs.
 An individual with a global score that is more than 19 is regarded as extremely high user of SNSs.

Note: A program is under preparation for easy scoring of the items in the SONTUS.

References

Andreassen, C.S., Torsheim, T., Brunborg, G.S., Pallesen, S., 2012. Development of a facebook addiction scale. Psychol. Rep. 110 (2), 501–517.
Benson, V., Saridakis, G., Tennakoon, T., 2014. Purpose of social networking use and victimisation: are there any differences between university students and
those not in HE? Comput. Hum. Behav. http://dx.doi.org/10.1016/j.chb.2014.11.034.
Bentler, P.M., 1990. Comparative fit indexes in structural models. Psychol. Bull. 107, 238–246.
Bentler, P.M., Bonett, D.G., 1980. Significance tests and goodness of fit in the analysis of covariance structures. Psychol. Bull. 88, 588–606.
Bollen, K.A., 1989. Structural Equations with Latent Variables. Wiley & Sons, New York, NY.
Browne, M.W., Cudeck, R., 1993. Alternative ways of assessing model fit. In: Bollen, K.A., Long, J.S. (Eds.), Testing Structural Models. Sage Publications,
Newbury Park, CA, pp. 136–162.
Byrne, B.M., 1989. A primer of LISREL: Basic Applications and Programming for Confirmatory Factor Analytic Models. Springer-Verlag, New York.
Carmines, E.G., Zeller, R.A., 1979. Reliability and Validity Assessment. Sage, Thousand Oaks, CA.
Cheek, J.M., 1983. The Revised Cheek and Buss Shyness scale. Wellesley College, Wellesley, Massachusetts, United States of America (unpublished
manuscript).
Chen, G.M., 2011. Tweet this: a uses and gratifications perspective on how active twitter use gratifies a need to connect with others. Comput. Hum. Behav.
27, 755–762.
Cheung, C.M.K., Chiu, P.-Y., Lee, M.K.O., 2011. Online social networks: Why do students use facebook? Comput. Hum. Behav. 27, 1337–1343.
CNNIC, 2011. Research Report on SNS Use by Internet Users in China in 2010. Retrieved from <http://www.cnnic.cn/hlwfzyj/hlwxzbg/201108/
P020120709345277353319.pdf> (accessed on 22.02.15).
CNNIC, 2013. Research Report on SNS Use by Internet Users in China in 2012. Retrieved from <http://www.cnnic.cn/hlwfzyj/hlwxzbg/mtbg/201302/
P020130219611651054576.pdf> (accessed on 22.02.15).
Cortina, J.M., 1993. What is coefficient alpha? an examination of theory and applications. J. Appl. Psychol. 78, 98–104.
Cramer, K.M., Barry, J.E., 1999. Conceptualizations and measures of loneliness: a comparison of subscales. Personality Individ. Differ. 27, 491–502.
Cronbach, L.J., 1951. Coefficient alpha and the internal structure of tests. Psychometrika 16, 297–334.
Cronbach, L.J., 1971. Test validation. In: Thorndike, R.L. (Ed.), Educational Measurement, Vol. 2. American Council on Education, Washington, DC, pp. 443–
507.
Crosby, P.B., 1979. Quality is Free. McGraw-Hill, New York.
Davis, F., 1989. Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Q. 13, 319–340.
Davis, F.D., Bagozzi, R.P., Warshaw, P.R., 1992. Extrinsic and intrinsic motivation to use computers in the workplace. J. Appl. Soc. Psychol. 22, 1111–1132.
DeVon, H.A., Block, M.E., Moyle-Wright, P., Ernst, D.M., Hayden, S.J., Lazzara, D.J., et al, 2007. A psychometric toolbox for testing validity and reliability. J.
Nurs. Scholarsh. 39 (2), 155–164.
DiTommaso, E., Brannen, C., Best, L.A., 2004. Measurement and validity characteristics of the short version of the social and emotional loneliness scale for
adults. Educ. Psychol. Measur. 64, 99–119.
Duggan, M., Ellison, N.B., Lampe, C., Lenhart, A., Madden, M., 2015. Social Media Update 2014. Pew Research Center, January 2015. Retrieved from <http://
www.pewinternet.org/2015/01/09/social-media-update-2014/> (accessed on 22.02.15).
Echeburúa, E., de Corral, P., 2010. Addiction to new technologies and to online social networking in young people: a new challenge. Adicciones 22, 91–95.
Ellison, N.B., Steinfield, C., Lampe, C., 2007. The benefits of facebook ‘‘friends:” social capital and college students’ use of online social network sites. J.
Comput.-Mediated Commun. 12 (4), 1143–1168.
Ellison, N.B., Steinfield, C., Lampe, C., 2011. Connection strategies: social capital implications of facebook-enabled communication practices. N. Media Soc.
Gilmore, H.L. (1974). Product Conformance Cost. Quality Progress, June, 16.
Gosling, S.D., Rentfrow, P.J., Swann Jr., W.B., 2003. A very brief measure of the big-five personality domains. J. Res. Pers. 37, 504–528.
470 Y. Olufadi / Telematics and Informatics 33 (2016) 452–471

Greenhow, C., Robelia, B., 2009. Old communication, new literacies: social network sites as social learning resources. J. Comput.-Mediated Commun. 14 (4),
1130–1161.
GreGory, P., GreGory, K., Eddy, E., 2014. The instructional network: using facebook to enhance undergraduate mathematics instruction. J. Comput. Math. Sci.
Teach. 33 (1), 5–26.
Grönross, C., 1983. Strategic Management and Marketing in the Service Sector, Marketing Science Institute, Cambridge (Ma.).
Griffiths, M.D., 2005. A ‘components’ model of addiction within a biopsychosocial framework. J. Subst. Use 10, 191–197.
Ha, Y.W., Kim, J., Libaque-Saenz, C.F., Chang, Y., Park, M.-C., 2015. Use and gratifications of mobile SNSs: facebook and KakaoTalk in Korea. Telematics
Inform. 32, 425–438.
Helms, J.E., Henze, K.T., Sass, T.L., Mifsud, V.A., 2006. Treating Cronbach’s alpha reliability coefficients as data in counseling research. Couns. Psychol. 34 (5),
630–660.
Helou, M.A., Ab Rahim, Z.Z., Oye, N.D., 2012. Students’ perceptions on social networking sites influence on academic performance. Int. Social Networking
Virtual Communities 1 (1), 7–15.
Higher Education Research Institute, 2007. How Many First-Year Are Using Social Networking Websites Such As Facebook and MySpace? Retrieved from
<http://www.gseis.ucla.edu/heri/PDFs/pubs/briefs/brief-091107-SocialNetworking.pdf> (14.12.14).
Hopko, D.R., Stowell, J., Jones, W.H., Armento, M.E.A., Cheek, J.M., 2005. Psychometric properties of the revised cheek and buss shyness scale. J. Pers. Assess.
84, 185–192.
Hu, L.T., Bentler, P.M., 1999. Cutoff criteria for fit indices in covariance structure analysis: conventional criteria vs. new alternatives. Struct. Equ. Modell. 25,
233–262.
Hughes, D.J., Rowe, M., Batey, M., Lee, A., 2012. A tale of two sites: twitter vs. facebook and the personality predictors of social media usage. Comput. Hum.
Behav. 28, 561–569.
Jacobsen, W., Forste, R., 2011. The wired generation: academic and social outcomes of electronic media use among university students. CyberPsychol. Behav.
Social Networking 14 (5), 275–280.
Ji, Y.G., Hwangbo, H., Yi, S.J., Rau, P.L.P., Fang, X., Ling, C., 2010. The influence of cultural differences on the use of social network services and the formation of
social capital. Int. J. Hum. Comput. Interact. 26 (11–12), 1100–1121.
Jo, J., 2013. KakaoTalk-Line, A Challenge to Facebook and Twitter. eToday. Retrieved from <http://www.etoday.co.kr/news/section/newsview.php?idxno=
772277> (accessed on 22.02.15).
Joinson, A.N., 2008. ‘Looking at’, ‘Looking up’ or ‘Keeping up with’ People? Motives and Uses of Facebook. In: Proceedings of the 26th Annual SIGCHI
Conference on Human Factors in Computing Systems (Florence, Italy, April 05–10 2008), CHI ’08. ACM, New York, pp. 1027–1036.
Junco, R., 2012a. Too much face and not enough books: the relationship between multiple indices of facebook use and academic performance. Comput.
Hum. Behav. 28 (1), 187–198.
Junco, R., 2012b. The relationship between frequency of facebook use, participation in facebook activities, and student engagement. Comput. Educ. 58 (1),
162–171.
Junco, R., 2013. Comparing actual and self-reported measures of facebook use. Comput. Hum. Behav. 29, 626–631.
Junco, R., Heiberger, G., Loken, E., 2011. The effect of twitter on college student engagement and grades. J. Comput. Assisted Learn. 27 (2), 119–132.
Juran, J.M., Gryna, F.M., Bingham, R.S. (Eds.), 1974. Quality Control Handbook, 3rd. ed. McGraw-Hill, New York.
KAA, 2012. 2012 KAA Media Research. Korea Advertisers Association. Retrieved from <http://www.kaa.or.kr/k/mag/2012/11_12/kaa1112_04.pdf> (accessed
on 22.02.15).
Kaiser, H.F., 1974. An index of factorial simplicity. Psychometrika 39, 31–36.
Kalpidou, M., Costin, D., Morris, J., 2011. The relationship between facebook and the well-being of undergraduate college students. CyberPsychol. Behav.
Social Networking 14 (4), 183–189.
Kang, Y.S., Lee, H., 2010. Understanding the role of an IT artifact in online service continuance. An extended perspective of user satisfaction. Comput. Hum.
Behav. 26, 353–364.
Katz, E., 1959. Mass communication research and the study of popular culture: an editorial note on a possible future for this journal. Stud. Public Commun.
2, 1–6.
Kim, H.W., Chan, H.C., Gupta, S., 2007. Value-based adoption of mobile internet: an empirical investigation. Decis. Support Syst. 43, 111–126.
Kirschner, P.A., Karpinski, A.C., 2010. Facebook and academic performance. Comput. Hum. Behav. 26, 1237–1245.
Kujath, C.L., 2011. Facebook and MySpace. Complement or substitute for face-to-face interaction? CyberPsychol. Behav. Social Networking 14 (1–2), 75–78.
Kuss, D., Griffiths, M., 2011. Online social networking and addiction – a review of the psychological literature. Int. J. Environ. Res. Public Health 8, 3528–
3552.
Kwon, O., Wen, Y., 2010. An empirical study of the factors affecting social network service use. Comput. Hum. Behav. 26, 254–263.
Lampe, C., Ellison, N., Steinfield, C., 2006. A face(book) in the crowd: social searching vs. social browsing. In: Proceedings of Anniversary Conference on
Computer Supported Cooperative Work. ACM Press, New York, pp. 167–170.
Levitt, T., 1972. Production line approach to service. Harvard Bus. Rev. 60 (5), 41–52.
Li, C., Bernoff, J., 2008. Groundswell: Winning In a World Transformed by Social Technologies. Harvard Business School Press, Boston, MA.
Li, X., Chen, W., 2014. Facebook or renren? a comparative study of social networking site use and social capital among Chinese international students in the
United States. Comput. Hum. Behav. 35, 116–123.
Lien, C.H., Cao, Y., 2014. Examining WeChat users’ motivations, trust, attitudes, and positive word-of-mouth: evidence from China. Comput. Hum. Behav. 41,
104–111.
Lin, C.P., Bhattacherjee, A., 2008. Elucidating individual intention to use interactive information technologies: the role of network externalities. Int. J.
Electron. Commerce 13, 85–108.
Lin, K.-Y., Lu, H.-P., 2011. Why people use social networking sites: an empirical study integrating network externalities and motivation theory. Comput.
Hum. Behav. 27, 1152–1161.
Lu, H.P., Su, Y.J.P., 2009. Factors affecting purchase intention on mobile shopping web sites. Internet Res. 19, 442–458.
Lubis, S.H., Ridzuan, S., Ishak, I.Y., Othman, H.F., Mohammed, N., Hamid, Z.A., Akmal, N.N., Farahana, N.N., Shanti, S., Zafirah, N., Hui, L.L., Izham, M., 2012. The
relationship between time spent on facebook and cumulative grade point average (CGPA) among third year biomedical science students in Faculty
Health Sciences, UKM. Proc. – Social Behav. Sci. 60, 590–595.
Lynn, M.R., 1986. Determination and quantification of content validity. Nurs. Res. 35, 382–385.
Marsh, H.W., Hocevar, D., 1985. Application of confirmatory factory analysis to the study of self-concept: first- and higher-order factor models and their
invariance across groups. Psychol. Bull. 97 (3), 562–582.
Mazman, S.G., Usluel, Y.K., 2010. Modeling educational usage of Facebook. Comput. Edu. 55, 444–453.
McKinney, B.C., Kelly, L., Duran, R.L., 2012. Narcissism or openness? college students’ use of facebook and twitter. Commun. Res. Rep. 29 (2), 108–118.
Memdouh, Saba., Taswir, Tarang., 2013. The effects of social networking sites on the academic performance of students in college of applied sciences, Nizwa,
Oman. Int. J. Arts Commerce 2 (1), 111–125.
Moon, J.W., Kim, Y.G., 2001. Extending the TAM for a world-wide-web context. Inform. Manage. 38, 217–230.
Mueller, R.O., 1994. Basic Principles of Structural Equation Modeling: An Introduction to LISREL and EQS.
Nadkarni, A., Hofmann, S.G., 2012. Why do people use facebook? Personality Individ. Differ. 52, 243–249.
Netemeyer, R.G., Bearden, W.O., Sharma, S., 2003. Scaling Procedures: Issues and Applications. Sage, Thousand Oaks, CA.
Nunnaly, J.C., Bernstein, I.H., 1994. Psychometric Theory, third ed. McGraw-Hill, New York.
Y. Olufadi / Telematics and Informatics 33 (2016) 452–471 471

Olufadi, Y., 2015. A configurational approach to the investigation of the multiple paths to success of students through mobile phone use behaviours.
Comput. Educ. 86, 84–104.
Panek, E.T., Nardis, Y., Konrath, S., 2013. Defining social networking sites and measuring their use: how narcissists differ in their use of facebook and twitter.
Comput. Hum. Behav. 29, 2004–2012.
Parasuraman, A., Zeithaml, V.A., Berry, L.L., 1985. A conceptual model of service quality and implications for future research. J. Marketing 49, 41–50.
Pempek, T.A., Yermolayeva, Y.A., Calvert, S.L., 2009. College students’ social networking experiences on facebook. J. Appl. Dev. Psychol. 30 (3), 227–238.
Pfeil, U., Arjan, R., Zaphiris, P., 2009. Age differences in online social networking: a study of user profiles and the social capital divide among teenagers and
older users in MySpace. Comput. Hum. Behav. 25, 643–654.
Pierce, T., 2009. Social anxiety and technology: face-to-face communication versus technological communication among teens. Comput. Hum. Behav. 25,
1367–1372.
Polsgrove, M.J., Frimming, R.E., 2013. A creative way to utilize social media to enhance fitness and health knowledge. Strategies 26 (2), 3–7.
ResonanceChina, 2013. Top 10 SNSs in China. Retrieved from <http://www.resonancechina.com/top-10-most-popular-china-social-media-networks/>
(23.02.15).
Rideout, V.J., Foehr, U.G., Roberts, D.F., 2010. Generation M2: Media in the Lives of 8–18 Year-Olds. Kaiser Family Foundation.
Ross, C., Orr, E.S., Sisic, M., Arseneault, J.M., Simmering, M.G., Orr, R.R., 2009. Personality and motivations associated with facebook use. Comput. Hum.
Behav. 25 (2), 578–586.
Ryan, T., Xenos, S., 2011. Who uses facebook? an investigation into the relationship between the big five, shyness, narcissism, loneliness, and facebook
usage. Comput. Hum. Behav. 27 (5), 1658–1664.
Ryff, C.D., Keyes, C.L.M., 1995. The structure of psychological well-being revisited. J. Pers. Soc. Psychol. 69, 719–727.
Schmitt, N., 1996. Uses and abuses of coefficient alpha. Psychol. Assess. 8 (4), 350–353.
Schultz, K.S., Whitney, D.J., 2005. Measurement Theory in Action. Sage, Thousand Oaks, CA.
Shaffer, H.J., LaPlante, D.A., LaBrie, R.A., Kidman, R.C., Donato, A.N., et al, 2004. Toward a syndrome model of addiction: multiple expressions, common
etiology. Harvest Rev. Psychiatry 12, 367–374.
Siemens, G., 2004. Connectivism: A Learning Theory for The Digital Age. Retrieved from <www.elearnspace.org/Articles/connectivism.htm> (accessed on
18.02.15).
Sledgianowski, D., Kulviwat, S., 2009. Using social network sites: the effects of playfulness, critical mass and trust in a hedonic context. J. Comput. Inform.
Syst. 49, 74–83.
Steiger, J.H., Lind, J.C., 1980. Statistically Based Tests for the Number of Common Factors. Annual Meeting of the Psychometric Society, Iowa City, IA, p. 758.
Tabachnick, B.G., Fidell, L.S., 2007. Using Multivariate Statistics. Pearson/Allyn & Bacon, Boston, MA.
Trochim, W.M.K., 2001. The Research Methods Knowledge Base, second ed. Atomic Dog Publishing, Cincinnati, OH.
Tucker, L.R., Lewis, C., 1973. The reliability coefficient for maximum likelihood factor analysis. Psychometrika 38, 1–10.
Tynes, B.M., 2007. Internet safety gone wild? sacrificing the educational and psychosocial benefits of online social environments. J. Adolesc. Res. 22 (6), 575–
584.
Valenzuela, S., Park, N., Kee, K.F., 2009. Is there social capital in a social network site?: facebook use and college students’ life satisfaction, trust, and
participation. J. Comput.-Mediated Commun. 14 (4), 875–901.
Valkenburg, P.M., Peter, J., Schouten, A.P., 2006. Friend networking sites and their relationship to adolescents well-being and social self-esteem.
CyberPsychol. Behav. 9, 584–590.
van der Heijden, H., 2004. User acceptance of hedonic information systems. MIS Q. 28, 695–704.
Voss, K.E., Stem Jr., D.E., Fotopoulos, S., 2000. A comment on the relationship between coefficient alpha and scale characteristics. Marketing Letters 11 (2),
177–191.
Walsh, W.B., 1995. Tests and Assessment. Prentice-Hall, New York.
Wang, J.-L., Jackson, L.A., Gaskin, J., Wang, H.-Z., 2014. The effects of social networking site (SNS) use on college students’ friendship and well-being. Comput.
Hum. Behav. 37, 229–236.
Widyanto, L., McMurren, M., 2004. The psychometric properties of the Internet addiction test. CyberPsychol. Behav. 7 (4), 445–453.
Xu, C., Ryan, S., Prybutok, V., Wen, C., 2012. It is not for fun: An examination of social network site usage. Inform. Manag. 49 (5), 210–217.
Young, K.S., 1998. Internet addiction: the emergence of a new clinical disorder. CyberPsychol. Behav. 1, 237–244.
Zigmond, A.S., Snaith, R.P., 1983. The hospital anxiety and depression scale. Acta Psychiatr. Scand. 67, 361–370.

Anda mungkin juga menyukai