Online Social Networking (SN) is rapidly growing and its popularity has ushered
in a flurry of new SN sites. However, due to the inordinate number of options
users have, many SN sites, including the famed Twitter, MySpace, Digg, Reddit,
have suffered from high abandonment rates (Kumar et al.,
2011). Given the rapid growth and proliferation of SN sites coupled with
an alarmingly high abandonment rate, understanding the factors that increase
user loyalty is critical for the survival and ultimate success of any SN site.
Despite the importance of user loyalty, there has been little systematic theorizing or empirical validation on how and why users develop a sense of loyalty in the SN context. Thus, this research develops and validates a model of antecedents of user loyalty in the SN context. Findings of this study will have direct implications for SN site owners in winning market share and creating sustainable competitive advantage. Recently, firms have also become increasingly interested in making significant investments in leveraging online communities for marketing, new product development and branding. As SN sites are one type of online community, understanding how and why users develop a sense of loyalty in the SN context could also provide profound implications for organizations initiatives in leveraging various types of online communities for competitive advantages.
Drawing on research in marketing (Chaudhuri and Holbrook,
2001), we define user loyalty in SN as users attachment towards a
social network site and their repeated usage behavior.
In the marketing literature, switching cost and satisfaction have been recurrently
identified as two important antecedents of loyalty. Thus, we expect them to
play a similar role in building user loyalty in the SN context. Despite the
importance of satisfaction in the formation of loyalty, researchers also recognize
that satisfaction does not always translate into loyalty and satisfying consumers
do switch between brands or products (Oliver, 1999).
Therefore, it is highly likely that loyalty can be developed through other mechanisms
(Oliver, 1999). While not widely studied in the marketing
literature, two other variables may also play a critical role in users
development of loyalty towards an SN site: trust and habit. In the SN environment,
various uncertainties exist, including SN sites opportunistic behavior
with the collection, disclosure and use of users personal information.
In the presences of such uncertainties and risks, trusting beliefs towards an
SN site could be an important mechanism through which user develop their loyalty.
Supporting this conjecture, IS researchers have repeatedly demonstrated that
trusting beliefs exert a significant effect on a variety of behavioral intentions
in uncertain environments (Pavlou, 2003). Additionally,
a SN site typically offers users with various services (photo and video sharing,
RSS, chat, etc.). As a user learns to successfully use those functions, s/he
tends to get better at them. The associated feeling of increasing competence
and confidence of using these functions may motivate the user to continue using
the same SN site. The action of using an SN site and the increasing confidence
of using it may reinforce each other, so much so that behavior of repeatedly
using the same SN site may guide itself in some automatic, habituated manner.
Once the habit of using a particular SN site is developed, it can prevent the
user from being distracted and from adopting other SN sites, leading to enhanced
loyalty towards the site. In the IS literature, habit has been found to be another
vital mechanism through which loyalty is developed (Lin and
Wang, 2006). Thus, we hypothesize:
||H1: The higher the switching cost, the more likely
a user will stay loyal towards the SN site
||H2: Users overall satisfaction towards an SN
site is positively related to their loyalty towards the site
||H3: Habit is positively related to user loyalty towards
an SN site
||H4: Users trusting beliefs towards an SN site
is positively related to their loyalty towards the site
EMPIRICAL ANALYSIS AND RESULTS
To test the above stated hypotheses empirically, data were collected via an electronic survey distributed to social networking website users. In total 304 surveys were distributed and 239 were successfully completed and returned yielding a response rate of 78.62%. The mean age of the respondents is 22.54 with the majority being females (66.9%) and rest being male (33.1%).
Non-response bias is a potential problem in survey research as it can threaten
the validity of the findings. In this study we adopt an approach advocated by
Jiang et al. (2002) who suggest using late respondents
as a proxy for non-respondents and comparing various characteristics of these
late respondents to those of early respondents. Accordingly, we divided our
sample into early and late respondents. Using a t-test, we compared these two
groups on age, network size, social network website visitation frequency and
the amount of time spent daily on social networking websites. The results revealed
no significant differences between the two groups implying that non-response
bias is not a major concerns in this study.
Common method variance is yet another potential issue in survey research. To assess common method variance Harman's single factor test was applied. The results revealed multiple factors and no one factor explained the majority of the variance, therefore common method bias is unlikely to be a concern in this study.
Partial Least Squares (PLS) regression was used to test the hypotheses. We
first tested the measurement model to ensure that the scales had adequate internally
consistent (Nunnally and Bernstein, 1994). Consequently,
we evaluated the Average Variance Extracted (AVE) and composite reliability
and Cronbachs alpha to ensure that they exceeded their minimum required
thresholds. All the scales exceeded their thresholds as illustrated in Table
1, suggesting adequate internal consistency.
We also assessed whether the scales demonstrated sufficient discriminant validity by checking whether the square root of AVE is greater than the correlations between latent variables. As illustrated in Table 1 the scales in this study demonstrated sufficient discriminant validity.
|| Measurement model
|CA: Cronbachs alpha, CR: Composite reliability, AVE:
Average variance extracted
|| Results of partial least squares analysis (structural model)
|***Significant at p<0.001
Three models were constructed to test the hypotheses:
||Model 1: A control variables only model
||Model 2: A control variables plus theoretical non-context
specific variables model
||Model 3: A control variables plus theoretical non-context
specific variables model plus theoretical context specific variables model
A bootstrapping procedure with 1000 subsamples was used to test the significance of the path coefficients. Paths were considered significant where p<0.05. Table 2 shows the results of the structural model analysis.
The results indicate that all the hypothesized paths are significant at the p<0.001 level. When the different models are compared the results also indicate a substantive change in R2 (36.8%) with a large effect size (f2 = 0.646) from Model 1 to Model 2 as well as a substantive change in R2 (8.3%) with a medium to large effect size (f2 = 0.170) from Model 2 to Model 3.
DISCUSSION AND CONCLUSION
The purpose of this study was to identify the factors affecting SN site user
loyalty. The results of our empirical study reveal that the factors identified
as antecedents of user loyalty to SN sites in this study do indeed explain a
significant portion (R2 = 51.3%) of the variance in user loyalty.
Specifically, Satisfaction and Switching Cost are found to have a positive impact
on SN site user loyalty. Perhaps the more interesting finding is that the twocontext
specific variables introduced in this study, Trusting Beliefs and Habit, also
have a positive impact on SN site loyalty. Moreover, the introduction of these
two variables not only has statistical significance, it also has practical importance
as reflected by the significant change in R2 and the medium to large
effect size (f2 = 0.17) when Trusting Beliefs and Habit are incorporated
into the model.
This study makes contributions to both theory and practice. From a theoretical
perspective the study is interesting and important because it not only provides
a systematic analysis of the antecedents of SN site loyalty, it also demonststrate
the significance of identifing and incorporating context specific variables
when examining loyalty in SN. Moreover, the study adds to the litereature by
operationalizing and empirically illustrating the pivital role played by Trusting
Beliefs and Habit in the SN context. From a practical perspective, the issue
of how to build and maintain user loyalty has been of great interest for many
SN site owners. The findings of this study suggests that increased focus on
the proposed four salient variables may substantially assist SN sites build
or maintain user loyalty. To improve customer satisfaction SN site owners should
ensure that they design their sites in such a way that are easy to use, give
users a sense of control over their usage of the site and provide users with
a perception of service value (Ba and Johansson, 2008).
The findings on the positive significant relationship between switching cost
and loyalty indicates that SN sites should invest time, money and effort into
developing features that permit users to personalize their SN site usage experience
in order to meet their idiosyncratic needs (Kim et al.,
2005). In so doing, the switching cost borne by users increases and consequently
loyalty towards the SN site would increase. This study found that Habit is yet
another strong predictor of user loyalty. Prior studies have suggested that
in order to nurture habitual use of a site it is essential for site owners to
implement any changes gradually and carefully (Kim et
al., 2005). Failure to do so could affect usage patterns and subsequent
loyalty to the site. Finally, it is also important for SN sites to build trust
by demonstrating that they are honest when dealing with users and issues related
to service. They should show concern about their users welfare and try
to understand how their actions will affect users (Beatson
et al., 2008).
Nevertheless, this study has a few limitations. First, the sample used in the study is relatively small compared to the population of SN site users. Second, the sample was relatively homogeneous and primarily consisted of respondents from the United States. Therefore, the influence of cultural on the relationship between each of the antecedents and loyalty was not considered. Finally, future studies may wish to consider the role of additional contextal variables such as users privacy concerns or the percieved benefits of using SN on loyalty.