International Journal of Business and Social Science
Vol. 5, No. 11(1); October 2014
Analyzing Factors Affecting Users’ Behavior Intention to Use Social Media: Twitter
Case
Ezgi Akar
Sona Mardikyan
Management Information Systems Department
Bogazici University Hisar Campus
Istanbul, Turkey
Abstract
Advancement of technology and the Internet proliferation have visible effects in the world. One of the most
important effects is the increased social media usage among the Internet users. For this purpose, factors having
impacts on users’ behavior intention to social media usage are investigated within the scope of the study. A
research model based on technology acceptance model 2 is proposed and revised by the taking social media
platforms into consideration. The effects of perceived ease of use, perceived usefulness, social influence,
facilitating conditions, playfulness, and trust are measured. Data collected from 462 respondents are analyzed by
structural equation modeling technique. According to results of the measurement and structural model validities,
a fit and acceptable model is achieved to measure the related effects of variables. The results of study reveal the
both direct and indirect positive impacts of the factors on users’ behavior intention to use social media.
Keywords: behavior intention, social media, structural equation modeling, twitter
1. Introduction
The importance and popularity of social media has risen with the technological developments and high Internet
penetration in the world. According to Global Digital Statistics report (2014), the population of the world nearly
7.1 billion and there are approximately 2.5 billion Internet users. About 1.9 billion of the Internet users are active
social media users. While the Internet penetration in the world is averagely 35%, social media is used averagely
26% in the world.
Social media refers to online platforms in where people share their own content such as photos, videos, music,
comments, and experiences etc. Popular online social media platforms are Facebook, Qzone, Google+, LinkedIn,
Twitter, and Tumblr. Global Digital Statistics report (2014) states that there are 1.814 million Facebook users,
632 million Qzone users, 300 million Google+ users, 259 million LinkedIn users, 232 million Twitter users and
230 million Tumblr users. These statistics validate the popularity of the social media platforms among the Internet
users. These statistics indicate that people communicate and collaborate through various social media platforms.
This study aims to focus on social media use by analyzing the impacts of relevant factors. In this manner, this
study finds answers to the question that why people use social media platforms frequently. Twitter is chosen as a
case study for analysis purpose. Twitter is released in 2006 as a microblogging platform in where usersshare their
opinions with texts including 140 characters. Additionally, they can add photos and videos to their tweets. A
research model based on technology acceptance model 2 is proposed. The effects of perceived ease of use,
perceived usefulness, social influence, playfulness, facilitation conditions, and trust are measured within the scope
of the study. In order to test the results, an online questionnaire is shared through social media platforms and data
are collected from 462 Twitter users. The proposed model is tested by structural equation modeling technique.
This study is divided into four parts. The first part covers the literature review. The second part deals with the
research model and methodology of the study. The third part includes the findings of the study. The last part is the
summary part which consists of the important and significant results of the study.
85
© Center for Promoting Ideas, USA
www.ijbssnet.com
2. Literature Review
It is obvious that social media has become a popular research field with the technological advancements and the
Internet proliferation. In this manner, many researchers focus on and conduct studies in social media from
different perspectives such as governmental and political, marketing, educational, and individual.
From the governmental and political perspective, both Twitter and Facebook have gained a popularity by
governments and adopted by various politicians in order to be in an effective and lasting communication with
citizens. Khan et al. (2014) define the governments as social governments in this new era and state that
governments should develop their social media strategies to become more efficient and effective in the social
media. In their study, Alam and Lucas (2011) conduct a study about Twitter use by Australian government by
analyzing the tweets of governmental agencies. They find that these agencies post tweets to share information
about news and updates about them or about external agencies. In another study, Alam et al. (2011) focus on the
usage of Facebook by the governments. They reveal that social media platforms are great opportunities for
governments to communicate and collaborate. Besides, Sobaci and Karkin (2013) analyze the use of Twitter by
Turkish mayors and find that Twitter is adopted and is widely used by mayors and so it improves the public
relations.
Social media is also widely studied from both students’ and educational institutions’ perspectives. Ivala and
Gachago (2012) measure the students’ engagement through Facebook and blogs and they find that these social
media platforms enhance the students’ performance. Lin et al. (2013) measure how students perceive Twitter as
an educational tool and they reveal that students are more interested in information sharing about the courses
through social media platforms. Prestridge (2014) also focus on students’ Twitter usage and indicates that Twitter
support engagement in learning. On the other hand, Palmer (2013) investigates the social media usage of
universities. Universities use social media in marketing, learning and teaching, student recruitment, alumni
communication, student services, and their libraries.
In the field of marketing, social media is a very popular topic. It is clear that social media platforms are a great
source of electronic word of mouth (Koo et al., 2011). Cheung et al. (2013) state that electronic word of mouth is
more important than traditional promotional and the Internet advertising tools. In this manner, Dlodlo and Dhulup
(2013) conduct a study on consumers to measure their social media usage. They propose a research model based
on technology acceptance model (TAM) and they try to test the effects of perceived enjoyment, perceived critical
mass, perceived usefulness, and perceived ease of use on intention to use social media and find the positive
effects of them except perceived usefulness.
Other than governmental and political, educational, and marketing perspectives, social media usage and analysis
of user behaviors are also one of the most popular topics in the academic world. Koçak and Oyman (2012) study
social media usage behaviors. They state that individuals use social media for watching videos, listening to music,
sharing photos, reading and writing comments, and sharing their own contents. They try to find that how users
behave in the social media. The results indicate that users prefer social media more to look at photos, listen to
music, and watch videos, read and follow webpages that they are interested in.
Hughes et al. (2012) analyze personality predictors of social media use. They try to investigate effects of
personality treats of big-five (neuroticism, extraversion, openness, agreeableness and conscientiousness),
sociability, and need for cognition. At the end, they find that personality is related to social media use. Moreover,
Ozguven and Mucan (2013) focus on five personality traits as in the study of Hughes et al. (2012). They
additionally measure the effects of income, age, gender, and life satisfaction. They find the significant effects of
conscientiousness, openness to experience, education, income level, and life satisfaction on social media use. In
addition, Rauniar et al. (2014) conduct a study on social media usage based on TAM. They select Facebook as a
case and try to find the effects of perceived ease of use, critical mass, capability, perceived playfulness, perceived
usefulness, and trustworthiness on social media use. They propose a revised TAM model by taking social media
into consideration. They validate the model and find the significant effects of the revised dimensions.
Scheepers et al. (2014) figure out the effects of sense of community which have four sub constructs (information
seeking behavior, hedonic behavior, sustaining strong ties, and extending weak ties) on social media use. Sense of
community refers to “a sense of affiliation and emotional connection, interaction and identification with a groups
of people”. They explain that these sub constructs identify the behaviors of social media users. Brooks (2013)
investigates both technological and personal characteristics on social media use.
86
International Journal of Business and Social Science
Vol. 5, No. 11(1); October 2014
He measures presenteeism and sociality as technological characteristics and information privacy sensitivity,
personal involvement, and cognitive absorption as personal characteristics. He reveals that these all characteristics
are highly associated with social media usage.
3. Research Model and Methodology
This part includes the proposed model, hypotheses, and the research methodology of the study.
3.1. Research Model
Many research studies that focus on technology acceptance are based on TAM. TAM was proposed by Davis in
1989 (Figure 1). The main dimensions of the model are perceived ease of use and perceived usefulness. Perceived
ease of use refers to “the degree to which a person believes that using a particular system would be free of effort”.
In addition, perceived usefulness is “the degree to which a person believes that using a particular system would
enhance his or her job performance”. Davis (1989) indicates that these two dimensions have effect on intention to
use a new technology.
Figure 1: Technology Acceptance Model (Davis, 1989)
After that, Venkatesh and Davis (2000) develop TAM2 model by adding social influence and cognitive processes
(Figure 2). While social influence processes include subjective norm, voluntariness, and image; cognitive
processes consist of job relevance, output quality, result demonstrability, and perceived ease of use. These all
processes have a significant effect on a new technology acceptance by users. In 1996, Venkatesh and Davis
developed a model to find the antecedents of perceived ease of use. They tested the effects of computer selfefficacy and object usability. Moreover, Venkatesh and Bala (2008) proposed TAM3 model including external
variables to test the antecedents of perceived ease of use. In other words, TAM3 model extends TAM2 model by
adding new dimensions which are computer self-efficacy, perception of external control, computer anxiety,
computer playfulness, perceived enjoyment, and objective usability. They find the significant effects of them on
perceived ease of use.
Figure 2: Technology Acceptance Model 2 (Venkatesh and Davis, 2000)
87
© Center for Promoting Ideas, USA
www.ijbssnet.com
Venkatesh et al. (2003) also develop a model called as unified theory of acceptance and use of technology model
(UTAUT). UTAUT includes eight dimensions: performance expectancy, effort expectancy, social influence,
facilitating conditions, voluntariness of use, experience, age, and gender. While performance expectancy is the
degree of that users believe that they will achieve high performance by using the new system, effort expectancy
refers to the degree of how much the using new system requires effort.
Figure 3 shows the proposed research model of the study and it includes the research hypotheses. The research
model that is based on TAM2, is revised for social media usage. Social influence, playfulness, facilitating
conditions, and trust are added as new constructs to the model. Davis (1989) says that perceived ease of use has a
direct effect on both perceived usefulness and behavior intention. In addition, perceived usefulness has an effect
on behavior intention. They are added as research hypotheses to the model:
H1: Perceived ease of use will have a positive effect on perceived usefulness.
H2: Perceived ease of use will have a positive effect on behavior intention.
H3: Perceived usefulness will have a positive effect on behavior intention.
Figure 3: Proposed Research Model of the Study and Hypotheses
Venkatesh et al. (2003), defines the social influence as “the degree to which an individual perceives that important
others believe he or she should use the new system.” They also state that social influence is related to subjective
norm in TAM2 and they are similar to each other. Venkatesh and Davis (2000) state that if a person suggests that
the system is useful, another person can believe that it is useful and have an intention to use the system.
H4: Social influence will have a positive effect on perceived usefulness.
Playfulness is an important dimension for social media platforms. Rauniar et al. (2013) state that if a person
enjoys them, he or she finds the service more useful. Moreover, they figure out that interactivity plays a key role
in social media platforms, especially for Facebook and Twitter.
H5: Playfulness will have a positive effect on perceived usefulness.
Facilitating conditions is added as other construct to the research model. Venkatesh et al. (2003) define the
facilitating conditions as “the degree to which an individual believes that an organizational and technical
infrastructure exists to support use of the system.” Facilitating conditions refer to that whether related social
media platform includes proper and enough instructions for users. Besides, the related social media platform
should have enough services and applications for its users. In this manner, facilitating conditions have effects on
perceived usefulness, playfulness, and trust. If facilitating conditions satisfy users, users become more
playfulness, trust the social platform, and believe that the social media platform enhances their performances.
H6: Facilitating conditions will have a positive effect on perceived usefulness.
H7: Facilitating conditions will have a positive effect on playfulness.
H8: Facilitating conditions will have a positive effect on trust.
Trust is one of the most important dimensions for especially online platforms. It is obvious that social media
platforms collect users’ information. Therefore, information confidentiality and not to misuse of information are
important for users. Users post tweets and share images and videos on Twitter.
88
International Journal of Business and Social Science
Vol. 5, No. 11(1); October 2014
Rauniar et al. (2013) explain that these activities are one of the examples of online behaviors of users. They
indicate that users should not be worry about privacy and safety concerns. Therefore, behavior intention to user
social media is influenced by users’ trustworthiness to the social media platform.
H9: Trust will have a positive effect on behavior intention.
3.2. Research Methodology
In the study, a questionnaire is prepared to collect data, measure the effects of each factor, and verify the research
hypotheses. The questionnaire includes nine questions about users’ Twitter usage behaviors. Descriptive
questions collected data about both users’ demographic information such as age, gender, education level, and their
follower and following numbers in Twitter. In addition, Twitter usage frequencies of users were gathered.
The last questions includes 26 sub items. It measures the dimensions having effects on users’ behavior intentions
to use social media. For each sub item, 7-point Likert-Scale is used. Appendix 1 shows the sub items of each
construct in the proposed research model. Some of the questions are adapted from the literature and revised for
social media. In addition, the questionnaire is designed in Turkish. For this purpose, English version of the
questionnaire is translated into Turkish and then Turkish version is translated into English in order to keep the
consistency among languages.
Data are collected using an online questionnaire service. Targeted sample is active Twitter users. Online
questionnaire is distributed through social media platforms such as Facebook and Twitter. As a result, 462 replies
are received. Hair et al. (2010) indicate that it requires minimum 150 sample size with a research model including
seven or less constructs, modest communalities, and no unidentified constructs for structural equation modeling
(SEM) technique.
4. Findings of the Study
This part of the study includes descriptive findings, results of the measurement and structural model validities.
4.1. Descriptive Statistics
Table 1 shows the descriptive findings of the study. According Table 1, While 60.2% of the respondents are
between 19 – 25 years old, there are 131 respondents who are 18 and less than 18 years old. The research sample
includes 394 males and 68 females. When the educational level is investigated, many Twitter users are either
university student or high school students. It can be concluded that Twitter is more popular among high school
and university students.
Moreover, most of the Twitter users have up to 199 followers and followings. 36.1% of the respondents have
between 0 and 99 followers and 34% of the respondents have between 100 and 199 followings. While followers
are the other Twitter users who follow the tweets of the user, followings are the users who are followed by the
user. Lastly, 44.2% of the respondents visit often Twitter and 25.8% spend about 3 – 5 hours in a day.
4.2. Confirmatory Factor Analysis
Confirmatory factor analysis (CFA) is developed by Karl Jöreskog in 1960s. CFA tests whether a set of items
defines a construct or not. Moreover, Swell Wright develop the path model with box and arrow diagrams (Wright,
1918, 1921, 1934, as cited in Blunch, 2007; Schumacker and Lomax, 2010). In fact, path model which tests more
complex relationships among factors, is the combination of correlation coefficients and regression analysis.
SEM that is developed by Karl Jöreskog, Ward Keesling and David Wiley (Jöreskog, 1969, 1973; Keesling 1972;
Wiley, 1973 as cited in Schumacker and Lomax, 2010), is the integration of path and confirmatory analysis. Their
model is known as JKW and becomes more popular with the development of linear structural relations model that
is also known as the first program to test SEM. In accordance with Hair et al. (2010) SEM have three different
characteristics when it is compared with other multivariate techniques. Firstly, SEM makes separate and
interconnected multiple regression equations simultaneously. Secondly, SEM measures unobserved variables
known as latent constructs by analyzing consistency among multiple indicators known as observed variables. The
last difference is that research theory including set of relationships and hypotheses are in a model.
SEM analysis has two parts: measurement and structural model validity. This part of the study includes the results
of CFA to test measurement model validity.
89
© Center for Promoting Ideas, USA
www.ijbssnet.com
These results are obtained by using AMOS 18.0 which is a software tool. Maximum likelihood estimation (MLE)
procedure is chosen to do CFA. MLE procedure finds the most likely estimates for the coefficients in an iterative
manner (Hair et al., 2010). Hair et al. (2010) state that factor loadings of the indicators should be at least 0.5 and
ideally 0.70 or greater. It is important that factor loadings should explain half of the variable even if at least 0.5
factor loadings are significant.
Table 1: Descriptive Statistics of the Study
In this manner, indicators of perceived usefulness (PU1), social influence (S3), behavioral intention (BI4), and
facilitating conditions (FC1) that have 0.58, 0.06, 0.60, and 0.61 factors loadings respectively, are deleted from
the model in order to represent the constructs well and increase the goodness of fit indices of the measurement
model. After that a new CFA is done. Table 2 shows the results of new CFA. It includes the name of the
constructs, indicators, and their factor loadings, square of factor loadings, measurement errors and p-values.
According to Table 2, all indicators are significant with 0.001 p-values. PE3, PU3, BI1, BI2, BI3, and FC2 have
0.66, 0.69, 0.67, 0.68, 0.65, and 0.68 respectively. These values are at least 0.50 and very close to 0.70. They do
not violate integrity of the constructs.
Table 2: Results of Confirmatory Factor Analysis
*not estimated when loading set to fixed value of 1.0
90
International Journal of Business and Social Science
Vol. 5, No. 11(1); October 2014
Table 3 shows goodness of fit indices of the measurement model. Chi-square of the model is 404.468 and degrees
of freedom is 188. In accordance with Engel, Moosbrugger and Müller (2003), Table 3 includes the acceptable fit
intervals and goodness of fit indices of the measurement model. All goodness of fit indices point out that the
measurement model is well-designed and indicators represent the related constructs well.
4.2.1. Construct Validity and Normality
In order to assess the construct validity, convergent and discriminant validities are investigated in the study. Hair
et al. (2010) and Bollen (1989) state that for construct validity, standardized factor loadings should be at least 0.50
and ideally 0.70 or greater. All the factor loadings in the measurement model are at least 0.50. In addition to factor
loadings, construct reliability should be at least 0.70 and average variance extracted should be at least 0.50 as a
rule of thumb.
Table 4 shows the construct reliabilities and average variance extracted results for each construct. All construct
reliabilities except for facilitating conditions are at least 0.70. On the other hand, construct reliability value of
facilitating conditions is relative to 0.70. Moreover, average variance extracted results for each construct except
behavior intention are greater than 0.50. The construct, behavior intention has 0.44 average variance extracted
which is close to 0.50 and does not violate the integrity of the measurement model. When these all values are
taken together and considered, they all support the convergent validity of the measurement model.
Table 3: Goodness of Fit Indices of the Measurement Model (Engel, Moosbrugger and Müller, 2003)
*good fit
Table 4: Construct Reliability and Average Variance Extracted Results
Venkatraman (1989) defines the discriminant validity that a measure does not highly correlated with another
measure. In order to establish discriminant validity, Hair et al. (2010) and Fornell and Larcker (1981) compare
AVE estimates for each factor with the squared correlations associated with that factor. When it is compared, all
AVE estimates in Table 5 are greater than the corresponding construct squared correlation estimates in Table 6.
While construct squared correlation estimates are shown above the diagonal, correlation estimates are shown
below the diagonal in Table 5. Results reveal that there are not any problem with discriminant validity for CFA
model. In addition to construct validity, normality is checked by Kolmogorov-Smirnov test. The results show that
all observed variables are normally distributed with 0,001 p-values.
91
© Center for Promoting Ideas, USA
www.ijbssnet.com
Table 5: Construct Correlation Matrix
Significance level: *** = 0.001
Note: Values below the diagonal are correlation estimates among constructs. Diagonal elements are
construct variances. Values above diagonal show the squared correlations.
4.3. Structural Model Validity
Structural model validity consists of hypotheses testing. The first results of SEM indicate that the first hypothesis
is not supported. In other words, the effect of perceived ease of use on perceived usefulness is not supported in the
model. Moreover, in order to achieve a better model, modification indices are examined. Modification indices
figure out that there are high correlations between social influence and facilitating conditions, and between
perceived ease of use and facilitating conditions. As a result, hypothesis 1 is removed from the structural model
and two correlations are added to the structural model to increase model fit. Figure 4 depicts the path diagram of
the final structural model.
Figure 4: Path Diagram of Structural Model
Table 6 includes regression and correlational weights. According to Table 6, all hypotheses are significant and all
p-values except hypothesis 9 are 0.001. These results indicate that hypotheses 2 3, 4, 5, 6, 7, 8, and 9 are
supported in the model. Besides, all correlations are significant, so it is concluded that variables social influence
and facilitating conditions, and perceived ease of use and facilitating conditions are positively correlated with
each other.
92
International Journal of Business and Social Science
Vol. 5, No. 11(1); October 2014
Table 6: Regression and Correlational Weights
Table 7 shows the total effects of exogenous constructs on endogenous constructs. Whereas, perceived usefulness,
behavior intention, trust, and playfulness are endogenous variables, social influence, perceived ease of use, and
facilitating conditions are exogenous variables in the structural model. According to Table 8, facilitating
conditions have both indirect (0,242) and direct (0,269) effects on perceived usefulness, so total effects of
facilitating conditions is 0,511. In addition, social influence has only direct effect on perceived usefulness. While,
facilitating conditions and social influence have only indirect effects on behavior intention, perceived ease of use
has direct effect on it. Lastly, facilitating conditions have only direct effects on both trust and playfulness.
Table 7: Total Effects of Exogenous Variables on Endogenous Variables
Table 10 shows the goodness of fit indices of the structural model. It includes the results of the proposed model
and acceptable fit intervals. According to Engel et al. (2003), GFI, RMSEA, Normed Chi-Square, NFI, CFI, and
AGFI values are in the acceptable fit intervals and all of them indicate that the structural model of the study is
well designed and acceptable. In addition, Chi-square of the model is 563,072 and degrees of the freedom is 199.
Table 8: Goodness of Fit Indices of the Structural Model (Engel, Moosbrugger and Müller, 2003)
5. Conclusion
The popularity of the social media is obvious among the Internet users. People share and express themselves
highly from these platforms. In this respect, a study is conducted to analyze the factors having impacts on users’
behavior intention to use social media. A research is done by taking the popular social media platform “Twitter”
as a case study. Proposed model of the study is based on TAM2 and it is expanded and revised by taking the
social media use into consideration. The proposed model includes seven constructs: perceived ease of use,
perceived usefulness, trust, playfulness, facilitating conditions, social influence, and behavior intention.
93
© Center for Promoting Ideas, USA
www.ijbssnet.com
These constructs are measured with an online questionnaire including 26 sub items. The questionnaire is
distributed through online channels and the data are collected from 462 respondents. SEM is applied as a research
technique to calculate the regression weights simultaneously. First of all, CFA is done to validate the
measurement model. According to results of CFA four indicators having less factor loadings are removed from
the measurement model. The new measurement results point out that the indicators represent the constructs well.
Furthermore, all goodness of fit indices are acceptable and so, the measurement model is valid and acceptable. In
addition, convergent validity and discriminant validity are measured. For this purpose, construct reliabilities and
average variance extracted values are calculated. All the values indicate a good model.
After that, hypotheses are tested within the structural model. Hypothesis 1 is not supported and removed from the
model. Besides, new correlations among constructs are added to model according to modification indices and
goodness of fit indices indicate that the structural model is valid. All the hypotheses and correlations are
significant.
The results figure out that behavior intent is positively affected by perceived ease of use, perceived usefulness,
and trust. It is obvious that the effect of perceived usefulness is greater than the two other constructs. It reveals
that if users believe that using social media platforms would enhance their own performances, they are more
intended to use these platforms. In addition, social influence, playfulness, and facilitating conditions have positive
impacts on perceived usefulness. These results indicate that if a user suggests that the social media is useful, users
enjoy it, and facilitating conditions satisfy their needs, then users can believe that social media is useful and have
more intention to use it. Lastly, it is concluded that facilitating conditions increase users’ trustworthiness to social
media platforms and their playfulness.
References
Alam, L., & Lucas, R. (2011, December). Tweeting Government: A case of Australian government use of
Twitter. In Dependable, Autonomic and Secure Computing (DASC), 2011 IEEE Ninth International
Conference on (pp. 995-1001). IEEE.
Alam, S. L., Campbell, J., & Lucas, R. (2011, December). Using Social Media in government: the Australian
taxation office e-tax Facebook page. In Dependable, Autonomic and Secure Computing (DASC), 2011
IEEE Ninth International Conference on (pp. 1002-1009). IEEE.
Blunch, N. (2008). Introduction to Structural Equation Modelling using SPPS and AMOS. Los Angeles, London:
Sage
Bollen, K.A. (1989). Structural Equations with Latent Variables, Wiley, New York, NY.
Brooks, S. L. (2013). Social Media Usage: Examination of Influencers and Effects (Doctoral dissertation,
WASHINGTON STATE UNIVERSITY).
Davis, F. D. (1989). Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information
Technology. MIS Quarterly, 13 (3), 319-340.
Davis, F. D., &Venkatesh, V. (2000). A Theoretical Extension of the Technology Acceptance Model: Four
Longitudinal Field Studies. Management Science, 46 (2), 186-204.
Davis, G. B., Davis, F. D., Morris, M. G., &Venkatesh, V. (2003). User Acceptance of Information Technology:
Toward a Unified View. MIS Quarterly, 27 (3), 425-478.
Dhurup, M., &Dlodlo, N. (2013). SELECTED SOCIAL MEDIA ANTECEDENTS: ATTITUDES TOWARDS
AND BEHAVIOURAL IMPACTS ON ITS USAGE AMONG CONSUMERS IN A DEVELOPING
COUNTRY. StudiaUniversitatis Babes Bolyai-Oeconomica, (3), 90-109.
Engel, KS.,Moosbrugger, H. & Müller, H. (2003). Evaluating the Fit of Structural Equation Models: Tests of
Significance and Descriptive Goodness-of-Fit Measures. Methods of Psychological Research Online, 8
(2): 23-74.
Fornell, C. and Larcker, D.F. (1981). Evaluating structural equation models with unobservable
variables and measurement errors. Journal of Marketing Research, Vol. 18 No. 1, pp. 39-50.
Hair, J., Black, W., Babin, B. & Anderson, R. (2010). Multivariate Data Analysis: A Global Perspective (7th ed.).
Upper Saddle River, NJ , London: Pearson Education.
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. Computers in Human Behavior, 28(2), 561-569.
94
International Journal of Business and Social Science
Vol. 5, No. 11(1); October 2014
Ivala, E., &Gachago, D. (2012). Social media for enhancing student engagement: the use of Facebook and blogs
at a university of technology. South African Journal of Higher Education, 26(1), 152-167.
Jöreskog, KG. (1969). A General Approach to Confirmatory Maximum Likelihood Factor Analysis.
Psychometrika, 34 (2), 183-202.
Jöreskog, KG. (1973). A General Method for Estimating a Linear Structural Equation System. Structural Equation
Models in the Social Sciences, 85-112.
Keesling, J. (1972). Maximum Likelihood Approaches to Causal Flow Analysis. Unpublished PhD Thesis,
University of Chicago.
Khan, G. F., Yoon, H. Y., Kim, J., & Park, H. W. (2014). From e-government to social government: Twitter use
by Korea's central government. Online Information Review, 38(1), 95-113.
KOÇAK, N. G., & OYMAN, M. Social Media Usage Behaviors of Individuals: An Application in Eskisehir.
International Journal of Business and Social Science, 3(22), 177-188.
Lin, M. F. G., Hoffman, E. S., &Borengasser, C. (2013). Is social media too social for class? A case study of
Twitter use. TechTrends, 57(2), 39-45.
Palmer, S. (2013). Characterization of the use of Twitter by Australian Universities. Journal of Higher Education
Policy and Management, 35(4), 333-344.
Prestride, S. (2014). A focus on students’ use of Twitter – their interactions with each other, content and interface.
Active Learning in Higher Education, 15 (2), 101-115.
Rauniar, R., Rawski, G., Yang, J., & Johnson, B. (2014). Technology acceptance model (TAM) and social media
usage: an empirical study on Facebook. Journal of Enterprise Information Management, 27(1), 6-30.
Scheepers, H., Scheepers, R., Stockdale, R., &Nurdin, N. (2014). The dependent variable in social media use.
Journal of Computer Information Systems, 54(2).
Schumacker, R. & Lomax, R. (2010). A Beginner's Guide to Structural Equation Modeling(3rd ed.). New York:
Routledge.
Sobaci, M. Z., &Karkin, N. (2013). The use of twitter by mayors in Turkey: Tweets for better public services?
Government Information Quarterly, 30(4), 417-425.
Venkatraman, N. (1989). Strategic orientation of business enterprises: the construct, dimensionality &
measurement. Management Science, 35 (8) , 942-962.
Venkatesh, V. and Davis, F.D. (1996). A model of the antecedents of perceived ease of use:
development of a test. Decision Sciences, 27 (3), 451-481.
Wiley, D. (1973). The Identification Problem for Structural Equation Models with Unmeasured Variables.
Structural Equation Models in the Social Sciences, 69-83.
Wright, S. (1918). On the Nature of Size Factors. Genetics , 3, 367-374.
Wright, S. (1921). Correlation and Causation. Journal of Agricultural Research, 20 (7), 557-585.
Wright, S. (1934). The Method of Path Coefficients. The Annals of Mathematical Statistics , 5 (3), 161-215.
95