To twitter or not to twitter?

To twitter or not to twitter?

What is driving consumer acceptance towards the use of Twitter?

2

To twitter or not to twitter?

What is driving consumer acceptance towards the use of Twitter?

By

Remko van Dijk

Groningen, June 2011

Master Thesis Marketing Management and Marketing Research Master of Business Administration

Faculty of Economics and Business University of Groningen

The Netherlands

Remko van Dijk R. Schuilingstraat 25

7815 BB Emmen Student number: s1808621 Email: remko_van_dijk@hotmail.com

Supervision University of Groningen, Department of Marketing Supervisor: Dr. S. (Sonja) Gensler

3

Management summary

4

Preface

This thesis is written to complete the Master of Science in Business Administration at the University of Groningen. After accomplishing my bachelor degree at Stenden University, I had no idea what I wanted to do. Therefore I decided to start the pre-Msc. at the University of Groningen. When I started studying at the University of Groningen I would have never imagined I would finish a double degree master program in Marketing and Marketing Research at the University of Groningen. Looking back at my period at the University of Groningen I am glad that I was able to push myself to this final result.

Of course I could not have written this thesis without the help of a number of people. First I would like to thank my supervisor dr. S. (Sonja) Gensler for her support, feedback and quick responding during writing my thesis. I also want to thank dr. ir. M.C. (Marjolein) Achterkamp for giving feedback and being co-assessor of my thesis.

Last but not least I would like to thank my parents for their support during my study period. Thanks to their support I have been able to finish this study.

Groningen,

5

Table of content

2.1 User acceptance of information technology ... 10

2.1.1 Theory of Reasoned Action and its development ... 11

2.1.2 Social Cognitive Theory... 14

2.1.3 Motivational Model... 15

2.1.4 Model of Perceived Characteristics of using an Innovation ... 16

2.1.5 Theory of Interpersonal Behavior ... 17

2.1.6 Unified Theory of Acceptance and Use of Technology (UTAUT) ... 19

2.1.7 Comparison of Models ... 20

2.1.8 Contribution to acceptance literature ... 24

2.1.9 Summary of literature review... 25

3. Formulation of the research model... 27

3.1 Proposed research model ... 27

3.2 Performance Expectancy ... 28 3.3 Effort Expectancy ... 30 3.4 Social Influence ... 31 3.5 Facilitating Conditions ... 34 3.6 Privacy Concerns ... 35 3.7 Behavioral Intention ... 37 4. Research design ... 38 4.1 Research method... 38

5. Results of empirical study... 41

5.1 Descriptive statistics ... 41

5.2 Factor Analysis ... 42

5.2.1 Deriving factors and assessing overall fit ... 43

5.2.2 Interpreting the factors ... 44

5.2.3 Reliability of factor analysis ... 44

5.3 Regression Analysis ... 45

5.3.1 Gender as moderator ... 49

5.3.2 Age as moderator ... 50

5.3.3 Internet experience as moderator ... 52

5.3.4 Personality as moderator ... 53

5.3.5 Usage as dependent variable ... 55

6. Summary and discussion ... 58

6

7.1 Recommendations ... 61

8. Limitations and directions for future research ... 62

Appendix A. Survey ... 67

Appendix B. KMO and Bartlett’s Test of Sphericity ... 69

Appendix C. Criterion for extraction of factors... 70

Appendix D. Output validity sample factor analysis ... 71

Appendix E. Normality test... 72

Appendix F. Normality test after log transformation ... 73

Appendix G. Assumptions moderating effect of gender ... 74

Appendix H. Assumptions moderating effect of age ... 75

Appendix I. Assumptions moderating effect of experience... 76

Appendix J. Assumptions moderating effect of personality... 77

Appendix K. Usage as dependent variable... 78

7

1.

Introduction

Micro blogging is a new form of communication which allow users to exchange small elements of content. Examples are describing current status, sharing information, or display video links. Micro blogging tools facilitate easily sharing status messages either publicly or within a social network (Java et al., 2007). A popular micro blogging tool is Twitter and has seen a lot of growth since it is launched in October 2006. Twitter.com is an online social network used by millions of people around the world. Eight months after its launch Twitter had about 94.000 users in April 2007 already (Java et al., 2007). Nowadays, it is estimated that Twitter has 190 million users, generating 65 million ‘tweets’ a day. ‘Tweets’ are short messages describing the current status of people within a limit of 140 characters. Topics range from new stories, current events, daily life, and other interests. These messages can be read by any other Twitter users who are declared by an individual user. These people are notified when that person has posted a new message. It allows people to stay connected to friends and family trough their computers and mobile phones. It is sometimes even described as the ‘SMS of the Internet’. This idea of sharing short messages using multiple access points (e.g. mobile phone, instant messaging, web) seems to be very appealing to people worldwide (Günther et al., 2009). Twitter is not only about exchanging messages, it has also the ability to spread news rapidly, thereby receiving lots of public attention. For example, in January 2009 an US Airways plane crashed on the Hudson river. Before it was actually in the news, it was already a hot item on Twitter. Because Twitter and similar micro blogging tools have become so popular, it is interesting to discuss the acceptance and utilization of Twitter.

1.1 Relevance

8 common users of Twitter. Also conversations, sharing information, and reporting news is Twitter most used for. Agrifoglio et al. (2010) also investigated the acceptance in a more individual setting. They divided the process of acceptance and use of Twitter by focusing on the role of intrinsic and extrinsic motivation in determining it. They find that perceived usefulness, enjoyment, and playfulness positively affect behavioral intention of using Twitter. Technology users feeling pleasure, joy, and fun are more likely to have a higher degree of intention to use Twitter according to Agrifoglio et al. (2010).

Besides literature about micro blogging for companies, there has been little research in the area of micro blogging phenomena. This is confirmed by Java et al. (2007) and Günther et al. (2009). They both state that to their knowledge there are no studies on micro blogging acceptance. It becomes clear in existing literature that there is a lack of empirical studies related to the acceptance of micro blogging. There is quite a lot of literature regarding acceptance of (internet) technologies. The Technology Acceptance Model seems to be used most in literature. Besides this model, there are several other models and theories that have been developed during the last years in the research of individual acceptance of technologies. Due to a lack of empirical studies related to the acceptance of micro blogging, a number of models related to technology acceptance will be reviewed with regard to their applicability to the research question. Information technology acceptance research has yielded many competing model, each with different sets of acceptance determinants (Venkatesh et al., 2003). Research of Venkatesh et al. (2003) has analyzed different acceptance models and integrated their results into one unified model. This model is called the Unified Theory of Acceptance and Use of Technology (UTAUT). This research will use a modified version of this UTAUT model in order to analyze which factors influence consumer’s acceptance of Twitter most. Factors derived from the original UTAUT model are performance expectancy, effort expectancy, social influence, and facilitating conditions. Other factors that are expected to influence intention to use Twitter are privacy concerns and personality. Building on insights from existing research about acceptance models, this research aims to formulate a model that seeks to predict the acceptance of Twitter. It is about the process before the decision is made to Twitter or not and has an focus on critical factors influencing the acceptance of Twitter. Interesting is to gain more insight in additional factors which helps to predict the intention to use Twitter, therefore the following research question is formulated:

9

1.2

Structure of thesis

10

2.

Literature review

This chapter contains an overview of the literature regarding models of consumer acceptance and determinants of acceptance towards the micro blogging tool Twitter. The first section describes the basic concept underlying user acceptance models. This is followed by the description of a number of models related to technology acceptance that are described in literature. This all leads to a comparison of models in which the findings are summarized and contributions to literature will be discussed.

2.1

User acceptance of information technology

Information Technology is a term that encompasses technologies in the field of telecommunication and computing. Information Technology plays an important role in our lives these days. Think about the role of computers and Internet in your own life. Micro blogging, or in this research Twitter, is an upcoming new technology. Information Technology research has long studied how and why individuals adopt new information technologies. A central focus in Management Information Research (MIS) has been to examine factors affecting user acceptance of computer technology. One stream of literature which is used in this research investigates the impact of these factors on the acceptance of information technology and has a focus on the individual acceptance. This stream of literature uses intention or usage as dependent variable (Venkatesh et al., 2003).

This research has a focus on the individual acceptance level of a new information technology, namely Twitter. A person’s acceptance of a technology is determined by his or her intention to use that technology (Yousafzai et al., 2010). Figure 1 gives an overview of a basic conceptual framework underlying the class of models explaining individual acceptance of information technology that forms the basis of this research.

11 A number of models related to technology acceptance are described in literature. The Technology Acceptance Model presented by Davis (1989) has been cited in most of the researches that deals with user acceptance of technology (Chuttur, 2009). Other theories discussed in literature are Theory of Reasoned Action developed by Fishbein and Ajzen in 1975, Theory of Planned Behavior (Ajzen, 1991), Social Cognitive Theory (Bandura, 1988), Model of Perceived Characteristics of using an Innovation (Moore and Benbesat, 1991), Motivational Model (Davis, 1992), Theory of Interpersonal Behavior (Thompson et al., 1991), and the Unified Theory of Acceptance and Use of Technology (UTAUT) model developed by Venkatesh et al. (2003). These models will be reviewed and compared. Based on this review, findings are summarized and contributions to literature will be discussed.

2.1.1 Theory of Reasoned Action and its development

The Theory of Reasoned Action (TRA) is a model developed by Fishbein and Ajzen in 1975. This model predicts consumer intentions and behavior and tries to identify where and how to target consumers’ behavioral change attempts. The model is displayed in figure 2.

Figure 2. Model of Theory of Reasoned Action by Fishbein and Ajzen (1975)

12 According to Bauer et al. (2005) overall acceptance can be predicted by measuring the attitude towards acceptance. The model of TRA is relevant to the development of a model for testing consumers’ acceptance. The TRA provides a useful model that could explain and predict the actual behavior of an individual (Yousafzai et al. 2010).

The TRA model is also the background of other models that predict consumer acceptance, namely the Theory of Planned Behavior and the Technology Acceptance Model. Davis (1989) used the TRA model and adapted it to the context of user acceptance of an information system. This model is called the Technology Acceptance Model (TAM). This model argues that acceptance of new technologies is predicted by the understanding of the users’ perceptions of the ease of use and usefulness of the new system (Chin et al. 2008). The TAM model suggests that users’ motivation can be explained by three factors: perceived ease, perceived usefulness, and attitude towards using. The model is specified as followed:

Figure 3. Technology Acceptance Model by Davis (1989)

13 Ajzen (1991) also used the Theory of Reasoned Action and extended the model by adding the factor of perceived behavioral control. This new model is called the Theory of Planned Behavior. It is represented in figure 4. This theory has been successfully applied to the understanding of individual acceptance and usage of many different technologies (Venkatesh et al., 2003).

Figure 4. Theory of Planned Behavior by Ajzen (1991)

14 Action in dealing with behavior in which people have incomplete control. This makes the Theory of Planned Behavior a useful tool to predict behavior of individuals.

2.1.2 Social Cognitive Theory

This theory is already mentioned when discussing the Theory of Planned Behavior. Social Cognitive Theory (SCT) has been developed of work in the area of social learning theory. SCT already has been proposed by Miller and Dollard in 1941. In short, they proposed that if an individual was motivated to learn a particular behavior, that behavior would be learned trough clear observations. This theory has been expanded by Bandura from 1962 until today. SCT is a widely accepted and empirically validated model of individual behavior (Compeau et al., 1995). According to this theory, watching others performing a behavior, influences the perception of the observer’s ability to perform the behavior, or self-efficacy, and the expected outcomes that they perceive. Bandura (1988) describes that this theory is a causal model in which cognitive and other personal factors, behavior, and influences from the environment all function as interacting determinants that influence each other. These determinants are not independent, they are reciprocal. Derived from SCT, Compeau et al. (1999) developed a model to test the influence of computer self-efficacy, outcome expectations, affect, and anxiety on computer usage. This is displayed in the research model of figure 5.

Figure 5. Model derived from Social Cognitive theory by Compeau et al. (1999)

The constructs in the model of figure 5 can be defined as followed:

15 • Outcome expectations, performance related; these outcomes are related to

improvements in job performance (efficiency) when using a computer.

• Outcome expectations, personal related; these outcomes are related to an individuals’ expectations of change in status or image, or to expectations of rewards (e.g. raises, or promotions).

• Affect; this is about the affective response of an individual toward using a computer. It represents the enjoyment a individual derive from using a computer.

• Anxiety; this is also about the affective response of an individual toward using a computer. It represents the feelings of unease, concerns, anxiety of using a computer. • Usage; this is about the level of use of computers at home and at work.

Research of Compeau et al. (1999) showed that SCT provides a strong confirmation that both self-efficacy and outcome expectations have an impact on an individual’s behavioral and affective reactions to information technology.

2.1.3 Motivational Model

16 2.1.4 Model of Perceived Characteristics of using an Innovation

A model that is going more into detail is the Model of Perceived Characteristics of using an Innovation (PCI). Perceptions of technology acceptance were initially based on the five characteristics of innovations derived by Rogers (1995) from the diffusion of innovations literature. According to Rogers (1995), diffusion is the process by which an innovation is communicated through certain channels over time among the members of a social system. An innovation can be defined as an idea, object, or practice that is perceived as new by an individual. Communication can be defined as a process by which participants create and share information with each other in order to reach mutual understanding. The idea of a diffusion process is the information exchange trough which individuals communicate new ideas to other individuals.

Moore and Benbasat (1991) adapted the characteristics of innovations of Rogers and refined a set of constructs in such a way that it could be used to study individual technology acceptance in Information Technology. Their model uses eight factors which provide a useful tool to study the initial individual adoption and diffusion of innovations. Difference between the research of Moore and Benbesat (1991) and Rogers (1995) is that Rogers’ construct definitions are based on perceptions of the innovation itself, and not on the perceptions of actually using an innovation. From this can be concluded whether an innovation diffuses or not is not about the potential adopters’ perception of an innovation itself, but rather about the perception of using an innovation. Moore and Benbesat (1991) their constructs can be defined as:

17 − Relative Advantage; the degree to which using an innovation is perceived as better

than using its forerunner.

− Compatibility; so the degree to which using an innovation is perceived as being consistent with the needs, existing values, and past experiences of potential adaptors. − Ease of Use; this construct used to be complexity in Rogers model. It can be defined

as the degree to which using an innovation is perceived as being difficult.

− Result Demonstrability; this construct focuses on the tangibility of the results of using an innovation. This used to be the observability construct of Rogers (1995)

− Image; Rogers (1995) included this construct in relative advantage. Moore and Benbesat (1991) found out that it should be used as a separate factor. It can be defined as the degree to which using an innovation is perceived as enhancing a individuals image or status within one’s social system.

− Visibility; can be defined as the degree to which an individual can see others using the system.

− Triability; the degree to which using an innovation may be experienced on a limited basis before adoption.

− Voluntariness; can be defined as the degree to which using an innovation is perceived as being of free will, or on a voluntary basis.

2.1.5 Theory of Interpersonal Behavior

18

Figure 7. Model of Thompson et al. (1991)

Thompson et al. (1991) defined the factors displayed in figure 7 as followed:

− Complexity of PC use; the more complex the innovation, the lower its rate of adoption. It is the degree to which an innovation is perceived as relatively difficult to understand and use.

− Job fit with PC use; whether the capabilities of a PC enhance an individual’s job performance.

− Long-term consequences of PC use; utilization of a PC should have a pay-off in the future.

− Affect towards PC use; the feelings of pleasure, joy, elation, or displeasure, hate, depression associated by an individual within a particular act.

− Social factors influencing PC use; the individuals internalization of a reference groups’ subjective culture and specific interpersonal agreements that the individual has made with others.

− Facilitating conditions for PC use; objective factors in the environment that several observers can agree make an act easy to do. For example, training in PC use.

19 conditions. The results show that social factors, complexity, long-term consequences, and job fit have a significant effect on PC use.

2.1.6 Unified Theory of Acceptance and Use of Technology (UTAUT)

Because there are many models in acceptance research, each with different sets of acceptance determinants, Venkatesh et al. (2003) have reviewed user acceptance literature and reviewed the previous discussed models. In their research, they empirically compare these models. Based on their research they formulated an unified model that integrates elements across the previous discussed acceptance models. This model represents a Unified Theory of Acceptance and Use of Technology (UTAUT). UTAUT has been empirically examined and is a good overall presentation of acceptance literature. This theory states that four constructs play a significant role as determinants of user acceptance and usage behavior. These four constructs are performance expectancy, effort expectancy, social influence, and facilitating conditions (Venkatesh et al., 2003). These constructs are based on conceptual and empirical similarities between the reviewed models. They also included four moderators in the UTAUT model, namely gender, age, experience, and voluntariness of use. This should lead to a better understanding of the complexity of technology acceptance by individuals. The UTAUT model is displayed in figure 8.

Figure 8. Unified Theory of Acceptance and Use of Technology by Venkatesh et al. (2003)

20 2003). Effort Expectancy can be defined as the degree of ease associated with the use of the system (Venkatesh et al., 2003). Social Influence is the degree to which an individual perceives important others think that he or she should use the new system (Venkatesh et al., 2003). Facilitating Conditions are defined as the degree to which an individual believes that there is a technical and organizational infrastructure that supports in using the system (Venkatesh et al., 2003). The UTAUT model is able to explain up to 70% of variance in predicting using intentions (Venkatesh et al., 2003), which makes it a very useful framework to measure technology acceptance.

2.1.7 Comparison of Models

After describing different models in the field of technology acceptance, these models are compared in this paragraph. What are similarities and dissimilarities?

The Theory of Reasoned Action (TRA) is a very general model, applicable in many areas. One of the comments in literature on this model is its generality. It does not specify the beliefs that are needed for a particular behavior. This means that the researcher first have to identify the beliefs that are salient for participants regarding the behavior that is investigated (Yousafzai et al., 2010).

A number of MIS specific models have been derived from TRA. The Theory of Planned Behavior (TPB) is one of them. Difference between TRA and TPB is that TPB accounts for conditions in which individuals do not have full control over the situation. TPB included behavioral control as direct determinant of intention. One comment that appeared a lot in literature was that TPB introduces only one new variable, while there is evidence that other factors also are likely to add predictive power. This is confirmed by Ajzen (1991) who developed the TPB. He states that the model is open for further expansion which might capture a significant proportion of the variance in intention.

The Technology Acceptance Model (TAM) is another extension of TRA. Difference between TRA and TAM is that TAM excluded subjective norm from the model. This exclusion of social variables is an important distinction between TRA, TPB, and TAM, but also for the model of Perceived Characteristics of using an Innovation (PCI). The original TAM model did not include social norms. According to Matthieson (1991) there could be social effects that are not directly linked to usefulness-related outcomes, and therefore should be included, like in TRA and TBP and the PCI.

21 control refers to skills, opportunities, and resources needed to use the system. TAM measures this by the construct ‘Ease of Use’. This relates to the respondents’ capabilities and skills required to use the system. The possession of these skills is important, but according to Ajzen (1985) there should be a distinction between internal and external factors. Internal factors are factors that are characteristics of the individual, and external factors depend on the situation. External factors relate more to time and opportunity. For example, the availability of a computer and Internet is important for someone to use a technology. Taylor and Todd (1995) tested TAM and TPB. Behavioral intention is the most important determinant in both models. TAM explains 52% of the variance in behavioral intention, TBP explains 57%. This indicates that the addition of perceived behavioral control and subjective norm provide more insight into behavioral intention.

Another model that shows a certain amount of parallelism with TAM is the Motivational Model. There is a close similarity between extrinsic motivation of the Motivational Model and perceived usefulness of TAM. In the UTAUT model Venkatesh et al. (2003) included both items in one construct, called performance expectancy. This in confirmed by literature (Davis, 1989) who tested both constructs and came to the conclusion that they are similar. Comment on the Motivational Model is that although there is empirical support for the Motivational Model, little is known about the underlying factors which influence extrinsic and intrinsic motivation (Venkatesh and Speier, 1999).

Noticeable when comparing the different models is that there is a difference in causal structures. The models of TAM, TPB and also PCI view the causal relationships in one direction, where the environment is influencing cognitive beliefs, and these beliefs influence attitudes and behaviors. The Social Cognitive Theory has a different view, there is a continuous interaction between the environment in which the individual is active in, his cognitive perceptions, and behavior (Compeau et al., 1999).

The PCI model provides insight in the impact of the innovation over time. It does not explicitly discuss user acceptance. It is about an individual’s perception of eight innovation attributes which are predictors for the individual’s acceptance of an innovation. Literature mention that the two TAM constructs are included in the constructs of the PCI. Ease of Use of TAM is equal to the complexity factor of PCI, and usefulness of TAM is more or less equal to relative advantage of PCI.

22 subjective norm should be explicitly considered in studying Diffusion of Innovations. An explanation for this is that subjective norms may overwhelm an individual’s attitude towards adopting an innovation.

The Theory of Interpersonal Behavior is according to Thompson et al. (1994) very complex. Some constructs are related in the sense that it may be difficult to develop the constructs in such a way that there are not too much measures.

It becomes clear that on one hand there are models that lack the comprehensiveness needed to consider them sufficient or complete, and on the other hand models are considered complex and not practical to apply. Therefore the UTAUT model was formulated. The UTAUT model is based on the core constructs of previous discussed models. The explanation of these constructs are displayed in table 1.

Construct Definition Core constructs Performance expectancy Performance Expectancy describes to

what extent an individual believes that using a new system will help him or her to attain benefits in job performance.

Perceived usefulness (Technology Acceptance Model), Extrinsic motivation (Motivational Model), Job-fit (Theory of Interpersonal Behavior), Relative advantage (Model of Perceived Characteristics of using an Innovation), Outcome expectations (Social Cognitive Theory)

Effort expectancy Effort Expectancy is defined as the degree of ease associated with the use of the system.

Perceived ease of use (Technology

Acceptance Model), Complexity (Theory of Interpersonal Behavior), Ease of use (Model of Perceived Characteristics of using an Innovation).

Social Influence Social Influence is the degree to which an individual perceives important others think that he or she should use the new system

Subjective norm (Theory of Reasoned Action, Theory of Planned Behavior), Social factors (Theory of Interpersonal Behavior), Image (Model of Perceived Characteristics of using an Innovation)

Facilitating conditions Facilitating Conditions are defined as the degree to which an individual believes that there is a technical and organizational infrastructure that supports in using the system

23 The constructs displayed in table 2 are the constructs which are mentioned in the discussed acceptance theories, but are not directly taken into account in the UTAUT model.

Table 2. Excluded constructs from UTAUT

Attitude toward using a technology, self-efficacy and anxiety are determinants of intention in several mentioned acceptance models. Attitude towards using a technology can be defined as an individual’s overall affective reaction to using a technology. Four constructs from different models closely align with this definition. Attitude towards using the technology (TRA and TAM), intrinsic motivation (Motivational Model), and affect toward use (Social Cognitive Theory). UTAUT does not include them as direct determinants of intention (Venkatesh et al., 2003). An explanation for this is that this construct is only significant when constructs related to performance and effort expectancy are not included in the model. Performance and effort expectancy are founded to be strong predictors of intention. This means that a relationship between attitude and intention results in omission of other key predictors. Therefore attitude is not included in the model. Self-efficacy and anxiety are also not modeled as direct determinants of intention. (Venkatesh et al., 2003). An explanation for this is that the effect is being captured by the existence of effort expectancy (Schaper and Pervan, 2004). This means that effort expectancy mediates the relation between self-efficacy and anxiety and intention. This makes self-efficacy and anxiety indirect effects of intention and are therefore not modeled. The other constructs in table 2 which are not modeled in the UTAUT model are

Core constructs Theory References

Attitude toward behavior Theory of Reasoned Action Fishbein and Ajzen (1975) Attitude towards behavior Technology Acceptance Model Davis (1989)

Result Demonstrability Visibility

Triability Voluntariness

Model of Perceived Characteristics of an Innovation

Rogers (1995) and

Moore and Benbesat (1991)

Attitude toward behavior Theory of Planned Behavior Ajzen (1991) Computer self-efficacy

Affect Anxiety Usage

Social Cognitive Theory Bandura (1962) and Compeau et al. (1999)

Intrinsic motivation Motivational Model Davis et al. (1992) Long-term consequences of PC use

Affect towards PC use

24 result demonstrability, visibility, triability, voluntariness, usage, and long-term consequence of PC use. These items were found to be non-significant, and therefore are excluded from the UTAUT model (Venkatesh et al., 2003).

2.1.8 Contribution to acceptance literature

25 The discussed models are originally developed to describe and explain organizational adoption of information technologies. Twitter is a more individual and personalized service, the acceptance is described and explained at an individual level. Why a focus on the individual acceptance level? Technology acceptance models have not paid sufficient attention to individual difference variables (Yi et al., 2006). This is confirmed by research of Wang et al. (2006), which mention that a general comment in acceptance literature is that it is necessary to examine the potential moderating effects of user technology acceptance. Most technology acceptance models did not include constructs related to individual differences. Individual differences can be defined as any dissimilarities between people, including differences in perceptions and behavior (Agarwal & Prasad, 1999). Individual differences refer to personal factors that include traits such as personality and demographic variables. The basic concept underlying the user acceptance model, places significant focus on individual reactions to new technology, in which personality can be expected to play a role (Devaraj et al., 2008). Moore (1987) also states that personal characteristics may have influence in adopting an innovation. Therefore personality is included as a moderator. One personality trait that is used a lot in research is an individual’s level of introversion and extraversion. Introversion and extraversion will be used to measure personality. Trait theorists propose that personality is composed of characteristics that describe and differentiate individuals. People have different behaviors, attitudes, beliefs, and cognitions. This is determined by their personality. Personality can be defined as the unique facets of each human being, the traits that define our essence, and it reflected in all of our thoughts and actions (Devaraj et al., 2008). Because traits play a big role in a persons’ cognition and behavior and differ per person, it is reasonable to expect that personality, divided into introversion and extraversion have an important role in the individual acceptance of Twitter.

2.1.9 Summary of literature review

3.

Formulation of the research model

In this part the research model will be described that will help to find out what drives the acceptance of using Twitter. The proposed research framework will be conceptualized based on research of Venkatesh et al. (2003).

3.1

Proposed research model

The proposed research model states that there are five constructs that play a significant role as determinant of user acceptance and usage behavior. These constructs are performance expectancy, effort expectancy, social influence, facilitating conditions, and as already mentioned in the previous chapter privacy concerns is included and expected to have an direct effect on intention in the proposed research model. Besides that four moderators are included. In the original UTAUT model, there is a distinction between voluntary and mandatory use of an information system. This is based on an organizational setting. Because this research has a focus on the individual acceptance level, there is no distinction between voluntary or mandatory use. Twitter is used on a voluntary basis, and therefore this moderator is left out the research model. Besides that, personality, measured by introversion and extraversion, is included as a moderator to measure individual differences. The other moderators are gender, age, and Internet experience. These modifications are taken into account when formulating the research model which should lead to a better understanding in explaining intention. This model seeks to predict the intention to use Twitter. This research model is displayed in figure 10 and will be discussed in this chapter.

28

3.2

Performance Expectancy

Performance expectancy is expected to be the strongest predictor of intention (Venkatesh et al., 2003). In the original UTAUT model, performance expectancy describes to what extent an individual believes that using a new system will help him or her to attain benefits in job performance. This construct is developed based on five constructs from different models. These constructs are perceived usefulness (Technology Acceptance Model), extrinsic motivation (Motivational Model), job-fit (Theory of Interpersonal Behavior), relative advantage (Model of Perceived Characteristics of using an Innovation), and outcome expectations (Social Cognitive Theory). As mentioned earlier, this research is about acceptance at an individual level, and not at an organizational level. At the individual level job performance it is more about interpersonal communication. From this can be derived that the job that Twitter should be performing is about interpersonal communication. This means that using Twitter would help to improve interpersonal communication. From this perspective it is interesting to investigate to what extent individuals belief that Twitter enhances communication with other individuals. Therefore hypothesized:

H1a: Performance expectancy positively influences behavioral intention to use Twitter.

The original UTAUT model states that age and gender have a moderating effect on the relation between performance expectancy and behavioral intention. In addition to that, the proposed model expects that personality also moderates this relation.

29 usefulness is related to performance expectancy, so it is reasonable to expect that this will be a stronger determinant for men than for women.

H1b: The positive influence of performance expectancy on behavioral intention to use Twitter will be moderated by gender, such that the effect will be stronger for men.

Also age is expected to play a moderating role. It is likely to assume that younger people are the most frequent users of the Internet. This is confirmed by a study of Thayer and Ray (2006). This research states that men and women between the ages of 20 to 30 are the most frequent users of the Internet. Consumers in this age group spend their time online chatting, emailing, meeting new people and play online games. Younger people also spend more time communicating online and building online relationships with friends and unknown consumers more than middle age and late age people (Thayer and Ray, 2006). These consumers in the middle age and late age group are relative less adaptive to changes in communication and relationship building that Internet provides (Thayer and Ray, 2006). There is a positive relationship between age and computer anxiety (Harrison & Rainer, 1992). This suggests that performance expectancy is a stronger determinant for younger people than for older people. This leads to the following hypothesis:

H1c: The positive influence of performance expectancy on behavioral intention to use Twitter will be positively moderated by age, such that the effect will be stronger for younger people.

30 H1d: The positive influence of performance expectancy on behavioral intention to use Twitter will be positively moderated by personality, such that the effect will be stronger for extravert people.

Internet experience is not used as a moderator for the relation between performance expectancy and behavioral intention. There is no reason to expect why more Internet experience would contribute to improving interpersonal communication by using Twitter. This is confirmed by the research of Venkatesh et al. (2003) which also not mentioned experience as a moderator between performance expectancy and behavioral intention.

3.3

Effort Expectancy

In the UTAUT model, effort expectancy is defined as the degree of ease associated with the use of the system (Venkatesh et al., 2003). In this research that would mean the ease of using Twitter. This construct is developed based on three constructs from different models that have a substantial similarity. These constructs are perceived ease of use (Technology Acceptance Model), complexity (Theory of Interpersonal Behavior), and ease of use (Model of Perceived Characteristics of using an Innovation). Expected is that the easier it is to use Twitter, the faster it is accepted and used by individuals, therefore hypothesized:

H2a: Effort expectancy positively influences behavioral intention to use Twitter.

The original UTAUT model states that age, gender, and experience have a moderating effect on the relation of effort expectancy and behavioral Intention. This is also expected in this research. Women typically display lower computer competences, and higher computer anxiety, compared to men (Venkatesh and Morris, 2000). According to Yi et al. (2006) there is a negative correlation between computer anxiety and computer self-efficacy. This indicates that, especially for woman, higher levels of computer anxiety lead to a lower level of self-efficacy, which can cause lower ease of use perceptions for Twitter. This suggests that effort expectancy is a more important determinant of intention for women compared to men.

31 As mentioned earlier, older consumers are relative less adaptive to changes in communication and relationship building that Internet provides. For this group of consumers Twitter might be difficult and complex to understand. Their acceptance might partly depend on the ease of use of Twitter. This would mean that younger people would accept and use Twitter earlier. This suggests that effort expectancy is more important for older people.

H2c: The positive influence of effort expectancy on behavioral intention to use Twitter will be negatively moderated by age, such that the effect will be stronger for older people.

Individuals with more computer and Internet experience will perceive Twitter easier to use. More experience with computers and Internet is associated with higher computer skills (Yi et al., 2006). It is also expected that individuals with much experience are more likely to explore new and useful functions offered via the computer (Yi et al., 2006). So for people with less experience, effort expectancy would be a more important determinant of intention than people with more experience. Therefore hypothesized:

H2d: The positive influence of effort expectancy on behavioral intention to use Twitter will be negatively moderated by experience, such that the effect will be stronger at early stages of Internet experience.

Extravert people tend to be more optimistic, enthusiastic and talkative compared to introvert people. Extravert people are more likely to participate in social channels compared to introvert people. Twitter might therefore be a way for extravert people to assert themselves more than introvert people would do. The easier it is to use Twitter, the more likely it is to expect that extravert people will join Twitter, therefore hypothesized:

H2e: The positive influence of effort expectancy on behavioral intention to use Twitter will be negatively moderated by personality, such that the effect will be stronger for introvert people.

3.4

Social Influence

32 this research it means the degree to which an individual perceives important others think that he or she should use Twitter. This construct is developed based on three constructs from different models that have a substantial similarity. These constructs are subjective norm (Theory of Reasoned Action, Technology Acceptance Model, Theory of Planned Behavior), social factors (Theory of Interpersonal Behavior), and image (Model of Perceived Characteristics of using an Innovation). A person’s perception of social influences is called subjective norm. Subjective norms can be seen as normative beliefs. This is the likelihood that important referent individuals or groups would approve or disapprove performing any behavior (Yousafzai et al. 2010). Studies have found normative beliefs to be an important determinant of intention (Yousafzai et al. 2010). Individuals’ perceptions of norms consist of informational and normative influences. Informational social influence can be defined as the influence to accept information obtained from another as evidence about the true state of some aspect of the individual’s environment (Burnkrant & Cousineau, 1975). Normative social influence can be defined as the influence to conform to the expectations of another person or group (Burnkrant & Cousineau, 1975). Many studies verified the effect of norms on intention to use Internet-related services (Hsu and Lin, 2008). Therefore the effect of social influences should not be ignored in the acceptance of Twitter. Therefore hypothesized:

H3a: Social influence positively influences behavioral intention to use Twitter.

The original UTAUT model states that age, gender, and experience moderate the relation of social influence and behavioral intention. In the proposed model this is extended with personality as moderator.

33 H3b: The positive influence of social influence on behavioral intention to use Twitter will

be moderated by gender, such that the effect will be stronger for women.

Literature suggests (Venkatesh et al., 2003) that older people are more likely to focus on the importance of social influences. Older people are expected to have less experience with new technologies. They will try to get notice of the new technology by attaining reactions from important referents. This implies that social influences would have the largest impact on older people.

H3c: The positive influence of social influence on behavioral intention to use Twitter will be positively moderated by age, such that the effect will be stronger for older people.

Individuals with experience with computer technology can better assess the benefits and costs associated with the use of Twitter. Individuals with more computer and Internet experience are less likely to be influenced by others, because these individuals start to rely more on their own experiences (Venkatesh & Morris, 2000) This indicates that people with less experience are more sensitive to social influences, therefore hypothesized:

H3d: The positive influence of social influence on behavioral intention to use Twitter will be positively moderated by experience, such that the effect will be stronger at early stages of Internet experience.

34 H3e: The positive influence of social influence on behavioral intention to use Twitter will be positively moderated by personality, such that the effect will be stronger for extravert people.

3.5

Facilitating Conditions

In the original UTAUT model facilitating conditions are defined as the degree to which an individual believes that there is a technical and organizational infrastructure that supports in using the system (Venkatesh et al., 2003). This construct is developed based on three constructs from different models that have a substantial similarity. These constructs are perceived behavioral control (Theory of Planned Behavior), facilitating conditions (Theory of Interpersonal Behavior), and compatibility (Model of Perceived Characteristics of using an Innovation). In the original UTAUT model, there is no relation between facilitating conditions and Behavioral Intention. According to Venkatesh et al. (2003) this can be explained by when both performance expectancy and effort expectancy constructs are present, facilitating conditions become non-significant in predicting intention. In the original UTAUT model facilitating conditions are described in organizational context. This would mean helpdesks, assistance etc. In the individual setting of acceptance of Twitter, facilitating conditions would mean that an individual should have the access to a computer and the Internet. Before Twitter can be used, an individual should have access to a computer and to the Internet, therefore hypothesized:

H4a: Facilitating conditions positively influences the usage of Twitter.

The UTAUT model describes age and experience as moderating effects on the effect that facilitating conditions will have on usage behavior (Venkatesh et al., 2003). This research also expect these effects. As discussed earlier, younger people are the most frequent users of the Internet. The more complex information technology becomes, the more difficult it become for older people to understand. So there are cognitive and physical limitations associated with age.

35 Individuals who own a computer are more motivated to familiarize themselves with computers and new technologies related to computers than individuals who do not own a computer (Harrison & Rainer, 1992). An increase of a users’ experience with a technology like the computer or Internet would mean that they find it easier to accept and use new technologies where these facilitating conditions are needed for, therefore hypothesized:

H4c: The positive influence of facilitating conditions on usage of Twitter will be positively moderated by Internet experience, such that the effect will be stronger for people with more Internet experience.

As mentioned earlier, facilitating conditions is defined as whether an individual has the access to a computer and the Internet to use Twitter. There are no differences expected between men and women, because both have equal opportunities to make use of computers and Internet. Venkatesh et al. (2003) also did not reported gender as a moderator between facilitating conditions and usage. Therefore this relation is not investigated. Also personality is not expected to be a moderator between facilitating conditions and usage. In an organizational context, facilitating conditions can be described as helpdesks, assistance etc. Extravert people are more open for assistance and opinion from others, therefore personality might have an effect in an organizational context. In the individual context, personality is not expected to have an effect because it is expected to have no influence on whether someone has access to a computer or Internet. Therefore this relation is also not investigated.

3.6

Privacy Concerns

As discussed earlier in none of the acceptance models privacy concerns is considered. Nowadays with an increasing exchange of electronic information, consumers are more concerned about their personal information available at the Internet. Research of Hoffman et al. (1999) found that consumer expectations of privacy depends on the medium. In electronic media, like Twitter, consumers have an intense need for control and protection. For example, many consumers want to have full control over demographic information websites capture. It becomes clear that privacy on the Internet has a negative influence on the consumer willingness to engage in relationship exchanges online. Therefore hypothesized:

36 It is expected that privacy concerns will influence intention in a negative way. It is expected that this relationship is moderated by age, gender, experience, and personality.

According to Sheehan (1999) more men than women say that they do not care whether a website maintains a customer profile about them. In the research of Philip and Suri (2004) differences have been observed in male and female web users’ online usage behaviors. There appeared to be gender differences in online communication and behavior. This is confirmed in the study of Sheehan (1999) where they indicated that women are more concerned about their personal privacy and information compared to men in information gathering situations online. These findings indicate that there is a gender difference in privacy concerns and that women may be more concerned about their privacy online than men.

H5b: The negative influence of privacy concerns on behavioral intention to use Twitter will be moderated by gender, such that the effect will be stronger for women.

As mentioned earlier, younger people are the most frequent users of the Internet. Men and women between the ages of 20 to 30 are the most frequent users of the Internet. Older people are less willing to accept changes in communication and relationship building that Internet offers. Privacy concerns appear to decrease when online users become more experienced with Internet. So individuals that are more familiar with Internet and how information is collected online would have less privacy concerns. Because younger people are most frequent Internet users, it is expected that privacy concerns are for them less relevant as for older people.

H5c: The negative influence of privacy concerns on behavioral intention to use Twitter will be negatively moderated by age, such that the effect will be stronger for older people.

37 Sheehan (2002) states that concerns with privacy even appears to decrease as online users become more experienced with the Internet. As an individual becomes more familiar with Internet and how information is collected online, privacy concerns decrease. Experience is expected to moderate the relation of privacy concerns in the relation with behavioral intention. This all leads to the following hypothesis:

H5d: The negative influence of privacy concerns on behavioral intention to use Twitter will be negatively moderated by Internet experience, such that the effect will be stronger for people in the early stages of Internet experience.

People who are introvert are interacting less with their environment than extravert people. Research showed that people who are introvert are more anxious, depressed, cynic, and generally more vulnerable. Introvert people have a stronger urge for anonymity and perceive higher intrusions of privacy (Junglas et al., 2008). As discussed earlier, introvert people are more shy, prefer to be alone, and are anxious in the presence of others. They attach more value to privacy. They tend to avoid social channels and may not find out about new products from others (Hoyer & MacInnis, 2008). They are also less motivated by social pressure and more likely to do things that please themselves. Therefore hypothesized:

H5e: The negative influence of privacy concerns on behavioral intention to use Twitter will be negatively moderated by personality, such that the effect will be stronger for introvert people.

3.7

Behavioral Intention

All the models that describe intention expect that behavioral intention will have a positive significant effect on the usage of Twitter (Venkatesh et al., 2003). This means that an individuals’ acceptance of Twitter is determined by his or her behavioral intention to use Twitter.

38

4.

Research design

This chapter describes the choice of research and the data collection methods used in this research. This chapter ends with a description of the analysis methods used to analyze the data.

4.1

Research method

This research will be a descriptive research to determine which factors drive consumer acceptance towards the use of Twitter. It is a single cross-sectional design, because it involves the collection of information from any given sample of population elements only once (Malhotra, 2007). An online survey will be used in order to collect the data for this research. The survey is developed to create insight in the constructs in order to test the hypotheses. The constructs personality, privacy concerns, performance expectancy, effort expectancy, social influence, and facilitating conditions are measured as displayed in table 3.

Construct Item Source Scale

Personality − _{I am the life of the party}

− I enjoy being in the center of attention

− I am skilled in handling social situations

− I like to be where the action is

− I make new friends easily

− I am quiet around strangers

− I don’t like to draw attention to myself

− I don’t like to party on the weekends

− I like to work independently

− I often enjoy spending time by myself

Hoyer & MacInnis, 2008

1 strongly disagree, 7 strongly agree

Privacy Concerns − _{To me, it is the most important thing} to keep my privacy intact when I am online.

− I am concerned about threats to my personal privacy when I am online.

− I am concerned that the information I submit on the Internet could be misused.

− Being able to control the personal information I provide to a website is important to me

Malhotra et al., 2004 and Dinev & Hart, 2004

1 strongly disagree, 7 strongly agree

Performance expectancy

− I would find Twitter useful for my interpersonal communication

− When using Twitter, it enables me to communicate online with others more quickly

− When using Twitter my productivity of communication increases

− If I use Twitter, I will increase my chances of getting a social boost

Venkatesh et al. (2003)

39

Effort Expectancy − _{My interaction with Twitter would be} clear and understandable

− It would be easy for me to become skillful at using Twitter

− I would find Twitter easy to use

− Learning to operate Twitter would be easy for me

Venkatesh et al. (2003)

1 strongly disagree, 7 strongly agree

Social Influence − _{People who influence my behavior} think that I should use Twitter

− People who are important to me think that I should use Twitter

− I use Twitter because of the

proportion of friends who use Twitter

− In general, my social environment has supported the use of Twitter

Venkatesh et al. (2003) 1 strongly disagree, 7 strongly agree Facilitating conditions

− I have the resources necessary to use Twitter

− I have the knowledge necessary to use Twitter

− Twitter is not compatible with other systems online systems I use

− A specific person (or group) is available for assistance with Twitter difficulties. Venkatesh et al. (2003) 1 strongly disagree, 7 strongly agree Behavioral Intentions

− I intent to use Twitter in the next month

− I predict I would use Twitter in the next month

− I plan to use Twitter in the next months

Venkatesh et al. (2003)

1 strongly disagree, 7 strongly agree

Table 3. Overview of items

Gender is measured by male/female, age is measured in categories ranging from 18-24, 25-34 etc. Internet experience is measured on a 5 item scale ranging from highly inexperienced to highly experienced. Usage is measured whether people are using Twitter currently or not. The complete survey is in appendix A. Besides the questions about age, gender and usage, all variables are ordinal scaled. These questions are answered on a seven point scale varying from disagree strongly to agree strongly. It allows one to determine whether a object has more or less of a characteristic than some other object, but not how much more or less (Malhotra, 2007). Age, gender, and usage are scale variables. There is also a question in the survey which is a reverse question. It is an item of facilitating conditions, namely Twitter is not compatible with other online systems I use. This item need to be recoded. This is done in SPSS where this variable is transformed and recoded into the same variable (1=7, 2=6 etc.)

40 among people who do not use Twitter yet. This survey has been send by mail to random selected people whether they would participate in this research and whether they would forward this mail to other people. Besides that it is distributed via other social media sites, like Facebook and Hyves. A total of 200 respondents is expected to be necessary for this research, which are randomly selected to participate in the online questionnaire.

41

5.

Results of empirical study

This chapter will discuss the analysis and the results of different statistical tests. First the data is checked for missing values. This is followed by some descriptive statistics to gain insight into the data. After that a factor analysis is conducted to test the constructs of the research model. To test the hypotheses several regressions are conducted.

5.1

Descriptive statistics

Before the data will be used for descriptive statistics, factor analysis, and regression analysis the data is checked for missing values. A total of 179 respondents participated in this research. After checking the data it appeared that 6 respondents did not completed their survey. These respondents are deleted from the dataset. A total of 173 respondents is used for data analysis.

Descriptive statistics and cross-tabs are used to gain more insight in the data. In this research a total of 173 respondents participated. Table 4 displays that 44.5% of these respondents are male and 55.5% are female.

Percentage Male 44.5 Female 55.5

Table 4. Gender

The largest group of respondents is between 18 and 24 (44.5 %), followed by 24.3% between 25 and 34. The other age categories are displayed in table 5.

Percentage 18-24 44.5 25-34 24.3 35-44 5.8 45-54 12.7 55-64 11.6 65+ 1.2 Table 5. Age

42

Percentage Highly inexperienced 2.9 Somewhat inexperienced 1.2 Neither experienced, nor inexperienced 17.9 Somewhat experienced 55.5 Highly experienced 22.5 Table 6. Experience

Table 7 displays whether an individual is using Twitter currently or not. A large group (71.1%) is not using Twitter. 17.9% does use Twitter, and 11.0% has an account but does not do anything with it.

Percentage

Yes 17.9

No 71.1

I have an account, but I am not using it at the moment 11.0 Table 7. Usage

5.2

Factor Analysis

A factor analysis is conducted, which primary purpose is to define the underlying structure among the variables in the analysis (Hair et al., 2006). The objective for this factor analysis will be summarizing data to identify underlying dimensions that describe data in a much smaller number of constructs (Hair et al., 2006). These constructs will be used in this research to conduct linear regressions to test the proposed research model.

The unit of analysis that is selected for this factor analysis is a R factor analysis. This R factor analysis is used to analyze a set of variables to identify dimensions that are latent (Hair et al., 2006). By using a R factor analysis, grouping is based on variables and not on respondents.

According to Hair et al., (2006) a preferred sample size should exceed 100 observations. Another general rule is that the minimum of observations should be at least 5 times as many as the number of variables to be analyzed. There are 20 variables selected in this factor analysis, which means 20 times 5 is 100. There are 173 observations which means this sample size is sufficient to conduct a factor analysis

43 of a factor analysis. One method to test this is the Bartlett’s test of sphericity, which provides statistical significance that the correlation matrix has significant correlations among at least some of the variables. The Bartlett’s test of sphericity performed on the dataset indicates a significant level of 0.000, which indicates that factor analysis is appropriate. Details are in appendix B. Another statistic is the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy. This index compares the magnitudes of the observed correlation coefficients to the magnitudes of the partial correlation coefficients (Malhotra, 2007). A value higher than 0.6 is desirable, because a lower value would mean that the correlations between pairs of variables cannot be explained by other variables which means that factor analysis is not appropriate. This research has a KMO value of 0.822, which means that factor analysis is appropriate. Details are in appendix B.

5.2.1 Deriving factors and assessing overall fit

A principal component analysis (PCA) is used. This method is most appropriate, because it has data reduction as a primary concern, focusing on the minimum number of factors needed for prediction purposes (Hair et al., 2006). To choose the number of factors to be extracted different selection criteria are used. First, the factors should have an Eigenvalue of at least 1,0. This is also called the latent root criterion. All factors that have latent roots larger than 1 are considered to be significant (Hair et al., 2006). This criterion indicates a 4 factor solution. The second criterion is that the factors have to add at least 5% of variance. This criterion indicates a 4 factor solution. A third criterion that is used, is that the factors should have a cumulative percentage of at least 60%. This criterion indicates also a 4 factor solution. The final criterion that is used is parallel analysis. This is computed by using the regression curve of Keeling (2000). If the regression line intercepts the line of the Eigenvalues, the extraction of factors should be stopped. This solution indicates a 4 factor solution. Details of these solutions are in appendix C.

44 5.2.2 Interpreting the factors

To make better interpretations of the factors, the factor matrix is rotated. By applying factor rotation, the factor matrix is transformed into a simpler one which is easier to interpret. The method used in this research is the varimax procedure, this is the most commonly used rotation method (Malhotra, 2007). To interpret the rotated factor matrix, variables should be identified which have high loadings on the same factor (Malhotra, 2007). When the factor loading is >0,30 it is significant, and >0,50 the variable is important to that factor. The output is visible in figure 11.

Figure 11. Output rotated component matrix

Overall, the variables have significant loadings related to their factors. Only remarkable is that effort expectancy and facilitating conditions are highly correlated.

5.2.3 Reliability of factor analysis

45

Factor Cronbach’s alpha Number of Items Performance Expectancy 0.909 4 Effort Expectancy 0.862 4 Privacy Concerns 0.785 4 Social Influence 0.742 4 Facilitating Conditions 0.761 2 Behavioral Intention 0.992 3 Extraversion 0.780 5 Introversion 0.565 2

Table 8. Cronbach’s alpha

Introversion and facilitating conditions were originally measured by respectively 5 and 4 measures. It became clear that these measures were not valid. After conducting analysis with ‘scale if item deleted’, it became clear that both constructs are represented by only two measures. Facilitating conditions improved to an acceptable level. Introversion has a remark of somewhat lower reliability and the need for future development of additional measures.

After checking reliability the validity should be checked as well. To validate the sample, the dataset is split into a random sample of 87 respondents. For this sample the factor analysis is performed again. Besides some small changes in loadings, there can be concluded that the factors are equal to the whole sample. Details are in appendix D.

5.3

Regression Analysis

To explore the interactions between the dependent variable behavioral intention and the independent variables performance expectancy, effort expectancy, social influence, and privacy concerns a linear regression is conducted. The basic model for this regression can be written as:

BI = α + β1PE + β2EE + β3SI + β4PC + ε , where:

BI = Value of behavioral intention α = Constant or intercept

PE = Value of the predictor variable performance expectancy

β1 = Regression coefficient that shows effect of performance expectancy EE = Value of the predictor variable effort expectancy

β2 = Regression coefficient that shows effect of effort expectancy SI = Value of the predictor variable social influence

β3 = Regression coefficient that shows effect of social influence PC = Value of the predictor variable privacy concerns