Master Thesis
What are the main factors responsible for the
continued use of a social media?
Koen Spreij S2128616 15-03-2019
MSc. Strategic Innovation Management Faculty of Economics and Business
University of Groningen
Thesis supervisor: W.W.M.E. Schoenmakers Co-assessor: P. Steinberg
Abstract
Table of content
Abstract ... 1
1 Introduction ... 3
2 Theoretical background ... 8
2.1 Social media in information system models ... 8
2.2 UTAUT2 ... 13
2.2.1 Moderators UTAUT2 ... 17
2.3 Extending UTAUT2 ... 18
2.3.1 Privacy and Trust ... 18
2.3.2 Uniqueness ... 19
2.4 Theoretical framework ... 21
2.5 Hypotheses ... 23
3 Method ... 29
3.1 Empirical setting ... 29
3.2 Data Collection Procedure ... 29
3.3 Survey measurements ... 30
4 Results ... 34
4.1 Descriptive Statistics and Correlations ... 34
4.2 Regression Results and Hypothesis Testing ... 37
5 Discussion ... 41 5.1 Hypothesis ... 41 5.2 Explained variance ... 47 5.3 Moderators ... 48 6 Conclusion ... 50 7 Theoretical Implications ... 52 8 Managerial Implications ... 55
8.1 Limitations and future research ... 56
9 Reference list ... 58
10 Appendix ... 68
10.1 Appendix I ... 68
10.2 Appendix II ... 71
1 Introduction
The rise of the internet has brought the introduction of major social media platforms, such as Facebook, Linkedin, Instagram, Twitter, YouTube, Reddit and Snapchat. Most of these companies have existed for over 10 years, so they are no longer a new phenomenon. People have gotten used to their existence, but users are also struggling with how to fit social media in their lives. A review of the literature shows that users struggle with the balance between privacy and openness (Ellison, Vitak, Steinfield, Gray, & Lampe, 2011), addiction to social media (Karaiskos, Tzavellas, Balta, & Paparrigopoulos, 2010), social media facilitating depression (Lin et al., 2016), insomnia and decreased wellbeing (Hunt, Marx, Lipson, & Young, 2018), and the spread of fake news (Allcott & Gentzkow, 2017).
As a result of these struggles, people are reframing their relationship with social media. Perin (2018) found that a significant share of US Facebook users have taken steps in the past year to change their relationship with social media. For example, 42% of US adults have taken a break from using social media for a period of at least several weeks (Perrin, 2018). Another study found that 59% of social media users think it would not be hard to give up social media (Smith & Anderson, 2018). Similar results were found in Europe, where Facebook, for example, lost 2 million monthly active users in the last two quarters (Edwards, 2018)—the first overall user losses in the history of the company. Also, monthly use in the US and Canada has been stable for almost four quarters (Facebook, 2018).
Anderson, 2018). With the Europe and US markets representing 73% of Facebook’s quarterly revenue (Facebook, 2018), the question for Facebook is not how to gain more users, but how to stay relevant to the current users in developed markets and prevent further EU and US user stagnation.
To deal with these challenges and opportunities, researchers are interested in why and how users use social media. Most research that has tried to explain the customer relationship to a social media company has been based on
information systems (IS) research (Ngai, Tao, & Moon, 2015). This knowledge on why users use social media is achieved through the identification of antecedents that lead to technology acceptance and use (Samaradiwakara & Gunawardena, 2014). Businesses can consider these antecedents, often called factors, when designing or innovating their services when people change their relationship with social media (Hallikainen, 2015). So, when researchers understand why a
particular user uses social media, superior business models can be designed (Osterwalder, 2004).
Examination of the IS literature shows that social media studies are predominately focused on factors such as perceived usefulness, social influence and perceived ease of use (Ngaia, Taoa, & Moon, 2015). These studies are predominately based on the prevailing literature stream for social studies: the technology acceptance model (TAM) (David, 1989), the theory of reasoned action (TRA) (Fishbein and Ajzen, 1975) and the theory of planned behavior (TPB) (Ajzen, 1985, 1991).
However, in a relatively short period, social media has gone through considerable changes. On the business side, social media companies have
generating content (Tang, Gu, & Whinston, 2012), functioning as knowledge sharing communities (Fernando, 2010, Kasavana et al., 2010, Yates and Paquette, 2011), used for collaborative product development (Mangold and Faulds, 2009, Porter and Donthu, 2008), used to connect and help organizations improve their customer retention and brand equity (Hanna et al., 2011; He et al., 2013; Kaplan and Haenlein, 2010; Laroche et al., 2013) and function as a marketplace and review site (Facebook, 2016).
Additionally, users have changed how they use social media, learned more about how the companies operate and developed a sense of when and how to use each social media platform (Smith & Anderson, 2018). The online
environment has made users better informed, less loyal and more involved (Prahalad & Ramaswamy, 2004). As such, user decision-making has become increasingly complex, however the IS models most often used to represent user decision have not been adapted (Samaradiwakara & Gunawardena, 2014).
The traditional models, TAM, TRA and TPB cannot represent this increasing complexity on their own and have additional limitations, such as only being applicable in organizational settings or are no longer considered to be accurate measurement methods (Samaradiwakara & Gunawardena, 2014). TAM, TRA and TPB focus on a specific set of factors of technology acceptance and use
(Hallikainen, 2015; Ngaia, Taoa, and Moon, 2015) and are unable to keep up with the rapid pace of change of social media technological developments (Ngaia, Taoa, and Moon, 2015).
perspective now that social media platforms play such an important role in people’s daily lives (Ngaia, Taoa, and Moon, 2015).
It is through theorizing and evaluating the users’ attitudes towards social media acceptance and use that researchers can develop insights for current and for future development of social media platforms and it will give directions to social media companies on how to develop new business strategies (March & Smith, 1995; Rauniar, Rawski, Yang, & Johnson, 2014). This led to the following research question:
RQ: What are the main factors responsible for the continued use of social media by social media users?
To achieve better user understanding of why consumers use a social media platform, the UTAUT2 model was used (an extension of unified theory of acceptance and use of technology in the consumer context).
2 Theoretical background
This section introduces relevant research streams on social media, the foundations of the UTAUT2 model and additional potential factors in a social media context. This describes the contemporary academic discussions in this research field and leads to the development of the hypotheses.
2.1 Social media in information system models
According to Dillon (2001), technology acceptance and use models are developed in information system literature to demonstrate the “willingness within a user group to employ information technology (IT) for the tasks it is designed to support” (Dillon, as cited in Samaradiwakara & Gunawardena, 2014, p.22). The term IT is used in the broadest sense, and thus includes technology acceptance and models from online banking, application adoption and the adoption of autonomous vehicles (Tamilmani, Rana, & Dwivedi, 2017). While there are many IS models (Ngai, Tao, & Moon, 2015), most social media studies are conducted on the basis of TRA, TBP and TAM (Ngai et al., 2015; Samaradiwakara & Gunawardena, 2014; Venkatesh, Morris, Davis, & Davis, 2003).
TRA was the first theory to gain general acceptance in technology acceptance research (Samaradiwakara & Gunawardena, 2014). It was a versatile theory, mainly due to its simplicity. The main goal of the model was to measure the “individual’s intention to perform a given behavior” (Ajzen, 1991, p.181). Intentions are constructed by the person’s attitude and subjective norms regarding that person’s behavior (Hsu & Lin, 2008). Attitude is defined as the person’s belief that the behavior will lead to a certain outcome (Fishbein & Ajzen, 1977). Subjective norms are the expectations of significant others, such as friends and family, regarding particular behaviors as perceived by the individual (Fishbein & Ajzen, 1977). It has been widely accepted by researchers that the TRA model has been too simplistic an approach and that there are most likely more explanatory variables that should be included in an IS model (Thompson et al., 1991; Webster and Martocchio 1992). For example, TRA does not account for behaviors users have no control over (Ajzen, 1991).
TAM, created by Davis (1989), is the most widely used model in social media IS studies (Ngaia, Taoa, & Moon, 2015). TAM expanded on TRA through the introduction and inclusion of two specific psychological factors: perceived ease of use and perceived usefulness (Samaradiwakara & Gunawardena, 2014). Research has shown that these two determinants are good at explaining social media use with similar scenarios, but different social media technologies (Davis, Bagozzi and Warshaw, 1989; Ngai et al., 2015). However, TAM does not include the subjective norms used in TRA and TPB. This reduces the model’s ability to explain the social dimensions and increases its focus on the technology imperative perspective (Yayla & Hu, 2007).
Previous studies have shown that TAM is a valid and robust model (King & He, 2006). While many studies have proven TAM’s usefulness and significance (Legris et al., 2003), TAM is believed to be too simplified (Bagozzi, 2007). While this makes it easier to use in various research settings, it also results in limitations, such as loss of detail (Kriponant, 2007). Rauniar et al. (2014) stated that the usability of the TAM model in social media should be further improved. Moon and Kim (2000), Hallikainen (2015) and Davis (1989) argued that future TAM research should explore additional explanatory variables. To address these issues, many authors have added additional factors or combined several models, leading to confusion on what TAM is and should be (Baenbasat & Barki, 2007).
Ngai et al., 2015; Samaradiwakara & Gunawardena, 2014). Several studies have suggested that the scope of social media research is too narrow and should aim to identify and gain broader understanding of the additional factors that play a role in social media acceptance and use (Bagozzi, 2007; Hallikainen, 2015; Ngai et al., 2015).
To measure how good an IS model is, the models are often compared based on their ability to explain a phenomenon using explained variance (R2) (e.g., Kwon,
Park, & Kim, 2014; Samaradiwakara & Gunawardena, 2014; Venkatesh et al., 2003; Yayla & Hu, 2007). TAM models have only managed to explain a proportion variance (Venkatesh, Morris, Davis, & Davis, 2003). Venkatesh et al. (2003) stated that the more variance a model can explain, the broader the researchers understanding of the phenomenon and the more explanatory factors are known (Kriponant, 2007).
According to Taylor and Todd (1995b) and Kriponant (2007), models should be evaluated on the balance between how many factors they have and their contribution to understanding of the phenomena. This means that fewer factors (parsimony) and higher explanation power is better. However, Venkatesh et al. (2003) argues that fewer factors is not desirable on its own, but only to the extent that it contributes to the understanding of the phenomena. If the goal of the study is to gain a complete understanding of a phenomenon, aiming for fewer factors can be discarded (Venkatesh et al., 2003).
Venkatesh et al. (2012) found for both UTAUT and UTAUT2 that the models have a significantly higher ability to explain behavioral intention and use than previous models, measured by the explained variance R2, while remaining useful
UTAUT scores significantly higher than adaptations of TAM (Samaradiwakara & Gunawardena, 2014). UTAUT, and more recently UTAUT2, thus have great explanatory power and are rapidly gaining interest as strong successors of TAM (Samaradiwakara & Gunawardena, 2014).
Theory / Model Constructs
(independent variables)
Moderators
TRA 1. Attitude towards
behavior 2. Subjective norm 1. Experience 2. Voluntariness TPB 1. Attitude towards behavior 2. Subjective norm 3. Perceived behavioral control 1. Age 2. Experience 3. Gender
TAM 1. Perceived usefulness
2. Perceived ease of use 3. Attitude
1. Gender 2. Experience
UTAUT 1.Performance expectancy 2. Effort expectancy
3. Social influence
1. Gender 2. Age
2.2 UTAUT2
UTAUT2, like TAM, was developed as a general technology use and acceptance model and is predomantly based on TAM. UTAUT2 tries to address the critiques of the previous models by introducing more factors found to be relevant in various studies (Venkatesh et al., 2003). This can also be seen in table 1 above, where the independent variables of the models are displayed. This shows that various models study different variables and that UTAUT2 adds several new factors. According to Samaradiwakara and Gunawardena (2014), UTAUT2 plays an essential role in technology acceptance research and is a solid basis for studying why users accept or reject technology in various contextual settings. The model contains more factors but reduces the interaction complexity that TAM, TRA and TPB focus on, as can be seen in Figure 2 below (Samaradiwakara & Gunawardena, 2014).
4.Facilitating conditions 4.Voluntariness
UTAUT2 1.Performance expectancy 2. Effort expectancy 3. Social influence 4.Facilitating conditions 5. Hedonic motivation 6. Price value 7. Habit 1. Gender 2. Age 3. Experience
Figure 2: UTAUT2 model as proposed by Venkatesh et al. (2012)
UTAUT2 is designed to understand technology acceptance and use in the consumer use context, specifically on the level of the individual user (Venkatesh et al., 2012, p.158). The UTAUT2 model has been used in at least 150 studies in a broad variety of contexts (Tamilmani, Rana, & Dwivedi, 2017). Venkatesh et al. (2012) suggest that the model should be used for various technologies. So far, UTAUT2 has been used to study online subjects similar to social media platforms, such as consumer acceptance behavior of mobile shopping (Marriott & Williams, 2016), online communication channel adoption (Nwanekezie, Choudrie, & Spencer, 2016), factors influencing the adoption of online shopping (Tarhini, Alalwan, Al-Qirim, & Algharabat, 2018) and the adoption of mobile internet (Venkatesh et al., 2012).
et al. (2012) use different terms for the factors to describe the same construct (see Appendix I). Table 2 shows that the four base factors of UTAUT2 all incorporate the aforementioned models and factors (TRA, TPB and TAM). The UTAUT2 model is therefore in line with previous social media models and capable of addressing the same factors as previous social media studies.
UTAUT2 base constructs Based on Source model
Performance expectancy Perceived usefulness TAM
Effort expectancy Perceived ease of use TAM Social influence Subjective norms TRA, TPB Facilitating conditions Perceived behavioral
control
TPB
Table 2: Base model UTAUT2 and models used in social media studies that form the meta constructs
In addition to the base model, UTAUT2 includes three new constructs and moderators that further refine the model’s relationships: hedonic motivation (e.g., enjoyment), price value and habit, and abandons on the moderator voluntariness as the model focuses on the consumer context (Venkatesh et al., 2012).
Various social media studies have also used the three factors addressed by Venkatesh et al. (2012). Lin and Lu (2011) found that hedonic motivation was the most important factor for why people use social media, stating that the social media platforms ability to arouse inner pleasure is crucial. Lu and Hsiao (2010) looked at price value and found that the perceived price value significantly influenced the percieve value of a social network platform. According to Venkatesh et al. (2012), habit (which can mean addiction, routine, or nature), is the least researched factor.
A lot of evidence supports the claim of users getting addicted to social media (Karaiskos, Tzavellas, Balta, & Paparrigopoulos, 2010; LaRose, Kim, & Peng, 2010) and use of social media is often considered routine behavior (Undiyaundeye, 2014). While habit is less used in IS literature, research found previous technology use is a strong indicator of future technology use (Kim & Malhotra, 2005). Futhermore, Ajzen and Fishbein (2000) state “repeated performance of a behavior can result in well established attitudes and intentions that can be triggered by attitude objects or cues in the environment” (Ajzen and Fishbein, as sited in Venkatesh et al., 2012). This means that repeated performance of a behavior can lead to automatic behavior, without the user first realizing that it has become a repeated behavior. According to Venkatesh et al. (2012), this repeated behavior is a habit and recognizable to many social media users.
IS studies (Appendix I). This means that there are enough indications to include all factors of Venkatesh et al.’s (2012) UTAUT2 model in this thesis.
2.2.1 Moderators UTAUT2
According to Venkatesh et al. (2012), technology acceptance and use is moderated by individual characteristics, such as age, gender and experience. Venkatesh et al. (2012) state that these moderators are not mandatory but can increase explanatory power. Previous social media studies did not encompass moderators or use a consistent and reliable system (Ngai et al., 2015).
A meta analysis showed that most researchers do not use the UTAUT2 moderators (Tamilmani et al., 2017). But the UTAUT2 research that did used curious approaches to the moderators; for example, several UTAUT2 studies used gender as a moderator but without incorporating a hypothesis. That is, they expected a relationship to be moderated, but did not state what they expect from the moderator before measuring the data. Yuan et al. (2015) measured age, gender and experience, but the potential moderating variables were not included in the model because they were not significant. Kim et al. (2008) also did not include a moderator hypothesis, however as the moderators were significant, moderator results were added in the research model. Correa, Hinsley and Zuniga (2010) incoperate two additional research questions, instead of hypothesis, to measure the effect of age and gender.
and experience can therefore help researchers to understand the user characteristics that play a role in the continued use of social media and help businesses understand their potential customers. Therefore, the moderating effect was tested in this thesis.
2.3 Extending UTAUT2
Venkatesh et al. (2012) suggest that additional variables should be considered, depending on the technology context. According to Venkatesh et al. (2012), the added variables should be based on careful theoretical and empirical consideration of the context being studied and should theoretically complement the factors already included in UTAUT2. For example, in the case of research about online gaming, fantasy and achievement were added as a factor because prior research in the online gaming context identified these factors as important (Xu, 2014). After a careful examination of the previous IS literature on Google Scholar with the terms E-business, social media and utaut2, three additional factors were identified by previous social media research: trust, privacy and uniqueness. These factors are discussed below.
2.3.1 Privacy and Trust
Dinev and Hart (2006) used the constructs privacy and trust in their privacy calculus theory as the phenomena where “individuals perform a calculus between the expected loss of privacy and the potential gain of disclosure” (Kokolakis, 2017, p. 14). This privacy tradeoff influence people’s final behavior (Jiang et al., 2013; Xu et al., 2011; Dinev & Hart, 2006). Researchers also looked at other factors. Gao et al. (2015) looked at privacy risk, Nysveen and Pedersen (2016) added privacy risk harm as a variable and Zhou (2012) looked at privacy concerns, trust and perceived risk. Dinev and Hart (2006) state that if individuals consider the gains greater than the losses, they will be willing to disclose personal information and as such continue to use the software. However, this does not mean that there cannot be contrary beliefs. Dinev and Hart (2006) argue:
“The influence of one belief [trust] might override another to the extent that the resulting probability favors one behavioral intention over another. However, the strength of the overriding beliefs influence does not eliminate the role or the importance of the contrary belief [privacy]” (p.62).
Privacy and trust are representations of the complexity of human decision-making and can both be significant at the same time. Dinev and Hart’s (2006) model allows researchers to investigate more precisely differently valued beliefs embedded in one person. This will result in a better understanding of the factors that play a role in consumers’ decision-making and should therefore be incorporated into a UTAUT2 social media model.
2.3.2 Uniqueness
Researchers study the factor uniqueness by comparing social media platforms. Two often-compared platforms in social media research are Facebook and Twitter. Davenport et al. (2014) suggest that features unique to Twitter may be more appealing to certain user groups than those on sites such as Facebook. The differences between the two social media platforms are summarized in Table 3. The table shows the differences between the two social media platforms and how these differences create a unique user experience for both user groups.
Facebook Twitter Author
Full array of functions One relatively light
function Kwon, Park, and Kim, 2014 Complex interface Simple interface Kwon, Park, and Kim,
2014
Versatile and dynamic Simple navigability Davenport et al., 2014
Evans, 2010 Tagtmeier, 2010
Multimedia Texts and links Mendelson and
Papacharissi, 2010 Stone et al., 2008
Passive use Active use Glasson, 2008
Complex and
customizable privacy measures
Content open to the
public Debatin, Lovejoy, Horn, and Hughes, 2009
Complexity makes it more suitable for desktop users
Simpler user interface ideal for mobile-based platforms
Kwon, Park, and Kim, 2014
imitable and non-substitutable resources. This competitive advantage through resources enables the firm to produce products or services that are perceived as superior by the customer and considered unique to the firm (Hunt & Morgan, 1995).
Due to the increase of e-businesses, uniqueness has gained more attention in research and businesses as customer options increased (Osterwalder & Pigneur, 2002). Today’s customers are better connected and more informed than customers in the past (Prahalad & Ramaswamy, 2004). Prahalad & Ramaswamy (2004) therefore stated that companies must deliver unique value for each customer.
Osterwalder and Pigneur (2002) also found that features are easily copied or changed in e-businesses (Osterwalder & Pigneur, 2002). This makes it hard for a company to stand out from the competition and create a sustainable competitive advantage. As a result, the consumer decides if being unique is a necessity (Steininger, Wunderlich, & Pohl, 2013). Uniqueness was therefore added to the UTAUT2 social media model.
2.4 Theoretical framework
While UTAUT2 focuses on technology acceptance and use of novel technologies, novel technologies is a broad interpretable concept and novel technologies in IS literature are often studied long beyond their introduction date (Kim, Park, & Oh, 2008). The term acceptance and use can also be a problem, as pointed out by Kim et al. (2008). Often the technology studied is already in the use phase, and interest is more focused on why people use it, instead of if people will use it.
to use is interchangeable with continued intention to use for social network platforms. Hallikainen (2015) and Lin and Lu (2011) state that while technology acceptance models are the most relevant for technology acceptance and use research, social media is not a new technology and thus the study should exchange “the continued intention to use as a proxy for the behavioral intention” (Hallikainen, 2015, p.12). This method to replace intention to use with continued intention to use has also been used in several similarly structured UTAUT2 studies that focused on topics besides social media, such as continued use of online games (Xu, 2014), keep using my health apps (Yuan, Ma, Kanthawala, & Peng, 2015) and continued use of SMS (Kim, Park, & Oh, 2008). The authors of these studies did not encounter any problems in exchanging intention to use with continued intention to use in the UTAUT2 model, so the latter measure was adopted in this study.
Figure 3: Proposed research model to study the intention to continue to use social media
This model aims to explain the user’s intention to continue to use of social media platforms by looking at previously developed factors in the acceptance and use of software technology, as described in UTAUT2. In addition, three newly identified factors were added (privacy concerns, trust and uniqueness).
2.5 Hypotheses
performance of a process through the use of a technology, as expected by the user (Venkatesh et al., 2003). According to Venkatesh et al. (2003), performance expectancy is the users’ belief that the software can help them accomplish things quickly, makes the users more productive and is helpful in their daily lives. This is in line with previous social media studies, which found evidence for usefulness in the decision to use social media (Lin and Lu, 2011). Users who think they are productive on Facebook and think that Facebook is useful and functional are more likely to continue to use the software. This leads to the hypothesis:
H1: Performance expectancy has a positive effect on the continued intention to use Facebook
Venkatesh et al. (2012) considers effort expectancy a strong indicator of software acceptance and use. Venkatesh et al. (2012, p.159) define effort expectancy as ”the degree of ease associated with consumers’ use of technology.” If users perceive social media as intuitive and easy to use, they are more likely to continue to use the software (Venkatesh et al., 2012). Many social media studies have found similar results (e.g., Hsu and Lin, 2008; Steyn et al., 2010; Hossain and de Silva, 2009). This leads to the following hypothesis:
H2: Effort expectancy has a positive effect on the continued intention to use Facebook
technology acceptance research, such as the adoption of smart kiosks (Chiu et al., 2010), mobile payment solutions (Oliveira et al., 2016) and RFID-based applications in the healthcare sector (Chong et al., 2015). There is strong evidence that friends and family influence the intention to use a particular technology in various ways. This leads to the following hypothesis:
H3: Social influence has a positive effect on the continued intention to use Facebook
Venkatesh et al.’s (2003) analysis of previous technology acceptance models found that facilitating conditions is an important factor in the acceptance and use of a technology. According to Venkatesh et al. (2003), facilitating conditions is the amount of organizational and technical infrastructure that the user believes the organization offers to support the use of a system. This means that it is the perceived available support that helps a user decide to use the technology. It includes the factors in the environment that facilitate or impede the use of social media.
According to Casaló et al. (2010) and Chang and Zhu (2011), facilitating conditions (perceived behavior control) is a significant factor in social media studies to predict users’ behavior from intention to action (Ngai, Tao, & Moon, 2015). In the context of this research, perceived behavior control means the availability of device support, an internet connection and users’ knowledge and skills regarding the application. If users have better access, more knowledge about social media and access to its support system, they are more likely to keep using the software. This leads to the following hypothesis:
According to Venkatesh et al. (2012), Lin and Lu (2011) and Hallikainen (2015), hedonic motivation is an important factor in the prediction of adoption and use of a technology. Venkatesh et al. (2012) describe hedonic motivation as the pleasure of using a novel technology. Previous social media studies used the terms enjoyment (Lin and Lu, 2011; Mathwick, 2002) and emotional value (Hallikainen, 2015) to define a similar concept. People who enjoy using social media are assumed to be more likely to use social media (Lin & Lu, 2011). This leads to the following hypothesis:
H5: Hedonic motivation has a positive effect on the continued intention to use Facebook
Venkatesh et al. (2012) found that the cost and pricing of a technology has a significant impact on consumers’ technology acceptance and use in previous studies. If consumers perceive the price value as positive, they are more likely to accept and use the technology. Previous social media studies have also stated the importance of price (Kohli & Jaworski, 1990) and price-quality (Chen, Fay, and Wang, 2011) in market orientated social media research. In this research, price value is used to describe suitability between the costs to be issued to use Facebook and the benefits that a user does perceive. This leads to the following hypothesis:
H6: Price value has a positive effect on the continued intention to use Facebook
use is a strong indicator of future technology use (Kim & Malhotra, 2005). Furthermore, several studies have pointed out that Facebook is addictive (e.g., Karaiskos, Tzavellas, Balta, & Paparrigopoulos, 2010; LaRose, Kim, & Peng, 2010). It is assumed that people who build a habit of using social media will be more likely to continue to use the platform. This leads to the following hypothesis:
H7: Habit has a positive effect on the continued intention to use Facebook
Several social media studies have indicated that privacy is an important factor in social media usage (Hugl, 2011; Kokolakis, 2017; Ngai et al. 2015). Dinev and Hart (2006) argue that privacy concerns are in line with the expectancy theory of Vroom (1964). This means that privacy concerns as a construct reflects the concern of opportunistic behavior related to the information provided by the user (Kowatsch and Maass, 2012). As a consequence, users will try to minimize the negative consequences of their information disclosure behavior. This leads to the following hypothesis:
H8: privacy concerns have a negative effect on the continued intention to use Facebook
recognize that they are willing to take some risk that is involved in using the technology. This leads to the following hypothesis:
H9: Trust has a positive effect on the continued intention to use Facebook
There has been an increased interest in IS literature on how social media platforms compare to each other (Kwon et al., 2014). In addition, businesses are increasingly concerned with delivering unique value to its users to keep them (Lee et al., 2006; Osterwalder, 2004). By delivering superior value through its offering, a company distinguishes itself from its competition in the eye of the user (Hunt & Morgan, 1995). Social media platforms aim to diversify themselves and have distinct advantages over other platforms. Unique value offering, if perceived by the customer, can lead to long lasting relationships with the customer. This leads to the following hypothesis:
3 Method
This section explains the setting used to study the new UTAUT2 model, how the sample data was collected for this study and how the variables were constructed. This section also discusses the methodological choices.
3.1 Empirical setting
To validate the UTAUT2 model in a social media context, this study targeted current users of Facebook. Facebook was selected as it is the most-used social media platform in the Netherlands and has been around for a long time, such that users are accustomed to it (Oosterveer, 2018). Facebook is also used throughout all demographics and all geographical regions of the Netherlands. Studying Facebook will provide good insight into how different people in the Netherlands use social media. In a consumer context, the use of Facebook is also a voluntary decision, which is important for the UTAUT2 model (Venkatesh et al., 2012).
3.2 Data Collection Procedure
To reach as many Facebook users as possible, an online survey was conducted that was promoted on the University campus and a technology forum. conducting a web-based survey has a number of benefits over traditional paper-based surveys: they are more inclusive and allow the researcher to access a larger population, they are cheap and fast to carry out, they can recruit large number of participants, and the data is captured in an electronic format, making analysis faster (Wyatt, 2000). Participants are also more likely to be drawn at random from the source population, reducing the possibility of selection bias.
respondents had a Facebook account. This means that 85% of the respondents had Facebook, compared to the Dutch average of 78% (Oosterveer, 2018). The demographics of the respondents are given in Table 4. To test whether the sample is a good representation of Facebook users, the sample characteristics were compared with available information about the population. According to the latest social media report in the Netherlands, Facebook is most used by people between the ages of 20–39 (89% of the population) and 40–64 (77% of the population) (Van der Veer, 2018). This is also represented in the sample, where the average age is 39 with a standard deviation of 14 years. The assumptions to test for outliers showed no irregularities in the scatter plot (Appendix III).
Measure Items Frequency %
Gender Male 80 62
Female 49 38
Age Min 19
Max 81
Average 39
Facebook usage Once a month or more 124 93
Less than once a month 5 7
Table 4: Demographic information of the respondents with Facebook
3.3 Survey measurements
After a rigorous literature review for the factor uniqueness, no existing questionnaire with questions which illustrate this factor was found. Following questionnaire design methodology (Fink, 1995) and using existing literature on uniqueness (e.g., Barney, 1991; Kwon et al., 2014), new items were designed. Extra attention was paid to focusing on simple language, avoiding multi-barrel question (questions that touches upon more than one issue), making the questions fit with the other questions and taking into consideration the formulated hypothesis (Fink, 1995).
The items of the construct were then first independently tested with 10 Facebook users to evaluate their validity and reliability and for internal consistency (Fink, 1995). All the item scales on the survey used for this research were adapted from Venkatesh et al. (2012). Items of the dependent and independent variable were measured using a seven-point Likert scale, ranging from “strongly disagree” to “strongly agree” (Venkatesh et al., 2012; Hossain & de Silva, 2009). An overview of the items can be found in Appendix II.
For the moderators, survey conductors were asked to indicate their age, gender and experience with Facebook. Experience was measured in years and frequency of use. Frequency was measured using a seven-point scale ranging from “never” to “many times per day.” Respondent who used Facebook once a month or more and had a Facebook account for a minimum of a year were considered experienced. All respondents had Facebook for more than a year.
(Venkatesh et al., 2012). The questionnaire was then tested among 10 Facebook users. These respondents were asked to comment on the format, the wording of the scales and the length of the instrument (Venkatesh et al., 2012). The respondents were also asked how hard it was for them to answer the questions in English on a scale from 1–10, with 10 being very hard. The average response was 3.
Respondents were then asked if there were any particular items they struggled with. Respondents reported that they had problems understanding the price value items. This led to minor adjustments in the descriptive text for the price value items compared to the original questions proposed by Venkatesh et al. (2012). The items can be found in Appendix II. The pre-test results were also used to see if the scales were reliable and valid. The pre-tests results were not included in the main survey.
4 Results
This section presents the statistical results derived from this study. First, the descriptive statistics and correlations are shown, after which the results of the multilevel regression analyses are described.
4.1 Descriptive Statistics and Correlations
Nunnally (1978) assumes that the average correlation of a set of items is a precise estimate of the average correlation of all items that relate to a certain construct. To measure the reliability of a scale that consists of a number of items, Cronbach's reliability coefficient alpha is often used. This is a measure of the internal consistency of the scale when it would be represented by an unweighted scale score (de Vocht, 2016). The average Cronbach’s alpha for each construct was calculated, and all the alpha values were above 0.7, as shown in Table 5. This means that the constituent items all relate to a single concept and thus are reliable (de Vocht, 2016).
Variables Cronbach’s alpha Number of items
Table 5: Reliability statistics and Cronbach's alpha
As shown in Table 6, there are no high correlations. As a result, it is likely that there are no issues with multicollinearity. Multicollinearity is the coherence between the independent variables (Vries & Huisman, 2007). A higher multicollinearity means that the independent variables are more dependent on each other
Correlations Intention to use Performance Expectancy Effort Expectancy Social Influence Facilitating conditions Hedonic Motivation Price Value Habit Privacy Concerns Trust Uniqueness Intention to use - Performance Expectancy .567** - Effort Expectancy .148 .037 - Social Influence .309** .408** .142 - Facilitating Conditions -.012 -.028 .417** .108 - Hedonic Motivation .536** .483** .191* .431** .008 - Price Value .322** .199* .316** .187* .219* .463** - Habit .532** .415** .136 .131 .066 .448** .347** - Privacy Concerns -.103 .002 -.163 .031 -.005 -.031 -.049 -.045 - Trust .401** .302** .137 .145 -.054 .223* .213* .271** -.351** - Uniqueness .368** .224* .200* .161 .065 .297** .279** .303** -.004 .231** -
**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
36 To further test for the presence of multicollinearity in the sample, the VIF test was conducted, as shown in Table 7. This test is suggested by Vries and Huisman (2007) and is used to test the assumption of no dependency between interdependent variables. The VIF indicates how much the variance of each regression coefficient increases compared to the situation in which all
independent variables would be uncorrelated and the tolerance is the inverse of the VIF (Vries & Huisman, 2007). Table 7 shows a mean VIF of 1,483 for all variables, with hedonic motivation presenting the highest value at 1,925, well below the suggested maximum threshold of 4 (Vries & Huisman, 2007). This means that the assumption of multicollinearity was not violated and that regression analysis could be conducted with the sample.
Variables Sig. Tolerance VIF
37 Table 7: Results of VIF test on the sample.
4.2 Regression Results and Hypothesis Testing
Table 8 below displays the results of the stepwise regression analysis performed to test the hypotheses. The significance of the independent variables and moderators varies between the models.
Model 1 is the baseline model to test the hypotheses. It includes the independent variables and measures their impact on continued intention to use. The research model is similar to the UTAUT2 model of Vankatesh et al. (2012), but has three additional factors. Several factors are significant and have a positive relation with continued use of Facebook, supporting the hypothesis. The first three factors are part of the original UTAUT2 model. Performance expectancy displays a positive and high significant result (b = .350, p = .001). Hedonic motivation is positively related and significant (b = .200, p = .032). Habit is also positively related and significant (b = .235, p = .003). Trust, a factor added to the research model, is also positively related and significant (b= .204, p =.042).
In model 2, all independent variables were tested in addition to single order interaction effects. This means that all independent variables were tested based on age (AGE), gender (GDR) and experience (EXP) independently of the other moderators. Performance expectancy is the only independent variable that is significant and has a positive relation with continued use of Facebook. Performance expectancy displays a positive and significant result (b = .528, p = .009), confirming hypothesis 1. None of the interaction effects are significant.
38 only difference is the addition of the three independent variables relevant in the social media context. None of the independent variables stand out, however, two interaction effects are significant. Performance expectancy x age is positive and significant (b = .058, p = .013) and higher-order interaction term performance expectancy x age x gender is also significant (b = -.079, p = .013). This means that older people and females (the min in b= -.079) show a higher effect of performance expectancy on continued intention to use of Facebook.
The analysis confirmed that R2 improved as more moderators were added
to model 1 (R2 = .521), model 2 (R2 = .630) and model 3 (R2 = .829). Adjusted R2
however has a dip in the second model going from R2 (adj) = .480 in the first
model to R2 (adj) = .458 in the second model and increases in the third model R2
(adj) = .601.
Finally, it should be stated that according to Green (1991) model 2 and 3 have a to low degree of freedom because there were not enough respondents. As a result, the findings of model 2 and 3 have very low power, this makes detecting differences within the data impossible (de Vocht, 2016).
Model 1 Model 2 Model 3
b error (t) b error (t) b error (t)
40 FC x AGE x GDR -.001 (-.033) HM x AGE x GDR -.021 (-.396) PV x AGE x GDR .036 (1.199) H x AGE x GDR -.009 (-.319) PC x AGE x GDR -.045 (-1.195) T x AGE x GDR .001 (.030) U x AGE x GDR .026 (1.164) EE x AGE x EXP -.057 (-1.031) SI x AGE x EXP .054 (1.043) HM x AGE x EXP -.138 (-1.277) PV x AGE x EXP .031 (.589) H x AGE x EXP .056 (.806) PC x AGE x EXP -.016 (-.197) T x AGE x EXP -.005 (-.195) U x AGE x EXP .027 (.527) EE x GDR x EXP -.037 (-.050) SI x GDR x EXP -.730 (-.924) FC x GDR x EXP -.162 (-.197) HM x GDR x EXP .380 (.302) PV x GDR x EXP .881 (.862) H x GDR x EXP -1.049 (-.978) T x GDR x EXP -1.424 (-.849) U x GDR x EXP 2.070 (1.707) PE x AGE x GDR x EXP .023 (.460) SI x AGE x GDR x EXP -.052 (-1.053) FC x AGE x GDR x EXP -.002 (-.066) HM x AGE x GDR x EXP .062 (.462) H x AGE x GDR x EXP -.030 (-.451) PC x AGE x GDR x EXP .051 (.654) R2 .521 .630 .829 R2 (adj) .480 .458 .601 F-change 12.616 3.658 3.631 Deviance (df) 116 86 54 * p < 0.05 ** p < 0.01 *** p < 0.001
41
5 Discussion
This section discusses and interprets the empirical results. First, the 10 hypotheses constructed using the theory will be discussed on the basis of model 1 of the regression analysis. Secondly, relevant results from model 3 with all moderators will be discussed.
5.1 Hypothesis
Hypothesis 1, performance expectancy has a positive effect on the continued intention to use Facebook, is supported. This means that users use Facebook because they find it useful, it helps them to achieve things or increases their productivity. Vankatesh et al. (2012) also found strong support for this factor, and it remains one of the main indicators of social media use (Hsu and Lin, 2008; Hossain and de Silva, 2009; Lin and Lu, 2011; Steyn et al., 2010). This study confirms the relevance of the focus on usefulness in social media studies from a broader perspective than studied before (Ngai, Tao, & Moon, 2015).
42 significance of ease of use holds up if intention to use is replaced with continued intention to use (Kim et al., 2008; Kwon & Wen, 2010). A possible difference could be that because this study had predominately experienced users, ease of use is less of a concern.
43 research models should therefor aim to include factors or items more suited to measure experienced users’ attitude towards social factors.
Hypothesis 4, facilitating conditions have a positive effect on the continued intention to use Facebook, is rejected. Venkatesh et al.’s (2003) analysis of previous technology acceptance models found that facilitating conditions is an important factor in the acceptance and use of a technology. It is based on TBP’s behavior control and it assumes that a user has the resources to perform the use of a social media (Venkatesh et al., 2003).
The factor facilitating conditions has been used with success to explain behavioral intentions in social media in the past (Casaló et al., 2010; Chang & Zhu, 2011). That the results are not significant in this research can partly be explained through the different items used for the factor facilitating conditions. Chang and Zhu (2011) used unusual items to measure facilitating conditions, focusing on having enough time and ease of use. These items are very different of that of what the items are and suppose to measure according to TPB (Ajzen 1991; Taylor and Todd 1995a) and how the items are used by Vankatesh et al. (2003). However, previous irregularities do not explain why it is not significant in this study. In the case of the social media platform studied in this research (Facebook), facilitating conditions might not be important or too farfetched for users, meaning that they are not interested in the availability of device support, have the right knowledge and skill or have an internet connection. Facebook is available for every device and has developed the user interface constantly to make it easier to understand (Bromwich & Haag, 2018).
44 Facebook stimulates the continued use of Facebook. This is an interesting new finding, as hedonic motivation is not part of TAM, TRA and TBP. Relatively few studies have looked at the subject of the joy and entertainment value of social media (Lin & Lu, 2011; Ngai et al., 2015), and similarly few IS studies in general have examined this subject (Venkatesh et al., 2012). This study shows, however, that it is a relevant factor that should be considered in a social media context.
Hypothesis 6, price value has a positive effect on the continued intention to use Facebook, is rejected. Given the results of previous research, this is an unexpected result. The result might be due in part to the survey items. As described in the method section, pre-testers struggled with the wording on these items. Terms like “worth the price” were confusing, because Facebook is a free service. The pre-testers asked how they should value something that is free. Paying zero euros and still gaining value does not necessary change the concept of price value (e.g., Schreiner & Hess, 2013). However, to make the questions less confusing, the research items were changed to incorporate the fact that Facebook is a free service. This was different than most surveys that use the UTAUT2 model, as most IT is not free, including the previous social media price study by Lu and Hsiao (2010).
45 Hypothesis 7, habit has a positive effect on the continued intention to use Facebook, is supported. Habit is the extent to which people perform behavior automatically (Venkatesh et al., 2012). This can mean that users use a social media platform solely because it is an addiction, habit or nature. This can be worrisome for Facebook developers, as there is no underlying reason to keep using Facebook in this case. While habits can be strong, they are susceptible to change (Venkatesh et al., 2012).
As stated in the theoretical section, relatively little literature has looked at the subject of habit in social media (e.g., Ngai et al., 2015). This factor is also not part of the traditional models (TRA, TBP and TAM). Research should therefore further investigate this factor and its relationship to other factors. This model only shows that habit plays a role, but not how it is influenced by other factors or potential moderators, such as the factor time on the platform.
46 However, if asked to elaborate on what the privacy application is, only 25% was able to explain the application. This points out that studies often expect knowledge of users that might not be there. Conclusively, the evidence from the above-mentioned studies, shine light on the complexity of the factor privacy which can lead to the suggestion of more extensive research on the topic of privacy in social media studies.
Hypothesis 9, trust has a positive effect on the continued intention to use Facebook, is supported. Trust is not a new factor in IS literature (Ngai et al., 2015), but to the researcher’s knowledge, this is the first time it has been added to UTAUT2. The items on the surveys used in this study measured user trust in Facebook. Users with high trust in Facebook as a company and how it handles sensitive user data are more likely to keep using Facebook. This is an interesting finding, considering that Facebook has faced many scandals recently that could jeopardize users’ perceived trust in Facebook. However, this model does not explain why people have such high trust. Kowatsch and Maass (2012) point out that trust can depend on the amount of risk a user is willing to take, something not measured in this model.
47 consumers are first asked to reflect on a product or service, consumers articulated preference will be more susceptible to change. In the case of Facebook or other social media platforms, this will mean that it is possible that when a user prior to a research, studies the characteristics of Facebook, the results could be different.
5.2 Explained variance
Previous IS models are compared based on their ability to explain a phenomenon using explained variance (R2). The use of moderators can have a
significant effect on explained variance (R2) (Van Raaij & Schepers, 2008). In line
with previous findings, the analysis confirmed that R2 improved as more
moderators were added between model 1 (R2 = .521), model 2 (R2 = .630) and
model 3 (R2 = .829). This is expected behavior of the computation model. R2
always increases as more variables are added to the model, no matter their relevance (de Vocht, 2016). De Vocht (2016) suggests therefor to look at adjusted R2. Adjusted R2 adjusts for the number of new variables in the model and only
increases if the explained R2 is greater than could be explained by chance (de
Vocht, 2016). Adjusted R2 has a dip in the second model, going from R2 (adj) =
.480 in the first model to R2 (adj) = .458 in the second model. This decrease means
that the moderators added in model 2 lower the explaining capability of model 2. This means that model 1 is a better predictor of Facebook use than model 2.
The third model has both the highest explained variance (R2 = .829) and
adjusted R2 (R2 = .601). The model, however, has fewer significant variables.
48 model 1 relationship disappearing and a new relationship taking its place. By adding so many variables, the variables lose test power in multiple regression (de Vocht, 2016). This means that while additional variables contribute to an increased adjusted R2, this comes at the cost of test power. Consequently, results are less
likely to be significant, as the model has to control for every additional variable (de Vocht, 2016).
5.3 Moderators
Venkatesh et al. (2012) found that moderators play a significant role in technology acceptance and usage. The results of this study on this point are however not easy to interpret.
If the results of model 1 and model 3 are compared, the results show that model 1, without the moderators of model 3 achieved good results. Model 1 found four significant factors and a R2 (adj) of 0.480, which is high compared to other findings (e.g., Bagozzi, 2007; Ngai et al., 2015). The moderator model (model 3) has the highest adjusted R2 (0.601) and thus explanatory power. However, the cost for increasing the adjusted R2 is a decrease in the number of significant factors. In model 3 only two factors are significant.
49 when you take into consideration the number of constructs and explanatory power. Venkatesh et al. (2016, p.338) has recognized that UTAUT2 scores badly on parsimony due to “the complex interactions among the attributes as implied by the moderation effects.”. Weber (2012) suggests in this case that moderators should only be considered if a parsimonious theory can be defined precisely so that only a select number of moderators can be added to the model. Weber (2012) suggests to only use 7 moderating effects. Thus, the research for the moderators in a social media context should be developed further and expanded so that hypotheses with precise estimates of the effect of age, gender and experience on social media in UTAUT2 model can be tested.
50
6 Conclusion
Social media has developed into an important role in the daily life of many people and provides a wide variety of functions. Previous social media IS literature highlighted the factors ease of use and perceived usefulness for social media (Ngai, Tao, & Moon, 2015). These studies were based on three dominant models: the theory of reasoning action, the theory of planned behavior and the technology acceptance model, and focused on a selective set of factors (Ngai et al., 2015; Samaradiwakara & Gunawardena, 2014). This thesis offers a wider view of social media research and provides additional factors that give valuable insights into why people use social media. To answer this research question, a more modern and inclusive IS research model (UTAUT2) was used. To measure the diversity of reasons to use social media, this study strived to answer the following research question: What are the main factors responsible for the continued use of social media?
This thesis model confirmed the importance of perceived usefulness and identified less researched factors, such hedonic motivation, habit and trust, as important indicators of why people use Facebook. A surprising finding was that ease of use (called effort expectancy in UTAUT2) is not a significant factor, contrary to previous studies. Social Influence was also not significant as experience users feel less pressure from peers.
51 that more extensive research models can give valuable results, although moderators should be used with caution and requires more research.
52
7 Theoretical Implications
This thesis underlines the complexity and variety of reasons why people use social media. Social media platforms, such as Facebook, have seen major changes in what they offer and how people use them (Ngai et al., 2015). However, the models (TAM, TBP and TRA) used to analyze these platforms have remained the same, focusing on a few specific variables (Ngai et al., 2015; Samaradiwakara & Gunawardena, 2014; Venkatesh et al., 2003). This research aimed to widen the view of research to include all relevant variables and moderators that affect users’ decision to use social media.
This research is based on the technology and acceptance model with the greatest capacity to explain IT phenomena: the UTAUT2 model (Venkatesh et al., 2012). In addition, three relevant independent variables were added to the UTAUT2 model. The UTAUT2 model has not been used in the social media context before now. This study was designed to provide new insights into the behavior of users of social media platforms to help social media businesses design new social media business models or improve existing ones. In particular, this thesis investigates the impact of different user factors in a wider context of factors that influence user decisions and examined how age, gender and experience influence social media user decisions.
The first contribution of this study is that the technology acceptance and use model used has significant results for three of the seven original UTAUT2 factors. In addition, one of the three newly added factors were significant
53 factors have been identified that have not received much attention in the traditional models (TRA, TBP and TAM) and in social media literature in general. Three factors in particular—hedonic motivation, habit and trust—should therefore be considered when researching social media. It also shows that factors that have been studied in the traditional setting, such as ease of use and social influence as main subjects of study, do not necessarily have similar meaning in a new and wider perspective. These factors may be relevant but require new constructs that correspond with social media users experience using social media.
The first minor contribution of this research to the social media research field is that the explained variance is acceptable for all three models. Most technology and acceptance models emphasize the explained variance (Samaradiwakara & Gunawardena, 2014). Model 3’s R2 score is high (R2 = 0.829),
however, the variables also have less predictive power as more variables inflate the R2,as discussed above. This is also one of the critiques on UTAUT in general,
that the high level of R2 is only achieved by adding all moderating variables (Van
Raaij & Schepers, 2008).
The second minor contribution is that the UTAUT2 model can be used in a social media setting and can identify factors in a wider perspective and find these to be relevant. Until now, the model has not been tested in a social media setting. This confirms Venkatesh et al.’s (2012) concept that universal technology acceptance is a use model for individual users that should be applied in a wide variety of technological settings.
55
8 Managerial Implications
This study provides several implications for practitioners. It supports the idea that consumer decision-making is based on a wider variety of reasons than represented by traditional research models. Managers should therefore not blindly focus on a selection of factors but consider a range of factors in their decision-making processes. They should also be aware of the complexity of relationships that factors might have. Another interesting finding is that habit is an important factor in customer decision-making to use Facebook. While it can be a strategic choice to try to addict users to Facebook, it could also mean that users have no reason to use the platform other than habit. This means that their relationship to Facebook is vulnerable to new, more meaningful or addictive alternative social media platforms. However, users perceive a high degree of usefulness in Facebook, so they currently have good reasons to keep using it and continue their habit.
The findings also underscored the importance of performance expectancy, which means that Facebook should keep focusing on adding more features that make it useful, such as a marketplace, events, reviews, and tools that enhance user productivity, such as collaboration tools and discussion groups. These kinds of features increase the likelihood of continued use of Facebook. This could also mean investing in social functions that are predominately experienced as useful because social influence itself is not an important factor (such as birthday calendar or group chat). However it is possible that a social identity is important.
56 providers of content. Finally, trust plays a major role in the engagement of social media, meaning that the company should try to focus on being trustworthy and should make well-considered decisions when dealing with user data.
8.1 Limitations and future research
The study is subject to limitations, which should be considered in future research in the field of social media studies. The first limitation is that this study analyzes one social media platform (Facebook) in one country (the Netherlands). Even though Facebook is the most accepted and used social media platform in the Netherlands, future studies using data from several social media platforms and examining several countries could help generalize the findings of this study.
The second limitation is that because Facebook is well established and not a new technology, the difference between intention and acceptance did not need to be measured. The study therefore only needed one measurement moment. However, using two measurement moments would have shown how consumers perception of factors changed over time as a platform grows and introduces new features (Hossain & de Silva, 2009).
57 UTAUT2. Future research could test, for example, the relationship between ease of use and usefulness (Hossain & de Silva, 2009).
58
9 Reference list
Ajzen, I. (1985). From intentions to actions: a theory of planned behaviour. Action control, 11-39.
Ajzen, I. (1991). The Theory of Planned Behaviour. Organisational Behaviour and Human Decision Processes, 179-211.
Ajzen, I., & Fishbein, M. (2000). Attitudes and the Attitude- Behavior Relation: Reasoned and Automatic Processes. European Review of Social
Psychology, 1-33.
Allcott, H., & Gentzkow, M. (2017). Social media and fake news in the 2016 election. Journal of Economic Perspectives, 211-236.
Baenbasat, I., & Barki, H. (2007). Quo Vadis, TAM. Journal of the AIS, 211-218. Bagozzi, R. (2007). The legacy of the technology acceptance model and a
proposal for a paradigm shift. Journal of the Association for Information Systems, 244–254.
Bagozzi, R., & Dholakia, U. (2002). Intentional social action in virtual communities. Journal of Interactive Marketing, 2-21.
Barney, J. (1991). Firm Resources and Sustained Competitive Advantage. Journal of Management,, 99-120.
Brandimarte, L., Acquisti, A., & Loewenstein, G. (2013). Paradox, Misplaced confidences privacy and the control. Social Psychological and Personality Science, 340-347.
Bromwich, J. E., & Haag, M. (2018, Januari 12). Facebook Is Changing. What Does That Mean for Your News Feed? Opgehaald van The New York Times: https://www.nytimes.com/2018/01/12/technology/facebook-news-feed-changes.html
Casaló, L., Carlos, F., & Guinalíu, M. (2011). Understanding the intention to follow the advice obtained in an online travel community. Computers in Human Behavior, 622-633.
Casaló, L., Flavián, C., & Guinalíu, M. (2010). Determinants of the intention to participate in firm-hosted online travel communities and effects on consumer behavioral intentions. Tourism Management, 898-911. Chang, Y. P., & Zhu, D. H. (2011). Understanding social networking sites
adoption in China: A comparison of pre-adoption and post-adoption. Computers in Human Behavior, 1840-1848.
59 Cheung, C., & Lee, M. (2010). A theoretical model of intentional social action in
online social networks. Decision Support Systems, 24-30.
Correa, T., Hinsley, A., & Zuniga, H. d. (2010). Who interacts on the Web? The intersection of users’ personality and social media use. Computers in Human Behavior, 247-253.
Davenport, S. W., Bergman, S. M., Bergman, m. J., & Fearrington, M. E. (2014). Twitter versus Facebook: Exploring the role of narcissism in the motives. Computers in Human Behavior, 212–220.
Davis, F. (1986). A technology acceptance model for empirically testing new end-user information systems: Theory and results. Massachusetts Institute of Technology.
Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly, 319–340. Davis, F. D., Bagozzi, R. P., & Warshaw, P. R. (1992). Extrinsic and intrinsic
motivation to use computers in the workplace. Journal of Applied Social Psychology, 1111-1132.
de Vocht, A. (2016). Basishandboek SPSS 24. Bijleveld.
Dienlin, T., & Trepte, S. (2015). Is the privacy paradox a relic of the past? An in-depth analysis of privacy attitudes and privacy behaviors. European
Journal of Social Psychology, 285-297.
Dinev, T., & Hart, P. (2006). An extended privacy calculus model for e-commerce transactions. Information systems research, 61-80.
Dodds, W. В., Monroe, К. В., & Grewal, D. (1991). Effects of Price, Brand, and Store Information on Buyers. Journal of Marketing Research, 307-319. Edwards, J. (2018, October 31). Facebook’s user base is declining in Europe, and
that ought to terrify its American bosses. Opgehaald van businessinsider: https://www.businessinsider.nl/facebook-revenue-and-user-growth-declining-in-europe-2018-10/
Ellison, N. B., Vitak, J., Steinfield, C., Gray, R., & Lampe, C. (2011). Negotiating privacy concerns and social capital needs in a social media environment. Privacy online, 19-32.
Facebook. (2016, October 3). Introducing Marketplace: Buy and Sell With Your Local Community. Opgehaald van Newsroom:
https://newsroom.fb.com/news/2016/10/introducing-marketplace-buy-and-sell-with-your-local-community/
