“The effectiveness of Sense of Community
on Intention to Use:
an extension of the TAM model”
University of Groningen
Table of contents 1. Introduction 3 2. Literature Review 6 2.1 Technology Adoption 6 2.2 Adoption Models 7 2.3 TAM2 9
2.4 Integrating Sense of Community 11
2.5 Research Model 13
2.5.1 Cognitive Instrumental Constructs 14
2.5.2 Social Influence Constructs 15
3. Methodology 18 3.1 Setting 18 3.2 Research Design 19 3.3 Sample 20 3.3.1 Sample Analysis 21 3.4 Measurements 23
3.4.1 Reliability and Validity 24
3.4.2 Independent Variables 24
3.4.3 Mediators 26
3.4.4 Dependent Variables 26
4. Data Analysis and Results 27
4.1 Structural Equation Modeling (SEM) 27
4.1.1 Exploratory Factor Analysis 28
4.1.2 Confirmatory Factor Analysis 28
4.2 Results 30 5. Discussion 32 5.1 Findings 32 5.2 Limitations 34 5.3 Implications 35 6. Conclusion 37 7. References 38
Appendix I - Original questions and adapted questions 44
Appendix II - Adapted questions and back-translated questions 46
“The effectiveness of Sense of
Community on Intention to Use:
an extension of the TAM model”
___________________________________________________________________________
Abstract
Community members share a common interest which comes with a certain feeling: sense of community. This feeling can change the behavior of the community members and how they interact with each other. This research wants to determine to what extent sense of community influences the intention to use new technologies and uses the Technology Acceptance Model 2 developed by Venkatesh and Davis (2000) in order to find this relationship. A quantitative research was executed amongst 168 supporters of Dutch soccer clubs. The findings are that sense of community does have an indirect effect on the intention to use and therefore should be added to the Technology Acceptance Model 2 as a predictor of intention to use.
___________________________________________________________________________
1. Introduction
Most people are part of a group or a community. Groups or communities exist of members that have shared interests (Crane, Matten and Moon, 2004). Because group
Next to shared interests of community members, a community is characterized by the influence that the community members have on each other. This influence of community members on each other can strengthen the cohesion of the group (McMillan and Chavis, 1986; Nistor, Daxecker, Stanciu, and Diekamp, 2015). In general, people may change their behavior in order to be liked and accepted by others (Aronson, Wilson, and Akert, 2010). This behavior is also present in communities. Therefore, community members can influence their fellow community members, because individuals want to belong to the group and feel accepted as mentioned by Aronson, Wilson and Akart (2010). Some community members will become important, meaning that they will have more influence on the behavior and intention of the community members than other members will have (Nistor et al., 2015). The influence of members on each other within a community can lead to the ability for community members to do things that they would not have done without the community. According to Hill (1996) there is a need for extensive research in a variety of contexts to understand the antecedents and outcomes of sense of community, to understand the whole concept.
As said earlier, sense of community has been related to a range of outcomes. A potential new outcome can be found in the technology adoption literature. Sense of community can be related to technology adoption. Sense of community is found when community members explore their shared interests. Moreover, an aspect of communities is that members influence each other and the impact of influence is higher regarding individuals that have no shared interest. People want to belong and are inspired by other members of the community to behave in a certain way. This might be of great importance when introducing new technologies to groups of people, because people behave likewise the ones they regard as important or influential.
Shah (2003) already discovered the importance of communities at the development of new technologies and products. For example, community members can provide innovation-related ideas and assistance and markets can be created via a community through word of mouth, newsletters, and observation (Shah, 2003). However, this relationship is about communities and the development of products, and not the implementation. Besides, Shah (2003) does not mention sense of community. The goal of this research is to better understand the role of sense of community in explaining technology adoption of new technologies by community members.
brought forward several models that describe the factors that influence the adoption of new technologies (Davis, 1989; Davis et al., 1989; Tornatzky and Fleischer, 1990; Ajzen, 1999; Rogers, 1995; Venkatesh et al., 2003). According to these models in the literature of
technology adoption, there are factors that can have a direct or indirect effect on the intention to use new technologies. However, these authors did not determine the role of sense of community in explaining technology adoption while it might be interesting to take this variable into account. This research will extent one of the existing technology adoption models with a new factor, sense of community. With a better understanding of the technology adoption process and how people choose to adopt new technologies, we could identify
strategies that would maximize adoption (Yu, Wong, and Woo, 2018). Therefore, the following research question is formulated:
RQ1. To what extent does sense of community influence the intention to use new
technologies for individuals?
This research aims to determine the relationship and the mechanisms between sense of community and the intention to use new technologies. If the relationship exists, it will be relevant for firms that develop new technologies. The outcomes of this study can create insights in technology adoption within communities. If sense of community has influence on technology adaption, this can cause a significant shift in the current manner of technology adoption. Firms can make use of the process of sense of community when implementing new technologies and products. Their products can be adopted quicker within communities with a high sense of community than communities with a low sense of community.
In the following section, hypotheses are developed based on the literature of
2. Literature Review
In this chapter the main model and concepts of this research will be discussed. First, the background of the technology adoption theory will be given and several technology adoption models will be elaborated on. Thereafter, the model of this research will be discussed and the hypotheses will be formulated in order to create a conceptual model.
2.1 Technology Adoption
Technology can be adopted into the lives of individuals. Adoption is the individual’s decision whether to integrate an innovation into his or her life (Straub, 2009, p.629). For instance, some individual choses to make use of a wearable that tracks their health such as a Fitbit. The process of adopting a technology consists of several steps (Karahanna, Straub, and Chervany, 1999). This process starts with some kind of knowledge about the technology. Then, a favorable or unfavorable attitude is formed towards the technology. Thereafter, the individual is making the decision to adopt or reject the technology. After the new technology is used or not, the individual is seeking reinforcement of the adoption decision that he or she made (Rogers, 1983). The individual may decide to reverse the decision depending on the information that he or she receives. The adoption process of technologies consists of several key constructs. These constructs are the innovation’s perceived attributes, the attitude and beliefs of the individual regarding the technology, and communications received by the individual from his or her social environment about the technology (Karahanna et al., 1999). These communications can be, for example, recommendations by others or a commercial that gets attention by an individual.
stages and want to learn more about the new technology. The two next groups are the largest groups: the (3) early majority and the (4) late majority. The first group is a curious group, individuals that are part of the early majority are aware of the new technology but first want to wait and see how others react to the innovation. The late majority is a group that has additional constraints and needs additional proof of the innovation. The last group is the group of the (5) laggards. Laggards do not adopt innovations and wait until new technologies are more common. Besides, laggards need help with the adoption and use of the new
technology.
Next to Rogers (1995), many authors have written about technology adoption with as a result that the user’s willingness to accept new technologies can be predicted by several models in the technology adoption literature. In the next chapter, these models will be discussed.
2.2 Adoption Models
The decision to adopt has been studied in various and numerous ways. Some studies took the ‘process’ approach and the in-depth process of technology adoption, as explained in the previous chapter. An example of this is the DOI model of Rogers (1995). Other studies focused on associations of technology adoption influencing variables. Several models will be discussed below to create insights in how adoption models can work and which approach is favorable to make use of in this research.
Fishbein and Ajzen developed the Theory of Reasoned Action (TRA) model in 1975. The theory of reasoned action is still useful and present in modern research, it recently
became the foundation of investigations of individual’s IT usage behavior (Taherdoost, 2018). Human behavior is predicted and explained by this model through three main cognitive
components: attitudes, social norms, and intentions (Fishbein and Ajzen, 1975). Ajzen (1991) extended TRA by creating the Theory of Planned Behavior (TPB). The variable Perceived Behavioral Control (PBC) was added to the existing TRA model. PBC can be defined as the capacity to act out certain behavior (Ajzen, 1991). PBC is determined by the availability of resources, opportunities and skills to achieve outcomes (White, Jimmieson, Obst, Graves, Barnett, Cockshaw, and Martin, 2015). Although both theories were developed to determine the individual’s behavior by the behavioral intention of the individual, TPB cannot be used in voluntary usage settings (Taherdoost, 2018).
Another model that is derived from TRA is the Technology Acceptance Model (TAM) by Davis (1989). TAM explains the motivation of users by three factors: perceived
usefulness, perceived ease of use, and intention to use. The Perceived Usefulness (PU) of a technology is defined as the extent to which a person believes that using the technology will enhance his or her task performance (Davis et al., 1989, p.320). A technology can be useful for individuals when performing a task, perceived usefulness describes how useful the technology will be when performing a task. When an individual perceives a technology as usefulness, the intention to use the technology will be more likely (Davis et al., 1989). It is valuable for an individual to use a useful technology and therefore the individual’s intention to use it will be higher. Next to perceived usefulness, perceived Ease of Use (PEOU) also can be a motivation to use a technology and perceived ease of use is defined as the extent to which a person believes that using the technology will be free of effort (Davis et al., 1989, p.320). A technology can be very complicated when using it, which indicates that the perceived ease of use is low. On the other hand, if a technology is easy and clear, the
perceived ease of use is high. When a technology is perceived as easy to use, it will increase the intention to use the technology (Davis et al., 1989). According to TAM, perceived ease of use has also a direct influence on the perceived usefulness. If the use of a technology is easy and clear to use, it will increase the usefulness of the technology for individuals (Davis et al., 1989).
(2000) added several external variables to TAM in their new model called TAM2. Venkatesh and Davis (2000) added two groups of constructs: social influence constructs (subjective norm and image) and cognitive instrumental constructs (job relevance, output quality, and result demonstrability). Next to these three cognitive instrumental constructs, perceived ease of use is also considered as a cognitive instrumental construct (Venkatesh and Davis, 2000). The new formation of constructs improved the predictive power of perceive usefulness because social and cognitive factors are taken into account. Therefore, TAM2 outperformed the traditional TAM (Taherdoost, 2018). The constructs will be defined in the next chapter.
This research will use TAM2 to determine the relationship between sense of community and intention to use. TAM2 is the preferred model in this research because it explains 40-50% of the variance in individuals’ intention to use an IT and actual usage (Venkatesh and Bala, 2008), it receives substantial empirical support (Taherdoost, 2018), and TAM2 is one of the most widely cited models in the field of technology acceptance (Wu, 2009; Taherdoost, 2018). Besides, TAM2 can be used in voluntary settings which is a criterion for this research at communities. Community members cannot oblige each other to use certain technologies, which is for example the case with an employer that can oblige an employee to use a certain technology. Another important reason to use TAM2 is that this model is easy to adjust. This research aims to add the variable sense of community to the technology adoption. The model was often changed by authors (Taherdoost, 2018). Therefore, TAM2 will be a suitable model for this research. In the next chapter, TAM2 will be further discussed.
Unfortunately, TAM2 was developed to estimate the intention to use in a working environment. This research focusses not on a working environment, but on a community environment. Therefore, the items of TAM2 have to be changed in order to use it in this research. This change will be explained later in this research.
2.3 TAM2
Venkatesh and Davis (2000) took TAM as a starting point during the development of TAM2. Across empirical tests of TAM, perceived usefulness has consistently been a strong determinant of intention to use. To understand the determinants of perceived usefulness, TAM2 was developed. A better understanding of the determinants of perceived usefulness would increase the overall understanding of the usage intentions of new technologies
that influence the perceived usefulness and divided them into two different groups: social influence constructs and cognitive instrumental constructs. Below, every construct will be briefly explained. In the methodology section, every relationship will be elaborated further.
Social influence constructs consist of subjective norm and image. Subjective Norm (SN) is defined as a person’s perception that most people who are important to him or her think he or she should or should not perform the behavior in question (Ajzen, 1991, p.188). So, other individuals’ can affect one’s behavior. Subjective norm is related to intention to use because others can directly affect the behavior in question. Also, subjective norm is related to perceived usefulness in TAM2 because others can affect the perception of the usefulness of a technology. The other social influence construct is Image (I). Image is defined by as the extent to which use of an innovation is perceived as enhancing one’s image or status (Mun, Jackson, Park, and Probst, 2006, p.351). A technology can increase the image or status of an individual, and therefore the technology can be seen as more useful by individuals.
Next to social influence constructs that affect perceived usefulness, cognitive instrumental constructs will now be discussed which also affect perceived usefulness. The first cognitive instrumental construct is Job Relevance (JR). Job relevance is defined as an individual’s perception regarding the degree to which the target system is applicable to his or her job (Venkatesh and Davis, 2000, p.191). So, if a new technology will help an individual with his or her job or task, the technology will become useful for the individual. Because of the relevance of the technology in job-related tasks, job relevance is included in TAM2 as a predictor of perceived usefulness. The second cognitive instrumental construct is Output
Figure 2. TAM2 with the social influence
Quality (OQ). The quality of the output is very important regarding new technologies
according to Venkatesh and Davis (2000). Satisfied users of products or services, perceive the products or services as useful (Lin, 2005). The third cognitive instrumental construct is Result Demonstrability (RD) and can be defined as the extent to which the tangible results of using an innovation can be observable and communicable (Mun et al., 2006, p.351). If a result of a new technology can be demonstrated to other individuals, they will consider the technology as more useful. Therefore, this variable is considered as a predictor of perceived usefulness because the outcomes of a technology are important regarding the usefulness of a technology. Next to these cognitive instrumental constructs, Venkatesh and Davis (2000) also consider perceived ease of use as a cognitive instrumental construct.
2.4 Integrating Sense of Community
The main question of this research is to what extent does sense of community influence the intention to use new technologies for individuals. To explore this relationship, we first need to define the concept and determine where the concept of sense of community can be placed in TAM2.
Sense of community is a difficult concept to describe, because it is rather intangible. Sense of community can be defined as a sense that individuals get from being part of a community. One of the first that defined community was Finnis (1980). Finnis defined community as a form of a unifying relationship between human beings (Finnis, 1980, p.136). This definition is rather holistic and broad. Other authors confirm that the concept of sense of community is difficult to describe. The definition of the concept remains broad, because even the members of a ‘community’ do not know how to describe the concept in a concrete manner (Wells, Ellis, Slack, and Moufahim, 2017). Kiernan (2017) may shed a different light on the concept of sense of community and states that sometimes the concept is best understood in terms of who is not in your community rather than who is (p.885).
This does not necessarily mean that there is any interaction between them. For example, think about your neighborhood, town, or city (Obst, Smith, and Zinkiewicz, 2002). Communities can also be characterized by (2) social interaction. People in communities regularly interact with each other, but these relationships are not always within a particular geographical area. Internet communities are an example of this. These communities provide a way for a group of peers to communicate with each other by discussions boards on websites, mailing lists, chat rooms, or newsgroups (Eysenbach, 2001). The last factor is (3) identity. Communities that are distinguished by identity share a sense of belonging by a shared set of beliefs or values. For example, a church community or a soccer community.
The term community often brings positive associations along. Occasionally, there is a ‘family’ atmosphere within a community (Brown, 2013). With a community comes a feeling, a sense of community. According to Hanna (1997), sense of community is a natural capacity to identify oneself with a greater community (p.23). So, it is very natural for individuals to have a feeling of belonging. McMillan and Chavis (1986) define sense of community as “a
feeling that members have of belonging, a feeling that members matter to one another and to the group and a shared faith that members’ need will be met through their commitment to be together” (p.8). Members’ needs will be met through the commitment of being together as a
community. Community members care about each other and share a feeling of belonging. This feeling of belonging and care can bring several benefits, such as adjustment, feel supported, have connections, set higher goals, and have stronger levels of social support (Bess, Fisher, Sonn and Bishop, 2002). These benefits can lead to several personal feelings, including the feeling of belonging, the feeling of support, the feeling of having more
connections and the feeling that an individual can do more within a community than alone (Bess et al., 2002).
The feelings that come along with the concept of sense of community can have a possible influence on the adoption of technologies of individuals. For example, community members feel more supported, which results in a buffer against threats, provides a place where individuals are free to express their true identities, and helps them to deal with changes and difficulties in society at large (Rovai and Wighting, 2005, p.100). Individuals behave in a way they normally will not do without the community. When a new technology is introduced in the market, most individuals wait with the purchase until there is more information
new technologies, because introducing a new technology can be viewed as a change to individuals. Individuals that are part of a community a better capable of dealing with changes because of the support of other community members.
Another example can be found in the feeling of belonging. Individuals want to belong to a group and feel accepted (Aronson et al., 2010). Individuals are influenced by other community members - consciously or unconsciously - and imitate behavior in order to become part of the group (Aronson et al., 2010). If certain community members are using a new technology, other community members can get influenced and also want to use the technology. More likely, community members can also get the impression that they should use the technology. Therefore, we can assume that individuals who have a strong feeling of belonging or that are part of a community with a strong feeling of belonging are more likely to adopt new technologies because they can get influenced by their fellow community members or want to be part of the group.
2.5 Research Model
In this research TAM2 is used to answer the main research question. In this chapter the hypotheses of this already existing model will be explained and thereafter the integration of sense of community will be explained. The integration of sense of community into this model can be best explained after the concepts of TAM2 are explained and clearly described.
Therefore, the model will be explained backwards. First, the hypotheses regarding the relationships with intention to use will be further elaborated on. Afterwards the hypothesis about the integration of sense of community will be discussed.
Perceived Usefulness - According to TAM, perceived usefulness has a direct influence on
intention to use (Davis, 1989). Other researchers have gathered empirical evidence that proves that perceived usefulness has impact (e.g. Venkatesh, 1999; Benbasat and Barki, 2007). If a technology is considered as useful by individuals when performing a certain task, the intention to use that technology will be more likely.
2.5.1 Cognitive Instrumental Constructs
According to TAM2 there are cognitive instrumental constructs that influence perceived usefulness of new technologies (Venkatesh and Davis, 2000). The cognitive instrumental constructs that Venkatesh and Davis (2000) distinguish are: perceived ease of use, result demonstrability, output quality, and job relevance. These constructs are seen as cognitive constructs because the constructs compare the capabilities of a technology with the outcomes of the technology (Venkatesh and Davis, 2000). These constructs and the
corresponding hypotheses will be discussed below.
Perceived Ease of Use - The perceived ease of use is related to intention to use according to
TAM (Davis, 1989). The ease of usage of a technology has significant added value for the user. Therefore, the intention of using the technology will be higher, a direct positive effect (Davis, 1989). Besides, there is also empirical evidence accumulated that perceived ease of use is indirectly linked to intention to use via its impact on perceived usefulness (Davis, 1989; Venkatesh and Davis, 2000). A technology that is not too complicated but easy and clear to use will be considered as more useful for individuals, which leads to a higher intention to make use of the technology (Davis, 1989; Kim and Forshythe, 2008). Therefore, we can formulate the following hypotheses:
H2. Perceived ease of use has a positive direct effect on intention to use. H3. Perceived ease of use has a positive direct effect on perceived usefulness.
Result Demonstrability - Result demonstrability is also considered as a predictor of perceived
usefulness according to TAM2. Potential adopters of new technologies view the new technologies and the tangible results of it. Because of the viewed outcomes, adopters are expected to form more positive perceptions of the usefulness of a new technology. Because of the results that are demonstrated by the technology, individuals are more likely to adopt the new technology (Venkatesh and Davis, 2000; Mun et al., 2006).
H4. Result demonstrability has a positive direct effect on perceived usefulness.
Output Quality - In order for an individual to make use of a new technology, the technology
quality is important regarding new technologies. Based upon the output quality of a new technology users can be satisfied or dissatisfied. Satisfied users of a new technology, perceive the new technology as useful (Lin, 2005). If individuals have to choose between different technologies, one would be inclined to choose the system that delivers the highest output quality (Venkatesh and Davis, 2000). According to TAM2, there is a positive direct
relationship between the output quality and the perceived usefulness (Venkatesh and Davis, 2000). Therefore, we can assume that output quality is a predictor of perceived usefulness.
H5. Output quality has a positive direct effect on perceived usefulness.
Job Relevance - Job relevance is defined as an individual’s perception regarding the degree to
which the target system is applicable to his or her job (Venkatesh and Davis, 2000). Because this research focusses not on jobs but on communities, the relevance of the task is used, rather than the job. A new technology can become useful for individuals when the new technology helps an individual with his or her task (Venkatesh and Davis, 2000). Job relevance becomes important when individuals can use the new technology to fulfil their task better or faster.
H6. Job relevance has a positive direct effect on perceived usefulness.
2.5.2 Social Influence Constructs
Besides cognitive instrumental constructs, Venkatesh and Davis (2000) also distinguish two social influence constructs that influence perceived usefulness: image and subjective norm. In this research we add sense of community the model, which is also considered as a social influence construct that has an indirect effect on the perceived usefulness.
Image - As mentioned before, a new technology can enhance the image or status of an
individual (Venkatesh and Davis, 2000; Mun et al., 2006) and individuals like to be
associated or identified with a positive image (Antón, Camarero, and Rodríguez, 2013). New technologies can have personal attributes or psychological associations that determine the image of individuals. Consumers are more likely to adopt new technologies that match their values, beliefs and lifestyle (Kleijen, Ruyter and Wetzels, 2004). The identification with a new technology based upon values, beliefs and lifestyle may lead to the believe of an
H7. Image will have a positive direct effect on perceived usefulness.
Subjective Norm - TAM2 included Subjective Norm (SN) as a predictor of perceived
usefulness. Also, the Theory of Reasoned Action (TRA) identifies subjective norm as one of the two determinants of behavioral intention (Mun, Jackson, Park and Probst, 2006). Some scholars found that subjective norm does not have a significant relationship with intention to use (Mathieson, 1991; Venkatesh and Morris, 2000), while other scholars found that
subjective norm has a significant relationship with intention to use (Hartwick and Barki, 1994; Venkatesh and Davis, 2000). And although Davis et al. (1989) did not found a
significant relationship, Davis et al. (1989) emphasize the need for additional research at this construct.
Venkatesh and Davis (2000) have additional insights on subjective norm and refer to ‘internalization’ when they describe the relationship between subjective norm and perceived usefulness. Internalization is described as “the process by which, when one perceives that an
important referent thinks one should use a system, one incorporates the referent’s belief into one’s own belief structure” (O’Reilly and Chatman, 1986, p.493). In other words, a new
technology is recommended to an individual by others and this individual might be influenced by a recommendation which could positively change the perceived usefulness of that
innovation to that person (Mun et al., 2006). Therefore, we can formulate the following hypothesis:
H8. Subjective Norm will have a positive direct effect on perceived usefulness. H9. Subjective Norm will have a positive direct effect on intention to use.
Sense of Community - As already mentioned in the previous sector, sense of community is
associated with a strong feeling of belonging. Individuals want to belong to a group and feel accepted (Aronson et al., 2010). Group members care about each other, and members can adjust their attitudes and behaviors towards the other members of the group because it is important to them to accept others and feel accepted themselves (Van Bavel and Cunningham, 2012).
kind of influence in the group (Peterson and Martens, 1972). On the other hand, cohesiveness comes together with the group’s ability to influence its members (Kelly and Volkart, 1952).
So, individuals want to be part of a group and are able to adjust their attitudes and behaviors in order to become part of a group. Besides, community members are able to influence the attitudes and behaviors of other community members. It is likely to assume that in order to become and stay part of the group, community members adjust their behavior to the other influential community members. Community members can get the impression that they should perform or should not perform a certain behavior in order to become or stay part of the group. This indicates the positive connection between sense of community and
subjective norm.
H10. Sense of community will have a positive direct effect on subjective norm.
Based on these hypotheses, the following research model can be developed (see Figure 3).
Figure 3. Research model - TAM2 with
3. Methodology
This chapter will start with explaining the setting. Afterwards, the research design will be discussed. Next, the sample method will be shown. Thereafter, the descriptive data of the sample will be viewed. In the last part of this chapter the survey and the survey questions will be explained.
3.1 Setting
In order to understand the context of the research the setting will be discussed. The setting consists of explaining the role of the instructing party and explaining their technology. This research is facilitated by BlockchainProjects B.V., which is located in Deventer, the Netherlands. BlockchainProjects B.V. is a team of Blockchain experts using Blockchain technology and build applications on Blockchain since 2014. BlockchainProjects B.V. focuses on consultancy, development and research. The blockchain technology that is used by this organization has impact on multiple industries such as the financial industry, public sector, supply chain and many more.
Blockchain is a technology that changes the way of making transactions. Nowadays, many transactions that are made are controlled by a third party such as a bank. This
transaction system is centralized, which means that all the information of transactions is controlled by the third party, and not exclusively by the two entities that are making the transactions. In addition, these third parties earn money with each transaction, paid by the entities. Blockchain technology has the possibility to change this manner of executing transactions (Yli-Huumo, Ko, Choi, Park, and Smolander, 2016). Blockchain technology is completely decentralized. Therefore, there is no need of a third party, which leads to quicker and cheaper transactions (Yli-Huumo et al., 2016).
new technology. A good example of this is the Community Wallet, build upon Blockchain technology and created by the instructing party of this research.
BlockchainProjects B.V. is building applications on Blockchain technology. One of the recent developed products of BlockchainProjects B.V. is the Community Wallet. The main idea of the Community Wallet is to bring soccer clubs, their fans, and other stakeholders together. Everyone that is in anyway linked to the club can use the application called
Community Wallet. The Community Wallet offers the production of special club tokens especially made for and produced by the club. Tokens are a substitute for money. With these tokens, supporters can pay their food, drinks, merchandise and so on in the stadium. Next to the use of the Community Wallet inside the stadium, supporters can also use the Community Wallet outside the stadium. For instance, the sponsors of the club can allow the Community Wallet in their transaction system. The Community Wallet offers accessibility for small businesses to become a sponsor of the soccer club. The new transaction system created by the Community Wallet makes it easier for supporters to pay. Besides, the sponsor does not need to pay a great amount of money to become a sponsor. Small businesses can allow their customers to pay with the special club token. These businesses can sell the club tokens back to the soccer club and support the club by making special price arrangements with the club. By using the Community Wallet and the repay system, the sponsor is supporting the club financially only if customers pay with the club tokens. An overview of all the sponsor that accept the special club tokens can be found in the application, the Community Wallet. Another advantage of these club tokens is that the sponsor can exactly see how many
customers pay with the token, and can therefore easily see if the sponsorship with the club is profitable or not. This concept is still in development, but there are numerous possibilities for in the future, also for other clubs and events. The Community Wallet is used in this research at technology adoption.
3.2 Research Design
structured and predefined answers, and the information is only collected from a fraction of the study population (Pinsonneault and Kreamer, 1993).
A survey is chosen as instrument because a survey is relatively low in cost and time, and can reach diverse segments of the population (Kanso, 2000). Various academic
researchers have favored surveys because of efficiency. Thereby, it also avoids interviewer bias, which means that the social nature of the interviewer cannot influence the answers of the respondent. Surveys encourage respondents to give socially undesirable information, such as personal experiences or financial details, and it gives the respondent more time to think about the answers to the questions (Kanso, 2000).
3.3 Sample
To establish a sample, it is necessary to determine which community members will be part of the sample, because there are different kinds of communities. As discussed in the literature review, communities can be distinguished by three different factors: geography, social interaction, and identity (Lee and Newby, 1983; Calvano, 2008; Wells et al., 2017). Communities that are based on identity share the same set of beliefs or values (Lee and Newby, 1983). This research will focus on community based on identity. Communities that are based on identity are associated with a high sense of community and this research tries to determine the influence of sense of community. Communities based on identity are
considered as community with a high sense of membership, influence, fulfillment of needs, and with an emotional connection (McMillan and Chavis, 1986). A good example of communities based on identity are soccer communities. Members of soccer communities share the same beliefs and values, and have experienced the same events together, which are the main characteristics of communities based on identity (Heere and James, 2007).
later, these associations received a reminder to distribute the survey. Unfortunately, only three supporters’ associations were willing to participate, which resulted in a response rate of
10.3%. The three supporters’ associations that were willing to participate consisted of a total of 1150 supporters. From this total amount, 23 supporters responded, which is a response rate of 1,5%. Concluding, a total response rate of 0.02% was achieved via this method.
Besides distributing the survey via supporters’ communities, the survey was also sent to the three soccer clubs itself. One club did not want to participate because they were already participating in a similar research. Another club, agreed to distribute the survey. However, after a while this club refused to participate.
Because the approach to contact the supporters’ associations of the clubs and the club itself was not effective, another approach was added to the procedure. The survey was distributed on Instagram via three different fan accounts from the same three clubs that were contacted. These accounts had a total of 6600 followers. This resulted in another 118
respondents, which is a response rate of 1,8%. Finally, the survey was distributed to another 27 supporters of clubs from several clubs in the Netherlands by means of word of mouth.
3.3.1 Sample Analysis
After establishing a sample consisting of supporters of various soccer clubs, a sample analysis was executed. The survey was filled in by a total of 168 respondents over a period of 20 days. Unfortunately, 42 respondents did not finish the survey. The analysis method of this research, the Structural Equation Model (SEM), prefers a sample size of approximately 200 (Boomsma, 1987). It was not possible to create a sufficient model fit with the data of the respondents that finished the survey. Therefore, data was imputed with the Multiple
Imputation technique. This technique has gained popularity as a statistical tool for handling missing data and is recommended by other researchers (Fishman and Cummings, 2003). Multiple Imputation technique is an attractive method which can be used for missing data in multivariate analyses and was executed in SPSS. It performs 5 data imputes and comes to one data set by averaging all the imputed datasets.
community between the complete respondents and the incomplete respondents. There was no significant difference between the two groups, t(168) = -1.55, p = 0,725. So, sense of
community does not differ between the sample of complete respondents (N=120, M = 2.57, SD = 3.87) and the sample of incomplete respondents (N=48, M = 2.46, SD = 4.32). Based on these results, the missing data was imputed with the Multiple Imputation technique.
In order to address the non-response bias, the dataset was split into two subsets: early and late responses. Late respondents were used as proxy for non-responses and were
compared with the early respondents to see if there were any differences between the two subsets (Wallace and Sheetz, 2014). The results showed that there was no significant difference between the first 42 (25%) respondents and the last 42 (25%) respondents before data imputation. Therefore, we can confirm that the non-response bias was not a concern (Wallace and Sheetz, 2014).
The descriptive data can be found in Table 1. As shown in this table, 73.2% of the respondents was male and 26.8% female. Respondents under 34 consists of the majority of the respondents (86,4%) and we can see that 59% of the respondents are already supporters for more than 10 years.
Item Category Percentage Item Category Percentage
Gender Male 73.2% Supporter
Duration (years) 1 - 2 1.8% Female 26.8% 3 - 4 6.0% Age (years) < 21 56% 5 - 6 8.4% 21 - 34 30.4% 7 - 8 11.9% 35 - 44 6.5% 9 - 10 11.9% 55 - 64 1.8% 11- 12 7.2% 65 > 0.6% 13 -14 14.3% Education None 15.5% 15 - 16 13.7% Mavo/Vmbo 11.9% 17 - 18 4.8% Havo 14.9% 19 - 20 > 22% Vwo 8.9% Money Spent (euros) < 5 30.4% Mbo 21.4% 5 - 10 29.2% Hbo 17.3% 11 - 15 16.7% Wo 6.0% 16 - 20 13.1%
3.4 Measurements
The questionnaire consisted of 9 variables with a total of 35 items. All the items were measured based on a 5-point Likert Scale ranging from Strongly disagree to Strongly agree. This was done because Long and Perkins (2003) strongly recommended to use a 5-point Likert Scale to increase variability and sensitivity of sense of community. The items of the other variables of TAM2 were measured with different scales in the past. In this research, a 5-point Likert Scale was used as used by Wu and Wang (2005) and Halawi and McCarthy (2007). The 5-point Likert Scale is the most used Likert Scale and will not cause confusion amongst respondents (Losby and Wetmore, 2012).
To measure the variable sense of community, the short version of the Sense of Community Index (SCI) was used (Chavis, Hogge, McMillan and Wanersman, 1986). The SCI scale is the most used scale to measure sense of community and can be adapted to different settings (Chipeur and Pretty, 1999). Further, the questions that were used in the survey to measure the other variables were developed by Venkatesh and Davis (2000).
First of all, the questions of the survey were adapted to create suitable questions regarding the setting in this research. Because the questionnaire was distributed amongst supporters of Dutch soccer clubs, the questions were translated into Dutch. Translation into the Dutch language was necessary in order to create understandable questions for the Dutch respondents. The responses that the respondents give, must be similar to the answers that would have been gathered from the questions in the original language (Del Greco, Walop, and Eastridge, 1987). To establish answers in the same manner to the original question, the back-translation method of Werner and Campbell (1970) was used. In back-back-translation, the
researcher asks an individual that is able to write and speak the two languages that were used during this research. This individual was asked to translate the questions into the target language, which was Dutch in this research. Another person translates the questions back into the original questions (English), without seeing the original questions. The researcher is now able to compare the two ‘original’ questions and this creates an impression of the possible answers to the questions in the target language. For the original and adapted questions, see appendix I. For the back-translated questions, see appendix II.
application can be used and what the benefits of its usage are. After the explanation about the wallet, the questions about the other variables were asked. For the survey, which is in Dutch, see appendix III.
Since some items were changed and adapted from previous studies, a pre-test was necessary. Because of the time limitations, this pre-test was conducted on a small scale without any analyses. This pre-test did not lead to big issues. Some items were too long to consist of one block and where changed into two smaller blocks. After the pre-test, the survey was distributed amongst the participants.
3.4.1 Reliability and Validity
Reliability is a measure of trustworthiness or stability of the result of a survey (Wu, Chou, Weng, and Huang, 2011). If the answers are consistent in the tests, the questionnaire’s reliability is high. If the answers are inconsistent, the reliability is low (Chen, 2004). To measure the reliability of the survey, the Cronbach’s alpha was determined. The closer Cronbach’s alpha coefficient is to 1.0, the greater the internal consistency of the scale (Gliem and Gliem, 2003). According to George and Mallery (2003) a Cronbach’s alpha of .8 is probably a reasonable goal. Wu et al. (2011) discovered that .5 - .7 the most commonly observed range of Cronbach’s alpha, followed by .7 - .9.
In the following chapter the original questions and the changed questions will be explained. An explanation for the general changes will be given with a few questions as an example. For the complete overview of the adapted questions, see appendix I.
3.4.2 Independent Variables
Sense of Community - To measure Sense of Community, Chavis, Hogge, McMillan
and Wandersman (1986), developed the Sense of Community Index (SCI). This SCI is
assessed in terms of the four dimensions of the Psychological Sense of Community (PSC) that was proposed by McMillan and Chavis (1986). These four dimensions are: Reinforcement of Needs, Membership, Influence, and Emotional Connection. According to Chipeur and Pretty (1999), various researchers have demonstrated that the SCI can be used in different settings. In this research, the short form of the SCI is used which consists of 12 items.
The SCI, the most popular instrument for empirically measuring the construct (Long and Perkins, 2003), was initially used for adults in the workplace (Pretty, McCarthy and Catano, 1992). But according to Chipeur and Pretty (1999), various researchers have
research to measure the Sense of Community in soccer communities. The scale was adapted to measure the Sense of Community in soccer communities. The statement “People in this
workplace do not share the same values” was reworded to “People in this community do not
share the same values”, and “Very few of my workmates know me” in “Very few of my
community members know me”. It is important that the respondents know that their
community consists of all the various supporters of the of the club. This will be mentioned to them at the beginning of the survey. The measure consists of four subscales according to McMillan and Chavis (1986): Reinforcement of Needs (items 1, 2, 3), Membership (items 4, 5, 6), Influence (items 7, 8, 9), and Emotional Connection (items 10, 11, 12). Sense of community was found to be highly reliable (12 items, α = .82)
Job Relevance - The questions of this variable are used from the study of Venkatesh
and Davis (2000). These questions were changed from “In my job, usage of the system is important” and “In my job, usage of the system is relevant” to “In my community, usage of this application would be important” and “In my community, usage of this application would
be relevant”. These two questions are the only ones in this measurement. Job relevance was
found to be highly reliable (2 items, α = .82).
Output Quality - The questions of this variable are used from the study of Venkatesh
and Davis (2000). These questions were changed from “The quality of the output I get from this system is high” to “The quality of the output I would get from this application is high”, and “I have no problem with the quality of the system’s output” to “I would not have a problem with the quality of the application’s output”. This was done because participants cannot already say what they think of the output of a product that doesn’t exist yet. They can only tell a prediction of what they think. This measurement also consists of two questions. The reliability of output quality was questionable (2 items, α = .66).
Result Demonstrability - The same was done with the Result Demonstrability, because
participants cannot say this already. The questions were taken from the research of Venkatesh and Davis (2000). The question “I would have difficulty explaining why using the system may or may not be beneficial” was changed to “I would have difficulty explaining why using this
application may or may not be beneficial”. The question “The results of using the system are
Image - The last control variable was used by Venkatesh and Davis (2000) and was
measured by the two questions “People in my community who use this application would have a high profile” and “Having this application would be a status symbol in my
community”. These questions were derived from the questions “People in my organization
who use the system have a high profile” and “Having the system is a status symbol in my
organization”. Image was found to be reliable (2 items, α = .79).
3.4.3 Mediators
Subjective Norm - The word ‘would’ was added to these two questions in this
measurement and the system was changed into this application: “People who influence my behavior would think that I should use this application” and “People who are important to me
would think that I should use this application”. The questions were taken from the study of
Venkatesh and Davis (2000) and were very reliable (2 items, α = .88).
Perceived Ease of Use - The questions of this scale were taken from Venkatesh and
Davis (2000). The question “My interaction with the system is clear and understandable”, was changed into “My interaction with the application would be clear and understandable”. And the question “Interaction with the system does not require a lot of my mental effort”, into “Interaction with this application would not require a lot of my mental effort”. The items were highly related (4 items, α = .84).
3.4.4 Dependent Variables
Perceived Usefulness - The same method is used for Perceived Usefulness. The
original questions of Venkatesh and Davis (2000) are changed to fit in this research. “Using
this system in my job would enable me to accomplish tasks more quickly”, was changed into
“Using this application in the stadium would enable me to accomplish tasks more quickly” and “I would find this system useful in my job” into “I would find this application useful in
the stadium”. This measurement had a high reliability (4 items, α = .91).
Intention to Use - The measurement of the intention to use the new technology is
based on the scale of Venkatesh and Davis (2000). It consists of the two items “Assuming I have access to this application, I intend to use it” and “Given that I have access to this
application, I predict that I will use it.”. These questions were changed from the original
4. Data Analysis and Results
4.1 Structural Equation Modeling (SEM)
Because of the complexity of the model in this research with both direct and mediating effects, Structural Equation Modeling (SEM) was used. SEM is a cross-sectional statistic modeling technique, which allows for more flexibility when analyzing data sets compared to many other statistical methods such as ANOVA and multiple regressions (Bollen, 1989, Meehan and Stuart, 2007). In SEM, first a path diagram is scanned to find the dependent variables, the variables with a one-way arrow aimed on them (Bentler, 2010). SEM also measures the overall model fit besides the individual relations between the constructs. To get a good model fit and a complete model, the independent variables are connected through two-way arrows, which implicates that there can be variance or even covariance between them (Bentler, 2010). In order to support hypotheses, the overall model fit must be sufficient. This indicates that if a relationship has a significant p-value, but the overall model fit is not
Factor 1 2 3 4 5 6 7 8
Sense of Community (SOC) (Cronbach’s alpha: .77)
SOC_1 .839
SOC_5 .854
SOC_12 .780
Job Relevance (RJ) (Cronbach’s alpha: .82)
JR_1 .739
JR_2 .922
Output Quality (OQ) (Cronbach’s alpha: .66)
OQ_1 .919
OQ_2 .701
Subjective Norm (SN) (Cronbach’s alpha: .88)
SN_1 .928
SN_2 .922
Image (I) (Cronbach’s alpha: .62)
I_2 .917
I_3 .759
Perceived Ease of Use (PEOU) (Cronbach’s alpha: .74)
PEOU_3 .821
PEOU_4 .851
Perceived Usefulness (PU) (Cronbach’s alpha: .85)
PU_2 .913
PU_3 .926
Intention to Use (IU) (Cronbach’s alpha: .90)
IU_1 .842
IU_2 .889
sufficient, the hypothesis of that relationship cannot be confirmed (Hooper, Coughlan and Mullen, 2008). As described, SEM measures both the individual relationships and the overall model fit, which gives a more complete model instead of other statistical methods (Hooper et al., 2008).
4.1.1 Exploratory Factor Analysis
Before SEM can be used, first an Exploratory Factor Analysis (EFA) was conducted in SPSS. The EFA measures the correlation between the items of the variables in the dataset. A total of 35 items were used to run the EFA. First of all, all the item communalities were above 0.6, meaning that they were not related to other items (Velicer and Fava, 1998). Next, the items that loaded on more than one factor had to be removed. Several items were removed because they had a load of 0.32 or higher on two or more factors (Tabachnick and Fidell, 2001). After these ‘cross loading’ items were deleted, items with a load of less than 0.4 were also removed (Osborne, Costello, and Kellow, 2008). The KMO statistic for this solution was 0.79 (Sign <0.05). These steps resulted in a stable solution with 17 items with 8 factors. Each factor consists of at least 2 items with a loading greater than 0.7. The remaining items of each factor can be seen in Table 2.
4.1.2 Confirmatory Factor Analysis
The first step of SEM is to conduct a Confirmatory Factor Analysis (CFA), which confirms the structure of the EFA. The items of the EFA were confirmed in the CFA because of a good model fit (CMIN = 148.130, df = 91 CFI = 0.95, RMSEA = 0.61, and PCLOSE = 0.148). The reliability and the validity of the CFA were measured and analyzed based on the criteria of Hair, Black, Babin, and Anderson (2010). The results of these tests are in Table 3. The reliability was measured by the Composite Reliability (CR) and the score must be higher than 0.7. The Convergent Validity was measured by the Average Variance Extracted (AVE)
CR AVE MSV
Sense of Community .770 .529 .032
Intention to Use .879 .784 .462
Perceived Usefulness .857 .751 .340
Perceived Ease of Use .736 .582 .375
Image .687 .549 .229
Subjective Norm .865 .762 .472
Output Quality .633 .464 .640
and must be higher than 0.5. At last, the Discriminant Validity was measured by the
Maximum Shared Variance (MSV) and this score must be lower than the AVE. The scores of the variables were all acceptable, except for image (CR = .0687) and output quality (CR = 0.633, AVE = 0.464, MSV > AVE). Because these variables were already narrowed down to two items each, these scores cannot be changed. Therefore, the analysis was continued with the scores of these two variables kept in mind. To measure the variance amongst all observed variables in the model, a method called the common method bias was used. This method uses a Common Latent Factor (CLF) to capture the variance (Richardson, Hettie, Simering, and Sturman, 2009). The unconstrained common method factor model was compared to the fully constrained common method factor model. The chi-square test came out to be significant (difference chi-square = 51.8, difference df = 17, p = 0.000). There was a significance shared variance which indicates that the CLF had to be included in the data imputation (Richardson et al., 2009).
To test the model of this research, SEM was used in the AMOS software. For the sample size of the SEM analysis, Boomsma (1987) suggested that the perfect sample size should be higher than 200, but he also mentions that the sample size between 100-200 would not generate incorrect results. The sample size of this research (N= 168) should not cause any trouble in the SEM analysis. To determine the model fit, several standards can be measured. However, there have to be more standards met to evaluate the whole model fit, not just one. Therefore, Hoyle and Panter (1995) created certain criteria for the evaluation of models. The criteria of Hoyle and Panter (1995) and the values of this research are shown in Table 4. The better the score, the better fit of the model, and the better the results can be generated. The model has a good fit, which means that the results of the model can be trusted.
Measure Value Recommended Value in Model
Χ2/d.f (Normed Chi-square) < 3.0 1.889
GFI (Goodness-of-Fit Index) > 9.0 0.979
AGFI (Adjusted GFI) > 0.8 0.883
NFI (Normed Fit Index) > 0.9 0.984
NNFI (Non-normed Fit Index) > 0.9 0.964
RFI (Relative Fit Index) > 0.9 0.926
IFI (Incremental Fit Index) > 0.9 0.992
RMR (Root Mean Square Residual) < 0.05 0.020
4.2 Results
Figure 4 shows the correlations between the variables. A regression analysis was done
to measure the statistical significance between the constructs The following conclusions can be made when analyzing these tests.
Perceived usefulness had a significant positive effect on intention to use (β = 0.14, p < 0.001), which confirms hypothesis H1. Perceived ease of use did have a significant effect on intention to use (β = 0.51, p < 0.001) but not on perceived usefulness (β = 0.29, p = 0.079). Therefore, hypothesis H2 was confirmed but hypothesis H3 was rejected. Paths that have an influence on the intention to use have a total explained variance of 0.64. Hypothesis H4 could not be tested because the items of this variable (result demonstrability) were not reliable according to the Cronbach’s alpha. Therefore, no conclusions could be drawn regarding to result
demonstrability in this research. The effect of output quality on perceived usefulness could not be supported (β = 0.72, p = 0.100) and the effect of job relevance on perceived usefulness
Figure 4. The proposed model’s test results.
could also not be supported (β = 0.38, p = 0.073). Therefore, hypotheses H5 and H6 could not be confirmed. Image had a significant positive effect on perceived usefulness (β = 0.36, p < 0.001), which confirms hypothesis H7. In contrast to the expectations, subjective norm had a negative effect on perceived usefulness (β = -0.35p < 0.001). Therefore, hypothesis H8 was rejected. Further, subjective norm had a significant positive effect on intention to use (β = 0.43, p < 0.001) and sense of community had a significant positive effect on subjective norm (β = 0.29, p < 0.001). Therefore, hypotheses H9 and H10 could be confirmed.
Another finding in this model was the significant strong positive relationship between job relevance and subjective norm (β = 0.94, p < 0.001). This relationship was not
hypothesized.
In Table 5, every direct, indirect and total effect of the research model is displayed. All the possible relationships are shown and all the shown relationships are significant, except for output quality on intention to use (β = 0.10, p = 0.068). The intention to use is a result
variable determined by sense of community, subjective norm, image, job relevance, perceived ease of use and perceived usefulness. The result show that perceived ease of use has the strongest direct effect (β = 0.51, < 0.001) and the strongest total effect (β = 0.55) on the intention to use. This suggests that the easier it is for people to use a new technology, the more they will intent to use it. The results also show the indirect effect of the sense of
community on the intention to use (β = 0.13, < 0.01), which answers the main question of this research. Further, the highest predictor of perceived usefulness is image (β = 0.36, < 0.001) and the only predictor of subjective norm is sense of community (β = 0.29, < 0.001).
Subjective Norm Perceived Usefulness Intention to Use
Direct Indirect Total Direct Indirect Total Direct Indirect Total
SOC 0.29*** 0.29 - .10** - 0.10 0.13** 0.13 SN - 0.35*** - 0.35 0.43*** - 0.05** 0.38 I 0.36*** 0.36 0.05** 0.05 JR 0.38 0.05* 0.05 RD OQ 0.72 0.10 PEOU 0.29 0.51*** 0.04* 0.55 PU 0.14*** 0.14
5. Discussion
The goal of this research is to find a relationship between the variable Sense of
Community and the TAM2 model of Davis et al. (1989). The results, as shown in the previous chapter, demonstrate that there are several hypotheses that can be supported. In this chapter the main conclusions, the limitations and the implications will be given.
5.1 Findings
The main research question of this research is: “To what extent does sense of
community influence the intention to use new technologies for individuals?”. According to the
SEM analysis, sense of community has a positive indirect relationship with intention to use new technologies (β = 0.13). This relationship is mediated by the subjective norm. This result is in line with the expectations of this research that if there is a high sense of community amongst community members, the feeling that they should perform a certain behavior increases, which leads to a higher intention to use new technologies. In other words, the feeling of belonging, the feeling that members matter to one another and to the group and a shared faith have influence on the intention to make use of a new technology. This
relationship is mediated by the social pressure to perform or not perform a behavior in question, namely to use a new technology.
On the other hand, the findings also suggest that there is a small negative relationship between sense of community and intention to use, mediated by subjective norm and perceived usefulness. This implicates that if there is a high sense of community amongst community members, the feeling that they should perform a certain behavior increases, which leads to a lower perceived usefulness and a lower intention to use. This finding is not in expectation of this research and is in contrast with previous studies and is remarkable (Davis et al., 1989; Mun et al., 2006).
According to Venkatesh and Davis (2000), a negative relationship between subjective norm and perceived usefulness can be explained my means of experience. An individual is gaining experience when using the technology. Experience can ensure that individuals rely less on social information of others and judge the usefulness of the technology on the basis of potential status benefits that individuals gain when using the technology. In the setting of this research this cannot be an explanation, because the application is still in development and cannot be used. It is also possible that respondents associated the Community Wallet with another application that had similar characteristics, and therefore respondents could have had prejudices about the Community Wallet.
Another reasonable explanation can be given by analyzing the rejected hypotheses, which are shown in Table 6. Job relevance, output quality, and perceived ease of use are not significantly related with perceived usefulness. Although the Cronbach’s alpha of the items that measured perceived usefulness was sufficient, there could have been something wrong with the translation and the back-translation of the questions. The questions that were used for the SEM analysis were changed into questions about enjoyment and pleasure of visiting the stadium. These adaption and translation of the questions could have caused misinterpretation by the respondents and could therefore be an explanation for the rejected hypotheses and the unexpected relationship between subjective norm and perceived usefulness.
Other possible reasons for the negative relationship between subjective norm and perceived usefulness are self-report bias and social desirability bias. Respondents had to reflect on their own behavior in the survey regarding subjective norm and perceived
usefulness which can lead to self-report bias. It can be hard for individuals to reflect on their
Hypothesis Result
H1. Perceived usefulness has a positive direct effect on intention to use. Supported H2. Perceived ease of use has a positive direct effect on intention to use. Supported H3. Perceived ease of use has a positive direct effect on perceived usefulness. Rejected H4. Result demonstrability has a positive direct effect on perceived usefulness. N/T H5. Output quality has a positive direct effect on perceived usefulness. Rejected H6. Job relevance has a positive direct effect on perceived usefulness. Rejected H7. Image will have a positive direct effect on perceived usefulness. Supported H8. Subjective Norm will have a positive direct effect on perceived usefulness. Rejected H9. Subjective Norm will have a positive direct effect on intention to use. Supported H10. Sense of community will have a positive direct effect on subjective norm. Supported
own behavior (Sjöberg, 2015). Social desirability is the individuals’ tendency to project themselves in the most favorable way (Huma, Hussain, Thurasamy, and Malik, 2017). In other words, when the respondents answered the questions they could have had problems reflecting on their own behavior and they could have answered the questions in a way to projects themselves differently, which could have led to other answers than was expected.
Besides the main findings of the research, there were also other relationships
confirmed that were based on TAM2. Image has a positive effect on the perceived usefulness, which was in line with the expectations of this research and other authors (Venkatesh and Davis, Wu et al., 2011). This implicates that if an individual’s status can be increased by a technology, the individual will have a higher intention to use that technology. For instance, if individuals have to make a new purchase decision they are more likely to purchase the product with the highest brand image in order for them to gain image amongst peers and others.
According to the conceptual model, intention to use was determined by perceived usefulness and ease of use. Both relationships were confirmed in this research. If a technology is useful and easy to use for individuals, the intention to use the technology will be higher.
A relationship that was not expected in this research was the strong positive relationship between job relevance and subjective norm. This finding assumes that if an individual has the perception that a technology helps him or her to achieve a task, the perception that the individual should perform a certain behavior also increases. This relationship was not found by other authors and can be an interesting topic for further research.
5.2 Limitations
individuals about the application and the results of the application. Respondents could not give clear answers on questions about recommending the application (the Community Wallet) to others. An explanation of this outcome of the survey could be that the application is still intangible and not in use. Fourth, the data collection method was insufficient. Therefore, other respondents were contacted via social media. This resulted in a very young sample
population. 86.4 % of the respondents were under 34. The sample population of this research is not representative for the Dutch soccer communities, were the average age of a supporter is 45. A possible explanation of a younger sample is the online distribution of the survey. Younger supporters have easier access to the Internet and social media. Future research could take a larger sample size with more variety in age in order to create a more generalizable sample. Fifth, self-report questions were used in the survey and this can possibly lead to biases (Sjöberg, 2015). The respondents filled in the survey and reflected on their own behavior. Respondents can be less likely to be honest about their own behavior and are less likely to give accurate answers. Especially when the questions are about their own behavior, numerous respondents would not admit their real behavior or even do not really know how to reflect on their own behavior. Participants may also vary regarding their interpretation of particular questions (Van de Mortel, 2008). Recommended for future research is conducting follow up interviews in order to prevent the self-bias (Van Mortel, 2008). Sixth, the setting of this research was very specific. The variable sense of community was measured amongst soccer communities of a few clubs in the Netherlands. Future research is recommended to broaden this setting. For example, a research within other sorts of communities and in other countries can be the basis for future research. Lastly, translating the original questions from existing research into Dutch was very challenging. To ensure that similar answers would be given to the translated questions compared to the original questions, the back-translation method of Werner and Campbell (1970) was used. However, differences in interpretations and word choices still exists. Therefore, future research should consider using questions that do not need translation in order to capture the actual meaning of the original question (Del Greco, 1987).
5.3 Implications
