The perceived value of Location-Based Services
by
Marcia Snijder
July 2012
University of Groningen
Faculty of Economics and Business
MSc Business Administration
Marketing Management & Research
MANAGEMENT SUMMARY
The main objective of this study is to identify the factors that influence the perceived value and consumers’ adoption intention of location-based services. Several authors researched some constructs of this study. However, not one author researched all constructs together. The variables that were used to determine the perceived value and adoption intention of LBS are: perceived service compability, perceived usefulness, perceived enjoyment, behavioral attitude towards LBS, perceived ease of use, subjective norms, perceived risk and privacy concerns. The research question is: Which variables influence the perceived value and adoption intention of location-based services?
In order to investigate the variables which might have an influence on the perceived value and adoption intention of LBS an empirical research is conducted. At first, a factor analysis is done to obtain a structure among the constructs. However, some constructs were grouped together in one component. So, the original constructs were used to analyze the constructs in the other analyses and to decide whether the hypotheses are consistent with the output of the analyses. Moreover, multiple regression analyses were done to investigate which variables have an influence on the perceived value and adoption intention. Besides, a moderator analysis is performed, which confirmed that there is no significantly effect of subjective norms on the adoption intention of LBS. Finally, a mediation analysis is done, the output of this analysis showed that perceived usefulness and enjoyment have a mediating effect on perceived value.
The outcomes of the empirical research showed that the variables that have an influence on adoption intention of LBS are behavioral attitude towards LBS, privacy concerns and perceived service compability and on perceived value are perceived usefulness and perceived enjoyment. This gives an answer on the research question of this study.
PREFACE
This thesis is written to complete the Master Marketing Management and Marketing Research within the study Business Administration at the University of Groningen. After graduating at the Hanzehogeschool in Communication, I decided to start with the Pre-Msc Marketing to increase my knowledge on Marketing. I am very glad that I took this opportunity. With this thesis, my time as a student finally comes to an end. Therefore, I would like to thank some persons who made it possible for me to graduate.
First of all, I especially would like to thank Dr. Sonja Gensler for her clear advices, help and feedback during the process. I also would like to thank the second supervisor, Daniela Naydenova, for her suggestions. Finally, I would like to thank my parents, sister, boyfriend and friends for their support during my study period. Thanks to their support and encouragement I have been able to finish this study. Thank you!
Marcia Snijder
TABLE OF CONTENTS
1. INTRODUCTION ... 5
2. LITERATURE REVIEW ... 10
2.1 Theoretical foundations in innovation diffusion ... 10
2.1.1 Theory of Reasoned Action ... 10
2.1.2 Theory of Planned Behaviour ... 11
2.1.3 Technology Acceptance Model ... 12
2.1.4 Comparison of the models ... 13
2.2 Perceived value ... 14
2.3 Adoption resistance ... 18
2.3.1. Subjective norms ... 18
2.3.2. Perceived ease of use ... 19
2.3.3. Behavioral attitude ... 20 2.3.4. Perceived risk ... 20 2.3.5. Privacy concerns ... 21 2.4 Adoption intention ... 23 2.5 Conceptual model ... 24 3. METHODOLOGY ... 25 3.1 Research design ... 25 3.2 Sampling technique ... 27 3.3 Measures ... 27 4. RESULTS ... 30 4.1 Descriptive statistics ... 30 4.2 Factor analysis ... 31 4.3 Correlations ... 34 4.4 Regression assumptions ... 35
4.4.1 Linearity and heteroscedasticity ... 35
4.4.2 Non-normality ... 36
4.5 Multiple regression analysis ... 37
4.6 Moderator analysis ... 40
4.7 Mediation analysis ... 40
5 CONCLUSIONS AND RECOMMENDATIONS ... 42
5.1 Recommendations ... 45
5.2 Limitations and suggestions for future research ... 46
REFERENCES ... 48
APPENDIX ... 63
Appendix 1 – Overview of Utilized Literature ... 63
Appendix 2 – Questionnaire ... 68
Appendix 3 – Factor loadings ... 74
Appendix 4 – Correlations ... 76
1. INTRODUCTION
The interactive marketing landscape has rapidly evolved over the past years with the increased usage of the location-enabled mobile web functionality of smartphones. There has been considerable growth in recent years both in the adoption of portable and wireless mobile devices and in the various ways to use those devices (e.g., text, email, video, navigation, and camera). Mobile phones which are more powerful than previous generations of desktop computers are becoming travelling companions for consumers, accompanying them wherever they go (Hennig-Thurau et al., 2010). Smartphones, defined as mobile phones with a mobile operating system and the ability to run applications, have exhibited strong growth rates with 53.8 million units sold worldwide in 2009, an increase of 41% year over year (Gartner, 2010). As of the end of 2009, 21% of US consumers owned a smartphone, and this growing segment mobile devices is forecast to overtake feature phones by the end of 2011 (Nielsen, 2010). Thus, consumers are rapidly adopting smartphone technology, which increases the interactive capabilities readily available on their mobile handsets via manufacturer included technology and add-on downloadable software applications.
do have control over type of ads (Tripathi and Nair, 2006). Market research studies have shown that prospects are only willing to receive ads on wireless devices when they have some control over the volume and type of messages (Andersson and Nilsson, 2000). This underscores the need to constrain the number of ads delivered. Barwise and Strong (2002) conduct a survey and find that prospects are willing to receive about three wireless ads per day.
simple location-based notification service to an integrated mobile platform for customer rewards, real-time mobile sampling, and loyalty programs, providing marketing practitioners innovative new ways to reach consumers based on their location (Humphrey and Laverie, 2011).
Moreover, location-based services are an interesting research topic (e.g. Balasubramanian, Peterson, and Jarvenpaa, 2002; Pura, 2005). Because companies have only begun to understand the economic potential that comes with location-based services. Effective marketing requires good knowledge of the underlying needs and value perceptions of the specific user segments. People who do not have any experience of using such services may have difficulty in comprehending the real value of location-based information (Pura, 2005). Implications are needed also to point out what type of value is important in the location-based context from a customer’s point (Pura, 2005). Additionally, perhaps the most interesting aspect of m-commerce is the potential for launching services that are, in essence, of value only through a mobile medium, as needs for such services predominantly arise when away from home and on the move. For example, the widely discussed location-based services such as routing and tracking/pinpointing. As these types of services are of value exclusively in mobile settings. Anckar and D'Incau (2002) believe that they are likely to constitute the core of the mobile commerce value proposition.
Empirical findings by Anckar (2002) indicate, in fact, that electronic channel adoption or rejection decisions by consumers are determined by their perceived value of a channel in comparison to existing alternatives, and thus that customer perceived value is a relevant construct in terms of channel adoption or rejection decisions. As noted by Keen and Mackintosh (2001) the demand side of m-commerce is a search for value, and hence there is a need to build an understanding of the elements and special features of wireless electronic channels that are value-adding from the consumer’s point of view.
Values define the key features of the services that are appreciated by the users. To study the adoption intention existing innovation adoption models will be discussed in the literature section. Several models will be used to give a complete literature review in order to determine the adoption intention of location-based services. Adoption intention instead of adoption behaviour is used as the dependent variable because adoption intention is more appropriate when the product is still in the early stages of its diffusion cycle (Hong and Tam, 2006). Furthermore, intention has been well established as a good predictor of behaviour that mediates the effect of other determinants on behaviour (Ajzen, 1991; Sheppard et al., 1988; Venkatesh and Brown, 2001; Venkatesh et al., 2003). Besides, the factor privacy concerns is an important barrier to adopt the service. Only little empirical research is conducted about privacy issues in location-based services. Customers need to understand how LBS bring value to their everyday performance in order to adapt to these new services (Kaasinen, 2005). Hence, it is essential to study which features customers find important to use location-based services. Examples of features are downloading maps, location aware directory services, emergency services, friend finder, locating potential dating partners/friends, route guidance, interface with points awarded, loyalty program, and real-time mobile sampling. Besides, the distinctive feature of mobile commerce is the significance of the user’s location, his situation, and his mission, and to gain an understanding of the drivers for consumer adoption and usage of m-commerce services (May, 2001).
2. LITERATURE REVIEW
This chapter contains the theoretical framework regarding the several focuses of this study. The chapter starts with a review of the literature about theoretical foundations in innovation diffusion. Then the literature of perceived value is described because it provides a good background for assessing location-based services and what consumers find important. After that the adoption intention and resistance will be described. The several factors of this study are concluded with a hypothesis of what the expected effect on location-based services will be. Finally, the chapter concludes with a conceptual framework of the literature review and in Appendix 1 a complete overview of the utilized literature can be found.
2.1 Theoretical foundations in innovation diffusion
Several theories which describe the factors influencing consumer behaviour exist in marketing, psychology and information systems literature. For the past two decades, a variety of theoretical perspectives have been advances to address individual adoption of IT innovations (Hong and Tam, 2006). The focus in technology related literature is on the theory of reasoned action (TRA) and the theory of planned behaviour (TPB), which are intention models that have proven to be good models for predicting and explaining behaviour in a large number of domains (Fishbein and Ajzen, 1975; Ajzen, 1991). Besides, another theory has come forward that is successful in explaining and predicting technology acceptance behaviour is the technology acceptance model (TAM) (Davis, 1989). With a few exceptions, the main focus of these research perspectives have been confined to understanding adoption processes within organizational settings, where IT has been regarded as a tool to improve task performance (Hong and Tam, 2006). All three models are widely applied to explain and predict behaviour of consumers with new technology. Therefore, the several models are in more depth described in the next paragraphs.
2.1.1 Theory of Reasoned Action
motivation to comply with the expectations of these referents. Subjective norms refer to an individual’s perceptions of other people’s opinions on whether or not he or she should perform a particular behaviour (Ajzen and Madden, 1986). TRA provides a relatively simple basis for identifying where and how target consumers’ behavioural change attempts (Sheppard, Hartwick and Warshaw, 1988). TRA is a general model, and as such, it does not specify the beliefs that are operative for a particular behaviour. With TRA, one must first identify the beliefs that are salient for participants concerning the behaviour under investigation. TRA is very general, which is designed to explain virtually any human behaviour (Ajzen and Fishbein, 1980). According to the TRA, displayed in figure 1, behavioral intention (BI) is the direct antecedent of an individual’s actual behaviour (B).
Figure 1: Theory of Reasoned Action (Fishbein and Ajzen, 1975)
2.1.2 Theory of Planned Behaviour
TRA is criticized for neglecting the importance of social factors that in real life could be a determinant for individual behaviour (Grandon and Mykytyn, 2004; Werner 2004). Social factors indicate all the influences of the environment surrounding the individual (such as norms) which may influence the individual behaviour (Ajzen 1991). To overcome this weakness in TRA, TPB was introduced by Ajzen (1991). The theory of planned behaviour is an extension of TRA. In TPB an additional factor in determining individual behaviour is added, which is Perceived Behavioural Control. Perceived Behavioural Control directly influences behavioural intention and actual behaviour (see figure 2).
Figure 2: Theory of Planned Behaviour (Ajzen, 1991)
2.1.3 Technology Acceptance Model
The Technology Acceptance Model (TAM) by Davis (1989) is inspired by the Theory of Reasoned Action of Fishbein and Ajzen (1975). TAM defines a framework to study user acceptance of new technology based on perceived usefulness and perceived ease of use (Davis et al., 1989). The reason why this model is so successful is that system designers have some degree of control over these constructs; usefulness and ease of use of products or services. Perceived usefulness (PU) is defined as the degree to which an individual believes that using a particular system would enhance his or her job performance. Perceived ease-of-use (EoU) is referred to as the degree to which a person believes that using a particular system would be free from effort. TAM helps to explain, as well as predict user behaviour by tracing the impact of external factors on internal beliefs, attitudes and intentions (Davis et al., 1989).
2.1.4 Comparison of the models
Figure 4: Compared model of TRA, TPB and TAM
2.2 Perceived value
Perceived value theories provide a good background for assessing mobile services and the value of the content from a customer’s point of view. There is a need to build an understanding of the elements and special features of mobile electronic channels that are value-adding from the consumer’s point of view. Currently, a consumer does not know the exact value of location-based services. That is why it is important to study perceived value. Moreover, perceived value is a significant factor affecting user adoption of location-based services. In the previous models (e.g. TPB and TAM) the dependent variable is actual use. However, not many consumers are using location-based services. So, the adoption rate of the location-based services is still very low and therefore it is interesting to study the reason of these low rates by using perceived value as the mediator and adoption intention as dependent variable. Adoption intention is more appropriate when the product is still in the early stages of its diffusion cycle (Hong and Tam, 2006). In addition, focusing on perceived value gives a good foundation to attract people who share similar value perceptions, not just an attitude to technology in general (Pura, 2005). Perceived value has been proven to be a reliable construct in predicting purchase behaviour. It also gives practical implications how to market services to end-users and demonstrate concrete benefits and value in specific contexts (Pura, 2005). Perceived value reflects a user’s overall assessment of location-based services based on perceptions on what is received and what is given (Zeithaml, 1988). In other words, value has been seen as the trade-off between benefit and sacrifice in an offering. Across a variety of service contexts service value consistently mediates the impact of service quality (i.e.,
Behavioral intention
TPB Control beliefs Perceived
benefits) on behavioural intentions (Cronin et al., 2000; Brady et al., 2005). Whereas value generically has been conceptualized as a cost or benefit trade-off, service value consistently is modelled as the difference between service quality attributes and sacrifice (Brady et al., 2005; Sweeney et al., 1999). In this sense, sacrifice represents a broad construct that comprises monetary and nonmonetary costs, such as effort and risk perceptions. Possible benefits which represents the relative advantages of mobile services are time convenience, user control, and service compatibility, whereas possible costs include technicality, perceived fees, risk, and cognitive effort (Kim, Chan, & Gupta, 2007; Kleijnen, de Ruyter, & Wetzels, 2007). The choice of benefits is corroborated by Hourahine and Howard (2004), who reveal that convenience, increased control, and need fulfilment are the main advantages reported by mobile consumers.
The definition of perceived value by comparing benefits with sacrifices is an indicator of adoption intention. On the other hand, Thaler’s (1985) model of consumer choice is a combination of economic reasoning and cognitive psychology. The value function is psychologically based and replaces the utility function from economics theory. The central principle of value function is that it is defined over perceived gains and losses relative to some natural reference point, suggesting that people tend to respond to cognitive comparisons rather than absolute levels, and that it is steeper for losses than for gains, signifying that sacrifices hurt more than the pleasure given by the benefits. There is strong empirical support that perceived value affects perceptual intention to use (Sweeney et al., 1997).
integration, consuming as classification and consuming as play. This typology offers ways to examine how perceived value is located in a possession. Thus, value is derived through individuals’ cognitive and affective perceptions, interpretations and responses arising from their contextualized meanings in the everyday consumption practices surrounding the possession (Andrews et al., 2011). Pura (2005) identifies that commitment and behavioural intentions towards location based services are strongly influenced by conditional and emotional value in specific use contexts, such as when a service is important to the consumer. By understanding what aspects of perceived value are important to their consumers, mobile services providers can market their products in ways that create value for customers in their daily activities.
views perceived usefulness as outcome expectancy and a measure of extrinsic motivation (Venkatesh, 1999). Individuals evaluate the consequences of their behaviour in terms of perceived usefulness and base their choice of behaviour on the desirability of the usefulness. Performance expectancies such as perceived usefulness, which focuses on task accomplishment, reflect the desire of an individual to engage in an activity because of external rewards (Venkatesh et al., 2003). For example, if consumers need to use a map then they find LBS useful. The usefulness construct has been used extensively in information systems and technology research, and has strong empirical support as an important predictor of technology adoption (e.g., Mori, 2001; Szajna, 1996).
Individuals, who experience immediate pleasure or joy from using a technology and perceive any activity involving the technology to be personally enjoyable in its own right aside from the instrumental value of the technology, are more likely to adopt the technology and use it more extensively than others (Davis et al., 1989). This notion is in line with popular definitions of emotional value. Sweeney and Soutar (2001) defined emotional value as the utility derived from feelings or affective states that a product generates. Enjoyment refers to the extent to which the activity of using a product is perceived to be enjoyable in its own right, apart from any performance consequences that may be anticipated (Davis et al., 1992). Thus, enjoyment represents an affective and intrinsic benefit. Past research has indicated that the benefit component comprises perceived enjoyment, in addition to perceived usefulness. Besides, enjoyment and fun have a significant effect on technology acceptance beyond usefulness (Davis et al., 1989 and Sweeney and Soutar, 2001). In this study it is suggested that the constructs usefulness and enjoyment has a positive effect on perceived value in contrast to the TRA, TPB and TAM models. Based on above literature one might believe that the constructs also have an effect on perceived value. Therefore, the following hypotheses are expected.
H1a: Perceived usefulness has a positive effect on perceived value of location-based services. H1b: Perceived enjoyment has a positive effect on perceived value of location-based services.
2.3 Adoption resistance
Despite company efforts to adopt consumer-oriented innovation development processes focused on delivering added value to the consumer, most commercial companies are faced with high rates of innovation failures (Danneels, 2003; Moore, 2002). This is puzzling, as innovation adoption research has stressed that relative advantage is a dominant driver of consumer adoption. Nevertheless, there is still resistance. One key issue of concern is that resistance includes not trying the innovation (Nabih et al., 1997; Ram & Sheth, 1989; Szmigin & Foxall, 1998). This is problematic because Rogers (2003) points out that initial objection toward an innovation can sometimes be overcome by offering consumers the opportunity to try the innovation for a certain period of time.
2.3.1. Subjective norms
when the use of technology is voluntary or not. When the usage of a system consists of mandatory users it would give greater weight to the opinions of others, since they think frequent use is appropriate. The intention of users under voluntary usage will be more based on their own attitudes (Hartwick and Barki, 1994). The usage of location-based services is voluntary, so in this research subjective norms have an indirect influence on LBS because most people use LBS to acquire information but it is not sure if they are influenced by others to use it. Therefore, subjective norms are a moderator. The social influence that leads an individual to use a technology has been found to affect adoption intention (e.g., Venkatesh and Morris, 2000; Venkatesh et al., 2003). Therefore, the following hypothesis is expected:
H2: Subjective norms will have an moderating effect on adoption intention.
2.3.2. Perceived ease of use
H3: Perceived ease of use is positively related to adoption intention of location-based
services.
2.3.3. Behavioral attitude
The behavioral intention is, according to Fishbein and Ajzen (1975), determined by attitude and subjective norm. Attitude can be defined as an individual’s positive evaluation of the performance effect of a particular behaviour (Ajzen and Madden, 1986). Harrison et al. (1997) showed that all three TPB constructs (subjective norms, attitude, perceived behavioral control) predict unique variance in adoption intention. However, the construct attitude had the strongest effect on the adoption intention. Attitude is a psychological tendency that is expressed by evaluating a particular entity with some degree of favour or disfavour (Eagly and Chaiken, 1993). High values of attitude would be indicative of an individual's intention to adopt the target technology (Nasco et al., 2008). The positive relationship between attitude and predicting adoption intentions has been well established in marketing literature (Harrison et al., 1997; Pavlou and Fygenson, 2006; Venkatesh et al., 2003). For example, if a consumer has a favourable attitude towards LBS than that the adoption intention of a consumer might be increased. Therefore, the following hypothesis is expected:
H4: Behavioral attitude has a positive influence on the adoption intention of location-based
services.
2.3.4. Perceived risk
consumer resistance toward innovations. Interestingly, a common consumer risk reduction strategy for consumers is the pursuit of information to increase knowledge about these risks and their solutions (Dowling and Staelin, 1994; Locander and Hermann, 1979). For example, new media can be particularly helpful in providing demonstrations of how new products work in virtual environments, illustrating how products can be incorporated in existing habits and living situations (Kleijnen et al., 2009). Kleijnen et al. (2009) describes also social and economic risk in their study. Economic risk is related to the costs of an innovation and that the innovation will be a waste of economic resources (Kleijnen et al., 2009). Location-based services does not require a high investment of the consumers, so this is not relevant for this study. Besides, social risk refers to the concern that the innovation will not be approved by others (Kleijnen et al., 2009). Subjective norms is already a moderator in this study, because it might has an indirect effect on LBS. Therefore, social risk will not be researched in this study. Hence, for this study only perceived product performance risk and psychological risk refers to location-based services. Perceived product performance risk (functional risk) suggests a negative relationship between performance risk and resistance (Ram and Sheth, 1989). Product performance risk is defined as the loss incurred when a brand or product does not perform as expected. Psychological risk may refer to disappointment, frustration, and shame experienced if one’s personal information is disclosed (Forsythe and Shi, 2003). Hence, perceived functional and psychological risk should have a negative effect on adoption intentions (Kleijnen et al., 2007). Therefore, the following hypotheses are derived:
H5a: Greater perceived product performance risk is associated with lower adoption intention
of location-based services.
H5b: Greater psychological risk is associated with lower adoption intention of location-based
services.
2.3.5. Privacy concerns
Location information often reveals the position of a person in real time, thus rendering the potential intrusion of privacy a more critical and acute concern (Clarke, 2001; Danezis et al., 2005). These concerns pertain to the confidentiality of accumulated consumer data and the potential risks that consumers experience over the possible breach of confidentiality (Clarke 2001; Xu and Teo 2004; Xu et al., 2005). The construct privacy concerns is one of the most widely used in research and it is often used as a proxy for the concept of privacy Today, privacy concerns often arise when new IT with enhanced capabilities for collection, storage, use, and communication of personal information arises (Culnan, 1993). Location-based services also collect and storage personal information by using their location, which is a great barrier for using LBS. However, it has been argued that users are typically unaware that their personal information will be used, and therefore the violation of personal privacy becomes a concern. Xu and Gupta (2009) conceptualize privacy concerns as the degree to which an individual believes that the organizational and technical infrastructure exists to prevent privacy breach. As individual consumers may not be able to fully exercise their beliefs regarding privacy and given its importance in sustained commercial activities, the safeguard of information privacy in commercial transactions has fallen into the domain of governmental entities such as the Federal Trade Commission (FTC). The FTC provides a set of guidelines known as the Fair Information Practices (Gillin, 2000). The guidelines incorporate rules that define how vendors should collect information, how they should fix any errors regarding personal information, how they should inform consumers regarding subsequent use of their information and how the vendors should prevent any unauthorized access to information. Similarly, consistent with findings of Culnan and Armstrong (1999) and Hoffman et al. (1999), the guidelines require that vendors should provide consumers control over all aspects of information collection and usage. The guidelines can be summed up into five principal actions: notice, choice, access, integrity, and enforcement.
secondary purposes without consent; 3) errors reflected the concern that protections against deliberate and accidental errors in personal data are inadequate; and 4) improper access reflected the concern that data about individuals are readily available to people not duly authorized to view or work with data. Moreover, the personal information can be used for nefarious purposes thus encroaching into a person’s personal life. Especially in today’s world, both private corporations and government agencies take advantage of the powerful surveillance means to track and profile consumers and citizens through mobile devices (Dinev et al., 2008; Levy 2004). Consumers are more fearful about disclosing personal information in the seamless electronic environment (Dinev et al., 2008). The negative impact of privacy concerns on behavioural intention has been empirically supported in the e-commerce context (Chellappa and Sin 2005; Dinev and Hart 2006a; Malhotra et al., 2004). In addition, consumers with a high level of privacy concerns tend to have a negative attitude towards location-based services. Hence, the following hypothesis is derived:
H6: Privacy concerns are negatively related to adoption intention of location-based services.
The next section describes the literature review of adoption intention.
2.4 Adoption intention
that will form a person’s attitude towards an innovation. This conceptualization of compatibility also has been used in relation to mobile technology adoption (e.g., Kleijnen et al. 2004). According to Laforet and Li (2005), consumers use mobile channels primarily because of the opportunity they offer to fulfil specific service needs. For consumers there are several needs, like the ability to access the service while they are walking along within in a city to receive the information they need. So, if the needs of the consumers are fulfilled with using a location-based service it is associated with a higher adoption intention. Therefore, the following hypothesis is expected:
H7: Greater perceived service compatibility is associated with higher adoption intention of
location-based services.
2.5 Conceptual model
Based on the literature review the conceptual model is constructed. The mediator perceived value consists of the construct usefulness and enjoyment. The constructs service compability, perceived ease of use and behavioral attitude relates positively to the dependent variable adoption intention. However, the construct privacy concerns and perceived risk are a barrier for adopting location-based services. Moreover, subjective norms is the moderator. Figure 5 displays the expected relationships between the constructs. Figure 5: Conceptual model
Perceived value of LBS H2 (+) H1a H1b (+) H5a H5b H6 (-) Subjective Norms
3. METHODOLOGY
This chapter contains the methodology part of this study. At first, the research method will be explained. Second, the sampling technique will be described. Finally, the indicated measures that will be used for the questionnaire are given.
3.1 Research design
There are several methods available for doing research. Some well-known methods to gather data are surveys, experiments and literature studies (Malhotra, 2007). For this research a survey (ThesisTools) will be used with questions about the constructs that are related to perceived value and adoption intention of location-based services. The reason for choosing a survey is that it enables to gather a large amount of data in a relatively short time span and it will help to test the relationships between the different constructs. It is useful to conduct a survey to find out what the thoughts of the consumers are about location-based services. Besides, the questionnaire is simple to administer and the data obtained are reliable because the responses are limited to the alternatives stated (Malhotra, 2007). The survey is based on one cell of analysis and consists of a structured direct questionnaire (Moutinho, 1998). A pre-test with 8 students of the University of Groningen is conducted to validate the appropriateness of the questionnaire. After completing the questionnaire, feedback from the participants is used to elect possible discrepancies. Moreover, the age of the group is between 18 and 65 years, to biasness towards a special age group and generation.
Besides, a multiple regression analysis will be applied to examine the relationships in the conceptual model. Hence, to test the hypotheses. Regression analysis can be used to statistically analyze associative relationships between a metric dependent variable and one or more independent variables (Malhotra 2007). The reason to use a multiple regression analysis is that in this study multiple independent variables need to be examined. The p-value from the ANOVA table will be used to determine whether the model is overall significant or not. When the model is indeed significant, the R-square will be taken to decide whether the independent variables have an effect on the dependent variable. The closer the number is to 1, the greater this effect is. The separate coefficients will be used to decide upon the level of influence of each variable and the significance number tells whether the level of influence is significant for the model that is tested. Throughout this thesis, a significance number of .05 will be held (Malhotra, 2009). The target group are consumers who have a mobile phone. Besides, multiple regression analysis can be complicated by the presence of multicollinearity, which are the intercorrelations among the predictors that are used in the model (e.g. de independent variables) (Malhotra, 2009). The higher the multicollinearity in the regression, the less precisely the partial regression coefficients can be estimated and the more difficult it is to assess the relative importance of the independent variables in explaining the variation in the dependent variable. Even though, each of the independent variables may be a relevant predictor, these variables can correlate among each other. If the value of tolerance is less than 0,1, so when the variance inflation factor (VIF) is higher than 10, the collinearity is problematic and one of the independent variables can better be excluded from the multiple regression analysis (Field, 2000).
Moreover, the following three models will be tested when conducting multiple regression analyses, mediation analysis and moderator analysis.
Model 1: Perceived value =
i ε i β i β α Enjoyment 2 Usefulness 1
Model 2: Mediating effect of perceived value =
i ε i i i β i β
α 1Usefulness 2Enjoyment 3Value 4Intention Model 3: Adoption intention =
i i i i i i i
i 2Attitude 3Use 4Risk 5Concerns 6Value 7Norms y
Compabilit 1
3.2 Sampling technique
Family, friends, Master students and acquaintances are invited by email and a message on social network site Facebook.com, to participate in the survey. In the email and in the message there is a link to the questionnaire. The self-administered online survey will be distributed with the help of a snowball system (Bloomberg, Cooper and Schindler, 2005). So, the respondents are asked to participate in the research themselves and to forward the email or message to people they know to participate in the survey as well. The condition to actually participate in the online survey is that the respondents should have a mobile phone.
3.3 Measures
Lastly, seven items for measuring privacy concerns are used from Smith et al. (1996) which are transformed by Xu and Gupta (2009) (questions 32-40, see Appendix 2). After the measurement of all constructs, several questions will be asked about location-based services and three demographic questions will be asked to the respondents.
Table 1: Measurement Scales
Constructs Source
Adoption intention (INT)
INT 1 Probability to plan to use location-based service in the future. Davis et al. (1989) adopted by Kim et al. (2007) and adapted by myself
Perceived value (VAL)
VAL 1 Compared to the effort I need to put in, the use of location-based service is beneficial to me.
VAL 2 Compared to the time I need to spend, the use of location-based service is worthwhile to me.
VAL3 Overall, the use of location-based service delivers me good value
Sirdeshmukh
et al. (2002) adopted by Kim et al. (2007)
Perceived service compatibility (PSC)
PSC 1 Using location-based service is compatible with my lifestyle.
PSC 2 Using location-based service is completely compatible with my needs. PSC 3 Location-based service fits well with the way I like to get things done.
Moore and Benbasat (1991), adapted by Meuter al. (2005)
Perceived usefulness (PU)
PU 1 Using location-based service would increase my chances of achieving things that are important to me.
PU 2 I would find location-based service to be useful in my daily life.
PU 3 Using location-based service would help me accomplish things more quickly.
Davis (1989) adopted by Hong and Tam (2006)
Perceived enjoyment (PENJ)
PENJ 1 I expect that using location-based service would be enjoyable. PENJ 2 I expect that using location-based service would be pleasurable. PENJ 3 I expect to have fun using location-based service.
PENJ 4 I expect that using location-based service would be interesting.
Behavioral attitude towards LBS (BATL)
Please express your attitude towards location-based service.
BATL 1 1 = Unfavorable, 5 = Favorable BATL 2 1 = Boring, 5 = Interesting
BATL 3 1 = Dislike very much, 5 = Like very much BATL 4 1 = Very worthless, 5 = Very valuable
Perceived ease of use (PEU)
Karson and Fisher (2005) and Gill et al. (1988)
PEU 1 I expect that learning how to use location-based service would be easy for me. PEU 2 I expect that my interaction with location-based service would be clear and understandable.
PEU 3 I would find location-based service to be easy to use.
PEU 4 I expect that it would be easy for me to become skilful at using location-based service.
Subjective norms (SN)
SN 1 People who are important to me would want me to use location-based service. SN 2 People who influence my behaviour would think I should use location-based service.
SN 3 People whose opinions I value would prefer me to use location-based service.
Davis (1989), adopted by Hong and Tam (2006)
Mathieson (1991) adopted by Hong and Tam (2006)
Perceived risk (PR)
PR 1 Difficult to judge quality of location-based service (perceived product performance risk).
PR 2 Do not trust that my personal information will be kept private (perceived psychological risk).
Forsythe and Shi (2003)
Privacy concerns (PC)
PC 1 It bothers me to disclose my personal information to service providers.
PC 2 I am concerned that other people may monitor my current location continuously. PC 3 Service providers are collecting too much information about me.
PC 4 Service providers may keep my private information (including my location) in a non-secure manner.
PC 5 Service providers may not take measures to prevent unauthorized access to my personal information.
PC 6 Service providers may reveal my personal information to unauthorized parties without my consent.
PC 7 Service providers may use my personal information for other purposes, e.g., analyzing my daily activities to derive information about me
4. RESULTS
In this chapter the results of this study will be discussed. First, the descriptive statistics will be described. In the next paragraph the results of the factor analysis will be discussed. Then the correlations, the assumptions and the output of the multiple regression analyses will be analyzed. The chapter will conclude by discussing the results of the moderator and mediation analyses.
4.1 Descriptive statistics
In total 242 respondents filled in the questionnaire. However 30 of them stopped the questionnaire after the first questions, so they were excluded from the analysis. Therefore, a total of 212 respondents will be used for the analysis. The demographics of the sample reveal that most of the respondents are between 18 and 35 (57.5%). Besides, 61.8% of the respondents are female and 38.2% of the respondents are male. The majority of the respondents are University students (49.5%) and HBO students (38.2%). Most of the respondents have a smartphone with internet (84.9%), 3.8% of the respondents have a smartphone without internet, 4.7% of the respondents have a normal phone with internet and 6.6% of the respondents have a normal phone without internet. The main purpose of the mobile phone can be seen in figure 6. As can be seen, the main purposes are to call (96.7%) and to text (91.5%).
Figure 6: Main purpose mobile phone 0% 20% 40% 60% 80% 100% 120%
Call Text Play games Shopping mp3 player Navigate Surf on websites Banking Social networks Make pictures
Most of the respondents do not have a location-based service application on their mobile phone (56.1%). The reason that they gave for this is that they are not familiar with the app, 25 respondents are not interested, 18 respondents indicated privacy concerns and 4.2% of the respondents do not have a smartphone. If they do have a LBS application, they spend less than 5 minutes on it (23.1%) or 5 minutes (13.7%) or they spend 10 minutes on the LBS application (13.7%). Figure 7: Features
Finally, the respondents indicated which features they (expect) to find most interesting. The majority of the respondents indicated downloading maps (71.7%), route guidance (86.8%), to find nearby restaurants (77.8%) as interesting features (see figure 7).
4.2 Factor analysis
When conducting a factor analysis with Principal Component Analysis (PCA) as selected factor model it appears that the Kaiser-Meyer-Olkin Measure of Sampling Adequacy is 0.857 and the Bartlett’s test of Sphericity is significant (.000), which means that the sample is adequate for a factor analysis. Moreover, the following criteria are used to determine the number of factors: the eigenvalue of the factors should be greater than 1.0, the percentage of variance explained per factor should be greater than 5.0% and the cumulative variance explained by all the factors should be greater 60% (Malhotra and Birks 2007). When looking at the factor criteria it shows that the Eigenvalue and the cumulative percentage indicates a six factor solution. However, when analyzing the percentage of variance it indicates a four factor
Emergency services Friend finder Dating partners Loyalty program Real-time mobile sampling Downloading maps Route guidance
Find nearby restaurants Read reviews of
solution. It is better to have more factors than too few and the rotated matrix also indicate six factors so it can be concluded based on the criteria of the Eigenvalue (> 1) and the cumulative percentage (> 60%) that a six factor solution is best (see table 2).
Table 2: Determining the number of factors
Component Eigenvalue % of Variance Cumulative %
1 7.678 29.531 29.531 2 3.448 13.260 42.791 3 1.968 7.568 50.359 4 1.890 7.270 57.629 5 1.347 5.179 62.808 6 1.059 4.074 66.882 7 .923 3.549 70.431 8 .858 3.302 73.733
The dependent variables and the construct behavioral attitude towards LBS are excluded from the factor analysis because it is a consequence of the other constructs so it is not included in the analysis. The purpose of conducting the factor analysis is to obtain a structure among the constructs. However, when analyzing the rotated component matrix it turns out that in the first component the following constructs are grouped together into one factor: perceived service compatibility, perceived usefulness and perceived enjoyment (see table 3). The reason for this is that the questions are related to each other (e.g. if it is useful, easy, interesting). Yet, the original constructs will be used, so the items will be separated in different components as in the original scales in the methodology chapter. In this case it is possible to analyze all the three constructs in the next analyses and to decide whether the hypotheses are consistent with the output of the analyses.
related and the item that belongs to perceived risk sounds like one item of privacy concerns. Although, the items will be separated as in the original scales in order to analyse them in the next analyses.
All items have a factor loading which are higher than 0.5 (see Appendix 3 and table 3). The Cronbach’s alpha for each of the constructs exceeds the lower limit of acceptability of 0.6 (Malhotra and Birks 2007). Only perceived risk will be excluded because it shows a low Cronbach’s alpha (0.210), which seems reasonable because the two items in this construct are very different from each other. So, one item of perceived risk which loaded in the sixth component and one item which loaded in the fifth component are excluded. The other constructs indicate a high reliability, which confirms the appropriateness of the items and further analyses. The constructs perceived value and behavioral attitude are used to indicate the reliability. Table 3 shows the Varimax rotated component loadings and Cronbach’s alpha of these items. Furthermore, the mean of the independent variables are computed for further analyses.
Table 3: Factor analysis
Privacy concerns PC 1 .769 .769 PC 2 .572 PC 3 .759 PC 4 .904 PC 5 .539 PC 6 .859 PC 7 .910
Behavioral attitude towards LBS .911 Perceived value .882
4.3 Correlations
The Pearson correlation coefficients indicate that most of the correlations are significant (p* < .05; p** < .01). The dependent variable is adoption intention and the independent variables are: perceived service compability, behavioral attitude, perceived ease of use, subjective norms, perceived risk, privacy concerns and perceived value. There is also a bivariate analysis conducted to see the correlations between perceived value, perceived usefulness and perceived enjoyment, they were all significant at the 0.01 level (see table 5) The complete outputs of SPSS can be found in Appendix 4. When analyzing table 4 it can be concluded that all the variables have a high correlation with adoption intention. Perceived service compability, behavioural attitude towards LBS, subjective norms and perceived value are not correlated with privacy concerns. In addition, subjective norms and perceived risk and privacy concerns and perceived value are also not correlated. The rest of the correlations are significant (see table 4).
Table 4: Pearson correlations with dependent variable adoption intention
Variable M SD 1 2 3 4 5 6 7 8 1 Adoption intention 61.56 25.07 1 2 Perceived service compability 4.81 1.457 .718 ** 1 3 Behavioral attitude towards LBS 7.12 1.637 .774 ** .778** 1 4 Perceived ease of use
Table 5: Pearson correlations with dependent variable perceived value Variable M SD 1 2 3 1 Perceived value 5.09 1.329 1 2 Perceived usefulness 4.93 1.359 .760** 1 4 Perceived enjoyment 7.10 1.808 .709 ** ,679 ** 1 Note: p* < .05; p** < .01 4.4 Regression assumptions
Ordinary Least Squares (OLS) regression estimation assumes specific behaviour of the disturbance terms of a model (Hair, 2010). For that reason, it is tested whether the underlying assumptions of the disturbance terms are satisfied. There are two models which are tested for the assumptions, one model with as dependent variable adoption intention and with the independent variables: perceived service compability, behavioral attitude towards LBS, perceived ease of use, subjective norms, perceived risk, privacy concerns and perceived value. In the other model the dependent variable is perceived value and the independent variables are perceived usefulness and perceived enjoyment. In this way it is consistent with the regression analyses in paragraph 4.5 and with the literature review. The three assumptions that will be tested in this paragraph are linearity of the measured phenomenon, heteroscedasticity and normality of the error term distribution. The assumption autocorrelation will not be tested because that is only a potential problem for time-series models (Leeflang et al., 2000). In the next paragraph the assumption multicollinearity will be tested. Testing these assumptions and addressing the violations in the best way possible, will give a better representation of the model (Hair, 2010).
4.4.1 Linearity and heteroscedasticity
To test for heteroscedasticity the same plots can be analyzed. It can be concluded that the residuals are randomly scattered around zero (the horizontal line), although there is a slight form of inconsistency in the variance, the residuals are not spreading out to one side and follow a relatively symmetrical distribution in both plots. So, the homoscedasticity assumption is not violated. The residual plot is conducted with the standardized residuals and standardized predicted values (see figure 8 and 9).
Figure 8: Residual plot model 1 Figure 9: Residual plot model 2
4.4.2 Non-normality
To test the normality of the disturbance terms both Kolomogorov-Smirnov and Shapiro-Wilk tests of normality are performed on the unstandardized residuals. Since the sample size is greater than 50 (n=212 < 50), the result of the Kolomogorov-Smirnov test will be used. The null hypothesis states that the sample follows a normal distribution. The test show in both models a significant value (.001 and .000) at a 95% interval, which means that the unstandardized residuals appear to be non-normally distributed (see table 6), so the hypothesis can be rejected. To remedy the non-normal distribution, transformations of the variables should be done. However, when using the log of the variables, the Kolomogorov-Smirnov test is still significant. Besides, it also possible to remedy it when deleting outliers only that is already done before doing the analyses and it does not help to remedy the non-normality. Therefore, it is important to be careful when interpreting the significance levels of the parameter estimates.
Table 6: Kolomogorov-Smirnov test Statistic Df Sig. Unstandardized residuals model 1 .085 212 .001 Unstandardized residuals model 2 .091 212 .000
To see if there is a relation between the residuals and the predictors, several plots are generated. In figure 10, 11, 12 and 13 the histograms and normal probability plots of the residuals of the two models can be seen. It can be concluded that the histograms of the residuals appears to have a bell shape and the residuals in the normal probability plots follow the diagonal of the normal distribution. However, in both figures a few small deviations from the normal distribution are visible.
Figure 10: Histogram model 1 Figure 11: Normal probability plot model 1
Figure 12: Histogram model 2 Figure 13: Normal probability plot model 2
4.5 Multiple regression analysis
variables (e.g. age, gender and education) are not significant so they are excluded from the analysis. Besides, the results of the regression analysis show that 67 percent of the variance of the probability of adoption intention and 64 percent of the variance of the probability of perceived value is determined by the independent variables, which indicates that the explanatory power of the model is strong. However, the results also show multicollinearity, which will be described later.
To test the hypotheses the results of table 7 and 8 will be analyzed because table 9 is only displayed to see what the effect is on the independent variables with a low VIF value when inserting the three variables with a high VIF value one by one. The first hypotheses states that usefulness and enjoyment has a positive effect on perceived value of location-based services. The results of the regression analysis of the two constructs reveal almost the same results (see table 8). The construct usefulness has a significant effect on the perceived value (p = .000). In addition, the construct enjoyment also have a significant effect on perceived value (p = .000). This indicates that hypothesis 1a + b can be accepted.
The second hypothesis which states that subjective norms will have an moderating effect on adoption intention can be rejected because it has no significant effect on the dependent variable, adoption intention (p = .514). Moreover, the third hypothesis suggests that perceived ease of use is positively related to adoption intention of location-based services. When analyzing table 6 it can be concluded that perceived ease of use has no influence on the adoption intention (B = .684, p = .419). Hence, hypothesis three is rejected. The fourth hypothesis states that behavioral attitude has a positive influence on the adoption intention of location-based services. When analyzing table 7, it can be concluded that behavioral attitude has a positive influence on adoption intention (B = 5.582). In addition, the hypothesis can be accepted because it is significant (.000).
This hypothesis can be accepted (p = .001). The last hypothesis states that greater perceived service compatibility is associated with a higher adoption intention of location-based services. When looking at the results of the regression analysis in table 7 it has a significant effect on the dependent variable, adoption intention (p = .035). So, the hypothesis can be accepted. Furthermore, the construct perceived value has a strong relationship with the dependent variable, adoption intention (B =5.595, p = .000).
Table 7: Multiple regression with adoption intention as dependent variable
Variables Hypothesis B Sig. VIF
(Constant) -9.421 .326
Perceived service compability H7 2.785 .035 3.743
Behavioral attitude towards LBS H4 5.582 .000 3.681
Perceived ease of use H3 .684 .419 1.379
Subjective norms H2 .544 .514 1.463
Perceived risk H5a+b -1.115 .446 1.117
Privacy concerns H6 -1.228 .001 1.118
Perceived value 5.595 .000 4.489
Overall model significance Adj. R^2: .672 F: 62.82 p-value: .000
Table 8: Multiple regression with perceived value as dependent variable
Variables Hypothesis B Sig. VIF
(Constant) .727 .002
Perceived usefulness H1a .506 .000 1.855
Perceived enjoyment H1b .263 .000 1.855
Overall model significance Adj. R^2: .643 F: 191.19 p-value: .000
case three variables indicate a high VIF value, which is high based on a total of 9 variables. Hence, the correlated variables will be included one by one, see table 9. After the correlated variables are one by one included in a regression analysis and compared with the uncorrelated items it can be said that the unstandardized regression coefficients and significant values change (see table 9). It seems that the construct privacy concerns are not significant anymore. Although, the constructs subjective norms and perceived risk are significant in table 7, which indicate that the relationship becomes stronger between these two variables and the dependent variable, adoption intention, by deleting the other three variables. In addition, the VIF values are decreased for the four variables.
Table 9: Regression analysis to discuss the multicollinearity
Variables B Sig. VIF
Perceived service compability 12.363 .000 1.000
Behavioral attitude towards LBS 11.861 .000 1.000
Perceived ease of use 6.159 .000 1.101
Subjective norms 6.668 .000 1.046
Perceived risk -4.526 .026 1.063
Privacy concerns -.603 .239 1.079
Perceived value 14.532 .000 1.000
4.6 Moderator analysis
To see if subjective norms will have a moderating effect on adoption intention, the variables will be first standardized and computed. The interaction with the moderator (subjective norms) and an independent variable (e.g. perceived ease of use) is computed. Unfortunately, the results of the regression analysis show that the interaction effect is insignificant (p = .285), which is consistent with the outcome of the regression analysis. Hence, subjective norms has no moderating effect on adoption intention.
4.7 Mediation analysis
discussed in Preacher and Hayes (2008) bootstrap confidence intervals are preferred over the normal theory-based Sobel test for inference about indirect effects because of the assumption the Sobel tests makes in regard to the shape of the sampling distribution of the indirect effect. Bootstrapping is performed at a confidence interval of 95%. When analyzing the results of the mediation analysis it can be concluded that mediation is established because the confidence interval does not include 0 and the p-values are all < 0.05, so the indirect effect a * b is significant (Preacher and Hayes 2008). Besides, also path c is significant and a * b * c is positive, so there is also complementary mediation present (see figure 14, 15 and appendix 5).
Figure 14: Mediation analysis model 1
Figure 15: Mediation analysis model 2
5 CONCLUSIONS AND RECOMMENDATIONS
The main objective of this study is to identify the factors that influence the perceived value and consumers’ adoption intention of location-based services. The investigation of these variables is urgent because the adoption rate of the location-based services is still very low and therefore it is interesting to study the reason of these low rates. Several authors researched some constructs of this study. However, not one author researched all constructs together. Hence, the aim of this study is to give a complete overview of the variables which determines the perceived value and adoption intention of LBS. The variables that are used to determine the adoption intention and perceived value of LBS are: perceived service compability, perceived usefulness, perceived enjoyment, behavioral attitude towards LBS, perceived ease of use, subjective norms, perceived risk and privacy concerns. The research question is: Which variables influence the perceived value and adoption intention of location-based services?
However, research indicate that the following variables are not supported: subjective norms, perceived ease of use and perceived risk (see table 10). The reason that can be given that subjective norms are not significant is that the usage of LBS is voluntary. Hartwick and Barki (1994) already researched that the intention of users under voluntary usage will be more based on their own attitudes, which is consistent with the outcome of this research. The social pressure for adopting location-based service is lower, which makes sense. This is different than for example with Facebook, if you do not participate and all your friends do and they are on Facebook all the time, then you miss a lot. This is not the case with location-based service, you will only use it if you are in an unknown city for example and you need some information where restaurants are located.
The results also indicate that perceived ease of use is not significant. This makes also sense because consumers that use a location-based service and know about it are the innovators, they have more interest in innovations than on average (Rogers, 1995). So, they will find it easy to find out how a location-based service works. Tornatzky and Klein (1982) find that when the degree of complexity is low a person is more likely to form a positive attitude about the innovation, which makes sense because the construct behavioral attitude is significant in this research. Besides, older people are said to be more resistant to technological change then younger people, while younger people tend to pursue innovativeness (Rogers, 1995). This is also the case with this research, the age group of the respondents is very young, 57.5% of the respondents are between 18 and 35. So, they grow up with these mobile applications and they will find it easier to use such an application than older people.
The constructs perceived usefulness and enjoyment have a positive effect on perceived value (see table 10). Individuals who experience immediate pleasure or joy from using a technology and perceive any activity involving the technology to be personally enjoyable in its own right aside from the instrumental value of the technology, are more likely to adopt the technology and use it more extensively than others (Davis et al., 1989). This notion is in line with the outcome of the results for hypothesis one.
So, the variables that have an influence on adoption intention of LBS are behavioral attitude towards LBS, privacy concerns and perceived service compability and on perceived value are perceived usefulness and perceived enjoyment. This gives an answer on the research question of this study.
Table 10: Overview of the tested hypotheses
Hypothesis Result
H1a: Perceived usefulness has a positive effect on perceived value of location-based services.
Supported
H1b: Perceived enjoyment has a positive effect on perceived value of location-based services.
Supported
H2: Subjective norms will have an moderating effect on adoption intention. Not supported H3: Perceived ease of use is positively related to adoption intention of location-based
services.
Not supported
H4: Behavioral attitude has a positive influence on the adoption intention of location-based services.
Supported
H5a: Greater perceived product performance risk is associated with lower adoption intention of location-based services.
Not supported
H5b: Greater psychological risk is associated with lower adoption intention of location-based services.
Not supported
H6: Privacy concerns are negatively related to adoption intention of location-based services.
Supported
H7: Greater perceived service compatibility is associated with higher adoption intention of location-based services.
Supported
are les popular are: 30.2% emergency services, 27.8% friend finder, 4.7% locating potential dating partners, 17.9% loyalty program, 22.2% real-time mobile sampling and 36.8% read reviews of restaurants.
5.1 Recommendations
At first, it is important that a location-based application stays useful and enjoyable, since these two constructs are significant. Next to this, most of the experienced respondents indicated that they spend less than 10 minutes on the application. It should be valuable if they can have some more benefits if they spend more time on the application to discover more features. Besides, the majority of the respondents indicated downloading maps (71.7%), route guidance (86.8%), to find nearby restaurants (77.8%) as interesting features of a location-based service. The advice to the developers of the services is to make the other features more interesting or to develop some new features which will attract consumers to use a LBS.
5.2 Limitations and suggestions for future research
Some limitations can be identified for this research, which in turn could result in opportunities for future research. The sample size of 212 is sufficient, but it would be recommendable to repeat the study with a larger sample size to make it more reliable and representative. Additionally, although a large age group was invited to fill in the questionnaire, most of the respondents were between 18 and 35 years old. So, a suggestion for future research is that it is important to get a more broad sample to analyze the differences between the age of the respondents with regard to the adoption intention. Besides, this study is only based on a Dutch sample, therefore future research should attempt to replicate this study in other countries to further validate the research model.
Furthermore, the output of the factor analysis showed that in the first component the following constructs are grouped together into one factor: perceived service compatibility, perceived usefulness and perceived enjoyment. The reason for this is that the questions are related to each other (e.g. if it is useful, easy, interesting). Besides, some items of the construct privacy concerns and perceived risk were separated and combined. However, the original constructs were used, so it was possible to analyze all the three constructs in the other analyses and to decide whether the hypotheses are consistent with the output of the analyses. A suggestion for future research is to use constructs with different questions who are not related to each other. The non-normality test showed in both models a significant value (.001 and .000) at a 95% interval, which means that the unstandardized residuals appeared to be non-normally distributed. However, when using the log of the variables, the Kolomogorov-Smirnov test were still significant. Therefore, it was important to be careful when interpreting the significance levels of the parameter estimates. In addition, the results also showed multicollinearity. To detect the presence of multicollinearity the complete model can be estimated by eliminating one of the predictors and re-estimate the model. However, in this case three variables indicate a high VIF value, which is high based on a total of 9 variables. Hence, the correlated variables were included one by one to analyze the effect on the variables with a low VIF value. A suggestion for future research is to solve the non-normality and the multicollinearity.
