MSc Marketing Management
Master Thesis
The Impact of Internet Privacy Concerns on Purchase
Intention and the Moderating Role of Prior
Experience and Trust
Florine van den Bent S 2048825 E: f.d.a.van.den.bent@student.rug.nl T: 06-‐45257951 Date: January 16th 2017
Supervisor: Prof. dr. P.C. Verhoef Second supervisor: Mr. Moeini Jazani
Faculty of Economics and Business University of Groningen
Abstract
Internet privacy is an important topic of concerns among consumers, in today’s world where data is collected on a daily basis and outside our awareness. Even though extensive research has been devoted to this issue, the rapidly developing technology asks for continuous updates in this field. This paper proposes four dimensions of
Internet Privacy Concerns (Collection, Secondary use, Control and Errors) and examines their direct influence on purchase intention. Moreover, the moderating effects of prior experience and trust were assessed in this relationship. By collecting surveys among 197 participants, it was found that the dimensions Collection and Errors had the
expected negative effects on purchase intention. Secondary Use and Control showed not to be of significant influence towards purchase intention. Prior experience and trust showed to moderate the negative effect for some of the dimensions and a main effect of trust on purchase intention was found. The managerial implications of these findings are discussed, as well as the limitations and directions for further research.
Keywords: internet privacy concerns, collection, secondary use, control, errors, purchase intention, prior experience, trust
Preface
Writing this Master Thesis is the final page to the chapter of my time as a student. I feel like I have come a long way and I developed myself in more ways than one. Without the help of others I could not have made it to this point, including my supervisor Peter Verhoef, all my professors and fellow students. The Rijksuniversiteit Groningen has given me all the support I needed and has provided plenty of opportunities to grow and make me feel prepared for the next chapter of my life. Lastly, I am very thankful that my parents supported me, whichever way I chose and always have so much believe in me.
Table of Content
1. Introduction ... 5
2. Theoretical framework ... 8
1.1 Internet privacy concerns ... 8
1.1.1 Collection ... 10 1.1.2 Secondary use ... 11 1.1.3 Control ... 12 1.1.4 Errors ... 13 1.2 Moderators ... 15 1.2.1 Prior Experience ... 15 1.2.2 Trust ... 15 1.3 Control variables ... 16 1.2.1 Age ... 16 1.2.2 Gender ... 17 1.3.3 Income ... 17 3. Methodology ... 18 3.1 Method ... 18 3.2 Data collection ... 18 Survey design ... 18 Moderator effect ... 19 3.3 Procedure ... 19 4. Results ... 21 4.1 Descriptive statistics ... 21
4.2 Internet Privacy Concerns ... 21
4.3 Regression Analysis ... 23
4.3.1 Control variables ... 23
4.3.2 Main effects ... 23
4.3.3 Moderators ... 25
4.3.4 Full model ... 25
5. Conclusions and Discussion ... 27
5.1 Conclusion ... 27
5.2 Managerial implications ... 28
5.3 Limitations and directions for further research ... 29
References ... 31
Appendix ... 37
7.1 Survey design ... 37
7.2 Factor Analysis ... 40
7.3 Regression Assumptions ... 41
7.4 Moderating effects of Prior Experience ... 43
7.5 Moderating effects of Trust ... 44
1. Introduction
Privacy has become a primary topic of concern among consumers, in a world that increasingly depends on the internet. Information privacy can be defined as ‘the extent to which an individual feels in control over when, how and by whom their personal information is communicated to others’ (Westin, 1967). Not only consumers are concerned about their privacy, other stakeholders such as scholars, governments and business leaders are subject to these same concerns (Smith et al., 2011).
Recently, these concerns for privacy have been accelerated even further, now that personal information is not just provided intentionally, but also unintentionally and outside our knowledge (Mai, 2016; Angst & Agarwal, 2009; Steijn & Vedder, 2015). The collection of our information happens during everyday routines like sending e-‐mails, doing grocery shopping, interacting with friends and family or listening to music (Mai, 2016). This personal information is increasingly used for data mining; the collection and analyses of personal data used for marketing purposes (Clemons & Wilson, 2015; Farid et al., 2016; Tsai & Huang, 2015). As a recent development, marketers have also used this collection of data to personalize the services offered to their customers (Song, 2016). These tailored services might improve the relationship between a firm and their customers and therefore increase trust and loyalty (Howard & Kerin, 2004; Alexander, 2015). However, this personalization of services and advertising might also lead to heightened internet privacy concerns by consumers (Shen and Ball 2009; Lee et al. 2011; Sutanto et al. 2013).
Next to that, social media is starting to play a bigger role in the sharing of
information by consumers (Steijn & Vedder, 2015; Hong & Tong, 2013). Social media is used for social goals, to keep up with current trends and to gather information (Quinn, 2016). These social media platforms need user-‐generated content and the way such information is used is not always clear (Quinn, 2016). Another possible threat of privacy through social media mentioned by Such & Criado (2016), is that items shared on social media by one person, may affect more that just that one person’s privacy.
The rise, existence and consequences of these internet privacy concerns have been extensively researched in the last couple of decades. A large amount of research connected the privacy concerns of consumers to the concept of trust and risks. Forsythe & Shi (2003) studied the effect of risk perceptions on internet shopping and to what extent these risk perceptions are affected by privacy concerns. According to Luo (2002), increasing the trust of the consumer is a viable solution to internet privacy concerns and can be used as an important tool to boost e-‐commerce.
dimensions, including collection, unauthorized secondary use, improper access and errors. Malhotra et al. (2004) focused on three main factors, namely collection, control and awareness. Most recently, Hong & Tong (2013) found six dimensions, which are collection, secondary usage, errors, control, improper access and awareness. After reviewing relevant and recent literature, my conceptualization of internet privacy concerns consists of four key dimensions, which are control, collection, secondary use and errors.
For marketers in this age it is of interest to see how these different elements of internet privacy concerns might influence the extent to which a consumer would have an intention to purchase with a particular online vendor. Kim, Ferrin and Rao (2008) state that trust towards a website is formed of a consumer's propensity to privacy and security concerns, and that this trust is a good indicator of purchase intention.
Moreover, a direct effect on purchase intention was found by Eastlick et al. (2006), who studied online business-‐to-‐consumer relationships and established that there is a strong negative effect of privacy concerns on purchase intent toward a services retailer. This same line of reasoning was found by Hong & Tong (2013), who confirmed that security and privacy risks are the main factors that negatively influence purchase intention. However, the effects of internet privacy concerns on purchase intention have not been studied extensively and not with the particular combination of dimensions that I intent to include in this paper. My main focus of this research is therefore to explore the effects of different dimensions of privacy concerns on purchase intention.
Two moderators are added to the model, one of which is prior experience with a website. A wide array of research confirms that previous positive or negative
experiences have an effect on trust, purchase intentions or risk perceptions (Elangovan et al. 2007; Goles et al. 2009; Pavlou & Gefen, 2005). Specifically, I would like to define prior experience with a website as the amount of purchases that have taken place with a specific online retailer in the past. I expect different levels of prior purchases to
moderate the relationship between the four dimensions of internet privacy concerns and purchase intentions. The second moderator of the conceptual model is trust. I expect that the negative relationship between internet privacy concerns and purchase intention is moderated by the extent to which a consumer trusts a particular online retailer.
The contribution of this paper is fourfold. Firstly, as Smith mentioned in the discussion part of their paper, “the dimensionality of internet privacy concerns is neither absolute nor static, since perceptions of advocates, consumers and scholars could shift over time”. Therefore it is of importance to keep these dimensions up to date, especially since the information technology changes at a rapid pace and the exposure of personal information is happening outside of consumer’s control.
Lastly, the moderating effects of prior experience and trust with an online retailer on the relationship between internet privacy concerns purchase intention will add additional insights. For instance, if a clear moderation is found, this shows the importance of creating loyal customers in reducing the negative effect of internet privacy concerns on purchase intention. The research questions I would like to answer in this research are formulated as follows:
RQ 1: To what extent do different dimensions of Internet Privacy Concerns; 1)
Collection, 2) Secondary Use, 3) Control and 4) Errors affect consumers’ Purchase Intention?
RQ 2: What are the most important dimensions of Internet Privacy Concerns in
relation to Purchase Intention?
RQ 3: To what extent do different levels of Prior Experience with a website
moderate the negative effect between different dimensions of Internet Privacy Concerns and Purchase Intention?
RQ 4: To what extent do different levels of Trust with an online retailer moderate
the negative effect between different dimensions of Internet Privacy Concerns and Purchase Intention?
For marketers, having a current overview of the most important factors driving privacy concerns could be of great value. Possessing this knowledge will make it easier to build in mechanisms or to give out information that reduces these concerns. Research showed that consumers are more likely to buy from online retailers that are more protective of their personal information (Tsai et al. 2011). Furthermore, these authors also showed that consumers are even willing to pay a price premium for products or services that are sold at privacy protective websites.
The rest of the paper is outlined as follows. The second chapter discusses
previous literature found on the four proposed dimensions of Internet Privacy Concerns and their relationship with the concept of Purchase Intention. Also, literature on the moderating effects of Prior Experience and Trust with an online retailer is discussed. In this same chapter the hypotheses are proposed and a visual representation of the conceptual model is presented. The subsequent chapter discusses the proposed
methodology that has been used to realize the research. Chapter 4 will cover the results of the conducted research, which is followed by the conclusion and discussion in chapter 5. The latter will include managerial implications and future avenues for research.
2. Theoretical framework
In the next section I will discuss the relevant research that has been conducted in the field of privacy concerns, regarding the proposed dimensions of internet privacy concerns and the variables purchase intention, prior experience with a website, trust and the control variables. Moreover, I will discuss the suggested relationship between these concepts. Following the outcomes of these theories and researches, I will express my expectations by means of stating the hypotheses. First I will shortly highlight the most relevant research that thus far has been done on the emergence and recent developments of internet privacy concerns.
1.1 Internet privacy concerns
Already three decades ago, Mason (1986) stated in his article that the protection of privacy is seen as one of the most serious ethical concerns of the information age.
Definitions of privacy have been conceptualized in many ways and developed over time. Before defining internet privacy concerns, I will first give an overview of the
conceptualizations of the more general concept of information privacy concerns that have been used across a substantial amount of previous literature.
One of the first definitions of information privacy concerns was developed by Westin (1967) who explained this concept as the extent to which a person has control over the way their personal information is acquired and used. This definition has been reinforced and adopted by many other authors (e.g. Stone et al. 1983; Warren and Brandeis 1890). Clarke (2002) explained information privacy as “the interest an individual has in controlling, or at least significantly influencing, the handling of data about themselves.” Even though this definition is relatively simple, much research has been done across different disciplines and different interpretations and ambiguities exist about the exact meaning. This is because it is a complex notion that can be viewed from many perspectives such as marketing, law or economics (Smith et al. 2011). Culnan & Bies (2002) add to this by saying that privacy is not absolute. They argue that the privacy interests of an individual are balanced with the information needs of the society as a whole. Translated to a business environment, there is a balance between the firm’s need to understand their customer needs and the individuals’ privacy needs (Winer, 2001).
Internet privacy concerns have been defined as the extent to which a user on the internet is concerned about the way their personal information is being acquired and used by a website (Malhotra et al 2004; Son & Kim, 2008). This definition describes the perception of consumers about the worries they might have about the way their private information is handled. Hong and Tong (2013) adjusted this definition slightly for their research, and specified internet privacy concerns as a dyadic relationship between a person and a digital entity, ‘which can either be a particular website or a category of websites, such as commercial websites’.
As was described in the introduction, internet privacy concerns have been
described as comprising of multiple dimensions. Smith et al. (1996) were one of the first to put together a set of dimensions that they claimed to be the most important factors of information privacy concerns. They developed a multidimensional scale, called concern for information privacy (CFIP), which included the four dimensions collection,
unauthorized secondary use, improper access and errors. Malhotra et al. (2004) followed up this research, but focused on internet users’ information privacy concerns (IUIPC) and proposed three main factors: collection, control and awareness. Their model was an extension to the online context and therefore complemented the traditional practice-‐oriented approach. Most recent work on this line of research was conducted by Hong & Tong in 2013. They made a conceptualization of internet privacy concerns (IPC) and identified six key dimensions that they found to be most commonly utilized in prior conceptualizations of IPC. These are collection, secondary usage, errors, improper access, and awareness.
As mentioned in the introduction part of this paper, these dimensions are not static and are changing over time. Based on the conceptualization of internet privacy concerns in earlier research, the aim in this paper is to focus on four drivers that are outlined as most prominent and relevant in today’s information age. When reading recent literature about internet developments such as data mining, personalization, identity theft and profiling (Dean et al. 2016; Sutanto et al. 2013; Milne et al. 2004), I categorized four relevant drivers, namely: collection, secondary use, errors and control. Below I will elaborate on why each of these dimensions should be included in my
conceptualization of internet privacy concerns, by discussing previous research that has been conducted on their characteristics and effects. In order to get a visual
1.1.1 Collection
Collection is the first discussed construct of internet privacy concerns, because the mere collection of personal information is a great source of concerns by consumers. (Malhotra et al. 2004). One of the first to mention this concern was when Miller (1982) made the strong statement: “There’s too much damn data collection going on in this society”. Malhotra et al. (2004) describe collection as “the degree to which a person is concerned about the amount of individual-‐specific data possessed by websites, relative to the value of benefits received”. This is being reinforced by Cohen (1987), who mentions that individuals will not be willing to give their personal information if they expect it to lead to negative outcomes.
Internet technologies and big data accelerate the collection of customer data by marketers. This raises concerns, since this data is merged to create comprehensive individual-‐level information and relational databases, leading to increased privacy concerns by consumers (Culnan, 1995). The relationship between these concerns and attitude towards the vendor was assessed by Milne and Boza (1998), who found that consumers have moderate concerns and therefore little of trust in marketing practices such as direct marketing.
When individuals’ personal information is collected by parties on the internet, they face several types of privacy threats. This is not only because the other party now has access to that information, but also because this information can be linked to
website are positively related to the online purchase likelihoods for those customers. This shows that there are concerns about collection of data by consumers, but that statements can reduce this, which increases purchase intention.
Milne & Boza (1999) also confirmed the relationship between collection of privacy
sensitive information and the decreased probability of purchase.
Based on the literature found on this construct, it appears that the collection of personal data is an important element of internet privacy concerns among customers. It is expected that the collection of personal information will lead to lower purchase intentions with an online vendor. Hence:
H1a: Collection is negatively associated with Purchase Intention. 1.1.2 Secondary use
This construct, found as one of the six dimensions of internet privacy concerns by Hong and Tong in their article in 2013, is described as the extent to which users are worried that their personal information is being used by (unauthorised) parties other than the party directly asking for their information (Smith et al. 1996). Secondary use could be referred to as the uncertainty that the provision of personal information will cause risks of exposing this information to unintended practices. (Bart et al., 2005). This potential of misuse of personal information makes customers more concerned about their internet privacy (Libaque-‐Saenz et al. 2016). These concerns are confirmed by a research by Hoffman et al. 1999), who show that 69% of people on the web do not want to share their personal information, because they do not know what will happen with that
information. Consumers are concerned that their personal data will be shared with third parties that use this information for marketing-‐related purposes (Mivazaki and
Fernandez, 2000). On the other hand, they are worried that this information will be used for unwelcome contact (Bart et al., 2005).
Even when personal information given is kept internally within the same online organization, people react negatively when this information is used in an unauthorized manner. (Milberg et al. 2000). Several cases of this type of internal secondary use have been raised in literature. One example is ‘sugging’, which means that the collected personal information is used at a later moment in time for marketing purposes (Cespedes and Smith 1993).
The emergence of the use of secondary data flows from the strategic advantage that companies can gain through information technology. Specifically, this comes from the effective results that can be obtained by using this secondary information (Wishart & Applegate, 1990). Of all existing data sources, customer data such as transactions and social media used are useful for marketers and are becoming the focus of this data trend (Chen, Chiang, and Storey, 2012). A common mistake of these firms is, however, that they do not communicate clearly to their customers what will happen to the information they collect (Schwaig et all. 2006). This leads consumers to be left with a feeling of distrust with these firms, because they worry that firms use their personal information for other secondary purposes without their informed consent (Andrew & Shen, 2000).
In other field such as healthcare, secondary use has also been found as a source of concerns by patients. In general, patients are cooperating to make their personal
information available for secondary use, but at the same time they are worried that this information might be used for different purposes, like for marketing or insurance purposes. (Willisan et al. 2016). A possible reaction to this concern could be to avoid giving personal information, which might also inhibit potential customers to purchase from a particular vendor.
The research discussed above seems to have a clear implication, which is that secondary use has been and still is a big concern for consumers. I therefore expect this factor to significantly reduce purchase intention. This leads me to hypothesize the following:
H1b: Secondary use is negatively associated with Purchase Intention. 1.1.3 Control
Control has been an important element of the privacy issues that have developed since the rise of the information technology (Hsu & Kuo, 2003).
Control as a dimension of internet privacy concerns has been defined in the context of a broader sense of privacy. Personal information privacy is referred to as “the ability of the individual to personally control information about one-‐self (Smith, 1994). Pollach (2005) defines the control of information as “the claim of individuals, groups, or institutions to determine for themselves when, how, and to what extent information about them is communicated to others”.
When this is definition is translated to the specific context of internet privacy concerns, Malhotra et al. (2004) and later Hong and Tong (2013) describe this dimension of IPC as the extent to which people are anxious that their personal information held by websites is out of their own control. According to the authors, control is one of the most important components in the internet privacy context,
shopping (Hoffman et al., 1999). When this perceived security is low, consumers are less comfortable moving to the next stage of the actual purchase of a product or service.
Dommeyer and Gross (2003) also mention control as a privacy-‐related construct and find that knowledge is an important factor in reducing the privacy-‐related anxiety. When consumers are knowledgeable about the privacy policies and ways to safeguard their personal information, they experience more control, which in turn less privacy concerns. This is in line with the principle of procedural justice, which in this context explains that a person will experience a procedure as fair when they feel they are in sufficient control of this procedure (Tyler, 1994). A logical implication of this is when a consumer does not experience a procedure as fair, this decreases purchase intention. In a research by Pollach (2007), she found that a factor determining user trust is the level of control users have over their personal information. The level of trust will influence on purchase behaviour and repeat visits. In this research a direct effect of perceived control on purchase intention is expected and tested.
A more recent development in the context of internet is personalization, which makes it possible to tailor online services to the specific characteristics, wants and needs of a customer (Vesanen, 2007). When consumers are exposed to these personalized web pages, they might feel like their personal information has been leaked. This loss of
control will lead them to believe that their privacy is at risk (Song et al. 2014). According to these authors, giving consumers a sense of control will reduce these feelings of risk. This finding was confirmed by Tucker (2014), who tested whether consumers with a greater sense of control over their personal information were more likely to click on personalized ads. Giving participants more perceived control over their privacy, actually significantly increased the likelihood of clicking on these ads.
As a consequence, people are more likely to get in touch with an online vendor, which could eventually lead to higher purchase intentions. Moreover, Nowak and Phelps (1995) showed that a person is less concerned about giving personal information, when they specifically give permission to entities on the web or when they are given the choice to opt-‐out. Offering consumers this kind of control over their disclosure of
personal information is widely found to influence trust in a website (Phelps et al., 2000; Eastlick et al., 2006). As mentioned earlier, I expect higher levels of control to also have a direct on purchase intentions.
Thus, based on the research above, it appears that control has been and still is an important part of internet privacy concerns. A loss of control leads to greater concerns and I expect that lower levels of control will decrease purchase intention. I will test the following hypothesis regarding control:
H1c: Loss of control is negatively associated with Purchase Intention. 1.1.4 Errors
worried that either accidentally or deliberately errors occur with their personal data (Smith et al 1996.). There are many individuals who are concerned that online parties are not making sufficient effort to reduce the problems associated with the potential errors that could occur when handling personal data. Bansal et al. (2016) show that a previous online privacy invasion has a positive effect on internet privacy concerns. Over the last decade, the intensity of consumer data exposure has increased the
frequency of privacy violation in cyberspace (Bansal & Gefen, 2015). As an example, one of the main errors that could occur is hacking. Numbers on hacking are quite large, for example in a research in 2011, 90% of people reported to have be hacked at least once and 50% of them said to have little confidence in preventing hacking from happening
again.
By now strong laws for privacy protection exist, but collecting and transferring customer’s personal data is getting easier with the growing amount of channels and devices (McFarland, 2012). Additionally, it has been increasingly argued that personal data can be anonymized and aggregated to still show trends. However, it is impossible to de-‐identify the multiple flows of data and the ability to re-‐identify individuals still makes that it will still violate the privacy laws (Britton, 2016). This shows that even though many actions and laws are being introduced to prevent hacking from happening, there are trends and forces that make this hard to achieve. Therefore, people still have these fears.
Liao et al. (2009) argue that the consequences of this fear of violation are that people are losing trust in firms, which reduces the extent of the customer cooperating with the firms. In order to purchase online, it is necessary to cooperate with the firm in providing personal information. This same line of reasoning was found by Pavlou & Gefen (2005), who discuss the buyer-‐seller relationships in online marketplaces. They distinguish several sources of Psychological Contract Violations with an individual seller. If the buyer feels like he or she has been treated wrongly, this will have a negative effect on trust and I expect this to hold as well for purchase intention. Further, Chen & Zahedi (2016) speak of errors in terms of a perceived threat, and found that an increase in perceived threat leads to an increase in avoidance. This avoidance behavior was also found as a reaction to privacy concerns by Sheehan & Hoy (1999), which implies that the perceived threat to errors can actually lead to avoidance behaviors. Avoiding an online retailer means that there is a decreased purchase intention.
Taken together, consumers show to be concerned about potential errors and therefore errors are included as relevant dimension of internet privacy concerns. If the concerns of this dimension are high, it is likely that consumers are less willing to disclose personal information and have a lower purchase intention. Therefore the hypothesis is formulated as follows:
1.2 Moderators
As mentioned briefly in the introduction of the paper, two moderators are included in the conceptual model. Both are expected to have a moderating effect on the relationship between the four dimensions of internet privacy concerns and purchase intention. The first moderator is prior experience, which reflects an experience component. The second moderator is trust and can be seen as an attitude component.
1.2.1 Prior Experience
The first moderator included is prior experience. In this context, I refer to prior
experience as the amount of purchases that have been done with a particular retailer in the past. In this sense, the different levels of purchase intention are defined by the number of purchases. Having purchased something from a retailer before implies that the barrier of disclosing personal information has already been overcome and that there were reasons to feel less concerned about privacy related matters. Feeling less
concerned in turn should increase purchase intention.
There has been a collection of research that found that prior experience with a website has a positive influence on consumers’ willingness to disclose information. For example, Bansal (2008) found that prior positive experience with a website increases consumers’ willingness to disclose personal information. The author explains this by means of prospect theory, which in this context means that the disutility related to giving personal information is decreased based on positive experience.
A study by Culnan and Bies (2003) showed that 71% of their respondents was willing to disclose personal information within an established relationship. I expect this same effect to apply to the relationship between internet privacy concerns and purchase intention.
Specifically, I expect different levels of prior experience influence the relationship between all four dimensions of internet privacy concerns and purchase intention. The four accompanying hypotheses are stated as follows:
H2a: Higher levels of Prior Experience with an online retailer will decrease the
negative relationship between Collection and Purchase Intention.
H2b: Higher levels of Prior Experience with an online retailer will decrease the
negative relationship between Secondary Use and Purchase Intention.
H2c: Higher levels of Prior Experience with an online retailer will decrease the
positive relationship between Control and Purchase intention
H2d: Higher levels of Prior Experience with an online retailer will decrease the
negative relationship between Errors and Purchase Intention.
1.2.2 Trust
an online retailer. Trust has been defined in many different ways and across many disciplines. Most widely used is the definition by Mayer et al. (1995) who describe trust as ‘the willingness to be vulnerable.’ Kumar et al. (1995) define trust as the expectation that other parties whom individuals decide to trust, do not take advantage of the
situation or behave in opportunistic ways. In a buyer-‐seller relationship, trust is of crucial importance. This is even more so when there is an element of risk involved, which is the case when this transaction is taking place online (Gefen et al. 2003). There is little guarantee that the online retailer will not take advantage of the situation by using personal information for unintended purposes. Luo (2002) found that most people hesitated to purchase online or even left the electronic market because of a lack of trust. As Bansal et al. (2016) stated, trust determines the extent to which an individual is willing to reveal private information to an online entity. Little is known about the physical location of the online vendor and websites have an impersonal nature, which may make customers vigilant of providing personal information online. As found by Moorman et al. (1992), trust would reduce “the perceived uncertainty and hence the perceived vulnerability”. This suggests that increasing levels of trust will decrease the negative effect of the four dimensions internet privacy concerns on purchase intention, and I expect this for all four dimensions. The hypotheses are therefore formulated as follows:
H3a: Higher levels of Trust with an online retailer will decrease the negative
relationship between Collection and Purchase Intention.
H3b: Higher levels of Trust with an online retailer will decrease the negative
relationship between Secondary Use and Purchase Intention.
H3c: Higher levels of Trust with an online retailer will increase the positive
relationship between Control and Purchase Intention
H3d: Higher levels of Trust with an online retailer will decrease the negative
relationship between Errors and Purchase Intention.
1.2 Control variables
To complete the model there are three control variables I add, to identify potential differences in the importance of the different dimensions of internet privacy concerns on purchase intention.
1.2.1 Age
According to Montgomery & Pasnik (1996) teenagers are more susceptible and do not know how to protect themselves when they engage in online activities.
After conducting their research, Clemons & Wilson (2015) found that teenagers might not be aware of the risks of online activity and do not take sufficient precautions to protect themselves from it.
These findings suggest that since the younger generation is more used to the information technology, they are less concerned about their privacy and will show less strong effects during the experiment.
1.2.2 Gender
Previous research on differences in perceived privacy between men and women shows that there is sufficient indication that this difference might indeed exist.
Men tend to provide personal information such as their address or telephone number on social media websites more often than women do (Tufekci, 2008).
Hoy & Milne (2010) conducted research on gender differences in terms of Facebook usage. They found that women were significantly more concerned about their private information posted on Facebook than men are.
These findings suggest that after conducting my research, I will find that the effects of the dimensions of internet privacy concerns will be of greater power on purchase intention for women than for men.
1.2.3 Income
Lastly, the control variable income is included. I expect that people with higher incomes will have more internet privacy concerns and therefore less purchase intention. Higher incomes suggest high involvement with their jobs, and being educated about internet usage is often a necessity. This education makes people more aware of the possible negative consequences and are therefore more concerned (Sheehan, 2002).
3. Methodology
3.1 Method
In order to test my research question, a survey was designed and distributed among 213 participants. 14 of the cases were removed, because of multiple missing values or a wrong answer to the control question. One case was removed as an outlier, which had a standardized residual of more than 4 and distorted the results significantly. The final sample therefore consisted of 197 participants. After collecting the surveys, a regression analysis was performed to test the proposed hypotheses regarding the relationships between the different dimensions of Internet Privacy Concerns and Purchase Intention. Regression was assumed to be the appropriate method, since it estimates relationships between a dependent variable (Purchase Intention) and one or more independent variables (dimensions of Internet Privacy Concerns). Specifically, regression analysis measures how the value of the dependent variables changes when one of the
independent variables is varied, given that the other independent variables are held constant.
Next to that, the moderating influence of both Prior Experience and Trust with an online retailer on the relationship between the discussed variables were tested. This was conducted by adding interaction terms between the moderators and the main effects to the regression model. Adding these interaction terms to the model was done for one moderator at the time, to see their individual effect on the relationship between the main effects and Purchase Intention.
Below I will explain the choices made in designing the survey and the collection of data. Further, I describe the procedure that participants followed and the
measurements that were used to analyse the data. 3.2 Data collection
Survey design
Before conducting the full research, a pre-‐test was performed. The survey was tested among 10 participants, to see whether the questions were understood correctly and whether any errors existed. After that, the final questionnaire was adjusted based on given comments and misunderstandings. Next to that, it was tested whether the several questions per dimension of internet privacy concerns actually all test the same thing. If one question fell out of line with the rest, this question was removed.
For Purchase Intention, I considered recent literature in an online context and adopted a mix of items from two researches by Yang et al. (2016) and Sahi et al. (2016). For all items mentioned, seven-‐point Likert scales were implemented with anchors ranging from “strongly disagree” to “strongly agree”. For several variables a reversed question was added, to increase the validity.
Moderator effect
To find out the influence of Prior Experience and Trust with a retailer on the
relationship between the different dimensions of Internet Privacy Concern and Purchase Intention, a couple of items were added to the survey. Prior Experience was measured by first asking the participants to state at which online retailer they most recently did a purchase. The next question was for the participants to state how often they had
purchased from this particular online retailer in the past. Four different levels were established: 1 time, 2-‐3 times, 4-‐5 times or more than 5 times. Since these categories had different widths, dummy variables were created with ‘1 time’ as a reference category. This made it possible to assess how different amounts of purchases might have had an influence on the relationship between the dimensions of IPC and Purchase Intention. The items for measuring Trust are based on research by Malhotra et al. (2004), who included trusting beliefs as a context-‐specific factor of internet users’ information privacy concerns (IUIPC). The questions were slightly adapted to make them specific to the particular online retailer, as mentioned by the participant. 5 items were adopted for to measure Trust, with seven-‐point Likert scales with anchors ranging from “strongly disagree” to “strongly agree”.
My control variables Gender, Age and Income were additional questions in my survey, in order to measure any differences in effect. Income consisted of four
categories: Below average income, average income, twice the average income, more than twice the average income. For this variable 3 dummy’s were created, with ‘average income’ as a reference category.
3.3 Procedure
As mentioned before, the survey was completed by 197 respondents. There was no specific education level required, the minimum age was 18 and both men and women were included. However, a diverse group of respondents in terms of gender, age and income was necessary for me to draw any possible conclusions about differences between these demographics.
I used Qualtrics as an online survey tool to collect my data. Participants could enter the questionnaire by clicking on a provided link. I distributed this link through channels such as Facebook, LinkedIn and e-‐mail networks. Since I sent out my survey mainly to my own known networks, there is a large chance of convenience sampling. That means that the participants were chosen because they were easy to access for me and therefore are less representative of the entire population.
gain insights on online purchasing behaviour. Lastly, the amount of time the survey would take was included. When the participants started the survey, they were first asked to fill in most recent online retailer they had purchased from. They were asked to keep that specific online retailer in mind for the remainder of the survey.
Next, there were asked to answer a few questions about this retailer, including the amount of times purchased from them before.
In the next section the participants were asked to continue to the next page, where they started filling out the questionnaire. The questionnaire started with
4. Results
4.1 Descriptive statistics
As displayed in table 1, the sample included slightly more women than men. A possible explanation is that I distributed the survey among my network of family and friends, which consists of more women than men. The respondent’s age ranged between 19 and 74 years old, with an average of 39,38 years old. This has a similar explanation as mentioned before; my network is mostly under the age of 40. In terms of income, the gross of the participants showed to be in the ‘less than the average income’ group or in the ‘average income’. This survey was partly filled out by respondents who got a monetary compensation for completing the survey, which suggests that these people could use the money well and are more likely to be in a lower income group. Almost all respondents had a Dutch nationality, except for 10 cases of the total sample. An
overview of the sample characteristics is summarized in table 1.
Table 1
Sample characteristics
Percentage Number of respondents
Gender Male Female 41.2% 58.3% 82 116 Age 19-‐25 26-‐40 41-‐60 >60 28.1% 30.7% 26.6% 14.6% 56 61 53 29 Income
Less than average Average
Twice the average More than twice the average 41.2% 34.7% 17.1% 7.0% 82 69 34 14
4.2 Internet Privacy Concerns
In order to see whether the four dimensions of Internet Privacy Concerns are actually different from each other and measure diverse elements of the construct, a factor analysis was performed. Applying factor analysis was found to be appropriate, with a KMO statistic of 0,890 and a significant outcome of the Barttlet’s Test of Sphericity (p=0,000). All communalities were above 0.4. Even though the eigenvalues of only the first two factors were above 1, the two following factors did each explain at least 5% (Appendix 7.2), which is why four factors are assumed to continue testing the proposed research question. Table 2 shows the results of the Pattern Matrix.