The interplay of retargeting, privacy risk, consumer and product characteristics and purchase intent

(1)

Amsterdam Business School

Master Business Administration.

Digital Business Track

The interplay of retargeting, privacy risk, consumer and product

characteristics and purchase intent.

MSc Thesis by

Djurre Heemskerk

10796940

Supervisor: Raoul Kübler, PhD.

Amsterdam, 22

th

of June 2018

(2)

1

This document is written by Student Djurre Heemskerk who declares to take full

responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that

no sources other than those mentioned in the text and its references have been used

in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of

completion of the work, not for the contents

(3)

2

Abstract

The aim of this research is to give a better explanation of the role of privacy concern on the relationship between personalization and purchase intent. Personalized ads can be fruitful for both companies and consumers due to being more relevant, however they are also experienced as intrusive and raise levels of privacy concern. Which in turn leads to lower purchase intent. Furthermore, the influence of product and consumer characteristics on the relationship between personalization and privacy concern has been investigated. The hypotheses were tested with a sample of 362 respondents. These respondents came from Amazon MTurk and survey groups. The main results are that there a significant effect of privacy concern on purchase intent, however there is no mediation effect of privacy concern on the relationship between personalization and purchase intent. Furthermore, product characteristics was found to have a negative effect on privacy concern, however both product and consumer characteristics had no moderating effect on the relationship between personalization and privacy concern.

(5)

4

1. Introduction.

The implementation of the new European legislation in May 2018 regarding the protection of personal data indicates that regulation is finally catching up to industrial giants. Who have been leveraging the anonymized data of consumers for several years now (European Commision, 2018). The General Data Protection Regulation (GDPR) made sure that there is one set of data protection rules that hold for every company that conducts business in the EU. Regardless of where they are based (European Commision, 2018). The GDPR had two main goals; giving people more control over their personal data and leveling the playing field for companies, so that there are no differences between companies caused by the country they are based in (European Commision, 2018). Personal data is the foundation where some of the biggest internet companies such as Google, Facebook and Twitter are built on (Esteve, 2017). The IDC (2017) projects that the revenue from big data and business analytics will reach $150 billion in 2017 and will increase to more than $200 billion in 2020. Companies see it as the new big thing and are betting big on it. Some even call personal data the new gold of companies (Esteve, 2017). But why is this personal data so valuable for companies?

For a couple of reasons. One of those reasons is the use of personal data for personalized advertising. Previously advertisers tried to reach a big public with one message, think of a tv commercial. Nowadays big data gives advertisers the ability to make a specific message for everybody. Combining personal data with developments in the information-processing technology enables messages which are now tailored to that one potential customer (Baek & Morimoto, 2012). This leads to bigger returns on marketing investments and allows them to target customers that are most likely to be influenced by the ads (Goldfarb & Tucker, 2011). The targeting of consumers which are most likely to be influenced is extremely valuable for

(6)

5

companies, as it saves them money and increases the chances of targeting consumers which are ready to convert. This leads to higher sales, lower costs and therefore more profit. The 2016 Jivox Benchmark Report analyzed performance data of personalized ads and found that personalized digital ads had a click-through rate that was 230% higher than the Google display ad benchmark, interaction rate was 84% higher than the Google benchmark, and average dwell time was 28% higher (2016 Jivox Benchmark Report). This shows that personalized ads are 3 times more effective than non-personalized ads when it comes to consumers interacting with them. Furthermore, a report from Adobe Digital (2017) shows that more than 33% of all age groups prefer personalization and over 50% of younger age groups prefer personalized ads and a staggering 83% of millennials like the personalization of ads. However, next to the advantages for companies, there are also advantages for consumers. As the ads are based upon personal data, they are likely to be more relevant and interesting to consumers (Srinivasan et al., 2002). This is because the ads represent the products or services which the consumer is looking for or interested in at that time.

Even though there are positive aspects of personalized ads for both consumers and companies, there are also concerns from consumers regarding the use of personal information to create such ads. Consumers are concerned with the level of intrusiveness of the ads and feel that their privacy could be violated. According to Estrada-Jiménez et al. (2017) consumers are seriously concerned with the increasing invasiveness and intrusiveness of digital advertising. Adobe Digital (2017) found that 83% from the millennials like personalization but 55% of them say that they can be improved and should be less creepy/ intrusive. Furthermore, two out of three internet users worry about the fact that their online behavior is monitored extremely closely without their knowledge and consent (Van Doorn & Hoekstra, 2013). This diminishes the value of personalized ads for the consumer, they gain the advantage of more relevant ads, but at the

(7)

6

cost of providing personalized information without their knowledge or consent. Furthermore, Van Doorn & Hoekstra (2013) showed that the higher levels of perceived intrusiveness caused by the personalized ads can have negative consequences for companies as well. They found that for higher levels of perceived intrusiveness and privacy concern purchase intent was negatively affected. So, the higher the levels of privacy concern caused by the ad, the lower purchase intention will be. Which could lead to lost revenue and wasted marketing efforts for companies. Van Doorn and Hoekstra (2013) defined it as a double-edge sword, on the one hand there are the positives of the personalization of ads. However, if consumers experience the ads as intrusive and the levels of privacy concern increase, personalization could lead to negative effects such as a decrease in purchase intent. Marketeers walk a fine line when it comes to working with personalized ads.

There are studies with only single measures such as purchase intentions (Goldfarb & Tucker, 2011a; Van Doorn & Hoekstra, 2013), click-through rate (Ansari & Mela, 2003; Tucker 2014) or online sales (Lambrecht & Tucker, 2013). But there is also a body of research that combined multiple measures such as click-through rate, purchase likelihood and brand

consideration (Urban et al., 2014). What they all had in common was that they found, in some way, a positive relationship between the personalization of ads and the chosen measures such as click-through rate or purchase intent. This shows that the positive relationship between

personalization and purchase intent is not only found in market research (2016 Jivox Benchmark Report; Adobe Digital, 2017), but also in the relevant literature (Lambrecht & Tucker, 2013; Goldfarb & Tucker, 2011a).

When we look at the overall picture, we see three different relationships which are likely to be related. We see that the personalization of ads has a positive effect on purchase intent (Van Doorn & Hoekstra (2013). Secondly, we see that the personalization of ads increases the levels of

(8)

7

privacy concern in some cases due to the ad being experienced as intrusive (Van Doorn & Hoekstra, 2013; Baek & Morimoto, 2012). Finally, we see that privacy concern has a negative effect on purchase intent (Goldfarb & Tucker, 2011a; Lambrecht & Tucker, 2013). This makes it likely that privacy concern has a significant role in this relationship. However, until this date research has only looked at the direct effects, and not at the relationship between all three components. This brings us to our first research question, namely:

What is the role of privacy concern in the relation between personalization and purchase intent?

When looking at factors that may influence the level of perceived privacy risk this paper already mentioned the degree of personalization. However, there are also other factors which are found to have a moderating effect on the relationship between personalization and privacy concern. Existing literature found four factors which decreased the level of privacy concern caused by personalization. Those factors were; fit (Van Doorn & Hoekstra, 2013; Thota & Biswas, 2009), trust (Martin, 2018), perceived risk (Eastin et al., 2016) and relevance (Zhu & Chang, 2016). This shows that it is possible to negate the privacy concern caused by

personalization and that there is room for improvement in the way personalized ads are designed and what factors to look at. To add to the understanding of this topic we introduce two more factors which are likely to influence this relationship.

The first factor which is lacking in existing research is the consumer itself. Consumer characteristics influence all sort of things, such as online game playing (Chen, 2010), social media use (Correa et al., 2010), privacy concerns (Acquisti and Gross, 2005) and a lot more. As different individuals experience different levels of privacy concern, it is also likely that

(9)

8

connect the personalization of ads to more privacy concern than others. This indicates that the characteristics of a consumer are likely to play a part in the relationship between personalization and privacy concern.

The second factor which is lacking in existing research is which product is advertised. Not every product is the same, and every product has different aspects. These differences in product characteristics can cause products to be accepted slowly, or very fast. Lian and Lin (2008) showed that the acceptance of online shopping differs between product types. Some products are more easily bought online than others due to different reasons. Compare a book to a car for example. A car has features which you want to see and test for yourself, as it is likely that a large sum of money is involved. While a book can’t really differ much, and the negative consequences of buying a book which is in a bad condition are smaller due to the price. If we continue this line of thought, we can argue that products could also differ in the speed of their acceptance when it comes to the personalization of ads. And how product characteristics influence privacy concern. As there could be specific characteristics which could make a product differ in how susceptible they are to privacy concern.

The above indicates that both consumer and product characteristics are likely to have an effect on the relationship between personalization and privacy concern in a similar way fit, relevance, perceived risk and trust effect this relationship. However, product and consumer characteristics are not yet researched with respect to the relationship between personalization and privacy concern. Which brings us to our second research question, namely:

How is the relationship between personalization and privacy concern moderated by consumer and product characteristics?

(10)

9

This paper will try to add to the literature on this subject by investigating the influence of privacy concern when personalization, privacy concern and purchase intent are research in the same setting. Previous research has only focused on these relationships on their own, but never in the grand scheme. Furthermore, this paper will add to the existing literature by introducing two factors that are likely to influence the relationship between personalization and privacy concern. Previous research has focused on fit, relevance, perceived risk and trust. However, product and consumer characteristics are new in this debate. Furthermore, they may help explain this relationship or what could be done to decrease the level of privacy concern caused by personalization.

The managerial implications of this paper are that it helps managers with decisions regarding personalized ads. A lot of consumers experience personalized retargeting ads as intrusive and research (Van Doorn & Hoekstra, 2013) shows that this side of personalized

advertisement decreases purchase intent. This paper will try to help managers understand the role of privacy concern in this relationship and adds two factors which could be important tools for managers to decrease this caused level of privacy concern. This helps them build an ad that raises less privacy concern and is therefore likely to be more effective.

To start, a foundation will be made using existing literature to build hypotheses on. This part will also be used to explain the variables and introduce the conceptual framework. In the part that follows, the methodology will be discussed. Subsequently, the results will be presented, followed by a section that discusses these results and draws conclusions from them. The final part of this paper will contain the limitations and suggestions for further research.

(11)

10

2. Literature review.

This section will be divided into four main parts. The first part will give a theoretical background on personalized advertising. This section will be a general introduction into personalized

advertising. The second part will discuss privacy concern and how it’s influenced by

personalization and how it influences purchase intention. The third section will discuss the effect of consumer characteristics. The fourth and final section will discuss the influence of product characteristics and its effect on the relationship between personalization and privacy concern. All these sections will start with a general explanation, followed by an explanation of why they are relevant regarding this research. Figure 1 shows how the concepts are related.

(12)

11 2.1 Personalized advertising

Personalized advertising has been around longer than the internet, however with the help of the internet the possibilities and the extent to which ads can be personalized increased massively (Baek and Morimoto, 2012; Lambrecht and Tucker, 2013). To show how personalized advertising has changed, this research will give a small historic overview. Personalized

advertising started by companies using demographics such as age and gender. From there they started to use more personal information such as buying history and interests. Nowadays

companies are able to track almost everything the consumer sees and interacts with on their site. Companies can now see how long a consumer stays at a page, where their mouse goes, how long their mouse is hovering over a specific product, and much more. This is done by using cookies (Ham, 2016). These cookies get dropped on a consumers’ hard drive and track consumers’ online behavior. This research defines personalized advertising the same as Baek and Morimoto (2012) “as a form of customized promotional messages that are delivered to each individual consumer

through paid media based on personal information (such as consumers’ names, past buying history, demographics, psychographics, locations, and lifestyle interests” (p. 59).

Personalization as we know it now finds its base in big data. Big data is a huge amount of data that is so massive that it is impossible to analyze its content using traditional methods

(Anshari et al., 2018). However, when using new tools this becomes possible, which leads to new sources of value. A single file of a consumer with the last items they bought is not valuable to companies as it doesn’t allow them to make predictions or personalized offers. Because of big data it is now possible to analyze everything which was mentioned above, and this from all the consumers and potential consumers of the company. Which sheds light on new patterns and creates new insight which can be very valuable (Anshari et al., 2018). Due to the aspect of big

(13)

12

data which is called velocity, it is furthermore possible to analyze this data in real time as it comes in. Personalization is made possible by big data as it relies on enormous amounts of information which make consumers anonymous, but still show buying behavior and who to target.

Personalized advertisements have significant cost efficiencies compared to the traditional mass media approach. The cost efficiencies are because personalized advertising allows the marketer to approach consumer who have been identified as viable customers with highly tailored messages (Kim et al., 2001). Furthermore, increasing the personalization of an ad could lead to decreased ad avoidance (Baek & Morimoto, 2012) and is associated with higher satisfaction and loyalty (Ball et al., 2006; Liang et al., 2006). Not only companies profit from personalized advertising, consumers also reap benefits. Consumers are now able to identify relevant ads more quickly and minimize the time they spent searching through alternatives and in turn find what they want quicker (Srinivasan et al., 2002).

However, since the beginning of personalized advertising it has been controversial due to concerns with privacy invasion (Awad & Krishnan, 2006; Xu et al., 2011). A Pew study (Zhu & Chang, 2016) found that many consumers are nervous about the collection of personal data through website, mobile devices and search engines. 68% of the respondents had problems with personalized advertising in general while only 28% felt comfortable with targeted advertising. Ham (2016) researched the imposed trade-off between the benefits for consumers, better and more relevant ads, to the negatives for consumers such as privacy concern. Ham (2016) found that the negatives of perceived privacy risk outweighed the positives of relevant ads. However, this was in contrary with Debatin et al., (2009) who found the opposite. The benefits of the use of Facebook were outweighing the perceived privacy risk. Ham (2016) explained this by addressing the context of the research. He researched perceived privacy in a setting of online

(14)

13

behavioral advertising while Debatin et al., (2009) looked at the use of a social media platform. Furthermore, the extent to which personalized ads result in privacy concern depends on the sensitivity of the information which is used in the ad (Nowak and Phelps, 1992). Therefore, it is expected that ads with a higher degree of personalization, so ads which use more personally sensitive information, cause a higher degree of privacy concern. This is in line with the argument of Bleier and Eisenbeiss (2015) who argued that it is likely that ads with a higher degree of personalization also cause more privacy concern. This results in the first hypothesis:

Hypothesis 1: The higher the level of personalization, the higher the level of privacy concern will be.

2.2 Privacy concerns.

As mentioned before, privacy concern is likely to have an effect on the relationship between the personalization of ads and purchase intent. (Van Doorn & Hoekstra, 2013; Baek & Morimoto, 2012). Privacy and technological advancements have been related since the beginning of the 19th century, sparked by print mass media and photography (Junglas, Johnson & Spitzmüller, 2008). Research has shown that new technologies and how they are used have a major influence on the concern for privacy (Westin, 2003; Armstrong & Ruggles, 2005). So was there in periods with little to no information-technology developments, such as the period after the Second World War, almost no interest in the concern for privacy (Westin, 2003). However, in the period after that, where there were a lot of technological advancements such as database management systems, concern for privacy was suddenly interesting and spoken about. This resulted for U.S. citizens in the Privacy Act of 1974 (Junglas et al., 2008). This series of events is similar to what we see now

(15)

14

with the previously mentioned GDPR in Europe. One of the reasons as to why this regulation was implemented was to protect citizens from companies which are gathering information about consumers without their consent or knowledge.

Privacy concerns affect consumers in the way they behave. When it comes to ads consumers with privacy concerns experience higher levels of ad avoidance (Baek & Morimoto, 2012), experience lower levels of trust (Martin, 2018) and when consumers perceive that their privacy is invaded they are less likely to use a service (Zhu & Chang, 2016). Research on advertising avoidance goes back to before the internet (Abernethy, 1990; Nuttall, 1962). Until Baek & Morimoto (2012), the focus of previous research was to give insights that would help marketeers to decrease advertising avoidance. Baek & Morimoto were the first authors that tried to understand the drivers of advertising avoidance in the context of personalized media.

Personalized media included: unsolicited commercial e-mail, telemarketing, postal direct mail, and text messaging. Baek & Morimoto found support for their hypothesis that perceived privacy concerns will be positively related to personalized advertising skepticism and that perceived privacy concerns are positively related to ad avoidance.

Furthermore, Van Doorn and Hoekstra (2013) found in one of their 2 studies that there was a significant negative effect of privacy concern on purchase intention. As click-through rate and likelihood of using a service can be seen as proxies or replacements for purchase intent, the findings of Van Doorn and Hoekstra (2013) are in line with what literature would suggest. Adding to this is that there is a negative relationship between trust and purchase intent

(D’Alessandro, Girardi, & Tiangsoongnern, 2012), and as consumers experience lower levels of trust due to privacy concerns this will in turn negatively impact purchase intention. This leads us to the following hypothesis:

(16)

15 Hypothesis 2: The higher the level of privacy concern, the lower purchase intention will be.

An advantage of personalization is that personalized advertisements may increase purchase intention due to a high fit with the preferences of the consumer (Goldfarb & Tucker, 2011b). However, this is a fine line as low fit offers can create irritation (Thota & Biswas, 2009). This is also shown by Van Doorn and Hoekstra (2013) as they found that personalized advertising increased purchase intention, however it also increased feelings of intrusiveness which in turn negatively impacted purchase intention. However, a possible moderator for these privacy concerns was found by Zhu and Chang (2016) in the form of relevance of the personalized ad. They showed that perceived relevance was positively related to consumer’s continuous use intentions, and that relevance mitigates privacy concerns. Furthermore, trust and perceived risk are found to be two principal components which contribute to privacy concerns from consumers, particularly in e-commerce (Eastin et al., 2016). Both also acting as moderators. Therefore, we see that fit (Van Doorn & Hoekstra, 2013; Thota & Biswas, 2009), perceived risk, trust (Martin, 2018; Eastin et al., 2016), and relevance (Zhu & Chang, 2016), all influence the relationship between personalization and purchase intent by influencing privacy concern. This implies that privacy concern has a mediating effect in the relationship between personalization of ads and purchase intent.

Furthermore, if we look at the previous literature we see that there is an increased likelihood of a personalized ad causing higher levels of privacy concern because the more personalized an ad is, the more personal information that is used. This in turn causes a higher amount of sensitive information, which induces higher levels of privacy concern. Furthermore, privacy concern has a negative influence on purchase intent (Van Doorn & Hoekstra, 2013). As we look at this relationship we see that personalization affects privacy concern, privacy concern

(17)

16

affects purchase intent and it is also likely that personalization affects purchase intention. This indicates that it is likely that privacy concern may have some explanatory power on the negative relationship between personalization on purchase intent. This brings us to the following

hypothesis:

Hypothesis 3: Privacy concern is mediating the relationship between personalization and purchase intent.

2.3 Consumer characteristics.

If we look deeper into the relationship between personalization and privacy concern, we see that there is quite some body of research which takes personality or personality traits into

consideration (Bansal, Zahedi & Gefen, 2016; Junglas, Johnson & Spitzmüller, 2008). Previously we’ve seen trust (Martin, 2018), relevance (Zhu & Chang, 2016), fit (Van Doorn & Hoekstra, 2013; Thota & Biswas, 2009) and perceived risk (Eastin et al., 2016) all having a moderating effect on the relationship between personalization and privacy concern. However, the effect of consumer characteristics on privacy concern haven’t been tested with relation to personalization. This can be very helpful for managers in the sense that they are likely to have a database which contains information on consumers. This database combined with shopping habits could provide a way to indicate what type of consumers they have, or which type of consumers they want to target. This could be done based on personality traits. For example, a travel company who offers adventurous trips is likely to target people who score high on openness to new experiences. If they want to approach new potential consumers with similar traits it is important to know if consumers with that personality trait react well to ads which are high into personalization or if they are the opposite.

(18)

17

attitudes towards situations (McCrae & Costa, 1987; Ajzen, 1988). “They reflect who we are and

in aggregate determine our affective, behavioral, and cognitive style” (Mount et al., 2005, p.

449). Furthermore, personality traits are expected to be stable and the change in them is limited, especially after adulthood. This means that personality traits are likely to stay the same across someone’s life. This implies that concerns about privacy are, to some extent, explainable by personality traits and therefore they are unlikely to change over time (Junglas et al., 2008). Not everyone likes the same things, and not everyone experiences the same level of privacy concern when they are targeted with a highly personalized ad. Which indicates that it is beneficial to target some consumers with more personalization than others, as they the level of personalization of the ad induces more, or less feelings of privacy concern. This makes it likely that the reception and effectiveness of the level of personalization differs between consumers. As it is likely that they differ in how receptive they are to those ads and how much privacy concern is experienced.

Research on which set of personality traits was most suitable for research was inconsistent until late in the 1980’s when the Big Five framework was introduced (McCrea & Costa, 1987). The five traits which were selected were agreeableness, conscientiousness, emotional stability, extraversion and openness to experience. Its validity has been tested over time in numerous domains such as job attitudes (Heller et al., 2002), well being (Costa et al., 1987), job performance (Barrick & Mount, 1991) and many more. The overarching names such as agreeableness are the same however, the questions by which they are tested differ between the length of the study.

Openness to experience is related to an individual’s intellect, curiosity, tendency to try

new things and experience new situations. People who are high on openness to experience were found to be exploring, curious, imaginative, empathic and unconventional (McCrae & Costa, 1991). They have been found to enjoy learning (Barrick et al., 2001) and perform well in training

(19)

18

programs. Furthermore, they show high levels of different thinking, liberalism and scientific creativity (Judge et al., 2002). Junglas et al. (2008) found that there was a significant positive relationship between openness to experience and privacy concern, which means that if individuals score high on openness to experience they experience higher levels of privacy concern. This is most likely since individuals who score high on openness to experience have learned a lot due to the high amount of life experiences they have. This gives them a better sense of awareness which leads to them being more sensitive to threatening things (Junglas et al., 2008). As openness to experience is positively related with privacy concern it is likely that individuals with higher levels of openness to experience will experience more privacy concern when it comes to personalized ads. Contrary, it is also likely that individuals with lower levels of openness to experience will experience less privacy concern induced by personalized ads. This leads us to the following hypothesis:

Hypothesis 4: Openness to experience will positively influence the relationship between personalization and privacy concern.

Extraversion is related to a persons’ pleasure in social interactions and how they experience

positive life events. Extraverted individuals are described as talkative, bold, sociable, outgoing and dominant in social situations (Bansal et al., 2016). They are more likely to engage in risky behavior due to their need for arousal (Gullone & Moore, 2000). If we look at introverted individuals we see that they are generally more depressed, cynic and anxious (Bermudez, 1999). Introverts have been found to experience higher intrusions of privacy (Stone, 1986) and place more importance on staying anonymous (Pedersen, 1987). Bansal et al. (2016) found that there was a significant negative relationship between extraversion and privacy concern in terms of e-commerce, the environment of personalized advertisements. However, Junglas et al. (2008) found

(20)

19

no such result when it came to location-based services regarding mobile phones. The explanation for this was that extraversion was not captured correctly in their study. As it not directed at the inside of an individual but rather at the outside, on their behavior, not their experience which is hard to capture for extraverted individuals. Both studies hypothesized a negative relationship between extraversion and privacy concern. Because extraverted individuals have an above average need for social interaction, they are more likely to be concerned with the well being of others than with the safety of their own personal information. If personal information has a minimal effect of the relationship with others, privacy concern is a lesser concern for extraverted individuals (Junglas et al., 2008). Because Bansal et al. (2016) found a significant negative relation in a field that is more related to the field in which this research is conducted, online transactions, we argue that it is more likely to be a negative relation in this study as well. Extraverted individuals are more likely to take risk, which makes them less anxious of negative consequences. Introverts are more concerned with their privacy and with staying anonymous. This means that the higher the level of extraversion an individual possesses, the less he or she is concerned with possible negative consequences. The lower the level of extraversion, the higher the level of concern with negative outcomes. Which makes it likely that individuals with high levels of extraversion experience less privacy concern induced by the personalization of ads, and it makes individuals low in extraversion likely to experience more privacy concern due to the personalization of ads. This leads us to the next hypothesis:

Hypothesis 5: Extraversion will negatively influence the relationship between personalization and privacy concern.

Conscientiousness represents how organized, dependable, and competent and individual is

(21)

20

they strive for excellence, efficiency and they do this by keeping high levels of self-discipline and deliberation (Costa et al., 1991). They are less likely to take risks or to become involved in

dangerous situations. Furthermore, they see deviant behavior as risky and dangerous (Bansal et al., 2016) and are likely to deeply thing about things, follow standards and pay attention to details (Junglas et al., 2008). Conscientiousness was found to have a significant positive effect on

privacy concern, which means that individuals high in conscientiousness experienced higher levels of privacy concern (Junglas et al., 2008). This positive relationship is explained on the basis that conscientious individuals are likely to pay close detail to behavior of others and see deviant behavior such as privacy invasion as threatening and hazardous (Junglas et al., 2008). Which are all aspects related to privacy concern. As conscientious individuals are likely to pay close attention to ads and remember which products they looked at, it is likely that individuals with high levels of conscientiousness will feel more privacy concern due to personalization than individuals with low conscientiousness. On the other hand, individuals with low levels of

conscientiousness are likely to experience less privacy concern due to the personalization of ads. Therefore, we come to the following hypothesis:

Hypothesis 6: Conscientiousness will positively influence the relationship between personalization and privacy concern.

Emotional stability is referred to as the extent to which individuals stay calm during stressful

conditions. When individuals score low on emotional stability it is likely that they are stressed, anxious, volatile or fearful (Goldberg, 1990). They view most of life as less positive compared to individuals with high emotional stability and see life as more stressful (Spector et al., 2000). Furthermore, they take less risk due to higher levels of anxiety, focus on negative events and possible losses (Junglas et al., 2008). This makes them more likely to concentrate more on the

(22)

21

negatives of technology than on the positives. Bansal et al. (2016) found that the emotional stability of an individual was negatively related with concern for privacy. Which seems logical as individuals with low levels of emotional stability are likely to experience more privacy concern as they look at negative aspects rather than at the positives, and if they are more anxious. However, Junglas et al. (2008) found no significant relation between the two. Both studies hypothesized a negative relationship. A negative relationship seems logical because of the specific traits which come with emotional stability or instability, which makes people who are emotionally instable more likely to look more at the negatives of personalization instead of the positives. With the negatives being privacy concern. This brings us to the following hypothesis:

Hypothesis 7: Emotional stability will negatively influence the relationship between personalization and privacy concern.

Agreeableness is the final personality trait of the big five. Agreeableness is the propensity to

conform to social norms. Individuals who score high in agreeableness are looking for harmony and they try to minimalize conflict in their interpersonal relationships (McCrae & Costa, 1991). Furthermore, they are described as warm, kind, generous and loving (Goldberg, 1990). They trust others and they show low levels of suspicion about other individuals or the environment. In contrast, individuals with low levels of agreeableness are found to be arrogant, aggressive and narcissistic (McCrae & Costa, 1991). When it comes to the relationship between agreeableness and privacy concern research is divided. Junglas et al. (2008) found significant evidence for their hypothesis that agreeableness and privacy concern were positively related while Bansal et al. (2016) found significant support for their hypothesis that agreeableness and privacy concern were negatively related.

(23)

22

influence privacy concern. Junglas et al. (2008) uses the protection motivation theory by Rogers (1975) as a foundation. This theory explains how individuals assess threats and how personality plays a role in this assessment. As privacy concern is defined in terms of threats, they feel that the protection motivation theory is suitable to explain the influence of personality traits. On the other hand, Bansal et al. (2016) based their hypothesis that agreeableness was negatively related on the prospect theory. The prospect theory (Kahneman & Tversky, 1979) suggests that the choices of individuals which are subject to uncertainty depend on the utility and disutility of outcomes. Furthermore, it shows that the influence caused by the joy of winning is valued less than the regret of losing.

This research mainly focusses on the negative aspects and threats of personalization and as it sees privacy concern as a mediator for the negative influence on purchase intent we see ourselves more in the protection motivation theory by Rogers (1975). Therefore, we argue that individuals who display high levels of agreeableness are therefore unlikely to see the actions of others as hazardous or harmful when it comes to privacy threats. This is because they are more likely to trust others and they are less likely to be suspicious (Junglas et al., 2008). Furthermore, it is shown that trust reduces the level of privacy concern (Martin, 2018), and as individuals with higher levels of agreeableness are likely to trust others more it seems logical that there is a positive relation. Therefore, we hypothesize that:

Hypothesis 8: Agreeableness will positively influence the relationship between personalization and privacy concern.

(24)

23 2.4 Product characteristics.

In a personalized ad there are more factors than only the level of personalization and who to target. Eventually it all evolves around the product that a company wants to sell. As different products have different characteristics and product characteristics have been found to influence levels of adoption (Lian & Lin, 2008) it seems possible that product characteristics also influence privacy concern.

There is no existing research on product characteristics and their effect on levels of experienced privacy concern. Therefore, we use the protection motivation theory by Rogers (1975) to explain the moderating effect of product characteristics. The protection motivation theory is based on three crucial pillars: the severity of a threatening event; the probability of the occurrence of that event; and the efficacy of a protective response (Rogers, 1975). As it is likely that some products are more privacy sensitive than others, consumers are likely to see some products as a more severe threat than others. For example, if we compare buying a sort of medicine for a private medical condition you want to keep secret to buying a book about

gardening, the severity of the consequences of people seeing you buy the medicine are far bigger than the severity of the consequences of buying a book. So, if we use the protection motivation theory we can say that experienced privacy concern will be higher for privacy sensitive products than for others because the severity of the threatening event is far higher. In the case of

personalized advertisements this would mean that a personalized ad about a privacy sensitive product causes higher levels of privacy concern than a personalized ad about a non-privacy sensitive product. This brings us to the following hypothesis:

(25)

24

As product characteristics are hypothesized to positively influence privacy concern, it is likely that the relationship between personalization and privacy concern is affected by the product and its characteristics which is shown in the ad. If it is a product which is high in privacy sensitivity it seems likely that being targeted with such a product based on personal information will positively moderate the relationship between personalization and privacy concern, and therefore will

increase the levels of experienced privacy concern significantly. It is also possible that a product low in privacy sensitivity will mitigate the effect of personalization on privacy concern. The positive effect of personalization on privacy concern may be lowered because people may be less concerned with the use of personal information in an ad which shows a product which is not privacy sensitive. This brings us to the last hypothesis:

Hypothesis 9b: Product characteristics will have a moderating effect on the relationship between personalization and privacy concern.

(26)

25

3. Methodology

This section will contain the research design through which the hypotheses were tested. First a pre-test will be discussed, after that the experimental design, sample and finally the measures which were used.

3.1 Pre-test.

To test whether the manipulations to the advertisements were strong enough a pre-test was

conducted (N=15). The pre-test consisted of 5 questions which made participants choose between the personalized and non-personalized ad. The final question asked the participants whether they felt the advertisements they saw were considered as personalized. All respondents chose the personalized ads as more personalized and 80% agreed that those advertisements were personalized, which allowed us to continue with the real experiment.

3.2 Experimental design and procedure.

This research conducted a scenario-based online experiment, or so-called vignette study, to test the hypothesis. This type of approach is commonly used in the domain of personalized

advertising (Bleier & Eisenbeiss, 2015). For a personalized and non-personalized ad, we investigated the effects of consumer characteristics and product characteristics on the level of privacy concern and purchase intent. Four conditions were made to which participants were randomly assigned. We made a highly personalized ad and a non-personalized ad and combined this with a privacy sensitive product or a non-privacy sensitive product. A highly personalized ad was made based on the description of Bleier and Eisenbeiss (2015b) which stated that a banner with a high degree of content personalization featured the most viewed category, and brand

(27)

26

combination. In this research we went a bit further and chose the specific product a consumer was shopping for. The non-personalized ad showed no signs of a product or brand. In this case a K-Swiss sneaker in white as non-privacy sensitive product and a Kruidvat Classic condom as a privacy sensitive product. We chose the Kruidvat Classic condom as a highly sensitive product because it has to do with sexual intercourse, which is still regarded as a highly private matter. This indicates that products associated with sexual intercourse are high in privacy sensitivity. This resulted in an experimental design which was a 2 (highly personalized vs. non-personalized) x 2 (privacy sensitive vs. non-privacy sensitive) between-subjects design.

A vignette study uses a short situation or description that is normally shown to respondents to elicit judgements about the scenarios (Atzmüller & Steinter, 2010). The

advantages of a vignette study in the form of a survey are the natural environment it provides to respondents, which eliminates laboratorial effects from the outcomes, and the ease of acquiring respondents.

The experiment consisted of 4 sections. The first section contained information on how their data would be handled and it had a short introduction about what the research would be. However, the aim of the research was not disclosed as this may have led to social desirable answers of participants (DeMaio, 1984). The second section contained some background information about personalized advertising. This was done to make sure that respondents knew what personalized advertisements were and how the information which is used in personalized advertisements is collected. If respondents didn’t know what personalized advertising was or how the data used in personalized advertising is collected, the results of this study would not be valid as the respondents would not know about the whole concept. The third section contained the situation the respondents had to put themselves into. It asked respondents to put themselves into a situation where they needed a specific product and that they wanted to buy this product online.

(28)

27

They specifically looked at the condoms or sneakers, but they were not yet sure if they wanted to buy them and exited the page to think about it. When a different website was opened an hour later the advertisement popped up. The situation was dependent on which condition the respondent was assigned to, the privacy sensitive product of the non-privacy sensitive product. The

difference being the product. The final section contained the survey questions regarding purchase intent, privacy concern and consumer characteristics.

3.3 Sample.

The experiment was conducted with a diverse group of respondents. The respondents came through survey groups, Amazon MTurk, or social connections. Those methods were chosen due to the constraint on resources such as time and money. 447 respondents completed the survey. Before analyzing the data, the data from 85 people was discarded due to them failing the attention check. In this question we asked participants which scenario they had to imagine themselves in. No respondents were removed based upon the completion time. Both these methods are

frequently used to detect participants who did not take the research seriously and therefore made an insufficient effort to answer the questions (Huang et al., 2012; Stenstrom and Curtis, 2012). After those responses were removed from the data 362 respondents remained (Mage = 31, 56,6%

male).

3.4 Measures.

Purchase intention was measured with 4 questions adopted from Moon, Chade and Tikoo (2008). The four items on the scale were: (1) I will purchase the Kruivat Classic Condoms/K-Swiss sneakers; (2) Given a choice, my friends will choose the Kruidvat Classic Condoms/K-Swiss sneakers; (3) There is a strong likelihood that I will buy the Kruidvat Classic Condoms/K-Swiss

(29)

28

sneakers; and (4) I would like to recommend the Kruidvat Classic Condoms/K-Swiss sneakers to my friends. All four items were measured on a seven-point Likert scale, ranging from

1=”Strongly disagree” to 7=”Strongly agree”. Alpha reliability was .872. No items were deleted. Privacy concern was measured with 5 questions adopted from Tolchinsky (1981). The five items on the scale were: (1) It was acceptable for the company to make use of personal information in this ad; (2) It was necessary for the company to make use of personal information in this ad; (3) I feel comfortable with personal information being used in this ad; (4) Greater internal controls are needed in the organization to limit this kind of use of personal information in ads; and (5) The above use of personal information is an invasion of privacy. All five items were measured on a seven-point Likert scale, ranging from 1=”Strongly agree” to 7=”Strongly

disagree”. Alpha reliability was .820. No items were deleted.

The test of consumer characteristics was based on the 10-item measure by Gosling, Rentfrow and Swann (2003). The Big-Five, the 5 most important personality traits, are measured by the TIPI which is a very brief measure which consists of only 10 statements. When compared to other, more extensive, personality test frameworks the TIPI did well, which makes it a viable measure of the Big-Five (Gosling et al., 2003). The 10 statements represented extraversion, agreeableness, conscientiousness, emotional stability and openness to experience. With 5 out of the 10 questions being scored reversely. Regarding extraversion the statements were: (1)

Extraverted, enthusiastic and (2) Reserved, quiet* (α=.603). For agreeableness they were: (3) Critical, quarrelsome* and (4) Sympathetic, warm(α=.225). For conscientiousness they were: (5) Dependable, self-disciplined and (6) Disorganized, careless*(α=.268). Emotional stability was measured with: (7) Anxious, easily upset* and (8) Calm, emotionally stable (α=.528). Finally, openness to experience was measured with: (9) Open to new experiences, complex and (10) Conventional, uncreative* (α=328.). Measures with an asterisk were reversely scored. The

(30)

29

Cronbach Alpha values seem low, however if we compare them to the Cronbach Alpha’s found by Gosling et al. (2003) they are quite similar.

Table 1.

Dimension This study Original research by

Gosling et al. (2003) Extraversion .60 .68 Agreeableness .23 .40 Conscientiousness .27 .50 Emotional Stability .53 .73 Openness to Experience .33 .45

(31)

30

4. Results.

This section of the research will present the results of the data that was collected and analyzed. First the adjustments to the data which were necessary for the analysis will be discussed. After that this section will be structured based on the different hypotheses.

4.1 Data preparation.

After the respondents who didn’t finish the study, and the respondents who answered the attention check question wrong, were deleted 362 respondents remained for data analysis.

According to Dooley (2001) this is a representative amount if an online research is conducted. To analyze the data SPSS was used, and missing values in the data were excluded.

Four dummy variables were created for the different groups regarding personalization and product characteristics. Those four groups were the conditions to which respondents were

assigned, namely: (1) Sneaker Personalized; (2) Sneaker Personalized; (3) Condom Non-Personalized; and (4) Condom Personalized. Another dummy variable was created for

personalization: (0) Non-Personalized ad and (1) Personalized ad. The final dummy variable was created for product characteristics: (0) Non-Privacy Sensitive and (1) Privacy Sensitive. Those dummy variables were created to make a distinction between the different groups and were used to compare the groups to each other.

As respondents were assigned to different groups, their answers were recorded at different questions which made it impossible to analyze purchase intent, privacy concern and consumer characteristics. To make this possible, new variables were created for all the questions regarding these measurers. This new variable included all responses of the different groups to this specific question. This only meant that the values which were previously spread over 4 questions were now collected under one variable and ready for analysis. After that the results of the questions

(32)

31

were added to create PurchaseTOT and PrivacyTOT. The questions regarding the TIPI model had to be combined to show the correct dimension (Gosling et al., 2003). Questions 1 and 6 were combined into “Extraversion”. Questions 2 and 7 were combined into “Agreeableness”.

Questions 3 and 8 were combined into “Conscientiousness”. Questions 4 and 9 were combined into “Emotional stability”. Finally, questions 5 and 10 were combined into “Openness to experience”.

Some questions were reversed and had to be recoded into different variables. Those questions were Privacy4 and Privacy 5 regarding the questions about privacy concern. Questions 2, 4, 6, 8 and 10 were recoded regarding the TIPI model. Finally, all the questions for purchase intention were reversed due to a mistake in the survey.

Table 2.

Hypothesis Prediction Conclusion

H1 The higher the level of personalization, the higher the level

of privacy concern will be. Rejected H2 The higher the level of privacy concern, the lower purchase

intention will be.

Supported H3 Privacy concern is mediating the relationship between

personalization and purchase intent.

Rejected H4 Openness to experience will positively influence the

relationship between personalization and privacy concern. Rejected H5 Extraversion will negatively influence the relationship

between personalization and privacy concern.

Rejected H6 Conscientiousness will positively influence the relationship

between personalization and privacy concern.

Rejected H7 Emotional stability will negatively influence the

relationship between personalization and privacy concern.

Rejected H8 Agreeableness will positively influence the relationship

between personalization and privacy concern. Rejected H9a Product characteristics will have a positive effect on

privacy concern.

Supported H9b Product characteristics will moderate the relationship

between personalization and privacy concern.

(33)

32 4.2 Hypothesis 1

The higher the level of personalization, the higher the level of privacy concern will be.

Hypothesis one assumed that a higher level of personalization of the ad would lead to a higher level of experienced privacy concern. To test this hypothesis a linear regression was performed, but first privacy concern was standardized. In the linear regression personalization was the independent variable and privacy concern the dependent variable. The model was not statistically significant. This tells us that the level of personalization does not significantly influence the level of privacy concern. As personalization does not significantly influence the level of privacy concern, it has no significant positive effect, as was hypothesized. Therefore, hypothesis 1 is rejected.

4.3 Hypothesis 2

The higher the level of privacy concern, the lower purchase intent will be.

Hypothesis two assumed that a higher level of privacy concern would lead to a lower level of purchase intent. To test this hypothesis a linear regression was performed. Purchase intent and privacy concern were first standardized. In the linear regression privacy concern was the independent variable and purchase intent the dependent variable. The model was statistically significant F (1, 360) = 52,0; p< .001 and explained 12,6% of variance in purchase intent. This tells us that a higher experienced level of privacy concern leads to a lower level of purchase intent. If privacy concern would go up by one, purchase intent would go down by .355 (B =.355,

p<.001). As there is a significant negative relationship between privacy concern and purchase

(34)

33 4.4 Hypothesis 3

Privacy concern is mediating the relationship between personalization and purchase intent.

Hypothesis three assumed that privacy concern has a mediating effect on the relationship between personalization and purchase intent. To test this hypothesis an SPSS macro of Hayes (2012) was used which is based upon a regression analysis. There was a significant negative relationship between privacy concern and purchase intent as mentioned in hypothesis two (B =.355, p<.001) There was no significant relationship between personalization and privacy concern as mentioned in hypothesis one. The final relationship between personalization and purchase intent, was also not significant. This shows us that there is no direct effect. If we look at the mediation effect of privacy concern we see that this is also not significant: BC95 = [-0,1320, 0,0647]. It is not

significant as the confidence interval is not different from zero. Therefore, we can conclude that there is no mediation effect. Which rejects hypothesis 3.

4.5 Hypothesis 4-9.

4. Openness to experience will positively influence the relationship between personalization and privacy concern.

5. Extraversion will negatively influence the relationship between personalization and privacy concern.

6. Conscientiousness will positively influence the relationship between personalization and privacy concern.

7. Emotional stability will negatively influence the relationship between personalization and privacy concern.

(35)

34 8. Agreeableness will positively influence the relationship between personalization and privacy

concern.

9. Product characteristics will have a positive moderating effect on the relationship between personalization and privacy concern.

Hypothesis 4 to hypothesis 9 are tested in the same way, as all hypotheses contain a moderating effect on the same relationship. The only difference is which variable is moderating. To test these hypotheses a linear regression was used including an interaction effect. According to Aiken et al. (1991) the use of interaction to test for moderation is an appropriate method. As concluded at hypothesis 1, there is no significant effect of personalization on privacy concern. Therefore, it will be left out of this part of the results. As it has already been tested and reported.

According to hypothesis 4 openness to experience will positively influence the relationship between personalization and privacy concern, thus a positive moderating effect. Openness to experience had a significant effect on privacy concern (B = ,157; p<,001) and this relationship was positive as was found by Junglas et al. (2003). However, the interaction effect between personalization and openness to experience had no significant effect on the relationship between personalization and privacy concern. Therefore, there is no significant evidence for a moderation effect of openness to experience. This means that hypothesis 4 is rejected.

Hypothesis 5 states that extraversion will negatively influence the relationship between personalization and privacy concern, thus a negative moderating effect. Extraversion had no significant effect on privacy concern. Which is in line with the findings of Junglas et al. (2003) who also found no significant effect. Furthermore, the interaction effect between personalization and extraversion had no significant effect on the relationship between personalization and privacy concern. Therefore, there is no significant evidence for a moderation effect of extraversion. This

(36)

35

means that hypothesis 5 is rejected.

Following hypothesis 6 there is a moderating effect of conscientiousness on the relationship between personalization and privacy concern, thus a positive moderating effect. Conscientiousness had a significant effect on privacy concern (B = ,211; p<,001) and this relationship was positive. Which is in line with the findings of Junglas et al. (2003) who also found a significant effect. However, the interaction effect between personalization and conscientiousness had no significant effect on the relationship between personalization and privacy concern. Therefore, there is no significant evidence for a moderation effect of conscientiousness. Hypothesis 6 is therefore rejected.

According to hypothesis 7 emotional stability will negatively influence the relationship between personalization and privacy concern, thus a negative moderating effect. Emotional stability had a significant effect on privacy concern (B = ,145; p<0.05) and this relationship was positive. This is not in line with previous research (Junglas et al., 2003; Bansal et al., 2016) which found no effect and a negative effect. Furthermore, the interaction effect between personalization and emotional stability had no significant effect on the relationship between personalization and privacy concern. Therefore, there is no significant evidence for a moderation effect of emotional stability. Which makes us reject hypothesis 7.

Hypothesis 8 states that agreeableness will positively influence the relationship between personalization and privacy concern, thus a positive moderating effect. Agreeableness had a significant effect on personalization (B = ,137; p<,001) and this relationship was positive. This is in line with Bansal et al. (2016) who also found a positive effect. However, it is contradicting Junglas et al. (2003) who found a negative effect. The interaction effect between personalization and agreeableness had no significant effect on the relationship between personalization and privacy concern. Therefore, there is no significant evidence for a moderation effect of

(37)

36

agreeableness. Which makes us reject hypothesis 8.

According to hypothesis 9 product characteristics will have a positive effect on privacy concern (a) and product characteristics will moderate the relationship between personalization and privacy concern (b). Product characteristics, whether a product was privacy sensitive of not, had a significant effect on privacy sensitivity (B = ,369; p<,001) and this relationship was positive. Thus, providing support for hypothesis 9. However, there was no significant interaction effect of personalization and product characteristics on the relationship between personalization and privacy concern. Not providing support for hypothesis 9b.

(38)

37

5. Conclusion and discussion.

This section will contain the findings of this research and how these findings contribute to

existing literature and the managerial implications of it. Furthermore, the results will be discussed and compared existing literature and the conceptual framework. Suggestions and possible

explanations will be given for results which are different from what was expected. Furthermore, this section will also contain the remarks on external validity and suggestions for further research.

The main contributions to existing literature are the findings on the interplay between the personalization of ads, privacy concern and purchase intention. Furthermore, this paper will contribute to the knowledge on the relationship between the personalization of ads and privacy concern. In specific, how personality traits and product characteristics influence this relationship. The findings and interpretations will be discussed more elaborately below.

This research started with the aim to give a better insight into the personalization of ads. The personalization of ads has been controversial. On the one hand there are the benefits for companies and consumers. Companies are now able to identify viable customers more easily and market more cost efficient (Kim et al., 2001). Consumers are now able to identify relevant ads more quickly and minimize the time they look for products (Srinivasan et al., 2002). However, the personalization of an ad is an art. As there is a fine line between being intrusive, relevant and consumers being nervous about the collection of data to make personalized ads possible (Awad & Krishnan, 2006; Xu et al., 2011). To give a better understanding this research has focused on two main aspects. Firstly, this paper hypothesized privacy concern to be a mediator in the relationship between the personalization of ads and purchase intent. Secondly, it tried to give a better

understanding of the relationship between personalization and privacy concern.

(39)

38

leads to a higher level of privacy concern. This is shows that the level of personalization does not affect privacy concern, which is interesting. If we look at existing literature, we see that Debatin et al. (2009) found similar results in the context of social media platforms. However, Ham (2016) found that a higher level of personalization caused higher levels of perceived privacy concern. The rejection of hypothesis 1 tells us that consumers do not experience higher levels of privacy concern when it comes to personalized ads. Possible explanations for this could be that

consumers have become more used to personalized ads which decreases the privacy concern they raise, or they believe they are more protected due to new laws being implemented. This happened recently in Europe, which was also the continent that provided most respondents. Furthermore, it may that personalized ads are subject to the privacy paradox. The privacy paradox is that

consumers are still willingly giving their personal information to others, despite reporting high levels of privacy concern (Smith et al., 2011). Consumers say that they feel that personalized ads are intrusive and infringe privacy, however when it comes to the actual reported levels of privacy concern, there is no difference.

Hypothesis 2 was supported which confirms that privacy concern has a negative impact on purchase intention. This is in line with existing literature (Van Doorn & Hoekstra, 2013) and shows managers that it is still worthwhile to make sure that privacy concerns are minimized. As this relationship was quite strong, it is recommended to make sure there are no factors in the marketing strategy contributing to privacy concern. As they could have detrimental effects on sales. Especially in an online setting.

The mediating role of privacy concern the relationship between personalized ads and purchase intent, as hypothesized, was not present. However, this is not conclusive due to a couple of reasons. This research tried to focus on the negative aspect of the relationship between

(40)

39

personalization and privacy concern. The same was true for the relation with purchase intent. This may be different for research which solely focusses on the negative effect of personalization on purchase intent. This research still feels that privacy concern may play a role as a mediator in this negative part of the relationship. However, if we look at the personalization of ads in total there is no mediating effect of privacy concern.

The results regarding the hypothesis that personality traits act as moderators on the relationship between personalization and privacy concern show that none of the personality traits act as moderators. When we look at the direct effect of personality traits on privacy concern we see that our findings give some clarity in a contradicting environment of existing research (Bansal et al., 2016; Junglas et al., 2003). We found significant positive relationships for agreeableness, emotional stability, conscientiousness and openness to experience. Our findings on agreeableness are in line with Bansal et al. (2016), while our findings on conscientiousness and openness to experience are in line with Junglas et al. (2003). These findings indicate that agreeableness has a positive effect on privacy concern, and not a negative effect as found by Junglas et al. (2003). However, it also introduces more confusion as extraversion has a positive effect on privacy concern, while previous research found that it had a negative effect (Bansal et al., 2016). One thing which was interesting was that all significant results found by this research had a positive effect on the privacy concern. This was not hypothesized and not found by other studies. This may be due to a change in consumer perception, however it is also possible that it had something to do with this specific study and the way it was conducted.

As different results are found in the same field, ecommerce, we can conclude that the effect of personality traits on privacy concern most likely differs to much between studies to be important enough to take into consideration at this point in time. This is also strengthened by the fact that no support was found for any of the personality traits having a moderating effect. Which

(41)

40

shows marketeers that personality traits should not be a high priority when it comes to reducing privacy concern in personalized ads.

Another new addition to the existing literature was the introduction of product

characteristics to the relationship of privacy concern and personalization. As this phenomenon was not yet researched, or minimally regarding this specific field. There was no foundation to build hypotheses on. We found support for our hypothesis that product characteristics positively influence privacy concern. Which tells us that characteristics of a product, in this case privacy sensitive or non-privacy sensitive, influence perceived privacy concern. This gives managers another aspect to think about besides trust (Martin, 2018), relevance (Zhu & Chang, 2016) and fit (Van Doorn & Hoekstra, 2016; Thota & Biswas, 2009) which they should take into consideration when promoting their product. It adds to literature in such a way that product characteristics are a new and valuable factor to look at when considering privacy concern. This research found no moderating effect for privacy sensitive or non-sensitive products in the relationship between personalization and privacy concern. However, the level of privacy sensitivity of the product is only one characteristic of a product. Products have more characteristics which could possibly influence this relationship such as tangibility or intangibility. Furthermore, it is almost impossible to change product characteristics, therefore it is not viable for managers to think of it as a way to reduce privacy concern. However, it could guide managers on how to promote their product or give an explanation on the level of privacy concern consumers experience when they are buying the product, which wasn’t thought about before. This research found a positive effect of product characteristics on privacy concern, and no moderating effect on the relationship between

personalization and privacy concern. However, this doesn’t mean that product characteristics should not be considered, they are a new factor in the field of privacy concern, just not when combined with the personalization of ads.