1
Personalization-privacy paradox in social media context:
the relationship between perceived ad relevance and
perceived privacy invasion
How perceived ad relevance and perceived privacy invasion interrelate to attitude towards the ad and the moderating role of perceived privacy invasion on the relationship between
perceived ad relevance and attitude towards the ad
by
Ngoc Huynh Bao Nguyen
2
Personalization-privacy paradox in social media context:
the relationship between perceived ad relevance and
perceived privacy invasion
How perceived ad relevance and perceived privacy invasion interrelate to attitude towards the ad and the moderating role of perceived privacy invasion on the relationship between
perceived ad relevance and attitude towards the ad
by
Ngoc Huynh Bao Nguyen
MSc Marketing Management Faculty of Economics and Business
University of Groningen
Master Thesis 8th January 2021
nguyen.1@student.rug.nl S4089553
3 ABSTRACT
Personalization-privacy paradox has received much attention in research in recent years. The paradox posits that consumers might respond favorably to personalization and personalized ads in particular because of its provision of useful and relevant information; however, they could respond unfavorably due to a perceived breach of privacy. Therefore, it is important to study when personalization will be effective by understanding the relationship of the paradox’s two facets: relevance and privacy invasion. Meanwhile, consumers' use of social media is constantly rising, especially during lockdowns due to the pandemic. This growing social media use together with the advancement of data tracking and mining technology have brought in promising yet challenging marketing opportunities, one of which is personalized ads. As a result, knowing how the paradox works in social media would provide not only theoretical insights but also practical implications. This present study aims to investigate how perceived ad relevance and perceived privacy invasion interrelate to attitude towards the ad, and how perceived privacy invasion moderates the relationship between perceived ad relevance and attitude towards the ad in social media context. For this purpose, a correlational design deploying an online survey was carried out (N = 622). The regression analysis results indicated that perceived ad relevance positively explained attitude towards the ad while perceived privacy invasion negatively related to attitude towards the ad. In addition, the moderation analysis result revealed that perceived privacy invasion strengthened the positive relationship between perceived ad relevance and attitude towards the ad. This finding challenges the commonly used theory of the psychological reactance, used to explain that perceived privacy invasion weakened this relationship (because consumers tend to react negatively to gain back their freedom when it is threatened by perceived privacy invasion). Instead, the theory of self-awareness could offer an alternative explanation. According to this theory, consumers attribute themselves, the “self” object, as the cause of all events occurring to revert back to normal state. A theoretical implication of this finding is that personalization’s effectiveness can be maximized by manipulating consumers’ self-awareness.
Keywords: personalization-privacy paradox, perceived ad relevance, perceived privacy invasion, attitude towards the ad.
4 PREFACE
As a marketing student who uses social media quite often, I believed that it is necessary and practical to study how marketing works in a social media context. More particularly, this study examines a situation that anyone using social media might at least encounter once: seeing a personalized ad. We might like a personalized ad because it provides catering information, saving us time and effort. However, simultaneously it raises a question of how our privacy was treated. I enjoyed reading the related materials and designing the research that investigated these two sides of personalized ads via consumers’ attitude towards them.
I would like to express my sincere appreciation towards my supervisor Dr. J.I.M de Groot for her guidance, advice and positivity that have helped me go through this whole thesis process. I would also like to wholeheartedly thank my parents and sister for always caring for and cheering me up along the way; my best friend P who has encouraged and made me believe more in myself. Their enormous support was a great source of motivation and I could not have made it without them. Finally, I am grateful for the classmates that I have met, the experiences that I have been through, the knowledge and skill sets that I have obtained during this Master program. All of them have contributed to the completion of my Master thesis.
I hope you enjoy reading this study.
Ngoc Nguyen
Groningen, 8th January 2021
5 TABLE OF CONTENT
I. INTRODUCTION ... 6
II. LITERATURE REVIEW AND HYPOTHESES ... 9
2.1. The importance of investigating consumer’s attitude towards the ad ... 9
2.2. Perceive ad relevance and attitude towards the ad ... 10
2.3. Perceived privacy invasion and attitude towards the ad ... 11
2.4. The moderating role of perceived privacy invasion on the relationship between perceived ad relevance and attitude towards the ad ... 13
2.5. Conceptual model ... 13
III. METHODOLOGY ... 15
3.1. Research design ... 15
3.2. Materials ... 16
3.3. Sample selection and sampling technique ... 17
3.4. Procedure ... 18
3.5. Measures ... 18
3.6. Plan of analysis ... 21
IV. RESULTS ... 23
4.1. Correlation, reliability and assumptions test ... 23
4.2. Relationships between perceived personalization, PAR and PPI ... 23
4.3. Direct relationships between PAR and PPI on Aad. ... 23
4.4. The moderating effect of PPI on the relationship between PAR and Aad... 24
V. DISCUSSION AND CONTRIBUTIONS ... 26
VI. LIMITATIONS AND SUGGESTIONS FOR FUTURE RESEARCH ... 29
VII.CONCLUSION ... 31
REFERENCES ... 32
6 I. INTRODUCTION
Social media has now become part of our daily lives shown by people creating and sharing numerous information and content via their social media accounts (Jung, 2017). Many social media companies now have billions of active users and this number does not show any sign of slowing down. A recent report by Statista (2020) reveals that there are currently 3.6 billion social media users worldwide and this number is projected to reach 4.41 billion in 2024. Likewise, the daily usage time spent on social media of internet users across the globe has been constantly growing since 2012, up to an average of 144 minutes in 2019 (Statista, 2020). Looking from a marketing and advertising perspective, social media seemingly offers a great opportunity to efficiently approach, communicate and maintain relationships with potential consumers. Not surprisingly, an annual growth rate of social media ad spending is expected to reach 7.6%, leading to a market volume of more than US$130,000 by 2024 (Statista, 2020).
The growing opportunities to use social media in marketing and advertising has resulted in a whole new field of social media marketing. Tuten and Solomon (2015) defined social media marketing as ‘‘the utilization of social media technologies, channels, and software to create,
communicate, deliver and exchange offerings that have value for an organization’s stakeholders” (p.21). In social media, this possibility happens partly because of the active
participation of users by creating profiles and sharing contents about their demographic, geographic and/or psychographics (Jung, 2017). Taking advantage of these readily available information and sophisticated social media technologies and software, companies can create personalized and relevant advertising content that is presumably fitting the right targeted audiences at the right time (Tam & Ho, 2006). Such employment is believed to bring value for consumers - one of the stakeholders of a company (Tam & Ho, 2006). In academic research, this practice is often referred to as personalization or personalized advertising (Sutanto et al., 2013; Aguirre et al., 2015; Zhu & Chang, 2016; Li, 2016; Jung, 2017).
7 Srinivasan, 2003). Therefore, this study focuses on attitude towards the ad to measure the effectiveness of personalized ads.
Research has pointed out that consumers typically respond in two opposite ways when exposed to personalized ads. Explanation for such response is provided by the privacy calculus theory (Laufer & Wolfe, 1977), which depicts the process of how consumers assess the potential benefits and the potential costs of personalized ads (Boerman et al., 2017). On the one hand, consumers respond favorably to the displayed personalized ad either because it is perceived to provide relevant information or because it addresses them personally (Jung, 2017). According to the privacy calculus theory, consumers could consider perceived ad relevance a benefit (Zhu & Chang, 2016) which could potentially result in a more favorable attitude towards the ad (Debevec & Iyer, 1988). On the other hand, consumers respond unfavorably to the seen personalized ad because it is perceived to violate their privacy (Zhu & Chang, 2016). According to the privacy calculus theory, consumers could view perceived privacy invasion as a cost (Zhu & Chang, 2016), which could result in a less favorable attitude towards the ad. This conflicting response is referred to as the personalization-privacy paradox (Sutanto et al., 2013; Aguirre et al., 2015).
9 II.LITERATURE REVIEW AND HYPOTHESES
2.1. The importance of investigating consumer’s attitude towards the ad
Attitude towards the ad is not a novel studied subject in advertising and marketing. In fact, it holds a long research history way back to 1925 (Muehling & McCann, 1993). Mitchell and Olson (1981) suggested that attitude towards the ad indicated an individual’s evaluation of the overall advertising stimulus. Likewise, Phelps and Thorson (1991) defined this construct as “a
viewer’s general liking or disliking of an advertisement” (p.202). In these conceptualizations,
attitude is perceived as a unidimensional construct (Muehling & McCann, 1993). In this research, attitude towards the ad is defined as a consumer’s general evaluation of the personalized advertisement.
Although attitude towards the ad does not immediately relate to e-loyalty, research in various fields has found its indirect relationship with e-loyalty. For example, in an online retail context, Srinivasan et al. (2002) and Anderson and Srinivasan (2003) showed that repeat purchase is a sign of e-loyalty. Additionally, Shimp (1981) found that attitude towards the ad could influence purchase intention. Thus, it can be inferred that a positive attitude towards the ad could increase purchase intention which eventually results in greater e-loyalty.
In addition, a positive attitude towards the ad is likely to generate a chain of positive outcomes. Firstly, the impact of attitude towards the ad has been found on brand attribute beliefs (Hastak & Olson, 1989), perceptions of ad credibility and ad persuasiveness (Gelb & Picket, 1983), brand cognitions (Homer, 1990) and brand recall and recognitions (Zinkhan, Locander & Leigh, 1986). Accordingly, a positive attitude towards the ad likely leads to a positive chain of cognitive responses which could eventually generate positive advertising outcomes. In today’s social media context where there are thousands of brands trying to occupy consumers’ minds, these factors appear to be particularly critical in making the brand image salient in people’s minds (Keller, 1993), which potentially enhances e-loyalty.
10 e-loyalty. In social media where consumers are free to share their opinions, this is survival because if they carry a good brand image, the brand can benefit from positive e-word-of-mouth and vice versa.
Lastly, there are evidences supporting the direct effect of attitude towards the ad on consumer behaviors such as purchase interest (Machleit & Kent, 1989; Machleit, Madden & Allen, 1990), brand consideration (Moore & Hutchinson, 1983) and repeat purchase (Shimp & Yokum, 1981). Again, as the purpose of an ad is to evoke purchase decision and increase e-loyalty, these effects emphasize the significance of attitude towards the ad being the study subject. However, this is not the case, specifically in relation to the personalization-privacy paradox in social media context. In particular, perceived ad relevance has been empirically found to be a contributing factor in the success of advertising in general (Jung, 2017). Therefore, the same relationship can be expected between perceived ad relevance and attitude towards the ad.
2.2. Perceive ad relevance and attitude towards the ad
Celsi and Olson (1988) defined relevance as “the extent (to which) consumers perceive the
object/situation/action to be self-related or in some way instrumental in achieving their personal goals and values” (p.211). Accordingly, in a personalized ad in social media context,
perceived ad relevance can be defined as the extent to which consumers perceive the personalized ad to be self-related or in some way instrumental in achieving their personal goals and values.
11 The relevance of a personalized ad is created by incorporating consumers’ self-related data such as demographics (Jung, 2017), past online behavior (Boerman et al., 2017), and past purchases (Tam & Ho, 2006). Therefore, a relevant ad could save consumers time and effort because it provides the right information (Zhu & Chang, 2016). Based on the privacy calculus theory (Laufer & Wolfe, 1977), in this case, the increased perception of ad relevance is considered a benefit that potentially could result in a positive attitude towards the ad. In this context, this positive effect is expected to be stronger because consumers have a tendency to think that they own their social media (Karahanna, Xu & Zhang, 2015). Thus, it could be argued that anything that is deemed relevant would be quickly spotted and adopted. All in all, the following hypothesis is proposed:
H1: The more consumers perceive a personalized ad as relevant, the more favorable their attitude towards the ad.
2.3. Perceived privacy invasion and attitude towards the ad
Personalized ad does not only have positive outcome but also negative one, which is the perception of privacy invasion (Zhu & Chang, 2016; Jung, 2017). In research regarding the personalization-privacy paradox, when referring to privacy related arguments, “privacy concern” appears to be used the most. Girondaa and Korgaonkarb (2018) state that “privacy
concerns would be worrying about a violation of one’s privacy happening, and invasiveness would be the realization or perception that a violation has actually happened” (p.68). Hence,
12 privacy invasion refers to the perception of a consumer that his or her privacy has been violated by the personalized ad.
A personalized ad indeed could evoke a perception of privacy invasion. This could be explained by two reasons. Firstly, marketers usually rely on consumer’s data to create relevant content, otherwise the so-called personalized ad will not work (Baek & Morimoto, 2012). This utilization potentially raises the issue of privacy invasion (Zhu & Chang, 2016; Baek & Morimoto, 2012). Studies about privacy invasion have shown its negative effect on ad’s effectiveness (Phelps, D'Souza & Nowak, 2001) in various contexts including online personalized ad (Baek & Morimoto, 2012). Moreover, as per Girondaa and Korgaonkarb’s (2018) definition of privacy invasion, it can then be expected that a perception of privacy invasion is likely to generate a less favorable attitude towards the ad. The reason behind this is that consumers are “informed” immediately that their personal data has been used as soon as they see the personalized ad (Boerman et al., 2017; Aguirre et al., 2015). Drawing on the privacy calculus theory (Laufer & Wolfe, 1977), consumers would consider this use of their personal data as a cost. Moreover, in social media, consumers tend to develop a much stronger sense of ownership throughout the usage time even though they do not actually own it (Karahanna, Xu & Zhang, 2015). Therefore, consumers might express a less favorable attitude if they see an (personalized) ad within their social media environment than other online contexts such as e-commerce websites or online news/magazines. Secondly, people use social media for a purpose. Whiting and Williams (2013) identified that there are ten different use and gratification motivations which explain why consumers engage with social media. Accordingly, when people are involving in a certain task to fulfill these motivations, they are more likely to feel negative about any interruption. In this research, such interruption is the personalized ad. In a context of social media which consumers “claim” to be their own possessions (Karahanna, Xu & Zhang, 2015), they might deem such interruption invasive because it shows up “uninvited” in that “online property”. According to the privacy calculus theory (Laufer & Wolfe, 1977), consumers would evaluate this interruption of the personalized ad as a cost. Consequently, consumers are likely to develop a less favorable attitude towards the ad. Based on the aforementioned arguments, the following hypothesis is proposed:
13 2.4. The moderating role of perceived privacy invasion on the relationship between perceived ad relevance and attitude towards the ad
According to the information boundary theory (Sutanto et al., 2013), when consumers consider the collecting and use of their personal information as harmful or too uncomfortable, its costs will not outweigh its benefits (Boerman et al., 2017). In an online context such as social media, privacy is related to the access to one’s personal information (Baek & Morimoto, 2012; Jung, 2017). Consumer’s perception of privacy invasion reflects the acknowledgement that their personal information has been accessed and/or used. This could mean to consumers that their control over personal data is threatened, causing them to consider the use of personal information as harmful or too uncomfortable (Boerman et al., 2017). Subsequently, they would express reactance to regain their freedom and autonomy (Boerman et al., 2017; Baek & Morimoto, 2012). Such reactance has its root in the psychological reactance theory (Brehm, 1966), which could be escalated because consumers tend to think that they own their social media account and everything appears in it (Karahanna, Xu & Zhang, 2015). This possessive desire can be explained by the psychological ownership’s underlying need elaborated as “being
able to control and be effectant in altering the environment” (Pierce et al., 2001; p. 300).
Therefore, consumers alter their attitudes to reaffirm the control position (Baek & Morimoto, 2012). With such an intense protective psychological process, even if the personalized ad is deemed relevant, it could be argued that perceived privacy invasion could mitigate the positive relationship of perceived ad relevance and the attitude towards the ad. Given the above arguments, the following hypothesis is proposed:
H3: Perceived privacy invasion moderates the relationship between perceived ad relevance and attitude towards the ad in such a way that the positive relationship between perceived ad relevance and attitude towards the ad is weakened, if consumers perceive a higher degree of privacy invasion caused by the ad.
2.5. Conceptual model
14 H1 (+) Perceived ad relevance Attitude towards the ad Perceived privacy invasion
Figure 1. Conceptual Model
15 III.METHODOLOGY
3.1. Research design
In the present study, there were two main purposes in understanding the relationships between perceived ad relevance (PAR), perceived privacy invasion (PPI) and attitude towards the ad (Aad). First, it was to understand how PAR (Hypothesis 1) and PPI (Hypothesis 2) interrelated to Aad, thus providing a potential explanation for the personalization-privacy paradox. Second, it was to explain how PPI moderates the relationship between PAR and Aad (Hypothesis 3), hence identifying the condition under which personalization worked. To analyze these three hypothesized relationships, a correlational research design was called upon. This was because of the following two reasons: (1) this particular research type is conducted to determine the degree to which marketing variables are related and (2) it is used when specific hypotheses have been developed (Malhotra, 2010).
16 and PPI was changed from fixed (first PAR then PPI) to randomized to reduce bias in how respondents form their final attitude towards the ad. In particular, questions about PAR were asked first followed by questions about PPI for one respondent, but for another respondent questions about PPI were asked first followed by PAR. Doing it this way could reduce respondents’ bias that could have been a result of how the questions were presented.
3.2. Materials
Traveling was chosen as the product in the survey because it offers a higher level of personalization than other products such as shoes, furniture, clothes or food/drinks. That is, consumers get to decide where they want to go in the first place while they can only choose from certain offerings with the other products. The scenario used in the survey was adopted from the paper by Li (2016) with adjustments to fit with the current research. Specifically, respondents were asked to imagine themselves in the following situation:
“It has been more than a year since the COVID-19 appeared, causing travel bans to be implemented almost everywhere in the world. As someone who likes to travel for leisure, you
can't wait to do it again as soon as the COVID-19 is over. Your destination of interest is [destination of choice] and thus you have been searching for trips to the place on Google.
Later in the day, when you are using your [social media of choice], you see the following advertisement about trips to [destination of choice] with special price offerings:
Hi [respondent’s name], been dreaming about visiting [destination of choice]. Dream no more as we’ve got what you need! Travel itinerary designed exclusively for you.
Contact us now to get up to 50% discount. [Image]”
(see Appendix A)
17 3.3. Sample selection and sampling technique
Given the scenario, the target population included people who owned a social media account such as Facebook, Instagram, Snapchat, etc. Moreover, respondents needed to have a good English reading skill because the survey was written entirely in this language. In order to easily reach such pre-conditions, the survey was distributed via the student researcher’s social media network using a convenience sampling strategy (Malhotra, 2010). Then, the survey was spread to a wider population by asking the first patch of respondents to share it via their own networks. This method is known as snowball sampling (Malhotra, 2010).
Using an a-priori sample size analysis by Soper (2020) with a small anticipated effect size (f2= 0.10), a minimal desired statistical power of 0.80, a probability level of 0.05, and eight predictors (perceived personalization, PAR, PPI, the interaction between PAR*PPI, age, income, frequency use of social media, and travel liking), a minimum sample of 158 participants was required (Cohen et al., 2003). After removing incomplete responses, there were 622 responses left including 206 responses that failed the attention check (i.e. When being
asked about which color appeared in the image, please select "purple". Based on the sentence above, please indicate which color appeared in the advertising image: 1- Orange, 2- Purple).
Since 206 was a large number, the analysis was carried out using 2 different sample sizes: N1 = 622 (including the 206 respondents that failed the attention check) and N2 = 416 (excluding the 206 participants that failed the attention check). Based on the minimum sample of 158, both samples were sufficient for the aim of the present study. These two samples generated the same results; therefore, the reported analyses in this thesis only included the results of the larger sample (N1 = 622).
18 3.4. Procedure
To start off the data collection process, the online survey was created using Qualtrics. The survey link was then spread through the researcher's social media channel as this allowed it to reach a number of respondents fast. The survey started by asking respondents for their socio-demographics information, if they liked traveling (if not, they were sent to the end of the survey because the advertised topic was traveling), how much they liked it and their preferred choice of travel destination. Then, it asked if they owned a social media account. If not, the survey ended. If yes, it asked which social media platform they used the most and the frequency of using it. Next, the scenario (that incorporated the social media and destination as provided by respondents) was shown. Respondents were asked to read the text describing the scenario in which the advertising was displayed. After this, respondents were asked to rate the items used to measure the constructs. The items that measured perceived personalization were shown first, followed by those that measured PAR and PPI shown in randomized order. Appearing last were those items measuring Aad. This way, respondents had some time to reflect on their own evaluation to enable them to develop their attitude towards the ad.
The survey was distributed on November 18th, 2020 and was left accessible for 7 days, until November 25th, 2020.
3.5. Measures
All the items used to measure the constructs in this study were adopted from previous studies so as to ensure reliability and validity. These items were measured on a 7-point Likert scale (with 1 being “strongly disagree” and 7 being “strongly agree”) and a 7-point bipolar semantic differential scale. A 7-point scale was chosen over a 5-point scale because the psychometric literature suggests that having more scale points is better but it should not be more than 11 points to avoid diminishing return (Nunnally & Bernstein, 1978).
19 adopted from Li (2016). They were 1- “The advertisement that I just saw on my [social media
of choice] seems to be designed specifically for me” and 2- “The advertisement that I just saw on my [social media of choice] targets me as a unique individual”. Mean scores revealed that
most respondents somewhat realized that the displayed ad was personalized (M = 4.34, SD = 1.47, Cronbach’s α = .71).
3.5.1. Perceived ad relevance (PAR)
The first independent variable of the study was PAR, which was measured on a 7-point Likert scale using ten items developed by Laczniak and Muehling (1993). In order to fit the question to the current study, it will be modified as shown below. Mean scores indicated that in general respondents somewhat did not perceive the ad to be relevant (M = 3.92, SD = 1.12, Cronbach’s α = .91).
“After seeing the advertisement, please indicate how much you disagree or agree with the following statements:
● The advertisement that I just saw on my [social media of choice] is useful to me ● The advertisement that I just saw on my [social media of choice] is created just for me ● The advertisement that I just saw on my [social media of choice] is of value to me ● The advertisement that I just saw on my [social media of choice] is relevant to my needs ● The advertisement that I just saw on my [social media of choice] is important to me ● The advertisement that I just saw on my [social media of choice] is meaningful to me ● The advertisement that I just saw on my [social media of choice] is worth remembering ● The advertisement that I just saw on my [social media of choice] is worth paying
attention to
● The advertisement that I just saw on my [social media of choice] is interesting to me ● The advertisement that I just saw on my [social media of choice] is likely to give me
new ideas”
3.5.2. Perceived privacy invasion (PPI)
20 order to help respondents understand the concept of privacy invasion in this research, a brief explanation was given before asking them to rate the items. The explanation was as follows:
“In this personalized ad, your personal information (in this case, your previous search about the trip to [destination of choice]) was used. After seeing that advertisement, please indicate how much you disagree or agree with the following statements:
● The above use of personal information is an invasion of privacy. ● That personalized advertising is an invasion of my privacy.
● My privacy is invaded by the way that personalized advertising was conducted. ● Personalized advertising violates my right to privacy.
Mean scores showed that a majority of respondents perceived that their privacy had been invaded (M = 4.83, SD = 1.35, Cronbach’s α = .93).
3.5.3. Attitude towards the ad (Aad)
The dependent variable of the study was Aad, which was measured on a 7-point bipolar semantic differential scale using ten items adopted from Muehling and McCann (1993). The following question was asked:
“Please indicate how accurately one or the other adjective describes your overall evaluation of the advertisement: ● Bad - Good ● Dislike – Like ● Unfavorable - Favorable ● Awful - Nice ● Unpleasant - Pleasant ● Uninformative - Informative ● Uninteresting - Interesting ● Unappealing - Appealing ● Unenjoyable - Enjoyable ● Not irritating - Irritating”
21 3.6. Plan of analysis
All of the analyses were performed in IBM SPSS version 25, provided by the University of Groningen. In addition, in all analyses, socio-demographics variables, specifically age, income, and frequency were controlled for based on past papers (Li, 2016; Zhu & Chang, 2016; Jung, 2017; Boerman et al., 2017); and as stated in section 3.1, travel’s liking was controlled for as well. Regarding age, previous research found that younger people tend to accept personalization more in general (Boerman et al., 2017). Moreover, it is realistic to control for income because it could play a fairly vital role in people’s lives and in deciding if and how much to spend on travel decisions in particular (Lai, Chen & Petrick, 2016). This is to say, income could be an influential factor when respondents perform the role-play in the scenario. As for frequency of using social media, it was included as an indicator of online experience, which has been found to influence perception of personalization (Lee et al., 2015; Miyazaki, 2008). That is, the more consumers use social media, the more they might get familiar with “seeing” advertisements, which could get to an extent in which they do not notice that it is personalized at all.
There were two stages in the analyses: the first stage was to inspect and prepare data; and the second stage was to test the hypotheses.
In the first stage, to inspect data, missing values and outliers were identified and replaced by mean values. Then, to prepare data, all assumptions to perform multiple regression analyses, the item's correlation and reliability were checked. The two assumptions for regression analysis that were checked included: (1) there was a linear relationship between the dependent and the independent variable, and (2) the residuals were normally distributed. A Pearson correlation checked if items measuring a construct correlated well with one another and hence could depict a whole construct up in one score (Malhotra, 2010), which needed to be positive and p-value needed to be smaller than 0.05. If the reliability measured by Cronbach’s alpha was acceptable (> 0.60), a sum variable was computed for each theoretical construct (Malhotra, 2010).
22 was perceived, and that the more personalized they perceived an ad to be, the more invasive it was perceived. Next, testing for the 3 hypotheses was executed.
To test Hypothesis 1 and 2, a multiple regression was executed with the independent variables PAR and PPI, and the dependent variable Aad. With this approach, the relationships between PAR, PPI and Aad were checked while taking into account the influences of PAR and PPI on each other. Then the controlled variables were added as secondary independent variables to check if they impacted the effect of PAR and PPI on Aad. In a multiple regression, the coefficient of determination (R2), which tells how much variance in the dependent variable (Aad) was explained by the independent variables, was looked at first. R2 values should be between 0 and 1, and the closer it was to 1 the better (but should not be too close e.g. 0.99 because this meant that there was an independent variable that is almost the dependent variable). p-values smaller than 0.05 were accepted as a significant effect.
23 IV.RESULTS
4.1. Correlation, reliability and assumptions test
To check if there was a linear relationship between the dependent variable and the two independent variables, a scatterplot was made. There, it showed that linear relationships could be drawn (see Appendix B). To check if the residuals were normally distributed, PP plots were made for the dependent variable and two independent variables. The dots on the PP plots were all in a straight line (see Appendix C). Therefore, both assumptions were met.
Correlation tests and reliability tests were carried out for perceived personalization, PAR, PPI, and Aad. The results showed that all items measuring these constructs correlated significantly with one another (Pearson’s correlation was all positive, p-value < .05) and that sum variables could be computed (Cronbach’s α > .60) and were averaged.
4.2. Relationships between perceived personalization, PAR and PPI
Two linear regression analyses with perceived personalization regressed on PAR and PPI were performed, respectively. One model explained 12% of variance in PAR and the other explained 1.5% of variance in PPI. Both the regression analyses were significant (for PAR: β = 0.28, p < .001; for PPI: β = 0.11, p < .01). After this, the controlled variables were added to check if they confound the results but none of them showed any effects (for PAR: R2 = 0.16, β = 0.26, p < .001; for PPI: R2 = 0.03, β = 0.12, p < .01). These results indicated that perceived
personalization significantly and positively related to both PAR and PPI: the more consumers perceived the ad as personalized, the more they perceived it as relevant, but at the same time, they also perceived that their privacy was invaded. The relationships were only shown because they were the baseline assumptions used to propose the 3 hypotheses. The results will not further extend on these relationships, as this was not the main aim of the present study.
4.3. Direct relationships between PAR and PPI on Aad.
The model explained 54.6% of variance in Aad and the regression was significant. The result showed that PAR positively explained Aad (β = 0.63, p < .001) while PPI negatively explained Aad (β = -0.17, p < .001). Then, the controlled variables were added to this model and the result revealed that they did not alter the effect of PAR and PPI on Aad (R2 = 0.56; For PAR: β =
24 VIFs were below 2.0 (Allison, 2012). This result indicated that the more consumers perceived a personalized ad as relevant, the more favorable their attitude towards the ad was, and that the more consumers perceived a personalized ad as invasive to their privacy (i.e. higher perceived privacy invasion), the less favorable their attitude towards the ad was, hereby supporting both Hypothesis 1 and 2.
4.4. The moderating effect of PPI on the relationship between PAR and Aad
The result of the moderation test can be found in Table 1. The overall model explained 55% of variance in attitudes towards the ad. In this model, an interaction variable was automatically computed between the independent variable (PAR) and the moderator (PPI) (Aiken & West, 1991). In line with Hypothesis 1 and 2, the result showed that there was a main positive effect of PAR on Aad (β = 0.32, p < .01) and a main negative effect of PPI on Aad (β = -0.41, p < .001). There was also a positive significant interaction effect (β = 0.62, p < .01). After this, the controlled variables were added to the model as covariates but they did not cause any changes (for PAR: β = 0.33, p < .01; for PPI: β = -0.39, p < .001; for moderator PAR*PPI: β = 0.67, p < .01). This result indicated that PPI was indeed a moderator but it strengthened rather than weakened the positive effect of PAR on Aad. The conditional effects of the focal predictor (PAR) at different values of the moderator (PPI) confirmed this direction: at a score of approximately 3.5, 5 and 6 on a seven-point Likert scale on PPI, PAR showed significantly increased effects with β equaled 0.53, 0.62 and 0.69 (p < .001), respectively. Therefore, it can be concluded that there was a moderator effect, but in the opposite direction as hypothesized, hereby rejecting Hypothesis 3.
Table 1
Moderation effect of PPI on the relationship between PAR and Aad
25 PAR*PPI 0.062 3.64 .0003 0.023 ; 0.095 R2 = 0.0095
26 V.DISCUSSION AND CONTRIBUTIONS
In a widely used and highly personalized environment like social media (Karahanna, Xu & Zhang, 2015; Tran et al., 2020), marketers are striving to enhance consumer’s attitude towards the ad because it indirectly determines consumers’ e-loyalty (Srinivasan et al., 2002; Anderson & Srinivasan, 2003). One of the most popular methods is personalized ads as it offers higher “matching” rates with consumers and therefore improves ad effectiveness. (Jung, 2017; Li, 2016; Zhu & Chang, 2016). However, the downside of it is that it might bring about a perception of privacy invasion (Van Doorn & Hoekstra, 2013; Zhu & Chang, 2016; Baek & Morimoto, 2012), which can then mitigate ad effectiveness (Baek & Morimoto, 2012; Aguirre et al., 2015). The current study aimed to investigate when personalization is effective in generating a favorable attitude towards the ad, while also checking where the boundary condition is for this effectiveness.
The results of this study confirmed that perceived personalization explains both perceived ad relevance and perceived privacy invasion. Specifically, the more consumers perceive an ad to be personalized, the more they perceive it to be relevant and to be invasive of their privacy. This finding is in line with the assumed explanation of the double-edged effect of personalized ads: the personalization-privacy paradox (Aguirre et al., 2015; Zhu & Chang, 2016; Jung, 2017; Boerman et al., 2017). The 3 hypotheses were proposed based on this assumption for which this study has provided empirical evidence. However, the discussion will not elaborate on it because it is not the research aim.
The present study also revealed that perceived ad relevance is greatly important in developing a favorable attitude towards a personalized ad. The more relevant consumers perceived an ad, the more likely they were to have a positive attitude towards it. The privacy calculus theory (Laufer & Wolfe, 1977) suggests that consumers evaluate both potential benefits and costs of personalized ad (Boerman et al., 2017). In line with this, consumers consider the provision of relevant information a potential benefit because it saves them time and effort (Zhu & Chang, 2016). Therefore, they showed a positive attitude towards the ad. Moreover, this study confirms this relationship in a social media context, which still has not been looked at much (Tran et al., 2020).
27 information relating to consumers (Jung, 2017), making them perceive that their privacy has been invaded. Based on the privacy calculus theory (Laufer & Wolfe, 1977), consumers consider this perception a cost (Baek & Morimoto, 2012). Because people generally avoid costs (Zhu & Chang, 2016), when ending up with one, they expressed a less favorable attitude towards the attitude object, which is the personalized ad in this case. This finding is in line with previous studies showing the negative relationship of privacy invasion on ad’s effectiveness (Phelps, D'Souza & Nowak, 2001). Again, this study confirms this relationship in a social media context.
Finally, the results showed that perceived privacy invasion moderates the relationship between perceived ad relevance and attitude towards the ad. This outcome is different from past papers that showed a mediating role of privacy invasion (Zhu & Chang, 2016; Jung, 2017). However, in contrast with the 3rd hypothesis which proposed that perceived privacy invasion weakened this relationship, it actually strengthened it. Therefore, this finding challenges theoretical arguments used in theories such as the psychological reactance theory (Brehm, 1966), used to explain what consumers tend to do when their freedom is threatened by perceived privacy invasion (Boerman et al., 2017; Baek & Morimoto, 2012). However, the positive effect of perceived privacy invasion on the relevance-attitude relationship might be understood by the theory of self-awareness (Duval & Wicklund, 1972). According to this theory, people can shift focus on themselves, termed the object “self”, and thus make it stand out in their consciousness (Duval, Silvia & Lalwani, 2001). In a social media context where people have a high tendency to think that they own everything there (Karahanna, Xu & Zhang, 2015), this possibility that this process happens would even be higher compared to other contexts. When such attention is triggered, people are more likely to "accuse" themselves as the cause of both favorable and unfavorable events. People view perceived privacy invasion as an unfavorable event (Zhu & Chang, 2016), and, based on the awareness process, consumers might take more self-responsibility for it by arguments such as "I traded my privacy for this personalized
information" (Zhu & Chang, 2016) or "I let this happen in the first place". This process is an
28 The same logic can be applied to perceived ad relevance but in the opposite direction to explain how perceived ad relevance could alleviate the less favorable attitude towards the ad caused by perceived privacy invasion. Indeed, a positive moderating effect of perceived ad relevance was found, meaning that it weakened the negative effect of perceived privacy invasion on attitude towards the ad. Specifically, personalized ads positively related to perceived ad relevance (confirmed by the result of this study: the positive relationship between perceived personalization and perceived ad relevance), which positively leads to high self-awareness (Zhu & Chang, 2016). Accordingly, the increased perception of relevance made people relate to themselves more i.e. all attention was redirected to the object “self”. Therefore, from this moment onwards, anything that occurred was because of one “self”. In other words, the “self” was the immediately available cause on which the internal attribution process could focus (Duval & Wicklund, 1972). Following this argument, consumers probably thought that "I'm
glad I searched about this on Google and now it is showing useful information on my newsfeed"
and thus they ended up in a more positive general state, leading to an enhanced favorable attitude towards the ad.
Therefore, the results of this study challenge commonly used theories such as psychological reactance and the privacy calculus to explain the relationship between relevance and privacy invasion, as they suggest that consumers weigh the potential pros and cons of personalized ad (Boerman et al., 2017; Zhu & Chang, 2016). Rather, they tend to come up with justifications for the current situation and try to identify a possible cause that can neutralize negative consequences (Duval & Wicklund, 1972).
29 VI.LIMITATIONS AND SUGGESTIONS FOR FUTURE RESEARCH
A potential limitation is that the sample was based on convenience, resulting in an over-representation of females in their late 20s, who used social media often and earned a low income. Because of these characteristics, generalization of the results should be done with caution. For example, previous research has found that younger people are less likely to oppose personalization (Boerman et al., 2017). Moreover, online experience (illustrated by frequency of using social media) also influences perceived personalization (Lee et al., 2015; Miyazaki, 2008): the more people use social media, the less likely they will notice personalized ads. The results of this study stayed unchanged when controlling for these confounding variables; however, the conclusion remains provisional until the model is further validated using a more generalized sample.
Another potential limitation is that respondents were explicitly informed that the survey was about personalized ad from the beginning, which could bias their evaluation. This drawback was controlled by randomizing the order in which the items measuring the constructs appeared, thus the results remained valid. Nonetheless, in reality, consumers usually will not be notified that the ad they are seeing is personalized. Therefore, future research could check if consumers develop different attitudes towards the ad when they are not made aware of the personalization. For instance, not mentioning that the survey is about personalized ad and not providing any explanation about it would be a way to do it.
31 VII.CONCLUSION
32 REFERENCES
Aguirre, E., Mahr, D., Grewal, D., de Ruyter, K., et al. (2015). Unraveling The Personalization Paradox: The Effect of Information Collection and Trust-building Strategies on Online Advertisement Effectiveness. Journal of Retailing, 91(1), 34-49.
Aiken, L. S., & West, S. G. (1991). Multiple Regression: Testing and Interpreting Interactions. Thousand Oaks, California: Sage.
Allison, P. (2012). When Can You Safely Ignore Multicollinearity? Available at https://statisticalhorizons.com/multicollinearity (accessed on 09-Dec-2020)
Anderson, R. & Srinivasan, S. (2003). E-satisfaction and E-loyalty: A Contingency Framework. Psychology & Marketing, Vol. 20 No. 2, pp. 123-138.
Baek, T. H., & Morimoto, M. (2012). Stay Away from Me. Journal of Advertising, 41(1), 59-76.
Boerman, S. C., Kruikemeier, S., & Zuiderveen Borgesius, F. J. (2017). Online Behavioral Advertising: A Conceptual Framework and Research Agenda. Journal of Advertising, 46(3), 363-376.
Brehm, Jack W. (1966). A Theory of Psychological Reactance. New York: Academic Press.
Burton, Scot & Donald R. Lichtenstein (1988). The Effect of Ad Claims and Ad Context on Attitude Toward the Advertisement. Journal of Advertising, 17 (1), 3-11.
Campbell, D. E., & Wright, R. T. (2008). Shut-up I Don't Care: Understanding The Role of Relevance and Interactivity on Customer Attitudes toward Repetitive Online
Advertising. Journal of Electronic Commerce Research, 9(1), 62-76.
33 Cohen, Jacob, Patricia Cohen, Stephen G. West, and Aiken, Leona S. (2003), Applied Multiple
Regression/Correlation Analysis for the Behavioral Sciences (3rd edition), Mahwah, NJ:
Lawrence Earlbaum Associates.
Debevec, Kathleen, & Easwar Iyer (1988). Self-Referencing as a Mediator of the Effectiveness of Sex-Role Portrayals in Advertising. Psychology & Marketing, 5(1), 71-84.
Duval, S., & Wicklund, R. A. (1972). A Theory of Objective Self-awareness. Oxford, England:
Academic Press.
Duval, T. S., Silvia, P. J., & Lalwani, N. (2001). Self-awareness & Causal Attribution: A Dual Systems Theory. Boston: Kluwer Academic Publishers.
Gelb, Betsy D. & Pickett, Charles D. (1983). Attitude-Toward-The-Ad: Links to Humor and to Advertising Effectiveness. Journal of Advertising, 12 (2), 34-42.
Girondaa, John T. & Korgaonkarb, Pradeep K. (2018). iSpy? Tailored versus Invasive Ads and Consumers’ Perceptions of Personalized Advertising. Electronic Commerce Research and
Applications, 29 (2018), 64-77
Hastak, M. & Olson, Jerry C. (1989). Assessing the Role of Brand-Related Cognitive Responses as Mediators of Communication Effects on Cognitive Structure. Journal of
Consumer Research, 15 (March), 444-56.
Hill, Ronald P. & Michael B. Mazis (1986). Measuring Emotional Responses to Advertising.
Advances in Consumer Research, 13, Richard J. Lutz ed., Provo, UT: Association for
Consumer Research, 164-169.
Hilton, C. E. (2017). The Importance of Pretesting Questionnaires: A Field Research Example of Cognitive Pretesting the Exercise Referral Quality of Life Scale (ER-QLS). International
Journal of Social Research Methodology, 20:1, 21-34, DOI: 10.1080/13645579.2015.1091640.
34 Jung, A. (2017). The Influence of Perceived Ad Relevance on Social Media Advertising: An Empirical Examination of A Mediating Role of Privacy Concern. Computer Human Behavior, 70, 303-309.
Karahanna E., Xu X. S., & Zhang, N. A. (2015). Psychological Ownership Motivation and Use of Social Media. Journal of Marketing Theory and Practice, vol. 23, no. 2 (Spring 2015), pp. 185–207.
Keller, K. L. (1993). Conceptualizing, Measuring, and Managing Customer-based Brand Equity. Journal of Marketing, 1-22.
Kelly, L., Kerr, G., & Drennan J. (2010). Avoidance of Advertising in Social Networking Sites: The Teenage Perspective. Journal of Interactive Advertising, 10 (2), 16–27.
Laczniak, R. N., & Muehling, D. D. (1993). Toward a Better Understanding of The Role of Advertising Message Involvement in Ad Processing. Psychology and Marketing, 10(4), 301e319.
Lai, Ying-Hsiao (Rebecca), Chen, Chun-Chu, & Petrick, James F (2016). The Economic Impact on Leisure Activities. Travel and Tourism Research Association: Advancing Tourism
Research Globally. 18. https://scholarworks.umass.edu/ttra/2010/Visual/18
Laufer, Robert S., & Maxine Wolfe (1977). Privacy as a Concept and a Social Issue: A Multidimensional Development Theory. Journal of Social Issues, 33(3), 22-42.
Lee, Seungsin, Younghee Lee, Joing In Lee, & Jungkun Park (2015). Personalized E-Services: Consumer Privacy Concerns and Information Sharing. Social Behavior and Personality: An
International Journal, 43(5), 729-740.
35 Machleit, Karen A. & Kent, Robert J. (1989). What is the Effect of Attitude Toward the Ad when the Consumer is Familiar with the Brand. AMA Educators' Proceedings, American
Marketing Association, 215-219.
Machleit, Karen A., Madden Thomas J., & Allen Chris T. (1990). Measuring and Modeling Brand Interest as an Alternative Aad Effect with Familiar Brands. Advances in Consumer
Research, 17, Marvin E. Goldberg, Gerald Gorn, and Richard W. Pollay, eds., Provo, UT:
Association of Consumer Research, 223-230.
Malhotra, Naresh K. (2010). Marketing Research: An Applied Orientation, 6th ed. Prentice
Hall, Upper Saddle River, NJ.
Marchand, J. (2010). Attitude Toward the Ad: Its Influence in a Social Marketing Context.
Social Marketing Quarterly, 16:2, 104-126.
Mitchell, Andrew A., & J.C. Olson (1981). Are Product Attribute Beliefs the only Mediator of Advertising Effects on Brand Attitude? Journal of Marketing Research, 1 (August), 318-32.
Miyazaki, Anthony D. (2008). Online Privacy and the Disclosure of Cookie Use: Effects on Consumer Trust and Anticipated Patronage. Journal of Public Policy and Marketing, 27(2), 19-33.
Moore, Danny L. & Hutchinson, J. Wesley (1983). The Effects of Ad Affect on Advertising Effectiveness. Advances in Consumer Research, 10, R.P Bagozzi and A. M. Tybout, eds, Ann Arbor, MI: Association for Consumer Research, 526-31.
Muehling, Darrel D. & McCann, M. (1993). Attitude toward the Ad: A review. Journal of
Current Issues and Research in Advertising. Volume 15, Number 2 (Fall 1993).
Nunnally, J. C. & Bernstein, I. H. (1978). Psychometric Theory 3rd Edition. New York:
McGraw-Hill, 1994.
36 Pierce, Jon L.,Tatiana Kostova, & Kurt T. Dirks (2001). Toward a Theory of Psychological Ownership in Organizations. Academy of Management Review, 26(April), 298–310.
Phelps, J., D'Souza, G., & Nowak, G. (2001). Antecedents and Consequences of Consumer Privacy Concerns: An Empirical Investigation. Journal of Interactive Marketing, 15(4), 2-17.
Phelps, J. & Thorson, E. (1991). Brand Familiarity and Product Involvement Effects on the Attitude Toward an Ad-Brand Attitude Relationship. Advances in Consumer Research, 18, Rebecca H. Holman and Michael R. Solomon, eds., Provo, UT: Association of Consumer Research, 202-209.
Rimer, Barbara K., & Kreuter, Matthew W. (2006). Advancing Tailored Health Communication: A Persuasion and Message Effects Perspective. Journal of Communication, 56 (1), 184–201.
Shimp, Terrance A. (1981). Attitude Toward the Ad as a Mediator of Consumer Brand Choice.
Journal of Advertising, 1981, Vol. 10, No. 2 (1981), pp. 9-15+48.
Shimp, Terence A. & Yokum, J. Thomas (1981). The Influence of Deceptive Advertising on Repeat Purchase Behavior. Proceedings: AMA Educators' Conference, Kenneth Bernhardt and Thomas Kinnear, eds. Chicago: American Marketing Association, 266-270.
Shiv, B. & Fedorikhin, A. (1999). Heart and Mind in Conflict: The Interplay of Affect and Cognition in Consumer Decision Making. Journal of Consumer Research, 26 (3), 278–92.
Soper, Daniel S. (2020). A-Priori Sample Size Calculator for Multiple Regression [Software]. Retrieved from https://www.danielsoper.com/statcalc (accessed on 12-Nov-2020).
Srinivasan, S., Anderson, R. & Ponnavolu, K. (2002). Customer Loyalty in E-commerce: An Exploration of Its Antecedents and Consequences. Journal of Retailing, Vol. 78 No. 1, pp. 41-50.
Statista (2020). Daily time spent on social networking by internet users worldwide from 2012
37 Statista (2020). Number of global social network users 2017-2025. Retrieved from https://www.statista.com/statistics/278414/number-of-worldwide-social-network-users/ (accessed on 10-Sep-2020)
Statista (2020). Social Media Advertising. Retrieved from https://www.statista.com/outlook/220/100/social-media-advertising/worldwide (accessed on 10-Sep-2020)
Sutanto, J., Palme, E., Tan, C.H. & Phang, C. W (2013). Addressing the Personalization-Privacy Paradox: An Empirical Assessment from A Field Experiment On Smartphone Users.
MIS Quarterly, Vol. 37 No. 4, pp. 1141 -1164/December 2013.
Tam, K.Y. & Ho, S.Y. (2005) Web Personalization as a Persuasion Strategy: An Elaboration Likelihood Model Perspective. Information Systems Research 16(3):271-291. https://doi.org/10.1287/isre.1050.0058
Tam, K. Y. & Ho, S. Y. (2006). Understanding the Impact of Web Personalization on User Information Processing and Decision Outcomes. MIS Quarterly, 30 (4), 865–90.
Taylor, David G., Jeffrey E. Lewin, & Strutton, D. (2011). Friends, Fans, and Followers: Do Ads Work on Social Networks? Journal of Advertising Research, 51 (1), 258–75.
Tolchinsky, P. D., McCuddy, M. K., Adams, J., Ganster, D. C., Woodman, R. W., & Fromkin, H. L. (1981). Employee Perceptions of Invasion of Privacy: A Field Simulation Experiment.
Journal of Applied Psychology, 66(3), 308.
Tran, Trang P. , Chien-Wei Lin, Sally Baalbaki, & Francisco Guzmán (2020). How Personalized Advertising Affects Equity of Brands Advertised on Facebook? A Mediation Mechanism. Journal of Business Research, 120, 1-15.
Tuten, T. L., & Solomon, M. R. (2015). Social Media Marketing. 2nd Edition, Sage.
38 Waites S.F. & Ponder N. (2016). May I Have Your Attention Please? The Effectiveness of Attention Checks in Validity Assessment. In: Kim K. (eds) Celebrating America’s Pastimes: Baseball, Hot Dogs, Apple Pie and Marketing?. Developments in Marketing Science:
Proceedings of the Academy of Marketing Science. Springer, Cham.
https://doi.org/10.1007/978-3-319-26647-3_97
White, Tiffany B., Debra L. Zahay, Helge Thorbjørnsen & Sharon Shavitt (2008). Getting Too Personal: Reactance to Highly Personalized Email Solicitations. Marketing Letters, 19(1), 39-50
Whiting, A. & Williams, D. (2013). Why People Use Social Media: A Uses and Gratifications Approach. Qualitative Market Research: An International Journal, Vol. 16 No. 4, 2013, 362-369
Zhu, Y.-Q., & Chang, J.-H. (2016). The Key Role of Relevance in Personalized Advertisement: Examining Its Impact on Perceptions of Privacy Invasion, Self-awareness, and Continuous Use
Intentions. Computers in Human Behavior, 65, 442–447.
https://doi.org/10.1016/j.chb.2016.08.048
39 APPENDICES
Appendix A: The scenario in the survey.
Assume that a respondent filled in a name of Mr. Queen, a choice of destination of
Switzerland and a choice of social media of Facebook. The scenario would appear as follows:
It has been more than a year since the COVID-19 appeared, causing travel bans to be implemented almost everywhere in the world. As someone who likes to travel for leisure, you
can't wait to do it again as soon as the COVID-19 is over. Your destination of interest is Switzerland and thus you have been searching for trips to the place on Google. Later in the day, when you are using your Facebook, you see the following advertisement about trips to
Switzerland with special price offerings:
Hi Mr. Queen, been dreaming about visiting Switzerland. Dream no more as we’ve got what
40 Appendix B: Scatterplot showing a linear relationship between the dependent variable
41 Appendix C: PP-plots
Attitude towards the ad
43 Appendix D: An example of how to manipulate consumers’ understanding of why the
personalized ad appears in their social media.
Assume that a consumer name is Princess, who uses Instagram the most and has been wanting to visit South Korea. The personalized ad would appear as follows:
*Note: this ad is suggested to you based on your previous search on Google
Hi Princess, been dreaming about visiting South Korea. Dream no more as we’ve got what you need! Travel itinerary designed exclusively for you. Contact us now to get up to 50%
