GIVING A MONETARY REWARD AS A METHOD FOR ACQUIRING CONSUMERS’ PERSONAL INFORMATION
Master Thesis, Msc Marketing Intelligence & Management,
University of Groningen, Faculty of Economics and Business.
June 26, 2017
BRAM SCHUNSELAAR Studentnummer: 2298694 Nieuwe Ebbingestraat 84a
2 Abstract
Firms try to gather as much personal information as possible by giving a reward in return for the information. However is this a good way to achieve this? This study focused on the potential ‘U’ shape moderation effect of monetary reward on the relationship between information sensitivity and privacy concern and the effect of privacy concern on the willingness to provide the information and misrepresentation of the information. An
experimental set-up with a survey was used to collect data and 146 responses were analysed. The process macro of Hayes was used to analyse a moderated mediation model. In addition a binominal logistic regression model was used to analyse actual information sharing
behaviour. The results showed that the more sensitive the information is the more privacy concern people become and resulting in people which are less willing to share and more inclined to misrepresent the information. Furthermore monetary reward did not moderate the relationship between information sensitivity and privacy concern. Privacy concern partially mediated the relationship between information sensitivity and willingness to share
information and fully mediated the relationship between information sensitivity and misrepresentation. Lastly the additional analysis showed that monetary reward did actual influenced consumers’ actual sharing behaviour. This is an interesting results, because monetary reward is not a moderator so does not influences consumers’ feeling of privacy but it does changes consumers’ actual behaviour. So giving a monetary reward can still be used as a good way to gather personal information.
3
Table of Content
MONETARY REWARD AS A METHOD FOR ACQUIRING CONSUMERS’ PERSONAL
INFORMATION ... 1
1. INTRODUCTION ... 4
2. THEORETICAL FRAMEWORK AND RESEARCH HYPOTHESES ... 7
2.1 Information sensitivity ... 9
2.2Effect of monetary rewards ... 10
2.3 Privacy concern and the consumers actions ... 12
2.4 Control variables ... 14 3 RESEARCH DESIGN ... 15 3.1Research design ... 15 3.2 Sample ... 15 3.2Procedure ... 16 3.3 Manipulations ... 17 3.4 Measurements ... 18 3.5 Data Analysis ... 19
4 DATA ANALYSIS AND RESULTS ... 22
4.1 Descriptive Statistics ... 22
4.2 Control variable checks ... 23
4.3 Information sensitivity measurements and manipulation checks ... 25
4.3.1 Information sensitivity measurements ... 25
4.3.2 Manipulation checks ... 26
4.4 Reliability of the Measurements ... 27
4.5 Hypothesis testing ... 28
4.6 Additional analysis ... 32
5 DISCUSSION ... 34
5.1 Implications of results ... 34
5.2 Theoretical and practical implications ... 36
5.3 Limitations and future research ... 37
REFERENCES ... 39
4
1. INTRODUCTION
Nowadays companies are focussed on collecting information about consumers. Companies are trying to collect as much personal information as possible, because it will help them in identifying who their consumers are. Accurate consumer information is therefore becoming more and more important asset for those companies. Accurate consumer information is needed for companies to be able to perform marketing activities and deliver the right goods and services (Han et al. 2007). The importance of consumer personal information is for example shown by McKinsey Global Industry, which states that the industry where personal information is being bought and sold is a $300-billion-a-year industry in the United States alone (Morris and Lavandera 2012). In addition companies today are trying to acquire the consumer information themselves, however some important aspects in the process of acquiring the consumer information have to be dealt with.
According to Hoffman, Novak and Peralta (1999) and Hallam and Zanella (2017) the willingness of consumers to share personal data online is the most important aspect. However, a consumers willingness to share personal data is based on an risk and benefit trade-off. A consumer needs to think about what the potential risks and benefits are and decide if the benefits outweigh the risks when they are asked to provide personal information. The risk part relates to privacy concern which makes consumers reluctant to share their personal
information online. Several surveys stated that throughout the years people’s privacy concern has increased and nowadays around 55% of the people are concerned about their privacy (ECP 2014; Turow, Hennessy and Draper 2015; Goldfarb and Tucker 2012). So what are the reasons why people are still privacy concern? According to Kobsa, Patil and Meyer (2011) and Steward and Segars (2002) privacy concern about the collection of personal information can come from several reasons such as; improper access, misuse of the data and the
possibility of the data being sold to a third party. When consumers perceive too much risk they may be inclined to misrepresent the information that they give or abandon the online activity (Milne et al. 2004).
5 Rewards come in different forms. Rewards can be monetary, non-monetary or other benefits which the consumer might value. The focus of this research is on rewards, because from all the potential ways to diminish privacy concern and acquire consumer personal information the use of rewards is the easiest to implement and most used method by companies and the use of this method becomes even more widespread (Gabish and Milne 2014). In addition e-mail marketing research has already looked at the effect of rewards and showed that monetary rewards works better to increasing response rates in comparison to non-monetary rewards (Baruch and Holtom 2008), but research on acquiring consumer personal information with the help of rewards is not as extensive and consistent in their results as in the e-mail marketing research. Xie, Teo and Wan (2006) argued that this inconsistency can come from the fact the effects of rewards might differ in different context for example different types and sensitivity of the information. Therefore it is interesting to see if monetary rewards will also have a positive effect for companies in acquiring sensitive consumer personal
information? How good is the quality of the information that those consumers give? And what is the best level of reward that a company should give in return for the information?
Prior researchers have started to make an important contribution in trying to
6 someone and this could harm people if the information is misused Lwin, Writz, and Williams (2007).So it is important to focus on information which is different based on their level of sensitivity, because consumers’ response will be different in such a situation. They might be more privacy concerned and less willing to share this information due to the sensitivity Xie, Teo and Wan (2006). Furthermore these types of information are important, because
companies can use financial and purchase information to identify and attract new consumers and to retain existing consumers and get a better insight on what consumers actually do. In addition with this information companies can also develop new ways to do up and cross selling and increase the customer life time value (Ngai, Xui and Chau 2009). Secondly, many of the studies focused on Asian consumers. According to these studies the Asian consumers scores high on privacy concern due to a lot of government regulations, therefore these results may not be very generalizable for other consumers like the European consumer. One
exception is Premazzi et al. (2010) who looks at consumers from the United States and
showed that consumers from the United States where less privacy concern than the Asian and that rewards decreased the concern. Therefore this study adds to the further understanding of how rewards might influence privacy concern in the western world. Lastly, the studies did not differ in the amount of their rewards, they only tested different rewards types. Andrade, Kaltcheva and Weitz (2002), Premazzi et al. (2010) and Xie, Teo and Wan (2006) tested the effects of monetary and non-monetary rewards versus no reward situation and they used only one amount of reward respectively a $10, $20 and $20 gift card. Therefore these studies did not test the effects of different amounts of monetary rewards. However, this is important especially because researchers expect a non-linear relationship between compensation and privacy concern, due to the possibility that a high amount of monetary reward might signal that a company really wants that data, which might make the consumer more concern about the purpose of the company to acquire that data (Lee et al. 2015). The studies suggested that further research should focus on these limitations.
The goal of this research paper is to examine the relationship of how different monetary reward levels might influence the willingness to share sensitive consumer information. Specifically it addresses the following research questions: Does the level of sensitivity of the consumer information effects the consumer’s privacy concern? How is this relationship influenced by the amount of a monetary-reward? Does privacy concern influence the willingness to share and the misrepresentation of the consumer information?
7 calculus theories (Andrade, Kaltcheva and Weitz 2002; Laufer and Wolfe 1977). Privacy calculus theory tells that people weigh their potential benefits and risks and that the outcome of the comparison of the risks and benefits will lead to a certain level of privacy concern (Andrade, Kaltcheva and Weitz 2002). The social exchange theory says that someone’s behaviour of self-disclosure can be described by looking at someone’s perspective on the potential rewards and costs (Laufer and Wolfe 1977). This research will focus more specifically on the benefit / reward side of these theories and how different amounts of the reward might help in getting consumer information which differs on the level of sensitivity. In addition this study will consider different amount of monetary rewards to further investigate the potential non-linear relationship due to the signalling effect. Furthermore, this study investigates the European consumer. In addition, prior research mainly focused on the willingness of the consumer in sharing the personal information, however the current study investigates the fact if people are also more honest in their shared information, so it combines both protection and sharing reaction of consumers. In practical perspective, the result of this paper will give managers a better understanding of how to acquire sensitive information and which amount of reward might work best in acquiring certain sensitive consumer information. This will help them in determining the right amount of reward and to not scare away the potential consumer when asking for sensitive consumer information.
The rest of the research paper is organized as follow. In the second part prior literature is reviewed and discussed and a theoretical model and hypotheses are constructed. In the third part, the process of data collection and the methods used to analyse the data for this research are discussed. In the fourth part, the results of the paper are shown and explained. The research paper concludes with the implications and limitations of the research and further directions for new research are given.
2. THEORETICAL FRAMEWORK AND RESEARCH HYPOTHESES
8 information. The cost are related to privacy concern which can occur when consumers might think that the company will not use their information properly which could harm them (Lee et al. 2015). Benefits can range from giving a compensation for the information or more
personalised offers and services (Chellappa and Sin 2005; Premazzi et al. 2010). In the end consumers weight the cost and benefits and decide if they want to share their personal information.
Two important theories that underlie the cost and benefit trade-off are the social exchange theory (Kelley and Thibaut 1978) and the privacy calculus theory (Laufer and Wolfe 1977). The social exchange theory and the privacy calculus theory are very similar to each other. Both theories look at privacy concern and try to understand a consumers’
information providing and disclosure behaviour, in addition the social exchange theory includes people’s perspective on privacy concern it involves their perspective on the rewards and risks related to privacy concern (Lee et al. 2015). The theories argue that before a
consumer is willing to disclose their personal information, he or she will first assess the potential benefits and risks of disclosing the personal information. According to Taylor, Davis and Jillapalli (2009) and Xie, Teo and Wan (2006) the assessment of the benefits and risks can take place both consciously and unconsciously, for example small privacy signs on a website can be unconsciously processed and work as a benefit. Benefits could be monetary, non-monetary or intangible benefits such as reward points or special services (Chellappa and Sin 2005; Premazzi et al. 2010). The potential risks consists of loss of control of information, possibility of misuse of the information or the information being sold to a third party (Kobsa, Patil and Meyer 2011). For a consumer to be willing to give away personal information the benefits need to be higher than the risks (Chellappa and Sin 2005). The consumer wants to get some economic or social value in return for the loss of control of their personal information (Culnan and Armstrong 1999). In the end, consumers will compare the potential benefits and risks and will determine whether or not to disclose their personal information (Gabisch and Milne 2014; Lee et al. 2015; Schumann et al. 2014).
9 FIGURE 1
Conceptual model
2.1 Information sensitivity
An important aspect in the transaction of personal information is the sensitivity of the information given by consumers to potential collectors. Sensitivity of the information is related to intimacy and risk of the information and is an important antecedent of privacy concern. Lwin, Writz, and Williams (2007) argued that consumers perceive information as more sensitive if the information is more intimate to them. Information is more intimate when it could harm them when someone else has that information. When it is more intimate to them consumers perceive more risk in disclosing the information because of potential losses. Potential losses that are related to the sensitivity of the information include psychological (harm to self-esteem due to embarrassment), physical (relates to life and health) and or material losses (related to financial harm or harm to other assets) (Moon 2000). Thus, information sensitivity is defined as the level of concern an individual perceives when asked to disclose a certain type of information (Castañeda and Montoro 2007; Gabisch and Milne 2014; Mothersbaugh et al. 2012).
In addition, researchers found that different types of personal information have
10 consumers see demographic or lifestyle information as the least sensitive information, more sensitive information is personal identifiable and most sensitive information is financial information. Several studies have shown that consumers are more concerned when asked for more sensitive information (Lee et al. 2015; Phelps, Nowak and Ferrell 2000). In addition, Malhotra, Kim and Agarwal (2004) showed that consumers where more concerned when asked to provide sensitive information compared to less sensitive information. Due to the potential risk which makes the consumer more cautious about the use of the information by the collector. These results are also consistent with the study of Lee et al. (2015) which showed that when consumers are asked to provide more sensitive information their privacy concern also increases. Therefore the following hypotheses is constructed.
Hypothesis 1. The level of information sensitivity is positively related to privacy concern.
2.2 Effect of monetary rewards
The use of monetary-reward is studied in several domains such as e-mail marketing, sales situations and behavioural studies. For example, studies have already shown that the use of monetary-reward can increase response rates or reduce dropouts for survey, because it gives people a motivation to respond (Göritz 2006; O’Neil and Penrod 2001). Giving a reward is also an often used method by companies to acquire consumer personal information and reduce privacy concern (Taylor, Davis and Jillapalli 2009). According to Lee et al. (2015) and
Taylor, Davis and Jillapalli (2009) there are two different forms of rewards: monetary rewards and non-monetary rewards. Monetary rewards includes cash, discounts, coupons, prizes or other forms of physical gifts and non-monetary rewards includes intangible benefits like convenience, personalization or access to exclusive content (Gabisch and Milne 2014; Lee et al. 2015). In this study the focus is on monetary rewards, which is defined as a currency or currency-equivalent compensation for the provision of personal information (Taylor, Davis and Jillapalli 2009), because this the most used sort of reward by companies in practice and even more and more companies make use of monetary rewards. So therefore it is interesting to see whether the method used by companies actually works.
11 signal that the company is willing to give something in return, which reduces the privacy concern, because a consumer will feel that there is an equal exchange when they are asked to provide information. Furthermore Xie, Teo, and Wan (2006) and Sheehan and Hoy (2000) showed that giving a reward will lead to less privacy concerned consumers. A reward encourages consumers to share their information and with less concern, because consumers see a reward as a fair and honest way of compensating for the exchange of personal
information. Therefore the following relationship is expected.
Hypothesis 2. An increase in monetary reward has a direct negative relationship with consumers’ privacy concern.
Furthermore, according to the privacy calculus theory one might expect that a monetary reward can help a company in acquiring consumers’ personal information and make them less privacy concerned especially in relationship with information sensitivity. The theory suggest that when more sensitive information is asked for consumer might feel more risk and will be more concerned. However, a monetary reward can counterbalance this, because it will give the consumer a benefit in return (Hui et al. 2007; Sheehan and Hoy 2000).
However, several studies have shown that monetary reward might not be a good way to reduce privacy concern especially when asked for sensitive information. According to Andrade, Kaltcheva and Weitz (2002) and Faja (2005) in the context of high sensitive information monetary rewards increase privacy concern, because the monetary rewards can make consumer worry about the motives and reasons why the collecting company offers a monetary reward for their personal information. Additionally, it can remind the consumer of prior negative experiences and increase the privacy concern (Lee et al. 2015). Furthermore, Yang and Wang (2009) argue that even within a certain level of information sensitivity different amounts of monetary rewards will have different effects on privacy concern. The differences occur due to potential signalling effect of the levels of the monetary rewards and the perspective of the people of what they think is a reasonable level of reward (Yang and Wang 2009). The authors tried to find an linear interaction effect between information
12 Thus, A simple linear moderation effect between monetary reward and information sensitivity on privacy concern is not expected, but a ‘U’ shape effect.
Hypothesis 3. The relationship between information sensitivity and privacy concern is
moderated by monetary reward in such a way that an increase from low to medium amount of monetary reward decrease privacy concern.
Hypothesis 4. The relationship between information sensitivity and privacy concern is moderated by monetary reward in such a way that an increase from medium to high amount of monetary reward increase privacy concern.
2.3 Privacy concern and the consumers actions
Privacy is a topic which is addressed in different research domains. The oldest and most simple definition comes from Warren and Brandeis (1890) who define privacy as ‘the right to be left alone’. However privacy is not just a simple concept, privacy has various different characteristics (Castañeda and Montoro 2007). Several research studies have classified four different types of privacy: informational privacy, social privacy, physical privacy and psychological privacy (Buchanan et al. 2007; Burgoon et al. 1989; Laufer and Wolfe 1977).
In the online context, which includes e-commerce, websites who sell products online, privacy ‘the right to be left alone’ primarily focusses on information privacy (Taylor, Davis, and Jillapalli 2009). Information privacy entails the consumers’ ability to control which personal information about one’s self is disclosed and how the personal information is used by the collector (Smith and Shao 2007; Stone et al. 1983). “Privacy concern exist when an individual feels threatened by perceived unfair loss of control over their personal information by an information-collecting body” (Lee et al. 2015, p. 46). According to Benamati, Ozdemir and Smith (2016) and Smith, Milberg and Burke (1996) privacy concern can come from concern about four specific internal elements: collection, which is concern about amount of personally information that is collected and stored; errors, concern that there is not enough protection to avoid errors; secondary use, concern that the information is used for other purposes than given permission for; and improper access, concern that information is
13 of the collection and use of the personal information acquired in an online transactions (Lee et al. 2015; Mothersbaugh et al. 2012; Taylor, Davis and Jillapalli 2009).
Prior research showed that privacy concern will have an influence on consumers behavioural responses (Castañeda and Montoro 2007). Behavioural responses can be grouped into different types. According to Son and Kim (2008) there are three main response
consumers might use when they face privacy concerns: information provision, which includes refusal of providing information and misrepresentation of the information; private actions, which entails removal of the information and negative word-of-mouth; and public actions, complain directly to the company and or complaining to third-party organizations about privacy issues. In this study, the focus is on the information provision responses, because according to Milne and Boza (1999) these responses are the first and most common responses that consumers have. The willingness to provide information can be defined as a consumers’ willingness to share personal information in an online setting (Mothersbaugh et al. 2012). Misrepresentation of the information is defined as consumers giving completely or partially false personal information when asked to provide their personal information(Son and Kim 2008).
Research on the influence of privacy concern and the provision responses is still inconclusive. Van Slyke et al. (2006) and Mothersbaugh (2012) do not find a negative relationship between privacy concern and the willingness to share information nor did they find a positive relation between privacy concern and misrepresentation of the information. However Benamati, Ozdemir, and Smith (2016), Son and Kim (2008) and Walrave and Heirman (2013) found an strong negative relationship between privacy concern and the willingness to share information. They argued that due to an increase in privacy concern people perceive more risk and are potentially less willing to share their information to an online company. In addition, Wirtz, Lwin, and Williams (2007) argued that there is a positive relationship between privacy concern and misrepresentation of the information and showed that around 50% of the consumers misrepresent their information. The reason for the
14 have a negative effect on information provision responses, because only in a very specific situation there is no significant relationship, but when you specifically look at privacy concern only there is a significant negative relationship.
Hypothesis 5. The level of privacy concern is negatively related to willingness to share personal information.
Hypothesis 6. The level of privacy concern is positively related to misrepresentation of personal information.
2.4 Control variables
The study includes six control variables. Research suggest several variables which also influences privacy concern and behavioural actions (willingness to share data and
misrepresentation of the information). The first control variable is trust. In this context trust means the general belief in the online company that the company will use and safeguard the personal information in the right way. According to Taylor, Davis and Jillapalli (2009) people who trust an online company more will be less privacy concern. The second variable is prior privacy invasion. Consumers who have had a bad experience in the past regarding privacy issues will feel more risk in the future when they are asked to provide personal information, therefore they will be more privacy concern and less willing to share personal information (Zhao, Lu and Gupta 2012). The next four variables concerns consumer characteristics; age, gender, education and income. Age will influence privacy concern in a way that older consumers are more privacy concerned (Benamati, Ozdemir and Smith 2016; Walrave and Heirman 2013). Women are more privacy concerned than men (Benamanti, Ozdemir and Smith 2016) and according to Phelps, Nowak and Ferrell (2000) an increase in education level and income will also increase privacy concern and behavioural actions.
15
3 RESEARCH DESIGN
3.1 Research design
In this study the effects of different levels of sensitive consumer information on privacy concern and the moderating effect of different amounts of monetary reward on this
relationship are investigated. Furthermore the effect of privacy concern on the willingness to share the consumer information and on misrepresentation of the consumer information are also examined. To test these relationships a web based experimental survey was created. The first part is a 2 (information sensitivity: low versus high) x 3 (amount of compensation: low, medium and high) between subject full factorial design experiment to test the effect of information sensitivity on privacy concern and the moderating effect of compensation. The second part is a structured survey to get the data needed to test the effect of privacy concern on the willingness to share the information and on misrepresentation of the information.
A factorial experiment is used, because it is a good method to manipulate the different variables and helps uncover causality, because you can influence the IV and measure the differences in DV. With this method I had full control over which variables were included in the manipulation, which made it able to fully focus on information sensitivity and amount of compensation and reduce the possibility of contamination by extraneous variables (Malhotra 2010)(Cooper and Schindler 2013). Furthermore this method made it also possible to
manipulate multiple variables at once within a manipulation (Cooper and Schindler 2013). In addition the web based survey part is used to gather the data, because this method gives the respondents enough time to think about their answer, therefore there is no time pressure and this method also diminishes potential interviewer biases. Furthermore it better ensures anonymity of the respondents (Malhotra 2010)(Cooper and Schindler 2013).
3.2 Sample
16 2010)(Cooper and Schindler 2013). Furthermore these methods made it possible for me to collect a lot of respondents in a short period of time, because it made me use my own social network and the social network of other. This is important, due to the limited time for
collecting data. However a disadvantage of these methods is that the results are not by default statistically generalizable for the whole population (Malhotra 2010). This might not be such a big problem, because we are mostly interested in people who participate in an e-commerce setting and the people in the sample have been in such a situation. The total sample consists of 146 respondents of which 94 people are male (64,4%) and 52 are female (35,6%).
3.2 Procedure
The experimental survey was distributed via internet and made with a program called qualtrics this program was very useful, because it also collected the data in an easy to
implement format for SPSS. When the participants started the survey they first had to read an introduction about the purpose of this study and some information about confidentiality. Then the respondents read a small scenario about a purchase situation online and a website was shown. On this website they sold electronic devices and mostly laptops and the respondents had to image that they were searching for a new laptop and finally found this website which was the best option for them. This set-up was created, because it is a common situation because many people have bought something online and personal information is often asked in such a situation. The scenarios where created by taking a screenshot from an online
webshop from which the brand name was deleted. These screenshots where used to make the purchase situation as realistic as possible. The next page they saw was one of the six
manipulations which were assigned randomly to the respondents. In the manipulation the same website was shown but now with a pop-up message which contained a customer survey. In the survey the variables information sensitivity and monetary reward were manipulated. After they saw the manipulated website they were asked to answer questions about their privacy concern, willingness to share the information and misrepresentation of the
information and manipulation checks were included. In addition to make the manipulations even more realistic the respondents had the option to actually provide their personal
17 Finally, the respondents were asked to answer a couple of questions about their
demographics such as age and gender and questions about the control variables income, education, trust and prior privacy invasion were asked.
3.3 Manipulations
The variable monetary reward was manipulated at three different levels, a low amount of monetary reward (5% discount), medium amount of monetary reward (30% discount) and a high level of monetary reward (70% discount) based on prior research by Yang and Wang (2009) and a pilot study. The following statement was used to manipulate the variable:
“Therefore we would like to ask you a few questions and in return you will receive a 5%, 30% or 70% discount for the next purchase.” The voucher statement is used, because prior research has already used this way of manipulating the monetary reward in their scenario’s and these studies showed that it is a good way to manipulate the monetary reward variable (Andrade, Kaltcheva and Weitz 2002; Xie, Teo and Wan 2006).
The second variables which was manipulated is the level of information sensitivity. In this research information sensitivity was manipulated at two different levels, low level of information sensitivity and high level of information sensitivity. The different levels of information sensitivity were based on prior research. Prior research had determined which sorts of information are more and which sorts of information are less sensitive. According to Mothersbaugh et al. (2012), Phelps, Nowak and Ferrell (2000) and Xie, Teo and Wan (2006) personal identifiable information is seen as more sensitive than non-personal identifiable information. The personal information asked in the scenarios is based on the categorization of Mothersbaugh et al. (2012), Phelps, Nowak and Ferrell (2000) and Xie, Teo and Wan (2006). Therefore in the low sensitivity scenario the respondents were asked to provide their marital status, occupation and education level. In the high sensitivity scenario the respondents were asked for their annual income, home address and phone number. The manipulations are shown in appendix A.
In order to test the validity of each scenario a pilot test was conducted with 12 students. The students were asked to state their perceived information sensitivity of the information asked for and their perception of the amount of the monetary reward which was given in return for the different scenarios. A paired t-test was conducted to see if the
18 each other. The results of the test showed that the respondents perceived the low sensitivity scenarios significantly different from the high sensitivity scenarios (MeanLowSensitivity= 2,17 and MeanHighSensitivity = 6,33; t(11)=37,081, p<0,00). The second test is to validate if the respondents perceive the amount of monetary reward as significantly different. There are three different levels therefore an One-way Anova test is conducted. The results shown that the different scenarios which showed the low, medium and high amount of monetary reward were seen as significantly different from each other on the perception of the amount of rewards (7-point metric scale) ( F(2,33)=193,147 , p<0,000). In addition the post hoc test showed that all the different amounts of rewards compared to each other were significantly different (p<0,000). So the pilot test showed that the scenarios for the experiment were statistical valid.
3.4 Measurements
The survey measurement items were designed to investigate the dependent variables privacy concern, willingness to share personal information and misrepresentation of the information. In addition questions about the control variables trust, prior privacy invasion, age, gender, income, education, online purchase tendency and general information sharing behaviour were made to be able to control for these variables. Furthermore additional data were collected, including manipulation checks, variables for perceived monetary reward and perceived information sensitivity and variables to be able to facilitate additional analysis. Table 1 shows all the measurements and items used in the survey to measure the different variables. All the scales and measurement items were based on prior literature to ensure the validity and reliability of the scales and measurement items. Privacy concern was measured using four items which were based on the measurements of Lee et al. (2015), Son and Kim (2008) and Taylor, Davis and Jillapalli (2009). The measurements were slightly adapted in their wording to link it more to the manipulation, for example the question “I am concerned that the
19 7 point semantic scale were used. The control variables prior privacy invasion is measured on a 7 point semantic scale and trust is measured by three items on a 7 point Likert scale (Taylor, Davis and Jillapalli 2009). The variables perceived monetary reward which is used for
additional analysis and as manipulation check was measured on a 7 point semantic and likert scales (Lee et al. 2015)( Mothersbaugh et al. 2012). Lastly perceived information sensitivity was measured individually for each different types of personal information with three different items on of the three items was reversed scales. All the three items were on a 7 point likert scale (Lee et al. 2015)( Mothersbaugh et al. 2012).
3.5 Data Analysis
After the data was collected via Qualtrics it was transferred to SPSS to analyse the data. The next step was to code the variables and give the proper values to the data to be able to interpret the data. Furthermore preparing the data for further analysis included, deleting unnecessary data, creating dummy variables and treating missing values and outliers. There were no missing values, because the respondents were forced to fill in an answer before they could proceed with the survey. There were five respondents who started the survey but only read the scenario and filled in one question. Because these respondents only answered one question they were deleted from the data set. Furthermore one respondent was deleted from the data set, because he or she did not meet the two attention checks. The respondent was asked to answer strongly disagree (1) on the questions, however both times the answers was neither agree nor disagree (4). Looking at the data which the respondent gave it made sense to delete it, because he or she answered on all the questions 4 which suggest that he quickly clicked through the survey, because it also took him only one minute to complete the whole survey while on average people spent at least 5 minutes on the survey before completing it. In addition a reliability test was conducted to test whether the multi item measurements could be put together as a sumvariable for the construct. After these preparation steps the
22
4 DATA ANALYSIS AND RESULTS
4.1 Descriptive Statistics
23
TABLE 2 Descriptive statistics
4.2 Control variable checks
In this research there are six different scenario’s which were randomly assigned to the respondents. Table 3 shows the number of respondents which were assigned to the different scenario’s.
TABLE 3 Scenario distribution
Frequency of the scenario’s
Low Reward (5%) Medium Reward (30%) High Reward (70%) High Information sensitivity 24 23 24 Low Information sensitivity 21 25 29
24 behaviour, age and trust. To test if the control variables gender, education and income differ for the six scenario’s a Kruskal-Wallis H test is conducted.
Firstly the assumptions for an one-way Anova are checked to see if it is possible to do one-way Anovas for the control variables on interval scales. The first three assumptions hold for all the control variables; prior privacy breach, likelihood to buy a new laptop, general information sharing tendency, online buying behaviour, age and trust. Because the dependent variables is on interval scale and the independent variables has more than two independent groups. Furthermore the participants are only in one of the six scenario’s so there is also independence of observations. In addition all the variables are homogeneous in the variance, however in some variables one or two of the six groups are not perfectly normally
distribution. For the variable prior privacy breach and likelihood to buy a new laptop two groups have a skewness which is a little bit higher than -1, they are between -1,2 and -1. So however the variables do not perfectly met all the assumptions still one-way anova tests are conducted.
The one way anova for prior privacy breach was not significant F(5,140)=1,801, p=0,117. However the posthoc showed that there were two groups which significantly differed from each other those were the low sensitivity low reward group with the high sensitivity medium reward group p=0,08 and the low sensitivity low reward group with the high sensitivity high reward group p=0,36. Likelihood to buy a new laptop F(5,140)=0,943, p=0,455, and the posthoc test showed no significant results. So the likelihood to buy a new laptop is not significantly different for the six scenario’s. General online information sharing tendency F(5,140)=1,186, p=0,319, however the posthoc test showed one combination of groups to be significant. Low sensitivity and high reward vs high sensitivity low reward p=0,026 was significantly different on the sharing tendency. Online buying behaviour
F(5,140)=1,162, p=0,331, however the posthoc test indicates that low sensitivity high reward group significantly differs from the high sensitivity high reward group p=0,04. Age
F(5,140)=1,334, p=0,253 and the posthoc test showed no significant differences between the groups. Trust F(5,140)=1,529, p=0,185, but two groups were significantly different from each other low sensitivity and medium reward vs low sensitivity high reward p=0,045, and the low sensitivity high reward vs high sensitivity low reward p=0,037.
25 which is the scenario has more than two categories in my case there are six categories,
observations are independent of each other and lastly the distribution of each group of the independent variable are the same shape. So therefore it is allowed to use the Kruskal Wallis test.
The first Kruskal Wallis test was performed on gender and the different scenario’s. The test was not significant, Chi-square(5)= 4,055 p=0,541 so the people in the different scenarios do not significantly differ on gender. The next test included education and the different scenario’s. This test was also not significant, Chi-square(5)= 9,455 p=0,092 so the people in the scenario’s did not significantly differ on their education level. Lastly we checked if the people in the different scenario’s significantly differed from each other based on income. The results showed that the people in the different scenario’s did not significantly differ on income level, Chi-square(5)= 2,485 p=0,779.
Based on these analysis the control variables prior privacy breach, general online information sharing tendency, online buying behaviour and trust are included in the model as covariates, because at least one of the combinations of scenarios were significantly different from each other on these variables. The variables likelihood to buy a new laptop, age, gender,education and income level are not included in the model, because those variables were not significantly different in the six scenario’s.
4.3 Information sensitivity measurements and manipulation checks
4.3.1 Information sensitivity measurements
26 Looking at the eigenvalue and the total explained variance of the three types of
information we could conclude that they all had 1 factor. Because for annual income only one item had an eigenvalue higher than 1, which was 2,941, and the total explained variance was 98% which is higher than 60%. The same results exist for address, only one item had an eigenvalue higher than 1, which was 2,868 and total explained variance of 96%. Lastly the results for phone number were not different, one item had an eigenvalue higher than 1, which was 2,908 and an total explained variance of that only item of 97%.
The next step was to check the internal consistency of the measurement this was tested by computing the cronbach’s alpha. The cornbach’s alpha should be higher than 0,7 to be able to make a sumvariable. For annual income α=0,990, address α=0,997 and phone number α =0,984. The cronbach’s alpha is higher than 0,7 so the three items per information type were summed together and dived by three to get the average perceived sensitivity per information type. The last step to get the total perceived sensitivity for the high sensitivity scenario is to add the scores of the annual income + address + phone number and divide it by 3 to get the perceived information sensitivity for the high information sensitivity scenario.
The same procedure was done for the low sensitivity scenario which contained marital status, occupation and education. For all the three different types of information a factor analysis was allowed, because the KMO for all was higher than 0,7 (marital status =0,766, occupation =0,768, education =0,70) and the Bartlett’s test was significant. In addition the three types have only one item with an eigenvalue of higher than 1 (marital status = 2,771, occupation =2,801, education= 2,828) and total explained variance of the 1 item is for all three types higher than 60% (marital status = 92,363, occupation = 93,354, education = 94,727).
The last check is to compute the cronbach’s alpha which is higher than 0,7 for the three different information types (marital status α = 0,956, occupation α = 0,962, education α = 0,967). So the three different questions per information type could be summed together and then will be summed together and divided by three to get the average perceived sensitivity for the low sensitivity scenario’s.
4.3.2 Manipulation checks
27 0,92) as significantly more sensitive than in the low sensitivity scenario’s (M=3,19, SD= 1,53). In order to test the validity of the different monetary rewards an one-way Anova is conducted to test if the respondents perceived the 5% (low), 30% (medium) and 70% (high) amount of discount significantly different from each other. The results show that F(2,143)= 81,250 p<0,00 the perceived monetary rewards significantly differ from each other. In addition the posthoc test shows that all the different amounts of rewards compared to each other significantly differed on their perceived monetary reward (p<0,000), which was
measured on a 7-point likert scale shown in table 1. So the one-way anova test shows the 70% scenario’s is perceived as the highest amount of reward (M= 6,26, SD=1,456) the medium as the middle amount of reward (M=4,96, SD= 1,166) and low amount of reward is perceived as the lowest amount (M=2,93, SD=1,214).These test showed that the validity of all the different scenario’s was good.
TABLE 4
Sensitivity of the information
4.4 Reliability of the Measurements
The variables in the model are measured by multiple items. To be precise the variable privacy concern is measured by 4 items and misrepresentation and willingness to share are both also measured by 4 items. To be able to reduce the amount of items measured for the variable a factor analysis is conduct. The KMO is 0,867 and the Bartlett’s test of Sphericity is
significant, furthermore the communalities of all the items is above 0,4 so factor analysis is allowed. Next the principal component analysis is conducted to determine the number of factors. Three factors have an eigenvalue of higher than 1 in total they explain more than 60% of the total variance namely 86,85% and they all explain at least more than 5% of the
variance. So we can conclude that the 12 items can be placed into three factors.
28 loading should be higher than 0,5. Table 5 shows the different factors and loading of the different items. From table 5 we can see that factor 1 consists of 4 items which measures willingness to share personal information. Factor 2 consists of 4 items which measures misrepresentation of the personal information and factor 3 consists also of 4 items which can be classified as privacy concern.
The next step in reducing the amount of items and testing the reliability of the measurement is by looking at the Cronbach’s alpha of the different factors. The Cronbach’s alpha for the four items of privacy concern is α =0,933, for the four items of willingness to share it is α =0,963 and for the four items of misrepresentation it is α=0,947. All the
Cronbach’s alphas exceed the threshold of 0,7 so the items will be summed together to reduce the number of variables. The same is done for the control variables trust in online companies which is measured by 3 items. The Cronbach’s alpha is α=0,879 so these three items are summed together to make one variable.
TABLE 5
Rotated component matrix
4.5 Hypothesis testing
29 of information sensitivity on the willingness to share personal information and
misrepresentation of the information which would be mediated by privacy concern and the relationship of information sensitivity on privacy concern which is expected to be moderated by amount of monetary reward. In addition, the control variables prior privacy breach, general information sharing tendency, online buying behaviour and trust are included as covariates. To test the significance of the effect the bootstrapping method is used and Confidence intervals are generated (bootstrapping sample =5000 and CI 95%), so all the variables which are significant based on their p-value also fall within the confidence interval.
The PROCESS Macro is used to make three models to investigate the moderated mediation model. The first model consists of the moderating variables monetary reward, the independent variable information sensitivity, dependent variable privacy concern and the covariates. The model itself is significant (Rsquare=0,4862, F(9,136)=14,302 p<0,000). The results show that the relationship between information sensitivity and privacy concern is significant and positive B=0,8112, t(145)=2,3722 p<0,05, which supports H1. So the more sensitive the information is the more privacy concern people become. Similarly it showed that medium monetary reward had a significant negative direct relationship on privacy concern B=-0.6605, t(145)=-1,9633, p<0,05, however the high level of monetary reward was not significant. So H2 is only significant for the change from low to medium level of monetary reward. The next step is to check whether there is a moderation effect between monetary reward (2 dummy variables for medium reward and high reward moderators) on the relationship between information sensitivity and privacy concern. The medium reward moderator (B=0,5686, t(145)=0,120, p=0,2307) and high reward moderator (B=0,3867, t(145)=0,8284 p=0,4089) are both insignificant. These results show that there is no moderation so H3 and H4 are not supported. In addition the overall test of moderated mediation for the model with willingness to share as dependent variable has for the medium monetary reward moderator a LLCI= -0,8161 and ULCI=0,1599 and for the high monetary reward moderator a LLCI= -0,7640 and ULCI=0,2815, because both ranges include 0 the moderated mediation model with willingness to share as dependent variable is not significant. Similarly the moderated mediation model with misrepresentation as dependent variable is also not significant the medium monetary reward moderator a LLCI= -0,1140 and ULCI=0,6456 and for the high monetary reward moderator a LLCI= -0,2530 and ULCI=0,6422.
30 0,99 which is good) so there are scenario’s which significantly differ on privacy concern. The posthoc test gives more information about the potential ‘U’ shape. By looking at the mean privacy concern for the different amounts of monetary reward within the low or high
sensitivity cases we see that there is no U shape, because within the low sensitivity scenario’s the low reward has a mean privacy concern of 4,2, the medium reward has a mean privacy concern of 3,9 and the high reward has a mean privacy concern of 3,9 and the posthoc test showed that these scenarios were not significantly different from each other p>0,05. The same results hold for the different amounts of monetary reward within the high sensitivity
scenario’s. The mean privacy concern for the low reward is 5,4, the medium reward has mean privacy concern of 5,3 and the high reward has a mean privacy concern of 5,2 and those scenarios were not significantly different from each other p>0,05.
Because there is no moderated mediation we still check whether there is only
mediation. To test this two mediation models are made by using PROCESS Macro model 4. the first one is to test if the relationship between information sensitivity and willingness to share personal information is mediated by privacy concern. Table 6 shows the results of the mediation model and the estimation power for both model is 1 which is very good.
TABLE 6
31 Table 6 shows that Path a, b and c’ are significant. In addition the total effect F(5,140)=19,99 p<0,01, R2= 0,4166 B=-1,3947, t(140)=-5,88 p<0,01 is also significant and based on these results we concluded that there is partially mediation, because the effect of c’ (-0,8108) is less strong than c (-1,3947), so the effect of information sensitivity on willingness to share
personal information is partially mediated by privacy concern controlled for prior privacy breach, information sharing tendency, frequency of online buying and trust. H5 is supported based on this analysis, because the results show that an increase in privacy concern lead to a decrease of 0,5061 units in willingness to share personal information. This model shows that an increase in information sensitivity and an increase in prior privacy breach (at 10%
significance) leads to an increase in privacy concern. Furthermore the more someone trust the online company the less privacy concern he will be. In addition an increase in privacy concern and information sensitivity leads to people being less willing to share their personal
information, however an increase in their general information sharing tendency and trust in the online company increases their willingness to share their personal information.
To test if the relationship between information sensitivity and misrepresentation of the personal information is mediated by privacy concern a similar model is made with
misrepresentation as dependent variable. The estimation power for these models is both 0,99 which is good. The results are shown in table 7 and table 7 shows that path a and b are significant and path c’ is not significant. The total effect is F(5,140)=3,947 p<0,01, R2
=0,1235 B=0,6829 t(140)= 2,485 p<0,05. Therefore we can conclude that privacy concern fully
32
TABLE 7
Mediation model misrepresentation
4.6 Additional analysis
In the questionnaire the respondents were given the possibility to actually provide their
information. The questions was “given the scenario you have seen would you actually provide your information” and then they got the possibility to fill in their personal information. This made it possible to measure the actual sharing behaviour of the respondents. So to get more insights into what characteristics and factors of the respondents influenced their actual
33
TABLE 8
Logit regression information sharing behaviour
The actual sharing of personal information depends on perceived information sensitivity, perceived amount of monetary reward, how likely you are to buy a new laptop soon, trust in online companies, privacy concern of the person.
34
5 DISCUSSION
5.1 Implications of results
Prior research tried to understand which elements might influence privacy concern and what kind of antecedents privacy concern will have. This research tried to give a deeper
understanding of how people react to different privacy sensitive information and how benefits like monetary reward might influence the process of privacy concern and the potential
antecedents like willingness to share their personal information and misrepresentation of the information and tried to see how the underlying cost and benefit trade-off might influence this process.
The first hypothesis looked at the effect of information sensitivity on privacy concern. As expected there was a significant positive relationship between information sensitivity and privacy concern. Meaning that in case the information which is asked by an online company is more sensitive, people will be more privacy concerned, because people perceive more risk when they are asked for more sensitive information. People are more concern because when this information is used wrongly or falls in the wrong hands it could be more harmful for them therefor they are more concerned.
Secondly, monetary reward was expected to directly reduces privacy concern. The results showed that only an increase from low to medium amount of monetary reward had a significant negative effect on privacy concern and the increase from medium to high amount of monetary reward were not significant. This effect shows that when 30% of discount was given in return for the personal information people were less privacy concern. This could be explained by the fact that there is now a reasonable benefit give for the costs of providing the information and that the 5% is seen as to little and the 70% might have created some doubts by the people of why the company gives so much in return.
35 sharing it no matter the monetary reward. However what is interesting is that the additional analysis showed that the monetary reward might not influences people’s privacy concern, but it significantly influences their actual information sharing behaviour. Because the results showed that an monetary reward was positively related to people’s actual sharing behaviour. So monetary reward does not influences how someone feels but it does influence their actual sharing behaviour.
Fourthly, privacy concern had a significant relationships with the antecedents
willingness to share personal information and misrepresentation of the personal information which was expected based on prior research. Privacy concern reduced the willingness to share personal information, people who were more concerned about their privacy are less willing to provide their information to the online company. In addition an increase in privacy concern also led to an increase in misrepresentation. The reason for this could be that people who are more concerned will not want to give their actual information, but still want to get the potential reward therefore they might misrepresent the personal information to still get the benefit and not have a cost because it is not their actual personal information. This is also supported by the fact that the perceived amount of monetary reward has a significant positive relationship with the actual information sharing behaviour. So people share at least some information even if it is false to get the reward.
Furthermore results showed that privacy concern partially mediates the relationship between information sensitivity and willingness to share information and fully mediates the relationship between information sensitivity and misrepresentation. The willingness to share still also depends partially on the direct effect of information sensitivity, because people just might be less willing to share more sensitive information no matter if they are very concerned or not.
In addition results showed that the people who have experienced prior privacy invasions are more privacy concern. This is a logical relation, because if you have
36 more often. This can be explained by the fact that if you share personal information more often there are also more possibilities that you might want to misrepresent the information, because you do not fully trust a company or have some concern then you might misrepresent a part of your shared personal information. Because if you don’t share your information generally there are also no possibilities to misrepresent the information.
Lastly the variable buying a new laptop soon is significantly positive related to sharing behaviour. People who are more likely to be in such a situation in the near future are likely to share their personal information. So it is important to target people who will potential be in such a purchase situation if you want to acquire personal information, because those people are more likely to share their personal information.
Looking back at the research question it is not possible to say that giving a monetary reward moderates the effect of information sensitivity on privacy concern, but it does actually influences people’s behaviour, So people´s feelings might not be influences but their actual behaviour does.
5.2 Theoretical and practical implications
The results from this study have both theoretical and practical implications. This study has several theoretical contributions for understanding privacy in the marketing literature. One of the main contributions is that this study tested a total model in which information sensitivity, monetary reward, privacy concern and the outcomes of privacy concern together with control variables was tested all at once in a moderated mediation model. This helps in better
understanding how all the different relations worked together. In addition in comparison to Andrade, Kaltcheva and Weitz (2002) and Premazzi et al. (2010) which included only two types of information it included three different types of personal information namely;
financial, personal identifiable and demographic information in the different scenario’s which gives a more complete view of the effects of the different information. The results showed that financial information was seen as the most sensitive information and that such
37 company. So if a company wants to acquire personal information and people need to fill in all options or nothing at all then companies need to be careful for asking financial
information, because this could lead to very privacy concerned people which do not want to share any information. Furthermore this study tried to further investigate the moderating effect of different levels of monetary reward, however the results did not show significant effects of the different levels of monetary reward. It is important, because it shows that a higher information sensitivity increases privacy concern, and that monetary reward cannot counter or influence this negative effect. Lastly an important contribution is that in this study the peoples actual sharing behaviour is also tested and which showed interesting results.
From practical perspective this study has some important implications for online companies who would want to acquire personal information from their potential customers. It is important for online companies to reduces the privacy concern and increase the willingness to share. Especially because privacy concern partially and fully mediates the outcomes
willingness to share and misrepresentation. So making sure that you are trustworthy is very important. According to Xie, Teo and Wan (2006) trust could be increased by showing privacy security labels and showing your policy and by having a good reputation. Another important element is that the online company targets the right people. Potential customers who are further in the purchase funnel so more inclined to actual buy, those people are more likely to share their personal information. Furthermore if you know that people might have experienced a privacy breach or invasion you should be cautious to ask them to share their information those people need to be convinced that the seller secures their information before they will give their information.
5.3 Limitations and future research
38 might not be generalizable for the whole population. However these people in this age
category make up a very large part of the people who buy products online. The last limitation is that this study only looked at an purchase situation and if then monetary rewards might work to reduce privacy concern. It might be that the effect will be different for different context like a service operator.
39
REFERENCES
Alharbi, Ibraheem Mubarak, Zyngier Suzanne and Hodkinson Christopher (2013), “Privacy by design and customers’ perceived privacy and security concerns in the success of e-commerce”, Journal of Enterprise Information Management, 26(6) 702 – 718
Andrade, Eduardo B., Kaltcheva, Velitchka, and Weitz, Barton (2002), “Self-Disclosure on the Web: The Impact of Privacy Policy, Reward, and Company Reputation,” in Advances in
Consumer Research (29), S. M. Broniarczyk and K. Nakamoto (Eds.), Association for
Consumer Research, Valdosta, GA, 2002, 350-353
Baruch, Yehuda and Holtom, Brooks C. (2008), “Survey Response Rate Levels and Trends in Organizational Research” Human Relations, 61(8), 1139-1160.
Benamati, John H., Ozdemir, Zafer D. and Smith, Jeff (2016), “An Empirical Test of an Antecedents – Privacy Concern – Outcomes Model.” Journal of Information Science, Forthcoming.
Buchanan, Tom, Pain, Carina, Joinson, Adam N., and Reips, Ulf-Dietrich (2007), “Development of Measures of Online Privacy Concern and Protection for Use on the Internet.” Journal of American Society for Information Science & Technology, 58(2), 157-165.
Burgoon, Judee K., Parrot, Roxanne, Le Poire, Beth A., Kelley, Douglas L., Walther, Joseph B. and Perry, Denise (1989), “Maintaining and Restoring Privacy Through Communication in Different Types of Relationships.” Journal of Social & Personal Relationships, 6(2), 131-158. Castañeda, Alberto J. and Montoro, Francisco J. (2007), “The Effect of Internet General Privacy Concern on Customer Behavior.” Electronic Commerce Research, 7(2), 117-141. Chellappa, Ramnath and Sin, Raymond (2005), “Personalization versus Privacy: an Empirical Examination of the Online Consumer’s Dilemma.” Information Technology & Management, 6(2/3), 181-202.
Cooper, Donald R. and Schindler, Pamela S. (2013), Business Research Methods: 12th
40 Culnan, Mary J. and Armstrong, Pamela K. (1999), “Information Privacy Concerns,
Procedural Fairness, and Impersonal Trust: an Empirical Investigation.” Organization
Science, 10(1), 104-115.
ECP (2014), “Onderzoek: data vragen in ruil voor karting? Geen goed idee.” (accessed February 23, 2017), [available at http://cddn.nl/nieuws/2014/11/27/onderzoek-data-vragen-in-ruil-voor-korting-geen-goed-idee].
Faja, Silvana (2005), “Privacy in E-Commerce: Understanding user trade-offs.” Issues in
Information Systems, 6(2), 83-89.
Gabisch, Jason Aaron and Milne George R. (2014), “The Impact of Compensation on
Information Ownership and Privacy Control.” Journal of Consumer Marketing, 31(1), 13-26. Goldfarb, Avi and Tucker, Catherine (2012), “Shifts in privacy concern.” American Economic
Review¸102(3), 349-353.
Göritz, Anja S. (2006), “Incentives in web studies: methodological issues and a review.”
International Journal of Internet Science¸ 1 (1), 58-70.
Hallam, Cory and Zanella, Gianluca (2017), “Online self-disclosure: The privacy paradox explained as a temporally discounted balance between concerns and rewards.” Computers in
Human Behavior, 68 (March), 217-227
Hann, Il-Horn, Hui, Kai-Lung, Lee, Sang-Yong Tom and Ivan, P.L. (2007), “Overcoming Online Information Privacy Concerns: an Information-Processing Theory Approach.” Journal
of Management Information Systems, 24(2), 13–42.
Hui, Kai-Lung, Teo, Hock Hai, Lee and Sang-Yong Tom (2007), “The Value of Privacy Assurance: an Exploratory Field Experiment.” MIS Quarterly, 31(1), 19-33.
Hoffman, Donna, Novak, Thomas P. and Peralta, Marcos (1999), “Building Consumer Trust Online.” Communications of the ACM, 42(4), 80–85.
Kelley, H. and Thibaut, J.W. (1978), “Interpersonal Relations: A Theory of Interdependence”, Wiley & Sons, New York, NY.
41 Laufer, Robert, S. and Wolfe, Maxine (1977). “Privacy as a Concept and a Social Issue: a Multidimensional Developmental Theory.” Journal of Social Issues, 33(3), 22-42.
Lee, Hwansoo, Lim Dongwon, Kim, Hyerin, Zo, Hangjung and Ciganek, P., Andrew (2015), “Compensation Paradox: the Influence of Monetary Rewards on User Behaviour.” Behaviour
& Information Technology. 34(1), 45-56.
Lwin, May, Wirtz, Jochen and Williams, Jerome D. (2007), ‘‘Consumer Online Privacy Concerns and Responses: A Power Responsibility Equilibrium Perspective,’’ Journal of the
Academy of Marketing Science, 35(4), 572-585
Malhotra, N. K. 2010. Marketing research: an applied orientation sixth edition, New Jersey, Pearson Education Inc.
Malhotra, Naresh K., Kim, Sung S. and Agarwal, James (2004), “Internet Users’ Information Privacy Concern (IUIPC): The Construct, the Scale, and a Causal Model.” Information System
Research, 15(4), 336-355.
Milne, George R. and Boza, Maria-Eugenia (1999), “Trust and Concern in Consumers’ Perceptions of Marketing Information Management Practices.” Journal of Interactive
Marketing, 13 (1), 5–24.
--- , Rohm, Andrew J., and Bahl, Shalini (2004). “Consumers' Protection of Online Privacy and Identity.” Journal of Consumer Affairs, 38(2), 217-232.
Moon, Youngme (2000), ‘‘Intimate Exchanges: Using Computers to Elicit Self-Disclosure from Consumers.’’ Journal of Consumer Research, 26 (March), 323-339
Morris, Jason and Lavandera, Ed (2012), “Why Big Companies Buy, Sell Your Data” (accessed February 23, 2017), [available at
http://edition.cnn.com/search/?sections=&text=data].
Mothersbaugh, David L. Foxx, William K., Beatty, Sharon E. and Wang, Sijun (2012), “Disclosure Antecedents in an Online Service Context: The Role of Sensitivity of Information.” Journal of Service Research, 15(1), 76-98.
Ngai, E.W.T., Xiu, Li and Chau, D.C.K. (2009), “Application of Data Mining Techniques in Customer Relationship Management: a Literature Review and Classification.” Expert Systems
42 O’Neil, Kevin M. and Penrod, Steven D. (2001), “Methodological variables in Web-based research that may affect results: Sample type, monetary incentives, and personal information”.
Behavior Research Methods, Instruments & Computers, 33 (2), 226-233.
Phelps, Joseph, Nowak, Glen and Ferrell Elizabeth (2000), “Privacy Concern and Consumer Willingness to Provide Personal Information.” Journal of Public Policy & Marketing, 19(1), 27-41.
Premazzi, Katie, Castaldo, Sandro, Grosso, Monica, Raman, Pushkala, Brudvig, Susan and Hofacker, Charles F. (2010), “Customer Information Sharing with E-vendors: The Roles of Incentives and Trust.” International Journal of Electronic Commerce, 14(3), 63-91.
Schumann, Jan H., Wangenheim, Florian von and Groene, Nicole (2014), “Targeted Online Advertising: Using Reciprocity Appeals to Increase Acceptance Among Users of Free Web Services.” Journal of Marketing, 78(1), 59-75.
Sheehan, Kim Bartel and Hoy, Mariea Grubbs (2000), “Dimensions of Privacy Concern Among Online Consumers.” Journal of Public Policy & Marketing 19(1), 62-73.
Slyke, Van Craig, Shim, J.T., Johnson, Richard and Jiang, James (2006), “Concern for Information Privacy and Online Consumer Purchasing.” Journal of the Association of
Information Systems, 7(6), 415-443.
Smith, Jeff H., Milberg, Sandra J. and Burke Sandra J. (1996), “Information Privacy: Measuring Individuals’ Concerns About Organizational Practices.” MIS Quarterly, 20(2), 167-196.
Smith, Rhys and Shao, Jianhua (2007), “Privacy and e-commerce: A consumer-centric perspective.” Electronic Commerce Research, 7(2), 89–116.
Son, Jai-Yeol and Kim, Sung S. (2008), “Internet Users’ Information Privacy-Protective Responses: A Taxonomy and a Nomological Model.” MIS Quarterly 32(2), 503-529. Stewart, Kathy A. and Segars, Albert H. (2002). “An Empirical Examination of the Concern for Information Privacy Instrument.” Information Systems Research, 13(1), 36-49.
