Executive Programme in Business Studies Marketing Track
Faculty of Business and Economics University of Amsterdam
Master Thesis
The Moderating Effect of Privacy Perception on the online
Intention-Behaviour Gap
Jonathan Kreleger Student Number: 10278710 1st supervisor: Karin Venetis 2nd supervisor: Ton Meulemans Version: Final Date of submission: 18 April 2014
Abstract
The main purpose and contribution of this paper to existing literature is to study the role of privacy perception in relation to the intention-behaviour gap, specifically in an online purchasing context. Previous (online) studies merely looked at the role of privacy perception during intention formation in an attempt to predict behavioural outcome. This study explores the intention-behaviour gap by researching what role privacy perception has during actual online consumer behaviour after intentions have already formed. To answer the research question multiple databases containing online purchase information gathered from a large Dutch bank were examined. In total the behaviour of more than 12.000 consumers was included in this study. Statistical analysis performed using SPSS resulted in several conclusions. First, the existence of the intention-behaviour gap was confirmed. Second, our results suggest that privacy perception has a moderating effect on the intention behaviour gap. Overall, our results indicate that privacy perception plays a substantial role during online purchase behaviour and should be recognized by organisations as an important aspect of the online purchase process.
Abstract ... 2
1 Introduction ... 4
2 Literature Review ... 7
2.1 Characteristics of e-commerce ... 8
2.1.1 The role of Perceived Privacy in e-commerce ... 10
2.2 Characteristics of the service industry ... 13
2.3 Defining behavioural intentions ... 14
2.4 Explaining the intention-behaviour gap ... 17
2.5 Known moderators of the intention-behaviour gap ... 19
2.5.1 Behavioural control ... 19
2.5.2 Degree of intention formation ... 20
2.6 Known effects of experience on intentions... 21
2.7 Research gap and hypotheses ... 24
2.8 Theoretical framework ... 27
3 Research Method ... 28
3.1 Data description ... 28
3.2. Data collection method ... 30
3.3 Measurement of variables ... 32
3.4 Data analysis and results ... 35
4. Discussion ... 49
4.1 Discussion of results ... 49
4.2 Theoretical contribution and managerial implications ... 50
4.3 Limitations and future research ... 51
References ... 53
1 Introduction
Intentions have been widely accepted as the best predictors of behaviour. Early models that were developed in the 70’s by Fishbein and Ajzen (1975) and Ajzen (1999) suggest intentional strength is the main determinant in predicting behaviour. Their theoretical models served as a foundation for many researchers. Most literature found today still tries to predict behaviour through known determinants of intentions: affection, normative believes and perceived control.
Despite multiple researches in the last three decades (Azjen, 1991; Notani, 1998; Liu ea., 2004; Chandon ea., 2005; Chen and Barnes, 2007; Sheeran, 2011) the gap between purchase intentions and actual purchase behaviour still remains largely unexplained. A recent comparison of multiple studies in this field carried out by Sheeran (2011) shows not even a third of the variance in behaviour can be explained by intentions, suggesting other factors are at play here as well. Still, many companies today try to lure in potential and existing customers by influencing their intentions. Without further understanding of the intention-behaviour gap, many of those efforts made to persuade consumers would remain ineffective and thus cost inefficient. Given the size of the remaining gap, much knowledge can still be gained which potentially enables more effective sales processes and increased customer satisfaction.
Hoffman ea. (1999) describe that a low conversion of intentions into behaviour have especially been found online. This online environment is developing rapidly. Consumers spend an increasing amount of their time online. With the expansion of online technologies and the introduction of new online devices, such as tablets and smartphones, both consumers and organisations will become even more reliant on the online environment in the future. These developments increase the importance of this sales channel and thus make it relevant to study online consumer behaviour.
Differences between off- and online sales channel characteristics could possible lead to differences in behavioural dynamics as well. Given the fact that e-commerce is relatively new and still developing, there is potentially still much knowledge to gain here. This provides us with good opportunities to contribute new knowledge to existing literature.
Recent literature (Liu ea., 2004; Chen and Barnes, 2007; Hui ea., 2004; Hoffman ea., 2007) on online behaviour mentions the importance of privacy perception extensively. Compared to offline purchasing processes, the Internet creates uncertainty and increased risk for consumers. These shortcomings of the Internet have raised privacy concerns among consumers. Many consumers feel current Internet laws and systems don’t protect them enough (Hoffman ea., 2007) They often believe to have no control over their personal data once it has been provided to an online vendor. This concern for privacy is even thought by Hoffman ea. (2007) to be the biggest barrier in online sales, preventing broad acceptance of e-commerce by the general public. Other found influences are: perceived usefulness, perceived security, perceived good reputation, and willingness to customize (Hui ea., 2004).
Together, the intention behaviour gap and the importance of privacy perception found online form an interesting field of research. Consumers that visit a website with the intention to purchase have at least formed some kind of trust base towards the vendor (i.e. through the vendors reputation, or overall trust in the online environment) But, even with this trust base, many consumers still opt out during the online sales process and by doing so, fail to behave according to their intentions. This phenomenon encouraged us to study whether privacy perception plays a role during purchase behaviour after intentions have already formed and has led to the formation of the main research question discussed in this paper:
What is the role of privacy perception in relation to the online Intention-Behaviour gap?
Chen and Barnes (2007) found that perceived privacy, amongst others, is specifically important to build initial trust in an organisation. Sufficient initial trust is needed in order for potential customers to eventually decide to purchase a service or product from the organisation. Post-purchase, existing or repurchasing customers can evaluate the organisation’s trustworthiness not only based on indirect factors, but also directly through the experienced security, privacy, process, product and service quality. This direct experience is found to be of great importance in trust formation and is key in forming robust and stabilized trust in an organisation. Stabilized trust is also believed to increase intentional strength, making it a better predictor of behaviour
(Sing and Sirdeshmukh, 2000; Liu ea., 2004) Once stabilized trust is formed the
costumer’s privacy perception is likely to be higher than during the initial trust phase. If privacy perception influences the intention-behaviour gap we would expect this gap to be smaller with experienced customers compared to new or potential customers.We try to answer our main research question by studying data of an online purchase process from a large Dutch financial service provider. We use multiple databases containing information that was collected from two different purchase process designs. Collection periods vary from 5 weeks to 3 months. The samples contain over 12.000 detailed interaction logs and purchase information of both potential and existing customers. We derive our conclusions from multiple statistical analyses that we applied on these data sets.
This paper unfolds as follows. First, we present a literature review to describe the characteristics of the industry and e-commerce channel in which this study is conducted. We then continue to discuss current literature about the intention-behaviour gap, known moderators of the gap and the core variable in this study: Perceived Privacy. Hypothesises formed will discuss the moderating effects of privacy perception on the intention-behaviour gap. We then continue with the methodology section, followed by the explanation of our research design and operationalization of variables. We then discuss the results of our research, their meaning and managerial implications. We end with the shortcomings of our study and suggestions for further research.
2 Literature Review
Multiple social psychological models exist proposing that intentions are the best predictor for ones behaviour. Most of them were developed during the 70’s until the late 90’s of the last century. Amongst them are the Theory of Reasoned Action (TRA) by Fishbein (1980) and the Theory of Planned Behavioural control (PBC) by Ajzen, (1985, 1991). At the base of these theories lies the assumption that a person does what he or she intends to do, and doesn’t do, what he or she doesn’t intend to do. But, as these and other studies show, intentions far from always result into behaviour.
In our current age researchers are particularly interested in online consumer behaviour (Hoffman ea.,1999; Chen and Barnes, 2007; Gentry and Galantone, 2002; Harris and Goode, 2004; Hui ea., 2004; Kim ea. 2004; Liu ea., 2004). With the arrival of new technologies and online devices (i.e. tablets, smartphones, smart TV’s) our online environment is rapidly expanding and becomes increasingly prominent in our lives. Organisations become more depended upon it, making it increasingly relevant
to understand online consumer behaviour. Special interests go out to the online intention-behaviour gap.
Because our research is conducted in an online purchasing context of financial services, we will start our literature review by explaining the main characteristics of both the services industry and e-commerce that are of relevance to this study. We then follow by discussing the definition of behavioural intentions in past and more recent literature and will continue to look at the intention-behaviour relationship. We end our review describing the research gap and our formed hypotheses.
2.1 Characteristics of e-commerce
E-commerce is a relatively new sales channel that has evolved rapidly over the last two decades. Its open character offers many advantages for consumers over offline purchasing. Liu ea. (2004) describe that the Internet is a global environment, providing consumers with unlimited access to a global assortment of products and services, 24/7. This global access gives consumers the ability to purchase products that are not available to them offline. Additionally, its open character allows for consumer friendly solutions to create structured and transparent tooling (i.e. search engines like Google) that assist consumers in their search for products. Making it possible for consumers to not only easily compare a great variety of product offerings, but it also enables consumers to purchase goods and services at the most favourable conditions (i.e. price, quantity, guarantee) while saving time.
However, compared to offline purchasing, e-commerce also offers several disadvantages for consumers. One disadvantage is the delay between the moment of purchase and delivery. This shortcoming in comparison to offline sales (where goods are generally received instantly) creates uncertainty and increased risk for consumers. Companies that offer online services and products often require consumers to provide
personal information, whereas offline sales can occur far more anonymous. This obligation to provide identification during online sales poses a threat to the consumer’s privacy, which could lead to an increased risk perception (Hoffman ea.,1999; Liu ea., 2004)
Research has shown that ‘perceived risk’ directly influences consumer attitudes towards online purchasing (Van der Heijden ea. 2003) Uncertainties and risks involved in e-commerce have led to a high-risk perception amongst consumers. This lack of trust in the e-commerce environment created a barrier that slows down the adaptation of Internet transactions by the masses (Cheskin, 1999; Hoffman et al., 1999 in Kim ea. 2004). Distrust in the online channel can even lead to consumers doubting other aspects of the transaction which are considered assured and normal with an offline store (Flavián and Guinalıu, 2006) It is widely agreed upon that only when the general public trusts the online environment, e-commerce can be a broad success. Therefor it is important to consider the role of trust in an e-commerce environment (Tan & Thoen, 2000) Given the low trust in e-commerce, risk perception is currently high and as a result many consumers opt out during the online purchase process.
A survey conducted by Hoffman ea. (1999) reports that a near 95% of web users that participated in their research had indeed refused to provide personal information to websites in the past. The reason why they declined to do so lies for a great deal in the fact that consumers don’t trust the organization running the website. Close to 63% of consumers that declined to provide personal information online said the reason to do so is because they do not trust the parties that are collecting their personal data. This lack of trust is partially created by the fact that consumers feel they lack control over the information that is provided by them to the web merchant
(Liu ea., 2004). While studying online behaviour, Hui ea. (2004) found that when consumers have no previous experience or information about the trustworthiness of the online vendor available, they rely on the online process information as a heuristic substitute to determine the vendor’s trustworthiness.This is primarily the case for first time customers, whereas existing customers already had previous off- or online experiences to determine the vendor’s trustworthiness. Experience is important because it mitigates the feelings of uncertainty consumers might initially hold when the website, its owners or the quality of the offered services is still unknown (Tan and Thoen, 2000).
It is notable that in contrary to the findings previously discussed, suggesting that concerns for privacy have a negative effect on behavioural intention, Berendt ea. (2005) came to a different conclusion. They found that even though many online consumers have strong opinions on privacy, they are unable to act accordingly. If the right circumstances appear, online users easily forget about their privacy concerns. They even are willing to provide very personal details without a clear reason to do so. They found this to be especially the case when the online environment is entertaining or when in offer for sharing personal information appropriate benefits are given. This indicates that there are multiple ways in which organisations can initially deal with privacy concerns in e-commerce.
2.1.1 The role of Perceived Privacy in e-commerce
Privacy perception is mentioned extensively in literature about online consumers behaviour (Hoffman ea.,1999; Chen and Barnes, 2007; Gentry and Galantone, 2002; Harris and Goode, 2004; Hui ea., 2004; Kim ea. 2004; Liu ea., 2004) In general, we see strong reactions by the public on topics that concern their privacy. For example, recently revealed information about NSA’s global phone
tapping activities have led to an explosion of public reactions believing these acts are a violation of their privacy. Heavy responses also occurred when privacy polices of social network sites were changed. This proves that there is a latent, if not always activated concern among the public about privacy. Organizations are increasingly using new technologies with the purpose of obtaining consumer data (i.e. characteristics, behaviour). These IT developments impact the consumer’s online privacy perception. Consumers are specifically concerned about how private data is used and treated by organizations and whether the information systems used are secure (Liu ea. 2004). Lardner (1999) found that over 40% of consumers feel that their privacy is in jeopardy. Also, 45% believes that current laws that were built to govern the Internet aren’t sufficient to protect their privacy (Lardner, 1999 in Flavián and Guinaliu, 2006)
There have been several studies that address the relationship of privacy and trust in an E-Commerce context (Hui ea., 2004; Hoffman ea., 1999), but Liu ea. (2004) were the first to point out privacy as the major antecedent of trust (see figure 1) Their research, carried out among American consumers, suggests that privacy perception has a strong influence on a person’s trust in an online store and in turn influences intentional behaviour (i.e. purchases or visits to the website). From this we arrive the assumption that privacy protection could very well be of great importance in online purchasing. Furthermore, the consumers perception that organizations will only use private data as expected and intended by the consumer is likely to influence a consumer’s online purchase decisions (i.e. actual behaviour). Flavián and Guinaliu (2006) defined Perceived privacy as followed:
“subjective probability with which consumers believe that their personal information (private and monetary) will not be viewed, stored, and manipulated
during transit and storage by inappropriate parties in a manner consistent with their confident expectations”
So far we pointed out multiple factors that have led to increasing concerns among consumers about their privacy in e-commerce context. We also noted the important role of privacy perception in gaining trust in an e-commerce business. As our online environment keeps expanding and finds new ways to infiltrate our lives we will become increasingly depended up on it. It is therefor relevant for online organisations to understand how privacy perception can be influenced. In an attempt to find answers to this question Liu ea. (2004) used the privacy dimensions provided by the US Federal Trade Commission (FTC) as the foundation for research model. It involves the following 4 dimensions:
Notice - Providing notice prior to collecting personal information Access - Giving people access to the data that is collected
Choice - People should be able to choose whether personal data can be shared or for what purposes it can be used
Security: - States that E-commerce practitioners should provide reasonable assurance that personal information is kept secure.
Together, these dimensions are thought to support the development of a transparent en secure online environment. While using these dimensions, Liu ea. (2004) found strong support for their theoretical model. Implicating that privacy perception has a strong influence on whether an individual trusts an e-commerce business. This trust will then influence the customer’s behavioural intentions to purchase from or visit the site. It also influences whether the customer will generate positive word of mouth about the organization.
2.2 Characteristics of the service industry
Our study focuses on online consumer behaviour in the service industry. Kotler and Keller (2007) described 4 key characteristics that distinct services from tangible goods: intangibility, inseparability, variability and perishability. Because services are intangible they cannot be evaluated by consumers prior to consumption (i.e. seen, tasted, felt, heard or smelled). This raises uncertainties for consumers about the services quality prior to their purchase. To reduce this uncertainty, consumers try to find clues that give indications about the service quality (i.e. organization reputation, website quality) Services are also inseparable, meaning that services are produced and consumed instantly. Tangible products are produced and consumed at distinct times. There is also a greater variability among services making them more heterogenic compared to tangible goods. Who offers the service, at what time and where all plays a vital role in the service evaluation, making it difficult to achieve consistence in service delivery. On the other hand, the quality of tangible goods has become more and more consistent as a result of continuous improvement in production systems.
Another difference between services and tangible goods is the expiration of the product. As services cannot be stored due to their intangible character, producing
them will depend on the availability of resources at any given time. When demand fluctuates, adjusting resources accordingly is often very challenging. When services are offered only at a specific time or place, unused services cannot be stored afterwards and perish.
In summery, the characteristics of services lead to an increased risk perception in the consumer’s eye compared to tangible goods. In the light of these uncertainties, experience becomes of greater importance as it helps consumers build trust in the organisation. Additionally, Flavián ea. (2005) describe corporate image is also an important factor in the degree of consumer trust formation. This is thought to be especially true in the financial service industry. In fact corporate image is considered by them to be a key element that enables the building of a sincere relationship of trust. Reputation is sometimes used in studies to determine corporate image and has great impact on online purchase of services (Flavián and Guinaliu, 2006) Reputation is believed to influence the consumer’s perception about the online store’s honesty, whether it’s truly concerned about its customers, and can be trusted (Liu ea. 2004) Since corporate image is hard to imitate and takes a long period of time to develop, it is considered to be of great strategic value in the service sector. Mainly because of it’s potential to enable both long-term objectives and help establish a competitive advantage (Flavián ea. 2005).
2.3 Defining behavioural intentions
In this paragraph we will take a closer look at how the current literature defines behavioural intentions. Triandis (1980) wrote that behavioural intentions are instructions people give to themself to behave in a certain way. Intentions are our decisions to perform certain actions. (Triandis, 1980 in Azjen, 1991) Psychology
refers to intentions as a person’s motivation to carry out certain behaviour. For example, if someone intents to have a drink with friends, this motivates that person to contact friends and plan to meet with them. Resulting in the behaviour to do so. The intensity of a decision depends on how much effort a person is willing to invest in order to achieve the intended outcome.
The theory of reasoned action (TRA) by Fishbein and Azjen (1975) considers intentions to be the most important predictor of behaviour. They position intentions as the core fundamental of attitude-behaviour relations. Their model suggests that the individual’s personal attitude and subjective norms are the two most important variables for behavioural intention strength. They also claim that if an individual has an increasing favourable attitude or social norm towards certain behaviour, intentions to perform that behaviour are expected to be greater. This will eventually lead to a greater probability of actual behaviour (see figure 2)
Figure 2: Theory of reasoned action model by Fishbein & Azjen (1975)
The theory of planned behavioural control (PBC) uses the same fundamentals, but this theoretical model also incorporates the idea that people are not always in a position from which they have sufficient control over performing the required behaviour to per sue their intentions. This phenomenon is described by Azjen (1991) as “ Perceived behavioural control” and refers to a buyer’s perception of the ease or
difficulty of performing a given behaviour. The buyer’s perception varies across situations and actions. Control beliefs are measured in terms of resources and opportunities possessed (or not possessed) by the individual. (Gentry and Galantone, 2002) We will discuss the role of control further in chapter 2.5.1.
Figure 3: Theory of planned behavioural control by Azjen (1991)
An important recommendation made repeatedly by Ajzen and Fishbein (1975) is that intention can change over time. New information for example can cause a change of mind, thus changing an individual’s intentions. It is therefor important to measure intention as close as possible to the point of the expected behaviour. Online, intentions are seamlessly converted into actual behaviour. As consumers browse through websites gathering information about products or services of their interests they unconsciously reveal their intentions to purchase these products or services. So studying intention-behaviour gap in an online environment can prove to be very valuable as intentions are measured as close as possible to the point of the expected behaviour.
As discussed, both TRA and PBC try to predict actual behaviour by looking at factors that influence behavioural intentions. What influences consumers during the actual behaviour remains until now an area of little research. The fact that prominent
theories like the TRA en PBC were developed in the 70’s and 80’s of the 20th century might explain why the focus on actual behaviour has been absent. During this period, the Internet was still absent from the masses and e-commerce didn’t exist yet. Basically all consumer activities happened offline. This made it hard to study actual behaviour (and what influences it) since often no data other than the outcome (i.e. actual purchase data) was available.
In the past it has also proven difficult to study intentions. Researchers often made use of consumer self-reported intentions in both academic and in commercial research. Chandon ea. (2005) describe this is caused mainly by the fact that self-reported intentions are an easy-to-collect proxy of behaviour, often collected through surveys. However, self-reported intentions are known to be poor predictors of actual behaviour.
2.4 Explaining the intention-behaviour gap
Sheeran (2011) performed a meta-analysis on past research on the intention-behaviour gap. He found that intentions have been studied to predict a great variety of behaviours, including: consumer and leisure decisions, smoking, weight loss, illicit drug use, driver behaviour, attendance behaviour, voting. On average, intentions accounted for 28% of the variance in these studies (see fig. 4). According to Cohen’s power primer, an established method to interpret variance, 28% can be considered as good. This supports the general believe that intentions are good predictors of behaviour.
Figure 4: Meta-analysis of meta-analysis of the intention-behaviour gap performed by Sheeran, 2011)
Still, no clarity exists on what causes the remaining gap of 72%. A 2 x 2 model designed by McBroom and Reid (1992) provides a good start to determine where to look for opportunities (see figure 5). Their basic model divides participants into two groups. Those who have a positive intention towards the studied behaviour and those with a negative intention towards the studied behaviour. Next, the model differentiates between those that acted according to their intentions and those who did not. This theory follows by assuming that the intention-behaviour gap lies within the two groups that failed to act according to their intentions, being: inclined abstainers and disinclined actors.
Subsequent behaviour Intention
Positive Negative
Acted Inclined actor Disinclined actor Did not act Inclined abstainer Disinclined abstainer
Fig 5: Attitude-behaviour by McBroom and Reid (1992)
Sheeran (2011) plotted results of several health related studies into this matrix and found that the biggest attitude-behaviour gap existed among those with a favourable attitude. Results showed a medium of 47% in this group for intenders that did not act as intended (inclined abstainers). Whereas the disinclined actors only showed a medium percentage of 7% for intenders that did perform the non-intended behaviour (disinclined actors) For this paper, we will look at behaviour of
inclined abstainers, as we will study the behaviour of individuals with a positive intention towards purchasing financial services from a leading Dutch bank. It is however fair to point out that the studies used in the Sheeran’s 2011 analysis were all health related and so the results of the meta-analysis might not be applicable to financial services. However, given the large difference between in the two groups, it is still likely to assume that, independent of the industry, the biggest gap is among the inclined abstainers.
2.5 Known moderators of the intention-behaviour gap
2.5.1 Behavioural control
Sheeran (2011) points out that one important factor in determining the intention-behaviour strength is “control”. One must be able to perform the behaviour needed to achieve the intended outcome. Many factors influence whether an individual is in control of the required behaviour. He or she will need to posses the required knowledge to carry out the behaviour (i.e. if one intends to purchase a service online, one must know how to do so), furthermore the individual must be able to perform the behaviour as well (i.e. physically able to use a computer). Next, the required resources should be available to perform the intended behaviour (i.e. time to visit the vendors website). There also needs to be an opportunity to perform the behaviour (i.e. you need to have access to internet and a computer to be able to purchase something online). A similar conclusion was made by Notani (1998) after performing a meta-analysis on previous studies on Planned Behavioural control. He wrote that:
“People intend to engage in behaviours if they perceive that they can carry them out. Similarly, intention alone is not sufficient to carry out behaviours. People need to
It is notable that research based on online behavioural data will only include those who comply with the variables discussed above. For those who have the intentions to purchase an online service but fail to comply with one or more control variables won’t end up online and thus won’t be included in research based on online behavioural data.
Another aspect of control is the certainty in which the needed behaviour leads to the intended outcome. There can be external, non-influence able factors that have impact on the intended outcome (i.e. you planned to go skiing but the ski slopes are closed due to insufficient snow fall). In case of online purchasing, which is a controlled environment, transparency of the process itself could possibly be a main determinant of whether the consumer feels he or she is in control of the expected outcome.
2.5.2 Degree of intention formation
The Degree of intention formation refers to how well a person has thought through the consequences of his or her decision to act. Sheeran (2009) wrote that if a decision has been thought through well, it is more likely the person’s intentions are well formed too and thus is more likely to behave accordingly. However, if intentions are poorly formed, it is more likely that the person will run into unforeseen situations that influence the stability of his or her intentions. Purchasing financial services often requires a well thought through decision-making process. The usefulness of many offered services, like lending or direct debit facilities, are often highly depended on the customers’ individual situation. In order to successfully fulfil a purchase, the requestor needs to determine which services suit him or her best. It is often also necessary to provide detailed information during the purchase. It is therefor fair to assume that the degree of intention formation influences the purchase of financial
services in general. However, during this study we don’t look at intention formation itself. We do acknowledge consumers need to posses a minimum of intentional formation required to start online interaction, or better, initiate a purchase process. This base of intentions that lead to customer’s initiation of a sales process is the starting point of our study.
2.6 Known effects of experience on intentions
Previous experiences with an organisation are thought to be valuable assets for consumers (Jones and George (1998) They act as a proxy for future experiences and help determine important characteristics of organisations, such as customer service, product and process quality. Another important future of experience for consumers is that it helps determine the organisations trustworthiness. (Kim ea., 2004)
Jones and George (1998) divided the trust evolution process in 3 stages: initial trust, trust stabilization, and trust dissolution. Sing and Sirdeshmukh (2000) distinguished two stages of trust: Pre-encounter and post encounter trust. The first stage corresponds largely with the initial trust stage defined by Jones and George while the latter corresponds with both trust stabilization and trust dissolution. Both theories have in common that they claim trust evolves as more experiences between the consumer and the organisation occur. They also found trust to be of great importance in forming purchase intentions.
We start by explaining the first stage of trust, which is described by Jones and George (1998) as initial trust. This trust is formed when purchase experiences with an organisation are still absent. This means perceived qualities about the vendor are formed indirectly. For example, by browsing the website, through media outings, or based on third party or public opinion about the vendor. Because true experience is
still lacking, this initial trust is fragile and easily disturbed. Additionally, Chen and Barnes (2007) found that perceived security, perceived privacy, and willingness to customize are specifically important determinants of online initial trust in an organisation. The first two relate to the risk perception of the consumer. Incomplete principles of security or privacy and media attention about online security issues may all be reasons for online consumers to be cautious about conducting online transactions. Providing online consumers with the guarantee that their private information will be safeguarded has been claimed to be key in developing online trust. We will address perceived privacy issues in e-commerce separately in the next chapter as it is of great importance to our study. Chen and Barnes (2007) also wrote that users in their study believed that a company’s willingness to customize proved the company’s capabilities, resources, and benevolence or concerns toward them (Koufaris and Hampton-Sosa, 2004 in Chen and Barnes, 2007). This believe attracted customers to purchase products or services and strengthened their online initial trust.
The next trust stage is stabilized trust. Kim ea. (2004) describe that this trust stage can only be achieved after completing a real transaction with the vendor. It includes one of the most basic trust building tenets pointed they point out as “experience with the trustee”. It is only after this experience that trust can potentially be stabilized. Positive experiences are likely to lead to an increased experience of trust in the other. Stabilization however happens only when the customer doesn’t feel he or she needs further evidence or reason for placing confidence in the trustworthiness of the store (Jones and George, 1998). Based on this we can conclude that customers who have already had a positive purchase experience with a (internet) store can be more confident in their trust belief and enjoy a higher privacy perception
because they have accumulated evidence of the store’s trustworthiness through direct experience.
Kim ea. (2004) compared the initial and stabilized trust stage to determine differences amongst the two. To do this, they looked at two groups of customers: potential and re-purchase customers (see fig. 6) When we observe their findings we conclude that how trust is invoked differs between the two groups. Structural assurance, as well as reputation and website quality are suggested to effect trust perception for both potential and repeat purchase customers. Additionally the repeat purchase customer builds trust based on customer satisfaction and previously experienced service quality. These two factors are suggested by Hui ea. (2005) to influence multiple dimensions in the trust building process. They help customers predict the vendors trustworthiness, its’ capability to deliver the promised services and its’ true intentions. These dimensions in the trust building process also influence the consumer’s privacy perception. (Liu ea., 2004; Chen and Barnes, 2007) So experienced or repeat purchasing customers are likely to have a higher privacy perception towards the vendor than new or potential customers.
Figure 6: Models of online trust building and the relevant factors by Kim ea. (2004)
2.7 Research gap and hypotheses
In summary our literature review shows that there is a general believe amongst researchers that intentions are good predictors of behaviour (Azjen, 1991; Fishbein and Azjen, 1975; Gentry and Galantone, 2002; Gollwitzer and Sheeran, 2009; Hui ea, 2004; Liu ea., 2004; Sheeran, 2011) As a result intentions have been used to predict behaviour in a great variety of fields. Nevertheless, a meta-analysis performed by Sheeran (2011) shows that less than a third of variance in behaviour is actually caused by intentions. This indicates that even when intentions have formed, consumers often still fail to act accordingly. In fact, current literature teaches us that this is especially the case online.
Current literature about online behaviour also stretches the importance of privacy perception amongst consumer extensively. Current privacy concerns amongst the general public have been reported to pose a serious threat to the acceptance of e-commerce. Although privacy perception has been found by Liu ea. (2004) to influence intention formation in online purchasing context, no research has been conducted yet into the role of privacy perception in online purchasing after intentions have already been formed. Because online purchasing often requires consumers to reveal private information, there is good reason to expect that privacy perception plays a role during the purchasing process. This assumption has led to the main research question of this paper:
What is the role of privacy perception in relation to the online intention- behaviour gap?
Before we can start to look for answers to our research question we must first establish confirmation that an intention-behaviour gap indeed exists in the online process. From this we derive our first hypotheses:
H1: There is a gap between purchase intentions and purchase behaviour found online.
Ones we have proof of the existence of the gap we can continue our research regarding the role of privacy perception towards it. Based on our literature review we assume privacy perception has a moderating effect on the intention-behaviour gap online in case one exists. We believe we can find evidence of this moderating effect by looking at differences in behaviour between potential and existing customers.
Privacy perception is most likely to be highest amongst existing customer that intent to make a re-purchase. Previous purchases have already given existing customers the opportunity to determine the vendor’s trustworthiness through direct experience. This gives them more confidence about the process outcome and the vendor’s true intentions. New customers however, still lack actual purchase experiences with the vendor and rely more heavily on indirect factors such as website security. Initial trust formed in the absence of direct experience, is less robust and fragile, privacy perception is therefor likely to be lower at this phase. Thus, if privacy perception influences actual behaviour after intention formation, we would expect experienced customers to provide privacy related information more often then new customers. From this we extract our first sub-hypothesis:
H2a: Existing customers provide privacy related information more often than new customers during the online purchase process.
Since we assume privacy perception is responsible for the expected difference in purchase process outcome between existing and new customers, we expect to see no difference in parts of the online purchasing process that don’t involve privacy related activities. This should especially be the case since behavioural control factors, also described in chapter 2.5.1, are equal for both existing and new customers. This brings us to our second sub-hypothesis:
H2b: Besides privacy related activities, existing and new customers show similar behaviour during the remaining steps in the purchase process.
If we find support for H2a and H2b, this would ultimately result in existing customers completing the purchase process more often then potential customers. Thus our last sub-hypotheses that could indicate the moderating role of privacy perception on the online intention-behaviour gap is formulated as followed:
H2c: Existing customers complete the purchase process more often than potential customers.
If we build further on the idea that privacy perception is of key importance during online purchasing, one might also assume that other determinants offer less relevance. Chen and Barnes (2007) found that an organisations willingness to customize products or services also influences consumer purchase behaviour
positively. Their main motivation is that customisation (i.e. the ability for consumers to choose) signals the vendor’s care for the customer and potentially leads to a greater fulfilment of the consumers needs. However based on our assumption that privacy perception plays an important role in the intention behaviour gap, customisation should be less relevant since there are no signs customisation of product or service offerings influences the consumer’s privacy perception. If we can find confirmation of the previous, this would provide us with additional supporting evidence that the intention-behaviour gap is for an important part moderated by privacy perception.
First we must find confirmation that customisation has an effect on purchase behaviour. Because Chen and Barnes (2007) found the effect to be positive our third hypotheses is formulated as followed:
H3a: Customisation has a positive effect on purchase behaviour
We found no evidence or indications in current literature suggesting customisation effects a consumer’s privacy perception. Therefor we believe that even when product offerings are customisable, its’ effect on the overall purchase behaviour process is smaller than the effects of privacy perception. This results in our last hypotheses:
H3b: A consumer’s privacy perception is of greater influence on online purchase behaviour than the consumer’s ability to choose.
2.8 Theoretical framework
The hypotheses discussed in the previous chapter lead up to the following theoretical framework.
Figure 7: Theoretical model
3 Research Method
In this chapter we describe the empirical part of his study. We start by describing the population in our samples. We then follow by describing how the data for this study was collected. Second the measurement of variables included in this study are discussed followed by the final paragraph in which we go into detail about the statistical procedures used in this research to test the hypotheses discussed in the previous chapter.
3.1 Data description
For this research 3 databases are used. All consist out of secondary data documented at a large Dutch bank. The databases all result from the bank’s online B2B client take on process. Database 1 and 2 contain interaction logs of customers that visited the bank’s website and initiated the customer on boarding process. Because only the website interactions were logged no demographic or geographic information about the visitors is available.
However, given the fact that the offered financial services serve a specific group of customers a general description of the visitor’s characteristics can be given. The services offered are targeted at both existing and potential customers who are owners or directors of small businesses with a projected maximum annual turn over of EUR 1 million. The business must be registered at the Dutch Chamber of Commerce and its Dutch legal form is either a “ Besloten Vennootschap” (BV) or Eenmanzaak (ZZP) Based on the limited information our databases do provide we know that the majority of visitors are most likely new business owners looking to acquire financial services that are needed to support their business. Around one third of the visitors already owned their business for a period of at least one year. This group is often looking to switch from their current financial service provider. To be more specific, database 1, contains interaction logs of 2.883 unique visitors that were taken over a period of 5 weeks. Of these visitors, 497 were existing customers of the bank and 376 visitors were potential customers. Of the remaining 2.010 visitors their current relationship status is unknown. Out of the 873 identified relationships we also know that 640 visitors owned a business that was founded less than a year ago at the moment of visit. 233 of the visitors owned their business for at least one year or longer.
Database 2 contains an overview of customers that completed the purchase process over a period of 3 months. All customers matched the general criteria that were described previously. 533 were existing and 1.258 were potential customers. Out of this group 1.210 founded their business less than a year ago and 581 had done so at least one year ago or longer.
Database 3 contains interaction logs of 9.537 unique visitors of which 1.592 are existing and 1.252 are potential customers. Of the remaining 6.694 visitors their current relationship status with the organisation is unknown. Out of the 2.844
identified visitors 1.891 founded their business less than a year ago and 953 visitors had founded their business at least one year ago or longer.
3.2. Data collection method
Most variables in the studied databases are categorical and for most part dichotomous. The first database consists out of interaction logs made during the banks current sales process (see fig. 8, process design 1). Each time a (potential) customer initiates the online sales process a log is made. This log contains a timestamp of the exact moment when the process was triggered. In total the sales process counts eight web form pages that need to be completed by the visitor. Each time a page is completed and the user continues to the next page a new log is made. This way we can follow the visitor throughout the entire process and discover exactly at what page a visitor decided to opt out or whether or not a visitor completed the process. Because each web form page serves a specific purpose during the sales process we can derive conclusions based on the customers response to that specific page.
Besides timestamps, the visitor’s device address is also registered. Since IP-addresses are unique numbers, we used this part of the log to determine the number of unique visitors in the process. By looking at unique visits instead of the total amount of times the process is triggered we erase, to some extend, duplicates from our database. For example, if a customer starts the process three times from his computer and opts out during the first two sessions but finishes the third, we only count the last session.
Figure 8: visual presentation of used data samples.
Our second database contains data of actual purchase requests made by the (potential) customer who completed the purchase process of process design 1. The purchase detail show which optional products or services the customer choose to purchase in addition to the obligatory current account and internet banking services. The following seven optional products are offered during the process: payment terminal, online payment services (iDeal), 2 types of saving accounts, short term deposit’s, insurance, financial lease. Depending on the customers organisation characteristic (i.e. brick and mortar store, internet shop) one or more service could potentially fit the organisation’s needs. The 1.791 records in the sample each represent a completed purchase request. Together they represent the total amount of purchase made over a period of 3 months in banks current process (design 1)
Last, our third database exists out of similar data as sample one, but from a previous process design (see fig. 8, process design 2) Key difference between both
process designs lies in the introduction page and their product offerings. Design 1 has two pages to obtain all necessary client information. The information provided by the customer during these two pages is needed to configure a customized offering at page 3. The customized offerings of design 1 include 2 obligatory services (current account and internet banking) plus a combination of the seven optional products discussed previously. The combined set of services is presented to the customers in three customized package deal. The offered packages vary in price and numbers of services, giving the customer the option to chose which package he or she prefers. Customers can also choose to modify the selection of services manually.
Design 2 only uses one webpage to gather client information, which is then followed by a page displaying a fixed service offering (current account and internet banking). No additional products were offered during the process.
Process design 1 was recently launched by the bank and temporally co-existed along side process design 2 during the initiation phase for a period of 5 weeks. Each time a visitor triggered the process an automatic router transferred the visitor to either the new or old process design. The router was set to route every fourth visitor to process design 1 (25%)
3.3 Measurement of variables
Behavioural intention
This variable is measured at the starting point of the purchase process, which is consciously triggered by the customer by selecting a link with the following phrase: “ Open an account ”. This link is placed in a self-service section of the banks webpage that gives an overview of services that can be purchased online directly. By selecting
this link the customer reveals his or her intention to open an account with the bank and thus purchase the service.
Actual behaviour
Actual behaviour is measured in 2 ways during this study. First we look at behaviour throughout the whole purchase process. This is done by looking at interaction logs that were created during the customer’s interaction with the purchase process (available in sample 1 and 3) Each page of the web form that is visited reveals information on how the customer behaved. Second, we use actual purchase data collected in sample 2. This data contains information about the type and number of optional products that were actually purchased by customers.
Perceived privacy
This variable is measured by the customer’s response to web form page 5 in process design 1 and page 4 in process design 2 (see fig. 9) This page requires the customer to provide personal information. Including:
• Gender • Name • Family name • Data of birth • Phone number • Email address
• Social security number
• Identification document number
Providing personal data has proven to be a barrier many online consumers are not willing to overcome. Some of the requested information in this purchase process can even be considered extremely private (i.e. social security number, identification document number). The response to this page therefor gives us a good indication about the users perceived privacy. If the user opts out at this page, the individual was not willing to provide private information and most likely had a low privacy perception. If the user does fill out the requested personal details and continues the process after this page it is clear that he or she does feel comfortable enough to provide private information and thus has a high privacy perception.
Customisation
The effects of customisation on purchase behaviour is measured by comparing conversion rates of the product offering page of process design 1: customised offering and process design 2: fixed offering (see fig. 8)
Business founding period (control variable)
During the purchase process the customer is asked to provide information about his or her business. One question specifically refers to how long the business is registered at the Dutch Chamber of Commerce at the time of the request. This question can only be answered as: < 1 year or > 1 year. Our assumption is that the longer a business exists the more experience the owner is likely to have with financial services that support it. We use this assumption to test to what extend any differences found between existing and potential customers are caused by the entrepreneurial experience of the customer.
3.4 Data analysis and results
Given the fact that data sample 1 and 3 consist out of categorical and dichotomous data our statistical options are limited. For our first hypotheses (H1: There is a gap between purchase intentions and purchase behaviour found online) we look for confirmation that a gap between intentions and behaviour indeed exists. For this analysis we simply compared the number of times customers initiated the process with the number of times those customers actually finished the process. We use de date available in Database 1 to analyse whether a gap is present. Database 1 shows that the purchase process was initiated 2,883 times by unique visitors and that only 436 of them or 15.1% completed the process (see also table 7). The remaining 84.9% of the visitors didn’t complete the purchase process and thus failed to act according to their intentions. This confirms the existence of the gap.
For our second hypotheses (H2a: Existing customers provide privacy related information more often than new customers during the online purchase process) we ran a cross table analysis (see table 1) We compared the variables “Customer Type” and “Personal identification”. Results show 334 existing customers provided personal identification details out of a total of 446 (74.9%) Out of the group of potential clients 166 identified themselves out of a total 266 (62.4%). The lowest number of expected count in this table is 79.2, allowing us to perform a Chi Square analysis.
The Chi Square analysis reports a Pearson Chi-Square value of 12,415. The found P-value is <0.01. Since we ran the test with an alpha level of 0.05, the Pearson Chi-Square value is significant. This means that existing clients filled out their personal details more often then potential clients. Thus we find support for our hypotheses (H2a).
Cust_type * Personal_Identification Crosstabulationa
provided personal
identification? Total
no yes
Cust_type Existing Count 112 334 446
Expected Count 132.8 313.2 446.0 % within Cust_type 25.1% 74.9% 100.0% % within Personal_Identification 52.8% 66.8% 62.6% % of Total 15.7% 46.9% 62.6% Potential Count 100 166 266 Expected Count 79.2 186.8 266.0 % within Cust_type 37.6% 62.4% 100.0% % within Personal_Identification 47.2% 33.2% 37.4% % of Total 14.0% 23.3% 37.4% Total Count 212 500 712 Expected Count 212.0 500.0 712.0 % within Cust_type 29.8% 70.2% 100.0% % within Personal_Identification 100.0% 100.0% 100.0% % of Total 29.8% 70.2% 100.0%
Table 1: Cross tabulation - Customer Type*Personal Identification
Chi-Square Testsa Value df Asymp. Sig. (2-sided) Exact Sig. (2-sided) Exact Sig. (1-sided) Pearson Chi-Square 12.415b 1 .000 Continuity Correctionc 11.825 1 .001 Likelihood Ratio 12.238 1 .000
Fisher's Exact Test
.001 .000
N of Valid Cases 712
a. Confirm_preselection = yes
b. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 79.20.
c. Computed only for a 2x2 table
Table 2: Chi Square analysis – Customer Type* Personal Identification
To test our third hypotheses (H2b: Besides privacy related activities, existing and new customers show similar behaviour during the remaining steps in the purchase process) we ran multiple analysis. We started by studying the interaction
logs of customers that provided their personal details and thus completed step 5 of process design 1. We did this by in two ways. First, we compared the actual completion rates of existing and potential clients during each proceeding step of the process to see if any differences occurred. Second, we compared the overall process completion rate of existing and potential clients that provided their personal details, neglecting possible difference in behaviour between each individual step. To clarify our approach we present a visual representation of the study including our test results below in figure 9.
Figure 9:Visual representation of research design and results of H2b analysis
Looking at the “ Providing Company info” step, we carried out a cross table analysis. The result show 95.8% or 320 of the 446 existing customers filled out their company’s details. For potential customers the percentage is 90.4% (150 out of 166).
The lowest expected count reported is 10 so we continue with the Chi Square analysis (see table 3)
The calculated results show a P-value of < 0.016 using a 5% alpha. (see table 4) This means the reported Pearson Chi-Square value of 5.833 is significant. Thus existing clients provided company details more often then potential clients. Our hypotheses suggesting behaviour after providing personal information is equal is not supported by this outcome.
Cust_type * Company_identification Crosstabulationa
Provided Company_identification?
Total
NO YES
Cust_type Existing Count 14 320 334
Expected Count 20.0 314.0 334.0 % within Cust_type 4.2% 95.8% 100.0% % within Company_identification 46.7% 68.1% 66.8% Potential Count 16 150 166 Expected Count 10.0 156.0 166.0 % within Cust_type 9.6% 90.4% 100.0% % within Company_identification 53.3% 31.9% 33.2% Total Count 30 470 500 Expected Count 30.0 470.0 500.0 % within Cust_type 6.0% 94.0% 100.0% % within Company_identification 100.0% 100.0% 100.0% a. Personal_Identification = 1
Table 3: Cross tabulation - Customer Type*Company identification
Chi-Square Testsa Value df Asymp. Sig. (2-sided) Exact Sig. (2-sided) Exact Sig. (1-sided) Pearson Chi-Square 5.833b 1 .016 Continuity Correctionc 4.907 1 .027 Likelihood Ratio 5.478 1 .019
Fisher's Exact Test .026 .015
N of Valid Cases 500
a. Personal_Identification = 1
b. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 9.96. c. Computed only for a 2x2 table
Table 4: Chi Square analysis – Customer Type*Company Identification
For the next step in the process in which the customer has to accept the service conditions and price we carried out the same analysis. Only this time we compared the behaviour of the remaining 470 customers that completed the previous step.
Results now show 305 of the 320 existing customers completed this step of the process (95.3%) Potential customers showed a 93.3%% completion rate as 140 of the remaining 150 customers completed the form (see table 5). The lowest expected count is still 10, thus we again follow by carrying out a Chi-Square analysis. The results now show a Pearson Chi-Square value of 0.794, which is not significant since the report P-value is 0.373 using a 5% alpha level (table 6) In this second step we find some support for our hypotheses suggesting similar behaviour between the two groups.
Cust_type * Acceptconditions_price Crosstabulationa
Accepted pricing & conditions?
Total
NO YES
Cust_type Existing Count 15 305 320
Expected Count 17.0 303.0 320.0 % within Cust_type 4.7% 95.3% 100.0% % within Acceptconditions_price 60.0% 68.5% 68.1% Potential Count 10 140 150 Expected Count 8.0 142.0 150.0 % within Cust_type 6.7% 93.3% 100.0% % within Acceptconditions_price 40.0% 31.5% 31.9% Total Count 25 445 470 Expected Count 25.0 445.0 470.0 % within Cust_type 5.3% 94.7% 100.0% % within Acceptconditions_price 100.0% 100.0% 100.0% a. Company_identification = YES
Table 5: Cross tabulation - Customer Type*Accept conditions and price
Chi-Square Testsa Value df Asymp. Sig. (2-sided) Exact Sig. (2-sided) Exact Sig. (1-sided) Pearson Chi-Square .794b 1 .373 Continuity Correctionc .450 1 .502 Likelihood Ratio .766 1 .381
Fisher's Exact Test .383 .247
N of Valid Cases 470
a. Company_identification = YES
b. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 7.98.
c. Computed only for a 2x2 table
We again carried out the same two analyses for the remaining 445 customers that entered the last step of the process: “Finalize Order” (see table 7). 298 out of the remaining 305 existing customers finalized their order (97.7%) against 138 out of 140 potential customers (98.6%) Because the expected count for potential customers that didn’t finalize the order is now 2 and thus < 5 we cannot use the results of the Chi Square test. Instead we look at the outcome of the Fisher’s exact test (see table 8) This test shows a P-value of 0.726. Because we used a 5% alpha level to perform the Fisher’s exact test the outcome difference between potential and existing customers for this final step is not significant, so we find some support for our hypotheses here.
Cust_type * Finalize_order Crosstabulationa
Finalized_order?
Total
NO YES
Cust_type Existing Count 7 298 305
Expected Count 6.2 298.8 305.0 % within Cust_type 2.3% 97.7% 100.0% % within Finalize_order 77.8% 68.3% 68.5% Potential Count 2 138 140 Expected Count 2.8 137.2 140.0 % within Cust_type 1.4% 98.6% 100.0% % within Finalize_order 22.2% 31.7% 31.5% Total Count 9 436 445 Expected Count 9.0 436.0 445.0 % within Cust_type 2.0% 98.0% 100.0% % within Finalize_order 100.0% 100.0% 100.0% a. Acceptconditions_price = YES
Chi-Square Testsa Value df Asymp. Sig. (2-sided) Exact Sig. (2-sided) Exact Sig. (1-sided) Pearson Chi-Square .364b 1 .547 Continuity Correctionc .058 1 .810 Likelihood Ratio .387 1 .534
Fisher's Exact Test
.726 .423
N of Valid Cases 445
a. Acceptconditions_price = 1
b. 1 cells (25.0%) have expected count less than 5. The minimum expected count is 2.83.
c. Computed only for a 2x2 table
Table 8: Chi Square analysis –Customer Type*Finalized order
To further test our hypotheses we also compared actual purchase information of existing and potential clients available in data sample 2. We ran an independent sample T-test to compare the means of optional products bought by both groups. The 533 existing customers in the sample show a mean of 0.49, whereas the mean for the 1,258 customers potential customers is slightly higher at 0.53 (see table 9) Levene’s test reported a P-value of 0.053 at an 5% alpha level, so we assume there is an equal variance amongst the two groups. This gives us a P-value of 0.314, which is greater than the used 5% alpha level for this analysis (see table 10) Thus we can conclude there is no significant difference between the number of optional products bought by either existing of potential customers. This again supports our hypotheses suggesting equal purchase behaviour once the privacy hurdle has been crossed.
Group Statistics Cust_Type_recode N Mean Std. Deviation Std. Error Mean Sum_of_prod Exsisting 533 .49 .767 .033 Potential 1258 .53 .819 .023
Table 10: Independent Samples Test –Customer Type* purchases optional products
Now that we found support for our hypotheses 2a and 2b, we should also find support for our third sub-hypotheses: (H2c: Existing customers complete the purchase process more often than potential customers) To confirm this, we ran a cross table analyse in SPSS on the variables “customer type” and “process completion” of data sample 1. A total of 873 records were analysed of which 377 were potential and 496 were existing customers. The cross table shows that 293 of the 496 existing customers finished the process (59.1%) whereas 124 of the 377 potential customers completed their purchase (32.9%) (see table 11) The lowest expected count in the table is 180.1. Since this is greater than 5, it allows us to perform a Chi Square test to test for significant differences.
The results of the Chi Square test show a Pearson Chi Square value of 58.846 and a P-value < 0.01 (see table 12). Because we ran the test with an alpha level of 5% the reported Chi Square value is significant. This means existing clients completed the process more often than potential clients. This outcome supports our hypotheses (H2c)
Cust_type * Finalize_order Crosstabulation
Finalized order?
Total
NO YES
Cust_type Existing Count 203 293 496
Expected Count 259.1 236.9 496.0 % within Cust_type 40.9% 59.1% 100.0% % within Finalize_order 44.5% 70.3% 56.8% Potential Count 253 124 377 Expected Count 196.9 180.1 377.0 % within Cust_type 67.1% 32.9% 100.0% % within Finalize_order 55.5% 29.7% 43.2% Total Count 456 417 873 Expected Count 456.0 417.0 873.0 % within Cust_type 52.2% 47.8% 100.0% % within Finalize_order 100.0% 100.0% 100.0%
Table 11: Cross tabulation - Customer Type*Finalized Order
Chi-Square Tests Value df Asymp. Sig. (2-sided) Exact Sig. (2-sided) Exact Sig. (1-sided) Pearson Chi-Square 58.846 a 1 .000 Continuity Correctionb 57.802 1 .000 Likelihood Ratio 59.724 1 .000 Fisher's Exact Test .000 .000 N of Valid Cases 873
a. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 180.08.
b. Computed only for a 2x2 table
Table 12: Chi Square test – Customer Type*Finalized Order
To strengthen our evidence involving the role of privacy perception in relation to the intention-behaviour gap we also ran a cross tabulation and Chi-Square analyses using the founding period of the customers business as control variable (see table 13. Results show that out of at total of 265 potential customers that founded their organisation less than a year ago 44.5% provided their personal identification. This percentage was 42.9% for the 64 customers that founded their business more than a year ago. (see graph 1 and 2)
Potential Customers
Graph 1: Cross tabulation results for potential customers: Business founding
period*Providing personal information
For existing customers the group sample exists out of 375 customers that founded their business less than a year ago and 121 that founded their business more than a year ago. Results show 67.7% of the first provided their personal identification while of the latter 66.1% did the same.
Existing customers
Graph 2: Cross tabulation results for existing customers: Business founding period *
Providing personal information
0,00% 10,00% 20,00% 30,00% 40,00% 50,00% 60,00%
Founded < 1 year Founded > 1
Client did not provide identification Customer provided personal identification 0,00% 10,00% 20,00% 30,00% 40,00% 50,00% 60,00% 70,00% 80,00% Founded < 1 year Founded > 1 Client did not provide identificatio n Customer provided personal identificatio n
