Building convenient self-service technologies : a closer look at internet-based SST design features

(1)

Building Convenient Self-service Technologies

A closer look at internet-based SST design features

Tycho Haukes 10106782 MSc BA Supervisor: dr. Evsen Korkmaz

Amsterdam Business School University of Amsterdam

(2)

(3)

Statement of Originality

This document is written by Student Tycho Haukes who declares to take full

responsibility for the contents of this document.

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

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

been used in creating it.

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

of completion of the work, not for the contents.

(4)

(5)

Abstract

The research objective of this thesis is to examine the impact of self-service technology (SST) design features on customer convenience perceptions and subsequent adaptation of SSTs. As there are several different types of SSTs, this study focusses specifically on internet-based SSTs. According to prior literature, three dominant SST design features can be distinguished. These three features are interactivity, vividness and content quality and they are the key constructs within this study. Their individual impact on convenience perceptions is aimed to be empirically tested. However, academics highlight the complexity of the construct of convenience. It is not easily understood what should be described as convenient. Therefore, literature distinguishes between time and effort perceptions. These two concepts combined encompass the phenomenon of customer convenience. Furthermore, in order to continue facilitating measurability, this research narrows down general customer convenience to the specific domain of accessibility convenience. Accessibility convenience can be defined as customer time and effort costs to request a service and eventually receiving it. Therefore, overall, this research empirically examines the relationship between the three SST design features and the accessibility convenience construct which consists of both customer time and effort perceptions.

Data is collected by means of self-completed and internet-mediated surveys. Subsequently, the data is analysed through several one-way ANOVA’s, 2x2 and 2x3 Factorial ANOVA’s as well as a multiple regression. Results indicate that different levels of vividness show significantly different perceptions regarding both time and effort expenditure. Meaning that vividness as an internet-based SST design feature significantly impacts consumer time and effort perceptions. No such results were found for interactivity and content quality. Additionally, used data shows that both time and effort perceptions significantly predict subsequent adaptation of SST. More precisely, once perceived time and effort expenditure decreases, the willingness to adapt to SSTs increases.

These results have severe managerial implications in the sense that they offer businesses the opportunity to optimize their effectiveness and decision-making regarding future development of SSTs. Theoretically, this research contributes to the general understanding of SST design and application. In implicit terms, it paves the path for further SST research to zoom in on the most concrete design aspects of SSTs. Developing adequate and efficient SSTs can only be achieved through a detailed focus on related design features.

(8)

(9)

Introduction

A publicly accepted assumption regarding technological enhancements is that they change our world rapidly. One particular change following technological inventions is that customers are becoming increasingly self-serving (Meuter et al., 2000; Campbell et al., 2011; Bitner et al., 2002). For instance, if you wanted to check into your flight ten years ago, you needed to join the queue and wait for staff members to help you out. Nowadays, at most airports you see digital interactive kiosks where you can check into your flight yourself. In academic literature these types of technologies are known as self-serving technologies or SSTs (Meuter et al., 2000; Bitner et al., 2002). In this day and age many companies are experimenting with all sorts of SSTs because they offer cost-cutting possibilities (Campbell et al., 2011). Consequently SSTs are becoming increasingly visible during our interaction with companies. Banking can be done through web-based portals, payment at gas stations via ‘on site’ pay-at-the-pump systems and tourist information can be gathered at interactive kiosks. These are just a few examples of how most technological advances are built in order to increase customer self-service (Meuter et al., 2000; Bitner et al., 2002). However it is not known to what extend SSTs are fully preferable from a service perspective (Meuter et al., 2000; Bitner et al., 2002).

The scientific domain of SSTs is broad and relatively young. Understanding of the phenomenon is growing, however still little exposed. Therefore, in this research I will try to add to the understanding of SST acceptance by focussing specifically on how these SSTs are designed and how this subsequently impacts the construct of customer convenience. Convenience of SSTs has been examined in prior research (Szymanski & Hise, 2000; Collier & Sherrell, 2010; Collier & Kimes, 2013; Brown, 1990; Berry et al., 2002). Convenience for example has been noticed as a driver of SST adaptation (Collier & Sherrell, 2010; Beatson et al., 2006; Chiang & Dholakia, 2003). However, convenience is difficult to narrow down as a construct. Yet most academics referred to it as time and effort customers need to undertake in order to gain a service (Szymanski & Hise, 2000; Brown, 1990; Berry et al., 2002; Collier & Sherrell, 2010). Although time and effort are therefore indicated to be drivers of SST adaptation, none of the prior literature puts explicit emphasis on how time and effort perceptions can be influenced. While Berry et al. (2002) argue that service system design is “instrumental in managing the time and effort costs required for customers to use a service” (Berry et al., 2002: p. 9), it is not clear which features such a system design should encompass in order to increase perceived time and effort in SSTs.

(10)

Still research on SST system design has shown several shared features. In an early paper, Steuer et al. (1995) distinguish between two dominant types of SST interface features: interactivity and vividness. Interactivity refers broadly to the two-way communication between customers and companies, regardless of time and distance (Blattberg & Deighton, 1991). Therefore, an interactive SST should enable customers and companies to communicate directly and clearly. Vividness, on the other hand, refers to the breadth and depth of an SST (Steuer et al., 1995). Breadth, in this sense, is meaning that an SST should encompass colours, animations, movies, etc., whereas depth refers to the resolution and responsiveness of an SST (Steuer et al., 1995). According to Steuer et al., (1995) this depth concept of SSTs can be best explained as ‘quality’ of the visualizations. In the real world, depth is taken for granted since our sensory mechanisms are almost always fully operating. However in the case of technology, audio or visual representations do not have the same capacities as humans, therefore the depth is dependent on bandwidth (Steuer et al., 1995). In addition to interactivity and vividness, a third SST interface feature was added by Huizingh (2000) and Dadzie et al. (2005), namely content quality. Content refers to all the information which is presented to customers via SSTs (Huizingh, 2000).

All in all, prior research indicates there is a lack of insights into self-service system design and its impact on customer time and effort when using SSTs (Berry et al., 2002; Collier & Sherrell, 2010; Collier & Kimes, 2013). Although three SST design features have been detected in academic literature, their subsequent impact has not been examined simultaneously yet. Therefore, the unique contribution of this study is that it examines how these three SST features (interactivity, vividness and content quality) impact time and effort customers need to undertake in order to obtain a service through technology. In addition, the influence of changing time and effort levels on subsequent adaptation of SSTs will be empirically examined.

The research question on which this thesis is built can be formulated as follows: How

do interactivity, vividness and content quality related SST features impact customer time and effort perceptions and subsequent adaptation of SSTs? Furthermore, five sub questions are

being examined in order to help generating a complete answer to the research question:

 Which elements encompass SSTs design?  How is self-service interactivity constructed?  How is self-service vividness constructed?  How is self-service content quality constructed?

(11)

In addition, the research objective is to understand the relationship between self-service system features and time and effort perceptions. Moreover, light will be shed on the impact of these time and effort perceptions on adaptation of SSTs.

Accurate answering of these questions will result in both theoretical as well as managerial insights. To this date, SST literature has focussed on the relationship between SSTs and convenience perceptions (Collier & Sherrell, 2010; Collier & Kimes, 2013; Berry et al., 2002, Weijters et al., 2007; Kallweit et al., 2014). However, none of the prior literature has put explicit empirical focus on the two components of perceived convenience: time and effort. Moreover, the subsequent impact of interactivity, vividness and content quality as SSTs design features on these time and effort constructs has not been examined simultaneously yet. Although SST design features have been highlighted as an important aspect in forming customer perceptions on time and effort (Berry et al., 2002). Therefore, the unique theoretical contribution of this research will be to examine the relationship between technological self-service system design and perceived time and effort expenditure. Furthermore, self-self-service technologies are indicated to be key in creating long-term business success (Bateson, 1985; Meuter et al., 2000). Moreover, turning customers in co-producers of value can reduce costs significantly (Campbell et al., 2011). Self-service technologies therefore have some clear advantages. This research will have severe managerial contribution since it will zoom in on one of the key drivers of customer adaptation of SSTs. Furthermore user-interface features which impact perceived customer time and effort will be examined. The results of which design features lead to significantly different outcomes in time and effort perceptions may subsequently be used by firms to optimize their SSTs.

What follows in the remaining of this research is a deepened understanding of the relationship between SST design features and both perceived time and effort expenditure. More precisely, based on the conducted statistical analyses, vividness of SST interfaces is found to be significant in decreasing perceived customer time and effort expenditure. This insight extends the current knowledge on SST design since the impact of vividness as an SST design feature on both time and effort perceptions has not been empirically examined before. It therefore offers the marketing science community a starting point to more explicit research on SST designing and its implications. Overall, based on the findings of this research, it can be argued that SST vividness is key in pursuing increased customer convenience. No such findings are found for interactivity and content quality. Additionally, this research confirms prior assumptions by Collier and Sherrell (2010), and Collier and Kimes (2013). Findings within this thesis namely indicate that when perceived time and effort expenditure decreases, the willingness to adapt to SSTs increases. Meaning that customer convenience can indeed be considered a key driver in customers’ adaptation of SSTs. Finally, two moderating effects are

(12)

being tested. However, it is concluded that, based on this research, gender is not significantly influencing the relationship between content quality and both perceived time as well as effort expenditure. In addition, age does not significantly moderate the relationship between vividness and both time and effort perceptions.

The following chapters are structured as follows. After this introduction section, an extensive literature review will be presented on self-service technologies, time and effort perceptions when using SSTs, interactivity, vividness and content quality. Following the literature review, a conceptual framework will be presented in chapter 2. Also, accompanying hypotheses will be formulated and discussed. Subsequently, in chapter 3, the methodology of this research will presented. Chapter 4 will provide the results and systematic answers to the hypotheses. Finally, within chapter 5, conclusions will be drawn while limitations will be discussed. Additionally, suggestions are made for future research.

(13)

1. Literature review

Within this chapter, the relevant and contemporary research involving SSTs, customer convenience and SST design features will be discussed. The chapter will be divided into six sections. The first section provides a general overview of technology based self-services, including different types of SSTs and their day-to-day application within business context. A specific type of SSTs which will be highlighted are internet-based SSTs. These internet-based SSTs will afterwards be the empirical center of attention for the remaining parts of this research. The second section of this chapter focusses on customer convenience. Different types of customer convenience will be presented and, more explicitly, two core elements of convenience will be discussed: perceptions on time expenditure and perceptions on effort expenditure. The third, fourth and fifth section within the chapter will, respectively, deal with: interactivity, vividness and content quality. According to scientific literature these three constructs are distinctive SST design features. Therefore, their characteristics and appearances will be individually discussed. Finally, the last section of this literature review, will present a schematic overview of all the discussed research and their points of focus.

1.1 Technology based self-service

Self-service technologies are broadly defined as technological interfaces which allow customers to produce services independently from the service provider (Meuter et al., 2000). According to Bitner et al. (2002), there are four types of these self-service technological interfaces. The first would be telephone-based or Interactive Voice Response (IVR) systems. These interfaces are widely used, for example when calling company service desks you often get a menu of options (Bitner et al., 2002). However, due to technological enhancements IVR systems are becoming increasingly intelligent in recognizing voices and sentences. Therefore, these systems can be used in more complex ways as well. A second type of SSTs according to Bitner et al. (2002) are Internet-based interfaces. These include, online banking or ordering products via company websites. As noted previously, these internet-based SST will be the main focus of this research. More precisely, the graphical user-interface design of Amazon.com will be used as a framework to empirically test different levels of interactivity, vividness and content quality. A third type of SSTs are interactive kiosks. Interactive kiosks are on-site computer terminals which are used for example as a possibility to check into flights (Bitner et al., 2002). Lastly, a fourth type of SST are video or CD technologies (Bitner et al., 2002). These are mostly used for educational purposes.

(14)

Campbell et al. (2011) state that self-service increasingly requires customers to become involved in the service process. Traditional boundaries between service producers and service users are shifting (Campbell et al., 2011). Whereas customers used to be largely isolated from the service production, nowadays they form an essential part of the production (Campbell et al., 2011). For instance, customers who wanted to withdraw money from their banking accounts back in the seventies had to go to banking offices where employees delivered a service and handed over the money. When the self-service ATM was introduced, customers could obtain their money without direct contact with service employees. Consequently, customers now perform many of the tasks formally done by service employees (Campbell et al., 2011). Self-service technologies therefore inherently require co-creation of value (Campbell et al., 2011). Turning customers into co-producers is one of the reasons for implementing SSTs (Meuter et al., 2005). Another reason might be to keep up with technological enhancements of competing firms (Meuter et al., 2005; Weijters et al., 2007). Furthermore, waiting time for customers can be reduced and costs can be saved for firms due to automation (Weijters et al., 2007). However, some authors suggest that the true value of self-service lurks in its potential to expand revenues through enhanced service and building profitable customer relationships (Rust & Kannan, 2003; Campbell & Frei, 2010). Overall, self-service purchasing has become a much used channel to serve customers.

Literature on SSTs generally distinguishes two forms of SSTs: transaction-related and information-related SSTs (Meuter et al., 2000). According to Kallweit et al. (2014), there is a recent change in predominance of these two forms. Traditionally, SSTs were designed in order to facilitate transactions (Kallweit et al., 2014). However, current self-service technologies are focussed more on providing information and delivering rich customer experiences rather than merely executing transactions (Kallweit et al., 2014). An explanation for this shift is the observed lack of satisfactory service by customers as a consequence of SSTs (Kallweit et al., 2014; von Wangenheim et al., 2015). A lack of service quality as a consequence of self-service technologies has also been noticed in prior research (Beatson et al., 2006; Meuter et al., 2000). Therefore, firms are continually trying to enrich customers online experiences through more enhanced SSTs (Kallweit et al., 2014).

As stated before, the scientific domain of SST is broad and relatively young. Current literature on self-service technologies for instance focusses on its impact on customer retention (von Wangenheim et al., 2015; Campbell & Frei, 2010) or the level of technology anxiety which influences customer behaviour in self-service (Meuter et al., 2003). These kind of studies contribute to our general understanding of self-service as a channel and its implications for customer experience. However, in order to narrow down this broad SST

(15)

domain, the continuation of this thesis will solely focus on literature concerning internet-based SST design in relation to convenience and both time and effort perceptions.

All in all, prior literature shows that SSTs can cut costs and increase revenues for firms (Weijters et al., 2007; Meuter et al., 2005) as well as providing a condition for long-term business success (Bateson, 1985; Meuter et al., 2000). However, rash SST implementation can have severe impact on service quality, customer defection and customer-relationships in general (Beatson et al., 2006; Meuter et al., 2000; Kallweit et al., 2014; von Wangenheim et al., 2015).

1.2 Customer convenience

The previous subsection provided an overview of self-service literature and its implications. Now the focus will shift towards customer convenience perceptions when dealing with self-service technologies. More precisely, literature on customer time and effort will be discussed and drivers for using technological self-services will be highlighted. As stated before, customer convenience is a broad and general construct. Therefore most academics refer to it as time and effort customers need to undertake in order to obtain a service (Szymanski & Hise, 2000; Brown, 1990; Berry et al., 2002; Collier & Sherrell, 2010). Although tightly connected, time and effort can be studied as individual constructs. Time is often considered a scarce resource. Time cannot be expanded and therefore is a limited resource that people try to allocate efficiently (Berry et al., 2002). Therefore the relationship between time and convenience can be found in the fact that time scarcity leads consumers to purchase convenient goods (Berry et al., 2002). On the other hand, effort is related to customer’s energy expenditure (Berry et al., 2002). This construct of customer effort has been undervalued by academics according to Berry et al. (2002). Convenience literature has namely mainly focussed on the time construct and assumed that saving time equals saving energy (Berry et al., 2002; Brown, 1990). However, according to Mohr and Bitner (1995), there are different forms of effort, including emotional, physical and cognitive effort and these cannot always be directly translated to the construct of time. For example, the emotional effort in order to perform a task might be much higher than the sole time it took someone to perform the task. Furthermore, Mohr and Bitner (1995) define perceived effort as “the amount of energy an observer believes an actor has invested in a behaviour” (Mohr & Bitner, 1995: p. 243). In addition to Berry et al. (2002), Mohr and Bitner (1995), highlight the role of enthusiasm as a valid determinant of perceived effort. All in all, it can be stated that convenience as a complex construct has to be understood deeply and consists of elements of time as well as effort.

(16)

However, in addition, convenience can also be narrowed down in specific types. Berry et al. (2002) argue there are five types of service convenience. First off, decision convenience involves the time and effort costs customers need to make in order to make a decision. For example decisions concerning which service to buy or which service provider to use (Berry et al., 2002). The second type of service convenience is accessibility convenience. This encompasses customer time and effort costs to request a service and eventually receiving it (Berry et al., 2002). As Collier and Sherrell (2010) noted, this accessibility convenience is fundamentally effected by the introduction of self-service technologies. SSTs namely drastically impact customers’ ability to initiate a service request and ultimately receiving it due to its capacity to be accessible ‘anywhere and anytime’ (Collier & Sherrell, 2010). A third type of service convenience is transaction convenience. This incorporates the time and effort expenditures in order to successfully complete a service transaction (Berry et al., 2002). Fourth, benefit convenience involves customer time and effort costs to receive core benefits of the service. This type particularly can have severe negative implications. For instance when customers put in time and effort but are unable to receive the core benefits of the service (Berry et al., 2002). Lastly, post-benefit convenience relates to time and effort expenditures when customers want to contact service providers in a post-purchase context for maintenance or repair issues (Berry et al., 2002). Distinguishing these types helps us to gain more understanding of the broadness of convenience. Because of the complexity of convenience as a construct, this study will narrow down convenience as accessibility convenience. As stated by Collier and Sherrell (2010), this specific type of convenience is drastically impacted by the introduction of SSTs. Therefore, this relationship is particularly interesting to examine.

When we translate these insights of convenience to self-service technologies, it becomes clear that there are important implications. The introduction of self-service technologies as a way to obtain your service has impacted our understanding of customer service (Collier & Sherrell, 2010). Brown (1990) for instance, argued that convenience could either mean that others would provide you a service without customers needing to interfere (high convenience), or people had to completely undertake the action themselves in order to obtain the service (low convenience). This early conceptualization of Brown (1990) shows that the introduction of self-service technologies has some conflicting implications for our general understanding of customer convenience. Self-service namely is inherent to people undertaking complete action on their own, which in Brown’s (1990) initial view stood for a low level of convenience.

(17)

1.3 Interactivity

In the previous chapter it has become clear that customer convenience manifests itself as a key reason why people turn to SSTs (Collier & Kimes, 2013). Internet allows us to visit several virtual stores within no time, and a few mouse clicks can initiate the process of buying and delivery right at your doorstep. However, which specific user-interface features contribute to SSTs becoming convenient? Prior research does not focus on this relationship between service system design and subsequent customer convenience. Still, there have been authors who have categorized several system features. In an early work by Steuer et al. (1995), system design features are classified as either interactive or vivid. Dadzie et al. (2005) and Huizingh (2000) argue to add a third class of system features, namely content quality.

Within this thesis I will build on these three constructs. I will zoom in on established scientific perspectives on interactivity, vividness and content quality. Subsequently an overview of measurement items for the three constructs will be presented.

The first construct of SST design features is interactivity. According to Steuer et al. (1995) interactivity can be defined as “the degree to which users of a medium can influence the form or content of the mediated environment” (Steuer et al. 1995: p. 10-11). In addition, website interactivity has also been defined more broadly as the communication between customers and companies, regardless of time and distance (Blattberg & Deighton, 1991). Therefore, interactive websites offer two-way communication and enable visitors to move swiftly and controlled through the site. Wu (1999) explains interactivity as “a two-component construct consisting of navigation and responsiveness“(Wu, 1999: p. 255). The concept of interactivity has been empirically examined in prior research. Coyle and Thorson (2001) for example, use interactivity of websites in order to prove that an increase in interactivity leads to higher levels of telepresence. Therefore, according to Coyle and Thorson (2001) interactive websites can establish a feeling of presence for the visitors. Additionally, many academics focus on the effects of personalized marketing in internet-based self-service technologies (Schafer et al., 1999; Goldsmith et al., 2004). According to Schafer et al. (1999) personalized recommendation systems can enhance sales opportunities in three ways. Firstly, customers can be assisted in finding the products they are looking for before leaving the website. Secondly, opportunities for cross-selling can be exploited. And thirdly, customer loyalty can be improved due to a value-added relationship (Schafer et al., 1999). Therefore, personalized marketing systems can offer substantial value to the two-way interaction between consumers and companies. Gaining customer insights and exploiting these via accurate personalized recommendations allows firms to enhance their interactivity with customers.

(18)

As shown, there are many perspectives on interactivity of websites. According to Liu (2003), this inconsistency among researchers may lead to inefficient use of websites as marketing communication channels. Therefore the conceptualization of interactivity is crucial. Wu (1999) provided an early list of measurement items (Appendix A, Table 1) to empirically examine the construct of interactivity. His list of items mainly focusses on the control and navigation aspect of interactivity which was also highlighted in his paper (Wu, 1999). In another paper, Liu (2003) presented a more complete conceptualization of interactivity. His measurement list also include the two-way communication aspect of interactivity.

1.4 Vividness

Where the interactivity construct inherits two-way action, vividness only consists of one-way communication (Fortin & Dholakia, 2005). As we have seen, interactive websites allow visitors to communicate with companies regardless of time and distance. Vividness however relates to a ‘flat’ level of communication in which visitors’ attitudes are formed on the basis of rich customer experience when surfing through the website (Coyle and Thorson, 2001). According to Steuer et al. (1995) the construct of vividness encompasses two forms: breadth and depth. Breadth refers to the number of sensory dimensions (Steuer et al., 1995). These may include colours, graphics, animation, movies, etc. Depth, on the other hand, relates to the resolution and bandwidth of the vividness (Steuer et al., 1995). Prior research by Coyle and Thorson (2001) shows that the level of vividness of websites can also have significant impact on enduring attitudes. Attitudes on websites with high levels of vividness were being measured immediately after exposure and two weeks after exposure. Consequently, Coyle and Thorson (2001) concluded that attitudes towards vivid websites were significantly higher for the group who was measured two weeks after exposure in comparison to the group who was measured directly after exposure. In addition, attitudes towards websites which express low levels of vividness were higher when measured immediately. This means that websites with high levels of vividness can have a long-term effect on peoples’ attitudes, whereas attitudes towards sites with low vividness are decreasing over time (Coyle and Thorson, 2001). All in all, prior research shows that vividness can be an important service system feature. However, its impact on perceived convenience is yet unknown.

Measuring the construct of vividness is difficult. Vividness namely is a rich construct which can be expressed in lots of ways. Moreover, technology is improving rapidly. Therefore websites which were perceived vivid a couple of years have now become old-fashioned. However one early set of measurement items was provided by Kelley et al. (1989) in Table 2 of Appendix A.

(19)

1.5 Content quality

The third service system feature is actually not a design feature. Rather, it concerns the content of websites. Or as Huizingh, describes it: “content refers to the information, features, or services that are offered in the Web site, design to the way the content is made available for web visitors” (Huizingh, 2000: p. 123). Although different, content quality is a vital part of websites. It is often said that ‘content is king’. Therefore website content quality, should be incorporated as a site feature when examining its impact on perceived customer convenience and subsequent attitude towards SSTs (Huizingh, 2000; Dadzie et al., 2005).

Huizingh (2000) distinguishes three categories of content: information, transaction and entertainment. Information content covers background information of the company as well as specific product information (prices, specifications, in-stock availability, terms of delivery, etc.) (Huizingh, 2000). In addition, transaction content refers to information on online buying options like ordering features (Huizingh, 2000). And lastly, entertainment content is mainly marketing communication on events, updates, trends and industry-related articles (Huizingh, 2000). Further research has shown that people are more interested in product information content on prices, in-stock availability, etc. than any other type of online content (Huizingh, 2000; Dadzie et al., 2005). Therefore this thesis will relate to product information content when mentioning content.

1.6 Schematic overview

In the previous five subsections some key constructs of this study have been discussed. Light has been shed on several academic perspectives on time and effort perceptions, consumer convenience, self-service technologies, interactivity, vividness and content quality. In order to establish a clear sense of this literature review, a schematic overview is presented in which the focus of key articles is highlighted.

Author Empirical/ theoretical Focus on self-service technology Focus on convenience Covariates

Interactivity Vividness Content quality Collier &

Kimes (2013)

Empirical Yes Yes - - -

Collier & Sherrell (2010)

(20)

Berry et al. (2002)

Theoretical Yes Yes - - -

Brown (1990) Theoretical No Yes - - - Bitner et al. (2002) Theoretical Yes No - - - Meuter et al. (2000) Empirical Yes No - - - Meuter et al. (2005) Empirical Yes No - - - Mohr & Bitner (1995) Empirical No Yes - - - Weijters et al. (2007)

Kallweit et al. (2014)

Jacoby et al. (1976) Theoretical No Yes - - - Steuer et al. (1995) Theoretical Yes No x x x Coyle and Thorson (2001) Empirical Yes No x x - Huizingh (2000) Empirical Yes No - - x Dadzie et al. (2005) Empirical Yes No - - x Fortin & Dholakia (2005) Empirical Yes No x x - Kelley et al. (1989) Empirical Yes No - x -

Liu (2003) Empirical Yes No x - -

Wu (1999) Empirical Yes No x - - Blattberg & Deighton, (1991) Theoretical Yes No x - - Haukes (2015)

(21)

2. Conceptual framework

This chapter will highlight the conceptual framework which is shaped according to prior research as discussed in the literature review. In addition, twelve hypotheses will be presented and their relationships argued.

As can be seen in Figure 1, the framework consists of six constructs. Three independent variables (interactivity, vividness and content) have been discussed in the literature review. According to Steuer et al (1995) and Huizingh (2000) these three constructs are dominant SSTs design features, which could have severe impact on time and effort perceptions according to established literature. Six hypotheses (H1, H2, H3, H4, H5, and H6) are defined to examine the relationships between the IV’s and subsequent time and effort expenditure perceptions.

H1 and H2 involve the relationship between interactivity of an SST and subsequent perceptions on time and effort expenditure. A key characteristic of SST interactivity is that it offers two-way communication (Steuer et al., 1995). By doing so, interactive SSTs allow users to “seek and gain access to information on an on-demand basis” (Fortin & Dholakia, 2005: p. 388). This should give users of interactive SSTs the sense that time and effort expenditures are minimal, since access to real-time assistance in interactive SSTs is abundant. Therefore H1 and H2 are defined as follows:

(22)

 H1: There is a significant difference between perceived time expenditure for different

levels of interactivity.

 H2: There is a significant difference between perceived effort expenditure for different

levels of interactivity.

The third and fourth hypotheses (H3, H4) zoom in on the relationship between vividness and perceived customer time and effort expenditures. Vividness as a concept is often mixed up with interactivity. As noted in the literature review, it differs however on possibility for two-way communication (Fortin & Dholakia, 2005). A TV for instance is highly vivid but non-interactive. Furthermore, vividness is characterized by its media richness (Steuer et al., 1995; Fortin & Dholakia, 2005). It can involve high number of visualizations, images and colours. According to research conducted by Frey and Eagly, these “colourful language, concrete images, and provocative metaphors contained in our vivid editorial distracted subjects from following its more abstract, logical (and possibly persuasive) line of argumentation” (Frey & Eagly, 1993: p. 33). Distraction therefore can be seen as an implication of highly vivid SSTs. As a consequence, it is assumed that perceived time and effort expenditures are significantly different for different levels of vividness.

levels of vividness.

Hypotheses 5 and 6 relate to the last independent variable, namely content quality. According to Huizingh (2000), the perception of content quality can be measured by the question whether it is informative. Providing value for customers is what defines strong content websites (Huizingh, 2000). This value includes information concerning product information (price, availability, etc.) as well as contact information (Huizingh, 2000). The function of content is therefore to inform customers. According to a study by Grotenhuis et al. (2007), customers desire information because of time and effort savings. Therefore, in this study it is hypothesized that perceptions concerning time and effort expenses are significantly different for different levels of content quality. This is formulated as follows:

(23)

levels of content quality.

Furthermore, it has been noted that convenience is a complex concept. It consists of several types and involves elements of time and effort. Collier and Sherrell (2010) argued that the specific type of accessibility convenience has been particularly influenced by the introduction of SST. As stated before, I will therefore narrow the construct of convenience down to accessibility convenience during this thesis. This type of convenience can be defined as customer time and effort expenditures in order to request a service and eventually receiving it. Finally, the outcome variable is customer adaptation of SSTs. As been acknowledged by Bateson (1985) and Meuter et al. (2000), self-service technologies are indicated to be key in creating long-term business success. Moreover, turning customers in co-producers of value can reduce costs significantly (Campbell et al., 2011). Therefore firms are keen to increase customer usage of self-service technologies. Testing the impact of perceived time and effort in the context of accessibility convenience on the outcome variable of customer adaptation of SSTs should therefore be of added value. In order to test this impact, two hypotheses have been formulated (H7, H8):

 H7: There is a negative direct effect between perceived customer time expenditure and

adaptation of SSTs.

 H8: There is a negative direct effect between perceived customer effort expenditure

and adaptation of SSTs.

Additionally, the conceptual model shows two possible moderating variables (age & gender). According to Cleveland et al. (2003), females show higher involvement and information processing while shopping than males. This conclusion might indicate that gender can have a severe impact on how SST users perceived time and effort expenditures. Specifically, the content quality feature of SSTs might be moderated by gender, since females have higher information processing while shopping and therefore the relationship between content quality and time and effort perceptions might be influenced. H9 and H10 are defined as follows:

(24)

 H9: The relationship between content quality and perceived effort expenditure is

different for gender.

 H10: The relationship between content quality and perceived time expenditure is

different for gender.

Another proposed moderating variable is age. According to research done by Morris and Venkatesh (2000), younger people are more driven to adopt the use of technology as compared to older people. Therefore age can be seen as a moderating variable. In this specific study it is hypothesized that age has a moderating impact on the relationship between vividness and perceived time and effort perceptions. As suggested by Steuer et al. (1995) SST vividness includes virtual visualizations and animations. And as stated by Frey and Eagly (1993), highly vivid SSTs might increase distraction. Consequently, especially older people could become distracted by this feature, hence, their time and effort perceptions might increase as compared to younger people, who should be able to adopt to a rise in visualizations. This leads to H11 and H12:

 H11: The relationship between vividness and perceived effort expenditure is different

for age.

 H12: The relationship between vividness and perceived time expenditure is different

for age.

All in all, in this chapter the conceptual model and hypotheses underlying this study have been presented. In order to gain an overview, Figure 2 offers a list of all the hypotheses as discussed within this chapter.

H1: There is a significant difference between perceived time expenditure for different levels of interactivity. H2: There is a significant difference between perceived effort expenditure for different levels of interactivity. H3: There is a significant difference between perceived time expenditure for different levels of vividness. H4: There is a significant difference between perceived effort expenditure for different levels of vividness. H5: There is a significant difference between perceived time expenditure for different levels of content quality. H6: There is a significant difference between perceived effort expenditure for different levels of content quality H7: There is a negative direct effect between perceived time expenditure and adaptation of SSTs.

H8: There is a negative direct effect between perceived effort expenditure and adaptation of SSTs. H9: The relationship between content quality and perceived effort expenditure is different for gender. H10: The relationship between content quality and perceived time expenditure is different for gender. H11: The relationship between vividness and perceived effort expenditure is different for age. H12: The relationship between vividness and perceived time expenditure is different for age.

(25)

3. Methodology

In this chapter the applied methodology will be discussed. Choosing the appropriate research design and sampling technique is crucial in order to reach satisfying results. Every design has its own strengths and limitations. The limitations of this research design will be rightly discussed. For now, this chapter has four sections. Firstly, the used research design will be described. Secondly, I will zoom in on the process of sampling. Thirdly, the used questionnaire and measurement items will be presented. Finally, the method of statistical analysis will be discussed.

3.1 Research design

This research is quantitative and explanatory. It tries to discover whether differences in SST design features matter for accessibility convenience perceptions of users. More specifically, this accessibility convenience is divided into time and effort perceptions and both of these constructs function as a dependent variable. As a data collection technique this study will use a cross-sectional questionnaire. The questionnaire will be self-completed and internet mediated by using Facebook, LinkedIn and mail. Questionnaires are the most common technique of data collection in the domain of SSTs, moreover it allows to retrieve information on perceptions of SST users.

Additionally, this research involves three independent experiments. The first experiment focusses on interactivity, the second on vividness and the third on content quality. Each of these experiments independently examines two different conditions: high interactivity vs. low interactivity, high vividness vs. low vividness, high content quality vs. low content quality.

In order to examine the two conditions of each individual variable (interactivity, vividness and content quality), six groups of respondents will be asked to view different pictograms. As said before this study is limited to internet-based SSTs. More specifically it involves webshops of online retailers. Therefore, the pictograms are graphical representations of six webpages showing a webshop with different levels of interactivity, vividness and content quality (see appendix B). I will not use actual websites since it would be too difficult to manipulate one SST feature while holding other variables constant. Subsequently, a static group comparison design will be used in which six conditions will be measured:

 Treatment group A: Graphical representation of a highly interactive website.  Control group B: Graphical representation of non-interactive website

(26)

Figure 3: Static group comparison design

(vividness and content quality held constant).

 Treatment group C: Graphical representation of a highly vivid website.  Control group D: Graphical representation of non-vivid website

(vividness and content quality held constant).

 Treatment group E: Graphical representation of a high content quality website.  Control group F: Graphical representation of a low content quality website

(interactivity and content quality held constant).

Figure 3 shows a schematic overview of how a static group comparison design works. Group A will be undergoing a manipulated treatment (X), whereas group B controls whether the treatment leads to significantly different outcomes (O). This set-up allows to consequently measure time and effort perceptions after different treatment of dominance in SST features.

3.2 Sample

The population of this study consists of internet-based SST users. This is a large and relatively young population. Most people between 18 – 50 years have visited self-service websites. Obtaining a random sampling frame is practically impossible. Therefore this thesis has used non-probability sampling. More specifically, data was gathered according to the convenience sampling technique and snowball technique. Although this technique is prone to biases such as overrepresentation, it is the most practical manner to gain a large sample size in a short timeframe.

Between November 12th_{and November 14}th_{a pilot study was conducted. This pilot} study involved 10 respondents (7 male and 3 female) of which the age ranged from 21 to 57 years. Based on the feedback received by these preliminary respondents some aspects of the survey were adjusted. For instance, an URL-link to the graphic of the website was placed on each page of the survey. This allowed respondents to easily take another look at the graphic of the site. Following this pilot study, actual data was collected between November 15th _and November 25th_{of 2015. On November 25}th_{the surveys were closed due to the fact that the}

(27)

number of daily respondents had reduced to 5% of the highest number of daily respondents. This indicated that the total amount of respondents would probably not become much higher over time. On the 25th_{of November 2015 a total of 310 respondents had been reached.} Subsequently 94 of these respondents had to be deleted due to incompletion of the survey which equals a drop-out rate of 30.3%. Overall, this resulted in 216 valid responses which were equally divided over the six conditions, resulting in 36 respondents per condition. Therefore, during the statistical analysis a total of n = 216 (n = 36 per condition) has been used.

3.3 Measurement items

As said previously, this study involved three independent experiment. One experiment focussed on interactivity, the second on vividness and the last experiment dealt with content quality. Subsequently, each experiment tested two conditions. A participant observed either a graphical representation of a highly interactive website or a non-interactive website, etc. Therefore, all in all six conditions were tested.

Consequently, six surveys were designed. Each survey asked participants for demographics such as gender (nominal), age (ratio) and education (ordinal). Additionally, different constructs were measured according to existing measurement items in prior research (Liu, 2003; Kelley et al., 1989; Aladwani & Palvia, 2002; Mohr & Bitner, 1995; Jiang et al., 2013; Seiders et al., 2007; Kleijnen et al., 2007; Van der Heijden & Verhagen, 2004). All items were tested based on a 7-point Likert-scale. The surveys are written in English. However, some complex English words were translated to Dutch in order to facilitate homogeneos interpretation of the respondents. An overview of the used measurement items can be found in Appendix A. The remaining part of this subsection will zoom in on each of the surveys.

 Survey 1 & 2: The first two surveys focus on respectively high and low interactive websites and their subsequent effect on participants perceptions of time and effort expenditure. In order to measure the construct of interactivity a list of measurement items was adopted from a prior research by Lui (2003) (see Appendix A). These items needed slight adjustments because participants in this experiment were not actually visiting real website, rather they were observing a graphic of a website. Therefore the questions needed to be explicitly referred to participants’ perceptions regarding the graphic. Furthermore, Appendix B shows the surveys as presented to the respondents. Respondents filling out surveys 1 or 2 were asked to answer the statements of part 3b (Appendix B) about interactivity before continuing to part 4 of the survey.

(28)

 Survey 3 & 4: These surveys are designed in order to measure the effect of vividness on perceptions of time and effort expenditure. In survey 3, participants were asked to observe a highly vivid visual representation of a website. Contrary, in survey 4, participants were asked to observe a lowly vivid graphic of a website. In order to test the construct of vividness a list of measurement items was adopted from Kelley et al. (1989) (see Appendix A). Here as well, items needed to be slightly adjusted because of the fact that respondents observed a graphic of a site rather than an actual website. Subsequently, respondents were asked to filled out part 3b (Appendix B) about vividness before continuing to part 4 of the survey.

 Survey 5 & 6: The final two surveys focus on the measurement of content quality. In order to measure this construct, measurement items by Aladwani and Palvia (2002) were used (see Appendix A). Similar to the previous surveys, these items needed slight adjustments since participants observed a graphic of a website. Consequently, respondents of surveys 5 or 6 were asked to fill out part 3c (Appendix B) in order to measure content quality before they could continue to part 4 of the survey.

3.4 Strengths and limitations

A strength of this study is the quantitative nature which allows statistical analysis of big sets of data. Also by using a questionnaire, perceptions of participants can fairly accurately be measured. Furthermore, incorporating an in-between quasi-experimental design with different conditions of SST features permits the possibility to test the pure effect of specific features (interactivity, vividness and content quality) on subsequent time and effort perceptions. Lastly, by including demographic items in the questionnaires moderating effects of gender, age and education can possibly be tested. On the other hand there are some limitations to the research design. The data collection technique being used is non-probability convenience sampling. Therefore conclusions about the data cannot be easily generalized across the population since there is no random sampling. Also, graphic representations of sites are used rather than actual sites. Consequently, participants’ experiences might be different when visiting actual websites. Furthermore, data may be skewed because convenience sampling does not allow control over who participates. Therefore certain groups may be under- or overrepresented. Finally, by using a static group comparison design the perceptions of respondents cannot be checked before treatment.

(29)

4. Results

In this chapter the results of this research will be presented. This will be done in five sections. The first section will examine the reliability and validity of the obtained dataset. Additionally, the second section provides a general correlation matrix of the used variables according to Pearson’s correlation coefficient. Section three tests hypotheses 1, 2, 3, 4, 5 and 6 by conducting a one-way between-subjects ANOVA. The fourth section checks hypotheses 7 and 8 based on a linear regression. Lastly, section five empirically examines hypotheses 9, 10, 11 and 12 through looking at the interaction effects within a 2x2 factorial ANOVA.

4.1 Descriptive statistics and reliability

Before any conclusions can be drawn an adequate check of the collected data has to be conducted. Within this section I will check frequencies as well as reliability, skewness and kurtosis. In addition, scale means will be computed and counter-indicative items recoded.

First off, in Table 1, an overview is presented of the frequencies, means, standard deviations, skewness and kurtosis per variable within this research. As can be seen, there are three different groups constructed. A total of n = 72 participants have been involved in either one of the three independent variables (interactivity, vividness or content analysis). However, participants of every group have indicated their perceptions on time and effort expenditure following observation of the graphical representations of either interactive, vivid or high content quality websites. Therefore, the score of n for these variables are 216.

Additionally, the skewness and kurtosis scores are presented within Table 1. A rule of thumb concerning these measures is that both skewness as well as kurtosis have to be between -1.96 and 1.96 (Field, 2000). According to the scores presented within Table 1, none of the variables exceeds these levels. However, there is some overrepresentation within the gender variable as can be concluded from the high level of kurtosis, K = 1.80. To be precise, within this sample, the amount of males (n = 132) is higher than the amount of females (n = 84). The consequences of this will be discussed when using this gender variable during the analyses of hypotheses 9 and 10.

(30)

Furthermore, before moving on to the statistical analyses, the measurement items were tested on reliability. This was done by checking the Cronbach’s Alpha of all the variables. Cronbach Alpha scores can differ from 0 to 1 and generally a scale can be considered reliable if α > 0.70 (Field, 2000). Meaning that, the used analysis procedures should yield consistent findings (Saunders et al., 2009). Table 2 provides an overview of the Cronbach Alpha scores of the used items within the research.

As can be seen in Table 2, the Cronbach Alpha scores for every construct measured within this research exceeds the generally accepted level of α > 0.70. Therefore, variables seem to have been measured accordingly.

Additionally, scale means of every variable were computed and counter-indicative items were recoded. Some used items namely involved reversed questions in order to reduce response bias. The reversed items can be found in Appendices A and B.

(31)

4.2 Correlation matrix

Within this next section a matrix will be presented based on Pearson’s correlation, r. The purpose of including this matrix is to get a general idea about the used variables and see how they are individually correlated.

The correlations matrix was constructed using SPSS. The results are shown in Table 3. Correlations between interactivity, vividness and content quality cannot be computed because these were not tested within the surveys. As said previously, this research was conducted in order to measure pure effects of interactivity, vividness and content quality on time and effort perceptions. Therefore, two surveys examined interactivity effects, the second set of surveys examined vividness and the last set examined content quality effects. For instance, participants within the interactivity experiment were not asked for their perceptions on vividness and content quality.

As shown in the correlation matrix of Table 3, the correlations between time and effort perceptions on interactivity, vividness and content quality, are all significantly negative correlated. Meaning that high scores on perceived time and effort expenditures are associated with low scores on interactivity, vividness and content quality are decreasing. Or, if reversed, when interactivity, vividness and content quality scores are high, the associated time and effort perceptions scores are low. However, this does not imply causality.

Though the strong negative correlations between perceived time and effort expenditure on one hand, and interactivity, vividness and content quality on the other is not the only observation which can be derived from this matrix. Additionally, can be observed that SST

(32)

adaptation is negative and significantly correlated with both time and effort expenditure,

r(216) = -.75, p < .05 and r(216) = -.71, p < .05. This indicates that, based on Pearson’s r,

adaptation of SSTs is increasing significantly when perceived time and effort expenditures are decreasing.

4.3 One-way ANOVA

In this section hypotheses 1, 2, 3, 4, 5 and 6 will be checked. This will be done in three subsections. The first subsection will dive into hypotheses 1 and 2 which concern interactivity. The second subsection empirically examines hypotheses 3 and 4 which are focussed on vividness and, finally, the last subsection will test hypotheses 5 and 6. These last two hypotheses concern the relationship between content quality and both time and effort perceptions. All hypotheses within this section will be tested by conducting a one-way in-between subjects ANOVA.

4.3.1 Hypotheses 1 & 2

Hypothesis 1 deals with interactivity and perceived time expenditure. The null-hypothesis is formulated as follows: H0= There is no significant difference between perceived time

expenditure for different levels of interactivity. In addition, hypothesis 2 focusses on

interactivity and perceived effort expenditure. The accompanying null-hypothesis states: H0=

There is no significant difference between perceived effort expenditure for different levels of interactivity.

In order to examine whether these null-hypotheses can be rejected, a one-way ANOVA analysis has been conducted using SPSS. More specifically, the ANOVA examined whether there was a significant difference between the retrieved scores on perceived time and effort expenditure for both the treatment group (1) and control group (2). In this experiment, the treatment group observed a graphical representation of an interactive website. Contrary, the control group observed a graphical representation of a non-interactive website.

Results of the one-way ANOVA show that the differences in perceived time expenditures between the treatment group (n = 36, M = 4.38, SD = 1.63) and control group (n = 36, M = 4.62, SD = 1.14) are not statistically significant, F(1,70) = 0.506, p = .479, η2_{= .007.} In other words, participants who observed a graphical representation of a website which was non-interactive had no significantly different perceptions regarding their time expenses then participants who observed a graphic which was interactive. Additionally, Levene’s test is found to be significant, F(1,70) = 6.37, p = .014, therefore the assumption of equally distributed

(33)

variances is violated. However, because the overall model is not significant this becomes irrelevant. Overall, based on the data collected within this research, the first hypothesis cannot be rejected.

Results of the second conducted one-way ANOVA show the differences in perceived effort expenditure between the treatment group (n = 36, M = 4.03, SD = 1.19) and the control group (n = 36, M = 4.10, SD = 1.18). Here again, the treatment group observed a graphic of an interactive website whereas the control group observed a graphic of a non-interactive website. Based on the output, this model is not significant, F(1,70) = 0.54, p = .818, η2_{= .001.} Meaning that participants within the two conditions did not have significantly different perceptions regarding effort expenditure. Furthermore, Levene’s test indicated that it can be assumed that variances are equally distributed, F(1,70) = 0.135, p = .714. Concluding from this analysis, hypothesis 2 cannot be rejected. For overview, all statistical levels are portrayed in Table 4.

Moving on to the next two hypotheses, both of these involved the relationship between vividness and both perceived time and effort expenses. Hypothesis 3 is formulated as follows: H0= There is no significant difference between perceived time expenditure for different levels

of vividness. Hypothesis 2 on the other hand, examined differences in effort expenditure. H0=

There is no significant difference between perceived effort expenditure for different levels of vividness.

Just as in the previous subsection, a one-way ANOVA has been conducted in order to check whether there are significant differences between the treatment group (1) and control group (2). However, in this experiment the treatment group observed a graphical representation of a vivid website rather than an interactive one. The control group observed a pictogram of a non-vivid website.

The output shows that there is a difference between the treatment group (n = 36, M = 4.12, SD = 1.11) and the control group (n = 36, M = 4.92, SD = 1.25) which is statistically

(34)

significant, F(1,70) = 8.308, p = .005, η2_{= .10. In other words, participants who observed a} graphical representation of a website which was non-vivid had significantly different perceptions regarding their time expenses then participants who observed a graphic which was vivid. Additionally, Levene’s test is found not to be significant, F(1,70) = 0.67, p = .41, therefore the assumption of equally distributed variances was not violated. All in all, based on the data collected within this research, the third null-hypothesis is rejected.

Because this experiment involves two conditions, a post-hoc test cannot be performed. However, we can take a look at the plot of the means (Figure 4). According to this plot, the average level of perceived time expenditure is lower for participants who observed the non-vivid website (control group) than for participant who observed the vivid website (the treatment group). As been concluded by the ANOVA this difference is significant. Additionally, this has been presented in Table 5.

Figure 4: Means plot of perceived time expenditure for both control group and treatment group vividness.

The second analysis within this subsection involves possible differences in perceived effort expenditure for different levels of vividness. Results of the conducted ANOVA show there is a difference between the treatment group (n = 36, M = 3.84, SD = 1.16) and the control group (n = 36, M = 4.67, SD = 1.21) which is statistically significant as well, F(1,70) = 8.824, p = .004, η2_{= .11. This shows that, based on the data retrieved, there is a significant difference in} perceived effort expenses for participants who observed a vivid website (treatment group)

(35)

and participants who observed a non-vivid website (control group). Furthermore, Levene’s assumption is not violated, F(1,70) = 0.402, p = .52.

Again, only two conditions were examined. Therefore, a post-hoc analysis is not possible. However, the means plot helps to visualize the phenomenon that is occurring (Figure 5). As this plot shows, the treatment group had a lower mean of perceived effort expenditure, illustrating that when participants in the treatment group observed a vivid website, they subsequently had lower perceptions regarding their effort expenses. All in all, data retrieved in this study indicates that hypothesis 4 is rejected.

Figure 5: Means plot of perceived effort expenditure for both control group and treatment group vividness.

The last two hypotheses of this section involve the implications of content quality. The accompanying null-hypotheses for these are: H0= There is no significant difference between

(36)

perceived time expenditure for different levels of content quality and H0= There is no significant

difference between perceived effort expenditure for different levels of content quality.

Similar to the previous two experiments, participants were equally divided over two conditions. The first condition (treatment group) were asked to observe a graphical representation of website which was manipulated in such a way that it portrayed high levels of content quality. On the other hand, participants within the second condition (control group) were asked to observe a pictogram of a website with low-content quality.

The results from the conducted one-way ANOVA show that differences in perceived time expenditure for the treatment group (n = 36, M = 4.30, SD = 1.50) and the control group (n = 36, M = 4.53, SD = 1.39) are not statistically significant, F(1,70) = 0.451, p = .504, η2_{= .006.} Meaning that, participants who observed a pictogram of a high content quality website did not significantly differ in their perceptions regarding their time expenditure than participant who observed a pictogram of a low content quality website. Therefore, based on the used sample, hypothesis 5 cannot be rejected.

The second experiment within this subsection involved perceptions regarding effort expenses. The one-way ANOVA analysis shows that differences between the treatment group (n = 36, M = 3.94, SD = 1.31) and the control group (n = 36, M = 4.27, SD = 1.20) are not statistically significant, F(1,70) = 1.229, p = .271, η2_{= .017. In other words, based on this study,} there is no significant difference in perceptions on effort expenses for different levels of content quality. Therefore hypothesis 6 cannot be rejected. Finally, Table 6 provides an overview of all the statistical levels within this subsection.

4.4 Multiple linear regression

Within this next section hypotheses 7 and 8 will be empirically examined. This will be done through a multiple linear regression. Hypotheses 7 and 8 test the relationship between perceived time and effort expenditures and adaptation of SSTs. Based on literature it is assumed that a decline in perceived time and effort expenses will result in an expansion of

Building convenient self-service technologies : a closer look at internet-based SST design features