Master thesis
Self-service technologies in the retail environment
– their effect on attitudes and behavior
Author:
E. C. W. Kraanen
Student number:
6 1 6 9 4 7 3
Date of Submission:
January 27, 2017
Qualification:
Master Business Administration – Marketing
Institution:
University of Amsterdam
1 | University of Amsterdam
Acknowledgements
This final work of my Master’s program at the University of Amsterdam would not exist in its current form without the help and support of a number of people. I would like to use this first page to express my gratitude to everyone who enabled the completion of this thesis. First and foremost I would like to thank Asics Europe for providing the opportunity to conduct a real-life research project. Martin Block, Jelle van Arendonk and Elseline ten Hope in particular, did everything they could to ensure the access to the necessary tools and data and to facilitate a smooth cooperation with the research locations. In turn, the different research locations were outstanding in their hospitality and support during the data collection – especially Barcelona.
Secondly I would like to thank my supervisor, Dhr. Alfred Zerres, for assigning me to the research project of Asics and trusting in my ability to successfully fulfill it. His ideas and enthusiasm allowed me to start off on the right foot and his continuous support, comments and help throughout the project and writing process of this document helped me to deliver a worthy piece of work. Last, but not least important, I am grateful for the support and companionship of my fellow researchers throughout the project – Lara Galka, Lukas Janssen and Nastie Schoenmakers.
Statement of originality
This document is written by Emma Kraanen 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,
2 | University of Amsterdam
Table of contents
Abstract
4
Introduction
5-6
Theoretical framework
7-16
Self-service technologies 7Stimuli in the retail environment 7-9
The SOR paradigm 9-10 SST presence – attitudes, behavior and emotional states 11-13
SST use – attitudes, behavior and emotional states 13-14
The role of emotional states 14-15
Technology aversion 15-16
Method
17-24
Research design and data collection
18Sample 19-20
Construct measurement 21-24
Analysis & results
25-35
Subsample
25Correlation analysis 27
The effect of the presence of SSTs 27-29
The effect of the use of SSTs 30
The mediating role of emotional dimensions 30-34
The moderating effect of technology aversion 34-35 Other relations between variables 35
Discussion of results
36-39
Managerial implications
40-41
Limitations & future research
41-42
3 | University of Amsterdam
References
44-48
Appendix
49-68
List of figures & tables
Figures
Figure 1: Conceptual model 16
Figure 2: Tourist vs. local per location 18
Figure 3: Mediation model 31
Figure 4: Conceptual model (*** p < .001, ** p < .01) 36
Tables
Table 1: Demographics 19Table 2: Factor (component) loadings 21
Table 3: TAv scale items & reliability 23
Table 4: Number of respondents per variable for research location Brussels 25
Table 5: Mean, Standard deviation and Correlations of Study Variables 26
Table 6: Notice of SSTs in the store 28
Table 7: Use of SSTs in the store 29
Table 8: Indirect effect of noticing the SST on store image 32
Table 9: Indirect effect of noticing the SST on brand image 33
4 | University of Amsterdam
Abstract
The past decades there has been an increase in the presence of self-service technologies –
‘technological interfaces that enable customers to produce a service independent of direct service employee involvement’ (Meuter, Ostrom, Roundtree, & Bitner, 2000) – in stores. Nonetheless, little is known about their actual effects on shoppers. The current study attempted to interpret the effects of self-service technologies in the retail environment, in a manner consistent with the environmental psychology approach (Mehrabian & Russell, 1974). Attitudes and behavior evoked by the presence and the use of information kiosks were examined at different locations of a sports equipment store. Store- and brand image of shoppers and their unplanned purchases were expected to be positively influenced by the presence and use of self-service technologies, through an
increased level of the various emotional states (i.e. pleasure, arousal and dominance). Results showed that a high level of pleasure or arousal reinforces the store- and brand image of shoppers, but that emotional states nor attitudes and behavior are significantly influenced by self-service technologies. This means that the (costly) introduction of these technologies should be very well considered by retail managers. Next to their managerial relevance, the findings of this study are a theoretical contribution to the existing literature on self-service technologies and the retail environment.
Keywords: self-service technology; emotional states; store image; brand image; unplanned purchases
5 | University of Amsterdam
Introduction
In order to please their customers, retailers spend millions of dollars each year on re-designing their stores (Grewal, Krishnan, Baker & Borin, 1998). The past two decades this re-design has been very much focused on technology. LEGO for example uses augmented reality technology and introduced a 3D simulation in its stores that shows the LEGO construction that is in the box (Zhu, Owen, Li & Lee, 2004; Westenberg, 2010). Rebecca Minkoff and UGG Australia introduced the digital mirror into their stores, allowing customers to try on virtual outfits (Milnes, 2015). Several supermarkets
introduced self-scanning systems that can answer to fluctuations in demand without the expensive adjustment of employee levels (Dabholkar, Babbitt & Lee, 2003; Curran, Meuter & Surprenant, 2003). Much of this technology can be categorized under self-service technology (SST), allowing customers to create a service outcome independent of direct service employee involvement, interacting with machines rather than with staff (Meuter et al., 2000). The main reasons for the introduction of SSTs in the retail environment are improving customer service, increasing customer touch points, providing information and generating revenue (Lee, 2014). However, there is little information on the exact way in which SSTs benefit retail organizations and it is still unclear whether they benefit organizations at all (Burke, 2002). So far, research on SSTs in the retail environment mainly focused on the willingness to use the SST or the preference for using SSTs over interpersonal contact (e.g. Eastlick, Ratto, Lotz, & Mishra, 2012; Gelderman, Ghijsen, & van Diemen, 2011; Simon & Usunier, 2007; Weijters, Rangarajan, Falk & Schillewaert, 2007). It for example showed that a shopper’s need for interaction negatively influences the use of SSTs in the retail environment (Lee, Cho, Xu & Fairhurst, 2010). A few studies have looked at the behaviors and attitude following the use of in-store SST devices (Lee & Young, 2013; Marzocci & Zammit, 2006; Meuter et al., 2000). These studies examined what happened to amongst others word of mouth and repurchase
intentions after a satisfying versus an unsatisfying SSTs experience in a store. Very few studies have examined the effect of SSTs on the shoppers’ perception of a store, their overall idea of a store, or ‘store image’ (Martineau, 1958; Baker, Grewal & Parasuraman, 1994). Nonetheless, the image that a store communicates is often more influential in a purchase decision than the store’s products itself (Kotler, 1973). Research has shown that first aesthetic impressions are very consistent and are likely to not only influence purchases but leave lasting impressions and influence later evaluations
6 | University of Amsterdam One study that examined the influence of SSTs on shoppers’ perceptions is that of Runyan, Kim & Baker (2012). Contrary to other SST studies, this study not only examined the effects of SST use, but also the effects of their mere presence. The setting of Runyan et al.’s (2012) study was a mall. The findings showed that SSTs in the mall environment influence shoppers’ perceived mall image. Even though SSTs are increasingly becoming a common sight in stores, so far no other studies have linked their presence to shoppers’ perceptions. Neither did a lot of studies on SSTs in the retail
environment take place in a real-life setting. For these reasons, this study will investigate how the presence and the use of SST in a retail environment affect a shoppers’ store- and brand image and their actual purchases. The study will be conducted in the field. Next to the relation between SSTs and store perceptions and purchase behavior, the role of emotional states or dimensions
(Mehrabian and Russell, 1974) will be examined. The research question of the study goes: What
influence do SSTs have on shoppers emotional states and attitudes and behavior at the point
of sales in a retail setting?
On the one hand there have been growing arguments that the pervasive installation of self-service technologies, results in a reduction in customer service and the depersonalized atmosphere (Alpert, 2008). Their presence in stores could thus negatively impact the shopping experience. On the other hand the interactivity of SST devices could elicit emotional states that will positively influence purchase rates and attitudes (Kiran, Majumdar & Kishore, 2012). With an eye on the increasing level of digitalization of our whole society the former argument might not hold for much longer. People are increasingly getting used to a computer preface (Curran & Meuter, 2007; Ernst & Young, 2016). It could very well be that SSTs in the retail environment are for various reasons adding to a positive shopping experience (Simon & Usunier, 2007; Weijters et al., 2007; Curran & Meuter, 2007). The results of this study will give new theoretical insides into the actual effects of the presence and the use of SSTs in a store. The outcomes can be especially interesting given the real-life setting and the breadth of the study. Next to a theoretical contribution, this study will make a practical contribution. The results will hold crucial information for managers and owners of retail organizations about the impact of the placement of SSTs in their stores. This information could guide them in their decisions concerning the implementation of SSTs.
In the following section the relevant literature will be discussed. After that, the method and data of the study will be presented, followed by the results of the data analysis. Subsequently, the results will be discussed and possible managerial implications, limitations and future research possibilities are presented. Finally, the conclusion of this study is offered.
7 | University of Amsterdam
Theoretical framework
Self-service technologies
The term ‘self-service technologies’ was first used by Meuter et al., (2000, p.50) who defined it as ‘technological interfaces that enable customers to produce a service independent of direct service employee involvement’. Self-service technologies (SSTs) allow customers to interact with a machine rather than with a person. SSTs can answer to fluctuations in demand without the expensive adjustment of employee levels and because of the technological interface lead to a more consistent service atmosphere independent of employees’ personality and mood (Curran et al., 2003; Hsieh, Yen & Chin, 2004). A very well-known SST is the ATM (Automated Teller Machine) or cash machine, enabling customers of a financial institution to perform financial transactions such as cash
withdrawal without the need for a human clerk. Since the introduction of the cash machine in the seventies there have been major developments in the service environment, or ‘servicescape’(Bitner, 1992). Racing technological development push retailers to accelerate when it comes to in-store service design. Different SSTs have been introduced in various service environments: self-service check-ins at airports, self-serve petrol pumps, self-scanning systems at supermarkets and
information kiosk at malls (Cho & Fiorito, 2010; Dabholkar et al., 2003; Hughes, Hilton, Little & Marandi, 2013). In the retail environment information kiosks and self-scanning systems in particular are becoming common features (Dabholkar et al., 2003). More advanced technologies are being used too, retailers are continuously looking for innovative ways of delivering service to their
customers (Bobbitt and Dabholkar, 2001; Dabholkar et al., 2003). LEGO for example uses augmented reality technology and introduced a 3D simulation in its stores that shows the LEGO construction that is in the box (Zhu et al., 2004). Rebecca Minkoff and UGG Australia introduced the digital mirror into their store environment, allowing customers to try on virtual outfits (Milnes, 2015).
This study will focus on information kiosks. Rowley and Slack (2003) define information kiosks as ‘computer workstations that are located in public concourses, and designed to provide public access to digital information and e-transactions’. According to the 9th Annual Kiosk Benchmark Study (2009) and Lee (2014) the most popular location for information kiosks is the retail store. Here they are mostly used to provide information and services directly to the customer, without
interference of a service employee (Rowley & Slack, 2003). Stimuli in the retail environment
8 | University of Amsterdam Environmental psychology studies have been various and date from back in the 70ies (Mehrabian & Russell, 1974; Ittelson, Proshansky & Rivlin, 1970; Ittelson, Proshansky, Rivlin & Winkel, 1974). Since Donovan and Rossiter (1982) first applied an environmental psychology approach to the retail environment, many researchers have followed (e.g. Bellizzi & Hite, 1992; Hirsch, 1995; Mitchell, Kahn & Knasko, 1995; Spangenberg, Crowley & Henderson, 1996; Turley & Milliman, 2000). Elements or stimuli in the retail environment can explain and predict the shopping behavior and attitudes of shoppers. They range from background music, to color, and floor plans. Music has shown significantly impact sales, arousal, time spent in the environment and in-store traffic flow (Turley & Milliman, 2000). The colors in a store impact time spent in the store, purchase behavior and pleasant feelings (Turley & Milliman, 2000; Bellizzi & Hite, 1992). Ambient scent is a more recent studied intangible stimulus and influences sales, processing time, variety seeking behavior and perceived time spent in a store (Hirsch, 1995; Mitchell, Kahn & Knasko, 1995; Spangenberg et al., 1996). All these studies showed that store elements can fundamentally impact shoppers’ attitudes and behaviors. It is therefore that manager continually plan, change, and control the tangible and intangible aspect of their stores. With the design of their store they try to create the most effective shopping environment (Kotler, 1973). The store design is a critical factor in the consideration of a shopper to enter or come back to a store (Varley, 2005). Though very often, the impact of a specific design or design change on the shopper is not fully understood (Bitner, 1992; Burke, 2002). This is all the more so now the effects of stimuli on the shopper often occur unconsciously (Epstein, 1994; Turley & Milliman, 2000). Besides impacting a shopper’s senses, the retail environment facilitates the accomplishment of a shopper’s goals. The floor plan of a store, the layout of the equipment and their design or the professionalism of the store personnel can all have a major impact on the ability of shoppers to achieve their shopping-goals. It can aid or hinder them in finding what they are looking for. Furthermore the retail environment communicates a meaning of a place and the norms and expectations for behavior in it. It also conveys the market segment of the retailer and its
distinctiveness from competitors (Groeppel-Klein, 2005; Alpert, 2008). Here, contrary to the effect on a shopper’s senses, it is the more tangible characteristics of a store that add to the shopper’s perception. Examples are the materials used in the store, the presence of artwork or technological tools (Bitner, 1992).
SSTs are stimuli in the retail environment and as such can fundamentally impact the store’s interior and the shopper’s senses. The SST can be experienced as an environmental stimulus in different ways. It facilitates the shopper in doing his shopping and helps him in achieving his goals. This can be by providing a payment service or a search tool. But as an object in space, the SST can
9 | University of Amsterdam also hinder a shopper in achieving his shopping goals. It might complicate the shopper’s way through the store. A SST can also be seen or experienced as a symbolic stimulus. It can be a signal of
modernity as it is associated with technology or it might signal an impersonal atmosphere (Alpert, 2008). The various ways in which the SST can be experienced as a stimulus in the retail environment will be discussed further in the paragraph on SST use and presence.
The SOR paradigm
How exactly environmental stimuli affect the decisions and attitudes of people in a store has been an important question in research. It has been long recognized that the shopping experience cannot be solely assessed by utilitarian benefits but that there are emotional aspects involved (Babin, Darden & Griffin, 1994). The role of emotions and affect - next to cognitive processes - has been increasingly underlined when it comes to behavior, whether in stores or elsewhere (Gutnik,
Hakimzada, Yoskowitz & Patel, 2006). There is laboratory evidence that people are being led by both a rational- and an experiential system when making certain decisions (Epstein, 1994). Emotions often drive certain behaviors even if people themselves are unaware of it (Sherman, Mathur & Smith, 1997). This can be illustrated by commonly known real-life phenomena such as conflict between the heart and the head; superstitious thinking; the ever present ubiquity of religion. Thus in a retail setting, a shopper’s interaction with a product, service or the retail environment itself can be intrinsically satisfying. Regardless of the utilitarian function of the stimuli (Holbrook & Hirschman, 1982).
One of the models that link environmental stimuli, emotions and human behavior is that of Mehrabian and Russell (1974). According to them, there are three emotional dimensions or states: pleasure, arousal, and dominance. These constitute the common core of human emotional
responses to all environmental stimuli. Pleasure is described as the degree to which a person feels joyful and ranges from extreme pain or unhappiness to extreme happiness. The emotional state of arousal can be described as an ‘alert state of mind’ and refers to the extent someone feels active - ranging from deep to moderate sleep to panic. Dominance is related to feelings of control and the extent to which an individual feels restricted in his behavior. Mehrabian and Russell link their dimensions of emotions to two generalized forms of behavior: approach and avoidance (Mehrabian & Russell, 1974). In retail, approach behaviors reflect all positive behaviors directed at a particular store such as physical approach, affiliation, exploration, shopping enjoyment, retail patronage, interactions with retail personnel, time spent in store and purchase behavior (Bitner, 1992; Sherman
10 | University of Amsterdam et al., 1997; Runyan et al., 2012). Avoidance behaviors are opposite to approach behaviors and include all negative reactions directed at a particular store (i.e. no physical approach, no affiliation, no exploration etc.) (Mehrabian &Russell, 1974). As illustrated in the previous paragraph, the retail environment is an environment full of stimuli, ranging from the music played in the background, to the color of the walls, sales personnel, information kiosk and digital mirrors (Turley & Milliman, 2000; Morimoto, 2001). It is therefore that the stimulus-organism-response model is very well applicable to the retail environment and repeatedly used to examine the effect of stimuli in the retail environment on shoppers’ behavior and attitudes (e.g. Donovan & Rossiter, 1982; Donovan, Rossiter, Marcoolyn & Nesdale, 1994; Sherman et al., 1997). In most of this research the emotional state of dominance is left out of the trinity because Russell himself argued to delete it from the original model (Russell & Pratt, 1980). Though various studies showed the importance of feelings of dominance in relation to behaviors (Seligman, 1975; Frijda, 1988; Morris, Woo, Geason & Kim, 2002). Bakker et al. (2014) even state that dominance should get more attention in the
environmental psychology. Another reason for the inclusion of dominance in the range of emotional states is the growing presence of SSTs in retail environments. They are being introduced among other reasons, to move resource inputs away from the store personnel towards the customer. This means that shoppers themselves become co-producers of a service. Co-production relates to specific tasks undertaken by customers prior to or during usage of a product or service (Lusch & Vargo, 2006). SSTs are not the only services that are influenced by the customer himself; when having a haircut the customer provides the hairdresser with input on how he would like to have his haircut and when booking an online holiday, the customer is the one filling out the search terms. In a way the customer is always co-producer. Though the extent to which the customer is active in adding to the desired outcome differs per service (Hughes et al., 2013). Once a SST is being used in the store, the level of co-production of service is increased through the transfer of task-performance from the store employee to the shopper. A shopper can check whether a certain product is still in stock or can pay without the assistance of a store employee. The fact that he can do so will give the shopper a feeling of control. Especially when he very well knows how to use the SST either because of his knowledge or because of the usability of the system (Hughes et al., 2013). Mehrabian and Russell (1974) connected dominance to feelings of control. That is, to the extent to which someone feels free and unrestricted in his behavior and thus controls his or her actions. As the SST will very probably add to a shopper’s feelings of control, this study will incorporate the emotional dimension of dominance.
11 | University of Amsterdam SSTs presence - attitudes, behavior and emotional states
The first paragraph stated that SSTs are increasingly becoming part of retail environments: there is a growing number of self-scanning systems at supermarkets and information kiosk at malls (Cho & Fiorito, 2010; Dabholkar et al., 2003; Hughes et al., 2013; Lee 2014). The SST is a new stimulus in the retail environment and can be experienced as such in various ways. A research on the consequences of the introduction of self-scanners in supermarkets showed that this SST positively impacted the overall opinion of the supermarket and the intention to patronize the store (Cho & Fiorito, 2010). This happens in different environments too: the type of office furniture of a psychologist or the apparel worn by a lawyer may influence a client’s perception about the quality of their services (e.g. Crane & Clarke, 1988; Pinto & Leonidas, 1995). So first and foremost the mere presence of a SST in the retail environment can be stimulating a shopper. Solomon (1985) described the store as a ‘visual metaphor’ for that what it has to offer. As such it can indicate to the shopper the potential and quality of service and products in the store. Thus the information kiosk as a part of the store as a ‘visual metaphor’, can convey an image and suggests the potential and quality of the store. It may be (unintentionally) dissipating a certain image about the store. Store image is described by Martineau (1958) as the way in which the shopper’s mind pictures the store, partly by its functional qualities and partly by its atmosphere of psychological attributes. The concept is of crucial importance for retailers. Shoppers use store image as an evaluative criterion in deciding which store to visit (Visser, Preez & Noordwyk, 2006). “In concentrated and relatively saturated retail markets, the position that a retailer etches out in the consumer mind is a vital element of its strategy. Customers must be given a good reason to shop with one retailer rather than another” (Varley, 2005, p. 19). In a highly
competitive and dynamic market place, a positive store image can give retail organizations competitive advantage over their competitors. The way in which information kiosks contribute to store image has found to be negative by Runyan et al., (2012). Though this was in a mall
environment rather than a retail environment. The same study also found that information kiosks in the retail environment are associated with information and newness (Runyan et al., 2012). This association with newness will presumably signal a certain level of modernity and a high level of service quality in a store (Meuter et al., 2003). So rather than a negative effect, the presence of information kiosks in a store may have a positive effect on the shopper’s store image.
12 | University of Amsterdam
SST presence & brand image
When a shopper is positively influenced by the presence of SSTs because they convey an image of service and modernity, their brand image of the retailer may also be positively influenced. Brand image has been an important construct in (retail) marketing and marketing research in the past decades (Gardner & Levy, 1955). From a retailer’s perspective a well-managed brand image is very important since it can predict shoppers’ behavior (Marks & Olson, 1981). In the long term the brand image of shoppers can significantly shape a retailer’s performance (Kim, Gon Kim & An, 2003).
It is often argued in marketing research that advertising attitude transfers to brand attitude (Gardner, 1985; MacKenzie, Lutz & Belch, 1986; Porat & Tractinsky, 2012). This comes from the assumption of a certain halo effect, in which an observer's overall impression of a person, brand, or product influences the observer's feelings and thoughts about that entity's properties (Long-Crowell, 2015). Similarly the image the shoppers perceived of the mall in Runyan et al.’s study (2012) was also assigned to the individual retailers in the mall (Chebat, Sirgy & St. James, 2006; Runyan et al., 2012). It can thus be expected that the perceptions of a shopper about the store are projected on the retail brand. A positive effect of the SSTs in the retail environment may be influencing both the store - and the brand image of a shopper.
H1b: The presence of a SST will positively impact shoppers’ brand image
Arousal & unplanned purchases
SSTs fill a space in the retail environment that was previously open. It thus affects the spatial density of a store, i.e. the amount of space available to a person (van Rompay, Galetzka, Pruyn & Garcia, 2008) and prevents a shopper’s free movement. This means that because of the presences of an information kiosk in a store, shoppers will feel as if the store is more crowded (Kim & Runyan, 2011). Several previous studies have identified relations between perceptions of the retail
environment - e.g. crowding - and the emotional state of arousal (Wakefield & Baker, 1998; Chebat & Michon 2003, Mehrabian and Russell, 1974). The perception of a crowded store can activate a shopper and cause him to feel stimulated or have a higher level of activity. Crowding is generally perceived as unpleasant in shopping situations (Michon, Chebat & Turley, 2005). This affects
shoppers in achieving their shopping goals. When shoppers get a feeling of crowdedness they reduce their shopping time to be able to get out of the crowded environment. Shoppers also deviate from their initial shopping lists and reduce their exploratory behavior when a store is very crowded
13 | University of Amsterdam (Michon et al., 2005). Crowding thus makes a shopper hasty and causes shoppers to make purchases that they did not previously plan to make, or, in other words conduct ‘impulsive buying behavior’ (Kiran et al., 2012).
H1c: The presence of a SST will positively impact shoppers’ unplanned purchases H1d: The presence of a SST has a positive impact on the shoppers’ feeling of arousal
SSTs use - attitudes, behavior and emotional states
Next to an effect of their presence, SSTs have another important way of influencing shoppers. They can be used. Within the area of environmental psychology a general system has been developed to describe various environments. At the core of this system is the concept of information rate, ‘the amount of information contained or perceived in the environment per unit of time’ (Mehrabian, 1976). The more information in the form of stimuli an observer, e.g. a shopper, must process, the higher the ‘load’ of the environment. Mehrabian and Russell (1974) refer to the load of an
environment as the combination of its novelty and complexity. The novelty refers to the degree of unfamiliarity and uncertainty. Complexity refers to the number of elements, features, or changes in an environmental setting. As SSTs are respectively new in the retail environment, their use will increase the ‘load’ of the environment. This can be experienced as positive and evoke associations modernity and high service levels (Meuter et al., 2003; Runyan et al., 2012). Similarly to when a shopper notices the presence of a SST in the store, the use of this technology may thus increase the shopper’s perceptions of the store and the brand. Actually using the SST will most likely make the shopper even more aware of the newness and modernity of a store. Therefore the effect on their attitudes is expected to be even stronger when using the SST.
H2a: The use of a SST will positively impact shoppers’ store image H2b: The use of a SST will positively impact shoppers’ brand image
SST use & emotional states
Mehrabian and Russell assume a direct connection between the environmental load and the emotional state of arousal. The higher the load, the higher a person's arousal level. So an environment that is unfamiliar, surprising, crowded, and complex will cause a person to be
stimulated, excited or ‘aroused’ (Billings, 1990). As the SST is impacting the shopper’s perception of crowding and is being related to information and newness it is very likely that the shopper will experience a higher level of arousal, especially when using it (Kim & Runyan, 2011; Runyan et al.,
14 | University of Amsterdam 2012). Thus the feeling of arousal will not only be sparked by the presence of the SST but also, or moreover, by using the SST.
H3a: The use of a SST has a positive impact on the shoppers’ feeling of arousal
When a shopper is using an SST to search for information, to order a product or to fulfill a payment, the shopper himself is adding to the respective service offered by the store - he is co-producer of the service (Hughes et al., 2013). The SST is thus aiding to the ability of the shopper to reach his or her shopping goals. As the shopper himself is the one controlling for the SST use he can experience an increased feeling of control as mentioned in the previous paragraph. This feeling of control can be connected to emotional dimension of dominance (Mehrabian and Russell (1974). It is therefore expected that the use of SSTs positively influences a shopper’s feeling of dominance.
H3b: The use of a SST has a positive impact on the shoppers’ feeling of dominance
People crave sensation and like movement and the sense of touch (Peck & Childers, 2006). SSTs are interactive and allow a shopper to experience movement and touch. This can increase their
perceived level of joy (Curran & Meuter, 2007). Mehrabian and Russell (1974) link pleasure to joy. It may therefore be that the users of a SST experience a ‘fun’ or pleasurable element in the use of an SST (Childers, Carr, Peck & Carson, 2001; Dabholkar, 1994; Dabholkar & Bagozzi, 2002). So apart from what possible consequences the use of a SST might have, the use itself can evoke feelings of pleasure (Weijters et al., 2007).
H3c: The use of a SST has a positive impact on the shoppers’ feeling of pleasure
The role of emotional states
The various emotional states of Mehrabian and Russell (1974) are influenced during a shopping experience by both the use and the presence of SSTs in the retail environment. Emotions are related to people’s mood. It has been proven that a store that induces a good mood leads to a good
shopping experience and has positive effects on shopping intentions (Swinyard, 1993). The three emotional dimensions from the MR-model can add positively and negatively to the mood of a shopper. Especially arousal can be experienced as negative. It is highly associated with the ‘load’ of the shopping environment (Mehrabian and Russell, 1976) which means that it can overstimulate a shopper with complexity and novelty. Novelty refers to the degree of unfamiliarity and uncertainty. Complexity refers to the number of elements, features, or changes in an environmental setting. A lot
15 | University of Amsterdam of those two concepts at once can be experienced as a negative feeling (Billings, 1990). On the other hand it also stimulates shoppers and gives them a feeling of excitement which can be translated into a positive mood. Pleasure is a positive feeling and is also associated with a positive mood (Swinyard, 1993, Mehrabian & Russell, 1976). The emotional state a shopper is in will reflect in his or her attitudes and behaviors. Shoppers’ who experience a high level of pleasure or arousal - caused by stimuli in the retail environment - for example exhibit higher approach behavior and have positive attitudes compared to those who do not (Donovan & Rossiter, 1982; Spangenberg et al., 1996, Wakefield & Baker, 1998). The approach behavior can be observed in an increase in purchase probabilities (Donovan and Rossiter, 1982; Kotler, 1973), willingness to buy (Baker, Levy & Grewal, 1992) and money and time spend in the store (Donovan et al., 1994; Sherman et al., 1997). Dawson et al. (1990) observed that positive store perceptions - or store image - are shaped by both the stimuli in the retail environment as well as the transient emotions induced by these stimuli. Since dominance has not been part of a lot of previous research in the retail environment there has been no measured effect of it on shopper attitudes and behavior. Though in general it is to be expected that the attitudes and behavior of shoppers are driven by emotions, that are evoked by stimuli in the environment.
H4a: The positive relation between the presence of a SST and store image, brand image is mediated through shoppers’ emotional states (i.e. pleasure, arousal and dominance)
H4b: The positive relation between the use of a SST and store image, brand image is mediated through shoppers’ emotional states (i.e. pleasure, arousal and dominance)
Technology aversion
Not every shopper will experience the presence or the usage of an SST in a store the same way. Personal traits and motivations influence the effect of SSTs in the retail environment on shopper emotions, attitudes and behavior (e.g. Weijters et al., 2007; Baltas & Papastathopoulou, 2003). There is a plausible mediating role of shopper’s characteristics varying from age or sex to personal preferences and motivations such as hedonic versus utilitarian (Babin, Darden & Griffin, 1994). It therefore is likely that in a highly technological retail environment familiarity with and preferences for technology will influence shoppers’ attitudes. Shoppers that have uncomfortable feelings toward technology and a high need for human interactions will be less positive about a retail environment that is high on technology (Curran et al., 2003). The construct of technology aversion (TAv) measures the level of antipathy towards technology tools (see method section) and can be seen as an
16 | University of Amsterdam important moderator of the relation between the presence and the use of SSTs and shopper
emotions, attitudes and behaviors. When shoppers are high on TAv, they are very unfamiliar with technology and tend to avoid technology for the lack of knowledge or interest. In this case the presence of a SST in a store might not be experienced as something positive. Neither the use of a SST, or at least to a lesser extent.
H5: TAv will negatively influence the relation between the presence of SSTs and the outcome variables of store image and brand image
H6a: TAv will negatively influence the relation between the use of SSTs and pleasure H6b: TAv will negatively influence the relation between the use of SSTs and arousal H6c: TAv will negatively influence the relation between the use of SSTs and dominance
A conceptual model of all the hypothesis is presented in figure 1.
17 | University of Amsterdam
Method
Research design and data collection
To test the hypotheses of this research a field study was conducted in cooperation with Asics Europe. This is a well-known sports brand that is operating globally. Thanks to this cooperation the opportunity was given to gather real-life data. During the period of a full week a survey was being dispersed in Asics stores in four different countries. This survey was translated in Dutch, French, Spanish and German, using back-translations to correct for errors. The stores in Amsterdam (the Netherlands), Barcelona (Spain) and London (United Kingdom) all had the common Asics look and feel. The store in Brussels (Belgium) just opened and featured a completely new look and feel, including various SSTs. This offered a unique opportunity to study different environmental effects in the retail environment. As this study captured life-time data, emotions and attitudes did not have to be recalled and were not influenced by memory error and post hoc attributions. This is contrary to most previous studies on the effect of the retail environment (and in particular the SST) on customer emotions, attitudes and behaviors, that predominantly took place in an experimental setting or used post-measures (e.g. Yalch & Spangenberg, 1990; Baker et al., 1994; Baker et al., 1992; Grewal, Baker, Levy & Voss, 2003; Runyan et al., 2012). The customers were approached while or right after they experienced certain feelings. To maximize the response rate, participants were given immediate incentives in the form of candy bars. Also, participants could win a pair of running shoes if they completed the survey, this was announced to them when they were approached for participation. The respondents were approached by the researcher either when they were finishing their purchase at the counter or when they were about to leave the store. The introduction of the researcher was always in the language of the country and went: “Hi there, could I ask you a
question? I am finishing my Master’s degree at the University of Amsterdam and currently writing my thesis. I am conducting a research on retail and it would be great if you have five spare minutes to fill out this survey. If you complete the survey you have the chance to win a pair of running shoes. Would you be willing to help?”. If the customer agreed to participate, he or she then had to fill out the survey on a tablet, supported by the researcher when desired. The fact that the researcher was always next to the participants during the completion of the survey, reduced the response bias and stimulated them to fill out the survey in a serious manner. To minimize possible acquiescence bias the researchers always emphasized that they were conducting independent research, unrelated to Asics, and that honest answers were most valuable.
18 | University of Amsterdam
Sample
The population of this study comprehends all consumers shopping at physical retail stores. The study draws its conclusions from a non-probability convenience sample, representing a subset of this particular population. The total number of respondents of the four locations was 1058 and the response rate was 70,6 percent. This is high compared to other research on the use of SST shows response rates between 49.1 percent and 24 percent (Beatson, Coote, & Rudd, 2006; Lin & Hsieh, 2006). Of the 744 responses 4 were uncompleted, these ‘missing values’ were deleted from the data set manually. The ages of the respondents ranged from 14 to 90 years, the average age being 37 and the most frequently appearing age 27 (N=740, M=36.72, mode=27). Just over half of the
respondents were male (54.6%) and 65% was local compared to tourist. Though the latter varied significantly per location. Most respondents preferred not to specify their income (29.5%). Of the
19 | University of Amsterdam ones who did the most frequently occurring income range was < €30.000 (25.9%). A considerable number of respondents completed a higher education, 36.1% obtained a Bachelor degree and 36.5% a Master’s degree (table 1).
Table 1: Demographics
Gender
N Percent Male 404 54.6 Female 327 44.2 Not specified 9 1.2 Total 740 100.0Income (€)
N Percent 0-30.000 192 25.9 20.000-60.000 157 21.2 60.000-100.000 104 14.1 100.000 + 69 9.3 Not specified 218 29.5 Total 740 100.0Education
N PercentLess than high
school degree 34 4.6 High school degree 110 14.9 Bachelor degree 267 36.1 Master’s degree 270 36.5 Doctoral degree 33 4.5 Not specified 26 3.5 Total 740 100.0
20 | University of Amsterdam
Construct measurements
The questionnaire that has been developed for this study held four different subjects; technology; body features; atmospheric elements and shopping company. To make the survey manageable for the participants the four different subject were divided over two different surveys. This means that the N varies running different test, using different constructs. The survey program1 randomized the sequence of the different surveys. The survey that captured emotional states and the SST specific questions was answered 372 times (N=372). The outcome variables of unplanned purchases, store image and brand image reappeared in every survey (N=740). All constructs, including the relevant constructs for this particular study, have been measured with either existing - or adapted existing scales. The adaptations were mostly made with regard to the longitude of the survey and are justified in this section.
Emotions
For the emotional dimensions of pleasure, arousal and dominance the original scale developed by Mehrabian and Russell (1974), adjusted by Tsai, Chang, Chuang and Wang (2008), was shortened. The improved scale of Tsai et al. (2008) consisted of seventeen items and had a Cronbach’s alpha of 0.91. The scale for the emotional states in the current study comprised a total of six adjective pairs and was semantic differential (e.g. Please indicate to what extend you feel happy versus unhappy). It had a Cronbach’s alpha of 0.85. A factor analysis (principal component) with varimax rotation was conducted to distract the different emotional states. The Kaiser-Meyer-Olkin measure verified the sampling adequacy for the analysis, the KMO value being .85 (Hutcheson & Sofroniou, 1999). The analysis yielded three factors, accounting for 59,1 %, 12,7% and 11% respectively, of the variance. The factors were extracted with a cumulative contribution rate of 82.8% to describe the participants’ emotional perceptions. The first factor comprised the emotional dimension of pleasure, consisted of
three items (e.g. unpleased, pleased) and had a Cronbach’s alpha value of .89. The second factor
comprised the emotional dimension of dominance and consisted of two items (e.g. influenced,
influential), this factor had a Cronbach’s alpha value of .67. Though this value is fairly low, it is still acceptable following Field (2013). The last factor consist of one item and holds the emotional state of arousal. As this factor only holds one item, the reliability was not tested for. Table 2 shows the
other Cronbach’s alphas. They suggest good reliability (Nunally, 1981; Field 2013).
21 | University of Amsterdam Table 2: Factor (component) loadings
Technology aversion
Two constructs that are related to shoppers’ preference or disfavor of technologies - such as SSTs in the retail environment - are ‘technology anxiety’(TA) and ‘personal innovativeness in the domain of information technology’ (PIIT) (Agarwal & Prasad, 1998). Technology anxiety is deducted from the construct of computer anxiety (Heinssen, Glass & Knight, 1987). It refers to consumers’ feeling of fear when they consider to use or actually use technological tools. In other words, technology anxiety rates the ability and willingness of consumers to use technological features like a SST
(Meuter et al., 2003, 2005; Eastlick et al., 2012). The higher a person’s technology anxiety the higher their tendency of avoiding the use of a SST (Meuter et al. 2003; Dew et al., 1984). Accordingly, for shoppers with great technology anxiety contact with store personnel, atmospherics or other store features will have a bigger positive influence on their attitudes towards the store than the presence
Rotated component Matrix
component1 2 3
Annoyed/pleased .876
Unhappy/happy .845
Unsatisfied/satisfied .810
Controlled for/in control .779
Influenced/influential .825
Aroused/unaroused .940
22 | University of Amsterdam or use of a SST. Meuter et al. (2003) found that TA is a better predictor than other demographic variables for the use of SSTs and it moreover influences the overall level of customer satisfaction. Contrary to TA, the construct of PIIT does not hold a negative weight but a positive weight related to the use of technology such as SSTs. PIIT is ‘the willingness of an individual to try out any new information technology’ (Agarwal & Prasad, 1998). Assuredly the higher a consumer’s PIIT, the lower his TA. The constructs are basically measuring opposite feelings on a similar scale: scoring high on one of the constructs automatically means scoring low on the other. The higher a shopper’s PIIT the higher the likelihood that SSTs in the retail environment will positively influence his attitudes towards the store. Whereas the higher a shopper’s TA, the effect will be reverse and his attitudes towards a store will be negatively impacted by the presence of SSTs. Therefore a new construct has been design combining the constructs of TA and PIIT. This construct was called ‘Technology
Aversion’(TAv) and (like TA) measures a negative weight. It can be seen as ‘the level of antipathy towards technology tools’. The construct holds the most important scale items of both TA and PIIT. The reliability of nine-item scale for TA, originally developed by Raub (1981), used and verified by Meuter, Ostrom, Bitner, and Roundtree (2003) has been confirmed with a Cronbach’s alpha value of .90. The four-item scale of Agarwal & Prasad’s (1998) for PIIT has a Cronbach’s alpha value of .84 and includes one reverse scale item to decrease potential acquiescence bias. The new construct of TAv is measured with five items on a 7-point Likert scale. Two of the items come from PIIT, including the reversed scale item. Three of the items were taken from the TA scale. The reversed scale item was recoded afterwards (‘reversed’ in table 3). A reliability check confirmed the reliability of the new 5-item scale with a Cronbach’s alpha of .77, shown in Table 4 (Nunally, 1981; Field, 2013).
23 | University of Amsterdam Table 3: TAv scale items & reliability
Item statistics
Mean SD N
I like to experiment with new technologies (reversed)
2.414 1.474 372
In general I am hesitant trying out new
technologies
3.199 1.972 372
When given the opportunity to use technology, I fear to damage it
3.194 1.963 327
I hesitate to use technology for fear of making uncorrectable mistakes
2.643 1.780 327
I have avoided technology because it is unfamiliar to me
2.737 1.847 327
24 | University of Amsterdam Store- & brand image
Store image has been measured on a two-item semantic differential scale based on the original 4-item scale of Baker et al. (1994), and was verified with a Cronbach’s alpha value of .77, which is acceptable (Nunally, 1981; Field, 2013). For brand image a semantic differential scale of five items was used as by Spears, Surendra and Singh (2004), verified with a Cronbach’s alpha value of .92. The original scale had a Cronbach’s alpha of .95.
SST presence and usage
The Brussels location featured two SSTs being an information kiosk on a wooden standard and one build in in the armchair of a couch in the middle of the store. To assume that differences in outcome variables are being caused by the mere presence of these tools would be overly simplifying a
complex situation. There are different variables such as day of the week, location, the weather but also the overall atmosphere and store personnel that may cause the differences. As it concerned real-life data gathering it was not possible to control for all these variables. It was thus necessary to at least (try) control the presence of the SSTs by introducing a question on whether the shopper noticed the SSTs (Appendix B, 1). The use of information kiosks was recorded by a different question (Appendix B, 2). None notice or use had the value of ‘0’ and notice or use the value of ‘1’. To be able to run the analysis with one variable for the noticing of a SST and one for the use, the noticing of the information kiosks (either one, the other or both) was computed into one variable and the use of the information kiosk (either one or both) too.
Unplanned purchases
The unplanned purchases were recorded with a question in the survey (logically following a question about whether the shopper made a purchase or not). After the shopper was done filling out the questionnaire the researcher would check the receipt at the counter and fill out what the respondent spend and how many items he/she purchased (Appendix B, 3).
25 | University of Amsterdam
Analysis and result
Subsample
The hypothesis of this study were all associated with SSTs. Thus the data relevant for the analysis was the data assembled in Brussels - Brussels was the only research location that featured SSTs. The number of respondents in this subsample was 177 (N = 321, response rate 76.6%). As the statistics program randomized the appearance of the two different surveys, the questions on emotions were answered only half of the time (N = 90). To test for the direct effect of shoppers’ emotional states on the various dependent variables the whole data set (N = 740) was being used and not the subsample as both this independent and the dependent variables were included in the questionnaire on every location. That is half of the time due to the randomization (N = 372).
*meaning the question ‘Did you use/Did you notice the SST’ was being answered with either yes or no
Table 4: Number of respondents per variable for research location Brussels
1 Noticed SST* 2 Used SST* 3 Arousal 4 Pleasure 5 Dominan. 6 Brandim. 7 Storeim. 8 Un.purch 9 TechAv N 177 177 90 90 90 177 177 32 90
26 | University of Amsterdam Table 5: Mean, Standard deviation and Correlations of Study Variables
M SD 1 2 3 4 5 6 7 8 1 Noticed SST 0.21 0.41 2 Used SST 0.10 0.30 0.20** 3 Arousal 5.24 1.60 -0.08 0.22* 4 Pleasure 5.84 0.93 -0.82 0.03 0.31** 5 Dominan. 5.34 1.11 -0.02 0.02 0.43** 0.55** 6 Brandimag. 6.18 0.80 -0.11 -0.11 0.30** 0.55** 0.41** 7 Storeimage 6.24 0.93 0.03 -0.03 0.23* 0.37** 0.40** 0.36** 8 Un.purcha. 1.44 0.50 0.22 0.22 -0.31 -0.12 -0.38 -0.26 -0.17 9 Tech. Av 3.04 1.29 0.00 0.00 0.11 0.14 0.20 0.04 0.06 -0.24
** Correlation is significant at the 0.01 level (2-tailed) * Correlation is significant at the 0.05 level (2-tailed)
27 | University of Amsterdam
Correlation analysis
An overview of the descriptive correlations of all the variables used for the analysis is presented in table 5. A first observation derived from the table is the strong significant correlation between the noticing of a SST and the use of a SST, r(175) = .20, p < .01. This means that most shoppers that noticed the screen by the couch or the information kiosk, very often also used it. The use of a SST shows a weak significant correlation with the emotional state of arousal, r(88) = .22, p < .05. Furthermore a number of strong significant correlations between the different emotional states - pleasure, arousal and dominance - can be observed. Pleasure and arousal, r(88) = .31, p < .01; pleasure and dominance, r(88) = .55, p < .01; and dominance and arousal, r(88) = 43, p < .01. There are also strong significant correlations between the emotional states and the dependent variables of store - and brand image. Both of these dependent variables correlate with each of the three
emotional states, the strongest two being store image and dominance, r(88) = .40, p < .01 and brand image and pleasure r(88) = .55, p < .01. Another observation derived from the this analysis is the strong significant correlation between the two dependent variables, store - and brand image, r(175) = .36, p < .01). Neither unplanned purchases nor technology aversion show any significant
correlations with one of the other variables in the model. The observed correlations indicate that some of the variables of this study move in the same direction. However, this does not explain which variable causes the other variable to change. In order to test the hypothesis presented in this study the relation between various variables will be examines in the next section. For all statistical tests an alpha level of .05 is used. Since such an extensive data set was at our disposition some additional tests (i.e. next to the statistical tests ran to test for the hypothesis presented in the previous section) were run in order to clarify or explain certain findings.
The effect of the presence of SSTs
Hypothesis 1a through 1d concern the various effects of the presence of a SST in a retail
environment. The presence of the SSTs cannot simply be assumed to be remarked by the shopper. Therefore the noticed presence of the information kiosks in Brussels is recorded by questioning whether the shoppers noticed the SSTs in the store (Appendix B, 1). Table 6 shows that the number of shoppers that actually noticed the SSTs is fairly low, 37 out of the 177 respondents. The
expectations were to observe an increase in shoppers’ unplanned purchases, store image and brand image when the SSTs in the retail environment were being noticed. As explained in the method
28 | University of Amsterdam section of this study, the notice of the two different SSTs was treated as one variable to test these expectations. This was notably done so to increase the reliability of the results, since the separate numbers of noticing a SST were too small for proper analysis. Several multiple regressions (enter method) were run. The first was conducted to see whether noticing the presence of SSTs – along with other store features - predicts the shopper’s store image. The independent variables presented in the regression model were all the store features the shopper could notice (Appendix A, 1). Since the N is quite high the addition of these arbitrary variables would not reduce precision – i.e. increase the standard errors – of the estimates of all the valid predictor variables. No significant regression equation was found (F(8, 168) = 0.870, p =.543). H1a, ‘The presence of a SST will positively impact shoppers’ store image’, is therefore not supported. A second multiple regression was conducted to predict brand image based on the store features noticed by the shopper. The same predictor variables were used. Here neither a significant regression equation was found (Appendix A, 2). The analysis shows that noticing certain store features, among which SSTs, did not significantly predict the brand image of a shopper (F(8, 168) = 0.941, p = .485). H1b is not supported.
Table 6: Notice of SSTs in the store
To test H1c, the relation between the noticing of SSTs and unplanned purchases, another multiple regression with the same predictors was run Appendix A, 3). The model did not proof to be significant (F(7, 24) = .627, p = .729). H1c is therefore not supported.
SST
notice
Frequency Percent Did not notice 140 79.1 Noticed 37 20.9 Total 177 100.029 | University of Amsterdam To test for H1d and to see whether noticing store features - such as the SSTs - predicts the level of arousal of a shopper, a multiple regression analysis (enter method) was conducted (Appendix A, 4). The model did not prove to be significant (F(8, 81) = .980, p =.458). As the number of people that noticed the SSTs was fairly low, the statistical power of the model may not be very high. This will be further discussed in the discussion section. A comparison of the arousal level at the different
research locations involves a far bigger sample and will have greater statistical power. Although here it is not possible to specifically account for the effect of the SST in the retail environment.
Nonetheless a one-way between subject ANOVA was conducted to compare arousal levels per location. Though a significance difference between locations was found (F(3, 368) = 9,302, p < .001) the post-hoc comparisons showed that Brussels (i.e. the location with SST presence) did not have a significant different arousal mean than the other locations. Either way, H1d ‘The presence of a SST has a positive impact on the shopper’s feeling of arousal’, is not supported.
Table 7: Use of SSTs in the store
SST use
Frequency
Percent
Did not use 159 89.9
Used 18 10.2
30 | University of Amsterdam
The effect of the use of SSTs
Table 7 shows the use of the SSTs in the store in Brussels. The number of people that actually used the SST in the store was very low (N = 18). Nevertheless the sample was used to test for the relation between the use of the SST and store - and brand image of the shoppers. A multiple regression analysis (enter method) was conducted. The use of various in-store features were presented as independent variables (Appendix B, 2) and store - and brand image as dependent variables. The regression model with store image as a dependent variable was significant (F(7, 169) = 2.078, p < .05). The R² was .079, which means 8% of the variance in store image was explained by the use of various in-store features. The use of the shoe wall and the apparel shelves were found as significant predictors of store image (Appendix A, 5). SST use was not one of the significant predictors of the model (β = .017; t(169) = .071, p = .943). H2a is not supported. The regression model with brand image as a dependent variable was not significant (F(7,169) = .521, p = .818). H2b is not supported. To test for the relation between the use of the SST in the store and the emotional
dimensions of arousal, pleasure and dominance various multiple regression analysis (enter method) were conducted. In each regression, the independent variables were the use of various in-store features (Appendix B, 2) and the dependent variables were the three emotional states (Appendix A, 7-9). The regression equation with arousal as a dependent variable was not found to be significant (F(7, 82) = 1.077, p = .386). Neither was the regression equation with pleasure as a dependent variable (F(7, 82) = 1.279, p = .271), nor the regression with dominance as a dependent variable (F(7, 82) = 1.733, p =.113). H3a - H3c are not supported.
The mediating effect of emotional dimensions
Multiple regression analysis (enter method) were run to see whether the different emotional states of the shopper predicted the dependent variables of store - and brand image and unplanned purchases. As explained in method section, emotions as well as the three dependent variables appeared in every survey (i.e. on every location). Therefore this analysis was conducted with N = 372. The model with store image as the dependent variable was proven to be significant (F(3, 368) = 64,901, p < .001, with an R² of .346 (Appendix A, 10). Thus 34.6% of the variance in store image can be explained by the different emotional states a shopper experiences. Pleasure (β = 0.457; t(368) = 8,961, p < .001) and arousal (β = 0.079; t(368) = 2.631, p < .01) are the variables that are significantly predicting store image. Dominance is not significant as a predictor of store image.
31 | University of Amsterdam The regression analysis with brand image as the dependent variable also showed emotional
dimensions to be significant predictors (F(3, 368) = 51,698, p < .001). The R² is .296, so 29,6% of the variance in brand image is explained by the different emotional dimensions. Here too pleasure (β = 0.439; t(368) = 8.442, p < .001) and arousal (β = 0.080; t(368) = 2.592, p < .01) contributed
significantly to the model and dominance did not (Appendix A, 11). Another regression analysis was run to test for the effect of the different emotional states on unplanned purchases (Appendix A, 12). This model did not prove to be significant (F(3, 368) = 1.313, p = .276).
Figure 3: Mediation model
Mediation as predicted in H4 can be tested through a series of regression analyses, which reflect the four conditions2 necessary to demonstrate mediation according to Baron and Kenny (1986). The regressions would have to demonstrate a direct effect of the independent variable on the
dependent variable and also the effect of the independent variable on the mediator (Zhao, Lynch & Chen, 2010; Field, 2013). The presence of the SSTs in the store did not predict the dependent variables (Appendix A, 1-3), nor did it predict the emotional state of arousal (Appendix A, 4). There might still be an indirect effect of the presence of SSTs on the dependent variables via emotional dimensions if there is complete mediation and all the effect on the dependent variable runs through the mediator. This is unlikely now the predictor variables of SST use did not have any significant effect on emotional states (Appendix A, 7-9). Nevertheless a mediation analysis was conducted.
2
(1) The predictor variable X must significantly predict the outcome variable Y (path c); (2) the predictor variable X must significantly predict the mediator M (path a); (3) the mediator M must significantly predict the outcome variable Y (path b); and (4) the predictor variable X must predict the outcome variable Y less strongly when included in the interaction model than when predicting the outcome variable on its own (path c’) (Field, 2013).
32 | University of Amsterdam Instead of looking for a significant relation recorded by different regression analysis (Baron & Kenny, 1986), an estimate of the indirect effect and its confidence interval was made to report the possible degree of mediation (figure 3; Field, 2013). This was done using Hayes’s (2012) Process tool in SPSS, with the subsample of Brussels, N = 90. Path a, the effect of the independent variable of noticing a SST, on emotional states was not statistically different from zero for any of the three dimensions (pleasure: t = .776, p = .440, arousal: t = .722, p = .4725, dominance: t = .166, p = .869). Path b, the effect of emotional states on store image, was statistically different from zero, but only for the emotional state of dominance, t = 2.121, p < .05 with a 95% confidence interval entirely above zero (.0123 to .383). The effect of b = .198 indicates that two shoppers who similarly notice a SST but differ by one unit in their level of dominance are estimated to differ by .198 units on store image: the higher their dominance level, the higher their evaluation of the store (Appendix A, 13). It is remarkable though that the only significant coefficient in this case was dominance, whereas in the regression that was run with the complete data set (N = 372) this was the only variable that not significantly predict store image (Appendix A, 10). There was no indirect effect of SST notice on store image either, table 8 shows that the effect of none of the emotional dimensions is statistically different from zero with a 95% BC bootstrap confidence interval. The direct effect of noticing the SST on store image, path c’, is not statistically different from zero, t = .143, p = .887 with a 95%
confidence interval from -.375 to .433. The total effect of noticing the SST and emotional dimensions on store image, path c, was not statistically different from zero, t = .377, p = .707 or between -.356 and .5230 with 95% confidence. This means there is no mediation of SST notice and store image by shopper emotions. As was already expected based on the previous multiple regressions, showing no significant relations between the independent variable and the mediator and the independent variable and the dependent variable.
Table 8: Indirect effect of noticing the SST on store image
Indirect effect of X on Y
Effect Boot SE Boot LLCI Boot ULCI
Dominance .0096 .0590 -.1030 .1565
Pleasure .0358 .0522 -.0388 .1822
33 | University of Amsterdam The same analysis was run with brand image as a dependent variable. Here the results were slightly different. Path a, the effect of the independent variable of noticing a SST, on the emotional
dimensions was not statistically different from zero for any of the three emotional dimensions again (pleasure: t = .776, p = .440, arousal: t = .722, p = .473, dominance: t = .166, p = .869). (Appendix A, 14). Path b, the effect of emotional states on brand image, was statistically different from zero, but only for the emotional state of pleasure this time. The effect of .405 indicates that two shoppers who similarly notice a SST but differ by one unit in their level of pleasure are estimated to differ by b = .405 units on brand image: the higher their pleasure level, the higher their evaluation of the brand. This effect was statistically different from zero, t = 4.115, p < .001 , with a 95% confidence interval entirely above zero (.209 to .600). There was no indirect effect of SST notice on brand image either (table 9). The direct effect of noticing the SST on store image, path c’, is not statistically different from zero, t = 1.944, p = .055 with a 95% confidence interval from -.008 to .744. Though the p-value is very close to significance. Lastly the total effect of noticing the SST and emotional dimensions on brand image, path c, was statistically different from zero, t = 2.081, p = .040 or between .021 and .912 with 95% confidence.
Table 9: Indirect effect of noticing the SST on brand image
Despite the fact that both path b and path c were statistically different from zero, no indirect effect occurred. Because a significant relation between the independent variable and the mediator of emotional dimensions was lacking, mediation was not recorded. H4a is not supported.
Indirect effect of X on Y
Effect Boot SE Boot LLCI Boot ULCI
Dominance .0046 .0377 -.0508 .1164
Pleasure .0768 .0993 -.1126 .3049
