Switching costs: A good strategy for firms?
A study about the effects of switching costs on customer’s emotions and
their perceived value about a firm.
Fleur Bastin
MASTER THESIS
MSc Marketing
Management & Intelligence
2
Switching costs: A good strategy for firms?
A study about the effects of switching costs on customer’s emotions and
their perceived value about a firm.
By: Fleur Bastin
University of Groningen Faculty of Economics and Business
Department of Marketing
MSc Marketing Management & Intelligence
June 2018
Supervisor: A. Bhattacharya
Second supervisor: J. van Doorn
C. H. Petersstraat 53a 9714 CJ Groningen
3
ABSTRACT
Customer loyalty is a central topic in marketing nowadays and customer retention is therefore an essential goal that firms want to achieve. Switching costs can lead to customer retention and is therefore often used by firms as a strategy to retain loyal customers. However, they can also generate negative consequences for the firm. Therefore, this study aims to provide more insights into switching costs from a customer’s perspective. More specifically, how they can influence customer emotions and their perceived value about a firm. In order to gain insights into customer’s emotions with regards to switching costs, text analytics with the program LIWC is performed on 91 individual customer narratives that discuss their reasons for (not) switching, in the telecommunication industry. In addition, a second study tests the effect of switching costs on the perceived value of customers with data from the American Customer Satisfaction Index. By combining both studies, this research aims to show that the effect of switching costs goes
through customer emotions. Different customer groups are identified with regards to their
perceived switching costs, which could provide as a basis for customer segmentation. Furthermore, a moderation effect of the type of industry has been taken into account, with differences between product and service industries. It was found that positive switching costs are related to positive emotions but not to negative emotions. In addition, they lead to an increase in customer’s perceived value about a firm. Consequently, switching costs can be seen as a viable strategy for firms to retain loyal customers, however, they should be implemented cautiously.
Keywords: Customer loyalty, retention, switching costs, emotions, perceptions of value,
4
PREFACE
In front of you lies my master thesis: ‘Switching costs: a good strategy for firms?’. I have been working on it from February until June 2018 as part of the final phase in the master Marketing Management & Intelligence at the University of Groningen. Even though I am glad that this journey now comes to an end, I enjoyed writing this thesis, diving deeper into this interesting topic and applying the skills that I have learned during my master. I learned about text analytics while doing this research, which was new to me. I am happy that in the end of my master I still got to learn a new tool which could maybe be useful in my future career.
I would like to thank my supervisor Abhi Bhattacharya for his excellent guidance during the process and for introducing me into a really interesting topic within customer loyalty: switching costs. Also, I want to thank him for being so involved in this project and for always providing me with valuable feedback. In addition, my thesis mate Johann needs a special thanks for the nice cooperation, all the nice coffees we had together and for always listening to my thesis struggles. Moreover, Harish was always helpful when I encountered struggles with the data, so thank you for that. Furthermore, I want to thank Jenny van Doorn in advance for reading and evaluating my thesis. Finally, I want to thank my family and friends for supporting me during this process.
I hope that you will enjoy reading my thesis! Fleur Bastin
5
Table of content
1. Introduction 6 2. Theoretical background 9 2.1 Switching costs 9 2.1.1 F/R/P 102.1.2 Benefits and costs 11
2.2 Emotions 12
2.3 Loyalty 14
2.3.1 Customer classifications 15
2.3.2 Perceived value 16
2.4 Industry (service versus product industries) 17
2.5 Time 19
2.6 Hypotheses and conceptual model 19
2.6.1 Study 1 20
2.6.2 Study 2 20
3. Research design 22
3.1 Study 1 22
3.1.1 Data 22
3.1.1.1 Human coding of the narratives 22
3.1.1.2 LIWC 23 3.1.2 Variables 24 3.1.3 Descriptive statistics 25 3.1.4 Methods 26 3.2 Study 2 27 3.2.1 Data 28 3.2.2 Variables 30 3.2.3 Descriptive statistics 31 3.2.4 Methods 31 3.2.4.1 Moderation 32 4. Results 33 4.1 Study 1 33
4.1.1 Financial, relational and procedural switching costs 33
4.1.2 Benefits and costs 34
4.1.3 The direct effect of switching costs on emotions 35 4.1.4 The direct effect of benefits and costs on emotions 36 4.1.5 The direct effect of benefits and costs on emotions per group 37
4.2 Study 2 39
4.2.1 The direct effect of switching costs on value 39 4.2.2 The direct effect of switching costs on value per group 40 4.2.3 The moderation effect of industry type 43 4.2.4 The moderation effect of industry type per group 44 4.3 Overview of the hypotheses 46
5. Discussion and conclusion 47
5.1 Findings and theoretical implications 47
5.1.1 Study 1 48
5.1.2 Study 2 49
5.2 Managerial implications 50
5.3 Limitations and future research 52
6. References 54
6
1. INTRODUCTION
Customer loyalty is an increasingly important topic of research in the field of marketing, especially since markets are becoming more competitive nowadays (Ou, Verhoef & Wiesel, 2017; Kumar & Shah, 2004). It is, among others, a result of the perceptions of value that a customer receives from using a specific product or service (Yang & Peterson, 2004). Many studies argue that customer loyalty consists of two constructs that are both needed for loyalty: repeated purchasing of the products or services of a firm and a favourable attitude towards the firm (Dick & Basu, 1994). These constructs are called behavioural and attitudinal loyalty, where it is argued that attitude drives behaviour (Kumar & Shah, 2004). Overall, customer loyalty has a strong impact on the performance of a firm and hence it is considered to be an important competitive advantage for companies (Lam, Shankar, Erramilli & Murthy, 2004). Consequently, customer retention - defined as the repeated purchasing of a firm’s products and services over time - is an essential goal for firms to achieve. Therefore, past literature has focused on identifying factors that can lead to customer retention. Switching costs are one of the main drivers of customer retention (Cullen & Shcherbakov, 2010). Switching costs can be defined as the costs or barriers that customers associate with the process of switching to another provider (Burnham, Frels & Mahajan, 2003). This can create a so called ‘lock-in’ effect, where customers feel hindered from switching between firms (Farrel & Klemperer, 2007).
7
for customers who feel forced into a relationship with a firm that it actually prefers not to be in (Jones, Reynolds, Mothersbaugh & Beatty, 2007). Nevertheless, several studies argue that switching costs from a customer’s perspective can consist of both positive and negative dimensions (Colgate & Lang, 2001; Jones et al., 2007). Positive switching costs refer to barriers that stimulate customers to willingly stay in the relationship, such as the advantages that they get from their relationship with a firm. Negative switching costs give customers a feeling of being locked in to their relationship with a firm. From a managerial perspective, it is important to understand how switching costs are perceived by customers, because this allows the identification of customers that are willingly staying or that feel locked in and actually want to end the relationship. This information may be helpful for firms with regards to their resource allocation (Roos & Gustafsson, 2007); they should invest more time and effort in customers with negative switching costs that are likely to spread negative WOM or to churn in the future, and less in customers that are already satisfied in their relationship with the firm and have positive reasons for staying (Neslin, Gupta, Kamakura, Lu & Mason, 2006).
Thus, perceived switching costs can be different for each customer and they will arise with a customer’s feelings about its relationship with a firm. According to the appraisal theory, emotions are derived from a customer’s evaluations about their circumstances (Ellsworth & Scherer, 2003). Hence, it could be said that switching costs can generate positive or negative emotions, depending on how they are perceived by customers (Jones et al., 2007).
Accordingly, this study aims to provide managers and academics with new insights on how to understand switching costs from a customer’s perspective and how they affect the emotions that arise from those switching costs. In addition, it studies the effect of switching costs on the perceived value that customers have about a firm – which in turn leads to loyalty. Thereby, it tries to prove that the effect of switching cost on loyalty, goes via the emotions felt by customers. It investigates several key questions: Which type of switching costs are perceived positively or negatively by customers? Do switching costs generate either positive or negative emotions and thereby increase/decrease customer loyalty? Can customers be segmented with regards to their perceived switching costs?
8
on customer loyalty goes via emotions. Hence, analysing emotions may be introduced as a relevant tool for firms that are concerned about switching costs and customer loyalty.
In addition to that, perceptions of switching costs do not only differ per customer, the degree of switching costs also differs per industry (Jones, Mothersbaugh, & Beatty, 2002). According to the de Ruyter, Wetzels & Bloemer (1998), the costs of switching are higher for services than for goods. This is mainly caused by the fact that services are based on interactions between the customer and firm, where a bond is created between them. This bond between firm and customer can lead to increased feelings of loyalty (Yim, Tse, Chan, 2008). As switching costs are discussed to generate customer loyalty - the relationship between these two variables is expected to be higher in service industries. Furthermore, Jones et al. (2002) found that the relationship between switching costs and purchase intentions differs per industry. Hence, the following question arises: Does the relationship between switching costs and customer loyalty differ per industry? More specifically, is the relationship between switching costs and customer loyalty higher in service industries?
This also contributes to existing literature, because past literature argues that switching costs are higher in service than in product industries, however, it has not yet been studied how this influences the relationship between switching costs and loyalty. If the effect of switching costs on customer loyalty differs per industry, this could generate important insights for managers that operate in either one of those industries, on how effective it is to use switching costs as a strategy to retain loyal customers.
9
2. THEORETICAL BACKGROUND
This chapter consists of an extensive discussion of interesting and relevant literature regarding the topic of this study. Important findings in this topic of research are addressed and provide as background information for two conceptual models. All variables in the models are discussed and their related hypotheses are presented.
2.1 Switching costs
As mentioned, firms are extremely concerned about customer retention and loyalty nowadays (Yang & Peterson, 2004; Caruana, 2003), because they have a positive effect on a firm’s profits (Lam et al., 2004). Also, retaining existing customers is less expensive than acquiring new customers (Han, Back & Barrett, 2009). One way in which firms can achieve customer loyalty and retention is by the creation of switching costs, which has been identified as one of the main drivers of customer retention (Jones, Mothersbaugh & Beatty, 2000). Switching costs are the perceived costs of a customer that will prevent him/her from moving to an alternative supplier (Yanamandaram & White, 2006). Switching cost consist, among others, of sunk cost investments made by customers that need to be given up and possibly created again at the new supplier (Farrel & Klemperer, 2007). Hence, by generating switching costs, firms create a so called ‘lock-in’ for its customers (Farrel & Klemperer, 2007). This ‘lock-in’ identifies a situation in which a customer feels bound to or forced into its relationship with the firm (Harrison, Beatty, Reynolds & Noble, 2012). In that case, the costs of switching to another provider will outweigh the benefits that customers can gain from switching. This phenomenon might be desirable for the firm, because it makes customers continuing the relationship with the firm and thereby making them loyal customers (at least in their behaviour). However, it should be noted that switching costs can also create negative effects for the firm.
As previously stated, when customers have the feeling that they are ‘locked in’ a relationship in which they are not happy or satisfied, this might lead to behaviours that have negative consequences for the firm, such as negative WOM (Haj-Salem & Chebat, 2013). These customers will experience negative switching costs, because they feel constrained from switching between providers. They will express ‘false loyalty’ rather than ‘committed loyalty’ (Yang & Peterson, 2004); they only show loyal behaviour due to their perceived switching costs and not because of their satisfaction with the firm. On the other hand, switching costs can also be a choice made by customers that are willing to continue the relationship with a firm, for example caused by feeling connected to a specific firm/brand, which may be classified as
10
of continuing the relationship, which are also seen as positive switching costs. Thus, switching costs can be perceived both positively (+) and negatively (-) (Jones et al., 2007). Additionally, according to Edward & Sahadev (2011) switching costs are seen as either positive or negative depending on the source of constraint that creates the switching costs. In this study, positive switching costs are associated with the willingness to continue the relationship whereas negative switching costs are associated with the constraint of ending the relationship, which forces customers to continue the relationship (Hirschman, 1970).
2.1.1 F/R/P
According to Burnham et al. (2003), switching costs is a multidimensional construct and there are three types of switching costs that could be distinguished: Financial (F), relational (R) and procedural (P). Financial switching costs are the monetary costs, or the loss (-)/gain (+) in resources that are associated with moving from one firm or brand to another. For example, when you are still in a contract with a provider and you want to end this contract, you might have to buy yourself out (-). However, if benefits are received from the relationship with a firm, there are perceived positive switching costs (+). Relational switching costs consists of the emotional discomfort (-) that comes from breaking ties with a firm, due to a personal relationship with or attachment to the firm. For instance, switching between dentists, when you have been with yours for ten years already. On the other hand, a relationship with a firm may be seen as a positive reason for continuing the relationship (+). Procedural switching costs involve the time and effort (or hassle) that comes from switching to an alternative, such as learning and setup costs (-). For instance, when switching between i.e. Apple and Android, customers have to learn how to work with a new software system, which could be seen as a source of constraint. Then again, procedural switching costs can also refer to the ease of not switching (+), and thereby the benefit of continuing the relationship. These three switching costs typologies are used within this research (Burnham et al., 2003; Caruana, 2003).
11
while procedural switching costs are only seen as being negative. However, it can also be the case that a customer perceives procedural switching costs as being positive, because of the convenience of not dealing with the hassle of switching. In addition, relational switching costs can also be perceived negatively (instead of only positive). This might happen when a customer is in a relationship with a firm that belongs to a family member or friend and therefore feels uncomfortable to switch to another firm – even though he/she can perceive more benefits at another firm. For these reasons, this study challenges this point of view and argues that each type (F/R/P) of switching costs can be perceived both positively and negatively. This will be illustrated with an example of the data later in this study.
2.1.2 Benefits and costs
Now it is clear that switching costs can have both positive and negative dimensions, the reasons for switching or staying are discussed in this section. These reasons are classified in this study as benefits and costs.
Positive switching costs can provide the customer with benefits of continuing the relationship with a firm, whereas negative switching costs provide the customer with disadvantages (or costs) when ending the relationship with a firm. A distinction will be made between the benefits of staying (BS) in the relationship with the firm, the benefits of leaving (BL) the firm, the costs of staying (CS) and the costs of leaving (CL). This distinction between benefits and costs can be explained by i.e. the prospect theory, where customers tend to calculate the net utility of their choices. Applying this theory to the switching costs phenomenon, customers will thus calculate whether the benefits of switching between firms outweigh the costs of switching (Yang & Peterson, 2004). The prospect theory argues that people make their decision between alternatives in terms of the utility they will get from that decision; based on the potential gains or losses that will follow that decision. This theory can be seen as a ‘descriptive model’ of decision making under risk (Barberis, Huang & Santos, 2001), and can thus be applied to the distinction of benefits and costs with regards to switching costs in this study. If customers perceive negative switching costs they will most probably experience costs of leaving and/or costs of staying in the relationship, whereas customers with positive switching costs will perceive benefits of staying. It should be stressed that benefits (BS/BL) do not represent positive switching costs and costs (CS/CL) do not represent negative switching costs. For example, customers can experience benefits of leaving the firm, however, if the costs of leaving still outweigh these benefits of leaving, this customer will overall perceive
12
customer then actually wants to end the relationship with the firm, because he/she thinks that he/she can perceive more benefits elsewhere. However, due to the costs of switching, this customer does not switch between providers and feels locked in to the relationship that it is currently in.
2.2 Emotions
In the past, several studies have commented that it is not appropriate to ignore the emotional components that are present within the feelings of satisfaction that people experience (Yu & Dean, 2001; Han et al., 2009). However, there is not yet much literature available in the field of customer satisfaction and loyalty that also includes the effect of emotions (White & Yu, 2005). This despite the big role that they actually play in human decision making (Haj-Salem & Chebat, 2013; Watson & Spence, 2007). Emotions can be defined as the arousal that is experienced in response to a stimulation, such as an event or situation (Barrick, Hutchinson & Deckers, 1989). It is generally accepted that emotions can be distinguished into positive and negative emotions (Aviezer, Trope, Todorov, 2012; Bagozzi, Gopinath & Nyer, 1999). Emotions can play an important role in explaining customer behaviour, for example their purchasing behaviour (Kim & Lennon, 2013), which also relates to loyalty. In addition, Mattson, Lemmink & McColl (2004) found that customer emotions are important motivators for customer loyalty. Hence, it could be said that customer emotion is an interesting variable to take into account when studying customer loyalty.
13
emotions (Jones et al., 2007). These commitment types can easily be linked to positive and negative switching costs, where calculative commitment is related to negative switching costs (lock in) and affective commitment to positive switching costs (desire to stay). This suggests that negative switching costs can thus lead to negative emotions and positive switching costs to positive emotions.
As mentioned, another explanation of how switching costs can lead to emotions and behaviour is in line with the appraisal theory. This theory argues that emotions will arise from people’s perception and underlying evaluations of their circumstances (Ellsworth & Scherer, 2003). This could explain that the way in which switching costs are perceived and evaluated by someone (either positive or negative) can influence the emotions that are felt by that person. In addition, the perceived benefits and costs (of staying/leaving) can then also influence the emotions of a customer. If switching costs feel as a constraint, this will generate negative emotions, whereas if customers perceive switching costs as positive and beneficial, this will generate positive emotions. Hence, the appraisal theory can be used in order to explain the emotions that are evoked by people in different situations, however, it may also be used to explain how emotions lead to behavioural responses (Watson & Spence, 2007). Consequently, the following hypotheses are derived:
H1a: Positive switching costs lead to positive emotions.
H1b: Negative switching costs lead to negative emotions.
H2a: Benefits (of staying/leaving) lead to positive emotions.
H2b: Costs (of staying/leaving) lead to negative emotions.
14
other hand, negative emotions have the opposite effect and therefore people are more likely to switch. Overall, it can be said that positive emotions lead to both behavioural and attitudinal loyalty and negative emotions do not. Thus, if a firm observes that a person is repurchasing its product/service (behavioural loyalty), but this customer is actually only loyal because of perceived (negative) switching costs, he/she is likely to express negative emotions. This is important for firms to understand, because they should be aware that not all their observed (behaviourally) loyal customers will express positive emotions. Therefore, this study aims to prove that the effect of switching costs on loyalty goes via emotions.
2.3 Loyalty
As mentioned, loyalty is an important topic of research and is regarded as one of the most important constructs in marketing (Caruana, 2003; Kumar & Reinartz, 2016). Traditionally, loyalty has been characterized by a behavioural measure, focusing mainly on purchasing behaviour of customers (Kumar & Shah, 2004). By this way, customers that are repeatedly purchasing from a specific firm are seen as loyal customers (Ndubisi, 2007). However, many researchers have discussed the essence of considering both behavioural and attitudinal loyalty (Dick & Basu, 1994; Kumar & Shah, 2004; Cheng, 2011). For example, it is discussed that the behavioural measures of customer loyalty are not sufficient to understand the underlying factors that generate brand-loyal purchasing behaviour (Bandyopadhyay & Martell, 2007). Therefore, according to Dick & Basu (1994), who have developed a framework for customer loyalty, a favourable attitude and repeated purchasing are required for loyalty. In addition, while repeated purchasing might be seen as an expression of loyalty, it is only complete if it is complemented with a positive attitude towards the firm (Amine, 1998), because this ensures that loyal behaviour will continue in the future. Thus, it could be said that attitudes are related to behaviours and therefore should also be considered when looking at customer loyalty. Attitudinal loyalty consist of the customer’s identification with a firm and their preference of this firm over other alternatives (Cheng, 2011; Kumar & Shah, 2004). It represents a higher-order commitment towards a firm that cannot be determined by only looking at repeated purchase behaviour (Shankar, Smith & Rangaswamy, 2003). Whereas attitudinal loyalty does not ensure purchasing in itself, it could lead to positive WOM – hence a positive image for the firm. According to Cheng (2011), switching costs influence both types of loyalty, and the effects on both attitudinal and behavioural loyalty are similar.
15
(continuously purchases from the same firm), this customer might not be satisfied with the firm and therefore does not feel loyal. This customer is then not attitudinally loyal towards the firm. For instance, a customer that is unwillingly locked in a relationship is behaviourally loyal towards the firm, because he/she continues the relationship with the firm. This customer might not be satisfied with the firm, but feels forced into the relationship due to perceived switching costs. Thus, this customer does not experience attitudinal loyalty towards the firm, which might in the end result in negative WOM about that company.
There is a difference between the amount of loyalty expressed and the amount of satisfaction experienced by this customer. In this study, this is referred to as ‘unexplained
loyalty’, which is caused by a customer’s perception of switching costs. Unexplained loyalty is
defined here as the degree to which customers exhibit loyalty towards a firm that cannot be determined by their level of satisfaction with the firm. Therefore, switching costs are argued to determine the degree of unexplained loyalty. Different customer groups will be identified in this study, according to the direction of their unexplained loyalty; in other words, with regards to their positive or negative switching costs.
2.3.1 Customer classifications
Different types of repeat buyers can be classified according to their ‘unexplained loyalty’ towards a firm, ranging from ‘prisoners to apostles’, where prisoners refer to the ‘locked in’ customers and the apostles consists of extremely loyal customers (Carolyn & Karen, 2002). Hence, based on that - in this study different customer groups are identified within the observations in the data. They are identified according to their loyalty towards and satisfaction with the firm. Customers that are more loyal than expected according to their satisfaction level with the show ‘excess loyalty’ whereas customers that are less loyal than satisfied show ‘deficit
loyalty’. The customers that are equally satisfied as loyal show the behaviour that is expected
when looking at their loyalty/satisfaction. These customers are classified as the rational stayers, who thus have similar levels of loyalty and satisfaction. They make a rational decision by weighing the benefits and costs of leaving/staying, but decide that it is ultimately the best choice for them to stick with the firm. Figure 1 below illustrates this visually, where the yellow lines represent the standard deviation in which the rational stayers should still belong.
16
towards the firm. Prisoners are customers that are not satisfied but despite that remain loyal towards the firm. They perceive costs (of staying and/or leaving) and therefore are stuck in the relationship or feel constrained from leaving (negative switching costs). Prisoners show behavioural loyalty towards the firm but they are not loyal in their attitude towards the firm.
The group that perceives ‘deficit loyalty’ are churners/variety seekers. Churners are the customers who end the relationship with a firm and are not so satisfied. Variety seekers are customers who may be satisfied with a firm, but still look around for variety in alternatives (Rohm & Swaminathan, 2004). They both show a low degree of loyalty towards the firm, even though their satisfaction with the firm can still be high. These customer groups do not have switching costs and can therefore leave the firm without feeling constrained. Hence, they are not the focus in this study, so it was chosen to classify all customers that experience deficit
loyalty as churners.
Figure 1: Switching costs customers classifications. 2.3.2 Perceived value
17
perceived value of customers about products or services from a firm leads to repeated purchase and loyalty towards the firm. This suggests that perceived value is a good indicator of loyalty and as this study investigates the effect of switching costs on customer loyalty, customer perceived value is also an important variable in this study.
Whereas perceived value leads to customer loyalty, on the other hand, switching costs may also have an effect on the perceived value of a customer directly. There are different levels of switching costs that are perceived by customers, and thereby these switching costs may also be associated with different levels of perceived value (Yang & Peterson, 2004). For example, sunk costs and learning costs (which are seen as switching costs) may lead to an increase in the perceived value that customers receive from a firm. This is because these are investments that have been made/knowledge that has been built in a relationship with a firm, which is of value in that specific relationship and will not be of value anymore if the relationship is ended (Burnham et al., 2003). In addition, customers that experience positive switching costs, perceive benefits of their relationship with a firm (Jones et al., 2007), which also leads to increased perceptions of value for these customers. On the contrary, if customers perceive negative switching costs and feel forced into the relationship with a firm, they might validate their reasons to stay and thereby increase the feelings of perceived value that they gain from the product/service. This can be compared to the theory that people change their own feelings in favour of the products that they have, in order to reduce their ‘cognitive dissonance’ (Klemperer, 1995). Cognitive dissonance is described as a psychologically uncomfortable state which motivates someone to reduce that specific dissonance (Sweeney, Hausknecht & Soutar, 2000). Hence, consumers that perceive negative switching costs and feel ‘stuck’ in the relationship with a firm, will seek justification for their decision to continue their relationship and thereby reduce their own discomfort (Stone & Cooper, 2001). This justification might result in increased perceived value of customers. Overall, it is argued in this study that switching costs will increase the perceived value of a customer about a firm. Therefore:
H3: Perceived switching costs have a positive effect on the perceived value of customers about a company.
2.4 Industry (service industry versus product industry)
18
consumed (Edvardsson, Johnson, Gustafsson & Strandvik, 2000). Where services consist of activities that are typically produced during an interactive process with the customer, goods are produced before purchase and consumption(Nilsson, Johnson & Gustafsson, 2001; Edvardsson et al., 2000). During these customer-firm interactions, trust is build towards the firm. Trust is positively associated with loyalty, and the management of customer trust is important in the marketing of services (Chiou & Droge, 2006), whereas it is less associated with the marketing of products. Therefore, it could be said that trust is a more important factor in service industries than in product industries. On the other hand, trust that is built in a relationship with a firm, can be also be seen as a psychological exit barrier, thus a type of switching costs (Sharma & Patterson, 2000). This suggests that in services industries, a higher degree of switching costs are perceived. Furthermore, having a relationship with firms is nowadays more important for customers than the physical products itself (Kandampully, 1998). As services are based on interaction with customers, relationship building will take place during these interactions, which in the end leads to loyalty (Gremler & Brown, 1996). Consumers might feel more connected towards providers that they have an emotional bond with, which is more easily created during interactions than with a physical product (Vlachos, Theotokis, Pramatari & Vrechopoulos, 2010).
Moreover, industries also differ in how the satisfaction of a customer influences the repeated purchasing and overall loyalty of that customer (Lee, Lee & Feick, 2001). For example, in the airline and banking industries, customers will stick with the same firm even when the service quality is decreasing (Cheng, 2011). However, this will not be the case with i.e. supermarkets, where customers will more easily switch to another one if they are not happy with the quality of the products anymore. So, the amount of loyalty expressed by customers towards a firm is in general higher in service industries than in product industries (Edvardsson et al., 2000).Also, service providers are better able to create stronger and loyal bonds with their clients, than suppliers of goods (Gremler & Brown, 1996).
19
relational switching costs are experienced by customers when changing doctors (de Ruyter et al., 1998). Overall, it is clear that there are significant differences between firms offering services or goods and therefore it seems reasonable to assume that this may also influence the relationship between switching costs and loyalty/perceived value. According to existing literature, both switching costs and loyalty are higher in service industries than in product industries. Therefore, it is expected that the link between switching costs and perceived value (which leads to loyalty) is higher within service industries than within the product industries:
H4: The relationship between switching costs and the perceived value of customers about the company is stronger in service industries than in product industries.
2.5 Control variable: Time
Switching costs can change over time. For example, at the beginning of a relationship with the firm a customer still has to face set-up costs and learning costs. When the relationship evolves over time, the costs of switching will increase because of perceived sunk costs, which are the customer perceptions of time, money and effort already invested in a relationship (Jones et al., 2002). If a customer switches between providers, after having already invested in a relationship with a firm, switching costs are perceived higher than when the relationship is not established at a further stage. In addition, according to Lee et al (2001), different types of switching costs evolve differently over time. For example, contractual costs are fixed, but only at the time of the contract, so if the contract ends the switching cost also disappears. Procedural costs, on the other hand, are more of a reflection of the long-term relationship between a customer and firm (Lee et al., 2001), because this consists of the effort that has been put into the relationship which should be done all over again if a customer switches to a different provider. In addition, if time passes by, more competitors can enter the market. If there are more alternatives available, customers have a broader choice and therefore might perceive lower switching costs, especially if the expected gains from switching outperform the costs of switching (Pick & Eisend, 2014). Because of the aforementioned reasons, time (in years) will be considered as a control variable in this research.
2.6 Conceptual models & hypotheses
20
perceived value of customers – which eventually leads to loyalty – with data from the American Customer Satisfaction Index (ACSI). By combining both studies it is hoped to illustrate that the effect of switching costs on loyalty (study 2) goes via customer emotions (study 1).
2.6.1 Study 1
Study 1 is used to test whether switching costs have a direct effect on the emotions of customers. The perceived switching costs in study 1 consists of the financial (F), relational (R) and procedural (P) switching costs that are identified in the narratives and the benefits and costs that are observed in each narrative (BS, BL, CS, CL). In study 1, there is no variable for loyalty available, however we know from past literature that there is a relationship between emotions and loyalty (Gracia, et al., 2011), which can be explained by i.e. the appraisal theory. The relationship between switching costs and perceived value (loyalty) is then tested in study 2. The goal of study 1 is to find out whether switching costs have an effect on emotions, and if that is the case, thereby prove that the effect of switching costs on loyalty can go through emotions. This leads to the following conceptual model and hypotheses:
Figure 2: Conceptual model – study 1.
H1: Perceived switching costs have an effect on the emotions felt by a customer.
H1a: Positive switching costs lead to positive emotions. H1b: Negative switching costs lead to negative emotions.
H2: Perceived benefits and costs have an effect on the emotions felt by a customer.
H2c: Benefits (of staying/leaving) lead to positive emotions. H2d: Costs (of staying/leaving) lead to negative emotions.
2.6.2 Study 2
21
costs and loyalty. The overall value in the ACSI dataset consists of both the quality given the price paid (QP) and the price given the quality of the products (PQ). Therefore, there are derived hypotheses about the effect of switching costs on all three variables of value, to be able to take into account the differences between them. Hence:
Figure 3: Conceptual model - study 2.
H3: Perceived switching costs has an effect on the perceived value (PQ, QP, overall value) of customers about a company.
H3a: Perceived switching costs have a positive effect on the overall perceived value. H3b: Perceived switching costs have a positive effect on the price given quality (PQ). H3c: Perceived switching costs have a positive effect on the quality given price (QP).
Another topic of interest in study 2 is whether the type of industry affects the relationship between switching costs and the perceived value. As discussed in section 2.4, this relationship is expected to be moderated by the type of industry, where it is predicted that this relationship is stronger in service industries. The following hypotheses about this relationship are derived:
H4: The relationship between switching costs and value is moderated by the type of industry.
H4a: The relationship between switching costs and perceived value is stronger in service industries than in product industries.
H4b: The relationship between switching costs and price given quality is stronger in service industries than in product industries.
22
3. RESEARCH DESIGN
This chapter elaborates how both study 1 and 2 are designed. First, the goal of each study is discussed, followed by the data that is used for that study. After that, a table with an overview of all the important variables in that study and descriptive statistics of the data are provided. This is followed by the methods used to tests the hypotheses in both studies.
3.1 Study 1
This study investigates the effects of perceived switching costs on emotions. It is expected that positive switching costs lead to positive emotions and negative switching costs to negative emotions. In addition, it is predicted that benefits will generate positive emotions and costs will generate negative emotions.
3.1.1 Data
The data provided for study 1 consists of 100 individual narratives that comprise pieces of text in which people discuss the last time they wanted to switch and why they did (not). Thus, it consists of qualitative data and it is generated via a questionnaire. It was collected at a college mall at the University of Indiana in 2016 and provided by the supervisor of this study. It consists of 40 females and 60 male students, between the age 19-21.
The data is relatively unstructured, because it is not numbered or separated as individual respondents. Therefore, the data was structured first and invalid narratives – that did not answer the question or were useless - were deleted. This leaves us with a total of 91 individual narratives that can be used for this study.
3.1.1.1 Human coding of the narratives
23
Customer group Definition Frequency
observed
Relative frequency (%)
Prisoner Group of customers that perceive negative switching costs. Wants to leave but feels constrained by switching costs.
33 36.26%
Positive stayer Group of customers that perceive positive switching costs. Is willingly staying in their relationship with the firm.
24 26.37%
Rational stayer Group of customers that weighs the pros and cons and decides that staying is the best option. This group experiences similar loyalty and satisfaction levels.
11 12.09%
Churner Group of customers that ends the relationship with a firm.
11 12.09%
Fist prisoner, then churner (FPTC)
Group of customers that was stuck in the relationship due to switching costs. After SC are gone, they switch to different provider.
12 13.19%
Table 1: Customer groups study 1.
These customer groups will be used to find out the differences in perceived switching costs, benefits and costs between the different groups. For example, prisoners are predicted to perceive more costs whereas positive stayers are expected to perceive more benefits from their relationship with the firm. In addition, it was manually coded if the respondent discusses their reasons for (not) switching as benefits of leaving (BL), benefits of staying (BS), costs of leaving (CL) and costs of staying (CS), to make a clear distinction between benefits-cots, and staying-leaving. Some respondents discuss more than one type of switching reasons (e.g. benefits of staying and costs of leaving), while others clearly discuss just one reason (e.g. only benefits of staying). Furthermore, the type of switching costs, financial (F), relational (R) and/or procedural (P) (Burnham et al., 2003) that are observed in each narrative have been coded.
Coding is seen as a significant step taken during analyses, to organize and make sense of textual data (Basit, 2003), in this case the narratives. The manual coding was done by two people working with this data. After the initial coding was done, a third independent person was asked to do this as well for validation of how each narrative has been coded. Finally, after verifying each other’s work and discussing the classifications, all three agreed upon the final coded dataset which will be used for the analyses in this study. To illustrate and give the reader an idea of how the coding was done, an example of a narrative that perceives several benefits and costs of leaving/staying, who has negative overall switching costs and is coded as a prisoner, is shown below in appendix I.
3.1.1.2 LIWC
24
computerized text analysis program that categorizes and quantifies language usage within written texts (Kahn, Tobin, Massey & Anderson, 2007). It counts the amount of words within specified psychologically meaningful categories (Tausczik & Pennebaker, 2010). Research shows that LIWC can accurately identify emotions within language use (Kahn et al., 2007) and that the ratings of positive and negative emotions by LIWC correspond with human ratings of written narratives (Tausczik & Pennebaker, 2010). Therefore, it is assumed that LIWC is a reliable source in identifying positive and negative emotions in the narratives.
The program reads a piece of text and counts the amount of words within that text that reflects each different category. It compares each word in the text with a dictionary that is set by the user. If the program has read the text, it will thus calculate the percentage of words that match the categories in the dictionary (LIWC, 2015). The dictionary ‘internal LIWC 2015
master dictionary’ is used in this study. Over 100.000 text files were already analysed with this
dictionary (LIWC, 2015) and therefore it is seen as a good choice for this study as well. All individual narratives were run through the program and the scores on all the categories are given per narrative.
The total outcome that LIWC provides about the narratives consists of many different categories (such as ‘I’, ‘we’, ‘social’, ‘family’ etc.), but the ones that are used for this study are the scores for both positive emotions and negative emotions. As some narratives are longer (contain more words) than others, the scores on positive and negative emotions should be accounted for wordcount, because this makes the scores relative and thereby comparable to each other. Therefore, both the positive and negative emotions scores for each narrative are divided by the total wordcount of that narrative, which is another variable provided by LIWC. The relative scores on positive and negative emotions are included in the dataset with the manual codes. The final dataset that will be used for the analyses in this study, consists of all the variables that are derived from the manual coding and the (calculated) emotion scores from LIWC.
3.1.2 Variables
There are different variables that are used for the analyses in study 1. Therefore, an overview of all these variables is provided in table 2 below. By this way, it is easier to keep an overview about the variables that are used per study.
Variable Definition Measurement items
Switching costs (SC) The costs that consumers associate with switching to another provider
25
Financial SC The monetary costs/the loss or gain in resources that are associated with moving from one firm or brand to another.
Manually coded from the narratives. (=1 if observed, =0 if not observed) Relational SC The emotional discomfort from breaking ties
with a firm or having a personal attachment towards a firm.
Manually coded from the narratives. (=1 if observed, =0 if not observed) Procedural SC The time and effort that are associated with
switching to an alternative. Or the convenience of not switching.
Manually coded from the narratives. (=1 if observed, =0 if not observed) Positive switching costs Positive reasons for continuing the
relationship – willingness to stay.
Manually coded from the narratives (=1) e.g. positive stayers have positive SC. Negative switching costs Negative reasons for continuing the
relationship – constrained from leaving.
Manually coded from the narratives (=2) e.g. prisoners have negative SC.
Benefits An advantage or profit gained from
something.
Manually coded from the narratives → classified into benefits of staying, benefits of leaving. (=1 if observed, =0 if not observed)
Costs A disadvantage or price that needs to be paid
for something.
Manually coded from the narratives → classified into costs of staying, costs of leaving. (=1 if observed, =0 if not observed) Wordcount (WC) The total amount of words observed in a
narrative.
Provided by the text analytic program LIWC.
Positive emotions Positive arousal in response to stimulation in the environment. For example: love, nice, sweet. (Tausczik & Pennebaker, 2010).
LIWC calculated scores on positive emotions.
Relative positive emotions The observed score of positive emotions relative to the amount of words per narrative.
LIWC score on positive emotions divided by the total wordcount.
Negative emotions Negative arousal in response to stimulation in the environment. For example: hurt, ugly, nasty. (Tausczik & Pennebaker, 2010).
LIWC calculated scores on negative emotions.
Relative negative emotions The observed score of negative emotions relative to the amount of words per narrative.
LIWC score on negative emotions divided by the total wordcount.
Table 2: Variables in study 1.
3.1.3 Descriptive statistics
The dataset consists of 91 respondents. The manually coded variables are all dummy
variables and the emotion variables from LIWC are continuous variables. Furthermore, there
are no missing variables observed within this dataset.
Because most variables are dummy variables, it is not so interesting to look at how they are distributed. However, how often each switching costs type, the benefits and costs (of staying and leaving) are observed within the 91 narratives is interesting and therefore shown in table 3. This reveals that within this data, more negative switching costs are observed and financial switching costs are mentioned in almost every narrative (80 out of 91). Furthermore, the costs of staying is the reason that is discussed most by the consumers in this data, followed by the benefits of leaving and then benefits of staying/costs of leaving.
26
that the scores on positive emotions are higher than the scores on negative emotions, so in general there are more positive emotions observed in the narratives.
Variable Level Frequency Relative frequency (%) Switching costs Negative (=1)
Positive (=2)
59 32
64,84% 35,16% Switching costs type Financial
Relational Procedural 80 32 29 87.91% 35.16% 31.87% Benefits Benefits of leaving
Benefits of staying 51 44
56,04% 48,35% Costs Costs of leaving
Costs of staying
44 64
48,35% 70,33%
Table 3: Descriptive statistics of the independent variables.
Variable Mean Min Max SD
Wordcount (WC) 199.8 59 348 52.596 Positive emotions 2.405 0.00 5.940 1.308 Relative positive emotions 0.0133 0.00 0.0625 0.0105 Negative emotions 1.103 0.00 2.940 0.7785 Relative negative emotions 0.0062 0.00 0.0298 0.0055
Table 4: Descriptive statistics of the dependent variables.
3.1.4 Methods
To get the quantitative data for this study, first manual coding of the individual narratives took place. In addition, the text analytics program LIWC was used to calculate the (relative) emotion scores for each narrative. By this way, a dataset is created that makes it possible to gain statistical insights into the effects of switching costs on customer emotions.
Several statistical tests are performed in order to test hypotheses 1a, 1b, 2a and 2b. It is key to find out if switching costs and benefits/costs have a significant direct effect on the emotions of customers. However, first it was tested whether each type of switching costs (financial, relational and procedural) can be both positive and negative. This was done by performing a t-test, where the mean of each switching costs type is mapped on both positive and negative switching costs. If this shows significant results, it can be concluded that each type of switching costs (F/R/P) significantly differs from each other with regards to positive and negative switching costs.
27
rational stayers, churners and first-prisoner-then-churners) actually differ in terms of the type of switching costs expressed (F/R/P) and in terms of their perceived benefits and costs (BL/BS/CL/CS). This was done by performing a one-way ANOVA, which tests if there are any statistically significant differences between the means of independent groups (Gelman, 2005). If the means significantly differ from each other, it is suggested that the customer classifications in this study can be validated, at least in terms of their perceived types of switching costs and benefits and costs. In addition to that, a multiple comparison, more specifically Bonferroni correction, is applied in order to gain insights into how the groups differ from each other.
Furthermore, linear regressions are performed to measure the direct effect that switching costs have on customer emotions. A linear regression is an easy way to determine a direct relationship between two variables, more specifically, how a variable influences another variable. To account for the small sample size in this study (N = 91), a bootstrap sample is created with 1000 resamples, to see if this increases the statistical significance (Koehn, 2004). It involves the resampling of points from one’s own data with replacements, to create a series of bootstrap samples of the same size as the original data (Felsenstein, 1985) After this was done, the linear regressions are performed with the bootstrap function as well. Since we are more interested in the conceptual and substantive conclusions, no assumptions are tested for this model.
3.2 Study 2
The second study in this paper investigates the effects of switching costs on the perceived value of customers about companies. The main goal of this research is to find out how switching costs affect customer loyalty and if this effect goes via emotions (study 1). In study 2, perceived value is chosen as the DV from the ACSI data, because as explained perceived value is a good indicator of customer loyalty. The total perceived value in this data consists of the rating of the price relative to the quality received (= PQ) and the rating of the quality given the price paid (= QP) (ACSI methodology report, 2005). These three indicators of value (value, PQ, QP) are used as DV’s in this study, to be able to see the differences between them. It is expected that switching costs have a positive effect on the overall perceived value, PQ and QP of customers.
28
3.2.1 Data
Data from the American Customer Satisfaction Index (ACSI) is used for this part of the study. The ACSI is ‘a uniform, national, cross-industry measure of satisfaction with the quality
of goods and services available to household consumers in the United States’ (ACSI
methodology report, 2005). The dataset originally consists of a total of 2984 observations between 1983 and 2013. However, all the older (1983-1993) and newer (2006-2013) observations did not contain any information on all the variables of interest and were therefore dropped from the analyses. Each observation consists of a firm, year, and their overall corresponding satisfaction scores from its customers in that specific year, such as value, satisfaction, WOM, etc. After cleaning the data from missing values, the ‘new’ dataset consists of a total of 1054 observations between the timeframe 1994 and 2006 – this dataset is used for the analyses in study 2. The important variables from this data, that are used in this study are: the total perceived value, PQ (price given the quality received), QP (quality given the price paid), the calculated switching costs and the type of industry. The switching costs variable for this dataset has been calculated with the variables satisfaction and loyalty, that are also present in the ACSI data. However, all the variables in this data are standardized first, before this calculation and any of the analyses were done, to account for the different scales that the variables have (e.g. loyalty is measured on a scale from 0-100, QP on a scale from 0-10).
The switching costs variable is calculated by regressing the variable loyalty on satisfaction, obtaining the residuals from that regression and adding these residuals to every observation in the data. After the switching costs have been calculated by this way, each observations in the dataset is assigned to one of the classified customer groups in this study (prisoners, positive stayers, rational stayers and churners).
Residuals are the deviations between the observed value and the estimated value, so the deviations from the (satisfaction-loyalty) regression line. Therefore, this is chosen as a metric for switching costs in this study, because switching costs is defined as the ‘unexplained loyalty’, resulting from the difference between loyalty and satisfaction (refer to figure 1 for visualization).
29
towards the firm even though they might be satisfied; these customers can leave the firm at any time. All the observations that perceive deficit loyalty are classified as churners, because the focus of this study are the customers that do perceive switching costs and thus show excess loyalty (prisoners and positive stayers) – however, it should be noted that there are thus also variety seekers present within this customer group. In addition, in this study we do not have information about whether the customers actually churned or not.
On the other hand, if the residuals are higher than zero, there are perceived switching costs, resulting in the ‘prisoners’ and ‘positive stayers’. These groups have a higher amount of loyalty towards the firm than that they are satisfied with the firm, which results in excess loyalty. This type of loyalty is caused by the (positive of negative) switching costs that customers perceive, which results in that both groups stay loyal to the firm even if their satisfaction level may not be on the same level as their loyalty towards the firm. This principle and the specified customer groups related to that were also explained and in section 2.2.1.
An upper and lower bound – for assigning the observations from the ACSI dataset to the customer classifications that they belong to, according to their perceived switching costs – are calculated via the following way:
- The upper bound: mean of all residuals + (0,25 * standard deviation all residuals) → this results in the groups that perceives switching costs and shows ‘excess loyalty’: prisoners and positive stayers.
- The lower bound: mean of all residuals - (0,25 * standard deviation all residuals) → this results in the groups that do not perceive switching costs and show ‘deficit loyalty’: churners and variety seekers (but they are all assigned to churner group).
The observations that fall within the calculated upper – and lower bound, thus who’s residuals are closest to the ‘fitted’ regression line, are the rational stayers. To make a distinction between positive stayers and prisoners, so the groups that perceive switching costs, the ratio between satisfaction and loyalty has been calculated (satisfaction/loyalty). If the ratio is greater than the median of all the observations, the perceived satisfaction is high, suggesting that this observation is a positive stayers. On the contrary, if the ratio is below the median of the ratio of all observations, the observation is classified as a prisoner. As mentioned, the customers that
do not perceive switching costs are all classified as churners. Thus, the classifications in study
30
Furthermore, to test the moderation effect of the type of industry (products versus services), all the firms in the dataset are manually coded as being either a product or a service industry (= dummy). For example, supermarkets, food, cars and electronics are classified as products (= 0) and banks, insurances, restaurants and hotels are classified as services (= 1). This creates the industry dummy variable.
Customer group Definition Frequency
observed
Relative frequency (%)
Prisoner Group of customers that wants to leave but is constrained by switching costs. This group shows
excess loyalty and has negative satisfaction with the
firm.
201 19.07%
Positive stayer Group of customers that is willingly staying in the relationship. This group shows excess loyalty and has positive satisfaction with the firm.
247 23.43%
Rational stayer Group of customers that weighs the pros and cons and decides that staying is the best option. This group experiences similar loyalty and satisfaction levels.
239 22.68%
Churner Group of customers that ends the relationship with a firm. This group shows deficit loyalty.
367 34.82%
Table 5: Customer groups study 2.
3.2.2 Variables
As within the first study, there are different variables that are used for the analyses in study 2. Therefore, also for this study an overview of all the variables and is provided again:
Variable Definition Measurement item
Year Year in which the satisfaction scores have been measured/collected.
This dataset consists of observations between the years 1983-2013. Switching costs The costs that customers associate with the
process of switching to another provider
Calculated by extracting the residuals from the regression between satisfaction and loyalty.
Positive switching costs Positive reasons for continuing the relationship – willingness to stay.
All residuals higher than the upper bound (see calculations in section 3.2.1). Negative switching costs Negative reasons for continuing the relationship
– constrained from leaving.
All residuals lower than the upper bound (see calculations in section 3.2.1).
Perceived value Combination of both PQ and QP. Combined score of PQ and QP, between
0-100. Price given Quality (PQ) A rating of the price paid relative to the quality
received.
Score of customers on this variable, between 0-10.
Quality given Price (QP) A rating of quality received relative to the price paid.
Score of customers on this variable, between 0-10.
Loyalty Consists of (1) repurchase likelihood and (2) the price tolerance.
Score of customers on this variable, between 0-100.
Satisfaction Consists of (1) overall satisfaction score, (2) (dis)confirmation of expectations and (3) comparison of performance with the ideal.
Score of customers on this variable, between 0-100.
Industry type Variable that represents whether a firm is operating in the service or product industry.
Each observation in the data is manually coded (0 = product industry, 1 = service industry).
31
3.2.3 Descriptive statistics
The dataset used for the analyses thus consists of 1054 observations between 1994 and 2006, collected by ACSI. The missing values have been removed from the dataset, because they did not add value at all and without them the dataset is still big enough to derive conclusions from, providing enough statistical power to be confident of the results. After the classifications of each observation into a customer type, subsets of all separate groups are created within the dataset to also be able to do the analyses per customer group. The independent variable, switching costs, in this study is calculated with the variables loyalty and satisfaction. The dependent variables in this study are value, PQ and QP. The descriptive statistics for these variables are shown in table 7 (note: these are based on the standardized variables). Looking at this table, it can be said that the satisfaction scores are lower than the loyalty scores in this study. This suggests that there will be more observations with ‘excess loyalty’ than ‘deficit loyalty’. Looking at the customer groups, this has been proven – as the prisoners and positive stayers taken together are more than the total amount of churners. Furthermore, the scores for QP are higher than the scores for PQ, suggesting that customers rate quality higher than price.
Variable Mean Min Max SD
Satisfaction -0.0012 -3.9964 1.1482 0.9922 Loyalty 0.0204 -3.1103 1.2175 0.9944 Switching costs (residuals) 0.000 -4.3745 2.9494 0.7314 Value 0.0029 -3.7935 1.0366 0.9923 PQ 0.0086 -3.2865 1.0219 0.9780 QP 0.0256 -3.6957 0.9381 0.9736
Table 7: Descriptive statistics of the variables in study 2.
In addition, considering the dummy variable ‘industry’, there are observed 754 (71,54%) product industries and 300 (28,46%) service industries within the data. The industries are almost equally distributed among the customer groups, relative to how much each industry type is observed in the total dataset (see appendix II). Thus, both industries are (almost) equally present in each customer group. This is slightly different in the rational stayers group, because within this group, the product industry is slightly more present than average.
3.2.4 Methods
32
significantly differ from each other (Gelman, 2005). If the groups are significantly different, a multiple comparisons table (Bonferroni correction) is created to gain insights into how the groups are different from each other.
Furthermore, linear regressions are performed in order to find out how switching costs affect the perceived value of customers.Three variables from the ACSI data are chosen as an indicator of value (overall value, PQ, QP), thereby the differences across them can also be discussed. To account for the different scales that are present in the variables in this data, all the variables that are used in the analyses are standardized by using the function ‘scale’. This is necessary, because otherwise variables with a large range (e.g. 0-100) are given more weight than variables with small ranges (e.g. 0.10) (Ketchen & Shook, 1996). After this was done, the linear regressions could be performed.
3.2.4.1 Moderation
To test the moderation effect of the industry dummy variable (product vs. service) on the relationship between switching costs and perceived value, the individual effects of both switching costs and industry type on the dependent variables (value, PQ, QP) are checked first to see whether one or both have a significant effect on the DV. After that, a new interaction
variable of the independent and the moderator variable is created, by multiplying these with
each other. In this case, the interaction effect between switching costs and the industry dummy is thus created. All variables were already standardized, so both variables are on the same scale (Frazier, Tix & Barron, 2004). For moderation to be allowed, there should be checked for
multicollinearity between the variables industry and switching costs first. This is done by
looking at the variance inflation factor (VIF)-scores for each model. If these are below threshold of 5 (Leeflang, Wieringa, Bijmolt & Pauwels, 2015), the variables are not correlated and the moderation analyses can be performed. After this is done, a linear regression including the interaction variable can be performed.
33
4. RESULTS
This chapter presents the results of both study 1 and 2. First, the effects of switching costs on both positive and negative emotions is evaluated. The effects of the perceived benefits and costs on emotions will also be taken into account here. Next, the second study tests if switching costs have a positive effect on the perceived value of customers about a firm. Furthermore, the moderator effect of the type of industry will be tested – where the relationship between switching costs and perceived value is predicted to be higher in the services industry.
4.1 Results study 1
This section discusses the outcomes of the tests that are performed, thereby whether switching costs have a significant effect on emotions. More specifically, it is tested whether positive switching costs lead to positive emotions and negative switching costs to negative emotions. Also, the effects of benefits and costs (of staying/leaving) are taken into account. Several ANOVA’s are performed to see if there are significant differences among the five specified customer groups in this study 1 (prisoners, positive stayers, rational stayers, churners, first-prisoner-then-churners). In addition, linear regressions are performed to see the direct effect of switching costs and benefits-costs on emotions.
4.1.1 Financial, relational and procedural switching costs
To see if the three types of switching costs that are exhibited by the customers in the narratives of study 1 - financial (F), relational (R), and procedural (P) switching costs - differ across the customer groups, a one-way ANOVA is performed between the type of switching costs (F/R/P) and the customer classifications. The results, financial (p = 0.143), relational (p = 0.367), procedural (p= 0.908), show that the means of all types of switching costs do not significantly differ across the customer groups, but that financial switching costs differ ‘the most’ among these groups. Thus, it could be said that there is not enough variation in the data regarding the types of switching costs identified.
