Do proposition conditions matter and is this moderated by switching costs? : a quantitative study on the perception of customers

(1)

Do proposition conditions matter and is this moderated by switching costs?

A quantitative study on the perception of customers

Author

ing. Roy Krukkert

| 10684492

University of Amsterdam

Faculty of Business and Economics Amsterdam Business School

Executive Programme in Management Studies

Track: Marketing

Program code: 2410E001

Thesis Supervisors | prof. dr. Ed Peelen and dr. Umut Konus

Date of Submission: 30.01.2016 Version 1.0

(2)

i

January 30, 2016

I Statement of originality

This document is written by Student ing. Roy Krukkert who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

II Thanks

Writing a paper is something you put effort and motivation in. You do not only need a topic which is inspiring, you need to stay inspired. Without the help and inspiration of others this paper would never have existed. First and far most of all a big thanks goes out to Tobias de Bont who made it possible for me to participate in this education. My wife Marina Krukkert and my kids also

contributed enormously by putting up with me being away and having to study. I received great help and understanding from them.

Studying isn’t something you do on your own. Luckily I had some amazing class mates, Humberto Coenen and Sander Bredewout, who were always prepared to spare a moment to help out and discuss ideas. Doing the analysis is one of the trickiest and most important parts of building up the arguments and to come up with some sensible conclusions. The man responsible for all this is Bobby den Bezemer who was prepared to answer all my questions and to discuss all the output of the tests. And last but not least I have to thank my supervisor Ed Peelen for all his feedback, prompt responses and time to discuss all my ideas. Writing a thesis is something you have to do yourself, but without great people behind you it gets a lot harder.

(3)

ii

January 30, 2016

III Abstract

Almost all advertorial content of no-frills (budget) telco (telecom) operators in the Dutch market is focussed on transactional communication. All operators try to have the most competitive offer in the market. The focus of this study is to see if proposition conditions can actually make a difference in offer preference, since adjusting your conditions is a less costly practice than the current practice of giving heavy subsidies on handsets. In academic literature this thesis contributes to the field of switching cost. Specifically in supporting the view of switching costs being a way to attract customers by lowering these costs (Jones, 2000), instead of being a way to keep customers in by introducing high switching costs. Furthermore this research makes a distinction between pre-switching costs and switching costs by incorporating a mediator variable (willingness to switch) in the model. This split isn’t found in earlier research.

The differential effect of proposition conditions is proven in this research. The offer preference for the simple conditions is significantly higher than for the complex conditions. This is something which can be used in the advertorial content of telco’s. The conclusion of this study means proposition conditions are a significant differentiator and thus should be communicated clearly to the customer. This would mean a change in advertorial practice for no frills operators who mainly communicate on a transactional level. The assumption was that this difference in offer preference would be

moderated by switching costs. This moderation isn’t found to be significant.

Switching costs do play a role in moderating the willingness to switch. The components which are proven to be significant are a switching barrier and brand relations. Higher switching barriers are negatively influencing the willingness to switch, thus stimulating people to switch. Brand relations positively influence the willingness to switch, thus reducing the willingness of respondents to switch with perceived higher brand relations.

The control variable age shows older respondents having a lower willingness to switch than their younger counterparts. Gender is found to have a significant influence on the willingness to switch. Men have a higher willingness to switch while they are not eligible to enter in a new subscription yet. When the respondents are retainable the difference in willingness to switch between men and women disappears. The last control variable current provider shows people having a lower

(4)

January 30, 2016

List of figures

Figure 1: Summary of results... 1

Figure 2: Switching cost components ... 3

Figure 3: Switching cost specified ... 8

Figure 4: Proposition conditions ... 9

Figure 5: specification of research model ... 10

Figure 6: Prompt for survey ... 14

Figure 7: Factor loadings and cronbach's aplha ... 16

Figure 8: Research model split ... 19

Figure 9: Research model evaluated ... 21

Figure 10: Significant research model ... 21

Figure 11: Bundle options ... 21

Figure 12: Process model ... 22

Figure 13: Switching cost components... 28

List of Tables

Table 1: Means, standard deviations, correlations (first construct) ... 17

Table 2: Means, standard deviations, correlations (second construct) ... 18

Table 3: Regression output (first construct) ... 19

Table 4: Regression output (second construct)... 20

Table 5: Process output ... 22

Table 6: Conditional effects ... 25

List of Graphs

Graph 1: Value of conditions ... 20

Graph 3: Brand relationship ... 23

Graph 2: Brand relationship * retainable ... 23

Graph 4: Switching barrier ... 23

Graph 5: Covariate age ... 24

Graph 6: Covariate gender ... 24

Graph 7: Covariate current provider ... 24

Graph 8: Offer preference ... 25

Graph 9: Hypothesis 1 ... 26

Graph 10: Hypothesis 1.1 ... 26

Graph 11: Brand relationship ... 26

(5)

January 30, 2016

1

January 30, 2016

1 Introduction

The products offered by the telecom industry are mostly in proposition form. This manner of offering products makes the product complex. The products mostly consist of several components like minutes, SMS, mb’s and a handset fee. These are offered for a certain price per month. So in fact the product telecom companies are selling, is a combination of a service, being able to connect and use the network, and a product being the sim card and/ or handset. To make it even more complex there are several kinds of proposition conditions. For instance calling abroad or calling to service numbers is charged separately and incurring substantial higher cost when exceeding the amount of units in your subscription (out of bundle charges). With all these conditions and the nature of this service product being technical with low repurchase rate, the offered product of telecom companies can be described as truly complex (Erasmus, 2004). With over 60 different mobile providers in the

Netherlands the competition on the Dutch market is intense. With an assumption of only 5

propositions per provider an orienting customer has over 300 propositions to choose from. In these price plans there is a high choice complexity present (Erasmus, 2014). This indicates that there is a high probability of customers experiencing the phenomenon of too-much choice (Greifeneder, 2010). This phenomenon indicates that less choice should increase your sales.

With this complexity, there should be a chance for operators to distinguish themselves from the rest of the market by being open and transparent without all these limiting proposition conditions and a transparent build-up of the monthly fee. Launching the new proposition at the no-frills (low budget) provider in the Netherlands was exactly aimed on enabling this opportunity. They launched a

subscription where all calling, internet or SMS costs are paid for from the same bundle. No additional billing for calling abroad or calling service numbers and no excessive charges when you go out of bundle. This should theoretically answer to a need of customers in the telecom market. When searching for literature on this specific topic no relevant research came up which exactly answered the doubts. These doubts were: do customers perceive value in a simple or less complex proposition? There is a lot of literature on switching cost where this question is closely linked to. Since the

literature (Khalifa, 2004; Bettman, 1998; Wernerfelt, 1985) describes switching cost as one of the components that builds up to the total customer costs (what does the customer have to pay and do to acquire your product or service, displayed in figure two). There are two directions of thought about switching cost.

The old way of thinking considered switching barriers as a way to keep the customers in (Tax, Brown, and Chandrashekaran, 1998). The new way of thinking doesn’t agree with this statement. For the retention of customers the focus should be on customer satisfaction instead of switching barriers (Jones, 2000) Customers should want to stay with you instead of being forced to stay. Switching

(8)

2

January 30, 2016

barriers also make it harder to attract customers (Khalifa, 2004). It’s of both managerial as academic interest to retest the effect of switching costs on customers. The other component tested by this thesis is perceived value in proposition conditions. In the academic literature this is interesting from a consumer behaviour perspective.

Putting better proposition conditions in place means a trade off on revenues. Higher out of bundle tariffs for instance means more revenues versus less revenues, when the out of bundle tariffs are equal to the within bundle tariffs. From a managerial point of view the relevance lies in adjusting products and proposition, to attract as many customers as possible versus creating more revenues. So this research focusses on testing if proposition conditions can be a differentiator in the telecom market and how this is moderated by perceived switching costs. This research gives insights in this question by asking customers, orienting on a mobile subscription on the homepage of the provider, how they perceive switching costs and if they perceive value in the more transparent and simple proposition of the no frills provider hosting this research. If this is proven to be true simplifying the propositions and thus lowering the switching costs, it would be a relevant way to attract customers. What this research does not answer is whether or not these conditions also help in retaining customers. Although you might reason that retainable customers enter into the acquisition process of a new subscription all over again. Therefore behaving according to the logic of new customers.

(9)

3

January 30, 2016

2 Literature review

2.1 Introduction to the literature

Companies need to increasingly be aware of how their products are perceived by customers and what triggers customers to buy their products. In the playfield of propositions, where there are multiple terms and conditions attached, it is even harder to be fully aware of how your proposition stands out from the competition.

Key to selling a product or proposition is the value the customer is willing to spend for a product (total value to the customer). This value can be measured in money but also in effort. If the total value to the customer is high enough, and the customer has a need for the product, the customer will want to have the product.

The total value to the customer is build up from different components. The model displayed in figure 2, inspired on a model from earlier literature (Khalifa, 2004), puts all these components together and displays the structure and relation between the different components.

2.1.1 Costs to supplier

Costs to supplier consists of all the costs a supplier has to incur to be able to offer the product. This can only be altered through better price negotiations with their suppliers or a more efficient internal process. So the size of this component can vary depending on the total incurred costs. Optimizing this component could result in a higher net customer value if the rest of the components remain stable. How to achieve this is out of scope for this paper.

2.1.2 Supplier margin

Supplier margin is the additional charge the supplier places on top of the cost price of the product to create profits. This component should be optimized to create both profit for the company and net customer value for the customer. This is a sales or marketing choice on the matter of how much profit the company wants to create. Normally this is a delicate balance volume and margin. The cost to supplier together with the supplier margin creates the price of the goods sold.

Total value to customer Net customer value Customer gain Total customer cost Switching costs

Price Supplier margin

Cost to supplier Fixed costs Room to manoeuvre

(10)

4

January 30, 2016

2.1.3 Switching costs

To get to the total customer costs the switching costs have to be included. These are the costs the customer experiences in selecting the right product to suit his needs. The switching costs can be split up in several components being procedural, financial and relational switching costs. Switching costs can be used to create a lock-in or they can be minimized to facilitate a transfer from one to the other provider. The room between the total customer costs and the total value to the customer is what makes a product attractive for the customer. Thus you would expect that by minimizing the switching costs you could create a bigger total value to the customer which makes your product more

attractive.

2.1.4 Net customer value

The net customer value is the benefit a customer perceives after deducting all the costs of searching, selecting and buying a new product. This is the real added value for a customer. The bigger this part is, the higher is the willingness to buy. Imagine being able to buy an iPhone 6S for 50 euros. This would immediately trigger you to buy, since the total value to the customer is much higher than 50 euros. So if there is a need, the willingness to buy could be stimulated by increasing the net customer value. With a higher net customer value the assumption is that customer has a higher purchase intention. Higher customer purchase intentions is a key goal of almost every company. That is what makes this question interesting as well as important and relevant to both the business and the academic world.

2.1.5 Proposition conditions

With the proposition conditions of your product of service you are able to control the switching costs, or at least lower them (Wang, 2012). Knowing this it has become of interest to see how these two factors interact with each other. Is there an influence of product or proposition complexity on the offer preference and is this moderated by the switching costs? (Wang, 2012)

2.1.6 Moving forward

With these components, briefly described in this introduction of the theory chapter, the goal is to find out whether proposition conditions really could make a difference in customer preference between a simple and a complex proposition and how this is affected by switching costs.

Moving forward this study includes nine chapters. This chapter will elaborate further on the relevant literature around the subject of switching costs and simple and complex propositions. Chapter three puts this literature into perspective within the company at which the research is done. Chapter four

(11)

5

January 30, 2016

states the research question, design, strengths and limitations. Chapter five shows the analysis, which is further elaborated in chapter six and seven, about what the analysis means for the research question and hypothesis. A discussion of the main findings, practical implications, limitations and ideas for future research are found in chapter eight. At the end, chapter nine offers the final conclusions of this study.

The introduction shortly introduced the most important components of this research. This chapter will provide a deeper understanding of the different components of switching costs and the relevant literature in the field. It goes into depth about switching cost and what exactly they entail. After this part of theory the case is introduced followed by the research question, hypothesis and tests.

2.2 So what are these switching costs?

If a proposition is too complex, the average customer will perceive higher switching costs. (Wang, 2012) There is surprisingly little to find on the topic of product complexity and the influence of this on the purchase intention. Some research is done in the online sales of assurances (Wang, 2012) which finds that a perceived product complexity has a negative effect on the post purchase intention. Although this is about post purchase intention instead of pre purchase intention the case is quite similar to this research. Both papers try to answer the same question: Do difficult propositions cause a lower purchase intention?

The telecom industry is a dynamic and complex marketplace. This is caused by the fact that telecom is technical, a service and has a low repurchase rate (Erasmus, 2014) These dynamics and complexity have a suspected significant effect on perceived switching costs and the ability for customers to choose a proper subscription. So the idea is that by reducing the switching costs compared to other companies, there is potential to create a competitive advantage, since customers experience less stress in choosing the right subscription for their needs. This advantage can be brief but if you create your advantage from proposition conditions it’s a lot harder for competitors to copy. It would take them at least a year to adjust their backend systems in such a way that they can offer the same conditions.

So is it beneficial for companies to simplify their propositions? Could it be that customers don’t fully comprehend the propositions and do they perceive a lower total customer value of it because of higher switching costs (Bettman, 1998)? If this is the case it would be highly beneficial for marketers to formulate an as easy as possible proposition around their products. This would lower the

perceived switching costs and with that increase the perceived net customer value like indicated in the customer exchange model (Khalifa, 2004).

(12)

6

January 30, 2016

Browsing through the different websites of companies you will find a wide variety of subscriptions and bundling options. So if support is found for the theory that by reducing switching costs (at least short term) a competitive advantage can be achieved, all the major companies in the market will act on this. And probably not only in the Netherlands, but this could be extrapolated worldwide which in turn will open the door to more research in proposition building with complex products. The

research question to be answered is: Do proposition conditions influence customer preference and do switching costs moderate the offer preference between a simple and a complex offer? The assumption is that a simple offer represents a higher total value to the customer then a complex offer.

This isn’t such a simple question to be answered. It could well be that customers experience a sense of value in the complexity of certain products.

There are several factors which build up these switch costs.

Firstly there are procedural switch costs. This includes the effort of searching, learning about the product and market (Wernerfelt, 1985), comparing alternatives to minimize the economic risk of a purchase and the set-up costs which entails the effort required to start using the new product. Like installing your new television and connecting it to all the equipment surrounding a television. These procedural costs should be minimized to prevent people from not switching due to all the effort they have to put in.

The second component is the financial part of switching costs. This includes losing benefits which, in the end, costs money to regain and monetary loss costs in the sense that you should spend

(potentially more) money. Again this should be minimized to attract new customers. If it’s clearly a better deal, people will come and buy.

The last component of the switching costs entails relational switching cost. These could be personal, like formed bonds with employees of suppliers, or brand relationship loss costs which means that customers can feel connected to the brand and don’t want to leave out of loyalty. Building up a relationship requires investments. When this relationship is terminated, these investments are lost. (Dwyer, Schurr, and Oh, 1987; Morgan and Hunt, 1994; Klemperer, 1995). The loss of these

investments is an important driver of switching costs (Blattberg and Deighton, 1996) Customers perceive higher investment in a supplier when they have different products with one supplier (Blattberg and Deighton, 1996). The same goes for modifications the customer might have made to the product. (Bhardawaj, Varadarajan, and Fahy, 1993; Robinson, 1988) These bonds, investments and modifications have to be recreated with the new supplier which again entails effort and

therefore costs. These relational switching cost can be a significant factor since people rely more on personal relation when products are perceived as more complex (Sheth & Parvatlyar,1995)

(13)

7

January 30, 2016

These switching costs should also be optimized to create net customer value and build a personal or brand relationship with the customer. With these formed bonds it should be easier to keep the customers with the company, because if they leave they lose the investment in the relationship. This net customer value is the last component of the total customer value. The net customer value is the perceived gain for the consumer.

To recap:

The supplier cost + supplier margin + switching costs = total customer costs. If the total customer value is bigger than the total customer costs the customer perceives net customer value. This is what the customer thinks to gain by the purchase after deducting all cost. When a supplier can optimize this net customer value the customer perceives more gain and therefore it is more likely that the customer will have a higher intention to buy the product. So what companies should do, is trying to optimize the price of the product, in the room to manoeuvre in the model in figure 2 mentioned earlier in this paper.

Of course this isn’t a very conscious process. But like mentioned earlier: If you can buy an iPhone 6s for 50 euros there is probably little to think about and you would purchase this immediately. In this case the supplier costs and margin should be really low and the switching costs are also pretty low since you can buy and try the phone. If it doesn’t suit your needs you can disregard it easily since the investment is only 50 euros. The perceived value of an iPhone 6s could be around 500 euros. Still cheaper than it really is although this would differ for each individual. All this would result in the simple sum of 500-50=450 euro net customer value. Again this is not a conscious process, but something we evaluate unconsciously.

Like indicated earlier complexity is a component which causes higher perceived switching costs. Because people have to investigate the differences of the several providers and the differences between subscriptions with a provider the telco market can be perceived as complex. (Erasmus, 2014) Switching costs directly affects loyalty, and has a moderating effect on both customer satisfaction and trust. Therefore, it plays a crucial role in winning customer loyalty. If a customer perceives switching costs to be high, a barrier to exit will be set up and the result will be apparent loyalty even in the absence of satisfaction or trust. Setting up this barrier to retain your customer is proven to be a bad choice. Customers could stay because of the high switching costs but can start expressing their negative emotions creating a lot of bad word of mouth. For the retention of customers the focus should be on customer satisfaction instead of switching costs. (Jones, 2000) Customers should want to stay with a company instead of being forced to stay. This is the more recent belief which is supported in this study. Switching costs could be used to lock customers in, making it difficult for customers to

(14)

8

January 30, 2016

switch. Some authors (Tax, Brown, and Chandrashekaran, 1998) support this way of using switch costs since it could mean that the retention rates go up.

Because of the potential importance of switching costs, GSM operators should focus on understanding and application of the switching costs phenomenon. (Aydin et al. 2005) This proves that customers tend to stick to a provider if they perceive high switching cost. This can be challenged by a provider with making the propositions as simple and transparent as possible.

2.3 Switching cost in a comprehensive model

A deeper insight in the matter of switching costs is provided by the theoretical model displayed in figure three. This model is inspired on Brunham (2003). This model incorporates all the different aspects of switching costs in a nice and comprehensive way.

After putting switching costs into perspective with all of its components and establishing a definition it’s relevant to see on what segments of the Telco market the effects are significant. If the total investment is bigger, the relative size of the switching cost component becomes smaller. So if you spend 2 euros a month you probably aren’t going to investigate all options very thoroughly. If the expenses lie around the 80 euros a month you are probably more willing to put in some time to get the best deal. According to Lee et all (2001) switching costs only play a significant moderating role in the satisfaction-loyalty link for the economy and standard groups. For mobile lovers, switching costs do not affect loyalty. The idea here is that heavy users have such a considerable investment in the specific product that it is worth to find out what the best option for them would be. To refer back to the Khalifa model in the heavy user scenario the price is a lot higher, which causes the switching costs to be a smaller percentage of the total customer cost. This means the switching costs could be especially relevant with a no frills provider.

Like indicated by Burnham et al. (2003) a component of switching cost is product complexity. Product complexity combined with a lot of alternatives, like the current situation in the Telco industry in the Netherlands, causes people to be less satisfied with the chosen alternative than when the attributes and alternatives are less. (Greifeneder , 2010)

Procedural switching costs

Economic risk costs Evalution costs Learning costs Set-up costs

Financial switching costs

Benefit loss costs Monetary lost costs

Relational switching costs

Personal relationship loss costs Brand relationship loss costs

Offers Purchase intention

Switching costs

(15)

9

January 30, 2016

This reduced satisfaction could be caused by people choosing the wrong product because of the too much choice effect. It is a known fact that a lot of Telco customers are having a non-optimal bundle like indicated by Miao and Jayakar (2014). The results of their analysis suggest that the large majority of customers chose non-optimal bundles. The results of this study showed that the customers were risk aversive in their bundling choices. They appeared to prefer bundles that provided them with usage limits well in excess of their current levels of usage. Hence, it could be inferred that customers prefer the convenience of fixed, though higher, subscription payments to the possible downside of surprise excess usage charges on their monthly bill. An alternative explanation is that customers overestimate their demand for usage and choose bundles that provided greater usage than their current levels justify. In short you could state that due to complexity, customers make the wrong choices which comes at a price premium.

3 Case: Switching cost in the no frills telco market

There is a no frills telecom provider in the Netherlands which recently changed its propositions in such a manner that you could state that their propositions have become less complex compared to the propositions used by the competitors. One of the drivers to do so is to increase sales without having to lower the prices. While exploring this phenomenon the Khalifa model came in sight, which isolates the different components of customer costs. Changing the offered proposition affects the switching costs component in the model. (Khalifa, 2004) By changing, the operator hopes to attract more customers. This study is therefore focussed on the acquisition of customers since this adds to your customer base.

In their effort to attract as many customers as possible mobile telecommunication firms (Telco’s) try to attract customers in all possible ways. For the 17 million people inhabiting The Netherlands there are 59 virtual network providers and 3 network providers. So in total there are over 60 different providers of mobile subscriptions in the Netherlands. This indicates that the competition in the Dutch market is intense. With the assumption that each provider offers only 5 different price plans there are 300 price plans to choose from. In these price plans there is a high choice complexity present (Erasmus, 2014). This indicates that there is a high probability of customers experiencing the phenomenon of too-much choice (Greifeneder, 2010). This phenomenon indicates that less choice should increase your sales.

The newly introduced propositions at the no frills provider entail several interesting elements which should lower the switching costs. These elements are displayed in figure 4. The combination of these advantages

(16)

10

January 30, 2016

bundle so you don’t have to know the distribution over internet use and calling minutes in advance. This bundle even includes calling service numbers and calling to and from foreign countries. Furthermore the out of bundle usage charge is lower than the within bundle charge. Which means there is no pressing need to choose a bundle exceeding your usage addressing the issue indicated by Miao and Jayakar (2014). So people are not surprised by unexpected bills. To make sure this won’t happen there is an optional capping. If wanted, you can incorporate a manual action (top up) to continue calling, texting or using the internet. This ensures no unexpected costs. When you unfortunately still have chosen a subscription too big or too small for your needs you can adjust this every month without a charge.

3.1 What does this study wants to prove about this proposition?

There is a lot of theory (Greifeneder, 2010; Aydin, 2005; Burnham, 2003; Khalifa, 2003; Lee, 2001; Miao, 2014; Chang, 1994; Wang, 2012; Erasmus, 2014; Wernerfelt, 1985) about switching costs and subscription choice. But what the paper focusses on is do customers perceive value in the conditions of a proposition? Are they willing to pay an additional charge to have these conditions? If so this would be a differentiator. Something which is often argued at the no frills operator but never really proven. Does this provider have a better subscription because of their conditions? After this the switching costs comes into play. The expectation is that the switching costs plays a role in the offer preference. So people who perceive higher switching costs are sooner inclined to choose a simple proposition. It

sounds logically from a theoretical point of view but now we are going to put it to the test. Do proposition conditions influence customer preference and do switching costs moderate the offer preference between a simple and a complex offer? Building on an earlier mentioned model inspired on the model of Brunham (2003) the pre switching costs have to be incorporated in the research model displayed in figure five.

H1.1 Procedural switching costs H2.1 Procedural switching costs

• H 1.1.1 Economic risk costs • H 2.1.1 Economic risk costs • H 1.1.2 Switching barrier • H 2.1.2 Evalution costs

• H 2.1.3 Learning costs • H 2.1.4 Set-up costs

H1.2 Financial switching costs H2.2 Financial switching costs

• H 1.2.1 Benefit loss costs • H 2.2.1 Benefit loss costs • H 1.2.2 Monetary lost costs • H 2.2.2 Monetary lost costs

H1.3 Relational switching costs H2.3 Relational switching costs

• H 1.3.1 Personal relationship loss costs • H 2.3.1 Personal relationship loss costs • H 1.3.2 Brand relationship loss costs

Retainable H1.4 Willingnes to switch

H1 Pre Switching costs H2 Switching costs

H2.4 Offer preference

(17)

11

January 30, 2016

This is contributes to academic literature by splitting up switching costs in pre switching costs and switching costs. This is a differentiation needed in this study since the industry works a lot with fixed term contracts. In the literature this split is not investigated yet and this model is new although inspired on earlier articles (Khalifa, 2003).

The assumption is that there is a need coming from being retainable (being able to enter in a new contract) or not. This need will have an expected direct effect on the willingness the switch. Which in turn will be moderated by the pre switching cost (construct H1). An example of this in a different market is the energy market. Here people know they can switch and potentially save money while still receiving the same thing. Despite of this a lot of people don’t switch because it’s perceived as too much of a hassle and it takes time to sort everything out. This is a very typical example of pre switching costs. Although people can switch they don’t because of (expected) effort they have to put in. The model above incorporates the factor retainable with pre switching costs to get to a high or low willingness to switch. Like illustrated in the example of the energy market it could be that people can, and should switch but don’t. This can be caused by the construct pre- switch costs.

The willingness to switch is de mediator in the model. This mediator has an expected direct effect on the independent variable in this model which is moderated by the construct switching costs. The assumption is that there is no direct effect between retainable and offer preferences since people who aren’t retainable or feel no need to switch probably don’t have an explicit offer preference. This is why the willingness to switch (Mediator) is introduced in the model. The willingness to switch can be low because of either the perceived switching costs or not being retainable.

Beside this moderator the expectation is that the simple or complex option also has a moderating effect on the offer preference. The switching costs should lower the offer preference where the simple option should result in a significantly higher offer preference than the complex offer.

(18)

12

January 30, 2016

4 Research question and hypotheses

Could an operator create an advantage over its competitors by offering less complex and more transparent conditions? Based on the theory above this could certainly be the case. With the switch costs being lower, the total customer costs decrease and with that it increases the total customer value. This research doesn’t only focus on one proposition being more or less attractive than a competing one. The switching costs in the process are also considered to be an important factor in the process of a consumer selecting a new subscription. These two topics of research concentrate themselves in the following main hypothesis:

Do proposition conditions influence customer preference and do switching costs moderate the offer preference between a simple and a complex offer?

This is an interesting question because this could be a real differentiator from other telco companies. The expectation is that people do perceive value in a simple offer. This expectation is based on earlier research (Khalifa, 2003) and the model presented in figure two.

To answer this main question there are several components which should be answered. The first construct in the model is whether or not being retainable (eligible to enter in a new subscription) has significant positive effect on the willingness to switch. This effect is expected to be positive and significant since people are expected to be more engaged when a decision directly affects themselves. This results in the first hypothesis of the model.

H1. Do people who are nearing the end of their contract by a month or less (retainable), have a higher willingness to switch than people who are further away from the expiring date of their contract?

The first hypothesis is expected to be moderated by the pre-switching costs. There could be a need (customer is retainable) but the willingness to switch could still be low. Like indicated earlier, with an example of the energy market, people could have a low willingness to switch because of the effort they have put in to invest a possible new subscription. A customer has to investigate the options available and choose what the best option is. This requires time and effort thus an investment. If the investment in the form of switching costs gets higher, the expectation is that the willingness to switch will go down. This results in the following hypothesis:

H1.1 Higher switching costs reduce the willingness to switch.

Having tested the first moderation variable the next construct comes into play. The effect of the mediator variable (willingness to switch) on the independent variable (offer preference). The

(19)

13

January 30, 2016

expectation is that a higher willingness to switch will result in a higher offer preference regardless of the offer. The reasoning behind this is that when the customer is really displeased with his current provider he wants to switch no matter what the costs or the alternative. This results in the following hypothesis:

H2 A higher willingness to switch will result in a higher offer preference.

This relation is expected to be moderated by the switching cost. Since the willingness to switch could be high the switching costs could withhold a customer from actually choosing a new subscription. The customer could for instance be so overwhelmed by the complexity of the product or the terms and condition that he doesn’t want to switch because of the total costs for the customer, because of switching costs, becoming more than the total value to the customer. Customers could also become indifferent, because of all the work required to switch. Of course this is an extreme example but the direction is set by this. Higher perceived switching costs are expected to result in a lower offer preference. This results in the following hypothesis:

H2.1 Higher switching costs result in a weaker (lower) offer preference.

To test whether or not customers perceive value in a proposition two different options are shown in an in-between samples design. This entails that one group of respondents fills in their offer preferences based on the simple proposition conditions of the no-frills operator where this research is conducted. The other group of respondents gets to see the more complex conditions of another operator in the Dutch no frills market. The expectation is that the offer preference of the group with the simple proposition conditions is significantly higher than the offer preference of the other group, since this would suggest that customers indeed perceive value in the newly introduced proposition conditions of the operator with the simple proposition conditions. This brings us to the last hypothesis:

H2.2 The simple proposition is expected to have a higher score on offer preference then the complex proposition.

4.1 Research design

The intended way of doing research on the matter of perceived switching costs influencing the offer preference in the telecom industry is by surveying people. This survey is going to be posted on the homepage of the no-frills operator. Everybody who enters the site will be prompted to fill in the survey. By asking people to take part in the survey on the instance of entering the homepage the group of people is randomly chosen out of people orienting online on telecom subscriptions.

(20)

14

January 30, 2016

4.2 Method

The best fit for this study regarding population is randomly selected subjects. This will be ensured by posting the survey on the homepage of a Dutch no frills provider. The relevant population are people interested in subscriptions with a no frills provider. When entering the site people will be asked in a pop-up window (Figure 6) at the bottom right of the screen if they are willing to take part in a scientific study about perceived switching costs and the effect of this on proposition choice. This entails simple random sampling. The survey will be

filled in online by the subjects without supervision. The intended sample will contain around 500 subjects to be sure of a representative group. Previous surveys on the site had a respondent rate of 40%. Considering these surveys where smaller and posed to a more engaged audience (people who had already bought a product) the assumption is the respondent rate is going to be lower on this survey. An estimation of 5% seems reasonable. With at least 20.000 visitors a day on the homepage this survey first is set to be displayed at 1 in 5 customers. This means 4000 people a day will be asked to fill in the survey. With the assumed respondent rate this means 200 results a day. The target of 500 subjects should therefore be reached in a timespan of around 3 days.

4.3 Strengths and limitations

4.3.1 Limitations

This research is done on the website of a no-frills provider. So is for the biggest part only applicable to no-frills offers and companies. Then we have the matter of self-administered surveys. Although there are control questions present it could be that people misinterpret the questions or lose interest while filling in the survey. Since the survey is posted on the website of a provider there is a high probability of getting a lot of respondents who are already a customer of the specific brand. Price and bundle size are components left out of the research. It would be interesting to see how strong price and bundle size would be as a predictor.

4.3.2 Strengths

Since this survey is launched on the website of a real provider the test subjects are real customers. This causes it to be a strong point. Furthermore the quantity of the surveys will probably be high, meaning a more representative sample of the population and a limited influence of outliers. Also, the results between the different variables are probably going to be significant if there is something to detect.

(21)

15

January 30, 2016

5 Analysis

In this section the results of the survey and following analysis are described. By the end of this paragraph you will find the conclusions about the hypothesis based on the gathered data.

5.1 Survey results

After having had the survey live for a week on the homepage of the no-frills provider the result is 1246 surveys. This means 178 surveys started a day, which comes close to the earlier estimation of the respondent rate. The dropped out rate is pretty high. Of the 1246 started surveys 638 where incomplete and therefore removed. The remainder is 607 completed surveys. Filling in the survey takes roughly 5 minutes. So all results with a completion time under 4 minutes are deleted since they can’t be filled in accurate. This concerns 37 results. Then a completion time of above 15 minutes is also suspicious and would suggest the attention wasn’t on filling in the survey. Therefore all results which took longer than 15 minutes are also deleted. This concerned 38 results. This leaves 532 completely filled-in surveys within the time bounds. To make sure respondents didn’t just randomly select answers without reading the questions 2 sets of reverse coded questions where incorporated. To filter out the people who are consistently selecting answers without reading the respondents are marked which answered the same on 2 sets of reversed coded questions. (H1.1.2 and H1.3.1) If the respondent had the same answer (apart from neutral) in both questions the results are removed because this suggest just selecting the same answer without reading. There were 38 of these cases filtered out with this method. To also check attention at the end of the survey the last 5 answers are checked to be unequal. If the answer was the same at all 5 questions the results are removed. 57 respondents did this and are therefore removed from the sample. This leaves a usable sample of 437 respondents. Which is close to the desired 500 surveys.

Questions H1.1.2_B, H1.2.1_B, H1.3.1_B, H1.3.2_C, H2.1.3_B, H2.1.3_D, H2.1.4_D we reverse coded to check for attention. These questions are recoded in the following manner 5=1, 4=2, 3=3. In this manner the question follow the same pattern as the rest of the questions. Where 1 represents high switching cost, intention or need.

(22)

16

January 30, 2016

Figure 7: Factor loadings and cronbach's aplha

5.2 Factor analysis and reliability analysis

In figure seven all the factor loadings are displayed of the different questions with the Cronbach’s alpha for each grouped item. The red marked items are deleted since they load on a group which wasn’t expected based on the theory, furthermore the Cronbach’s alpha goes up by deleting them. Therefore they are removed. The group names and the group abbreviation as used in SPSS are displayed in the last 2 columns.

5.3 Independent, dependant and mediator variable.

To start this analysis we need to determine the independent variable, mediator and dependent variable. The independent variable is retainable. This is determined by how long the contract of the respondents runs. The responses are recoded from the variable contract duration. The question here is how far away are you from the ending date of your current contract? I coded more than a year and half a year to be “not retainable” (2). A month, week and already expired is coded to be “retainable” (1).

To get the willingness to switch I grouped the questions measuring this together. The translated questions are “it’s likely that I will switch mobile subscription in the near future” and “it’s likely that I will also switch provider”. These questions both measure the same construct according to the factor analysis. The Cronbach’s alpha of the willingness to switch construct is .764. So the grouping of these question forms a reliable construct to measure the willingness to switch.

Kijken of 1.2 nog te verklaren valt aan de hand van velengbaar of niet? Cronbachs op groepen maken.

Factor loadings Crombach's Rotated Component Matrixa Group name Group abbreviation

0,884 0,784 H1.3.1_A Ik heb veel geïnvesteerd in de relatie met mijn huidige mobiele aanbieder (9) 1.3.1 Monetary lost cost F1.3.1 0,843 H1.3.1_B Ik heb R(niet veel)R geïnvesteerd in de relatie met mijn huidige mobiele aanbieder (10)

0,731 0,853 H1.2.2_A Ik heb veel tijd en geld geïnvesteerd in mijn huidige mobiele abonnement (7)

0,873 0,768 H1.1.2_A Het gaat mij veel tijd kosten om een nieuw mobiel abonnement te selecteren (3) 1.1.2 Switching Barrier F1.1.2 0,872 H1.1.2_B Het kost mij R(weinig)R tijd om een nieuw mobiel abonnement te selecteren (4)

0,64 0,831 H1.2.2_B Het kost mij teveel moeite om te wisselen van mobiel abonnement (8)

0,849 0,623 H1.3.2_B Ik ondersteun mijn huidige mobiele aanbieder als bedrijf (12) 1.3.2 Brand relationship loss costs F1.3.2 0,792 H1.3.2_A Ik vind het publieke imago van mijn huidige mobiele aanbieder erg leuk (11)

0,582 0,66 H1.2.1_A Mijn mobiele abonnement biedt mij voordelen die ik nergens anders krijg (5)

0,86 0,656 H1.1.1_B De service van een nieuw mobiel abonnement zou slechter kunnen zijn dan de service die ik nu krijg (2) 1.1.1 Economic risk cost F1.1.1 0,809 H1.1.1_A Ik weet niet zeker wat het niveau van de dienstverlening is als ik naar een nieuwe mobiel abonnement overstap (1)

0,77 0,217 H1.2.1_B Ik zou R(geen enkele)R voorkeursbehandeling verliezen als ik wissel van mobiel abonnement (6) 0,633 H1.3.2_C Het maakt mij R(niet)R uit welk merk mobiele aanbieder ik gebruik (13)

0,796 0,853 H2.1.2_E Het is moeilijk om mobiele abonnementen te vergelijken (12) 2.1.2 Evaluation costs F2.1.2 0,777 H2.1.3_A De opties van een nieuw mobiel abonnement, net zo goed gebruiken als nu, gaat mij veel tijd en moeite kosten (13)

0,772 H2.1.2_C Het kost veel tijd en moeite om de informatie te krijgen die je nodig hebt om zelfverzekerd een goede keus te maken tussen de verschillende mobiele abonnementen (10) 0,716 H2.1.2_D De voordelen van mijn huidige mobiele abonnement met andere mobiele abonnementen vergelijken, kost mij teveel tijd en moeite (11)

0,643 H2.1.2_B Ik heb de tijd niet om te zoeken naar de informatie die ik nodig heb om te wisselen van mobiel abonnement (9) 0,431 H2.1.2_A Ik weet niet wat ik allemaal moet doen om te wisselen van mobiel abonnement (8)

0,42 H2.1.4_A Het kost tijd om alle stappen te doorlopen bij het wisselen van mobiel abonnement (17)

0,884 0,907 H2.3.1_C De mensen bij mijn huidige mobiele aanbieder doen er toe voor mij (29) 2.3.1 Personal relation loss costs F2.3.1 0,869 H2.3.1_D Ik praat graag met de mensen bij mijn huidige mobiele aanbieder (30)

0,848 H2.3.1_B Ik vind het prettiger om te communiceren met de mensen van mijn huidige aanbieder dan met de mensen van een andere aanbieder. (28) 0,831 H2.3.1_A Ik zou het contact missen met de mensen van mijn huidige mobiele aanbieder (27)

0,775 0,823 H2.1.1_B Als ik van mobiel abonnement wissel zou het kunnen dat ik een tijdje last heb van slechte service (2) 2.1.1 Economic risk costs F2.1.1 0,738 H2.1.1_C Wisselen van mobiel abonnement brengt waarschijnlijk verborgen kosten met zich mee. (3)

0,73 H2.1.1_A Ik maak me zorgen dat de service van andere mobiele aanbieders niet zo goed werkt als verwacht (1) 0,691 H2.1.1_E Wisselen van mobiel abonnement brengt waarschijnlijk onverwacht gedoe met zich mee (5)

0,57 H2.1.1_D Waarschijnlijk krijg ik een slechte deal als ik wissel naar een ander mobiel abbonement (4)

0,779 0,763 H2.1.4_D Het installatieproces van een nieuw mobiel abbonement is R(snel en gemakkelijk)R (19) 2.1.3 Learning costs F2.1.3 0,748 H2.1.3_D Wennen aan een nieuw mobiel abonnement is R(makkelijk)R (16)

0,677 H2.1.3_C Zelfs na het wisselen van mobiel abonnement, kost het moeite om alles in te stellen (15) 0,611 H2.1.3_B Het begrijpen van een nieuw mobiel abonnement stelt R(niet)R veel voor (14)

0,825 0,788 H2.2.1_A Een nieuwe mobiele aanbieder nemen betekent dat ik punten, credits, services en dergelijke verlies, die ik heb opgebouwd bij mijn huidige aanbieder (21)2.2.1 Benefit loss costs F2.2.1 0,794 H2.2.1_B Ik verlies de voordelen van het lang klant zijn als ik weg ga bij mijn huidige mobiele aanbieder (22)

0,54 0,702 H2.2.2_C Wisselen van mobiele aanbieder betekent nieuwe opstartkosten. (aansluitkosten, verzendkosten, lidmaatschapskosten, borg enz..) (23) 0,462 0,788 H2.1.4_E Er komen veel formaliteiten kijken bij het wisselen van mobiele aanbieder (20)

(23)

17

January 30, 2016

Then we need the final construct (dependent variable) being offer preference. To measure this construct four questions are posed.

1. It’s likely I will enter in a new subscription

2. There is a big chance I choose for a subscription with the conditions as mentioned before 3. I’m prepared to pay 1 euro extra for the showed proposition

4. I think that the proposition contributes to the predictableness of the bill

To test if these questions measure the same construct a factor analysis is performed. With Eigen values all above one this set of question measures the same construct which explains 58,3% of variance. To see if the scale is reliable the Cronbach’s alpha is calculate for these questions. This group has a Cronbach’s alpha of .746 which can be slightly improved with .041 by deleting question one. Since it’s only .041 and the Cronbach’s alpha doesn’t reach the .8 even if question one is deleted the question is kept in the construct. Which formed the construct “offer preference” (dependent variable).

5.4 Correlations

5.4.1 Construct 1 pre switching cost

SPSS provides us, after computing the scale means, with a table of correlation coefficients as

displayed in table 1, for all of the combinations possible in the construct pre switching costs. The test here is to see which components of pre switching costs affect the probability of selecting a new subscription. Derived from this analysis the conclusion is that pre switching costs components are no real predictor for willingness to switch. All correlations found are tendencies. So no strong

correlations in construct one. Significant correlations are found between brand relationship loss cost and Willingness to switch (-.357) which would suggest that the willingness to switch has a tendency to correlate negative with brand relation loss cost. Other significant correlations are Economic risk costs with switching barrier and retainable with willingness to switch.

Table 1: Means, standard deviations, correlations (first construct)

Table1 Means, Standard Deviations,Correlations (First construct)

M e a n St d . D e v ia ti o n W ill in g n e s s t o s w it c h N u m _ Ve rl e n g b a a r_ 1 /Y 1 .3 .1 M o n e ta ry lo s t c o s t 1 .1 .2 Sw it c h in g Ba rri e r 1 .3 .2 Bra n d re la ti o n s h ip l o s s c o s ts 1 .1 .1 Ec o n o m ic ri s k c o s t G e n d e r Ed u c a ti o n Willingness to switch 2,86 1,05 (.764) Retainable 1,61 0,49 ,208**

1.3.1 Monetary lost cost 3,64 0,86 -,025 -,067 (.85)

1.1.2 Switching Barrier 3,14 1,00 ,156**

,141**

,098* _(.77)

1.3.2 Brand relationship loss costs 2,99 0,74 -,357**

-,100* _,057 _-,043 _(.62)

1.1.1 Economic risk cost 2,73 0,88 ,018 ,028 ,064 ,348** _-,043 _(.66)

Gender 1,36 0,48 ,063 -,058 -,058 -,072 -,076 ,013 Education 3,47 0,96 -,024 ,047 ,022 -,017 ,110* _-,024 _-,067

Age 15,65 3,69 ,117* _,058 _-,052 _-,059 _,064 _,089

-,260**

,220**

**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).

(24)

18

January 30, 2016

5.4.2 Construct 2 switching cost

SPSS provides us, after computing the scale means, again with a table of correlation coefficients as displayed in table 2, for all of the combinations possible in the construct switching costs. The test here is to see which components of switching costs affect the probability of selecting a new

subscription. Derived from this analysis the conclusion is that switching costs components are no real predictor for offer preference. Since all the r values in the column are around 0 and no bigger than 0.2. In this construct all the different component are significantly positively related to each other with a significance value smaller than p=0.01.

There are high positive relations between evaluation costs and economic risk costs and between learning costs and evaluation costs. All of the other constructs have a tendency of a positive relation with r value varying between .438 and .178.

Table 2: Means, standard deviations, correlations (second construct)

So to summarize these results of the correlation of switching costs with intention to switch there is no significant relation between perceived switching costs and willingness to switch.

Furthermore the switching costs also don’t appear to affect the offer preference.

Now we concluded that switching costs are no real factor in the choice of a subscription. In correlation testing we should test if the complexity of the product does create a significant difference. M e a n St d . D e v ia ti o n G e n d e r Ed u c a ti o n Ag e O ff e r_ Pre fe re n c e Will in g n e s s t o s w it c h 2 .1 .2 Ev a lu a ti o n c o s ts 2 .3 .1 Pe rs o n a l re la ti o n l o s s c o s ts 2 .1 .1 Ec o n o m ic ri s k c o s ts 2 .1 .3 L e a rn in g c o s ts 2 .2 .1 Be n e fi t lo s s c o s ts Gender 1,36 0,48 Education 3,47 0,96 -,067 Age 15,65 3,69 -,260** _,220** Offer_Preference 2,96 0,85 ,011 ,102* ,051 (0.746) Willingness to switch 2,86 1,05 ,063 -,024 ,117* _,132** (.764) 2.1.2 Evaluation costs 3,24 0,72 -,073 ,009 -,027 ,089 ,120* _(.85)

2.3.1 Personal relation loss costs 3,88 0,80 ,001 ,141** _-,137** _,011 _-,283** _,200** _(.91)

2.1.1 Economic risk costs 3,16 0,74 -,045 -,008 -,063 ,066 -,089 ,525** _,310** _(.82)

2.1.3 Learning costs 3,40 0,64 -,075 ,070 -,183** _-,014 _,015 _,576** _,178** _,438** _(.76)

2.2.1 Benefit loss costs 3,38 0,95 ,003 ,059 -,057 -,008 -,070 ,261** _,346** _,236** _,203** _(.79)

**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).

(25)

19

January 30, 2016

5.5 Regressions

To test which moderators are significant, linear regressions are performed on both parts of the model since it isn’t possible to test the full model with Process.

5.5.1 Regression first part (Left and Red)

The first regression following is about the first part of the model displayed in figure 8(left and red square). Here we test the effect of our control variables and moderators on the relation between need for a new proposition and willingness to switch.

Hierarchical multiple regression is preformed to investigate the ability of economic risk costs, switching barrier, monetary loss costs and brand relationship to predict the level of willingness to switch, after controlling for gender, age and current provider. Education level, monetary loss costs and economic risk costs are found to be not significant and therefore left out of the model. The results of the significant factors are displayed in table 3.

In the first step of the multiple regression, three predictors were entered: gender, age and current provider. This model is statistically significant F (13.812); p< 0.001 and explained 8,7% of variance in willingness to switch. After entry of retainable, switching barrier and brand relationship loss costs in step 2, 3 and 4 the total variance explained by the model as a whole is 21,4% F(20.831) p<0.001. In this model 6 out of 9 predictor variables were statistically significant. With brand relationship loss costs recording a higher Beta value (β .327, p < 0.001) than current provider (β .173, p < 0.001), age (β -.145, p < 0.01), switching barrier (β.119, p < 0.01), retainable (β -.103, p < 0.05) and Gender (β-.095, p <0.05).

H1.1 Procedural switching costs H2.1 Procedural switching costs

• H 1.1.1 Economic risk costs • H 2.1.1 Economic risk costs • H 1.1.2 Switching barrier • H 2.1.2 Evalution costs

Retainable H1.4 Willingnes to switch

Hierachical regression model of willingnes to switch

Adjusted R Square model = 21,4% R R² R² Change B SE β t

Step 1 Gender .296 .087*** -.206 .096 -.095* -2.134 Step 1 Age -.041 .013 -.145** -3,264 Step 1 Current provider .522 .136 .173*** 3.833 Step 2 Retainable .325 .106** .018 -.221 .097 -.103* -2.286 Step 3 Switching barrier .348 .121** .015 .125 .045 .119** 2.739 Step 4 Brand relationship

loss costs

.475 .225*** .104 -.460 .061 -.327*** -7.603 Note statistical significance *p <.05; ** p <.01 ; *** p <.001

Figure 8: Research model split

(26)

20

January 30, 2016

5.5.2 Regression second part (right and green)

The next step is to analyse the right part (green line) of the model. The test is to see if offer preference is significantly moderated by one or more of the switching costs components or by one of the control variables. The outcome is offer preference for either the simple or the complex proposition.

Hierarchical multiple regression is preformed to investigate the ability of different switching costs components to predict the level of offer preference, after controlling for Gender, Age, Education level. All of the moderating switching costs constructs and control variables in the right part of the model are found to be not significant and therefore left out of the model. The significant factors are displayed in table 4.

In the first step of the multiple regression willingness to switch was inserted and in the second step the option variable. This model is statistically significant F (100,51) ; p< 0,001 and explained 31,7% of variance in offer preference. In this model 2 out of 10 predictor variables were statistically significant. With the option variable (Option 1 simple and Option 2 complex) recording a higher Beta value (β -.547, p < 0.00) than willingness to switch (β .153, p < 0.00.

An additional interesting question to analyse is if people are willing to pay an additional charge for the showed conditions. This question is imbedded in the variable offer preference but since this is such an interesting component it was worth to focus separately on this. Since one half of the sample shows the simple conditions and the other half the complex conditions it’s nice to know if there is a significant difference. The results are F = 17.619 with p < .001 in the Levene’s test. So equal variances are not assumed across conditions. The results t (409.7) = -6.978, p < .001 show that respondents do perceive value in the offer conditions since they are willing to pay an additional charge for option 1 (M = 3.36, SD = 1.02) versus option 2 (M = 4.00, SD = .87). The graph (Graph 1) and means here are inverted, meaning that low scores should be interpreted as a higher agreement with the statement.

Hierachical regression model of offer preference

Adjusted R Square model = 31,3% R R² R² Change B SE β t

Step 1 Willingness to switch .132 .017** .017 .124 .032 .153*** 3.85 Step 2 Option variable .563 .317*** .299 -.928 .067 -.547*** -13.78 Note statistical significance ** p <.01 ; *** p <.001

Table 4: Regression output (second construct)

(27)

21

January 30, 2016

5.6 Selection of process model

In above mentioned linear regressions we concluded which variables have significant explanatory power in each of the models. Now the goal is to test the entire model with the relevant variables in process. Following from above linear regressions the conceptual model is reduced from 13 possible switching cost variables to 2 that matter. These variables only play a role in the relation between retainable and willingness to switch. The switching cost displayed at H2 are all lacking sufficient explanatory power. A visualization of the results is shown in figure 9.

Figure 9: Research model evaluated

All the not significant moderators removed leaves us with the model displayed in figure 10 left to test.

Figure 10: Significant research model

With this we need a model with two moderators before the mediator and one moderator after the mediator. This one moderator isn’t displayed in the research model but has to be incorporated in the conceptual model because it’s the choice option in offer preference. Half of the sample answered the questions on the right part of the model (red square), which are based on the simple proposition and the other half of the respondents answered based on a more complex proposition. Details of the two propositions are displayed in figure 11.

Figure 11: Bundle options

H1.1 Procedural switching costs Relevant H2.1 Procedural switching costs

• H 1.1.1 Economic risk costs Not significant • H 2.1.1 Economic risk costs • H 1.1.2 Switching barrier Dropped out in factor analysis • H 2.1.2 Evalution costs

Retainable Y/N H1.4 Willingnes to switch

H1.1 Procedural switching costs

• H 1.1.2 Switching barrier

H1.3 Relational switching costs

• H 1.3.2 Brand relationship loss costs

Retainable Y/N H1.4 Willingnes to switch H2.4 Offer preference

H1 Pre Switching costs

Option 1 (Simple option) Option 2 (Complex option)

- Bundel includes sms, minutes and mb’s - Separate bundle for mb’s and minutes/sms - Bundle including service numbers and foreign callsrate - Out of bundle is average 5 times in bundle - During contract free contract adjustments - Foreign calls are charged seperatly - Out of bundle same rate as in bundle usage - Service number are charged seperatly - Never unexpected out of bundle (capping) - Onlimited out of bundle calling - Price and quality comparable with your - Price and quality comparable with your

(28)

22

January 30, 2016

The correct process model (model 23) with the relevant variables indicated in the model. X = Retainable Y/N

Mi = Willingness to switch Y = Offer preference W = Switching barrier Z = Brand relation lost cost V = Options (Simple or complex)

Besides the variables displayed above we have 3 covariates with significant explanatory power.

These are age, gender and current provider. They will also be inserted in the model displayed in figure 12. The assumption is that X only has significant explanatory power over Y through M, since this is the Willingness to switch. This variable is moderated by the W (Switching barrier) and Z (Brand relation loss cost) components. In the relation between M and Y only the options variable has significant explanatory power.

5.7 Process Results

Being retainable (X) has a significant effect (p<.0001) on M. So for each point X changes M changes 1.239 points. Thus there is an effect. The same goes for W (p=.009). Z is only significant (p=.009) in interaction with retainable. Significant covariates in the first part of the model are Gender (p<.05) Age (p<.001) and current provider (p=.0001). In the second part of the model X isn’t significant anymore which means that X has an effect on Y but only through M which is in line with earlier assumptions. M has a significant effect (p=.023) on Y and the option variable is significant with (p=.002). The gender, age and current provider covariates lose their significance in the second part of the model.

Antecendent Coeff. SE p Coeff. SE p

X (Retainable) 1.239 .4585 .0072 -.0800 .0734 .2761 W (Switching barrier .3979 .1509 .0087 --- --- ---Z (Brand relation lost cost) .0170 .1936 .9302 --- --- ---M (Willingness to switch) --- --- --- .2341 .1072 .0295 V (Option) --- --- --- -.6662 .2146 .0020 i1 X*W -.1762 .0907 .0527 --- --- ---i2 X*Z -.3162 .1203 .0089 --- --- ---i3 M*V --- --- --- -.0832 .0650 .2008 Gender (Covariate) -.2264 .0959 .0187 -.0357 .0729 .6251 Age (Covariate) -.0417 .0125 .0010 -.0065 .0096 .4975 Current provider (Covariate) .5281 .1354 .0001 .0908 .1047 .3862 constant 2.659 .7979 .0009 3.897 .4517 <.0001

R²=.2439 R²=.3246 F=17.259, p <.0001 F=29.448, p<.0001

Consequent

M (Willingness to switch) Y (Offer preference)

Figure 12: Process model

Do proposition conditions matter and is this moderated by switching costs? : a quantitative study on the perception of customers