The willingness to pay for innovative product-service systems : testing personalization, innovativeness and firm effort

(1)

The Willingness to pay for innovative

Product-Service Systems

Testing Personalization, Innovativeness and Firm Effort

Amsterdam Business School

Christian Stork

10090428

Amsterdam, 04.07.2014

Master Thesis Topic: Willingness to Pay/Prdouct-Service Systems Program: MSc in Business Studies

Specialization: Entrepreneurship and Management in the Creative Industries (EMCI) Supervisor: Frederik Situmeang (Amsterdam Business School)

Academic Year: 2013/2014 Semester: 2nd

(2)

Abstract

A product-service systems (PSS) is an integrated solution of a product and a service to satisfy

the increasingly high demands of consumers. The importance of PSSs for companies has

therefore increased largely in the last decade. Furthermore, innovativeness, personalization

and the effort a company puts into delivering a service or manufacture a product are also of

high relevance in research in the last decade. Therefore, this wants to find out how a

consumer’s willingness to pay (WTP) changes with respect to the mentioned product and service features. This is done by auctioning a PSS at an auction platform that uses the

second-price sealed-bid or Vickrey auction system. The results show that personalization and

innovativeness communicated together in one ad has the highest WTP compared to the other

features. This goes in line with previous research. However, the other auction treatments that

tested personalization, firm effort and innovativeness alone, did either show no significant

effect or contradicting effects. In fact, communicating firm effort and communicating

innovativeness had a significant lower mean bid amount then the control treatment that

signaled nothing. These none supported findings have two main implications. First, literature

that found evidence that firm effort, personalization and innovativeness increase the WTP

may be questioned. Secondly, there needs to be further research done on the fairly new and

ambiguous concept of PSSs and how consumers value them with respect to different concepts

(3)

List of Tables and Figures

Tables

Table 1 – Treatments and their Cues 19

Table 2 – Age Sample compared to Age Dutch Population 22

Table 3 – Descriptive Statistics of Treatments with respect to Bid Amount (in € Cent) 25

Table 4 – Test for Normality – Auction Bid Data 27

Table 5 – Trimmed Descriptive Statistics of Treatments with respect to Bid Amount (in €

Cent) 28

Table 6 – Descriptive Statistics for the Control Questions 29

Table 7 – One-Way ANOVA Test Results – Above Median Percentile 31

Table 8 – Multiple Comparison Post Hoc LSD Test – Above Median Percentile 32

Table 9 – Linear Regression Analysis Bid Amount and Questions Treatment 1 34

Figures

Figure 1 – Mean Bid Amount 26

Figure 2 – Box Plot for each Treatment 26 Figure 3 – Histogram for Bid Amount 26 Figure 4 – Box Plot for Bid Amount 26

(5)

5

1 Introduction

Organizations, especially manufacturing companies, increasingly develop and deliver

product-service combinations. This development is due to the consumer’s rising demand for more

comprehensive solutions that contain both a product and a service element (Morelli, 2006). This

trend has also been recognized by many scholars in the past decade which resulted in a broad

range of research articles (e.g. Mont, 2012; Spring & Araujo, 2009). The most recognized and

attention drawing product-service combination is called Product-Service System (PSS). It has

been identified as a combination of one or more product and service elements that jointly fulfill

a user’s need while decreasing the environmental impact (Mont, 2012). The trend towards PSSs is associated with different factors. The most important one with respect to consumer products is the shift from society’s interest in solely functional fulfillment products and services to a

customer experience. Other factors are related to sustainability and reducing the ecological

footprint by longer life-cycles of the products through added services or by renting and leasing.

A successful example of the sustainability factor is the all-in one solution that Rolls Royce plc

offers. Instead of selling the aircraft turbines directly to airlines, Rolls Royce maintains

ownership and leases the engines. This includes spares and maintenance service and is paid by

the hourly usage of the turbines (Baines et al., 2007).

Companies that adopt the PSSs business model allow them to be more competitive and

add value to their offerings. This is due to quicker innovation processes through such solutions,

as they know their clients better. Furthermore, they can built better client relationships with such

solutions and strengthen customer loyalty as well as fulfil client needs more accurate through

personalization (Riemer & Totz, 2003; Tukker, A., 2004). As both, personalization and

(6)

6

they increase the WTP. In this context the WTP is the participant’s maximum price at which she

buy or demand one unit of a product or service (Mankiw, 2012; Varian, 1992).

A third concept that is incorporated in this research is firm effort. Different scholars

(e.g. Kruger et al., 2004; Morales, 2005) have identified that the effort a firm performs in order

to satisfy customer’s demands, may be a good indicator for the valuation of products and

services. That is, the more effort a firm provides while creating products and services and

communicating this to the consumer, the higher is the willingness to pay (Morales, 2005).

Despite the growing interest in PSSs and due its novelty, no research has empirically

tested how consumers perceive and value them. Furthermore, earlier research (e.g. Sandström,

2008) has only investigated WTP with respect to services but not PSSs. It is also fair to assume

that effort plays a big part in creating personalized PSSs for customers. Therefore, this research

wants to answer the following question:

To what extent does communicating the product and service elements,

personalization, innovativeness and firm effort, influence the willingness to pay for product-service systems?

The main objectives of this research paper are twofold. First, it empirically tests and further

investigates on the concept of PSSs from a customer perspective. Secondly, it gives implications

for product developers and marketing managers on how to design and promote PSSs in order to

achieve the highest WTP from customers.

To answer the research question the relevant literature about PSSs is reviewed first.

Morelli (2006) and Manzini and Vezzoli (2002) identified the need of how to design

(7)

7

personalization, innovativeness and the firm effort concept are reviewed. The literature review

leads up to several hypotheses that are tested in an experimental auction of a PSS. The

experiment set up is the Veylinx auction website of the University of Amsterdam. The platform

uses a sealed-bid, second-price auction to research the actual WTP. One experiment with five

treatments and real bidders are conducted to receive information on the willingness to pay for

PSSs. The product is auctioned by promoting them in different visual and textual set-ups. The

data obtained is analyzed and discussed, in order to give an answer to the research question and

(8)

8

2 Literature Review

This part of the research critically discusses the literature that has already been written on

product-service combinations, firm effort, personalization/customization and innovativeness.

First, the term product-service combination is broadly explained and then narrowed down to

PSSs. Secondly, the literature on personalization and customization is critically assessed. The

third part reviews the literature about innovativeness (newness). In the last part firm effort is

discussed.

2.1 Product - Service Combinations

The concept of product-service combinations has received much attention in the last decade.

Scholars (e.g. Mont, 2004; Tukker, 2006) have developed different frameworks and definitions

for product-service combinations. This is due to two main developments. First, companies are

increasingly looking for ways to differentiate and earn additional revenues by offering joint

solutions of services and products to their customers. Most companies do initially not offer

product-service combinations but they are growing into it (Mont, 2004). Traditional

manufacturing firms now offer services because it can be beneficial for them, with respect to

increased revenues and more predictable cash inflows. By offering services they can lock

customers into long-term relationships (Vandermerwe & Rada, 1988; Tukker, 2004).

Furthermore, product-service combinations may reduce costs for both the producer and the

consumer (Penttinen & Palmer, 2007; Gebauer and Friedli, 2005) or increases (perceived) value

(Ulaga & Reinartz, 2011). Secondly, consumer demands have changed from the need of solely

functional fulfillment of products and services to customer experiences. The end users of today

do not only want to see and use the products or services, but they want to have a whole

(9)

9

product or service because they can derive more pleasure from it (Mont, 2004). However, not

only product manufacturing firms can benefit from the trend towards entire product-service

solutions. Service firms can bundle products and services by replacing service providers with

products such as robots (Oppedijk van Veen, & Schoormans, 1999). However, PSS are not only

beneficial for companies to differentiate themselves and earn extra revenues but also for the

consumer who is the focus in this research. Especially the service part of the PSS can be

personalized to the individual demand of the consumer. This gives the consumer more flexibility

and choice which in turn increases satisfaction. Companies can do this by collecting data about

how the customer uses the PSS and can thereby continually improve the offering (Aurich et al.,

2010). A more in depth review of personalization is presented in section 2.2.

As mentioned at the beginning of the literature review scholars have identified different

product-service combinations. Spring and Aurajo (2008) differentiate from an operations

management, marketing and economics point of view, between six product-service

combinations namely: support services, systems integration, performance-based logistics,

bundling, offerings and finally PSSs. The concept of product-Service systems (PSSs) has

received the most attention among the six combinations. Goedkop et al. (1999) defines it as the

‘marketable set of products and services, jointly fulfilling a client’s need’ (p. 3). However, due to its novelty, there is no generally accepted definition of PSS (Mont, 2004). The essential part

of PSSs is the integrated combination of products and services. Most PSS literature is concerned

with the decrease of environmental impact because PSSs may include a change of ownership

and rental and lease agreements that increase a product’s life cycle (Mont, 2002). Baines et al.

(2007) go beyond the definition and purpose of the sole reduction of environmental impact and

(10)

10

2007, p. 1). On top of that, Manzini & Vezoli (2002) describe PSSs as an innovative strategy

that integrates products and services that together fulfill client demands. The relation to

innovation is further reviewed in section 2.3. This notion is used in this article because the

research investigates consumer products from a customer perspective with. It further focuses on

service elements where the main concern is not on environmental issues but on the integration

and joint customer experience of a product combined with a service.

2.2 Personalization

First of all, the concepts of personalization, customization and customerization (Wind &

Rangaswamy, 2001) have to be clarified, as there exist different definitions for each concept in

the literature. Secondly, the differences between customization, customerization and

personalization are discussed and the concept of personalization, which is used for this research,

is introduced. Subsequently, its benefits and role towards PSSs is outlined.

All three concepts are about individualized offers for the customer. Customizations or

mass customizations are offerings by which the company provides a range of possible

configurations from which the consumer can choose (Wind & Rangaswamy, 2001). Hart (1996)

adds another component to the concept. He says that the production processes are flexible in

order to adapt to the individual offerings. On top of that, the organizational structure must go in

line to be able to produce these individualizations. Mass customization implies that the

customized products and services have to be produced at the costs of standardized

mass-produced products and services in order for the company to reach a competitive advantage.

Examples range from mass-customized cars to mass-customized shoes and apparel in general.

The common note in most customization and mass-customization literature is the fact that

(11)

11

means that mass-customization is performed by the customer (e.g. Imhoff et al., 2001; Allen et

al., 2001; Cöner, 2003). In contrast to that stands personalization. It is the individualization of

offerings through information collected about the consumer. This means that companies initiate

personalization (e.g. Cöner, 2003). However, not all authors agree with these explanations of the

concept and differentiations between personalization and customization. Wind & Rangaswamy

(2001) argue that personalization can be either company initiated or customer initiated. Customer

initiated personalization include personalized web content whereas company initiated

personalization is more related to individualized offerings. Allen et al. (2001) and Imhoff et al.

(2001) agree with the notion that personalization is mainly related to the individualized content

and experience that the company tailors according to user information and data. However, they

also argue that customization is web content related but the only difference is that it is initiated

by the customer. These differences in defining the concepts of personalization and

mass-customization show, that there are two different dimensions. First, the direction of initiation

(firm vs. customer) and secondly the scope (content related vs. entire offering).

The third concept that has to be differentiated is customerization. It was first mentioned

by Wind & Rangaswamy (2001) and introduces the next stage that builds upon customization. It

essentially includes more functions and activities than mass-customization. That is, companies

do not only produce products and services that are individualized for and by the consumer but

the whole value chain of activities is customized. Previously, companies offered different

options, so that consumers could put together their own, preferred product or service which is

called mass-customization. With customerization this offering has changed. Now the whole

process is consumer-centric. This goes from marketing, specifically tailored towards individual

customers to the end product and delivery. It is important to note that companies do not need to

(12)

12

offers customerized solutions in the nursery sector. Customers can create a profile and arrange a

garden of their liking from the scratch, including such things as the consideration of their local

climate. The website also gives useful insights into gardening through an encyclopedia. After

completion, the customer orders the products and garden.com arranges an on time delivery of all

products at the same time through its network of 100 suppliers. This example shows that

customerization is a buyer-centric strategy. After having identified and explained the three

different concepts, personalization is further outlined and the definition which is used in this

research is given.

Due to the rapid sophistication of internet technology and the need for companies to

target customers and deliver products and services more cost-effective, personalization has

increasingly received a lot of attention from scholars and managers alike. Furthermore, managers

have realized that competition, especially through e-commerce, is increasing due to similar

offerings and price transparency. Consumers do not want one-fits all products and services but

have highly heterogeneous demands. These preferences for heterogeneity change quickly in

consumer’s markets (Franke & Piller, 2004). A study by Franke, Hoppe & Reisinger (2009)

found significant evidence that consumers are dissatisfied with standardized offerings even in

matured markets. Therefore, the need to personalize products and services is essential for

businesses in order to increase customer satisfaction, loyalty and stay successful. Through the

personalization of products and services, companies are less comparable and can differentiate

themselves from competitors, to achieve a unique position in the market place (Riemer & Totz,

2003). As mentioned earlier in the literature review there exist different definitions of

personalization. One point all authors agree with, is that companies collect data to individualize a

product, content or a service accordingly. This is often the case for PSSs as they aim to give the

(13)

13

such as the iPod that is connected to iTunes which allows for example for personalized playlists,

this research defines personalization as both company and user driven.

2.3 Innovativeness

Innovation has been widely discussed in the literature over the last decades. The complexity and

ambiguity of the concept has led to an abundance of definitions for the terms innovation and

innovativeness. Furthermore, different types of innovation have been identified ranging from

imitative to radical innovations. Calantone & Garcia (2002) developed a common definition

from different types of literature, such as micro- and macroeconomics as well as from

marketing, management and engineering literature. It describes innovation in two parts. First,

the innovation process is a combination of the technological development of an invention, and

introducing this to the market through adoption and dispersal. Secondly, the process of

innovation is gradual which means that an innovation is introduced and then replaced by a new,

improved innovation that builds upon the old one. In the same research paper, the authors define

innovativeness as the degree of ‘newness’ of the innovation. That is, innovations with a high degree of ‘newness’ are more innovative than products or services with a low degree of ‘newness’.

2.4 Company Effort

A study by Kruger et al. (2003) found evidence that consumers perceive a product or service of

higher quality and would be willing to pay more when a company primes its effort. For instance,

if artists claim it took them 26h to finish the painting compared to claiming it took 5h, people

would most likely pay more for the painting that took 26h. However, ambiguity about the

(14)

14

quality of the product they value firm effort higher (Kruger et al., 2003). Morales (2006) did a

similar study to find out whether advertising effort increases the WTP. The research gave

similar findings, namely that consumers reward companies that put extra effort in producing and

delivering products and services. However, this only holds if people feel gratitude as the

gratitude is the driving force for higher a WTP and an overall better evaluation of the company.

This is also shown by the research as this is true even though the consumer does not expect a

better quality due to the extra effort. When consumers infer that companies only signal extra

effort to persuade people to buy their product or service, consumers do not reward this behavior

by a higher WTP. Another research about priming effort was conducted by Cho & Schwarzer

(2008). They investigated whether priming talent would influence how people perceive the

product and ultimately how much they would be willing to pay. The researchers found that if

people get primed with the high talent of for example a painter, signaling low effort increases

perceived value as the consumer infers that a talented painter should not need as much time to

finish a painting then a not so talented painter. This shows that depending on the situation and

the primes, people derive different valuations. Transferred to an offering buy a company, this

could mean that people are not willing to pay more if for instance a well-regarded service firm

advertises extra effort. However, this study builds its hypothesis on the research by Kruger et al.

(15)

15

3 Theoretical Framework and Hypotheses

This section discusses the three concepts with respect to WTP and outlines the hypotheses. The

WTP is the monetary equivalent to how much a consumer values a certain product or service

(Mankiw, 2012). With respect to personalization Ball, Coelho & Vilares (2006) found evidence

that personalizing a service increases loyalty. This means if customers are offered a personalized

service they are more likely to come back to the same company. Furthermore a study by

Homburg, Koschate, Hoyer (2005) revealed that there is a strong positive correlation between

loyalty and the WTP. That is, customers that showed loyalty to the company in the experiment

also indicated a higher WTP. Therefore, the first hypothesis is as followed:

H1: Customers’ willingness to pay more for PSSs increases if personalization is communicated compared to not communicating it.

PSSs are an innovative way for both traditional manufacturers but also new companies to

deliver customer value and gain a competitive advantage of their competitors. The reason for

that is the fact that PSSs are new offerings and strategies that integrate products and services to

satisfy the demanding customer needs (Manzini & Velozzi, 2002). A study by Szymanski, Kroff

& Troy (2007) found that on average a higher degree of newness and product performance is

positively related. However, the effect is small to medium and therefore they conclude that it

doesn’t show evidence whether innovativeness is a key driver for product success. A recent study by Schreier, Fuchs & Dahl (2012) found that consumers show higher loyalty, more

enthusiasm, higher purchase intentions and are willing to pay more for products of companies

that they perceive as innovative. Therefore, the second hypothesis is as follows:

H2: Customers are willing to pay more for PSSs that communicate innovativeness compared to not signaling innovativeness.

(16)

16

Previous studies (Franke & Hippel, 2003; Franke & Piller, 2004) investigated the effect of

customizable, innovative toolkits on customer satisfaction and the WTP. They found evidence

that a combination of both increases consumers purchase intention and ultimately their WTP for

the product. Taking these studies into account and building up on hypothesis one and two, the

following is hypothesized:

H3a: Customers are willing to pay more for PSSs that communicate personalization and innovativeness compared to PSSs that communicate nothing.

H3b: Customers are willing to pay more for PSSs that communicate personalization and innovativeness compared to PSSs that only communicate personalization.

H3c: Customers are willing to pay more for PSSs that communicate personalization and innovativeness compared to PSSs that only communicate innovativeness.

As described in the literature, (e.g. Imhoff et al., 2001; Allen et al., 2001) personalization comes

from the company to tailor offerings and content according to the consumers data and

preferences that were obtained by the enterprise. This implies that the company makes efforts to

deliver the best possible fitted product or service, according to the consumer’s information.

Krugers’s et al. (2003) study only investigates perceived value of higher effort with respect to products. Nevertheless, it is fair to assume that this also holds for services. Perhaps customers

value service effort even higher because services often involve direct interaction between the

provider and the consumer so the effort may immediately be noticed. The following hypothesis

reflects the findings by Kruger (2003):

H4: Customers are willing to pay more for PSSs that communicate firm effort compared to no firm effort.

(17)

17

4 Methodology

This section describes how the research is conducted. First, the research design is outlined. The

second part explains the design of the experiment, the respondents and the different treatments.

The last part discusses the variables and measures that are used to run the experiment.

4.1 Research Design

The purpose of this research is to find a causal relationship between personalization,

innovativeness/newness and firm effort on the one hand and the willingness to pay for PSSs on

the other hand. A research that investigates the causalities between two or more variables is

called an explanatory study (Saunders et al, 2009, p. 140). Since the purpose of this study is to

find whether consumers are willing to pay more for PSSs that indicate firm effort, innovation

and/or personalization, the former (i.e. willingness to pay) is the dependent variable and the

latter (i.e. firm effort, personalization and innovativeness) are the independent variables. In

order to test the hypotheses one experiment is conducted that provide the data for further

analysis. The set-up is the auction platform Veylinx which was developed by the University of

Amsterdam. There are currently around 5400 users of which 70% are 30 years and older in order

to ensure that the sample is more representative for the whole population. One product is

auctioned and presented to the participants of the auction in different ways, to see whether there

is an effect of firm effort, personalization and innovativeness on the willingness to pay. After all

the data is collected, it is analyzed and results are drawn.

4.2 Method

The experiment will be conducted using the online auction platform Veylinx. With this platform

(18)

18

experiment in which a product is presented in different ways to different bidders with a panel

size of around 5400 members and an average responds rate of 25%. The auction platform is

based on the second-price sealed-bid or Vickrey auction. With this type of auction, every bidder

places one bid and does not know what the others placed. The person with the highest bid pays

the price equal to the second-highest bid (Vickrey, 1961). This type of auction is especially

suited for individual demand in experimental situations. However, it does not immediately

inform about the WTP for products and services (Coursey, Hovis & Schulze, 1987). Coppinger,

Smith & Titus (1980) describe in their paper that the best strategy to reveal the true WTP of

participants is to first have trial runs of auctions. However, due to time constraints this is not

possible in this study. Therefore, every participant is thoroughly informed about the auction

process and conducts two trial auctions to get familiarized with the platform.

4.3 The Auctioned Product

The PSS that is auctioned combines a health tracker called H2O-Pal that measures the daily

water consumption of a user, with a personalized app that gives further information. In this case

the health tracker is the product and the personalized app the service. The product has been

chosen because it is a PSS that is not on the market yet and is highly innovative which gives the

right prerequisites for the research. To conduct the experiment, each member that has signed up

at Veylinx receives an invitation to the auction and if participating, is presented with one of four

treatments that are explained below. Participants have six minutes in order to place a bid, with

the highest amount they are willing to pay. After that, they are asked to fill in a short survey of

(19)

19

4.4 Overview of the Experimental Design and Treatments

The research wants to test the whether there is a causal relation between company effort,

innovativeness or personalization and the WTP for Product-Service-Systems. Table 1 shows the

different treatments and their cues.

Table 1

Treatments and their Cues

Treatment Personalization Firm Effort Innovativeness No Cues

Nr. 1 (Baseline) x

Nr. 2 _x

Nr. 3 _x

Nr. 4 x

Nr. 5 x x

There are 5 treatments in total. Each participant is randomly assigned to one of the five

treatments. That is, one group of participants is presented with a screen that promotes the PSS in

one of the five ways, a second group is presented with a screen that promotes the PSS with

another cue and so on. Each case in Table 1 represents one treatment. Participants that are

assigned to treatment 1 are presented with a screen that only describes the basic features of the

PSS. The participants assigned to treatment 2, see a screen that highlights the product

development and research time. Bidders designated to treatment 3, view a screen that highlights

the personalization features of the PSS. The group that is assigned to treatment 4, is presented

with an ad that highlights the innovativeness of the PSS. Finally, participants assigned to

(20)

20

regarding personalization and innovativeness. All five treatments and the respective

advertisements are shown in the Appendix B.

Additionally, after the participants placed their bids, they are asked to fill in a short

survey in order to finalize the auction. The following 4 questions were asked (see Appendix B

for original questions in Dutch):

1. Do you find it important to know the development time of a product? a. Strongly agree

b. Agree c. Neutral d. Disagree

e. Strongly disagree

2. Are you familiar with this product? a. No

b. Yes

3. Do you find the product innovative? a. Strongly Agree

e. Strongly Disagree

4. Do you value products more that can be personalized? a. Strongly agree

e. Strongly disagree

Question 1 is asked to see whether there are any differences or similarities between the bidding

of participants that are assigned to the firm effort ad and the answer they give. This may give

stronger support for hypothesis 4. Question 2 is used to observe any bias that might occur if

(21)

21

whether participants that are presented with the innovativeness are coherent with their placed bid

and their perception of innovativeness, which may give additional support for hypothesis 2.

Furthermore, it may also give support for hypothesis 3a and/or 3b if people placed a relatively

high bid and find the product innovative. Question 4 may also give additional support for

hypothesis 3a and/or 3b if participants that placed a high bid at the innovativeness ad, indicate

that they value the personalization feature of a product. Furthermore, question 4 may also give

further support for hypothesis 1.

4.5 Data Collection and Respondents

In order to use the Veylinx platform, every student had to recruit either 100 new members if the

product was not provided for free or 50 new members if the product was provided for free. After

contacting the founders of the H2O-Pal, one sample for each treatment will be provided as soon

as the product is launched. This means that 50 panel members were recruited through online

social networks. To give a sample that does not only represent students, as it is the case in many

research experiments, 70% of the recruited members had to be over 30 years old. Since this is

the requirement for each student using Veylinx to conduct experiments, the average member age

should be over 30 years old. At the time of the auction, approximately 5400 Dutch people had

signed up. For all panel members that signed up, the participation in an auction is still voluntary.

After running the auction for two consecutive days 689 (12,7%), members placed a bid. This

relatively small number may be due to the fact that the auction was indicated as a concept

auction which means that the product is not available yet. Table 2 shows the participation age

compared to the Dutch population with respect to age. This is divided in <20, 20-40, 40-65 and

>65 in order to see whether the sample is representative. Three outliers regarding age were

detected and removing these outliers gives an average age of 41,9 years for all participants.

(22)

22

Dutch population (CBS, 2013). However, when looking at the different age intervals one can

observe that there are differences. The under 20 group (1,2% compared to 23,1%) is

underrepresented in the sample data whereas the 20-65 group (91,3% compared to 60,1%) is

overrepresented. This could be due to the fact that 70% of the recruited members have to be

over 30 years old and that there are few young people (<20) in the social networks of the

students that use Veylinx. However, considering that the auction was about a consumer product

that targets the age groups of 20-65, the sample can be assumed as representative. Furthermore,

405 men (58,8%) and 284 women (41,2%) participated in the auction. The relative number

differs slightly from the 49,5% men and 50,5% women of the Dutch population (CBS, 2013).

This might be due to the fact that more male (53%) then female (47%) subjects have signed up

at Veylinx or that the product appeals more to men than to women.

Table 2

Age Sample compared to Age Dutch Population

Age in Years <20 20-40 40-65 >65 Sample (N=686) 1,2% 44,4% 48,2% 6,2% Dutch Population 23,1% 24,6% 35,5% 16,8%

4.6 Variables and Measures

The research tests the effect of communicating company effort, personalization and

(23)

23 3.6.1 Independent Variables

The independent variable ‘firm effort’ is measured by using the R&D time of the PSSs. That is, the R&D is either not indicated or with 1,5 years. ‘Personalization’ is measured by either

indicating that the PSS has a personalization feature or not. The last independent variable

‘innovativeness’ is measured by whether the PSS is advertised as innovative or not.

Furthermore, all three independent variables are each included in one of the questions after the

participants placed a bid and are measured on a 5 point Likert Scale with 1 being ‘Strongly agree and 5 being ‘strongly disagree’.

3.6.2 Dependent Variable

The dependent variable in the research is the WTP. This is measured by the auction bid of each

participant because this way, each bid relates as closely as possible to the actual floor

reservation price the participants would give (Wang, Venkatesh & Chatterjee, 2007).

3.6.3 Control Variable

To control for the eventuality that participants are familiar with the auctioned PSS, they have to

answer the question at the end of the auction whether they are familiar with it or not. This

(24)

24

5 Results

The last section described the methodology that is used in this research. The following section

gives the results that were obtained by analyzing the data with SPSS. First, some descriptive

statistics and correlations are given. Due to the messiness of the data, trimming was conducted

to arrive as close as possible at a normal distribution. After that, the results of the ANOVA test,

the regression analysis and the chi-squared test that were conducted are stated and explained.

5.1 Descriptive Statistics

5.1.1 Bid Amount

This section gives an overview of the most important descriptive statistics and lays the

foundation for further analysis. On top of that, the assumptions of the parametric tests that are

used subsequently to test the hypotheses are tested. That is, the assumption of normality and

homogeneity of variances is explored. The central limit theorem states that if the sample data are

approximately normal, the sample distribution will also be normal. A common rule of thumb is

that if a sample size of N>=30 is given, normality can be assumed. Even though the sample size

of each treatment is above 30, it is still useful to test for normality (Field, 2009).

First, the statistics for checking whether participants were familiar with the PSS were

obtained. The results show that 643 out of 689 users answered the question. Ten participants

were familiar with the PSS. Due to the low number, it can be concluded that the auction bids are

not influenced by users that know the product.

Section 3.6 showed that the sample is not fully representative of the Dutch population

but can still be a good representation of the group (20-65 years old) that most likely buys the

(25)

25

the five treatments with respect to the bid amount. All numbers that are related to the bid

amount are indicated in cent. 689 participants placed a bid in total, of which 266 placed an

amount of 0. The average bid amount is 559,71 whereupon the lowest bid is 0 and the highest

bid is 7500. The results show that treatment 5 has the highest average bid amount (M = 639,75)

as well as the highest maximum bid amount (Max. = 7500) but the lowest total amount of bids

(N = 123). Treatment 1 received the most bids (N = 158) whereas treatment 4 had the lowest

average bid amount (M = 502,40) and treatment 3 had the lowest maximum bid amount (Max =

3500). In general it can be seen that all treatments have a relatively high standard deviation with

an average of 825,718. Figure 1 visually shows the mean bid amount for each treatment and

figure 2 shows the box plots for each single treatment. It can be observed that there are multiple

outliers in each treatment.Additionally, all treatments are heavily skewed to the right and as

Table 3

Descriptive Statistics of Treatments with respect to Bid Amount (in € Cent)

Variable N Mean Std. Error Mean Std.

Deviation Min. Max.

Treatment 1 158 596,61 63,165 793,974 0 4500 Treatment 2 127 541,78 72,854 821,019 0 4500 Treatment 3 145 521,06 58,882 709,031 0 3500 Treatment 4 136 502,40 68,347 797,058 0 5000 Treatment 5 123 639,75 91,384 1013,497 0 7500 Treatments Total 689 559,71 31,457 825,718 0 7500

(26)

26

mentioned before, have a high amount of 0 and low bids respectively. Therefore, the assumption

of normality is visually and numerically explored (Field, 2009). Figure 3 shows the histogram

with the normal distribution curve of the bid amounts and figure 4 the box plot. Both figures

confirm the high amount of 0 and low bids. The box plot also shows multiple outliers that

influence the distribution from being normal. To test whether the distribution is normal with

numbers, a Kolmogorov-Smirnov Test for each treatment and the treatment total is conducted

and the skewness and kurtosis is reported (Table 4). It can be seen that all statistics are

significant (p < 0,001) which means that normality cannot be assumed. A normal distribution

Figure 1 .Mean Bid Amount Figure 2. Box Plot for each Treatment

(27)

27

***Significant at the 0,001 Level

would have values of zero or close to zero for the skewness and the kurtosis (Field, 2009).

However, for all treatments there is a relatively heavy skewness and kurtosis. Most of the data is

to the right of the distribution and therefore it is said to be positively skewed. The positive

values of the kurtosis indicate a spiky and heavy-tailed distribution. These observations are

coherent with the visual observations. Therefore, based on these findings normality cannot be

assumed.

Another assumption that may be tested in order to run parametric tests such as ANOVA

is the homogeneity of variances. That is, if going through the levels of one variable the

variances of the other variables may not change. This can be tested using the Levene’s test.

Based on the this test, homogeneity of variances can be assumed because the test statistic is

non-significant.

In order to normalize or bring it closer to normality, the data can be trimmed (Field,

2009). This means, that part of the data will be left out. The observation of the histogram and the

box plots above showed that the data is heavily skewed to the right due to the high amount of

zero or low bids. Furthermore, the multiple outliers may also impact the results. For that reason

Table 4

Test for Normality – Auction Bid Data

Variable N

Kolmogorov-Smirnov

Shapiro-Wilk Skewness Kurtosis

Treatment 1 158 0,226*** 0,761*** 1,953 4,890 Treatment 2 127 0,255*** 0,709 *** 1,981 4,401 Treatment 3 145 0,236*** 0,750*** 1,822 3,204 Treatment 4 136 0,264*** 0,667*** 2,767 10,101 Treatment 5 123 0,264*** 0,656*** 3,170 16,518 Treatments Total 689 0,249*** 0,707*** 2,526 10,493

(28)

28

all detected outliers (bid_amount<2223) will be left out. Additionally, due to the high amount of

low values only 50% of the data above the median are used, so that the means are calculated

from the remaining sample data. Table 5 shows the descriptive statistics for the trimmed means.

The new, trimmed statistics can be compared with the old ones. It can be seen that the mean, the

most important statistic for the subsequent ANOVA test, has increased in all treatments and the

treatments total. However, the ranks of the different treatments with respect to the bid amount

have not changed. That is, treatment 5 has still the highest mean (M = 996,59) and treatment 4

the lowest mean (M = 784,43). In fact, when calculating the percentage change of bid amount

for each treatment and the treatments total, the increase in bid amount ranges between 48,5%

(treatment 2) and 64% (treatment 1). A similar effect can be observed for the number of

participants for each treatment. Treatment 1 (N = 71) has still the highest number and treatment

5 (N = 54) the lowest.

Table 5

Trimmed Descriptive Statistics of Treatments with respect to Bid Amount (in € Cent)

Variable N Mean Std. Error Mean Std.

Deviation Min. Max.

Treatment 1 71 978,90 59,640 502,539 300 2100 Treatment 2 54 804,56 72,825 535,149 100 2099 Treatment 3 66 852,44 63,287 514,143 200 2100 Treatment 4 61 784,43 61,211 478,071 200 2000 Treatment 5 54 996,59 85,086 625,254 200 2000 Treatments Total 306 885,21 30,498 533,491 100 2100

(29)

29

To test the new data set for normality, the Kolmogorov-Smirnov test was conducted again but

the results still showed significant statistics. Therefore, based on the repeated test for normality

it cannot be concluded that the data is normally distributed. However, the ANOVA test is said to

be a robust test which means that the test can still be run even though the data might not have a

normal distribution (Field, 2009).

5.1.2 Control Questions

Further statistics were calculated for the control questions that were measured with a Likert

scale at the end of the auction (Table 6). The differences in the sample size is due to the fact that

participants sometimes do not answer the questions on Veylinx or only answer a few of the

questions. Since the mean is relatively unimportant when analyzing a Likert Scale, only the

Mode and the percentages of the chosen answers are given (Field, 2009). It can be observed that

1,6% of all participants that answered the question are not familiar with the product and

therefore it can be assumed that their actual WTP is revealed. Additionally, the other three

Table 6

Descriptive Statistics for the Control Questions N Mode Likert Scale 1-2 (in %) Likert Scale 3-5 (in %) Product Familiar 643 0 1,6 98,4 Firm Effort 641 3 15,8 84,2 Personalization 641 3 33,7 66,3 Innovativeness 638 3 47 53

(30)

30

questions show that more participants are either neutral towards firm effort, personalization and

innovativeness or do not find those aspects important at all.

5.2 Correlations

Before testing the hypotheses the most interesting and significant correlations are outlined and

explained. The Pearson correlation coefficient matrix including means and standard deviations

can be found in Appendix A.

There is a weak, positive correlation (r = 0,164; p < 0,01) between age and the question

whether participants find the product innovative. This implies that younger users perceive the

product as more innovative then older users. Furthermore, there is a positive correlation (r =

-0,254; p < 0,01) between age and bid amount. That is, older people placed on average higher

bids than younger people. With respect to gender it can be seen that men compared to women,

find it slightly more important to have a product or service that can be personalized (r = 0,116; p

< 0,05). Interestingly, participants that perceived the PSS as innovative also find it important to

be able to personalize it (r = 0,212; p < 0,01) and to know the R&D time (r = 0,189; p <0,01).

Additionally, there is a positive correlation ( r = 0,315; p < 0,01) between the answers of the

development time question and the personalization question. This indicates that users who

appreciate knowing the development time also prefer products and services that can be

personalized. With regards to bid amount it can be observed that bid duration is positively

correlated (r = 0,126; p < 0,05). That is, participants that spent more time on placing a bid

(31)

31

5.3 One-Way ANOVA Test

In this section the four hypotheses are tested. To test for significant differences, a one-way

ANOVA test is used that measures the differences in the mean bid values of the different

groups. To conduct a parametric test such as ANOVA, the assumptions of normality,

homogeneity of variances, the observations are independent and the data has to be measured at

an interval level (Field, 2009). The latter two assumptions are given due to the nature of the

experiment conducted and the homogeneity of variances has been tested in the previous section.

However, the normality assumption was visually and numerically not confirmed. Nevertheless,

it is still possible to use

*Marginally significant at the 0,1 level

the ANOVA to test the data because it is said to be a robust test (Field, 2009). Table 7 shows the

results of the ANOVA test. The outcome of the test shows a marginally significant result (F(4,

305) = 2,080; p<0,1). This means, that there is some variation in the means between the five

treatments. However, the ANOVA only tests whether there is variance in the means between

any group. Therefore, in order to test the hypothesis a post hoc test has to be carried out to see

where the differences in the means of the bid amounts lie. Table 8 shows the results of the LSD

post hoc test for significant results.

Table 7

One-Way ANOVA Test Results – Above Median Percentile

Bid Amount in Top 50 Percentile Sum of Squares df Mean Square F

Between Groups 2334927,337 4 583731,834 2,080* Within Groups 84471943,86 301 280637,687

(32)

32

*Marginally significant at the 0,l level ** Significant at the 0,05 level

There is a marginally significant difference between the bid amount of the baseline treatment

and the firm effort treatment 2 (MD = 174,346; p < 0,1). This means that the mean bid amount

of the baseline treatment is higher than the mean bid amount of treatment 2. Therefore, it can be

concluded that H4, stating that communicating firm effort will increase the WTP, is rejected.

Furthermore, there is a significant difference between the baseline treatment and the

innovativeness treatment 4 (MD = 126,462; p < 0,05). Therefore, H2, stating that

communicating innovativeness will increase the WTP, is rejected. There happen to be a

marginally significant difference (MD = -192,037; p < 0,1) between the firm effort treatment

and treatment 5 (innovativeness + personalization), however, this was not part of any

hypotheses. The last significant difference (MD = -212,166; p < 0,05) that was observed is

between treatment 4 (innovativeness) and treatment 5 (innovativeness + personalization). This

means that H3c, stating that communicating innovativeness and personalization will have a

higher WTP, compared to only communicating innovativeness, is supported. Since there has

Table 8

Multiple Comparison Post Hoc LSD Test – Above Median Percentile

Bid Amount in Top 50 Percentile (I) Treatment (J) Treatment

Mean Difference (I-J) Std.Error 1 (Baseline) 2 174,346* 95,654 4 126,462** 90,580 2 5 -192,037* 101,951 4 5 -212,166** 98,983

(33)

33

been no significant difference between the other treatments, H1, H3a and H3b are rejected. The

next section analysis the answers from the four control questions that were asked at the end of

the auction.

5.4 Regression Analysis

The following part shows the regression analysis. Although only one hypothesis (H3c) was

supported there could still be an effect whether participants value firm effort, innovativeness and

personalization. This is tested with the baseline treatment because this is the only treatment that

did not contain any cues and therefore does not influence the subsequent answers.

Before running the linear regression analysis it is useful to look at the correlations

visually. Figure 5 shows the scatter plot for treatment 1 with bid amount as the dependent

variable and the firm effort scale as the independent variable.

(34)

34

It can be seen that there is a negative relationship between bid amount and the firm effort Likert

Scale (1 strongly agree – 5 strongly disagree). Based on the correlation, this implies that on

average people that were presented with the baseline treatment, value firm effort more when

they placed higher bids. To further investigate this relationship a linear regression analysis is

conducted. Table 9 shows the regression analysis for the baseline treatment and the three

questions related to firm effort, personalization and innovativeness respectively.

*** Significant at the 0,01 level (1-tailed)

In total 69 participants placed a bid at treatment 1 and answered the questions. The results show

a significant, negative correlation (r = -0,348; p < 0,01) between bid amount and the question

about firm effort. This indicates that on average people that placed a relatively high bid also

indicated that they value firm effort. Multiplying R-Squared (=0,121) with 100 gives the

percentage (12,1%) of variation due to firm effort. That is, 12,1% of the variation in bid amount

can be accounted to firm effort. Furthermore, the F-ratio indicates the accuracy of the predicted

Table 9

Results - Linear Regression Analysis Bid Amount and Questions Treatment 1 Variable (Questions) N Pearson's Correlation R-Squared F-Ratio

B-Value Std. Error t-value

Firm Effort 69 -0,348*** 0,121 9,262 1457,061/ -133,341 165,293/43, 815 8,815/ -3,043 Personalization 69 -0,080 0,006 0,437 1093,979/ -41,802 175,167/ 63,263 6,245/ -0,661 Innovativeness 69 0,027 0,001 0,049 948,688/ 15,792 177,106/ 71,194 5,357/ 0,222

(35)

35

outcome compared to the inaccuracy of the model. A relatively large F-ratio (F > 1) predicts a

fairly accurate model. The results show an F-ratio of 9,262 (p < 0,01) so it can be concluded

that there is less than a 0,01% chance that an F-ratio of that size would appear if the null

hypothesis was true. The second B-value (B = -133,341) shows the gradient of the regression

line. Therefore, for each point going up on the Likert scale the bid amount goes down by 133,341 € cent. Both the personalization (r = -0,080; p > 0,05) and the innovativeness (r = 0,027; p > 0,05) question showed no significant correlation with bid amount in the baseline

treatment.

5.5 Chi-Squared Test

The last part of the results section shows the control analysis for several variables. The different

cues regarding firm effort, personalization and innovativeness are controlled for and compared

to the answers of the questions, to see whether the cues influenced the answers.

The first chi-squared test analyzed whether participants that were exposed to the firm

effort treatment indicated higher levels of agreement at the control question about the

development time. Since the questions could be answered by every participant all zero bids have

been included. The sample number may differ because of some participants only answering

some questions or none at all. The test for independence revealed that there is no significant

difference between participants of the different treatments and their answer of the question ( (4; N = 609) = 7,908; p = 0,095). That is, the distribution of participants that value firm (‘strongly agree’ and ‘agree’) is equal over the different treatments and participants of the firm effort treatment were therefore not influenced by the cues. The same Chi-Squared test was

conducted for personalization and innovativeness as well. The test for personalization revealed ( (4; N = 654) = 3,194; p = 0,526) that the amount of participant that answered the

(36)

36

personalization question with ‘strongly agree’ or ’agree’ is equally distributed among the five treatments. Therefore, participants in the personalization treatment did not get influenced by the cue. The results for the innovativeness question ( (4; N = 609) = 7,908; p = 0,095) also

showed no significant difference between the five treatment and therefore it can also be assumed

(37)

37

6 Discussion

The following section discusses the results from the previous analysis. First the main findings

and the supported and not supported hypotheses are discussed. This is then compared to the

existing literature and followed by theoretical and managerial implications. Lastly, limitations

and suggestions for future research are given.

6.1 General Discussion

It was the purpose of this research to find out how consumers react to different advertising cues

and whether this translates in to differences in the WTP with respect to PSSs.

The first look at descriptive statistics showed that opposed to what was hypothesized, the

baseline treatment had the second highest mean and trimmed mean bid amount (M = 596,61; M

= 978,90). The treatments for firm effort (M = 541,78; M = 804,56), personalization (M = 521,06; M = 852,44) and innovativeness (M = 502,40; M = 784,43) all show a lower mean bid

amount and therefore a lower WTP than the baseline treatment.The only treatment that showed

a higher mean and trimmed mean bid amount (M = 639,75; M = 996,59) was treatment 5

(personalization + innovativeness).

These first observations were then partly confirmed by the subsequent ANOVA analysis.

A marginally significant effect (p < 0,01) was found between the baseline treatment and

treatment 2 (firm effort). However, in contrary to what was hypothesized, the average bid

amount of the baseline treatment was 174,34 € cent higher than the firm effort treatment. This result partly contradicts with the findings of previous research (e.g. Kruger et al., 2003; Morales,

2006). Morales (2006) found that WTP, likelihood of store visit and overall evaluation increased

when firms exerted extra effort. This holds true even if the additional effort does not directly

(38)

38

the effort information, customers do not reward this by an increase in WTP, store choice or

overall evaluation. This might have been the case in this research. The advertisement indicated a

development time of one and a half years. Participants that read this extra piece of information

might have thought that it is used to persuade them. Therefore, the WTP did not increase.

Another explanation for the rejection of the firm effort hypothesis may be linked to the research

by Kruger et al. (2003). The study found in an experiment that participants overall valuation of

the product and perceived quality only increased when the quality of the product was difficult to

determine. This may have played an influencing part in this research. Over the last two years

there was a multitude of health trackers released to the market and advertised on many

platforms, media and TV (Taylor, 2014). Therefore, people have been exposed to health trackers

recently and might therefore have an idea about the quality and thus are not inclined to place a

higher bid on the firm effort advertisement. When looking at the results from analyzing the

follow up question about knowing the development time, a different conclusion can be drawn.

Concerning the participants in the baseline treatment, there was a positive correlation between

bid amount and agreeing with the statement that knowing the development time is important.

That is, participants who find it important to know how long it took to research and develop the

product, placed on average a higher bid amount. This does not directly support hypothesis 4

because they were not presented with any cues about firm effort in the advertisement.

Nevertheless, this shows that consumers might still be interested in knowing the R&D time. One

explanation for the difference between the outcome of the firm effort treatment and the firm

effort question might be due to the presentation of the ad. It was indicated in the ad that it took

1,5 years to research and develop the product but participants could have missed this

information or interpreted it differently. Furthermore, participants only have 6 minutes to place a

(39)

39

pressure and users might therefore spend more time on reading and answering the questions

properly.

The second significant effect (p < 0,05) was found between the baseline treatment and

treatment 4 (innovativeness). The difference between the mean bid amount was 126,462 € cent. This finding also contradicts with the stated hypothesis that communicating innovativeness

should increase the WTP compared to the baseline treatment. Both, Szymanski, Kroff & Troy

(2007) and by Schreier, Fuchs & Dahl (2012) find evidence that innovativeness may increase

the overall valuation of a product and the WTP. However, a study by Calantone, Chan & Cui

(2006) questions these findings and suggest that the view on innovativeness is too holistic. They

provide a framework that proposes a distinction between product advantage, product

innovativeness and customer familiarity. They claim that researchers too often take product

advantage misleadingly as an innovative factor. Furthermore, according to them, it is often not

distinguished between how the customer and how the firm see and define innovativeness. Firms

generally refer to innovativeness as a change in technology and compare it to competitive

offerings whereas customers judge innovativeness based on how it can alter their mental state

and behavioral habits. If this is not given consumers do not value innovativeness higher

(Danneels and Kleinschmidt, 2001). That might explain the lower WTP in this research.

Participants that saw the advertisement with the H2O-Pal could have inferred that they cannot

derive any value or change in their mental state and behavioral habits from the product. Another

explanation for the contradicting result concerns the uncertainty about new products because it

inhibits consumers from adopting a new innovation (Castano et al., 2008). If consumers were

ambiguous and not certain about the benefits of the H2O-Pal they might have reacted with a

(40)

40

There was a marginally significant effect (p < 0,1) found between the firm effort

treatment and the personalization + innovativeness treatment. However, this was not part of the

research. The only hypothesis that was supported stated that communicating personalization and

innovativeness will have a higher average WTP than only communicating innovativeness. This

result is interesting in light of the observed average bid amount of the other treatments. As

shown above both personalization and innovativeness have a lower mean bid amount than the

baseline treatment. However, communicating personalization and innovativeness together

increases the WTP to an amount that is higher than the baseline treatment. These findings

confirm the results from Franke & Piller (2004). They investigated the likelihood of purchase

and the WTP of innovative user toolkits. These kinds of toolkits are products that are developed

and personalized by consumers. They found a positive effect on the WTP and purchase

intention. This indicates that neither personalization nor innovativeness itself seems valuable for

the customer but combining both increases the WTP.

With respect to the hypotheses 1,3a and 3b there has been no significant effect found and

they are therefore rejected. The reason why there has been no significant effect between the

baseline treatment and treatment 3 (personalization) might be explained by the fact participants

did not perceive the feature of personalization as product enhancing and therefore did not want

to pay more than participants from the baseline treatment. Furthermore, the study by Ball et al.

(2006) found evidence that personalizing a service may increase loyalty and in turn the WTP.

However, this research investigated on PSSs. Therefore, participants might have inferred that

the feature of personalization might more of a product feature and therefore no effect was found.

(41)

41

that has already been discussed in the literature review. A PSS is a fairly new concept and type

of offering. Thus, the consumer might not understand the benefits of a PSS yet.

The last paragraph of the general discussion concerns the Chi-squared test to find out

whether participants were influenced by the different cues in the advertisements. The results

showed that users who answered with ‘strongly agree’ or ‘agree’ were equally distributed over the three treatments. This may be an indication that they did not get influenced by the cues.

6.2 Implications

The current research supports and contradicts with previous research in the fields of PSSs,

personalization, firm effort and innovativeness in different ways. The three hypotheses that

tested each concept individually did not receive any support. This implies that the theory about

those concepts might be questionable. The theory about all three concepts with respect to PSSs

is very limited. Therefore, this research implies that even though previous research found

evidence that all three concepts increase the WTP, this might not hold for PSS. However, this

study also supports the results of previous papers (e.g. Schreier, Fuchs & Dahl, 2012; Franke &

Piller, 2004) in that personalization and innovativeness increase the WTP and overall customer

purchase intentions.

The nature of the auction experiment may give very close to reality outcomes and

therefore interesting implications for businesses. The research did not find evidence that

personalization, firm effort and innovativeness itself increase the WTP. This means that

marketing managers should be careful with signaling those kinds of features or product and

service specifications. With respect to innovativeness, as mentioned, it is important to reduce the

(42)

42

think and feel ambiguous about it (Castano et al., 2008). When the product or service

communicates that both personalization and innovativeness are part of it the WTP increases.

This implies that producers of PSSs as well as marketers should put focus on that if it is relevant

for the PSSs in questions. That is, producers of PSSs could add an additional function to make

the PSS personalizable and thereby increase the value for consumers. Marketers on the other

hand may focus on both signaling that the PSSs is highly innovative as well as that it has the

ability to be personalized.

6.3 Limitations and Suggestions for Future Research

The results of the research should be interpreted with care. First of all, even though the data

collection method of the Veylinx auction has proved to be good for previous research there may

be still limitations. The platform is based a second-price sealed-bid or Vickrey auction (Vickrey,

1961). This reveals very close estimates to the actual WTP of a person. However, since nobody

is forced to really bid the amount she is willing to pay, there might still be biased bid amounts.

This could explain the high amount of low bids. This is connected to the second limitation.

There was a high number of zero or low bids that caused a highly positively skewed distribution.

Furthermore, the distribution was also far from being normal. After taking out the outliers and

trimming the mean down by only using data from the top 50 percentile, the data was still

slightly positively skewed and not normal. Even though the ANOVA is said to be a robust test

the results have to be looked at carefully. The third limitation concerns the sample size. In the

beginning the sample size was adequate but due to the high amount of zero and low bids and the

downsizing to receive a closer to normal distribution the sample decreased to N = 306.

However, on the other hand decreasing the sample size is usually not beneficial to reach a

(43)

43

reason for the high amount of zero and low bids could also be due to the fact that participants

were not attracted by the advertisements or the product. Additionally, this research only tested

one product on the experiment due to time and capacity constraints. Therefore, the results are

only based on one PSS which might be a problem for generalizability.

As discussed in the literature (e.g. Mont, 2004; Tukker, 2006), PSSs are currently

research and debated about in many journals. Due to the nature of their theoretical but also

highly practical contributions they can deliver to research they are an interesting field to

investigate. Especially when looking at different features and how consumers value and perceive

them, is interesting to look at in future research. Furthermore, innovation and innovativeness are

highly discussed topics and especially interesting when researching PSSs because of their

importance in the future for producers and consumers alike. With respect to the auctioning of

products and services, future research could auction at least two products or services from

The willingness to pay for innovative product-service systems : testing personalization, innovativeness and firm effort