The New Service Development Process
The influence of Stage-Gate Activities and Process Formalisation on the Firms’ Innovative Performance
By
Remco van der Zee S2025051 Heresingel 20a 9711 ET Groningen remco.vdzee@gmail.com
MSc. BA: Strategic & Innovation Management University of Groningen
Faculty of Economics and Business
June 2016
Supervisors:
Dr. Wim Biemans
&
Dr. Hans van der Bij
Abstract
The Dutch specialised healthcare organisations are recognised worldwide for providing top quality care. Nevertheless, these organisations operate in a complex environment and face critical challenges such as rising costs, ageing populations and retiring workface. Hence, the necessity of finding innovative solutions is evident. This research focuses on the New Service Development (NSD) process by examining how the stages and activities influence new service innovation performance. Moreover, the direct influence and the moderating effect on the NSD stages of formalisation are accounted for.
Finally, an inter-industry comparison is made with the development process of financial service institutions. The results indicate that the majority of stages do not have a direct, positive influence on new service innovation performance. Additionally, the stage business analysis and marketing strategy is beneficial for both explorative and exploitative innovations. Next, the results suggest that managers should be careful in exerting control through formalisation, as it may do more harm than good.
Nevertheless, delicate implementation of such techniques depending per stage may improve service innovation performance. Lastly, a remarkable result is that no significant differences were found for the development of new services between financial institutions and specialised healthcare organisations.
Word count: 11.322
TABLE OF CONTENTS
List of figures ………...………3
List of tables ……….3
Introduction ……….4
1.1 Research questions ……….6
Theoretical Background………..7
2.1 Service innovation ………..7
2.2 Healthcare service innovation ……….7
2.3 The new service development process ……..………..8
2.4 New service innovation performance………10
2.5 Hypotheses and conceptual model ………10
Methodology ………..16
3.1 Sample ………...………16
3.2 Data collection ……….. 17
3.3 Method of analysis ………19
3.4 Measures ………19
3.5 Controllability, validity and reliability ……….22
Results ………22
4.1 Data reduction results ………22
4.2 Descriptives and correlations ……….……….. 23
4.3 Hypothesis testing ………. 25
Discussion ………..……… 31
Conclusion ……….34
6.1 Implications for management ………34
6.2 Limitations ………35
6.3 Future research ………36
Bibliography ………..………37
Appendices ……….... 44
List of figures
Figure 1: Stage gate model ………9
Figure 2: Conceptual model ………. 16
List of tables Table 1: Sample statistics healthcare ……….18
Table 2: Sample statistics healthcare and financial industry ……….. 19
Table 3: Overview of variables ……….. 20
Table 4: Cronbach alpha scores ……….……….23
Table 5: Descriptive statistics ………..……….. 24
Table 6: Correlations ………..24
Table 7: Regression analysis (DV = explorative innovation) ………. 26
Table 8: Regression analysis (DV = exploitative innovation) ………27
Table 9: Rotated component factor analysis DV ………48
Table 10: Rotated component factor analysis IVs ………..………48
Table 11: Rotated component factor analysis formalisation ………49
Table 12: Product newness ………49
INTRODUCTION
The importance of services in developed western economies receives more and more attention from scientific scholars. This is rightly so, because in 2014 74% of the economic added value in the Euro Area was generated by services, and even over 80% of the added value in the US economy was produced in service related industries (European Central Bank, 2015). Nevertheless, scholars are inconsistent in assessing the current, ambiguous state of literature on services, which is considered both scarce and dispersed (De Jong & Vermeulen, 2003; Story & Hull, 2010) as mature (Bryson
& Monnoyer, 2004; Cainelli, Evangalista & Savona, 2004). As a result, managers planning to improve their services will only find limited support from the existing literature on services and specifically regarding the New Service Development (NSD) process (Biemans, Griffin & Moenaert, 2016). Consequently, scholars call for further development of service the literature by gaining a deeper understanding of how the service context impacts the development of new services.
To move the field of NSD forward and to account for the strong variation in the services landscape, scholars emphasize a need for basic research (Droege et al., 2010; Griffin et al., 2013). Due to the large variation in the services landscape, studying different service contexts will deepen the understanding of NSD theory (Biemans et al., 2016). Despite the differences between contexts, the majority of studies have examined NSD in financial services in single industry research and as part of multiple industry research, although more recent research expanded to telecom/ICT and hospitality services (Biemans et al., 2016)
Most studies regarding NSD have concentrated on studying critical success factors and as such scholars seem to have neglected to study the details of NSD stages and how these stages are managed (Alam & Perry, 2002). In addition, most studies on NSD do not distinguish between radical and incremental innovation. Nevertheless, the importance of following structured NSD stages in order to be successful has been emphasized (De Brentani, 1998; De Brentani & Ragot, 1996; Froehle, Roth, Case &
Voss, 2000). Thus, a detailed look into the activities and stages of the NSD process as
well as examining how formalisation affects innovative performance for previously
under-investigated industries contributes to the literature on service innovation and
will help practitioners to improve their development process.
This paper studies NSD in the Dutch specialised healthcare industry. The Dutch healthcare system is ranked as one of the best in the world, and is even considered the best in the European Union (Health Consumer Powerhouse, 2014).
Despite the well-perceived quality, the industry is in dire straits as costs are increasing at an unsustainable rate. According to a World Economic Forum report from 2013, the majority of developed countries experience faster growing healthcare costs than annual GDP growth. The report argues that when this trend continue, regardless of any form of economic crisis, unsustainable levels will be reached in the second half of this century (World Economic Forum, 2013). Next to the rising costs, the industry also faces challenges such as ageing population, a retiring workforce, and patients demanding the highest quality of care that exploits all of the latest technology developments and the adoption of related external knowledge (Länsisalmi et al., 2016). These particular challenges demand innovative solutions, involving all aspects of healthcare in terms of delivery to customers, technologies, and business models (Herzlinger, 2006).
Especially in the healthcare industry, where organisations are expected to be highly reliable and customers are emotionally sensitive, providing new and innovative services matters a great deal in shaping the reputation of the organisation and in distinguishing from competitors (Berry, Davis & Wilmet, 2015). However, changes in healthcare are often politically and socially sensitive, and very complex in terms of policy-making (World Economic Forum, 2013). This sensitivity and complexity raises the influence of stakeholders such as the government and regulating agencies, which increases the degree of difficulty for healthcare innovation practitioners to manage the development of service innovations.
The complexity of the healthcare sector makes the development of new
services through NSD processes ‘chaotic’ (Sethi & Iqbal, 2008). A chaotic NSD
process increases the importance for managers to exert control over the process and
bring discipline to the NSD process (Sethi & Iqbal, 2008). A well-known and frequent
research mechanism that managers may apply to deal with ‘chaotic’ process of
developing new services is the stage gate model. This model enables managers to
systematically guide the innovation process in order to increase efficiency, improve
performance and reduce the cycle time of the development process (Sethi & Iqbal,
2008). This research applies the stage gate model in order to gain understanding of the
process and the activities per stage, and examines its impact on the development of explorative and exploitative innovations.
The purpose of this research is threefold: first, the research fills a gap by taking a detailed look into the process and stages of the development of new services as according to Papastathopoulou and Hultink (2012), this research topic in the field of NSD is underdeveloped. Second, this study compares the findings from the investigated healthcare industry to the more frequently studied financial industry.
Therefore, this research contributes to NSD literature by doing basic research into how the service context impacts the development of new services.
Research Questions
Based on the introduction of the topic and the goals of the research, the following research question is developed to demonstrate the relationship between the activities and stages of the NSD process and the impact on service innovation performance:
RQ
1: ‘‘What is the influence of the NSD process on service innovation performance?’’
In order to answer the research question, the following sub questions are formulated:
- What is the influence of formalisation on the NSD process and on service innovation performance?
- Which stages (activities) are important in developing exploitative innovations, and which stages (activities) are important in developing explorative innovations?
RQ
2: ‘‘Are the insights of the NSD process, which are usually generated from the financial industry, similar in the healthcare industry?’’
The remainder of this paper is structured in the following manner: first, a literature
review is conducted to highlight the relevant theoretical concepts. Next, the
hypotheses are derived, building upon expectancy theory, bounded rationality, and
rational decision making and garbage can theory. Following, the methodology is
described extensively after which the results are presented. Finally, the results are
discussed, theoretical conclusions are drawn, practical implications are suggested and
the limitations of this research are stated.
THEORETICAL BACKGROUND
2.1 Service innovation
Some authors distinguish service innovation from service development (Bettencourt, 2010), while others treat them as synonyms (Biemans et al, 2016) A broad definition is provided by Toivonen and Tuominen (2009), who define service innovation as ‘‘a new service or a significant renewal of an existing service which is put into practice and which provides benefit to the organisation that has developed it.
Additionally, the renewal must not only be new to its developer, but in a broader context’’. In a similar but more concise vein, the definition from Biemans et al., (2016) is: ‘the process of devising a new of improved service, from idea generation to launch’’. Some interesting aspects of innovation can be derived from these definitions: first, there is a clear separation between innovative outcome and innovative process. Second, the invention only becomes an innovation when it is used and put into practice. Third, it is an explicit, step-wise process and finally, the invention must be new to the relevant actors and create some sort of added value (Witell et al., 2015). Scholars have identified not only financial benefits as a measure of innovative performance (e.g. De Brentani, 1989), but also the creation of customer value (Narver & Slater, 1990) and increasing strategic success (Kay, 1993).
2.2 Healthcare Service Innovation
In 2006, the Dutch healthcare system, as several other western countries, has reformed in order to encourage ‘managed competition’ between healthcare providers and health insurers to increase quality and decrease costs (Ikkersheim & Koolman, 2012). A critical change for healthcare providers was the increasing freedom for patients to select the provider of their choice. Providers were forced to publicly publish the quality of the care they offered annually (Ikkersheim & Koolman, 2012) and as a result, the competitiveness between providers increased. Consequently, the necessity for healthcare providers to distinguish themselves by offering better services or through greater service offerings receives more importance.
Innovation has become a crucial capability for all healthcare organisations
(Länsisalmi et al., 2006), to balance cost containment and the quality of healthcare
(Omachonu & Einspruch, 2010). Digital information, nanotechnology,
semiconductors and genetic engineering are revolutionizing healthcare and creating
huge opportunities for innovation (Govindarajan, 2007; Omachonu & Einspruch, 2010). Hence, innovation in healthcare is a topic of great interest yet research on healthcare innovation, especially from a business perspective, is limited (Omachnu &
Einspruch, 2010).
Innovations in healthcare organisations typically involve newly introduced services, new ways of working to improve efficiency and/or new technologies (Länsisalmi et al., 2006) Thus, innovations are related to product innovation (i.e. what the patient pays for), process innovation (i.e. the production or delivery methods) or structure innovation (i.e. internal/external infrastructure and business models) (Varkey et al., 2008). However, most healthcare organisations do not have the luxury of an R&D department and depend on the creativity of practitioners and feedback from patients and practitioners (Omachnu & Einspruch, 2010)
To understand innovation in healthcare, an in-depth analysis of the related challenges and the innovation process is needed (Omachnu & Einspruch, 2010).
Research has shown that it is difficult to change the behavior of practitioners and healthcare organisations (Greco & Eisenberg, 1993; Shortell, Bennett & Byck, 1998).
Moreover, attempts to innovate are strongly regulated by laws and regulations (Faulkner & Kent, 2001). As such, ‘really new practices’ are thoroughly scrutinized in the early stages of development, to prevent the development of potentially harmful innovations (Faulkner & Kent, 2001).
2.3 The New Service Development Process
The literature on NSD stems from research on New Product Development (NPD) and consequently the innovation process for both products and services was assumed to be similar. Nevertheless, services have unique characteristics – intangibility, heterogeneity, inseparability and perishability, to distinguish goods from services (Lovelock, 1983; Zeithaml & Bitner, 2000). Despite these unique characteristics, many scholars have used concepts, frameworks and methods of NPD to explain the process of NSD, resulting in dispersed findings and a lack of consensus (Biemans et al., 2016).
The NSD process can be defined as ‘‘activities executed and decisions made
to generate ideas, develop the concept, analyse the opportunity and test and launch the
service’’ (Cooper et al., 1994). Although innovation usually does not follow a linear
process and is more ad-hoc, research indicates the importance of following the NSD
process structurally through the stages in order to be successful (De Brentani, 1989, Froehle et al., 2000, Alam, 2006). One of the more common models that is frequently applied is the Stage-Gate model (Cooper, 1990). However, the absolute majority of research on Stage-Gate models focuses on product innovation instead of service innovation. The stage gate model is visualized in figure 1.
Figure 1: Stage Gate Model
Stage gate controls explicitly breaks up the innovation process into a set of discrete and identifiable stages (i.e. idea generation, business analysis and marketing strategy, technical development, testing and launch), whereby each stage is executed through a set of activities. Before the development can move to the next stage, the gates control whether or not the development may proceed (Cooper, 2001). Thus, managers have the possibility to control the development of new services through two means: first, through influencing the gates by making them more rigid or loose, and second, by formalizing the activities in each respective stage, exerting control over the behavior of the employees by specifying methods and procedures to be adopted for performing tasks (Sethi & Iqbal, 2008).
The idea generation stage (or, discovery) involves activities that are designed
to discover opportunities and activities to generate new ideas for innovation. This
stage is characterised by low levels of formalisation and unstructured processes
(Khurana & Rosenthal, 1997). This stage is seen as very crucial as it determines the
success of further stages (Alam, 2006). The business analysis and marketing strategy stage (or, scoping and build business case) examines the attractiveness of the ideas in terms of technical, marketing and business feasibility. Market conditions are analysed, as a basis to develop the new service upon and it is the foundation for successful innovation efforts (Sundbo, 1997). The third stage, (technical) development, translates plans into actual deliverables. In this stage, the development and design of the new service occurs, the operation plan is determined and the testing plans are developed.
These first three stages are also referred to as the ‘fuzzy front end’ (Alam, 2006). The last two stages, testing and launch are designed to provide validation of the new service, train personnel, identify the acceptance of the customers, and eventually the full commercialization of the new service.
2.4 New service innovation performance
Exploitative innovations are incremental and are designed to meet the needs of existing customers and/or markets (Benner and Tushman, 2003). These innovations enlarge existing knowledge and skills to improve existing designs, expand existing services and increase efficiencies (Abernathy & Clark, 1985). Conversely, explorative innovations are radical innovations and are designed to meet the needs of new, emerging customers and/or markets (Benner and Tushman, 2003). These innovations are building upon new knowledge or depart from existing knowledge (Levinthal &
March, 1993).
A crucial difference between exploitative innovation and explorative innovation is that there is more uncertainty involved in explorative innovation. This implies that because the uncertainty in incremental innovation processes is lower, the information about the whole development process is more complete, which increases the likelihood that innovators make rational decisions. Because uncertainty for radical innovation output is high(er), the whole process is less rational and innovators rely more on ‘following their guts’ (Saarinen & Rilla, 2009)
2.5 Hypotheses and conceptual model
The NSD literature evolved over the last 30 years (Biemans et al., 2016), from
the early studies (1980s) which mainly used qualitative methods to explore the nature
and stages of the NSD process, followed by later studies (1990s) that applied
quantitative methods where the emphasis of research involved the identification of
key NSD success factors. As a result, most research on NSD is empirically driven.
Yet, when applied, the theoretical perspectives are not consistently used to explain how and why organisations follow certain strategies to develop their services. In this study, hypotheses are formulated based on the expectancy theory (Vroom, 1964), garbage can theory (Cohen, March and Olsen, 1972) and the rational system perspective, and bounded rationality (Simon, 1978).
Expectancy Theory (Vroom, 1964) is one of the major theories in the study of work motivation. Vroom (1964) developed a model of Valence-Instrumentality- Expectancy (VIE) to study how individuals behave or act, driven by the individuals’
expectation that the selected behavior will result in performance and outcome. The theory has been applied in fields such as organisational behavior, leadership and compensation (Van Eerde & Thierry, 1996). To develop the first hypothesis, the line of thought from the expectancy theory is shifted from the individual level to the organisational level.
Expectancy theory describes why actors make certain decisions and asserts that the choice of behavior is a function of expectancy (Chen, Ellis & Suresh, 2016) Thus, the expectancy theory proposes that organisations act in a certain way because they are confident that these activities lead to expected performance (Expectancy).
More specifically, organisations will perform the selected activities when they expect that there is a positive relation between the efforts of the activities and performance.
This performance will ultimately lead to an outcome (Instrumentality). Valence refers to the value that the organisation attaches to the outcome (i.e. if the outcome is not desired, the activities will not be performed)
In this particular study, the efforts are the activities in each stage of the stage gate model and the performance relates to the service innovation performance (i.e.
exploitative or explorative innovations). Moreover, as an outcome of the service
innovation performance the organisation may be better able to distinguish itself in the
market, leading to some sort of (financial) reward such as a larger market share or
higher profit. Even though these outcomes may seem beneficial at all times, the
theory holds that not all outcomes are directly beneficial and thus desirable for the
organisation. For example, an increasing market share could make the organisation a
more tempting target for actual and potential competitors or the organisation may
attract attention from government agencies and consumer associations that monitor
the operations and management of the organisation (Bloom & Kotler, 1975)
Based on the logic of the expectancy theory, assuming the expectation that activities are positively related to performance, hypotheses 1
a-1
eare formulated:
H1
a: Performing the activities of the Idea Generation stage positively influences service innovation performance
H1
b: Performing the activities of the Business Analysis & Marketing Strategy stage positively influences service innovation performance
H1
c: Performing the activities of the Technical Development stage positively influences service innovation performance
H1
d: Performing the activities of the Testing stage positively influences service innovation performance
H1
e: Performing the activities of the Service Launch stage positively influences service innovation performance
A key element of the expectancy theory is the notion of behavioral choice, accounting for why certain activities are selected when other alternative courses of action are also available. Research found that not all organisations accurately follow the activities of the stages of the NSD process and as a result different performance outcomes are recognized (Stevens & Burley, 2003). Moreover, the failure rate of new services is high (Stevens & Burley, 2003). The notion that different performances are recognized implies that decision makers assess the value of activities in relation to the expected performance differently. The fact that managers differently assess the value of activities is because decision-makers are highly bounded by their subjectivity, individual preferences and their cognitive ability (Zafar, 2010). Simon (1978) stresses in his notorious theory of bounded rationality that the cognitive ability of humans is limited. The theory argues that there is a significant difference in the complexity of the real world and the ability of the human to understand and assess all of the complex information thoroughly. In order to deal with this phenomenon, managers may implement methods such as the stage gate model to manage this complexity and to control the developments in the organisation. The stage gate model is a linear process functioning as a step-by-step guideline in which each stage has explicit activities, and may serve as a tool for organisations to cope with the complexity.
Rational choice theory is a decision-making theory that is related to the
concept of bounded rationality and is frequently applied to assume the behavior of
individuals in microeconomics, but is also found in philosophy, political science and sociology research. This model argues that the behavior of actors is assumed to be rational when it is goal oriented, whereby actions and process are evaluated, and tasks are consistently executed (i.e. high levels of formalisation) instead of random and impulsive behavior. As a result of rational behavior, actions result in more predictable outcomes. In section 2.4 it was argued that exploitative innovation is more likely for low levels of uncertainty, and when uncertainty is low predictions are easier to make.
Therefore, by means of the rational choice theory, an organisation that controls behavior through formalising procedures, rules and structures is likely to develop exploitative innovations.
On the other hand, formalized procedures, structures and rules may make the organisation inflexible (Sethi & Iqbal, 2008). This is especially detrimental for the development of radical innovations as flexibility and fast development processes is required. The Garbage Can theory (Cohen, March & Olsen, 1972) is the opposite of rational decision-making theory in the sense that ‘‘specific decisions do not follow an orderly process from problem to solution, but instead are outcomes of several relatively independent streams of events within the organisation’’ (Daft, 2006). Thus, in line with this idea, the development of new services does not necessarily have to follow an orderly process. Instead, the activities of the stage gate model may be executed in a less structured fashion and do not necessarily follow sequential steps, but are executed based on the characteristics of the situation. This approach makes organisations much more flexible, increasing the probability of developing explorative innovations. Based on these arguments, hypothesis 2 is developed.
H2: Formalisation has a positive/negative influence on the development of exploitative/explorative innovations
Research suggests that, despite the importance of each stage in developing new service innovations, the development of either exploitative or explorative innovations may stress more importance on one stage over the other due to the nature of the innovation and the unique characteristic of the service (Avlonitis et al., 2001;
Dolfsma, 2004).
Exploitative innovations are relatively easy for competitors to copy as the
improvements in the existing service are minor (De Brentani, 1989). Therefore, it is
important to be superior in positioning the service as a means of distinguishing yourself to the relatively similar offering of the competitor. Avlonitis et al., (2001) found that the stages business analysis and marketing strategy, and service launch are particularly important stages in this respect. These stages focus on designing services that differentiate, at least in the perception of customers, and the building of a unique selling proposition (Avlonitis et al., 2001), as a means to overcome the threat of imitation by competitors. Hence, the stages that involve commercialization activities become more important as competitors may have substitute services
Explorative innovations are characterised by high uncertainty. As explorative innovations involve the emergence of new customers and new markets, the needs are not easily defined or even unknown (Avlonitis et al,. 2001; Danneels, 2002). Due to the higher risk that is associatied with developing explorative innovations, more financial and human resources are required (Danneels, 2002) In the research of Avlonitis et al., (2001) it is found that idea generation, business analysis and marketing strategy, and testing are particularly important stages to develop explorative new services. These stages involve important activities in exploring and defining the latent needs of new emerging customers or markets and translating them into pioneering services which offer unique and novel solutions (Dalgic, 2000).
Proficiency in the activities in these stages increases the confidence of the organisation that the innovation faces a certain degree of acceptance should it be introduced to the market (Dalgic, 2000). This is also in line with Faulkner & Kent (2001), who found that radical innovations (in healthcare) are scrutinized in the early stages of development to avoid developing harmful new services. In line with this argumentation, hypotheses 3
aand 3
bare developed.
H3
a: The stages Business Analysis & Marketing Strategy and Service Launch are particularly important stages for the development exploitative innovations H3
b: The stages Idea Generation, Business Analysis & Marketing Strategy and
Testing are particularly important stages for the development of explorative innovations
Besides the direct effect of formalisation on service innovative performance as
argued in hypothesis 2, it can be expected that formalisation also has a moderating
influence on the relationship between the individual stages of the NSD process and
service innovation performance. Research has shown that following the activities in each stage of the NSD process in a structural manner increases the effectiveness of the organisation and consequently reduces the number of NSD failures (De Brentani, 1999; De Brentani, 2001). Moreover, the organisation may become more cost- efficient (Droege et al, 2010). However, the influence of formalisation on the activities in each stage and the structural following and execution of this process makes the organisation inflexible, slow (Sethi & Iqbal, 2008) or even overloaded with information (Papastathopoulou et al., 2001). Based on the argumentation here and from the development of hypothesis 2, it can be expected that the influence of formalisation is only facilitating the development of exploitative innovations and has detrimental effects on the development of explorative innovations. Thus, hypothesis 4 argues:
H4: The moderating effect of formalisation on the relations between NSD stages and service innovation performance is (a) negative for developing explorative innovations and (b) positive for developing exploitative innovations
Research suggested that the development of new services is very dependent on the context of the industry (Biemans et al., 2016, Droege et al., 2010). This may be attributed to the characteristics of services. For example, services are very heterogeneous and are often developed to meet the individual needs of the customer.
Therefore, not only within service industries variation may be expected but especially between service industries the variation in development processes should be strong.
Therefore, the likelihood that the development of new services in healthcare organisations is different from the development of new services in the financial industry is assumable. To account for this variation, hypothesis 5 is developed.
H5: The development of new services in healthcare is different from the development of new services in the financial industry.
Based on the formulated hypothesis, a conceptual model is provided. This model can
be found in figure 2.
Figure 2: Conceptual Model
METHODOLOGY
3.1 Sample
In order to test the hypotheses, data was necessary which could not be obtained from public databases. Therefore, data was collected by means of a survey.
Through online websites (Zorgatlas, 2014; ZKN, 2016) all relevant specialised
healthcare organisations were identified and categorized per sector: academic
hospitals, general hospitals and specialised hospitals were identified as public
institutions and for private institutions I distinguished between institutions offering
insured and uninsured care. As a result of this search, 8 academic hospitals, 103
general hospitals and 29 specialised hospitals were identified. Further research
learned that as a result of fusions, hospitals being subsidiaries of academic hospitals
and care groups, and bankruptcy, the category general hospitals was reduced to 68
and specialised hospitals to 20. The list of private institutions (i.e. independent
treatment centres) consists of 143 organisations. Similarly, further research reduced
the sample to 119 private healthcare providers as a result of cooperation with academic hospitals, fusions, and bankruptcy. This resulted in a total population of 215 specialised healthcare providers in the Netherlands.
3.2 Data collection
To collect the data, I used a survey that was used for an NSD study in the financial service industry (Rademaker, 2014), which was adapted from the study of Avlonitis et al., (2001). This facilitates the comparison of the present study to the findings of the financial industry. To adapt the questions of the survey to the context of healthcare, I consulted with two professionals in the field. As a result, minor adaptations were necessary that were only implemented if the content of the questions would not be altered so that a valid comparison with the financial industry is still enabled. For example, firm translated to organisation and customer was changed into patient. Moreover, 10 questions involving health care insurers and health care professionals were included to get a more complete understanding of the process. For example, the existing question: we test the new service with potential patients was expanded with two questions to indicate whether or not the service was tested with healthcare insurers and health care professionals. In addition, two questions were included trying to get feedback regarding financial indicators. More specifically, the respondents were asked about the % of new services that were insured by insurance companies and the % of new patients after the introduction of a new service. The questionnaire can be found in Appendix A, in which the added questions are in italic.
After the survey was prepared for distribution, I reached out to all 215
organisations by mail, explaining the purpose of this research and asking for contact
information of the employee responsible for innovation. This approached prevented
that directly mailing the survey marked the mail as spam. In total, I received 78
positive responses of people who were willing to participate, and to whom the survey
was mailed via the Qualtrics mailing option. After multiple reminders 38 respondents
completed the survey. 59 people started the survey, but did not complete for several
reasons: some found the approach of the survey too different from their particular
innovation process in practice and decided not to complete the survey. Others were
planning to participate but lacked the time, and some did not respond after several
reminders. At the end of the survey, 34 (89,5%) respondents indicated that they
wanted to be informed about the results, suggesting that the respondents had a positive attitude towards the study (Hultink & Robben, 1995).
Moreover, 35 organisations indicated that they were not willing to participate, for reasons of lacking time, receiving too many similar requests and not participating in any of them, simply not being interested or having the explicit policy of devoting all attention to the patients.
The 38 completed surveys (n=215) resulted in a response rate of 17,7%. The distribution of the participants can be found in table 1. The high standard deviations indicate that there is a relatively high variation in the sample statistics. From the 38 responses 23 are men and 15 are women; and the respondents held functions such as service/product/marketing/general manager (31,6%), NSD/innovation manager (23,7%), and director/owner/board member (44,7%)
The survey questions were asked in random order. The respondents were not explicitly made aware of the stages belonging to the activities. As the original survey is in English the survey was translated to Dutch (Rademaker, 2014) using the parallel translation/double translation method (Adler, 1983; Sekeran, 1983). As the sample only includes Dutch specialised healthcare organisations, translating the questions facilitated the ease of understanding for respondents. Also, the survey in Dutch can be found in the Appendix (B).
Type
Amount of
Respondents Avg. Age Avg. Size
Avg. Innovation Experience
Academic hospitals
2 76,00 (σ 33,94) 7000 (σ 0,00) 29,00 (σ 32,53) General
hospitals
13 53,54 (σ 49,51) 3630,77
(σ 2196,35)
18,33 (σ 31,43) Specialised
hospitals
6 120,67 (σ 188,41) 785,83
(σ 1006,66)
18,20 (σ 16,83) Private clinic
insured
16 11,44 (σ 8,65) 72,19 (σ 126,04) 8,60 (σ 5,64) Private clinic
uninsured
0 - - -
Other 1 116 9000 -
Total 38 49,24 (σ 85,23) 2001,84
(σ 2676,59)
14,37 (σ 20,85)
Table 1: sample statistics healthcare
The collected data from the healthcare industry is added to the existing dataset of
Rademaker (2014). This results in a dataset of 152 respondents (healthcare N=38;
financial industry, N=114). This combined dataset facilitates the testing of hypothesis 5. Moreover, because of the relatively low response from the healthcare industry, more meaningful results regarding hypothesis 1 until 4 can be drawn from the combined dataset. The distribution of the total dataset can be found in table 2. For specific information regarding the NSD study in financial services the author refers readers to Rademaker (2014)
Type Amount of
Respondents
Avg. Age Avg. size Avg. Innovation Experience
Finance 49 39.49 (σ 46,616) 2687,06
(σ 9238,147)
31,93 (σ 42,592)
Fintech 35 3,34 (σ 2,903) 20,20
(σ 43,571)
3,34 (σ 2,903)
Insurance 30 88,50 (σ 66,511 1423,40
(σ 3087,70)
57,77 (σ 61,66)
Healthcare 38 49,24 (σ 85,23) 2001,842
(σ 2676,593)
14,37 (σ 20,85)
Table 2: sample statistics healthcare & financial industry
3.3 Methods of Analysis
The program SPSS was used to execute the analyses that are described below.
Before the hypotheses could be tested, a variety of tests were performed to assess the validity of the results. First, a factor analysis (the Principal Component Analysis) was used to determine the constructs that explain most of the variance in a much larger set of variables. I suppressed for small numbers (<0,4), since Stevens (1992) recommends interpreting only factor loadings with a higher score than 0,4 (explains around 16% of the variance). After identifying the constructs and determining the reliability through Cronbach alpha scores, a multiple regression was executed to determine the overall fit of the model and the relative contribution of all independent variables, as well as the moderating influence of formalisation and the control variables.
3.4 Measures
All the measures and questions are directly adapted from the survey conducted
by Rademaker (2014). This survey questions are based upon previous studies by
Avlonitis et al., (2001), Janssen et al., 2006 and Despandé & Zaltman (1982). An overview of the variables can be found in table 3.
Table 3: overview of variables (Adapted from Rademaker, 2014
3.4.1 Dependent variable
The dependent variable new service innovation performance is measured through two types of indicators, financial and non-financial measures. To measure the non-financial indicators explorative and exploitative new service developments, Rademaker (2014) consulted the scales about exploratory and exploitative innovation from Jansen, Van den Bosch and Volberda (2006). An example question of
Variable Measure Operationalization Scale
Dependent Service innovation
performance
An explorative and exploitative 1-7 Likert Scale (1 = never; 7=always) with both 9 questions, and financial indicators (% new sales; % new profit,
% new patients, % new services insured
Ordinal
Independent Idea Generation (IG) 1-7 Likert scale, (1 = never; 7=always) with 6
questions
Ordinal
Business Analysis &
Marketing strategy (BA)
1-7 Likert scale, (1 = never; 7=always) with 10 questions
Ordinal
Technical Development (TD)
1-7 Likert scale, (1 = never; 7=always) with 5 questions
Ordinal
Testing (TE) 1-7 Likert scale, (1 = never; 7=always) with 5 questions
Ordinal
Service Launch (LA)
1-7 Likert scale, (1 = never; 7=always) with 9 questions
Ordinal
Control Size (S) Open ended question concerning the number of
employees
Ratio
Age (A) Open ended question concerning the period from founding date in number of years
Ratio
Innovation Experience (IE)
Open ended question about the innovation experience of the firm in number of years
Ratio
Formalisation (FO) 1-7 Likert scale, (1 = never; 7=always) with 5 questions
Ordinal
Product Newness (PN)
Three questions concerning the type of service that is launched (explorative or exploitative)
Binary
Industry Public (academic-, general-, specialised hospitals) and Private (insured, uninsured clinics)
Nominal
explorative innovation development is ‘we introduce services to the market that are completely new to the organisation’ and an example question of exploitative innovation development is ‘we improve the efficiency in which we deliver existing services’. For both exploitative and explorative innovation development, nine questions were developed to answer with a 1-7 Likert scale (where 1= never and 7 = always).
Moreover, the financial indicators were measured with four questions and concerned the percentage of new sales, percentage of new profit, percentage of new patients and percentage of new services insures, whereby the latter two were added to better take into account the particular characteristics of the healthcare industry.
3.4.2 Independent variables
The independent variables are the stages from the NSD process, yet also the direct influence of formalisation on innovative performance is considered. The measurement of the NSD stages was based on the 29 NSD activities of Avlonitis et al., (2001). Again, all questions were measured with a 7-points Likert scale.
An example question of the first stage, idea generation is ‘we filter the new service ideas and make an initial evaluation’, an example question of the stage business analysis and marketing strategy is ‘we identify unique characteristics that distinguish the new service from competitors’, for the third stage technical development an example question is ‘we make a prototype of the service’, an example of the fourth stage testing is ‘we test the new service with the employees of our organisation’ and for the final stage service launch an example question is ‘ We launch the service to the market (including promotion, distribution, etc.)’.
3.4.3 Control variables
In order to clarify and assess the relationship between the independent
variables and the dependent variable, it is important to control for the effects caused
by the sample statistics. In line with Rademaker (2014), first the size of the
organisation is assessed by means of an open-ended question about the number of
employees. Second, the years of existence in the industry, which relates to the
experience a firm has within the industry, is measured by an open-ended question
about age (in years). Related to this is the question concerning innovation experience,
which assesses the number of years the organisation is actively innovating. Finally, to
measure the last hypothesis a binary variable was created to distinguish between
financial services and healthcare services (whereby financial = 0 and healthcare is 1).
3.5. Controllability, Validity and Reliability
In order to reach inter-subjective agreement on the research results, controllability is secured by means of a detailed description of the study (van Aken, Berends & van der Bij, 2012). The validity was obtained by using an existing survey.
Rademaker (2014) has developed the original survey based on existing and statistically proven dependent variable measurements and control variables. This is in line with Dunn, Seaker and Waller (1994), who argue that validity is assured by deriving the items from existing literature. Furthermore, the validity of the independent variables was secured by means of a pre-test and checking if the responses were socially desirable (Rademaker, 2014). Finally, as the results should be independent from the used survey, reliability was provided through standardisation of the data collection (Rademaker, 2014).
RESULTS
4.1 Data reduction results
The first step of analysis is conducting a factor analysis. In Appendix C, the outcomes of the factor analysis can be found for the independent variables of the NSD process and formalisation, and for the dependent variable new service innovation performance.
In this research, principal component analysis (PCA) is used as extraction method, with a varimax rotation as selected rotation method. According to Hair et al., (2006) rotation is the most important tool to interpret the factors. The orthogonal
‘varimax’ rotation was selected as it more clearly separates the factors. In the case of the dependent variable, PCA was selected with a fixed number (2) of factors to extract. Only one code was distracted (RAD3) due to the 0,4 suppression limit.
In the case of the independent variables of the NSD stages, PCA was selected with a fixed number (5) of factors to extract. The 5 constructs correspond to the 5 stages of the NSD process. In order to facilitate a correct comparison between the financial industry and healthcare, the added questions as established in section 3.2 are not included for this analysis. From the 29 activities of all stages, 19 activities were extracted and 10 activities remained.
In order to conduct factor analysis on Formalisation, first FO
3had to be
reversed because the question was stated in a negative fashion. When the data is
reversed (i.e. 7=1, 6=2), the scores of FO
3mean the same as other FO data. Again, PCA was executed to create the formalisation construct, but no items had to be deleted.
The new variables were constructed from the remaining components. The Cronbach Alpha was calculated for all components to test the internal consistency.
Internal consistency measures if all the variables in the construct measure the same construct (Tavakol & Denninck, 2011). As a general rule, a Cronbach Alpha score higher than 0,7 is considered acceptable, which means that 70% of the variance in the score is reliable and 30% of the score is error variance (Tovakol & Denninck, 2011).
The scores of Cronbach Alpha for each construct can be found in table 4. Idea Generation and Testing are excluded as these are based on a single item. As visualized in table 4, only Exploitation and LA score under the acceptable standard of 0,7. However, these scores are fairly close to the acceptable standard of 0,7 and Nunnaly (1978) argues that scores above 0,6 may still be reasonable demonstrations of internal consistency. Thus, recognising that the internal consistency of these constructs may be questionable, the data is considered useful and further analysis was conducted without the validity of results being seriously harmed.
Variable Exploration Exploitation BA TD LA Formalization Cronbach
alpha
0,758 0,696 0,705 0,807 0,686 0,750
Table 4: Cronbach Alpha scores per construct
4.2. Descriptives and Correlations
Table 5 provides an overview of the descriptive statistics of the combined dataset, after the reductions from the principal component analyses. Some of the descriptives provide an interesting insight. First, the respondents gave higher scores to exploitative compared to explorative innovations. These scores are confirmed by the responses on the three general questions regarding product newness (Appendix D).
This is also consistent with research proposing that most organisations aim for
exploitative innovation, as most successful innovations are based on incremental
changes (Berry et al., 2006; Tushman & Nadine, 1986). Second, the standard
deviation of the control variables is high, demonstrating the strong variation in the
characteristics of the organisations.
Variable Mean Std. Deviation (σ) Minimum Maximum
1. Exploration
4,5305 1,1449 1,83 6,67
2. Exploitation
5,5150 0,7813 3,00 6,71
3. Idea Generation
4,6513 1,6368 1,00 7,00
4. Business Analysis
& Marketing Strategy
5,2248 1,1340 2,33 7,00
5.Technical Development
4,9693 1,4310 1,00 7,00
6. Testing
4,6316 1,8546 1,00 7,00
7. Service Launch
5,0163 1,4448 1,00 7,00
8. Formality
4,6980 1,3031 1,00 7,00
9. Age
43,46 64,291 0,00 500,00
10. Innovation Experience
25,89 42,3251 0,00 204,00
11. Size
1638,51 5581,837 2 45000,00
Table 5: descriptive statistics
Table 6: correlation Statistics
As can be seen from the correlation matrix that is provided in table 6, there are six negative Pearson correlations with exploration (of which two are significant) and only one significant Pearson correlation with exploitation. Thus, should exploitation or exploration increase, these variables decrease. Also, the independent variables from the NSD stages are all, besides service launch for exploration, significantly correlated.
Formalisation is only significant for the dependent variable exploration.
4.3 Hypothesis Testing
In order to test the hypotheses, multiple analyses were conducted with the constructed variables after the factor analysis. In order to improve the outcome, only inter-industry differences are accounted for. The control variable innovation experience is excluded from the analysis, as the data from healthcare organisations did not properly display the actual experience of organisations. Next to missing values, the question was not answered in a consistent manner as respondents differently interpreted the question (individual experience vs. experience of the organisation). Additionally, due to many fusions in this industry the respondents indicated the experience of the newly established firm, yet the experience with innovation of the former organisations was not taken into account. Furthermore, the control variable product newness was excluded as Rademaker (2014) suggested that this scale was too subjective. Besides, the product newness is also captured in the dependent variables. In order to test the moderating effects the independent variables were mean centered. This is necessary to be able to interpret the results of the regression analysis.
The results of the multiple regression analyses can be found in table 7 and 8.
Model 2 for exploration and exploitation without the influence of moderating effects,
explain 37,3% (exploration) and 20,6% (exploitation) of the total variability based on
the adjusted R
2. In addition, these models are significant (P<0.001) based on the F-
value. The model shows that all stages of the NSD process except service launch for
explorative innovation have a positive direct relationship on the dependent variables
exploration and exploitation. However, only the factors business analysis and
marketing strategy (P<0.001) technical development (P<0.05) and testing (P<0.1) are
significant for explorative innovations, while it is interesting to mention that
formalisation has a significant negative effect (P<0.1) on exploration.
Model 1 EXPLOR
ß
Model 2 EXPLOR
ß
Model 3 EXPLOR
ß
Model 4 EXPLOR
ß
Model 5 EXPLOR
ß Constant 4,627*** 2,135*** 5,289*** 4,666*** 4,640***
Control
Age -0,002 -0,003** -0,003** -0,002* -0,002*
Size 0,000 0,000 0,000 0,000 0,000
Independent Variables
Idea Generation (IG)
0,048 0,060 0,034 0,055
Business Analysis &
Marketing Strategy (BA)
0,388*** 0,374*** 0,351*** 0,334**
Technical Development (TD)
0,186** 0,176** 0,198** 0,166*
Testing (TE) 0,73* 0,078* 0,073 0,096*
Service Launch (LA)
-0,72 -0,050 -0,099 -0,062
Formalisation (FO)
-0,139* -0,148** -0,148* -0,180**
Moderation Effects
FO - IG -0,041 -0,043
FO - BA 0,049 0,041
FO - TD -0,029 -0,032
FO - TE 0,082** 0,086**
FO - LA 0,067* 0,045
Health - IG 0,092 0,078
Health - BA 0,136 0,129
Health - TD -0,061 0,011
Health - TE 0,034 -0,043
Health - LA 0,034 -0,021
Health - FO 0,371* 0,386*
Model
Summary
F 1,066 12,168*** 8,683*** 6,894*** 5,825***
R2 0,14 0,407 0,452 0,434 0,473
Adjusted R2 0,001 0,373 0,400 0,371 0,391
N 150 150 150 150 150
*P<0.1, **P<0.05, ***P<0.001
Table 7: Regression analyses (DV = explorative innovation)
Additionally, only business analysis and marketing strategy is significant for
exploitative innovations. Multicollinearity is not an issue as for all models the VIF
value (<5) and tolerance (>0,10) were in good condition. The control variables age
and size in model 1 for both exploration and exploitation are not significant.
Model 1 EXPLOIT
ß
Model 2 EXPLOIT
ß
Model 3 EXPLOIT
ß
Model 4 EXPLOIT
ß
Model 5 EXPLOIT
ß Constant 5,449*** 3,670*** 5,431*** 5,501*** 5,403***
Control
Age 0,002 0,001 0,001 0,001 0,001
Size 0,000 0,000 0,000 0,000 0,000
Independent Variables
Idea Generation (IG)
0,057 0,057 0,082* 0,085*
Business Analysis &
Marketing Strategy (BA)
0,160** 0,170** 0,113 0,123*
Technical Development (TD)
0,079 0,102* 0,029 0,028
Testing (TE) 0,055 0,045 0,072* 0,054
Service Launch (LA)
0,055 0,076 0,079 0,147*
Formalization (FO)
-0,045 -0,013 -0,071 -0,029
Moderation Effects
FO - IG -0,009 -0,021
FO - BA -0,039 -0,053
FO - TD 0,004 -0,020
FO - TE -0,047* -0,043*
FO - LA 0,091** 0,116*
Health - IG -0,064 -0,053
Health - BA 0,159 0,216
Health - TD 0,093 0,132
Health - TE -0,047 0,042
Health - LA -0,062 -0,280**
Health - FO 0,264 0,220
Model
Summary
F 1,303 5,875*** 5,434*** 3,559*** 4,040***
R2 0,017 0,249 0,340 0,283 0,383
Adjusted R2 0,004 0,206 0,278 0,204 0,288
N 150 150 150 150 150
*P<0.1, **P<0.05, ***P<0.001
