Int. J. Technology Management, Vol. 61, No. 1, 2013
Exploring the exploratory search for innovation: a structural equation modelling test for practices and performance Davide Aloini* and Antonella Martini Department of Energy and System Engineering (D.E.S.E.), University of Pisa, Largo Lucio Lazzarino 1, 56100, Pisa, Italy E-mail:
[email protected] E-mail:
[email protected] *Corresponding author Abstract: Identifying, developing and commercialising innovations through traditional approaches are often ineffective or inefficient when discontinuous conditions occur. New or additional capabilities and practices are desirable for organisations under such complex conditions. Based on a comprehensive literature review about search practices and drawing on the empirical background of a survey to 500 Italian mediumand high-tech companies (by the Discontinuous Innovation Lab), the relations between search practices, exploration activities and firm performance are tested by structural equation modelling. Evidences shows that higher levels of search practices lead to higher levels of exploration activities and this, in its turn, can affect firm innovation performance. Keywords: exploratory search; search practices; performance; structural equation modelling; discontinuous innovation. Reference to this paper should be made as follows: Aloini, D. and Martini, A. (2013) ‘Exploring the exploratory search for innovation: a structural equation modelling test for practices and performance’, Int. J. Technology Management, Vol. 61, No. 1, pp.23–46. Biographical notes: Davide Aloini is an Assistant Professor at the Faculty of Engineering, University of Pisa. He received his BS and MS in EconomicManagement Engineering at the University of Pisa and then his PhD in Management Engineering in 2008. His research interests include supply chain management, information management (ERP and e-procurement systems), risk management and innovation management. Antonella Martini is an Associate Professor of ‘Business Economics and Organisation’ at the Faculty of Engineering, University of Pisa. She graduated in Business Economics and obtained her PhD in Managerial Engineering in 2000. Involved in national and international research projects on innovation and knowledge management, she is a member of the international board of the CINet and author of more than 70 international publications. Her research interests refer to the role of ICT in innovation (Enterprise 2.0, social software, community management, co-creation) and ambidexterity.
Copyright © 2013 Inderscience Enterprises Ltd.
23
24
D. Aloini and A. Martini This paper is a revised and expanded version of a paper entitled ‘Search practices and performance for discontinous innovation: first results of the test’ presented at 11th International CINet Conference ‘Practicing Innovation in Times of Discontinuity’, Zürich, Switzerland, 5–7 September 2010.
1
Introduction
Innovation can be defined as a process made up of three subsequent phases named respectively search, select and implement (Bessant et al., 2005). The success of every innovative action is rooted in the very early phase of the process where firms look both inside and outside in search for new ideas to renew themselves. Most innovation literature is about steady state conditions, where firms are focused on ‘doing what we do but better’, while discontinuous conditions – i.e., ‘doing differently’ – is less researched (Bessant, 2008). At the same time, the old exploration/exploitation question is challenged by the scope of changes in environmental elements and by the level of interactivity amongst them, which means that discontinuities incidence is likely to rise. In other term: to survive firms have to satisfy, at the same time, the requirements of today’s customers (in terms of functionality, price, time, quality, place and service) and the needs of tomorrow’s customers, in a context that cannot be expected to have the same characteristics as in the past. The first requirements develop along a defined pathway – i.e., existing trajectories – and produce incremental as well as radical innovation. The second ones instead also result in small – i.e., incremental – or large – i.e., radical – innovation but refer to new trajectories – i.e., new directions. Therefore, it can be argued that the exploit/explore binary divide is referred to the direction of the search activity: ‘same direction’ – i.e., along a defined pathway – or ‘different direction’ – i.e., beyond the envelope, out-of-the-box; whilst in both situations we have small or large steps in terms of incremental and radical innovation. Discontinuous innovation involves a fundamental change in the approach or technology. This article refers to innovation under the general context of discontinuous conditions – i.e., innovation beyond the envelope, which occurs under environmental complexity – and it refers specifically to the search phase of the innovation process: the exploratory search. A questionnaire submitted to a 500 high-tech Italian firm sample was developed and exploratory search practices and innovation performance were investigated by structural equation model (SEM) methodology. Specifically, the article provides the first evidences of the exploratory innovation process, analysing: a
the search phase
b
its outcomes in terms of developed competences and innovation
c
the final results in terms of obtained performance.
On the contrary, previous contributions focused on single aspects of the whole process. The article is organised as follows: Section 1 is the Introduction; Section 2 reports the research design, the conceptual model, constructs and hypothesis; methodology is in Section 3 and analysis and results are in Section 4. We discuss contributions and limitations in Section 5.
Exploring the exploratory search for innovation
2
25
Research design
Innovation can be described as a problem of search, selection, implementation and capture (Bessant et al., 2005). Search focuses on how to find opportunities for innovation, while selection refers to what to do and why; finally, implementation and capture conceive respectively. how to make it happen and how to get the benefits from it. Search phase resides at the fuzzy front end (FE), and specifically at the ‘early’ front-end, where activities are related to generate ideas, managing ideas and locate opportunities (Crawford and Di Benedetto, 2005) while selection and concept development refer to the ‘later’ front-end – i.e., selection phase. Figure 1
Search phase of the innovation process (see online version for colours)
SEARCH
SELECTION
IMPLEMENTATION
Late front end
Early front end Fuzzy front end
NPD
According to literature, in the search phase there are two main activities: idea generation and idea management. Idea generation refers to environmental scanning, idea seeding and opportunity identification; idea management is the process of capturing, storing, and organising ideas to be adopted in the late FE processes. Under discontinuous conditions, firms need to develop the capacity to ‘see’ weak early warning signals, extending their natural steady state search space (Day and Schoemaker, 2006). But it is often firms that excel at managing innovation in a steady state environment that suffer most when discontinuous shifts occur (Christensen, 1997; Leonard-Barton, 1995). These firms typically deploy ‘best practice’ steady state routines (Kahn et al., 2005) – i.e., they work closely with customers/suppliers, and they make use of sophisticated resource allocation mechanisms in order to select a strategically relevant projects portfolio. They also, use advanced project and risk management approaches in developing new products, services and operative processes. These ‘exploit’ search routines are the product of well-developed adaptive learning processes that give firms a strong position in managing innovation under steady-state conditions. However, the same routines may also act as barriers to detecting and responding to innovation threats and opportunities, associated with discontinuous shifts. In fact, in such context, the ‘explore’ search behaviour involves radical, ‘do different’, generative learning. These two search activities are different in nature thus requiring very different routines to be enabled (Von Stamm, 2008). For this reason it is often new entrant firms who are best able to exploit the ‘fluid phase’ and develop innovations to take advantage of the changing conditions (Christensen, 1997). While a ‘deep search’ (or ‘exploit’ search) seems to secure success in a ‘steady state’, it is generally a ‘wide search’ (or ‘explore’ search) that it is required to succeed in discontinuous conditions. Different search practices are required for these two different types of search and this has only captured researchers’ attention in the last decade (McDermott and Colarelli O’Connor, 2002).
26
D. Aloini and A. Martini
Based on a literature review and on the results of more than 80 case studies developed by the DI-Lab researchers (Bessant, 2008), we identified bundles of search practices to be adopted under discontinuous conditions (from here on the term, ‘search practices’ refers only to discontinuous conditions – or DI). These practices do not yet appear as ‘routines’ but are rather indicators of emerging patterns and trajectories around which such routines may form. The process of developing and codifying routines for discontinuous conditions will require extensive experimentation – essentially a trial and error learning phase aimed at building a relatively structured set of approaches for dealing with them.
2.1 Literature review It is possible to review literature on search topics according to two main perspectives: where to search and how to search. Contributions on the first perspective (where) refer to the choice of: a
knowledge boundary (internal and external)
b
knowledge domain (market and technology)
c
knowledge proximity (local and distant)
d
search intensity and scope (depth and breadth).
Literature on the second perspective (how) investigates the practices used to organise search. These practices are behaviours as well as structures or processes that deal with search for innovation.
2.1.1 ‘Where’ to search: the directions Contributions on the boundary area (a), consider whether an organisation is drawing on internal or external knowledge sources. External sources for innovation are customers (Von Hippel, 1988), suppliers (Pavitt, 1984), competitors (Dussauge et al., 2000) and research centres (Link et al., 2006). For the first two sources of knowledge the impulse is market-driven, while in the others is technology-driven (Sofka and Grimpe, 2010). Other external sources are: conferences, trade, fairs, press (Frost et al., 2002). Internal sources instead are individuals who can act as boundary spanning, champions, scouts, idea generators, gatekeepers (Reid and De Brentani, 2004). Here, issues such as knowledge management (KM), idea generation, incentives have great impact. Recent literature pointed out the importance of Openness in the search field (Chesbrough, 2003c) and refers to external knowledge identification as search strategy (Laursen and Salter, 2006; Sofka and Grimpe, 2010). The openness to external knowledge has been defined according to the search areas (d) breadth and depth. The breadth is measured by the diversity of external inputs and represents how widely a firm explores external knowledge. The depth instead is defined in different ways by different authors: for Katila and Ahuja (2002), it is the degree to which existing knowledge is reused (or exploited); for Laursen and Salter (2006), it represents how deeply a firm draws knowledge from external sources. On the same line, Sofka and Grimpe (2010, p.4) argue that firms need to specialise their search strategies “as a balancing act between fruitful diversity in potential knowledge impulses and the efficiency of how to access it”.
Exploring the exploratory search for innovation
27
They found that firms need to expand the search scope, while, on average, they only draw deeply from one source. Knowledge domain (b), is used in literature to refer either to the search source [Sofka and Grimpe (2010) distinguish between science-driven, market driven and supply driven sources] or to the generated knowledge (Lavie et al., 2011, 2006). Katila and Ahuja (2002, p.1184) add the local vs. distant search areas (c) to refer to how the new knowledge acquired by firms is closely related to its pre-existing knowledge base: local search (or ‘exploit’ search) on one end of the spectrum, and distant search (or ‘explore’ search) at the other end when “search behaviours involve a conscious effort to move away from current organizational routines and knowledge bases”. To sum up: in current literature there are different dimensions and many areas to analyse search; empirical articles focus on some of them and sometimes definitions adopted are different for the same concept. As above explained, when dealing with DI, the problem is not the search lack-or-presence but the overall search direction. This can be constrained as bounded exploration if a firm, in order to extend the boundary of a known business frame, continues to direct its efforts along a defined pathway. On the contrary, it can be unconstrained by the old rules of the game and be directed towards a re-framing of different elements in the environment or even directed towards an open space where completely new game can emerge. Whilst consideration is given to question of exploit vs. explore search, our goal is to focus on the successful behaviours, structures and processes experimented by firms in dealing with DI search. In other term, our analysis object is the search practice. Given this, in order to map these practices, the where-dimension search areas are not so useful, since a practice can be referred to not just a single dimension and search area but to many of them. Considering that the search phase resides at the early FE of the innovation process, it can be useful instead, to refer to its macro activities. On this line, O’Connor (2008) suggests taking a holistic view when studying a (radical innovation) capability as it develops from a complex system of interdependent elements. A single aspect should not be analysed as isolated from the others. Therefore, we reviewed early FE literature, adopting an holistic approach on search practices, in order to identify its scientific domain.
2.1.2 ‘How’ to search: the practices According to the early FE literature, activities can be broken up into two broad categories (Crawford and Di Benedetto, 2005): the first one groups activities dealing with the process of idea generation while the second one includes those related to the idea management. Idea generation refers to opportunity identification and analysis carried out by environmental scanning (Flynn et al., 2003; Kim and Mauborgne, 2005), seeding ideas (Hargadon and Sutton, 2000; Gamlin et al., 2007), application exploration (Thongpapanl et al., 2008). It can occur inside or outside a business. Idea management is the process of capturing, storing, and organising ideas adopted in the late FE process. It can be used also for preliminary idea evaluation and screening as well as idea diffusion across the company (Gorski and Heinekamp, 2002; Van Dijk and Van de Ende, 2002). It integrates activities, such as ideas generation, screening, collaboration and idea development, from the early and late innovation FE.
28
D. Aloini and A. Martini
Glassman (2009) highlights some of the current literature shortcomings on idea generation and management: particularly the need to integrate the two categories and the need to include KM, which is transversal to both categories (Hargadon and Sutton, 2000; Flynn et al., 2003), in the fuzzy FE process. The how-to-search dimension refers to organisational practices used for idea generation and management activities (KM included). It is possible to identify in the literature a number of recurrent themes within these interlinked categories. These are summarised in Table 1, with their main references. Table 1
Search domain
Dimension Learning about markets
Openness to external sources
Managing idea generation
Network management system
Measurement focus Main references Idea generation Practices such as lead Thomke (2001), O’Connor (1998), users, experimentation, O’Connor and Veryzer (2001), Tidd et al. unconventional tools (2005), Christensen et al. (2005), Von Hippel and Katz (2002), Hienerth (2006), and Kim and Mauborgne (2005) Practices that enable the Chesbrough (2003a, 2003b), Chesbrough search breadth (2004), Gassmann (2006), and West and Gallagher (2006) Idea management and KM A system for capturing Leifer et al. (2000), Flynn et al. (2003), ideas to look both inside Andriopoulos and Gotsi (2005), and and outside the firm, Hargadon and Sutton (2000) idea hunters, dedicated teams Practices related to Hargadon and Sutton (2000), O’Connor internal organisation et al. (2008), and O’Connor and McDermott (2004)
2.2 Conceptual model and hypotheses The conceptual framework (Figure 2) is composed of three key conceptual boxes: 1
search
2
exploration activities
3
performance.
Figure 2
Conceptual model and hypotheses (see online version for colours)
Firm performance
Innovation performance
Performance
H2
Exploratory innovation Competence exploration
Exploration activities
H1
Search dimensions and practices
Search
Exploring the exploratory search for innovation
29
Search box refers to the four search dimensions drawn from the literature (Table 1). These dimensions contain specific search practices derived from the literature as well as, empirically, from the 80 case studies developed by the DI-Lab and discussed with industry experts in focus groups (Bessant, 2008; Aloini et al., 2010). These bundles of search practices, which support each other, do not yet appear as ‘routines’. They are rather indicators of emerging patterns and trajectories around which such routines may form. In fact, the process of developing and codifying routines for discontinuous conditions will require extensive experimentation. In specific they will need a learning though trial and error, leading to a relatively structured set of approaches for dealing with innovation in complex environments. The practices in the box are used to ‘explore’ searching: i.e., to search in new directions – beyond the envelope. The rationale of the conceptual model posits that search practices have an impact on the exploration activities box, which specifically refers to competence exploration and to exploratory innovation. Competence exploration is “… the tendency of a firm to invest resources to acquire entirely new knowledge, skills, and processes. Its objective is to attain flexibility and novelty in product innovation through increased variation and experimentation” [Atuahene-Gima, (2005), p.62]. It reflects search behaviour processes that engender procedural knowledge or skill (Kogut and Zander, 1992). Exploratory innovation “… denote technological innovation activities aimed at entering new product-market domains” [He and Wong, (2004), pp.483–484]. Exploratory innovations are radical innovations that offer new designs, create new markets, and develop new channels of distribution (Jansen et al., 2006). In other words, the explore search is linked with distant search (see the type of knowledge/innovation generated discussed in § 2.1). Performance box is composed of innovation and firm performance. We analyse exploratory activities, therefore the reference is to radical innovation performance. These activities involve fundamental changes in technology for the firm, typically address the needs of emerging customers, who are new to the firm and/or industry, and offer substantial new benefits to customers (Chandy and Tellis, 1998). By exploiting existing competences a firm can develop successful innovation. Exploratory innovations require new knowledge or departure from existing knowledge (Jansen et al., 2006; Benner and Tushman, 2002). As competence exploration involves experimentation and focuses on emerging markets and technologies for ideas, it is supposed to generate exploratory innovation (Attuahene-Gima, 2005; Miller et al., 2007). Therefore, we posit the following hypothesis: H1 High level of search leads to high level of exploration activities. We modelled the impact of exploratory activities on firm performance as a two-stage process in which activities affect innovation performance that, in turn, affects firm performance. This path model approach is also consistent with Eisenhardt and Martin’s (2000, p.1106) argument that “dynamic capabilities are necessary, but not sufficient, conditions for competitive advantage”, hence their impacts on firm performance must be measured through their effects on the firm’s resource configuration (new products in our case). Therefore, we posit the following hypothesis: H2 High level of search leads to high level of exploration activities.
MKT
MNG
NET
OPEN
INN_EXR
KW_EXR
PERF_INN
Managing idea generation
Network management system
Openness to external sources
Exploratory innovation (He and Wong, 2004; Lubatkin et al., 2006; Cao et al., 2009)
Competence exploration (Zahra et al., 2000; Atuahene-Gima, 2005)
(Radical) innovation performance (Atuahene-Gima, 2005; Chiang and Hung, 2010)
Item
Learning about market
DI search practices
Construct
3
3
4
2
3
Over the last three years: Number of new product innovation respect to main competitors % of sales of new product innovations Business innovation degree (products are new or improved significantly)
Learned product development skills and processes (such as product design, prototyping new products, timing of new product introductions, and customising products for local markets) entirely new to the industry Acquired entirely new managerial and organisational skills that are important for innovation (such as forecasting technological and customer trends; identifying emerging markets and technologies; coordinating and integrating R&D; marketing, manufacturing, and other functions; managing the product development process). Strengthened innovation skills in areas where it had no prior experience.
Introduction of new generations of products Extension of product range Opening up new markets Entering new technological fields
We have a website where outsiders can submit their suggestions and ideas for new markets, products and/or services. We use an open innovation system in which technology-related challenges are posted online by our R&D staff so that a community of registered scientists anywhere in the world can propose their solutions.
We consciously hire people who are different to encourage diversity within our organisation. In our organisation, we encourage radical innovation teams to expand their resource network by tapping into the knowledge of any employee in our firm. In our organisation, we have ‘network ambassadors’ who can help radical innovation teams connect with other people company-wide when new knowledge or insight is needed.
We encourage people to come forward with ideas, even if they have only a vague idea of the potential market applications for the idea. Our organisation supports corporate entrepreneurship as part of our discontinuous innovation efforts.
We use scenarios to help understand and influence our organisation’s future. We currently have someone (full or part-time) officially charged with scouting for new ideas outside the organisation and looking for trends and developments that might have implications for our organisation’s future. We have a dedicated group of people (e.g., from marketing, sales, R&D) that explores new ways to apply our existing technology to new industries and new customers. When we have a very new and different technology, we search for multiple applications and conduct several market experiments to discover promising markets.
Variable description
Table 2
2
4
N.
30 D. Aloini and A. Martini
Construct and variable description
Exploring the exploratory search for innovation
31
2.3 Constructs operationalisation For construct operationalisation, we used scales which are well consolidated in the literature for exploration activities and firm performance (see Table 2 for specific references), while for search practices we used the measurement scale previously validated in other of our own papers (see, for details, Aloini et al., 2010). Search dimensions are modelled as a first-order latent construct (factor) consistent with their conceptualisation as a ‘bundle of practices’. Each set of practices (i.e., the specific search dimensions) is measured using a set of items (i.e., the specific search practices). The search dimensions measure the search capacity, modelled as a second-order latent construct. We define search capacity, which is yet a new and not consolidated construct in the innovation management field (Bessant, 2008), as “the capacity to perform the ‘explore’ search activities under discontinuous conditions”. We also compared the measurement scale for search with the results from 80 case studies developed by DI Lab researchers. The case analysis led to the identification of 12 search strategies (more details in Bessant and Von Stamm, 2008). Each search strategy is a coherent bundle of search practices. Table A (in the Appendix) reports the description of these strategies and the comparison with the search construct previously validated and used in this study. All the search strategies are represented in the validated search construct: some strategies (n. 2 and n. 7) refer to two search dimensions, as the specific search practices they are composed of report to one or the other dimension. To measure innovation performance we included two intermediary variables: product innovation intensity and number of radical products introduced by the firm in the last three years. Product innovation intensity is measured as the percentage of total annual sales of new/highly improved products introduced over the last three years. Natural log of product intensities were used to compensate for skewness. Prior research has shown that product innovation intensity has positive impacts on firm performance in general and on sales growth in particular (e.g., Skinner, 1992; Zahra and Das, 1993; Zairi, 1992), thus justifying our choice for it as intermediary variable. Finally, firm performance takes into account the sales trend over five years and controls for the effect of trends in the sector. Data were gathered by the AIDA dataset (2010) and are computed as the logarithmic growth rate of sales revenues between 2006 and 2009 since it is recognised to be a reliable proxy indicator of other dimensions of superior firm performance, including long-term profitability and survival (Timmons, 1999; Henderson, 1999).
3
Methodology and data collection
The research model has been evaluated using a structural equation modelling (SEM) approach. In last decade, the use of structural equation modelling has gained increasing attention in both operation management and economic-managerial research (Shah and Goldstein, 2006) since SEM presents substantial advantages over the first-generation techniques, such as principal components analysis, factor analysis or multiple regression (Chin, 1998).
32
D. Aloini and A. Martini
3.1 Methodology The SEM methodology main phases are briefly described here (Hair et al., 1998): a
Specify a theoretically-based model. Definition of the theoretical hypothesis as a structured model of constructs and links. To this end, the first part of this paper has presented the current RI theory and grounded the different model variables and relations in the emerging literature.
b
Define the structural and measurement models. Formal model specification through a series of equations (or a graph) that define structural equations linking constructs. The model formal specification is presented in Section 2.
c
Test the model: 1 Collect data for model testing. Determining the minimum sample size for adequate model fit and power is complex and at the same time essential to the parameter estimates reliability, the model fit and the equation statistical power. In general, structural equation modelling requires large sample size. For example, Hair et al. (1998) recommends a minimum 100–200 sample size when using the maximum likelihood estimation (MLE) procedure. 2 Test the measurement model. A confirmatory factor analysis (CFA) is applied in order to assess the construct dimensionality, reliability and validity (Gerbing and Anderson, 1988) including content validity, convergent validity and discriminant validity. 3 Test the structural model. It aims at testing the causal links between the theoretical variables, by analysing the sign, magnitude and statistical significance of the structural path coefficients. 4 Evaluate the model fit. Use goodness-of-fit measures to determine how adequately the model accounts for the data. It is often recommended that the model be evaluated from different perspectives, so that multiple conformity (fit) measures are desirable (Shah and Goldstein, 2006).
d
Interpret the results and refine the model. For this purpose modification index and other useful indications are provided by SEM software.
3.2 Sample and data collection process An online cross-sectional survey was used for data collection. The survey was built on DI phenomenon theoretical knowledge, concepts, models and propositions. Its objective is to describe the purposes and practices of DI search. A structured questionnaire was developed in order to measure the theoretical constructs while five-point Likert scales, with ‘strongly disagree’ and ‘strongly agree’ end points, were used to measure the items. A test of the resulting questionnaire was conducted on two groups of subjects: colleagues and target respondents. These two tests were conducted independently and lead to improvement and update of the survey instrument.
Exploring the exploratory search for innovation
33
The target sample frame consisted of Italian medium and high-tech companies selected according to the international OECD science classification. 500 firms were randomly selected from all the AIDA (2010) companies with more than 50 employees and covering the specific two-digit ATECO (2007) codes 20, 21, 25, 26, 27, 28, 29, 30, 32. Analysed sectors include the areas reported in Table 3. The data collection process was supported by the use of Survey Monkey® online service. Respondents were typically R&D department vice president or director or CEO. In the effort to increase the response rate, all target firms were firstly contacted by phone in order to present the initiative. Then, an e-mail including a cover letter of the survey and the survey access (Monster account) were sent to the selected respondents. Two weeks after the initial contact, reminders were sent to all potential respondents by both phone and e-mail. Of the 500 surveys mailed in Italy, 112 responses were received, resulting in a 22.4% response rate. A total of 16 were discarded due to incomplete information, resulting in an effective 19.2% response rate. Statistics on the employees number in surveyed companies are reported in Table 4. Table 3
Business sectors in the sample
High-tech
Aerospace
3%
Computers, office machinery
10%
Electronics-communications
3%
Pharmaceuticals
10%
Medium-high tech
Table 4
Scientific instruments
3%
Motor vehicles
9%
Electrical machinery
22%
Chemicals
15%
Other transport equipment
3%
Non-electrical machinery
22%
Number of employees of companies in the sample
Number of employees
%
Less than 100 employees
35%
100–200 employees
14%
200–500 employees
20%
500–1,000 employees
11%
More than 1,000 employees
19%
Missing data were automatically managed by the software AMOS 7.0, which uses a procedure known as full information maximum likelihood (FIML, also known as ‘raw maximum likelihood’). FIML seems to outperform most common methods of handling missing data, including list-wise and pair-wise data deletion, mean substitution and the similar response pattern imputation (SRPI) procedure, implemented in LISREL 8.30 and higher.
34
D. Aloini and A. Martini
Finally, in order to test the non-response bias (Lambert and Harrington, 1990) we compared the early and late respondents by a t-test. The early wave group consisted of 66 responses while the late wave group consisted of 46 responses. The t-tests performed on these two groups responses yielded no statistically significant differences (at 99% confidence interval).
4
Analysis and results
In the initial part of the analysis we assessed the measurement model; after that, causal relationships were introduced in order to evaluate the full structural equation model.
4.1 Measurement model test The test of the measurement model aims to validate unidimensionality, reliability and scale validity (the last includes: content validity, convergent validity and discriminant validity) of adopted constructs and measures. We have assumed content validity to be maintained, since the constructs are well-grounded in the literature. Reliability of constructs was then tested adopting the internal consistency method estimated by using Cronbach’s alpha. Typically, reliability coefficients of 0.70 or higher are considered adequate (Cronbach, 1951; Nunnally, 1978). Moreover, Nunnally (1978) states that permissible alpha values can be slightly lower (0.60) for newer scales. This is certainly also the case of the search construct reported in this analysis. Finally, an AMOS-based CFA was applied to the entire set of constructs, in order to assess ‘convergent validity, discriminant validity and uni-dimensionality’, since it is considered a more powerful tool and it requires fewer assumptions than the traditional Campbell and Fiske MTMM matrix method (O’Leary-Kelly and Vokurkar, 1998). •
Convergent validity is considered verified if individual item’s path coefficient is greater than twice its standard error (Gerbing and Anderson, 1988), or alternatively by examining the loadings and their statistical significance through t-values (Dunn et al., 1994). It is verified when the factor loadings of each construct are significantly high. Carr and Pearson (1999) consider the observed variables proportion of variance (R2) in order to estimate the indicator reliability (R2 values above 0.30 are considered acceptable).
•
Discriminant validity is established (after convergent validity) verifying that correlations for all possible pairs of latent constructs are significantly different from 1.
•
Uni-dimensionality was established by assessing the overall model fit of the general model by AMOS. As reported in the following part, all the selected indexes respect the goodness threshold for a very good fit so that the test can be considered successful.
These tests were applied to validate the measurement models of both search and explorative activities (EA) construct. In Tables 5 and 6, we present the discriminant validity test output for each construct, while the CFA’s outputs are reported in Tables B and C of the Appendix.
Exploring the exploratory search for innovation
35
The model fit indexes (χ2 53.75; RMSEA 0.06; NFI 0.87; N-NFI or TLI 0.93; CFI 0.96; IFI 0.96; PNFI 0.53; PCFI 0.58; χ2/df 1.34) together with the previous results, show that all the theoretical constructs exhibit acceptable levels of reliability, validity, and unidimensionality. Results also show that quite all the indexes respect the selected threshold thus confirming the constructs reliability and scale validity. After the preliminary analysis and the model refinement, two critical point still remain in the measurement scale. The first point concerns the search construct and particularly the ‘openness to external sources’ factor, which present a .56 Cronbach’s alpha Index for the group. The second one is about the innovation construct and the firm performance: while the first presents a quite low Cronbach’s alpha Index (0.52), the latter shows two items (items 3 and 4) with a considerably lower significance than the other ones. At first we considered dropping off these items and modifying the constructs in order to improve the measurement scales. However we finally decided not to further modify the model because of the adopted scale novelty which allows us to accept lower cutting thresholds and it also leaves us the possibility to improve indicators by using a larger sample. Table 5
Search construct
Assessment of discriminant validity: chi-square differences between fixed and free models A A
B
C
D
-
B
9.55**
-
C
11.49***
6.28*
-
D
17.93***
6.33*
12.95***
-
Notes: *Significant at the 0.05 level, **significant at the 0.01 and ***significant at the 0.001 level (for 1 d.f.) Table 6
EA construct
Assessment of discriminant validity: chi-square differences between fixed and free models Innovation exploration Innovation exploration Competence exploration
Competence exploration
20.505***
-
Notes: *Significant at the 0.05 level, **significant at the 0.01 and ***significant at the 0.001 level (for 1 d.f.)
4.2 Structural model test The full structural model was finally tested. Results are summarised in Figure 3 which also reports the most relevant SEM parameters and significance (construct covariance, regression weights and factor loadings). The hypothesised relationships between search, exploration activities and firm performance constructs are supported by the data. Regression weights are respectively .73 (significant at the .001 level) and .84 (significant at the .001 level). Details about the estimated impact are reported in Figure 3 and in the Appendix.
36
D. Aloini and A. Martini
We also checked for the significance of the link between the innovation performance and the financial performance. The test was performed out from the SEM framework. The standardised regression weight is .21 and significant at the .01 level. Furthermore, as recommended, a multiple fit indexes set (Table 7) was used to check the measurement scale goodness-of-fit with data. Goodness-of-fit criteria evaluate how well the collected data fit the proposed model. They are generally categorised into three groups representing (absolute) model fit, (incremental) model comparison, and model parsimony (Schumacker and Lomax, 1996). In conclusion, these evidences provide an overall good support for the results to be deemed an acceptable representation of the hypothesised constructs. Structural model test
.70
.70
Note: Error terms are omitted for clarity purpose.
.59
.77
OPENNESS TO EXTERNAL SOURCES
.59 NET_4
.56
.71
NET_3
.49
.81
NET_2
.52
NETWORK MANAGEMENT
.77 MNG_3
.53
.98
.96
.75
MNG_2
.51
.70 MKT_10
.38
.72 MKT_9
MKT_7
.73
MKT_6
.72
KW_EXR_4
.69
.61
.69
IDEA GENERATION
KW_EXR_3
.83
ABOUT MARKET
.84
.75
LEARNING
.98
KW_EXR_2
.56
COMPETENCE EXPLORATION
SEARCH
.83
INN_EXR_4
.71
.84 .73
INN_EXR_3
.66
.59
.34
.64 OPEN_6
INN_PERF_2
INN_PERF_1
INN_PERF_2
EXPLORATION ACTIVITY
INN_EXR_2
.50
.95
INN_EXR_1
.45
.84
.67
INNOVATION PERFORMANCE
.54
..73 ..69 .69
NNOVATION
.58
EXPLORATORY
.30
.70
.90
.48
.33
.48
.09
.50
.49
OPEN_4
Figure 3
.40
.46
Exploring the exploratory search for innovation Table 7
37
Fit indices for the full structural model
Index
Value
Recommended values for a good fit
Recommended values for very good fit
Sources
χ2
223.3
-
-
-
RMSEA
0.05
< .08
< .05
Byrne (1998)
NFI
0.76
> .8
> .9
Byrne (1998) and Zhang et al. (2002)
N-NFI or TLI
0.92
> .8
> .9
Byrne (1998) and Zhang et al. (2002)
CFI
0.94
> .8
> .9
Byrne (1998) and Zhang et al. (2002)
IFI
0.94
> .8
> .9
Byrne (1998) and Zhang et al. (2002)
PNFI
0.59
-
> .5
Byrne (1998) and Mulaik et al. (1989)
PCFI
0.74
-
> .5
Byrne (1998) and Mulaik et al. (1989)
χ2/df
1.23
> 1 and < 5
> 1 and < 3
Bollen (1989), Carmines and McIver (1981), Hair et al. (1995) and Jöreskog (1969)
5
Discussion and conclusions
In this study, we examined how firms search ‘out-of-the box’ for new products (i.e., explore search) and how this may have an impact on EA and on firm performance. The structural test analyses and quantifies links among search practices, exploratory activities and performance. Two hypotheses are confirmed: H1 High level of (explore) search leads to high level of exploratory activities. H2 High level of EA leads to better firm performance. We discuss results on the basis of the two following points: 1
What is old and what is new in H1 and H2 evidences? •
Old. The (unexpected) result of Katila and Ahuja (2002) is the linear effect of search scope (which corresponds to the theoretical notion of new knowledge exploration) on new product innovation, instead of the expected curvilinear relationship. Laursen and Salter (2006)’s work links search strategy, measured in terms of depth and breadth, to innovative performance, which s proves that searching widely and deeply is curvilinearly related to performance. Sofka and Grimpe (2010) support the work of Laursen and Salter (2006) and Katila and Ahuja (2002), founding a positive relation between search strategies and innovation performance. Chiang and Hung (2010) found that open search
38
D. Aloini and A. Martini breadth (measured in terms of number of external channels opened by the innovating company to draw knowledge for innovation) is positively related to radical innovation performance. Chen et al. (2011) focus on the impact of openness scope, depth and orientation on innovative performance. They found that both scope and depth of openness have a positive impact on innovative performance. However, in line with Laursen and Salter (2006), they also argued that too many external relationships can worsen innovative performance for STI (science-technology-and innovation) firms. •
2
New. Prior contributions on search (Katila and Ahuja, 2002; Laursen and Salter, 2006; Sofka and Grimpe, 2010; Lin and Wu, 2010) have frequently adopted the where-to-search perspective (see § 2.1 for details): the choice of knowledge boundary (internal and external), knowledge domain, knowledge proximity (local and distant) and search intensity and scope (depth and breadth). In addition, these contributions operationalised the search construct by using patents (Katila and Ahuja, 2002; Lin and Wu, 2010) or knowledge source actors (Sofka and Grimpe, 2010; Chiang and Hung, 2010). On the contrary, we adopt the how-to-search perspective, and investigate the organisational practices used for searching. These practices are behaviours and accompanying structures and processes that deal with search for innovation. While ‘steady state’ innovation involves the problem of systematic search within known or ‘knowable’ selected environments, discontinuous innovation requires a much more open ended and agile approach to manage both innovation and emergent business fields where search strategies are difficult to predict in advance. Currently, there is not a well-codified ‘best practice’ model for this. Rather, firms find themselves in the ‘pre-routine’ stage of capability development, using trial and error experimentation to approach practices which work and may become routine in the future. This phenomenon has only found researcher’s attention in the last decade. The how-to-search perspective allows us to take into account the richness of a search strategy that the where-to perspective can only explore superficially.
The structural test analyses and quantifies links among exploratory activities and performance, so including the role of competence exploration. What is old and what is new? •
Old. The majority of contributions analysed the relationship between exploration and performance, without considering the competences role. Atuahene-Gima (2005) and Miller et al. (2007) explored the role of competence exploration (and exploitation) on innovation performance and found a positive relation.
•
New. SEM modelling was chosen to perform the test of the overall model, as it represents a more powerful tool than factor analysis and traditional techniques: it enables contemporaneous dimensionality checks, convergent validity, discriminant validity and construct reliability, which are all necessary for a secure (not disputable) evaluation of the structural analysis.
Exploring the exploratory search for innovation
39
The final model has a number of interesting implications, both for research and practice: it is, to the best of our knowledge, the first time that a practice-based (DI) search measurement scale is empirically validated and used. Even if improvements are still desirable, findings can be interpreted as a recommendation for future model refinement. The study is not normative (different generalisation problems occur) and firstly must be interpreted in order to offer some insight for future research or to set the stage for further analysis and model improvements. In order to get a more complete perspective of the phenomena, future empirical investigations should aim at exploring the presence of mediating variables and their effect on performance, as well considering the overall role of contextual variables. The influence of the context in which a company operates may have a deep impact on the model. The present study may also develop to shed light on the search strategies-performance link, which is not completely understood, by using the search practice construct.
Acknowledgements We would like to thank the anonymous reviewers for their valuable feedbacks which have contributed to improve the final version of this work and Dr. Niccolò Casini for his effort in supporting the analysis.
References AIDA Italian dataset (2010) Available at https://aida.bvdep.com/version-2012426/cgi/template.dll? product=26 (accessed on 31 December 2011). Aloini, D., Cantarello S., Martini, A. and Nosella, A. (2010) ‘Search practices for discontinuous innovation: a test for constrcuts and measurement in Italian high tech companies’, R&D Management Conference, Manchester, 30 June to 2 July. Andriopoulos, C. and Gotsi, M. (2005) ‘The virtues of ‘blue sky’ projects: how lunar design taps into the power of imagination’, Creativity and Innovation Management, Vol. 14, No. 3, pp.316–324. ATECO (2007) Coding sheet, available at http://www.istat.it/it/archivio/17888 (accessed on 30 October 2011). Atuahene-Gima, K. (2005) ‘Resolving the capability-rigidity paradox in new product innovation’, Journal of Marketing, October, Vol. 69, No. 10, pp.61–83. Benner, M.J. and Tushman, M.L. (2002) ‘Process management and technological innovation: a longitudinal study of the photography and paint industries’, Adminstrative Science Quarterly, Vol. 47, No. 4, pp.676–706. Bessant, J. (2008) ‘Dealing with discontinuous innovation: the European experience’, International Journal of Technology Management, Vol. 42, No. 1, pp.35–50. Bessant, J. and Von Stamm, B. (2008) ‘Search strategies for discontinuous innovation’, in Bessant, J. and Venables, T. (Eds.): Creating Wealth from Knowledge: Meeting the Innovation Challenge, Elgard Publishing, UK. Bessant, J., Lamming, R., Noke, H. and Phillips, W. (2005) ‘Managing innovation beyond the steady state’, Technovation, Vol. 25, No. 12, pp.1366–1376.
40
D. Aloini and A. Martini
Bollen, K.A. (1989) Structural Equations with Latent Variables, Wiley New York. Byrne, B.M. (1998) Structural Equation Modeling with Lisrel, Prelis, and Simplis: Basic Concepts, Applications and Programming, Lawrence Erlbaum Associates, Mahwah, New Jersey. Cao, Q., Gedajlovic, E. and Zhang, H.P. (2009) ‘Unpacking organizational ambidexterity: dimensions, contingencies, and synergistic effects’, Organization Science, Vol. 20, No. 4, pp.781–796. Carmines, E.G. and McIver, J.P. (1981) ‘Analyzing models with unobserved variables’, in Bohrnstedt, G.W. and Borgatta, E.F. (Eds.): Social Measurement: Current Issues, pp.65–115, Sage, Beverly Hills, CA. Carr, A.S. and Pearson, J.N. (1999) ‘Strategically managed buyer-supplier relationships and performance outcomes’, Journal of Operations Management, Vol. 17, No. 5, pp.497–519. Chandy, R.K. and Tellis, G.J. (1998) ‘Organizing for radical product innovation: the overlooked role of willingness to cannibalize’, Journal of Marketing Research, Vol. 35, No. 4, pp.474–487. Chen, J., Chen, Y. and Vanhawerbeke, W. (2011) ‘The influence of scope, depth and orientation of external technology sources on the innovative performance of Chinese firms’, Technovation, Vol. 51, No. 8, pp.562–573. Chesbrough, H. (2003a) ‘A better way to innovate’, Harvard Business Review, July, Vol. 81, No. 7, pp.12–13. Chesbrough, H. (2003b) ‘The era of open innovation’, MIT Sloan Management Review, Spring, Vol. 44, No. 3, pp.35–41. Chesbrough, H. (2003c) Open Innovation: The New Imperative for Creating and Profiting from Technology, Harvard Business School Press, Boston. Chesbrough, H. (2004) ‘Managing open innovation’, Research Technology Management, January to February, January, Vol. 47, No. 1, pp.23–26. Chiang, Y.H. and Hung, K.P. (2010) ‘Exploring open search strategies and perceived innovation performance from the perspective of inter-organizational knowledge flows’, R&D Management, Vol. 40, No. 3, pp.292–299. Chin, W.W. (1998) ‘Issues and opinion on structural equation modeling’, MIS Quarterly, Vol. 22, No. 1, pp.vii-xvi. Christensen, C. (1997) The Innovator’s Dilemma, Harvard Business School Press, Cambridge, Mass. Christensen, C., Cook, S. and Hall, T. (2005) ‘Marketing malpractice: the cause and the cure’, Harvard Business Review, December, Vol. 83, No. 12, pp.74–83. Crawford, C.M. and Di Benedetto, C.A. (2005) New Products Management, McGraw-Hill/Irwin, New York. Cronbach, L.J. (1951) ‘Coefficient alpha and the internal structure of tests’, Psychometrika, Vol. 16, No. 3, pp.297–334. Day, G.S. and Schoemaker, P.J.H. (2006) Peripheral Vision: Detecting the Weak Signals that will Make or Break Your Company, Harvard Business School Press, Boston, MA. Dunn, S.C., Seaker, R.F. and Waller, M.A. (1994) ‘Latent variables in business logistics research: scale development and validation’, Journal Business Logistics, Vol. 15, No. 2, pp.145–172. Dussauge, P., Garrette, B. and Mitchell, W. (2000) ‘Learning from competing partners: outcomes and duration of scale and link alliances in Europe, North America and Asia’, Strategic Management Journal, Vol. 21, No. 2, pp.99–126. Eisenhardt, K.M. and Martin, J.A. (2000) ‘Dynamic capabilities: what are they?’, Strategic Management Journal, Vol. 21, No. 10–11, pp.1105–1112.
Exploring the exploratory search for innovation
41
Flynn, M., Dooley, L., O’Sullivian, D. and Cormican, K. (2003) ‘Idea management for organisational innovation’, International Journal of Innovation Management, Vol. 7, No. 4, pp.417–442. Frost, T.S., Birkinshaw, J.M. and Ensign, P.C. (2002) ‘Centers of excellence in multinational corporations’, Strategic Management Journal, Vol. 23, No. 11, pp.997–1018. Gamlin, J.N., Yourd, G. and Patrick, V. (2007) ‘Unlock creativity with ‘active’ idea management’, Research Technology Management, Vol. 50, No. 1, pp.13–16. Gassmann, O. (2006) ‘Opening up the innovation process: towards an agenda’, R&D Management, Vol. 36, No. 3, pp.223–228. Gerbing, D.W. and Anderson, J.C. (1988) ‘An updated paradigm for scale development incorporating unidimensionality and its assessment’, Journal of Marketing Research, Vol. 25, No. 2, pp.186–192. Glassman, B.S. (2009) ‘Managing idea generation and idea management in-order to better manage the fuzzy front end of the innovation process’, PhD dissertation, Purdue University. Gorski, C. and Heinekamp, E.J. (2002) Capturing Employee Ideas for New Products, pp.219–242, The PDMA ToolBook for New Product Development, John Wiley & Sons, Inc., New York. Hair, J.F., Anderson, R.E, Tatham, R.L. and Black, W.C. (1998) Multivariate Data Analysis, 5th ed., Prentice Hall, Upper Saddle River. Hardagon, A. and Sutton, R.I. (2000) ‘Building and innovation factory’, Harvard Business Review, Vol. 78, No. 3, pp.157–166. He, Z-L. and Wong, P-K. (2004) ‘Exploration vs. exploitation: an empirical test of the ambidexterity hypothesis’, Organization Science, Vol. 15, No. 4, pp.481–494. Henderson, A.D. (1999) ‘Firm strategy and age dependence: a contingent view of the liabilities of newness, adolescence, and obsolescence’, Administrative Science Quarterly, Vol. 44, No. 2, pp.281–314. Hair, J.F., Jr., Anderson, R.E., Tatham, R.L. and Black, W.C. (1995) Multivariate Data Analysis, 3rd ed., Macmillan Publishing Company, New York. Jansen, J., Van den Bosch, F.A.J. and Volberda, H.W. (2006) ‘Exploratory innovation, exploitative innovation and performance: effects of organizational antecedents and environmental moderators’, Management Science, Vol. 52, No. 11, pp.1661–1674. Jöreskog, K.G. (1969) ‘A general approach to confirmatory maximum likelihood factor analysis’, Psychometrika, Vol. 34, No. 2, pp.183–202. Kahn, K.B., Castellion, G. and Griffin, A. (2005) The PDMA Handbook of New Product Development, 2nd ed., John Wiley & Sons. Inc., Hoboken, New Jersey. Katila, R. and Ahuja, G. (2002) ‘Something old, something new: a longitudinal study of search behavior and new product introduction’, Academy of Management Journal, Vol. 45, No. 6, pp.1183–1194. Kim, W.C. and Mauborgne, R. (2005) Blue Ocean Strategy: How to Create Uncontested Market Space and Make the Competition Irrelevant, Harvard Business School Press, Boston. Kogut, B. and Zander, U. (1992) ‘Knowledge of the firm, combinative capabilities, and the replication of technology’, Organization Science, Vol. 3, No. 3, pp.383–397. Lambert, D.M. and Harrington, T.C. (1990) ‘Measuring non-response bias in customer service mail surveys’, Journal of Business Logistics, Vol. 11, No. 2, pp.5–25. Laursen, K. and Salter, A. (2006) ‘Open for innovation: the role of openness in explaining innovative performance among UK manufacturing firms’, Strategic Management Journal, Vol. 27, No. 2, pp.131–150. Lavie, D. and Rosenkopf, L. (2006) ‘Balancing exploration and exploitation in alliance formation’, Academy of Management Journal, Vol. 49, No. 4, pp.797–818.
42
D. Aloini and A. Martini
Lavie, D., Kang, J. and Rosenkopf, L. (2011) ‘Balance within and across domains: the performance implications of exploration and exploitation in alliances’, Organization Science, ePub ahead of print, 11 March, available at http://orgsci.journal.informs.org/cgi/content/abstract/orsc.1100. 0596v1. Leifer, R., McDermott, C.M., O’Connor, G.C., Peters, L.S., Rice, M.P. and Veryzer, R.W. (2000) Radical Innovation: How Mature Companies can Outsmart Upstarts, Harvard Business School Press, Boston. Leonard-Barton, D. (1995) Wellsprings of Knowledge: Building and Sustaining the Sources of Innovation, Harvard Business School Press, Boston, Mass. Lin, B-W. and Wu, C-H. (2010) ‘How does knowledge depth moderate the performance of internal and external knowledge sourcing strategies?’, Technovation, Vol. 30, Nos. 11–12, pp.582–589, doi:10.1016/j.technovation.2010.07.001. Link, A.N., Siegel, D.S. and Bozeman, B. (2006) ‘An empirical analysis of the propensity of academics to engage in informal university technology transfer’, Industrial and Corporate Change, Vol. 16, No. 4, pp.641–655, SSRN, available at http://ssrn.com/abstract=902207. Lubatkin, M.H., Simsek, Z., Ling, Y. and Veiga, J.F. (2006) ‘Ambidexterity and performance in small- to medium-sized firms: the pivotal role of top management team behavioral integration’, Journal of Management, Vol. 32, No. 5, pp.646–672. McDermott, C.M. and Colarelli O’Connor, G. (2002) ‘Managing radical innovation: an overview of emergent strategy issues’, Journal of Product Innovation Management, Vol. 19, No. 6, pp.424–438. Miller, B.K., Bierly, P.E., III and Daly, P.S. (2007) ‘The knowledge strategy orientation scale: individual perceptions of firm-level phenomena’, Journal of Managerial Issues, Vol. 19, No. 3, pp.414–435. Mulaik, S.A., James, L.R., Van Altine, J., Bennett, N., Lind, S. and Stilwell, C.D. (1989) ‘Evaluation of goodness-of-fitness indices for structural equation models’, Psychological Bulletin, Vol. 105, No. 1, pp.430–445. Nunnally, J.C. (1978) Psychometric Theory, McGraw-Hill, New York. O’Connor, G.C. (1998) ‘Market learning and radical innovation: a cross case comparison of eight radical innovation projects’, Journal of Product Innovation Management, Vol. 15, No. 2, pp.151–166. O’Connor, G.C. (2008) ‘Major innovation as a dynamic capability: a systems approach’, Journal of Product Innovation Management, Vol. 25, No. 4, pp.313–330. O’Connor, G.C. and McDermott, C.M. (2004) ‘The human side of radical innovation’, Journal of Engineering and Technology Management, Vol. 21, Nos. 1–2, pp.11–30. O’Connor, G.C. and Veryzer, R.W. (2001) ‘The nature of market visioning for technology-based radical innovation’, Journal of Product Innovation Management, Vol. 18, No. 4, pp.231–246. O’Connor, G.C., Paulson, A.S. and DeMartino, R. (2008) ‘Organizational approaches to building a radical innovation dynamic capability’, International Journal of Technology Management, Vol. 44, Nos. 1–2, pp.179–204. O’Leary-Kelly, S.W. and Vokurka, R.J. (1998) ‘The empirical assessment of construct validity’, Journal of Operations Management, Vol. 16, No. 4, pp.387–405. Pavitt, K. (1984) ‘Sectorial patterns of technical change: towards a taxonomy and a theory’, Research Policy, Vol. 13, No. 6, pp.343–373. Reid, S.E. and De Brentani, U. (2004) ‘The fuzzy front end of new product development for discontinuous innovations: a theoretical model’, Journal of Product Innovation Management, Vol. 21, No. 3, pp.170–184.
Exploring the exploratory search for innovation
43
Schumacher, R.E. and Lomax, R.G. (1996) A Beginner’s Guide to Structural Equation Modeling, Lawrence Erlbaum Associates, Publishers, New Jersey. Shah, R. and Goldstein, S.M. (2006) ‘Use of structural equation modeling in operations management research: looking back and forward’, Journal of Operations Mangement, Vol. 24, No. 1, pp.148–169. Skinner, W. (1992) ‘The shareholder’s delight: companies that achieve competitive advantage from process innovation’, International Journal of Technology Management, Vol. 7, Nos. 1–2, pp.41–48. Sofka, W. and Grimpe, C. (2010) ‘Specialized search and innovation performance. Evidence across Europe’, R&D Management, Vol. 40, No. 3, pp.310–323. Thomke, S.H. (2001) Enlightened Experimentation: The New Imperative for Innovation, Harvard Business School Press, Boston, Mass. Thongpapanl, N., O’Connor, G.C. and Sarin, S. (2008) ‘Understanding the role of early application exploration processes in the realization of major innovations’, PDMA Conference Proceedings, September, Orlando, Florida. Tidd, J., Bessant, J. and Pavitt, K. (2005) Managing Innovation – Integrating Technological Market and Organizational Change, 3rd ed., John Wiley and Sons, Chichester. Timmons, J. (1999) New Venture Creation: Entrepreneurship for the 21st Century, 5th ed., Irwin/McGraw-Hill, Boston, MA. Van Dijk, C. and Van Den Ende, J. (2002) ‘Suggestion systems: transferring employee creativity into practicable ideas’, R&D Management, Vol. 32, No. 5, pp.387–395. Von Hippel, E. (1988) Sources of Innovation, Oxford University Press, New York. Von Hippel, E. and Katz, R. (2002) ‘Shifting innovation to users via toolkits production and operation management’, Management Science, Vol. 48, No. 7, pp.821–833. Von Stamm, B. (2008) Managing Innovation, Design and Creativity, Wiley, London Business School, UK. West, J. and Gallagher, S. (2006) ‘Challenges of open innovation: the paradox of firm investment in open-source software’, R&D Management, Vol. 36, No. 3, pp.319–331. Zahra, S.A. and Das, S.R. (1993) ‘Innovation strategy and financial performance in manufacturing companies: an empirical analysis’, Journal of Production and Operation Management, Vol. 2, No. 1, pp.15–37. Zahra, S.A., Duane Ireland, R. and Hitt, M.A. (2000) ‘International expansion by new venture firms: internal diversity, mode of entry, technological learning, and performance’, Academy of Management Journal, Vol. 43, No. 5, pp.925–950. Zairi, M. (1992) Competitive Benchmarking: An Executive Guide, Technical Communications, (Publishing) Ltd., Letchworth, UK. Zhang, Q., Vonderembse, M.A. and Lim, J. (2002) ‘Manufacturing flexibility: defining and analyzing relations among competence, capability, and customer satisfaction’, Journal of Operations Management, Vol. 40, No. 3, pp.1–19.
[2] Exploring multiple futures Use futures techniques to explore alternative possible futures; and develop innovation options from that
[6] Probe and learn Early prototyping to explore quickly
[5] Deep diving In consumer research, study what people actually do, rather than what they say they do
[4] Working with active users Team up with knowledgeable product and service users to see the ways in which they change and develop existing offerings
[1] Sending out scouts Dispatch idea hunters to track down new innovation triggers
Learning about market
Search dimensions
[12] Idea generators Operate some form of creative idea generation to support DI
[2] Exploring multiple futures Stimulate innovation by periodically commissioning teams to generate ideas around major platforms or themes
[11] Deliberate diversity Create diverse teams to help challenge assumptions
[7] Mobilise the main stream Bring mainstream actors into the product and service development process
Network management [10] Use brokers and bridges Build cross-sector, cross-disciplinary and geographically dispersed linkages
[9] Corporate entre/intra Stimulate and nurture the entrepreneurial talent inside the organisation
Idea generation management
Openness to external sources
[7] Mobilise the main stream System or procedures in place that ensure insights from outside facing people are being collected systematically
[3] Using the web Harness the power of the web, through online communities, and virtual worlds, for example, to detect new trends
Table A
12 search strategies
44 D. Aloini and A. Martini
Appendix
Search construct and case study result comparison
Exploring the exploratory search for innovation Table B
45
CFA on the search construct
Indicator (Cronbach’s alpha; eigen value)
Principal component factor loading
Measurement model Standard coefficient
Unstandard coefficient
R2
t-value2
A – learning about markets (α = 0.77; eigen value = 2.37) MKT_6
0.63
0.59
1.00
MKT_7
0.76
0.69
1.25
0.47
***
MKT_9
0.71
0.72
1.31
0.51
***
MKT_10
0.65
0.73
1.21
0.53
***
0.53
***
B – managing idea generation (α = 0.69; eigen value = 1.52) MNG_2
0.75
0.72
1.00
MNG_3
0.66
0.73
0.97
C – network management system (α= 0.84; eigen value = 2.26) NET_2
0.76
0.78
1.00
NET_3
0.74
0.82
1.05
0.67
***
NET_4
0.86
0.78
1.16
0.61
***
0.38
***
D – openess to external sources (α = 0.56; eigen value = 1.38) OPEN_4
0.72
0.63
1.00
OPEN_6
0.84
0.61
0.86
Notes: *Significant at the 0.05 level, **significant at the 0.01 and ***significant at the 0.001 level (for 1 d.f.)
46 Table C
D. Aloini and A. Martini CFA on EA, and on innovation performance
Indicator (Cronbach’s alpha; eigen value)
Principal component factor loading
Measurement model Standard coefficient
Unstandard coefficient
R2
Significance
Exploratory innovation (α= 0.787; eigen value = 2.45) INN_EXR_1
0.80
0.73
1.00
INN_EXR_2
0.78
0.69
0.95
0.48
***
INN_EXR_3
0.77
0.68
1.05
0.48
***
INN_EXR_4
0.77
0.67
1.10
0.45
***
Competence exploration (α = 0.800 eigen value = 2.24) KW_EXR_2
0.81
0.70
1.00
KW_EXR_3
0.85
0.75
1.08
0.56
***
KW_EXR_4
0.87
0.83
1.21
0.69
***
Innovation performance (α = 0.52; eigen value = 1.37) INN_PERF_1
0.83
0.70
1.00
INN_PERF_2
0.52
0.30
11.76
2.31
**
INN_PERF_3
0.63
0.58
1.12
0.54
***
Notes: *Significant at the 0.05 level, **significant at the 0.01 and ***significant at the 0.001 level (for 1 d.f.)
1.03
