Master Thesis
Effects of Routine Formalization on Innovation Performance &
the Mediating Role of Knowledge Sharing
Master Thesis:
Author : M.J.L. (Michel) de Bruijn Student number : 10884181
Date : 28-01-2017
Version : Final v1.0
Faculty : Economics & Business - International Strategy & Marketing Programme : Master of Science - Executive Programme in Management Studies
Track : Strategy
Institution : University of Amsterdam, Amsterdam Business School Supervisor : Dr. Sebastian Kortmann
EPMS Strategy 2015 -2017
University of Amsterdam, Amsterdam Business School. 2
Statement of Originality
This document is written by Michel J.L. de Bruijn, who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
M.J.L. (Michel) de Bruijn 10884181
University of Amsterdam, Amsterdam Business School. 3
Abstract
While the effects of formalization on organization performance have been examined
extensively, the effects of routine formalization on innovation performance nor the effects of knowledge sharing as a mediator have not. Based on the knowledge-based theory of the firm and formalization as a variable of organizational structure, this study aims to explain the quantitative relations between routine formalization and innovation performance within organizations and how this is mediated by knowledge sharing. The research model, a literature review and six hypotheses are formed positing that formalized routines and informalized non-routines not only influence innovation speed and quality directly, but also through explicit and tacit knowledge sharing. The data was collected from 94 Dutch
organizations varying in age, size and economic sectors. The research model was empirically tested using multiple hierarchical regressions. Results show that formalized routines and informalized non-routines positively influence innovation quality directly. Moreover, results show that both explicit and tacit knowledge sharing have significant mediation effects, but only with regard to the relations between formalized routines and innovation speed and innovation quality, respectively.
University of Amsterdam, Amsterdam Business School. 4
Inhoudsopgave
Abstract ...3
1. Introduction ...6
2. Theoretical foundation and hypotheses ... 10
2.1. Routine formalization ... 10
2.1.1. Formalized routines ... 11
2.1.2. Informalized non-routines ... 11
2.2. Knowledge sharing ... 12
2.2.1. Explicit and tacit knowledge sharing ... 13
2.3. Innovation performance ... 15
2.3.1. Innovation speed and quality... 15
2.4. Research model and hypotheses... 16
2.5. Formalization and innovation performance ... 17
2.6. Formalization and knowledge sharing ... 21
2.7. A Knowledge sharing and innovation performance ... 23
3. Research Methodology ... 26
3.1. Research and survey design ... 26
3.2. Data collection and sample ... 27
3.3. Descriptive statistics ... 28
3.4. Data analysis method... 29
3.5. Measurement instruments ... 30
3.5.1. Constructs ... 30
3.5.2. Control variables ... 32
3.6. Construct validity and reliability ... 33
3.7. Common method variance ... 34
3.8. Non-response bias ... 35
4. Results ... 36
4.1. Correlations... 36
4.2. Hypotheses testing for direct effects ... 37
4.3. Indirect effects and mediation results ... 44
3.4.1. Informalized non-routines and mediation effects on innovation performance of organizations ... 45
3.4.2. Formalized routines and mediation effects on innovation performance of organizations ... 46
5. Discussion and limitations. ... 49
5.1. Conclusions... 49
5.2. Discussion ... 50
5.2.1. Formalization and innovation performance ... 50
University of Amsterdam, Amsterdam Business School. 5
5.2.3. The mediating role of explicit and tacit knowledge sharing ... 52
5.3. Theoretical implications ... 53
5.4. Practical Implications ... 55
5.5. Limitations and future research ... 56
REFERENCES ... 58
APPENDIX A Online survey introduction ... 66
University of Amsterdam, Amsterdam Business School. 6
1. Introduction
Various theories and many researchers have accepted formalization as part of organizations and it is often explained by, and related to, two types of organizational processes. First, early researchers explained formalization as a stable and inflexible mechanism to accomplish organizational goals and as the primary means why organizations work like they do (March & Simon; 1958; Cyert & March, 1963; Thompson, 1967; Nelson & Winter, 1982). By that means they represent formalized processes or routines within an organizational context. Further, its stability factor is a defined charateristic of bureaucracy (Stinchcomb, 1959). Implementation of formalized routines in organizations, also referred to as institutionalization, has on one hand proven to be an important building block for the structuring of work and accountability (Weber, 1947; Crozier, 194; Kaufman, 1977). On the other hand, they are also recognized as a source for stagnation, inertia (Hannan & Freeman, 1983), inflexibility or mindlessness (Gersick & Hackman, 1990; Asforth & Fried, 1988). From this point of view, formalization empowers the organization of expertise and makes exercising power efficient, while simultaneously stimulating bureaucracy. Second, more recent research has argued that formalization based on tacit knowledge, utilized in informalized processes or non-routines, enhances new knowledge creation and acts as a mechanism for generating innovativeness and dynamic capabilities (Teece and Pisano, 1994; Tranfield & Smith, 1998). Informalized non-routines can therefore be considered as an important source of flexibility, adaptibility, innovation and organizational change (Feldman, 2000; Miner, 1990; Pentland & Rueter, 1994).
Feldman & Pentland (2003) emphasized that routines matters to deal with
environmental dynamics because they are both ongoing processes and subject to modification by people who work with them. Routines even change in stable environments of old
University of Amsterdam, Amsterdam Business School. 7 mechanisms as meta-routines (Hackman &Wageman, 1995). Hence, the routines underlying formalization are a two-sided phenomenon (Feldman & Pentland, 2003): one part is the objective structure for guiding work processes and accountability, the other is the
performative part relating to agency for creating, maintening and modifying organizational processes. Both enhance operational and financial performance as well as innovation performance.
Recent decades the knowledge-based theory and researchers have recognized and investigated the importance of knowledge resources for organizations, as drivers for innovation performance, strategic flexibility and enhancement of adaptive capabilities. Nowadays, knowledge resources are thus considered to be crucial for any organization, especially those in dynamic environments (Grant, 1996; Subramaniam & Youndt, 2005; Teece, Pisano, & Shuen, 1997). Drucker (1993) emphasized traditional resources would be replaced by knowledge as most important driver for competitive advantage. The knowledge-based view of the firm (Grant, 1997) increased interest of researchers and practioners to investigate the processes of creating, identifiying, capturing, sharing, accumulating and applying explicit and tacit knowledge (Alavi and Leidner 2001; Jang, Hong, Bock, & Kim, 2002; Kogut & Zander, 1996; Michailova & Husted, 2003; Nonaka 1994; Nonaka & Takeuchi, 1995). These processes are considered important building blocks within organizations (Cummings 2004; Rico et al., 2008) to achieve sustainable competitive advantages over other organizations based on knowledge management.
To benefit and capture knowledge within the organizational context many
organizations spend much financial resources to support distribution and knowledge sharing processes among employees by implementation of information technology, quality systems and knowledge management (Bock et al. 2005; Wasko & Faraj 2005). This shows explicit and tacit knowledge sharing are more and more embraced by organizations, especially since
University of Amsterdam, Amsterdam Business School. 8 researchers argue that knowledge sharing behavior among organizational members
contributes to the generation of vital effectivity and innovation capabilities. Since knowledge resources are needed for organizations to survive changing environments and functions as an amplifier for organizational performance (Kogut & Zander, 1996), the effectiveness and innovativeness of organizations particularly depend on how well knowledge is shared on individual, group or organization level (Alavi & Leidner, 2001; Argote & Ingram, 2000; Huseman & Goodman, 1998; Pentland, 1995).
The effects of formalization on organization performance as well as knowledge sharing (mostly from behavioral perspectives) on innovation performance have been studied separately in many ways. Formalization showed to improve organization performance (Adler & Borys, 1996; Schwenk & Schrader, 1993) and knowledge sharing is positively related to production costs reduction, faster completion of new developments, innovation capabilities and organization performance (Arthur & Huntley, 2005; Collins & Smith, 2006; Hansen et al., 2005; C.P. Lin, 2007; H.F. Lin, 2007). Yet, there is very few literature about the specific links and quantifiable outcomes concerning the relation of formalization and innovation
performance on organizational level and none for knowledge sharing as a mediator in between (Lee & Choi, 2003 ). Recently, there is growing conviction among researchers that
application of knowledge sharing does not directly impact organizational performance. Rather, researchers argue organization performance is indirectly impacted through
intermediate outcomes caused by knowledge sharing such as innovation performance (Lee & Choi, 2003; Davenport & Prusak, 1998; Hsu & Lin, 2008; Law & Ngai, 2008; Liebowitz & Chen, 2001; Wang & Wang, 2012). Although recent decade research is continuously being conducted, there is still plenty of empirical evidence for effects on innovation performance to be discovered. At the same time, there is consensus that innovation speed and quality are revealed as its most important dimensions to achieve it (Kessler & Chakrabarti, 1996).
University of Amsterdam, Amsterdam Business School. 9 Therefore, building on this discussion, the main research question of this study is:
What are the effects of routine formalization on innovation performance within organizations and how is this relation influenced by knowledge sharing in organizations?
This study aims to explain the direct relationships at organizational level of formalized routines and informalized non-routines on innovation speed and quality respectively, as well as the indirect effect of explicit and tacit knowledge sharing on these relations. This research used multiple hierarchical regressions to investigate the hypotheses, based on a survey of 94 organizations in the Netherlands.
University of Amsterdam, Amsterdam Business School. 10
2. Theoretical foundation and hypotheses
This chapter presents a theoretical foundation regarding the relations between routine formalization and innovation performance. After explaning routine formalization, knowledge sharing, innovation performance and their dimensions, the proposed direct relationships between these variables are discussed and hypotheses are formulated. The proposed relations are shown in the research model (figure 1).
2.1. Routine formalization
Formalization and its underlying processes are a central feature for any organization’s
performance. They inform about how work is done and appear continuously in descriptions of organizational actions (Marsh & Simon, 1959; Nelson & Winter, 1982; Levitt & March, 1988). In the extant literature formalization consist out of formalized and non-formalized processes, also referred to as formalized routines and informalized non-routines, to describe work activities (Prusak, 2001). These forms of organizational routines caught plenty attention in academic debates, although strictly defining them in an explicit definition has been proven difficult because of the ambiquous character. Formalized routines appear more visible and concret compared to informalized non-routines and are therefore easier to define. The latter are more abstract and less visible, but both are considered important processes in
organizations.
The majority of early research on one hand, defined organizational routines as stable and assigned them as essential parts of organized work to achieve higher efficiency and effictivity through standardization of processes. Yet, these processes may also retain history and cause inertia (Gersick & Hackman, 1990; Hannan & Freeman, 1983). More recent research argued on the other hand, that routines are stable, but flexible and changeable at the same time and thus important essentials for organizational flexibility, adaptivity and
University of Amsterdam, Amsterdam Business School. 11 continuous change (Feldmand 2000; Feldman & Pentland, 2003, Becker et al., 2005) leading to sustainable competitive advantage.
2.1.1. Formalized routines
Formalized routines are present in organizations as highly structured, documented, codified
information or knowledge. They are locked in policies, organization charts, procedures, rules, plans, guidelines, standard operational procedures (SOP’s) etc, and are often considered contributers for formalized structures and guides for work processes within the organizational context (Fredrickson, 1986). Early research shows investigators approached formalized routines as bureacratic and inflexible, because they inhibit flexibility and leave little room for
dividing decision-making or flexibility in tasks execution (Marsh & Simon, 1959; Nelson &
Winter, 1982; Levitt & March, 1988). Yet, they enable accountability, facilitate cognitive efficiency and are an expression of selective and subjective knowledge retention (e.g. organizational memory) of historical organizational knowledge (Huber, 1991). Hence, formalized routines are structures to help organize expertise and efficient power exercise within organizatons (e.g. top-down processes) and positively enhance performance (March, 1991; Schwenk & Shrader, 1993). From this point of view formalized routines are functional and therefore reduce costs and increase managerial control.
2.1.2. Informalized non-routines
Informalized non-routines are not as structured and visible as formalized routines, but are
considered to be organizational processes as well (Lillrank & Liukko, 2004). Because
informalized non-routines are vague and cannot easily be classified into categories compared to formalized ones, it is difficult to link them directly to structured elements of organizations
like, guidelines, procedures, SOP’s and actions.Seen from this perspective, informalized
non-routines are more suited to address non-predicted, surprising or unfamiliar events through inquiry and learning systems, including the capacity for problem-solving (Lillrank, 2003).
University of Amsterdam, Amsterdam Business School. 12 This in turn enhances the achievement of strategic flexibility and adaptability of
organizations. Informalized non-routines occur when people enact and participate in social processes with each other, between units (Gold et al., 2001) or even across organizations. During informalized non-routines (e.g. bottom-up processes) (specialized) individuals freely share and integrate their knowledge which drives product or service development and thus creativity and innovation.
2.2. Knowledge sharing
Knowledge sharing in a widely researched part of knowledge management and, like in this study, mostly defined as exchanging explicit and tacit knowledge in organizations (Kogut & Zander, 1992; Nonaka & von Krogh, 2009). Knowledge sharing processes utilize knowledge and create possibilities for maximizing organization’s ability to meet their needs by the generation of efficiencies and solutions (Reid, 2003). To meet those needs knowledge sharing has shown to be essential for organizations, since it provides them with opportunities to improve innovation performance and reduce redundant learning efforts (Calantone et al., 2002; Nonaka & Takeushi, 1995; Scarbrough, 2003). Knowledge sharing will outline as a social interaction culture, involving the exchange of knowledge, experiences, and skills throughout the organization. It includes a group of shared understandings and is associated with providing employees acces to relevant knowledge and facilitates construction and utilization of networks in organizations (Hogel et al., 2003). Further, to fully benefit and utilize knowledge in organizations, it needs to be shared on individual and organizational level (Sabherwal, 2006). In Nonaka & Takeuchi’s (1995) SECI-model four processes of knowledge sharing are emphasized and these explain how to turn organizational knowledge into individual and group knowledge and vice versa, while tacit and explicit knowledge sharing processes are highlighted as important resources to achieve this.
University of Amsterdam, Amsterdam Business School. 13 Sharing of knowledge on individual level gets things done better, faster and more efficiently by interaction among employees. This involves the provision or receipt of (technical) information, know-how and skills (Haas & Hansen, 2007), as well as interaction and communication among team members. Two requirements on individual level are needed to achieve knowledge sharing. First, the transfer of acquired individual (specialized)
knowledge from one individual to another within the organization is needed (Cohen & Levinthal, 1990; De Long and Fahey, 2000; Alavi & Leidner, 2001; Cabrera & Cabrera, 2002; Ipe, 2003). Second, individuals must actively participate in collecting and donating knowledge to fully exploit existing knowledge and to explore new knowledge creation opportunities (Argote & Ingram, 2000).
Knowledge sharing on organizational level entails capturing, organizing, reusing and transferring of previoulsy gained experience-based knowledge and subsequently make it assessable to fellow employees within the organization. Organizations of any size therefore promote knowledge sharing more and more by the use of relevant tools, such as knowledge management, processes and systems. These tools intentionally stimulate distribution of knowledge throughout the organization with the purpose to transform it into collective knowledge (Alavi & Leidner, 1999, 2001; Cabrera & Cabrera, 2002; Kuo & Young, 2008).
Currently, organizational promotion of knowledge sharing is altering traditional concepts of managing intellectual resources by provision of new processes, disciplines and cultures, and hence shaping organizational innovation (Darroch and McNaughton, 2002). The outcomes of these processes, disciplines and cultures unveil the effectively achieved degree of knowledge sharing on firm innovation capability.
2.2.1. Explicit and tacit knowledge sharing
Extant literature expresses knowledge sharing in several ways, but most used are two dimensions: explicit and tacit knowledge sharing. These dimensions are distinct (Alavi &
University of Amsterdam, Amsterdam Business School. 14 Leidner, 2001) but can sometimes be misleading, when treated as polarized elements.
Therefore I rely on Polanyi’s (1966) and Tsoukas’s (2005) explanation of knowledge, who
state tacit and explicit knowledge are not polarized ends but two sides of the same coin.It
means new knowledge is not only created by transforming tacit knowledge into explicit knowledge, but is also created by reflecting on our practice and using knowledge in a new or improved way under different circumstances. Phrased like this, explicit and tacit knowledge sharing can be a source for knowledge creation. A further explanation of the characteristics of both knowledge dimensions is needed though for understanding the differences between explicit and tacit knowledge sharing.
Explicit knowledge is often defined as ‘know-how’ and involves practically all the institutionalized knowledge within organizations. It is comprised from individual prior experience, skills and knowledge and is easiers to capture, codify and faster to transmit compared to tacit knowledge (Nonanka, 1994). Explicit knowledge is embedded in
management mechanisms, information (technology) systems, handbooks, routines, SOP’s, guidelines, etc. Explicit knowledge sharing is therefore the transfer of knowledge among organizational members via the aforementioned items. The ‘explicitness’ makes it easier to collect and donate because of its simplified and concrete character, hence facilitating distribution easiness. (Coakes, 2006; Huang et al, 2010; Zander & Kogut, 2005).
Tacit knowledge is referred to ‘know-what’ and is knowledge that resides within individuals. It is developed over time by learning from experiences and combining knowledge. It is inferred from individual action, although people do not always explicitly recognize it in themselves. For instance, people in organizations can solve cross-unit complex product errors, but still remain tacitly unaware of how they interact as resolvers (Kane et al., 2005). Tacit knowledge is based on individual experiences and exists as a semi-unaware composition in our brains instead of codified structure. Hence, it is accessable, but harder to
University of Amsterdam, Amsterdam Business School. 15 express compared to explicit knowledge. Tacit knowledge sharing occurs during face-to-face social interaction in which people’s capacity and willingness to share what they know and to use what they learned are key principles (Holste & Fields, 2010; C.P. Lin, 2007; H.F. Lin, 2007; Endres et al., 2007). When people comply to these principals they are able to transfer tacit knowledge across different contexts. Explicit knowledge does this also, but is easily communicated, codified and ís in our awareness such as factual process steps or executing physical tasks. Either way, people are knowledge carriers and can restructure and share both knowledge forms in new contexts (Kane et al., 2005).
Thus, knowledge sharing can be abstract, explicitly represented, codified, and accessed by individuals on different levels throughout the organization, while both tacit and explicit knowledge occur throughout organizations and are transferable between individuals and across units. When explicit and tacit knowledge sharing are properly used as parts of knowledge management, they become important elements to stimulate innovation
performance.
2.3. Innovation performance
Innovation is a neccesity for organizations in dynamic environments, because increasing competition, changing customer demands and shorter product or service life-cycles put emphasis on faster innovation and product differentation (Heirman & Clarysse, 2007). Innovation performance is therefore often expressed in the dimensions innovation speed and quality because both are supposed to enhance the speed-to-market and increase differentiated product or service offering (Kessler & Chakrabarti, 1996).
2.3.1. Innovation speed and quality
Innovation speed reflects the acceleration degree of an organization and the time elapsed
between innovative initiatives and new product or service introductions and support the first-mover advantage (Allocca & Kessler, 2006; Kessler & Bierly, 2002). Innovation quality is
University of Amsterdam, Amsterdam Business School. 16 recognized as a comparison of the potential of innovation(s) and the realized result of it in new or renewed products or services. Although, it is not yet a fully developed concept because of its complexity and the difficulty to define catalists over the different domains of procedure, process and organization (Haner, 2002; Lanjouw & Schankerman, 2004). Despite the latter, it is suggested as denominator for customer value in which knowledge sharing is the way forward as part of the organizational learning platform (Ng, 2009; Tseng & Wu, 2006).
2.4. Research model and hypotheses
This study focusses on the direct relations of formalized routines and informalized non-routines on innovation speed and quality respectively, and the mediating role of explicit and tacit knowledge sharing. Figure 1 shows the research model for this study. The sections in this chapter discuss the hypothesized relations between variables. The hypotheses are shown in the research model in figure 1.
Innovation quality Innovation speed Explicit knowledge sharing Formalized routines Tacit knowledge sharing Informalized non-routines H2b H5a H1a H4b H3a H6b H3b H4a H1b H1b H2a H5b H6a
Control variables: Firm age, firm size, economic sector, educational level
Routine formalization Knowledge sharing Innovation performance
University of Amsterdam, Amsterdam Business School. 17
2.5. Formalization and innovation performance
Having accepted knowledge as most important (innovation) resource, every organization needs to manage its knowledge to fully benefit the potential independent from economic sector (Baskerville & Dulipovici, 2006). Knowledge management is thus crucial and
comprises organizational learning processes in order to gain competencies driving innovation performance. It is therefore important for organizations to keep the knowledge flow going by designing structures, systems and processes that create, integrate and manage knowledge resources effectively (Nonaka, 1994; Nonaka & Takeuchi, 1995) to stimulate innovation performance and enhance innovation speed and quality. Yet, where organizational structures and methods are well researched, very few researchers examined the relation between routine formalization and innovation performance (Pertusa-Ortega, 2010).
In widely studied hybrid organizational designs emphasis is put on strategical
management of knowledge resources by linking horizontal and vertical characteristics of their structures in order to create, recombine, accumulate and spread knowledge resources
effectively with the purpose to enhance organization performance in dynamic environments. (Nonaka, 1994; Nonaka & Takeuchi, 1995). Organizations adopting these organizational designs can outperform other organizations, because linking horizontal and vertical structures results in combining static formalized structures with dynamic aspects of flexibility (Gold et al., 2001). Further, it is known that every organizational setting contains static and dynamic elements which determine the degree of knowledge utilization and creation (Beesley, 2004). Hence, taking these resource-based organizational design structures as a reference for static formalized routines and more flexible informalized non-routines, the similar positive effects on innovation performance may occur as for hybrid organizational designs on organization performance.
University of Amsterdam, Amsterdam Business School. 18 Some researchers have argued formalized routines inhibit knowledge creation (Von Krogh,1998) due to inhibiting interaction among employees (López et al., 2006) and high degrees of formalization restricts the amount of new idea development (Lee and Choi, 2003). Subsequently, they claim enhancement of flexibility and non-formal behavior within the organizational structure stimulates knowledge creation. More recent studies argue both formalized routines and informalized non-routines show permanent similarities when they are applied and therefore must be approached as patterns of behavior, action or interaction
(Feldman & Pentland, 2003). They claim both enable knowledge utilization and creation (Kern, 2006), improve process quality and the generation of new product (Ahn et al., 2006; Söderquist, 2006). Though the dimensions of routines are distinct: formalized routines are a form of codified explicit knowledge, whereas informalized non-routines are a form of tacit knowledge. (Reynaud, 2005).
The achievement of operational efficiency is often related to procedural or explicit knowledge (‘know-how’) and is embedded in formalized processes, the so-called standard
operational procedures (SOP’s)(Lillrank & Liukko, 2004). Yet, SOP’s have the down-side of
only accepting pre-defined input and people must follow procedures for turning input into output without stimulating change of the procedure. This could inhibit gradual change, while not stimulating assessment and classification of input. Though, more recent research shows that SOP’s as formalized routines are important elements of change and they are considered to be tacit as well instead of only being explicit (Reynaud, 2005). This is based on Feldman & Pentland’s reconceptualization of organizational routines as source of flexibility and adaption to the environment. They state when SOP’s are not solely ostensive but allow agency and subjective assessment to influence them, they become formalized routines within the organization and a source of change providing flexibility and opportunities for innovation. Lillrank & Liukko argued and demonstrated this in a hospital setting. Employees, following
University of Amsterdam, Amsterdam Business School. 19 SOP’s to perform skin treatment before surgery, experienced that SOP’s cannot be fully standardized in a procedure or guideline without interpretation of the skin condition first. Assessment of skin conditions led to different treatments for patients and thus adjusted the formalized process. Thus, when assessment and classification of input for formalized
processes as well as selection from different alternatives (e.g. other procedures, guidelines or actions) are possible, they become more flexible as a formalized routine and a source for improvement, change and innovation. March (1991) confirms this by stating these adjustments can be considered as organizational learning that stimulates variability and reduces too much standardization, while avoiding failure. Hence, formalized processes are positively associated with innovation performance.
Besides formalized routines, informalized non-routines are also an important part of organizational learning (Pavitt, 2002) and forms of knowledge utilization and creation platforms enhancing innovational activity and performance. Informalized non-routines are considered dynamic processes rather than static ones (Becker et al., 2005) while their tacit knowledge base (Cepeda & Vera, 2007) generates new knowledge. These non-routines promote knowledge creation through double loop learning (Cohen & Levinthal, 1990; Zahra & George, 2002). Double loop learning occurs when multiple actors interact during
development of organizational routines resulting in a diversity of knowledge, different goals and interpreting organizational schemes differently. Actors subjectively handle knowledge differently based on their prior knowledge and experience. Their appropriate choice of action will therefore be divergent and result in a fitting modification of knowledge for their goals and thus creating and utilizing new knowledge at the same time (Nonaka & Takeuchi, 1995). These arguments support the dynamic and non-formalized side of routines and can be similar for formalized routines in organizational structures because, as reported above, both relate to
University of Amsterdam, Amsterdam Business School. 20 processes established within organizations and entail patterns of behavior, action and
interaction for change and improvement.
Further, organizations have rules and policies, which have an abstract and formal character and steer different work related actions to accomplish organizational objectives in different situations (Feldman and Pentland, 2003). Although useful in that manner, they lack practical guidance and detailed information about the execution of work processes. Similar to informalized non-routines, employees must interpret the existing rules and policies in a different manner with regard to their prior knowledge and experience to yield new knowledge. Therefore, routine formalization, expressed as formalized routines and
informalized non-routines, not only gains new knowledge, but can also be applied to improve innovation performance for several reasons. First, they improve cooperation and collaboration and thus form structure and scope for interaction (Kern, 2006) which produces new ideas for knowledge development. Second, by means of rules they support and ease sharing of explicit and tacit knowledge originated in different organizational units (Cohendet et al., 2004; Cordón-Pozo et al., 2006). Last, they decrease ambiguity (Cordón-Pozo et al., 2006) in organizations and allow individuals to handle contingencies more effectively, since best practices from experience can be codified and incorporated in the organizational memory (Adler & Borys, 1996). This allows for conservation and easy acquirement of knowledge by organization members and subsequently stimulates innovation performance. Building on the above discussion and citing innovation speed and innovation quality as most important dimensions to achieve innovation performance (Kessler & Chakrabarti, 1996), I follow Kern (2006) and Feldman & Pentland’s (2003) approach on routines and propose the following hypotheses for the effects of formalization on innovation performance:
H1a. Formalized routines are positively related to innovation speed. H1b. Informalized non-routines are positively related to innovation speed.
University of Amsterdam, Amsterdam Business School. 21
H2a. Formalized routines are positively related to innovation quality. H2b. Informalized non-routines are positively related to innovation quality.
2.6. Formalization and knowledge sharing
As discussed previously, routines contain explicit and tacit knowledge embedded in formal documents and information systems or within individuals. Further, explicit and tacit knowledge are input for adjusting routines and are output from knowledge management. (Postrel, 2002). When knowledge is applied through formalization processes, it improves process quality and the generation of new product development in order to fulfill customer needs (Ahn et al., 2006; Söderquist, 2006). It is therefore obvious organizations must focus on strengthening this relationship, which can be achieved by sharing of explicit and tacit
knowledge among employees.
As touched upon in the elaborated discussion of section 2.5, formalized routines and informalized non-routines appear to positively influence explicit and tacit knowledge sharing in organizations (Cohendet et al., 2004; Cordón-Pozo et al., 2006) as well as cooperation and collaboration among employees (Kern, 2006). However, the effectiveness of knowledge sharing may vary between organizations. This can be caused by motivational dispositions of organizational members, the value of knowledge or the existence, quality and cost of sharing mechanisms (Gupta & Govindarajan, 2000b). Further, it is known organization members share knowledge and ideas they have already in common more easily than unshared ideas that
are unique to individuals (Lu et al., 2012; Wittenbaum et al., 2005).For example, in job
interview evaluations organizational members look for information and knowledge they all share, which lowers the possibility for new information and knowledge infusion. Also, the probability of mentioning information or knowledge increases when organizational members already possess it (Stasser & Titus, 1987). Explicit knowledge in formalized routines is therefore expected to be shared more than tacit knowledge in informalized non-routines. This
University of Amsterdam, Amsterdam Business School. 22 unfortunately also suggests organizational members do not always recognize the value of tacit knowledge and different information when individuals share or donate it.
Research results of knowledge sharing points out observability and complexity effect the success of knowledge sharing and the degree of ‘sharing easiness’ (Galbraith, 1990; Meyer & Goes, 1988). The shareability of explicit knowledge from formalized processes is easy among organizational members, because it is well-understood and easy to apply through codification in documents and information systems (Coakes, 2006; Huang et al, 2010; Zander & Kogut, 1995), while tacit knowledge in informalized non-routines proves to be complex and more difficult to share. The reasons lie in its complexity and higher ‘causal ambiguity’ (Szulanski, 1996), compared to well-understood knowledge. Though, during research in a hospital environment Meyer & Goes proved transfer of tacit knowledge through informalized non-routines has positive effects on adopting innovations because of its complexity and necessity for accurate treatments. It leads to higher innovational performance and was more likely to lead to knowledge creation, combination and utilization. They further found that the more observable innovations are, the easier knowledge is shared. Difficult observable
innovations are therefore harder to use for organizational purposes than easy observable ones. Research of Laughlin & Ellis (1986) found that knowledge demonstrability influences
knowledge sharing in such a way, that when demonstrability is high it is more easy to explain and transfer to fellow employees. When low, it makes it harder to explain and harder to persuade others of its suitability. Kane (2010) found interaction between knowledge demonstrability and the degree to which groups of employees in organizations share an overarching social identity. When knowledge demonstrability is high, knowledge is transferred between groups even when no identity is shared. But, when low, the degree of knowledge sharing was higher when employees do share overarching identity compared to non-shared identity. Thus, although the degree of formalized routines is expected to be higher
University of Amsterdam, Amsterdam Business School. 23 than from informalized non-routines, both can be associated in a positive manner with explicit as well as tacit knowledge sharing. Building on this discussion and examples, the following hypotheses are formulated for the effects of formalization on knowledge sharing:
H3a. Formalized routines are positively related with explicit knowledge sharing. H3b. Informalized non-routines are positively related with explicit knowledge sharing. H4a. Formalized routines are positively related with tacit knowledge sharing.
H4b. Informalized non-routines are positively related with tacit knowledge sharing.
2.7. A Knowledge sharing and innovation performance
Last decade research on the effects of sharing and (re)combining knowledge has shown it is a driver for value creation in organizations, since it leads to better organization performance, as well as enhancement of innovation (e.g., Arthur & Huntley, 2005; Collins & Smith, 2006; Cummings, 2004; Hansen, 2002; Lin, 2007d; Mesmer-Magnus & DeChurch, 2009;
Srivastava, 2006). This underlines the importance of knowledge sharing as part of knowledge management for innovation performance. Based on theories of knowledge a new typology of two types of innovation (e.g. product and process innovation) was proposed by Gopalakrishan & Bierly (2001), who explicitly used the dimension ‘explicit-tacit’ knowledge as predictor for innovation. Their quantitative results from a cross-sectional study among small, medium and large banks found explicit innovations to be more effective than tacit innovations. This was followed by Abou-Zeid & Cheng (2004) arguing that sharing explicit and tacit knowledge resources must be associated with the creation and utilization of knowledge, although this relation is still hardly empirically investigated. They conclude by pointing out that both explicit and tacit components stimulate knowledge management processes, which are important for the success of innovation within organizations.
To innovate organizations dependent on employees who share their knowledge, skill
University of Amsterdam, Amsterdam Business School. 24 input for innovation, since they are firm-specific and socially complex making it hard to imitate or replicate by other organizations (Chiang & Hung, 2010; Dimitris, Konstantinos, Klas Eric & Gregory, 2007; Gachter, von Krogh & Heafliger, 2010). Moreover, sharing of knowledge resources in organizations is increasingly essential because it reduces redundant learning efforts and, more importantly, it increases their innovation speed and quality. Success of organizations regarding innovation speed and quality, is the result of the degree in which organization succeed developing in new problem-solving methods, process quality
enhancement, new product development and the speed of introduction (Calantone et al., 2002; Marina du, 2007; Scarbrough, 2003; Tidd et al., 2005). In an empirical study among 50 organizations Lin (2007d) found evidence for improvement of innovation capabilities when knowledge sharing is supported by individuals and the upper echelon. More concrete, they emphasize organizational members enjoy knowledge sharing by donating and collecting knowledge for self-efficacy and to help others solve problems. Lin also found explicit and tacit knowledge sharing enhances innovation performance, which makes them important enablers for organizations to adjust and create knowledge with the intention to improves their innovative speed and quality (Heirman & Clarysse, 2007). Recently, Wang & Wang (2012) empirically investigated explicit and tacit knowledge sharing effects on organization’s operational and financial performance and how innovation speed and quality mediate this relation. They argued organizational performance is not directly, but indirectly influenced by knowledge sharing and innovation performance is an intermediate outcome. Their results concluded both explicit and tacit knowledge sharing directly affect innovation speed and quality in a positive manner. The effects on innovation speed and quality showed to be larger for explicit knowledge sharing than from tacit knowledge sharing.
Hence, organizations that facilitate explicit and tacit knowledge sharing between employees within an organization, will likely cultivate new ideas for improving process
University of Amsterdam, Amsterdam Business School. 25 quality, new product development and faster product introduction, thus enhance innovation performance. Based on this discussion, I propose the following hypotheses:
H5a. Explicit knowledge sharing is positively related to innovation speed. H5b. Tacit knowledge sharing is positively related with innovation speed. H6a. Explicit knowledge sharing is positively related with innovation quality. H6b. Tacit knowledge sharing is positively related with innovation quality.
University of Amsterdam, Amsterdam Business School. 26
3. Research Methodology
The empirical part of the study is represented in this chapter. First, the research and survey design are explained as well as the data collection and descriptive statistics. The middle part reports the data analysis method and the measurement of constructs. Last, the validity and reliability are discussed.
3.1. Research and survey design
A deductive research approach was used to study the relationships between routine
formalization, knowledge sharing and innovational performance at organization level. To test the hypotheses a cross-sectional study and quantitative analyses were used. A questionnaire was conducted and pretested before the formal data collection. I built and applied a structured online-questionnaire with closed questions (Appendix B) for data collection using Qualtrics online survey platform. The electronic questionnaire is preferred above a paper response method, because of fewer missing responses and more coding and distribution flexibility (Boyer et al., 2001). Besides electronic answering, a Likert-type scale was used for the convenience for respondents since it saves time and increases the response rate (Jamieson, 2004). I invited most of the respondents personally to participate before the questionnaire was distributed by email to stimulate a higher response rate (Yu & Cooper, 1983). Only one potential respondent refused in person and was deleted from the distribution list. There were no refusals received by email. The distributed email contained a hyperlink to the
questionnaire and a commitment to participate as respondent in research of others whenever suitable. Before respondents started filling out the questionnaire, they were informed about the purpose of the survey and an illustration was given about how to fill out the questionnaire (Appendix A). Due to language diversity the email and questionnaire were parallel translated from English to Dutch and vice versa to prevent lexical and idiomatic errors, as well as grammatical and syntax errors (Hofstede, 1980; Adler, 1983). The translation was also for
University of Amsterdam, Amsterdam Business School. 27 convenience of respondents since the majority is Dutch and a minority foreign. Furthermore, the questionnaire contained general instructions, sub instructions per section, an example question and assurance of anonymous recording and processing of responses (Appendix A). Participation was voluntary and anonymity was guaranteed by using an untracked open link to the survey, which was protected against indexation by online search engines and secured uploading of files. Also, to secure validity and reliability, respondents were unable to
complete the questionnaire more than once. After the initial email two follow-up emails were sent in the second and third week to stimulate participation (Yu & Cooper, 1983). Although the availability sample may lead to lower generalizability, the composition of the respondents group is diverse because respondents work at heterogeneous organizations in different
economic sectors. To account for differences among firms, three control variables were included in the research model: firm size and age as well as economical sector. To be able to act as key informant for their organization and to account for appropriate professional and intellectual ability of respondents, their educational level as fourth control variable was added since this often reflects one’s position in an organization.
3.2. Data collection and sample
A non-probability (convenient) sample was used in which fellow part-time working and studying master students acted as key informants of their organizations. Using key
organizational informants has proven to be effective in many research areas, especially when key informants have access to, and use of, organization’s knowledge. For fellow students to be admitted to the master programme they must at least have a higher vocational education. Therefore, their educational and current work level as well as the knowledge about, and of, their organizations are representative to act as key informant (Huber & Power, 1985). The respondents work in different economic sectors and their organizations are geographically based in the Netherlands. The online questionnaire was sent to 161 potential respondents.
University of Amsterdam, Amsterdam Business School. 28 Only one reply email was received, mentioning the respondent found it hard to understand a few items. 94 respondents out of 113 respondents who started filling out the questionnaire fully completed the questionnaire (response rate 58,4%), while 16 respondents were excluded because of partly missing data or not completing the questionnaire. The missing data problem can be handled and solved by a ‘Hot deck imputation’, a tool for SPSS used by some
researchers. It enables to handle missing data in a practical way (Myers & Mason, 2011). Although by some considered effective, it replaces missing data based on calculations and acts as artificial input. And, we still do not really know why data is missing or why
respondents did not answer. Therefore, data imputation can also cause inaccuracy of measures and was not used. Moreover, the amount of 94 respondents multiples more than fifteen times the six constructs and is considered to be sufficient for performing quantitative analyses.
3.3. Descriptive statistics
Table 1 shows descriptive statistics derived from the sample including educational level, organization age, organization size and economic sector. The educational level of respondents was used for reliability reasons, as previously mentioned. On average 100% of the employees had at least a higher vocational education degree which is consistent with part time student’s admission requirements. This makes controlling for differences in educational level
redundant, since all respondents have high educational levels. Respondents with low and middle level education did not participate, as desired. The average firm age was 90,2 years and more than half of the organizations (57,4%) have over 500 employees. As shown in table 1, most of the responding organizations are active in industries that belong to the tertiary (50,0%) sector, followed by the quaternary economical sector (39,4%). The primary and secondary sectors were underrepresented.
University of Amsterdam, Amsterdam Business School. 29
3.4. Data analysis method
The data of useful surveys (N=94) were imported in the Statistical Software Package for Social Sciences (SPSS) and was prepared for statistical analyses. First, one reverse coded item of the construct of formalized routines (‘In my organization our front-line people are 'on their own, even with routine tasks’) and one item of the construct of informalized non-routines (‘In my organization personnel must follow formal procedures for non-routine processes’) were recoded to avoid the risk of acquiescence response bias (e.g. ‘yea-saying’). Second,
descriptive statistics, skewness, kurtosis and normality test were computed. Results showed a bit of skewness in the data and a few outliers. This can be explained by the use of a non-probability sample, a cross sectional survey approach and the diversity in organization size, age and economic sector. Because the scale of construct items was relatively small (1-7 Likert scales), outliers were not removed. Subsequently, I evaluated the internal reliability of scales by analyses of Cronbach’s alpha (CA) since it enables examination of internal consistency of measurements. After sufficient CA scores were met, scale items were computed into scale means and a full correlation matrix was assembled (table 2). Last, multiple hierarchical
Table 1
Profile of responding organizations
Overall (N=94) Percentage Educational Level (Employee with at least a Higher Vocational education) 94 100.0%
Firm Age (average years) 90.2
Firm Size 0-10 employees 3 0.2% 11-50 employees 8 8.5% 51-100 employees 9 9.6% 101-250 employees 14 14.9% 251-500 employees 6 6.4% > 500 employees 54 57.4% Economic Sector Primary sector 1 1.1% Secondary sector 9 9.6% Tertiary sector 47 50.0% Quarternary sector 37 39.4%
University of Amsterdam, Amsterdam Business School. 30 regressions analyses were performed to test for direct and mediation effects. Because
including all four control variables showed fuzzy outcomes or diffuse explanations of variance in the regressions analyses, the control variables economic sector was removed.
The hypothesized direct relations were tested in three steps. In step one, the control variables were inserted in the multiple regression models to control for organization age and size. In step two, the proposed antecedents were included as well to test for direct effects. I applied (Cohen et al., 2013) multiple hierarchical regressions, because they allow for effect testing of hypotheses and simultaneously for controlling other independent variables in a model. Further, it is common and widely used for hypotheses testing in behavioral science and business. Subsequently, in step 3, the hypothesized direct effects of the proposed
mediators explicit and tacit knowledge sharing where measured by entry of one mediator per multiple hierarchical regression. Though this measures the increased difference in explained variance of the total model, it does not compute the combined paths of the indirect effect directly. Hence, bootstrapped multiple hierarchical regressions were performed by using a PROCESS-macro from Hayes (2016). This macro does compute the size or degree of indirect effects including their confidence intervals and determines the presence of mediation effects, after controlling for organization age and size. Effects were tested by adding the mediators separately.
3.5. Measurement instruments
In this this paragraph the used constructs of the study are introduced. The constructs were all derived from previous validated constructs and marginally adapted to fit the circumstances of this study. The last subsection includes the explanation of the control variables.
3.5.1. Constructs
The questionnaire included measures for six constructs: formalized routines, informalized non-routines, explicit knowledge sharing, tacit knowledge sharing, innovation speed and
University of Amsterdam, Amsterdam Business School. 31 innovation quality. All items per construct were presented as statements throughout the questionnaire (see Appendix B). To answer construct items a seven-point Likert-type scale
(Allen & Seaman, 2007; Brown, 2011) ranging from “Entirely disagree” (1) to “Entirely
agree” (7) was used. Respondents answered by registering the most appropriate response and only one answer per item. All measured items per construct were adopted from previously validated scales of English studies from the field of strategic management and innovation research. Since the majority of respondents speak and read Dutch as their native language, every item was translated in Dutch followed by a back-translation in English. A third person was asked to carry out the translation to prevent misinterpretation by respondents. A small number of discrepancies was corrected in the Dutch questionnaire.
For measuring explicit and tacit knowledge sharing, as well as innovation speed and
quality, the items from Wang & Wang (2012) were used. Each construct has between five and
seven items. Their research was, like this one, across industries and items per construct showed high composite reliability and Cronbach’s Alpha (CA) loadings (0.96; 0.97; 0.94 and 0.93).
Explicit knowledge sharing is the exchange of knowledge enshrined within
organizations through standardized processes, methods and practices. Explicit knowledge items contain items of organizational practices concerning the collecting and sharing of formal documents, availability of education and development programs, as well as IT systems. Tacit knowledge sharing is the exchange of knowledge that is not captured in organizations. It occurs through contact between employees exchanging their (unique) knowledge and experience. Tacit knowledge items contained measures for sharing or collecting knowledge of employee’s experiences, proficiency or lessons learned, as well as collecting and sharing of know-where and know-whom. The construct items of innovation
University of Amsterdam, Amsterdam Business School. 32 launching and development, new processes and problem solving. Respondents were asked to compare their organizations with competitors or equivalent organizations on these items. The same items account for the construct of innovation quality, but instead of innovation speed the novelty and creativity of innovational activities were measured compared with competitors or equivalent organizations.
To measure the constructs of formalized routines and informalized non-routines, the items of Baum & Wally (2003) were adopted. These were used in an empirical study across industries in which, (among other organizational dimensions) effects of routine formalization on organizational performance were measured. Each construct had a high CA (0.76; 0.84). Formalized routines and tasks are common and well known in organizations and occur repeatedly. Formalized processes had three items and measured the presence of formal communication regarding routine processes and the use of standard operational procedures. One item was reverse coded which means a relative low score indicates a relatively high level of formalized routines. Informalized non-routines are not known or common in organizations and are more related to informal social interaction processes between employees. The
construct of informalized non-routines consisted out of four items of which also one was reverse coded. It measured the lack of routine tasks and processes, as well as the freedom employees have in the execution of informalized tasks and processes. Items of each construct are represented in Appendix B.
3.5.2. Control variables
Four control variables were included to account for differences among organizations; organization size and age, economic sector and educational level. As indicated in previous research these may have a potential impact on innovational performance as well as the level of formalization. First, small organizations have higher creative accumulation than larger firms, whereas the size and age of (large) firms show signs of past success and ability to
University of Amsterdam, Amsterdam Business School. 33 market new innovational products. They both may influence the present innovational
performance. (Ravichandran & Lertwongsatien, 2005). Second, as the count of employees for young organizations grows as well as the need to shift from informal interactions into
formalized routines, I controlled for firm size and age to account for differences in levels of
formalization and coordination (Davila, 2005; Greiner 1972, 1998). I also controlled for
economic sector and educational level, to account differences in innovation speed and quality.
This has two reasons. On the one hand industry structure, especially the concentration and degree of unionization, may restrain innovation, while on the other hand the educational level of employees, in particular skilled workers in fast developing industries, enhances innovation
(Acs & Audretsch, 1998; Dolfsma & van der Panne, 2008). However, as reported in the
descriptive statistics, there were insufficient responses from the primary and secondary economic sector and all respondents have at least a high educational level. Including these control variables gave fuzzy analytical outcomes in regressions. Therefore, both were excluded.
3.6. Construct validity and reliability
In order to assess the quality of the research model I measured convergent validity of individual constructs. To analyze the internal reliability of items from individual constructs Cronbach’s Alpha (CA) analyses were performed. CA represents the estimator of the internal consistency of items within a construct. This test is performed to verify whether all items per construct contribute to its measurement and shows whether an item should be removed to improve internal consistency. A threshold of 0.7 assures sufficient internal reliability
(Carmines & Zeller, 1979; Nunnally & Bernstein, 1994). Only the scale items of informalized non-routines showed insufficient internal consistency (CA = 0.60). Therefore, the item ‘In my organization I can get information that I need when I face unusual problems without going through channels’ was dropped. The internal-correlation of items within the construct also
University of Amsterdam, Amsterdam Business School. 34 showed confirmation for this removal since it was the only item with a negative correlation with two other items, respectively -0.06 and -0.01, and absence of internal-correlation (≤ 0.20) with one other item. Removal improved CA by 0.13 to a sufficient level of 0.73. Further removal of other items of the construct did not increase CA. The construct of innovation speed showed possibilities for improving CA when 1 or 2 items were deleted. However, with a high initial CA score of 0,91 and removal of 1 or 2 items it leads to a minimum
improvement of CA by respectively 0.02 or 0.03. This does not compensate for the loss of data. The same applies for the construct of tacit knowledge sharing. Removal of one item improves the initial CA of 0,89 by less than 0.01. Other constructs were considered good since CA’s were 0.78 up to 0.94 and no further improvement was possible by item deletion. CA scores are represented in table 2. Furthermore, for all final adjusted and unadjusted constructs applies that the internal-correlation per individual construct showed correlations varying between 0.35 up to 0.87 indicating tendencies to, or high, positive relations between items of individual constructs.
3.7. Common method variance
A non-probability sample was used for data collection from single key informants per organization. Therefore, the potential risk of common method bias exists and may endanger the validity of the results. To lower the risk of common method bias the independent website of ‘Quatrics’ was used to create an untracked open link to the survey which was protected against indexation by online search engines and secured uploading of files. This link guarantees respondents’ anonymity and confidentiality. To improve validity only their IP-addresses were collected, which also prevented respondents to fill in the survey more than ones, although pausing and continuing the survey later was made possible. Another important closely related danger is social desirability (Podsakoff et al., 2003; Podsakoff & Organ, 1986). Therefore, in addition to the above, the sequence of the constructs in the survey was modified
University of Amsterdam, Amsterdam Business School. 35 in a non-structured order (Podsakoff et al., 2003) compared to the position of variables in the research model. This should reduce the effect of respondents giving socially desirable
answers.
3.8. Non-response bias
Although the response rate is relatively high (58,4%), the non-response rate may distort the results and lower the quality of the survey, especially in small sample sizes. Two main reasons for non-response are unwillingness to participate or non-reachable respondents. Therefore, it is important to test the quality of the respondents answers by performing a key informant check. To assess non-response bias a multiple analyses of variance was conducted (Armstrong and Overton, 1977) to check for non-response bias and differences between the early and latter respondents. All participants were divided into two groups, early respondents (N=30) and late respondents (N=64), and independent sample T-tests were performed for the mediating and dependent variables. All the results were non-significant (P > 0.05), meaning that there is statistically no significant variance between early and late respondents.
University of Amsterdam, Amsterdam Business School. 36
4. Results
This chapter reports the quantitative results derived from analyses. It starts by the correlations between variables. Second, the quantitative results of all the hypothesized direct associations between formalization, knowledge sharing and innovational performance are outlined. Last, the analyses of the mediating roles of explicit and tacit knowledge sharing are outlined.
4.1. Correlations
In table 2 an overview is presented of the means, standard deviations, correlations and scale reliabilities of the model variables. The first noticeable observation derived from the table is that formalized routines shows significant strong positive correlation with explicit knowledge sharing (r = .56, p < .01) and a significant positive relation with tacit knowledge sharing (r = .38, p < .01), but non-significance and absence of correlation on both innovation speed and innovation quality. This might indicate there is no significant direct or indirect effect of formalized routines on innovational speed or quality, but formalized routines may positively affect both explicit and tacit knowledge sharing. The second noticeable observation is that informalized non-routines show no significance or correlation with explicit and tacit knowledge sharing directly, but does show significant tendency to positively correlate with innovation speed (r = .23, p < 0.5). This signals that there might be direct influence of
informalized non-routines on innovation speed. The most interesting and striking observations are the tendencies for positively correlation of both explicit and tacit knowledge sharing on respectively innovation speed and innovation quality. Explicit knowledge sharing shows significant positive correlation with innovation speed (r = .29, p <.01) and innovation quality (r = .39, p < .01) as well as tacit knowledge sharing does on innovation speed (r =.33, p < .01) and innovation quality (r = .42, p < .01). These significant correlations might indicate full or partially mediation effects of knowledge sharing between the informalized non-routines and innovational performance of organizations. Last, as expected firm size shows two significant
University of Amsterdam, Amsterdam Business School. 37 positive correlations (p = < .01) for both formalized routines (r = .32) and informalized non-routines (r = .38) which indicates that levels of routine formalization go up when firms grow. Significant negative correlations (p < .01) occur for both control variables firm age and size on innovation speed (r = -.28 and -.24) signaling that innovation speed is slower for large and older organizations compared to young and small ones.
4.2. Hypotheses testing for direct effects
Multiple hierarchical regressions for direct effects were independently performed for both independent and mediating variables and their effects on innovation speed and quality, after controlling for firm age and size. This allows to investigate the influence effects in isolation. Results (table 3, 4, 5, 6, 7 and 8) for all tested direct effects show significance for 8 out of 12 hypotheses (p < 0.05), when organization age and size as predictors for innovation speed and innovation quality are included in the models. Yet, organization age and size are not
confirmed to have a significant statistical influence (p > 0.05) on the research model. For hypotheses 1 (H1) and 2 (H2), I hypothesized that formalized routines and
informalized non-routines predict a positive association with innovation speed and innovation quality, respectively. After inserting the antecedents formalized routines and informalized non-routines in the H1 model (table 3), for testing effects on innovation speed, the model does not significantly improve (p > 0.05), whereas the model for effects on innovation quality (H2)
Table 2
Mean, Standard Deviations, Correlations and Reliability
Construct Number of items ME SD 1 2 3 4 5 6 7 8 9 10
1. Educational level 1 5.39 .85 - 2. Firm size 1 4.85 1.55 -.04 -3. Firm age 1 90.21 88.80 -.02 .48** -4. Economic sector 1 3.28 .68 .02 .12 .35** -5. Formalized routines 3 5.42 1.33 .02 .32** .16 -.03 ( .78) 6. Informalized non-routines 3 4.43 1.36 -.07 .38** -.20 -.04 -.30** ( .73) 7. Explicit knowledge sharing 6 5.27 1.14 .05 .17 .09 -.12 .56** -.20 ( .85) 8. Tacit knowledge sharing 7 5.43 .98 .00 .00 .03 .17 .38** -.00 .60** ( .89) 9. Innovation speed 5 4.43 1.44 .09 -.28** -.24* -.24* .05 .23* .29** .33** ( .94) 10. Innovation quality 5 4.51 1.24 .00 -.12 -.03 -.20 .16 .20 .39** .42** .80** ( .91) Notes : N = 94. Reliabilities (Crombach's alpha) are reported along the diagonal
ME: Mean; SD: Standard Deviation.
*
. Correlation is significant at the .05 level (two-tailed)
**
University of Amsterdam, Amsterdam Business School. 38 as shown in table 4 does significantly improve (p = 0.005). The explained variance increases by 9% from 1.4% to 10.4%. Statistically significant support is found for the positive direct effect of formalized routines on innovation quality (H2a) with b-value 0.25 and p = 0.015. Contrary to H2a, formalized routines have no direct significant positive effect on innovation speed (H1a) (b = 0.21; p > 0.079). The same applies for the predicted direct positive
associations of informalized non-routines on innovations speed and quality, respectively. Hypothesis H2b, measuring the effect of informalized non-routines on innovation quality is supported (b = 0.22; p = 0.034), but no significant positive effect was found for hypothesis H1b (b = 0.19; p = 0.108). With b-values of 0.25 and 0.22, formalized routines have a little higher predictive effect than informalized non-routines have on innovation quality. Hence, organization’s formalized routines and informalized non-routines do not directly enhance innovation speed of organizations, but they both do enhance their innovation quality.
Table 3
Results of Formalized routines and Informalized non-routines as predictors for Innovation speed
Innovation Speed
Variable R R² R² change b SE Beta t Sig.
Model 1 0.308 0.095 0.095 0.011 Organization size -0.20 0.106 -0.22 -1.900 0.061 Organization age 0.00 0.002 -0.14 -1.219 0.226 Model 2 0.376 0.141 0.047 0.096 Organization size -0.20 0.114 -0.21 -1.724 0.088 Organization age 0.00 0.002 -0.14 -1.232 0.221 Formalized routines 0.21 0.115 0.19 1.778 0.079 Informalized non-routines 0.19 0.115 0.18 1.621 0.109
Note: N =94; all test are two tailed p ≤ 0.05
Table 4
Results of Formalized routines and Informalized non-routines as predictors for Innovation quality
Innovation Quality
Variable R R² R² change b SE Beta t Sig.
Model 1 0.118 0.014 0.014 <0.001 Organization size -0.10 0.096 -0.13 -1.084 0.281 Organization age 0.00 0.002 0.03 0.231 0.817 Model 2 0.322 0.104 0.090 0.005 Organization size -0.10 0.100 -0.12 -0.999 0.321 Organization age 0.00 0.002 0.03 0.247 0.806 Formalized routines 0.25 0.102 0.27 2.470 0.015 Informalized non-routines 0.22 0.101 0.24 2.150 0.034
Note: N =94; all test are two tailed p ≤ 0.05
