Current rapidly-changing markets ask for firms that can both innovate incrementally, as well as radically to survive. To be able to simultaneously perform these seemingly conflicting activities, this article poses that a firm needs to have certain strategic orientations. Building on recent literature regarding strategic orientations, ambidexterity in general and innovative ambidexterity specifically, this paper identifies the mediating role of innovative ambidexterity in the relationship between market orientation (MO), entrepreneurial orientation (EO) and firm performance. This paper contributes to existing research as it identifies the combination of MO and EO as antecedents of innovative ambidexterity and for the first time tries to assess the mediating role of innovative ambidexterity in the relationship between MO, EO and firm performance. Using data from 109 respondents from various industries in the Netherlands, the results show that though MO and EO have a significant effect on innovative ambidexterity, the latter does not mediate their relationship with firm performance. In the end, rejecting the claim that innovative ambidexterity fully mediates the EO - firm performance relationship and partially mediates the market orientation - firm performance relationship.
Keywords: market orientation, entrepreneurial orientation, innovative ambidexterity, firm
performance.
University of Amsterdam – UvA Amsterdam Business School Executive Programme in Management Studies – Strategy Track Thesis supervisor: Dr. Sebastian Kortmann Student: Daan Schouten – 10730737 Original date of submission: 30-06-2015
Statement of Originality
This document is written by Daan Schouten, 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.
Introduction p. 4
Proposed conceptual model p. 8
Literature review and hypotheses p. 9
Strategic orientations p. 9
Market orientation p. 11
Entrepreneurial orientation p. 15
Ambidexterity p. 19
Innovative ambidexterity p. 20
Data and method p. 23
Data collection p. 23 Sample p. 24 Measures p. 26 Control variables p. 28 Statistical procedure p. 28 Results p. 30 Correlations p. 30 Regression analyses p. 32 Discussion p. 39 Managerial implications p. 41
Limitations and future research p. 42
Conclusion p. 43
References p. 44
Appendices p. 50
Appendix A: exploratory factor analysis MO p. 50
Appendix B: exploratory factor analysis EO p. 51
Appendix C: residual plots independent variables p. 52
4
INTRODUCTION
Since Porter (1979) wrote his paper on competitive forces and their effect on strategy, business scholars are looking for the strategic directions firms need to take to gain a competitive advantage. Current literature suggests that not only the choice for a certain generic strategy is important for gaining and sustaining this advantage. The attention now focusses on the mechanisms and processes within the firm to execute the strategy and adapt it to the rapid changing environment of nowadays (Gatignon & Xuereb, 1997). Failure to do so leads to quick demise of even the biggest and most successful companies, with such famous examples as: Kodak, Nokia and Xerox.
So on the one hand, the choice for a certain strategy is important as a starting point for a firm in a certain market. On the other, the implementation of this strategy into context, processes and mechanisms within the firm is equally important to gain and sustain a competitive advantage. Without these mechanisms to adapt to changing circumstances in the market a firm will quickly lose its advantage (Menguc & Auh, 2006). Strategic orientations are described as such strategic implementation mechanisms (Noble, Sinha, & Kumar, 2002). These ‘orientations’ form the utterly important bond between the chosen strategy of a firm and the actual execution of this strategy and consist of all the “the strategic directions implemented by a firm to create the proper behaviors for the superior performance of the business” (Gatignon & Xuereb, 1997, p.78).
Having strong strategic orientations in a firm could according to many researchers thus lead to improved performance (Miles, Snow, Meyer, & Coleman, 1978), both in the short- and long-term (Menguc & Auh, 2006) because it helps firms adapt to their environment. From a competitive forces standpoint this makes sense, because strategic orientations are the directions in which a firm focusses because of certain market conditions. But it also agrees with the resource-based-view, in that these orientations are of great value to the company, intangible and thus difficult to imitate
5 (Wiklund & Shepherd, 2003). According to a meta-study done by Grinstein (2008a), the strategic orientations with the most effect on firm performance are market orientation, innovation orientation and entrepreneurial orientation. Especially the combination of market orientation (MO) and entrepreneurial orientation (EO) is much focused upon, mostly because of their augmenting effects (Grinstein, 2008a). MO is argued to be directed at satisfying current customers and finding new ones in the current market (Narver & Slater, 1990), where EO is directed at risk-taking, expanding to new markets and exploring new technologies (Lumpkin & Dess, 1996). This paper takes great interest in the combination of these two strategic orientations in relation to firm performance, as they create the potential within a firm to both exploit the current resources and strategy and explore new ones.
This dual focus on these two seemingly paradoxical processes is called ambidexterity (Birkinshaw & Gibson, 2004) and according to many authors should be treated as vital for firm success in current rapidly changing markets (He &Wong, 2004). Coming from the Latin words “ambos” and “dexter”, meaning literally “both hands”, it depicts the capability of organizations
to both exploit current resources and explore new ones (March, 1991; Tushman & O’reilly III, 1996; Simsek, 2009; Raisch, Birkinshaw, Probst, & Tushman, 2009), or pursue both adaptation and alignment (Gibson & Birkinshaw, 2004; Berghman, 2012; De Clerq, Thongpapanl, & Dimov, 2013). Within ambidexterity literature there are roughly two streams of thought on how to accomplish this state, either structurally through differentiation and integration mechanisms (March 1991; Tushman & O’reilly III, 1996; Jansen et al., 2009) or contextually through creating the right culture where ambidextrous actions can be taken (Birkinshaw & Gibson, 2004; Gibson & Birkinshaw, 2004). Both streams claim that ambidexterity nowadays is a major reason for firm success, but while the need for ambidexterity in the current rapid-changing markets is commonly accepted, there is a general lack of empirical findings on the subject and especially antecedents to
6 ambidexterity are not that well established (Raisch & Birkinshaw, 2008; O’reilly III & Tushman, 2013). This paper would like to deepen that understanding and sees MO and EO as two potentially important starting points.
In their article, Baker and Sinkula (2009) test three different models which incorporate MO, EO, innovation success and firm performance. Their results show that MO and EO are both mediated by innovation success in their effect on firm performance. With the potential ambidextrous effect of the two orientations, this paper takes a slightly different approach in stating that not innovative success mediates this relationship but innovative ambidexterity. Innovative ambidexterity, is the state where firms both innovate incrementally (mostly seen as product of exploitation) and radically (mostly seen as a product of exploration) and therefore can be defined as a realized state of the aforementioned organizational capability (Li, Lin, & Chu, 2008). As stated before, this dual focus is essential for being and staying successful as a company. If for example a firm only launches very successful incremental innovations, it could endanger its performance in the long run when new breakthrough innovations threaten the status quo. When a firm only launches successful radical innovations it could endanger the immediate cash-flow and bottom-line of the firm. Therefore it is adamant that MO and EO not only lead to innovation success, but lead to this simultaneous duality in incremental and radical innovations.
Prior work has identified that cost orientation, innovation orientation (Kortmann, 2014) and market orientation (Li et al., 2008) all have significant effect on innovative ambidexterity. This paper extends along the ideas of the paper by Li et al. (2008), who found a positive relation between MO and innovative ambidexterity, but adds EO as it enhances the explorative/radical innovation side of it. In the research done by Kortmann (2014) cost orientation interacts with the exploitative side of ambidexterity and innovation orientation to both. The combination of market orientation and entrepreneurial orientation follows the same logic, because it is generally believed that MO
7 tends more to the exploitative side and EO more to the explorative side of the construct (Baker & Sinkula, 2009).
As found from earlier research, both orientations have effects on: various measures of (direct) performance and innovation, either on their own (Zhou, Yim, & Tse, 2005; Hult, Hurley, & Knight, 2004; Lumpkin & Dess, 1996) or together (Baker & Sinkula, 2009; Grinstein, 2008a). Many authors also made a case for considering innovativeness or innovation success as a mediator between MO, EO and performance (Slater & Narver, 1995; Baker & Sinkula, 2009). With so much research done along these lines, it is remarkable that there is so little research on the relation between both orientations and innovative ambidexterity (Li et al., 2008).
First of all, it is argued in this study that there is need for clarification and further research on what could be possible antecedents for innovative ambidexterity. Combining earlier research on strategic orientations with the existing ambidexterity research, it could potentially shed a new light on the contextual vs. structural debate and could also substantiate the claims that strategic integration mechanisms are important to obtain innovative ambidexterity (Kortmann 2014, Li et al., 2008). Where market orientation and entrepreneurial orientation have shown to be related to various measures of performance and innovation/innovativeness, together they have not yet been addressed as antecedents of innovative ambidexterity. Secondly this paper claims that having strong MO and EO together makes a firm ambidextrous and should lead ultimately to a higher performance. Where in earlier literature innovation success is mediating this relationship between MO, EO and performance, nowadays innovation success is not enough to gain and sustain superior firm performance. The main question which arises from the current lack of clarity in the relationship between antecedents in relation to innovative ambidexterity and performance is: does the combination of MO and EO lead to superior firm performance and does innovative ambidexterity mediate this relationship?
8
Proposed research model
The proposed research model as described in figure 1 depicts the first concept of the research model. In the literature review part, based on earlier research, hypotheses will be drawn and added to this fist crude version of the model. This elaborated variant of the model can be found directly following the literature review part.
Figure 1: proposed research model containing variables Market orientation
Entrepreneurial orientation
Innovative
9
LITERATURE REVIEW AND HYPOTHESES Strategic orientations
Articles on strategic orientations in general make it clear, that because of the vast amount of work on the subject, one single definition of the construct seems to be out of the question (Noble et al., 2002). One of the initial articles on strategic orientations is the well-known paper by Miles et al., (1978). In their study they classify three types of strategic orientations: defenders, analyzers and prospectors. Miles et al. (1978) see these orientations as adaptation devices for dealing with entrepreneurial, engineering and administrative problems, which firms tackle in the so-called adaptive cycle. While Miles et al. (1978) take a fairly responsive approach to strategic orientations and see them as answers to problems coming forth from certain business processes, another key article by Gatignon and Xuereb (1997) defines the construct as: “the strategic directions implemented by a firm to create the proper behaviors for the superior performance of the business”
(Gatignon & Xuereb, 1997, p. 78). Here strategic orientations are more of a choice from top management to steer the organization to superior performance. Kohli and Jaworski (1990) take a third approach in defining it as an important element of the firm’s total culture. The discussion mimics the discussion of structural ambidexterity (the malleable, top-down variant) vs. contextual ambidexterity (the cultural, holistic view).
This paper agrees with Gatignon and Xuereb (1997) that before an orientation can become part of the culture it needs to be introduced and implemented by top-management. Their definition also shows the more dynamic relationship between a firm’s strategic direction(s) and its behavior, plus it names superior performance as its ultimate goal. The literature regarding strategic orientations focusses much on this well-established relationship between certain strategic orientations and firm performance (e.g. Noble et al., 2002; Grinstein, 2008a; Baker & Sinkula, 2009; Lumpkin & Dess, 1996). The choice for a certain strategy does not exclude a combination
10 of strategic orientations though. In their meta-analysis, Noble et al. (2002) found strong evidence for an enhanced effect when combining certain strategic orientations. These effects were also established in other papers, either as interaction effect with certain orientations as independent variables (e.g. Grinstein, 2008a; Baker & Sinkula, 2009), or as independent variable and moderator (Bhuian, Menguc, & Bell, 2005).
Another set of consequences which is much researched by scholars in the various fields of strategic orientations is the effect of the different strategic orientations on innovations and innovativeness (e.g. Gatignon & Xuereb, 1996; Baker & Sinkula, 2009; Hult et al., 2004). Though the positive results of certain strategic orientations on innovations and innovativeness are generally found and supported, Zhou et al. (2005) argue that different orientations are needed for different kinds of innovations. In their article they show that for example, for breakthrough market-innovations, market orientation is not a suitable strategic direction. According to them, the effect of a strategic orientation depends a) on the sort of innovation (market vs. technology based) and b) on the radicalness of the innovation. Kortmann (2014) used cost-orientation and innovation orientation as mediators between ambidextrous decisions and innovation ambidexterity and found evidence that cost-orientation tends more to the exploitative side of innovative ambidexterity. Besides his article though, the combined effect of strategic orientations on (innovative) ambidexterity remains a fairly unexplored terrain. This paper agrees with Zhou et al. (2005) and Kortmann (2014) that some orientations have a more exploitative nature and some a more explorative one. And it will distinguish between these two sides of the (innovative) ambidexterity construct.
11
Market orientation
Market orientation is either described as a cultural phenomenon (Narver & Slater, 1990) or as a set of behaviors (Jaworski & Kohli, 1993). Proponents of the cultural explanation see market orientation as an important part of the total culture of the firm, while those in favor of the behavioral explanation address market orientation as a fixed set of behaviors implemented to focus the firm on getting the most useful information out of their respective environment. Either way, the construct usually has three dimensions: customer orientation, competitor orientation and inter-functional coordination (Grinstein, 2008a). Customer orientation includes the understanding and measurement of (latent) customer needs. Competitor orientation focusses on knowing the firm’s (in)direct competitors and their strengths and weaknesses. Inter-functional coordination is the integration of the marketing concept in all areas within the firm (Han, Kim, & Srivastava, 1998). Along these lines, MO can be defined as a distinct form of a strategic orientation, focused on using market intelligence to satisfy current and latent future customer needs, in this way creating superior value (Narver & Slater, 1990). This “superior value part” is seen as an important element of the definition, as much research has been done on understanding the consequences of MO on performance related attributes (Noble et al., 2002).
Market orientation and performance
Effects of MO on performance (e.g. Noble et al., 2002; Slater & Narver, 1994; Grinstein, 2008a; Deshpandé & Farley, 1998) have been widely established and researchers showed this performance enhancing effect of MO, with both subjective (e.g. Jaworski & Kohli, 1993; Hult et al., 2004) and objective performance measurements such as profitability (e.g. Narver & Slater, 1990; Baker & Sinkula, 2009). The effect is further solidified by various meta-studies done by: Cano, Carrillat and Jaramillo (2004), Noble et al. (2002) and Kirca, Jayachandran and Bearden (2005). Some
12 researchers even argue that MO can be seen as the most important orientation (Hult et al., 2004; Grinstein, 2008a; Hult & Ketchen, 2001) because of its large and commonly found effect on firm performance. According to them, the strength of the relationship between MO and firm performance stems from intelligibly using market information to adapt quickly to changing market-circumstances (Grinstein, 2008a).
Innovativeness is seen as a culture characteristic which seems to be an important moderator or interacting independent variable of the MO – performance relationship. Day (1994) and Slater and Narver (1995) argue that MO is actually a learning orientation and is part of a firm’s culture. Slater and Narver (1995) furthermore describe that innovativeness could mediate the relationship between MO and various measures of performance, as it enhances this learning effect of MO. Menguc and Auh (2006) also see innovativeness as the key for market orientation to become a firm-level dynamic capability. They argue that MO has become more or less common-place and that firms need to couple it with high levels of innovativeness to obtain and keep their competitive advantage. Their results show that when complemented by innovativeness the results of MO on performance are significantly stronger.
Though the effect of MO on performance seems to be well developed, there are some articles in which the relationship could not be established. For example, Han et al. (1998) could not replicate a direct effect of MO on firm performance and instead found that organizational innovation fully mediates this relationship. Bhuian (1997) also found no evidence for enhanced performance within Saudi-Arabian banks. Furthermore, Jaworski and Kohli (1993) could establish a positive effect when firm performance was measured subjectively, but did not get a significant result when firm performance was measured objectively in the form of market share. So apparently, MO does not always lead to higher firm performance. For the greater part, these less positive results are assessed in the meta-study done by Kirca et al. (2005). In their paper, the relationship between
13 MO and performance is elaborated and mediating and moderating variables are assessed. In their analysis they establish an overall significant correlation between MO and performance and found significant evidence for factors influencing the MO – performance either as mediator or as moderator. In correspondence with the articles that could not establish the MO – performance relationship, they conclude that this relationship is stronger for 1) production firms, 2) firms with certain cultural traits, 3) firms operating in dynamic markets and 4) when performance is measured subjectively (Kirca et al., 2005). Some of these variables will be taken into account when performing this study. With the majority of the articles, including at least three meta-studies acknowledging the positive effect of market orientation on firm performance, this paper formulates the following hypothesis:
H1: market orientation has a significant positive effect on firm performance.
Market orientation and innovation measures
Another well-developed consequence of MO is its effect on innovations and innovativeness. Various measures pertaining to the realm of innovations have been researched as consequences of MO, such as: innovation success (Baker & Sinkula, 2009), new-product development (Atuahene-Gima, 1996) and types of (radical) innovations (Lukas & Ferrell, 2000; Zhou et al., 2005). Most of the time, the relationships between MO and innovation measures are either established with the innovation-measure as outcome (Zhou et al., 2005; Atuahene-Gima, 1996) or as mediator (Baker & Sinkula, 2009; Hult et al., 2004; Han et al., 1998) for firm performance. As with the MO – performance relationship, innovativeness is shown to be an important enhancing factor (Verhees & Meulenberg, 2004). Grinstein (2008b) has done another meta-analysis on the effects of MO on innovations. He found that all three dimensions have positive effects on innovation consequences, but did not assess MO as a complete construct.
14 The results of the aforementioned papers are agreeing on a positive MO – innovation relationship. When innovation is further split up into different kinds of innovations, the effect of MO on these innovations is mixed though. In their paper on the effect of MO on different types of performance, Lukas & Ferrell (2000) for example found, that the different dimensions of MO (customer orientation, competitive orientation and inter-functional coordination) have different effects on innovations varying in degree of radicalness. Especially the effects of MO on the most-radical ‘new-to-the-world innovations’ are mixed. Where the dimension customer orientation shows a positive effect, competitor orientation shows a negative effect on this type of radical innovations and inter-functional coordination shows no significant effect at all (Lukas & Ferrell, 2000). Zhou et al. (2005) further show, that MO in a model together with EO could have a negative effect on breakthrough market-based innovations. Reason for this negative effect is that the focus of a firm with strong MO lies more on satisfying current customer needs, than on finding and developing new markets for itself. When modeled with EO, which has a pro-active and explorative posture, this paper reckons that this contrast will be even stronger.
Along the same tracks, Li et al. (2008) argue that market orientation can further be divided into pro-active market orientation and responsive market orientation. Where responsive market orientation stands for: the way that existing customer needs are fulfilled and pro-active market orientation is the way a firm tries to respond to the latent needs of these customers through experimentation and discovery. In the study done by Li et al. (2008) it was found that each type of market orientation has a higher effect on either exploration or exploitation, where pro-active market orientation has a larger effect on exploration and responsive market orientation has a larger effect on exploitation. This research uses the more traditional definition of market orientation (Deshpandé & Farley, 1999), which is similar to the construct of ‘responsive market orientation’ described by
15 Li et al. (2008) and models MO together with the more explorative concept of EO. Therefore this paper hypothesizes that:
H2a: market orientation has a significant positive effect on incremental innovations (exploitative
side of innovative ambidexterity) when modeled together with entrepreneurial orientation.
H2b: market orientation has no significant effect on radical innovations (explorative side of
ambidexterity) when modeled together with entrepreneurial orientation.
Entrepreneurial orientation
Another strategic orientation which through the years has gotten a lot of attention, is entrepreneurial orientation. EO is focused on discovering and using untapped possibilities within and outside of the firm (Lumpkin & Dess, 1996). EO consists of three to five dimensions: pro-activeness, innovativeness and risk-taking (Covin & Wales, 2011) and additionally: autonomy and competitive aggressiveness (Lumpkin & Dess, 1996). Pro-activeness is the propensity of a firm to move forward and actively introduce new product or services before competitors do so (Rauch, Wiklund, Lumpkin, & Frese, 2009). Innovativeness has already been addressed in the MO literature as an important moderator or mediator for that construct. It pertains to exploration, research and development and experimentation to introduce new (technological) developments to the market (Lumpkin & Dess, 1996). Renko, Carsrud and Brännback (2009) compare innovativeness to explorative learning as used by March (1991) and as such seems to be strongly related to the exploration side of innovative ambidexterity. Risk-taking is using resources to take certain steps with an unverifiable outcome (Rauch et al., 2009). Furthermore, competitive aggressiveness is the will to win over competitors and autonomy is the degree of independence used for entrepreneurial actions by the leaders and their teams (Rauch et al., 2009). There is still a lot of debate over the use
16 of these dimensions as either one higher-order construct like the Covin and Slevin (1989) scale or as separate and independent dimensions, as described by Lumpkin and Dess (1996). This paper takes after Covin and Slevin (1989) and uses the dimensions as a higher-order construct.
Entrepreneurial orientation and firm performance
Just as MO has a vast amount of research done on its relation to performance, within the realm of EO the link with firm performance is also often assessed. According to Rauch et al. (2009), the results of these studies vary quite a bit. From studies where the positive effects of EO on firm performance are strong (e.g. Wiklund & Shepherd, 2003; Keh, Nguyen, & Ng, 2007) to studies where this link is weaker or even not-established (e.g. Madsen, 2007; Lumpkin & Dess, 2001). The meta-study by Rauch et al. (2009) of 55 different articles on the link between EO and firm performance confirmed the positive effect as moderately large though (r = .242).
In Grinstein’s (2008a) meta-analysis, it is argued that MO could be a good addition to EO and that their combined effect on firm performance is stronger than when observed separately. While this effect is established when EO moderates the MO – firm performance relationship (Li, Zhao, Tan, & Liu, 2008), the combined effect of both orientations on performance is still under debate. Bhuian et al. (2005) for example also research EO as a moderator between MO and performance and argue that the level of EO should be “just enough” to have the most positive effect. They describe the EO - firm performance relationship as an inverted U-curve. If entrepreneurial orientation is too low, the firm is too selective, risk-averse and not pro-active enough to exploit knowledge gained from the market. If entrepreneurial orientation is too high, it tends to be overly dominant within the firm, which in turn will not listen to the market and start exploring on its own. Furthermore, Hughes and Morgan (2007) did not find a positive effect of EO on firm performance in their research under recent start-ups. They argue that taking more risks and
17 being more pro-active in some situations may not be the right path to choose.
Even more contradictive is the paper by Baker and Sinkula (2009), which reports direct effects of EO and MO on innovation success. Their results show no direct relationship of EO on firm performance when modeled together with MO as independent variables. They pose that MO and EO have overlapping underlying dimensions and that when modeled together, the effect of these dimensions within EO are removed together with its positive effect on performance. This lack of effect from EO on performance is also acknowledged by Slater and Narver (1998). Subsequently, Matsuno, Mentzer and Özsomer (2002) even found a negative effect for EO when combined with MO in one model. Baker and Sinkula’s (2009) paper also tested three different models of MO, EO, innovation success and firm performance and found that the one with MO and EO as independent variables had the best fit. This paper therefore models both constructs in the same fashion and takes after Baker and Sinkula (2009) in hypothesizing that:
H3: entrepreneurial orientation has no significant effect on firm performance when modeled
together with market orientation.
Entrepreneurial orientation and innovation measures
Where it was assumed that MO (in its traditional form) adheres more to the exploitative side of innovative ambidexterity, it will be interesting to see whether EO also tends more to one side of the continuum. As Rauch et al. (2009) describe EO as having a proclivity for driving change and moving forward (the pro-activeness dimension) through creativity, experimentation and new product development (the innovativeness dimension), it would be a logical step to first discover the relationship between explorative innovation and EO. The article by Zhou et al. (2005) identifies EO as an important antecedent of breakthrough innovations. They establish positive effects for the effect of EO on both market-based breakthrough innovations and technology-based breakthrough
18 innovations. In their research EO has the strongest effect of the three tested orientations (MO, EO and technology innovation) and is the only orientation which has a positive effect on both types of breakthrough innovations. Atuahene-Gima and Ko (2001) describe EO as mirroring generative and exploratory learning and found that EO leads to faster market entry than MO due to its exploratory and forward-leaning posture. New products by firms with a high measure of EO also made relatively more money than those of firms with a high measure of MO. This can potentially also be attributed to the exploratory value of EO, as it gives the new products more competitive advantage and thus generates more money (Atuahene-Gima & Ko, 2001). Avlonitis and Salavou (2007) found support for their hypothesis that active entrepreneurs (whom instill a high degree of EO in their firms) have a significant effect on new-product uniqueness, which is contributed to exploratory research and innovation. Baker and Sinkula (2009) also distinguish MO as clearly related to exploitative innovations, while EO pertains more to radical innovations.
The link between exploratory innovation and EO may be clear, but what about EO and exploitative innovations? While it is logical to describe EO as an explorative construct, there are some authors who also describe the exploitative side of it. Kraus, Rigtering, Hughes and Hosman (2011) for example argue that entrepreneurship in general is equally about exploring and exploiting opportunities, as it is about generating opportunities as well as taking them. Lisboa, Skarmea and Lages (2011) empirically tested the potential that EO holds for both explorative and exploitative capabilities and found for both exploitative and explorative capabilities a mediating effect between EO and new-product differentiation. Furthermore, it is argued that the different dimensions of EO have different effects on innovation (Pérez-Luño, Wiklund, & Cabrera, 2011) and that for example, the dimension pro-activity can lead to exploiting innovations done by competitors (adoption) and those generated by the own firm. MO could furthermore augment the exploitative effect, because of the intelligence gathered from customers and competitors. With this information, the firm could
19 use its EO posture to adapt innovations from competitors or actively introduce incremental innovations based on current customer needs. In all the aforementioned papers, EO has coupled both exploitative and explorative capabilities and characteristics though. It is this paper’s view that when coupled with only exploitative innovations, the promise EO holds for explorative innovation goes forlorn. This paper thus agrees that EO should be assessed as both affecting the incremental (exploitative) and radical (explorative) innovation side of ambidexterity, especially when modeled together with MO. Therefore it is hypothesized that:
H4: entrepreneurial orientation has a significant positive effect on both incremental innovations
(exploitative side of innovative ambidexterity) and radical innovations (explorative side innovative
ambidexterity) when modeled together with MO.
Innovative ambidexterity
The concept of ambidexterity was first described by Duncan (1976). Later on, the seminal articles by March (1991) and Tushman and O’reilly III (1996) further investigated the promise ambidexterity holds for positive effects on firm performance. Al three authors describe the structural separation of exploration and exploitation as the way a firm becomes ambidextrous. This structural ambidexterity usually takes place one unit down the structural ladder. So within the level of the firm it is achieved by giving business units either the task to exploit or to explore, while within business units it is created by division between teams (Raisch et al., 2009).
With the growing interest in this fairly recent field of research, a different view on how to attain ambidexterity focuses on creating the right organizational context (Adler, Goldoftas, & Levine, 1999; Birkinshaw & Gibson, 2004). This contextual ambidexterity, as opposed to the structural variant, is inseparable, simultaneous and is achieved within all levels of the organization:
20 within the firm, business units and teams (Berghman, 2012). Originally following Ghoshal and Bartlett’s (1994) article on how organizational context mediates the relationship between managerial action and the behavior of individuals in the firm. Firms that score high on both the performance context and the social context are considered to show a form of “high performance context” which is argued to have a positive relationship with contextual ambidexterity (Birkinshaw & Gibson, 2004). Organizational ambidexterity is a distinct dynamic capability as it consists of the processes, routines and resources bundles which firms use to both exploit and explore (Jansen et al., 2009). Innovative ambidexterity is using the organizational ambidexterity concept to explain how firms can simultaneously innovate incrementally (exploitation) and radically (exploration) and it can be seen as a tangible effect of ambidexterity as described earlier.
The mediating role of innovative ambidexterity
In the previous paragraphs, their separate effects on exploration and exploitation have been described extensively and it is argued that MO pertains more to exploitation while EO pertains to both with more evidence for its effect on exploration. This is in congruence with Baker and Sinkula (2009), who acknowledge the need for both types of innovations and argue that EO influence exploration and MO influences exploitation, combining them to explain innovation success and subsequently firm performance. It also takes after Kortmann (2014), who describes the balancing of cost orientation and innovation orientation as having a significant positive effect on innovative ambidexterity. Strategic orientations can form an important link as cultural implementation mechanisms between a firm’s strategy and its innovative ambidexterity (Kortmann, 2014). These mechanisms are critical to overcome difficulties coming from the aforementioned structural separation of units in ambidextrous firms (Jansen, Tempelaar, Van den Bosch, & Volberda, 2009). Strategic orientations in general thus make it possible for firms to simultaneously address
21 exploration and exploitation issues from a structural ambidexterity point of view.
From a contextual ambidexterity viewpoint, strategic orientations supply the firm with a distinct set of cultural traits which form the context of the firm. In the article by Brion, Mothe and Sabatier (2010) for example, creativity, risk-taking and autonomy are all addressed as cultural traits that influence innovation ambidexterity within the firm. These three items combined are also part of the EO construct, thus strengthening the evidence for a potential effect between EO and innovative ambidexterity. MO have been described as an antecedent of innovative ambidexterity by Li et al. (2008), though they split market orientation in two parts. The part that is used in this article affects only the exploitative part. This paper argues though that the pro-active part of market orientation (which influence the exploration side of innovations) can be substituted by EO, as it also represents a pro-active posture and stimulates experimentation and creativity just as pro-active MO does. Based on these papers it is thus thought that MO and EO, with their already established effects on the separate sides of ambidexterity, also supply the firm with the means to balance exploitation and exploration and pursue them in simultaneous fashion.
But to be considered a potential mediator, the effect of innovative ambidexterity on firm performance need to be established as well. This effect has been established multiple times, and has been found to be a promising way to adapt to changes in the environment and to strengthen performance (He & Wong, 2004; Lubatkin, Simsek, Ling, & Veiga, 2006). This claim of the potential positive effect on firm performance has been substantiated over the years, both in manufacturing companies (He & Wong, 2004), service-firms (Marabelli, Frigerio, & Rajola, 2012), in large organizations (He & Wong, 2014) and smaller ones (Chang, Hughes, & Hotho, 2011; Chang & Hughes, 2012; Lubatkin et al., 2006). Because of these clear effects on firm performance it is hypothesized that:
22 H5: market orientation and entrepreneurial orientation have a positive effect on innovative
ambidexterity.
H6a: innovative ambidexterity has a positive effect on firm performance and thus partially
mediates the relationship between market orientation and firm performance.
H6a: innovative ambidexterity has a positive effect on firm performance and thus fully mediates
the relationship between entrepreneurial orientation and firm performance.
Figure 2: final research model including hypotheses and assumed relationships Market orientation
Entrepreneurial orientation
Innovative
ambidexterity Firm performance
H1 H3 H2 a,b H6 a,b H4
II
H523
DATA AND METHOD
Data collection
This study has taken a quantitative empirical approach to assess the different variables in this study and their relationships. Quantitative was chosen over qualitative study design because of the already established measurements of the constructs and some of the relations within the conceptual model. The survey contained 40 Likert-scale questions, but was combined with three other studies done on relating subjects. This brought the total number of questions to 134, which took most respondents approximately between 11 and 30 minutes to fill in. The study had a “runtime” of a little over a month and was distributed April 3rd and closed on May 6th. The questionnaire was sent out to circa 700 potential respondents, all of which were acquaintances, co-workers, family or friends of the researchers. The sampling method used is convenience sampling (use of acquaintances) together with snowball sampling (it was asked to send the questionnaire along to more potential respondents). This gives a potential risk of selection bias, but on the other hand is helpful for an efficient use of time. Also, because the sample contains respondents from a diverse set of firms (both in size, industry, product vs. service firms and performance), we expected the external validity to be less biased. The only requirement for filling in the survey is basic understanding of the context (culture and structure) and performance of the organization the participant is working in, so potentially the tenure of one working in a firm could bias the results. This was added to the general questions section and included as a control variable in the bivariate correlation matrix.
24
Sample
The number of respondents which completed the questions concerning the main constructs of this study was 137. In the sample taken, 20 responses were found to have all of the general (descriptive) questions missing and 8 more were found to have only the organizational tenure question missing. These cases were excluded list-wise, so that in the final model 109 complete responses remained. This brought the response rate for the study to 15,6%. The survey was sent mostly to relatives, co-workers, peers and acquaintances of the author(s). This should have led to a stronger response rate, but was offset against the main reason respondents discontinued their efforts, which was the length of the study. This was shown in the dropout rate of 49%, which is quite high.
Of the complete sample 26% were female and 74% male, their age ranging from 19-24 to 65 and older, with the mode lying within the 45 – 54 years category. The sample which was used for this research consisted mostly of people with a higher education, for 6% had a bachelor’s degree, 40% had a master’s degree or doctoral degree and another 39% finished the University for Applied Sciences. Employees (non-management) represented the biggest group within the sample (41%), followed by middle management (27%) and entrepreneurs (20%). No top managers took part in the survey, which could potentially be contributed to the high level of entrepreneurs in the sample. For it is common, especially in smaller firms that the entrepreneur also is the top manager or takes part in top management. Organizational tenure ranged from 1 and under to 49 years (Mtenure
= 10.76, SDtenure = 9.31) and was quite dispersed. The majority of the respondents were employed
within firms either sized 501 -1000 employees (27%), 5001 and more employees (17%) and 6 to 25 employees (16%).
25
Measures
Innovative ambidexterity
For incorporating the construct of innovative ambidexterity into the survey, there are not many well-developed scales available. This study will make use the measurement used in the article by Jansen et al. (2009). They asses innovative ambidexterity via its two parts, exploitative and explorative innovations, measured independently with each four 7-point Likert items, ranging from 1 (strongly disagree) to 7 (strongly agree). These scales are then combined for the effect of innovative ambidexterity. The original study found a Cronbach’s alpha of α = 0.70 for exploitative innovations and α = 0.86 for explorative innovations, making them both reliable measures. Because this paper hypothesizes that the MO and EO have different effects on the explorative and exploitative sides of innovative ambidexterity, the scale by Jansen et al. (2009) is first used in said two parts. Then the two subscales are added two each other to test for the complete effect of both independent variables on innovative ambidexterity. This method of adding both scales had according to Jansen et al. (2009) the most explanatory power compared with multiplying or subtracting the two subscales.
In contrast to innovative ambidexterity, market orientation has been tested thoroughly. In their article, Deshpandé and Farley (1998) studied three different measurements of market orientation. They found that all three earlier measurements are indeed reliable with very good Cronbach’s alphas and can be used interchangeably. In their study the authors make a case for synthesis of the three scales, for means of using the true core when measuring market orientation and thus shortening part of the questionnaire. Their combination and distillation of the three scales resulted in a new scale named “MORTN” a series of 10, 5-point Likert-scale questions ranging 1 (disagree) to 5 (agree). This synthesized scale will be used in this study, both because it is short
26 and sweet, has a managerial focus (Deshpandé and Farley, 1998) and has a good reported Cronbach’s alpha of (α = 0.88). The original scale is transformed to a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree), after which reliability and validity were assessed and still found to be good.
For testing entrepreneurial orientation, first it have to be decided which sub-scales to include. While the original scale existed of three subscales (Covin & Wales, 2011) and was described in the article by Covin and Slevin (1989). Lumpkin and Dess (1996) added autonomy and competitive aggressiveness as two other subscales. When added to the existing dimensions, namely: innovativeness, pro-activeness and risk-taking, this makes five different subscales. This complete scale containing five dimensions, is developed by Hughes and Morgan (2007) and contains twenty-two 7-point Likert-scale questions, ranging from 1 (strongly disagree) to 7 (strongly agree). The scale is originally treated in a disaggregated way, were all subscales are tested independently. This research chose to combine the subscales into one larger EO-scale, because the research question of this thesis is different from the Hughes and Morgan (2007) paper. Where Hughes and Morgan (2007) wanted to see which of the independent subscales of EO had a significant effect on firm performance, this research would like to test the construct as a whole. It is also chosen to incorporate all five subscales though and not test only three of them, mostly because autonomy could be important for a firm in being ambidextrous. As explained in the Gibson and Birkinshaw (2004) article and the Birkinshaw and Gibson (2004) article, shaping the right context also means leaving room for employees to experiment and find optimal ways to do things. So leaving autonomy out of the equation would in this paper’s view omit a potential important factor in the EO – innovative ambidexterity relationship.
27 Firm performance was tested using a self-reported scale by McDougall et al. (1994) from the Kreiser, Marino and Weaver (2002) article. The scale asks respondents to reflect on four performance items in comparison to their foremost competitors, over the last three years. The original scale has a Cronbach’s alpha of α = 0.89. The four items are 5-point Likert type statements, ranging from 1 (disagree) to 5 (agree) and was adapted to a 7-point Likert scale. Just as for the MO scale, reliability and validity were assessed and found sufficient. A self-reported scale was preferred instead of an objective measurement of performance, because of the ease of use. Also the more holistic view of measuring performance with four items was preferred over a simple objective figure like net profit or EBIT, to get a more complete picture of a firm’s performance. Furthermore, measuring performance over the last three years could give an indication of the sustainability of the performance.
Control variables
The study incorporates five control variables, of which two were found to correlate significantly with the independent variables. The control variables incorporated in the final model were: organizational tenure and firm size. Organizational tenure was used as a continuous measure, and rounded to a full year of work at that particular firm. Firm size was measured using an 11-point scale ranging from 1 to > 5.000 employees currently working for the firm. Because it was expected that there would be many respondents working for either bigger or smaller companies, the used range of 11-points was explicitly chosen for its extensiveness.
Statistical analysis
Statistical Package for the Social Science edition 23.0 for Windows was used to run the statistical analyses of the data and to test the hypotheses in this paper. First of all, the data was examined and given appropriate labels. Answers in two scales (firm performance and firm size) needed recoding,
28 after which the frequencies showed no irregularities. Scale reliabilities were checked, psychometric tests were performed to assess the descriptive statistics of the different constructs and normality tests were performed. Kurtosis (-.421 to .334) and skewness (-.508 to -.025) all fall within the ‘moderate’ range, so normality could be established. When checking the reliability of the constructs, in both the EO scale and the exploratory part of the innovative ambidexterity scale, one item was found to negatively influence the overall Cronbach’s alpha. The item from the innovative ambidexterity scale (our organization regularly uses new distribution channels) was deleted, which brought the Cronbach’s alpha from α = 0.68 to α = 0.70. The item in the EO scale moved the Cronbach’s alpha slightly from α = 0.92 to α = 0.93 and it was decided to keep the item in the scale. The Cronbach’s alpha’s found in the reliability checks were sufficient to very good (see table 1: means, standard deviations and correlations), some even higher than the ones found in the original studies and reported in the ‘data and method’ section of this paper.
Then, exploratory factor analyses (EFA’s) were established to check the construct validity of the different scales. Both explorative and exploitative innovations and firm performance were found to have loadings on one factor. The MO items loaded on two factors (see Appendix A: exploratory factor analysis MO), which can be best explained as one factor pertaining to the gathering of customer data (customer orientation) and one factor to the execution of the strategic direction from the firm’s point of view (competitor and inter-functional orientation). For EO another exploratory factor analysis was performed where the number of expected factors (5) was fixated, the results are shown in Appendix B: exploratory factor analysis EO. Though not all items loaded as expected, from the analysis it can be derived that autonomy and risk-taking load well on their respective factors. The other suspected factors are less clear, where items from innovativeness and pro-activeness load onto one factor and other pro-activeness items load together with competitive aggressiveness items. Because of the reliability of these items and the complete scale,
29 it was chosen not to exclude these items though. Only one item was excluded (employees have access to vital information), as it solely loaded on a factor and exclusion would not lead to a lower Cronbach’s alpha. After all scales were assessed, a bivariate correlation matrix was established to check for significant relations between the various constructs (see table 1: means, standard deviations and correlations). Control variables were entered as well, to see which of them correlate significantly with the independent variables and thus should be incorporated into the final regression model.
Multicollinearity could be an issue in multiple regression analysis. Therefore, the reported VIF-values were checked to see if multicollinearity could be influencing the regression model. When using linear regression, following (Osborne & Waters, 2002) four assumptions should be fulfilled first. Normality is already assessed via kurtosis and skewness levels, and could be established. Linearity and homoscedasticity were then assessed via the residual plots and could be established for all constructs as well (see Appendix C: residual plots). Statistical independence only needs to be checked when performing multiple tests over time.
After these preliminary tests, a multiple regression analysis with two mediators was performed using the PROCESS macro by Hayes (2012), instead of following the causal steps approach (Baron & Kenny, 1996). The advantage of using the PROCESS macro is that it performs all subsequent analyses automatically and that it makes use of bootstrapping. According to Hayes (2009) bootstrapping has much more power than the aforementioned causal steps approach and is less influenced by data which violates assumptions regarding its distribution. Because PROCESS uses bootstrapping it does not supply us with a regular confidence interval, but uses a bias-corrected confidence interval (Hayes, 2009) and accompanying p-value. If the zero does not fall within that interval, the effect of the test is significant. PROCESS uses unstandardized B-values for assessing the slopes of the separate variables. First, both independent variables were tested in two
30 independent models, secondly the two independent variables were tested in one model and ultimately the independent variables were tested together with the control variables. This way, as PROCESS does not have the option to feed the variables in different ‘blocks’ like the regular linear regression module in SPSS, the effect size (R2) of both the independent and control variables could
be assessed. Hayes (2009) advises to use at least a bootstrapping sample of 5.000, which this paper did.
RESULTS
Correlations
First of all the results from the correlation matrix will be discussed, then various regression analyses (direct and indirect) will be assessed which will lead to the discussion section of this paper. Two correlation matrices were devised. One to test for correlations between the independent variables and the two separate scales of innovation as used by Jansen et al. (2009). The second matrix to test for correlations between the independent variables and the higher-order construct of innovative ambidexterity (the sum of both innovation scales).
At first glance, the first correlation matrix (see table 1: means, standard deviations and correlations) shows us that all independent variables have significant positive correlations with each other and with the dependent variable, firm performance. MO correlates strongly with firm performance (r = .34, p < 0.01), as does EO (r = .46, p < 0.01). When assessing the independent variables and potential mediators, it can be seen that EO is strongly correlated with both innovation constructs, respectively r = .52, p < 0.01 and r = .50, p < 0.01. MO also correlates quite strongly with both innovation constructs, respectively r = .36, p < 0.01 and r = .45, p < 0.01. This could mean that both strategic orientations have an effect on both the explorative and exploitative side of innovations, but needs to be confirmed through performing a linear regression. Because of the high
31 correlations between the independent variables, VIF-values will be checked for high levels of multicollinearity, as this could harm the overall regression model.
In the second correlation matrix, the higher-order construct of innovative ambidexterity replaces the two separate innovation scales (explorative and exploitative). Innovative ambidexterity correlates strongly with MO (r = .46, p < 0.01) and EO (r = .58, p < 0.01). The correlation matrix shows that, innovative ambidexterity also correlates with firm performance (r = .29, p < 0.01). From the correlations there seems to be enough evidence to proceed with the regression analyses.
Table 1: means, standard deviations and correlations
Variables M SD 1 2 3 4 5 6 7 8 9 10 1. Gender 1.26 0.44 -2. Age 4.23 1.16 -.14 -3. Education level 5.63 1.57 .06 -.32** -4. Firm size 6.62 3.22 -.01 -.12 .24* -5. Organizational tenure 10.76 9.31 -.17 .55** -.19* .03 -6. Market orientation 4.80 1.01 -.01 .18 -.12 .18 .20* (0.87) 7. Entrepreneurial orientation 4.24 1.04 -.13 .09 -.12 -.31** .07* .41** (0.92) 8. Explorative innovations 4.78 1.22 .02 -.00 -.09 -.11 .04 .36** .52** (0.70) 9. Exploitative innovations 5.06 1.00 -.15 .01 -.00 .07 .10 .45** .50** .56** (0.76) 10. Firm performance 4.48 1.17 .08 -.11 -.13 -.12 -.08 .34** .46** .26** .24* (0.91)
Please note: N = 109, Cronbach's alpha's for scale reliability reported on the diagonal between brackets.
**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
Table 2: means, standard deviations and correlations
Variables M SD 1 2 3 4 5 6 7 8 9 1. Gender 1.26 0.44 -2. Age 4.23 1.16 -.14 -3. Education level 5.63 1.57 .06 -.32** -4. Firm size 6.62 3.22 -.01 -.12 .24* -5. Organizational tenure 10.76 9.31 -.17 .55** -.19* .03 -6. Market orientation 4.80 1.01 -.01 .18 -.12 .18 .20* (0.87) 7. Entrepreneurial orientation 4.24 1.04 -.13 .09 -.12 -.31** .07* .41** (0.92) 8. Innovative ambidexterity¹ 9.83 1.96 -.06 .00 -.06 -.03 .08 .46** .58** n/a
9. Firm performance 4.48 1.17 .08 -.11 -.13 -.12 -.08 .34** .46** .29** (0.91)
Please note: N = 109, Cronbach's alpha's for scale reliability reported on the diagonal between brackets.
**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
32 Finally, two control variables need to be included in the regression model. Firm size has a negative correlation with EO (r = -.31, p < 0.01). Which could make sense, as smaller firms are usually more entrepreneur-driven. Organizational tenure has positive, but smaller, correlations with both MO (r = .20, p < 0.05) and EO (r = .07, p < 0.05). One explanation could be that, the longer people work for a certain firm, the more knowledgeable they are of the culture and context of the firm, so the higher they value it.
Regression analyses
Four regression analyses were ran to test the different hypotheses formulated in the literature review part of this thesis. First a preliminary analysis was performed to check for multicollinearity. Second, a regression analysis was performed to test the effect of the independent variables on the two separate scales of innovation. Third, because the PROCESS macro used for the regression analysis does not allow for ‘blocks’ of variables to be included in the model, only the control variables were introduced in the model. This way the separate explained variance (R2) of both the control- and independent variables could be assessed. Ultimately, the complete model was included in the analysis. The PROCESS macro allows for testing the direct and indirect effects via bootstrapping and bootstrapped confidence intervals, as is described in the method section of this paper.
From table 3, the VIF-values and tolerance levels can be checked of the different independent variables and potential mediators. From the table we can conclude that all VIF-values fall within the acceptable range (VIF <10), so do the tolerance levels (>.01). Multicollinearity should not be an issue in the final regression model.
33 Table 3: collinearity statistics
Variable Tolerance VIF
Market orientation .739 1.353
Entrepreneurial orientation .629 1.596
Innovative ambidexterity .607 1.642
Direct effects
As shown in tables 4, 5 and 6, direct effects could be established between some of the independent variables and dependent variables. Table 4 presents the results from the regression analysis with explorative innovation as dependent variable. A multiple regression analysis was conducted to predict explorative innovations by means of MO and EO. The regression model was found to be significant (F (4, 104) = 11.18, p < 0.01), with an explained variance of R2 = .30. The results show that only EO has a significant direct effect on explorative innovations (B = .52, p < 0.01), this is confirmed by the bootstrapped confidence interval (BC95 = [.29, .25]). MO does not have a
significant direct effect on explorative innovations, as goes for the two control variables (firm size and organizational tenure). In table 4 the results are also shown from the regression analysis with exploitative innovation as dependent variable. Again, a multiple regression analysis was conducted to predict exploitative innovations by means of MO and EO. The overall regression model was found to be significant (F(4, 104) = 13.76, p < 0.01), with an explained variance of R2 = .35. The results indicate that both EO (B = .44, p < 0.01) and MO (B = .22, p < 0.01) have a significant effect on exploitative innovations, as is acknowledged by their respected bootstrapped confidence intervals (BC95 = [.04, .42]) and (BC95 = [.26, .62]). For both control variables a significant effect
could not be established.
Table 5 shows the results of the multiple regression analysis with innovative ambidexterity as dependent variable. The multiple regression analysis was conducted to predict innovative
34 ambidexterity by means of MO and EO. The regression model was found to be significant (F(4, 104) = 17.25, p < 0.01), with an explained variance of R2= .40. The results indicate that both MO (B = .45, p < 0.01) and EO (B = .96, p < 0.01) have a significant effect on innovative ambidexterity, as is confirmed by their respective bootstrapped confidence intervals (BC95 = [.10, .80]) and (BC95
= [.62, 1.31]). Again, both control variables do not seem to have a significant effect on the dependent variable.
Furthermore, table 6 shows the results of the multiple regression analysis with firm performance as dependent variable. The multiple regression analysis was conducted to predict firm performance by means of MO, EO and innovative ambidexterity. The regression model was found to be significant (F(5, 103) = 7.33, p < 0.01), with an explained variance of R2 = .26. From the results can be concluded that both MO (B = .26, p < 0.05) and EO (B = .44, p < 0.01) have a significant effect on firm performance, as is again confirmed by their respective bootstrapped confidence intervals (BC95 = [.02, .50]) and (BC95 = [.18, .70]). Innovative ambidexterity is included
as an independent variable, but has no significant effect on firm performance, which applies to both control variables as well.
Finally, table 7 shows the total effect model where firm performance is again used as dependent variable. The multiple regression analysis was conducted to predict innovative ambidexterity by means of MO and EO. The regression model was found to be significant (F(4, 104) = 17.25, p < 0.01), with an explained variance of R2 = .26. The results indicate that both MO
(B = .25, p < 0.05) and EO (B = .42, p < 0.01) have a significant effect on innovative ambidexterity, as is subsequently confirmed by their respective bootstrapped confidence intervals (BC95 = [.02,
.49]) and (BC95 = [.19, .65]). The difference between the total effect model and the multiple
regression shown in table 6 is the exclusion of the possible mediator, innovative ambidexterity. From the differences in coefficients between both models, the mediating effect of innovative
35 ambidexterity can be inferred. The separate effect (R2) of the control variables, as compared to the main variables was assessed first. From it, could be deduced that the main variables (MO, EO) explain a large part (R2 = .20) of the variation in de dependent variable (firm size) as opposed to the control variables where the effect is only small (R2 = .02) and non-significant.
Indirect effects
This paper hypothesizes that the direct effect between MO, EO and firm performance is mediated by innovative ambidexterity. This potential mediation effect is tested through regular regression analysis and bootstrapped regression analysis, via the PROCESS macro by Hayes (2012). The results in table 8 pertain to the mediation effect of innovative ambidexterity on the relationship of MO and EO on firm performance.
From the data shown by the PROCESS macro, it can be concluded that innovative ambidexterity has no mediating effect on the relationship between MO, EO and firm performance. This can be inferred from the bootstrapped confidence intervals showing the indirect effect from MO (BC95 = [-.08, .04]) and EO (BC95 = [-.14, .10]). As the zero falls within the interval in both
instances, innovative ambidexterity does not have a significant mediating effect. This conclusion is also supported by the regular regression analysis (or causal steps method as used by Baron & Kenny, 1986). When comparing the total effect model (table 7) with the direct effect model (table 6), the changes in both the B-coefficients for MO and EO are very small. Furthermore, innovative ambidexterity is non-significant in the direct effect model, also an indication that innovative ambidexterity does not function as a mediator (full or partial) in the relationship between MO, EO and firm performance.
36 Table 4: summary of bootstrapped multiple regression analysis for variables predicting explorative and exploitative innovations
Variables B SE B p LLCI ULCI B SE B p LLCI ULCI
Market orientation .23 .12 .06 -.01 .46 .22* .09 .02 .04 .41 Entrepreneurial orientation .52** .12 .00 .29 .75 .44** .09 .00 .26 .62 Firm size -.00 .04 .93 -.07 .07 .05 .03 .06 -.00 .11 Organizational tenure -.00 .01 .78 -.02 .02 .00 .01 .81 -.02 .02 R² .30 .35 F 11.17** 13.76**
Note: listwise exclusion (N = 109), bias-corrected confidence intervals (95%), bootstrapped sample 5000
* p < .05, ** p < .01
Explorative innovations Exploitative innovations
Table 5: summary of bootstrapped multiple regression analysis for variables predicting innovative ambidexterity
Variables B SE B p LLCI ULCI
Market orientation .45** .18 .01 .10 .80 Entrepreneurial orientation .96** .17 .00 .62 1.31 Firm size .05 .05 .34 -.05 .15 Organizational tenure -.00 .02 .95 -.03 .03 R² .40 F 17.25**
Note: listwise exclusion (N = 109), bias-corrected confidence intervals (95%), bootstrapped sample 5000
* p < .05, ** p < .01
37 Table 6: summary of bootstrapped multiple regression analysis for variables predicting firm performance
Variables B SE B p LLCI ULCI
Market orientation .26* .12 .03 .02 .50 Entrepreneurial orientation .44** .13 .00 .18 .70 Innovative ambidexterity -.02 .07 .77 -.15 .11 Firm size -.01 .03 .68 -.08 .05 Organizational tenure -.02 .01 .10 -.04 .00 R² .26 F 7.33**
Note: listwise exclusion (N = 109), bias-corrected confidence intervals (95%), bootstrapped sample 5000
* p < .05, ** p < .01
Firm performance
Table 7: summary of bootstrapped total effect model for variables predicting firm performance
Variables B SE B p LLCI ULCI
Market orientation .25* .12 .03 .02 .49 Entrepreneurial orientation .42** .12 .00 .19 .65 Firm size -.02 .03 .66 -.08 .05 Organizational tenure -.02 .01 .10 -.04 .00 R² .26 F 9.22**
Note: listwise exclusion (N = 109), bias-corrected confidence intervals (95%), bootstrapped sample 5000
* p < .05, ** p < .01
38 Table 8: summary of bootstrapped indirect effects of MO and EO on firm performance, mediated by innovative ambidexterity
Variables Effect SE LLCI ULCI
Market orientation -.01 .03 -.08 .04
Entrepreneurial orientation -.02 .06 -.14 .10
Note: listwise exclusion (N = 109), bias-corrected confidence intervals (95%), bootstrapped sample 5000
* p < .05, ** p < .01
