Ambidexterity: A Meta-Analysis of Organizational
Antecedents
Paradoxical Tensions Revised
Master thesis
Business Administration - Business Development
University of GroningenFaculty of Economics and Business Department of Business Development
Nettelbosje 2
9747AE Groningen, The Netherlands
By Marissa Gubler
Studentnumber: s2040484
m.gubler@student.rug.nl
Date: August 2012
First supervisor: dr. J.D. van der Bij
Second supervisor: dr. C. Reezigt
Preface
This preface is for the reader the beginning of the research. For me this is almost the end. My thesis period started with the search for an interesting topic. After struggling with my thoughts of what I perceive as interesting, the possibility to participate in a project concerning ambidexterity came along. The University of Groningen gave the opportunity to partake in a challenging research project. This appealed to me because of its trending and provocativecharacter. In a previous course I had chosen this topic to investigate. However, this investigation was rather brief and left my mind puzzling. Therefore, the ambidexterity project gave me the opportunity to dive into the topic and try to solve the puzzle.
Before starting this thesis, I had never conducted a meta-analysis. With this thesis, I learned the protocol of a meta-analysis and the interpretation of its results. Furthermore, it taught me how to properly perform academic research and writing. The process of this project also improved my collaboration and communication skills. I learned how to systematically store and share a great amount of data with my colleague and supervisors. Not only did I learn about performing academic research, I also learned about the field of ambidexterity and its paradoxical tensions.
My sincere thanks go out to dr. J.D. van der Bij and dr. C. Reezigt for their guidance and sharing of knowledge during the process. Without their experience and critical reviews, the thesis could not reach its current academic level. In addition, I would like to thank Gerko Telman for his pleasant teamwork and work climate. Furthermore, I would like to thank my parents in supporting me during the previously year and giving me the possibility to obtain my Master degree. Finally, my thanks go out to my friends and family for their support during my thesis, in special Harlinda Gubler for her critical view regarding my academic writing.
Marissa Gubler
Abstract
Ambidexterity is essential for long-term survival in today’s fast changing market. Ambidextrous organizations are able to simultaneously explore new opportunities and improve existing capabilities. Studies found a positive significant relationship between firm performance and ambidexterity. Therefore, there is a raising interest on the organizational antecedents of ambidexterity in empirical studies. However, the results of these studies are often provocative and fragmented in terms of nature of the relationship and the type of antecedents. The purpose of this study is to eliminate these discords that exist in current academic literature by conducting a analysis. Through this meta-analysis, this paper contributes to a comprehensive view of how organizations should manage and achieve an ambidextrous strategy. Based on extensive literature search, this study identifies 39 different independent variables organized into eleven conceptual constructs. After correcting for artifacts and sample size effects, of the eleven potential ambidexterity’s antecedents identified in current literature, only market orientation is homogeneous related to ambidexterity. Two of these eleven potential antecedents are homogenous significant antecedents for exploitation and exploration; respectively market orientation and output control, and learning and market orientation. For the remaining ten heterogeneous antecedents of ambidexterity, a moderator analysis was conducted. Of this set, one appeared to be an antecedent of ambidexterity, namely formalization. However, it seems to be only an antecedent of exploitation. This study concludes with future research directions.
Keywords: Ambidexterity, exploration, exploitation, tensions, meta-analysis
Introduction
A fast pace of change is required of organizations in order to compete in the nowadays rapid developing market. Due to this, organizations are confronted with the tension of exploiting existing capabilities and exploring new ones (Levinthal & March, 1993; March, 1991; Jansen, Van Den Bosch & Volberda, 2006). Developing existing capabilities are needed in order to perform more efficient and consequently keep costs low. On the other hand, exploring new opportunities are needed in order to meet the changing market demands. Focus on one guarantees short-term success but also long-term failure (Tushman & O’Reilly, 1996). As Li, Lin and Chu (2008) state, rapid continuous change is an essential capability that organizations need to attain for long-term survival. He and Wong (2004) go a step further in saying that simultaneously exploitingthe development of existing capabilities as exploring new ones is a prerequisite for sustained performance. The ability to do this is called ambidexterity. Duncan (1976) introduced this term and described ambidexterity as the ability of organizations to design dual structures that facilitate both the initiating stage and implementation stage of innovations (Gibson & Birkinshaw, 2004). However, March (1991) was the first that linked ambidexterity to the two related concepts of exploration and exploitation (Cao, Gedajlovic & Zhang, 2009). Some academic literature refers to this concept as the capability-rigidity paradox which states that the exploitation crowds out exploration, because organizations fail to exploit both strategies at
the same time (Leonard-Barton, 1992; Atuahene-Gima, 2005; Tushman & O’Reilly, 1996). The consequence of not being ambidextrous is the increased risk of falling in a trap of a downward spiral of mediocrity (March, 1991; Chang, Hughes & Hotho, 2011).
The importance of ambidexterity to organizations is stressed by the investigation of the relationship between firm performance and ambidexterity, which resulted in a positive outcome (e.g. Beckman, 2006, Chang et al., 2011; Jansen et al., 2006; Menguc & Auh, 2008; De Visser, De Weerd-Nederhof, Faems, Song, Van Looy & Visscher, 2010). Due to this importance of building ambidexterity in today’s businessit is essential to identify how organizational factors need to be managed to create a successful ambidextrous organization. However, no consensus and conformity is attained in current academic literature on how to achieve this. Studies identify different organizational antecedents of ambidexterity. If researchers do identify the same organizational antecedents of ambidexterity, discussion concerning the nature of the relationship exists. That is whythis study focuses on analysing and eliminating these discords. These discords can originate from methodological problems, different study design, different measurements, omitted variables in the regression models, and noncomparable samples (Song, Podoynitsyna, Van Der Bij, and Halman, 2008). To preclude these methodological problems, a meta-analysis is used because it operates independently of model composition. This meta-analysis gives a transparent overview of the relationship between organizational antecedents and ambidexterity. The analysis is done on firm-level in order to provide insight on how organizations should manage ambidexterity.
The study contributes to the ambidexterity literature in several ways:
1) The research provides an integrated theoretical foundation of the organizational antecedents for ambidextrous organizations.
2) It provides a new theoretical framework for these organizational antecedents.
3) It identifies homogeneous and significant organizational antecedents of ambidexterity. 4) It identifies antecedents that are heterogeneous and conducts a quantitative evaluation to
ascertain moderators for those organizational antecedents.
5) It gives recommendations for necessary future research in the field of ambidexterity.
6) It provides a grounded basis for managers on which control mechanisms are needed to achieve a balance between tensions related to ambidexterity and thereby increasing long-term survival of their organization.
In discussing the influence of the organizational antecedents on ambidexterity, the next section presents a theoretical background on the concept ambidexterity and the different organizational antecedents identified in current academic literature. Then, the data collection and methodology are outlined. Next, the results of the meta-analysis are presented, and conclusions and implications are given. This paper concludes with its limitations and recommendations for further research.
Theoretical background
Ambidexterity
Ambidexterity is a well-known subject in the field of innovation, but is often defined differently. For example, Gibson and Birkinshaw (2004) outline the trade-off between conflicting demands that organizations manage by putting in place dual structures of alignment and adaptation. Tushman and O’Reilly (1996) cite revolutionary an evolutionary change when referring to ambidexterity. Researchers agree that ambidexterity is an ambivalent concept that stresses the ability of an organization to manage paradoxical strategies simultaneously. The most common views of ambidexterity include exploitative and exploratory innovations (e.g. Brion, Mothe & Sabatier, 2005; Cao, Gedajlovic & Zhang, 2009; Chandrasekaran, Linderman & Schroeder, 2012; Chang et al., 2011; Jansen et al., 2006) incremental and radical innovation (e.g. Germain, 1996; Koberg, Detienne & Heppard, 2003; Li et al., 2008; Menguc & Auh, 2010; Subramanian &Youndt, 2005) and initiation and implementation (e.g. Damanpour & Schneider, 2006; Grover & Goslar, 1993; Matta, Koonce & Jeyaraj, 2012). What makes ambidexterity such a debatable topic is the need for paradoxical structures and activities in managing these twin concepts concurrently. It is emphasized that exploitative innovations involve more cost efficient strategies in stable and existing markets and technologies. In such circumstances, mechanistic structures are necessary to control the tightly coupled systems in the organization. In contrast, exploratory innovations involve more uncertainty and focus on emerging markets and technologies. To manage such high uncertainty levels, organic structures are required, involving risk-taking, experimentation and flexibility (Brion et al., 2005; Gibson & Birkinshaw, 2004; He & Wong, 2004; Jansen, Tempelaar, Van Den Bosch &Volberda, 2009; Li et al., 2008; Tushman & O’Reilly, 1996; De Visser et al., 2010). These paradoxical strategies are also conflicting because of their competition for scarce resources, their activities are inherently self-reinforcing which cause failure, and the requirement of different routines (Gupta, Smith & Shalley, 2006; Im & Rai, 2008).
Due to the discord on how to define ambidexterity, this paper takes a global description of ambidexterity. Hence, ambidexterity is defined as the ability of an organization to simultaneously exploit paradoxical strategies regarding innovations. This definition is sufficiently broad to include all different tensions of ambidexterity pertaining to innovations in organizations.
Organizational Antecedents of Ambidexterity
Studies examined the relation between firm performance and ambidexterity and concluded that there is a positive significant relationship (e.g. Beckman, 2006; Chang et al., 2011; Jansen et al., 2006; Menguc & Auh, 2008; De Visser et al., 2010). Given this positive relation between firm performance and ambidexterity, it is important to examine how toachieve ambidexterity in an organization.Numerous studies focus on the antecedents of ambidexterity in an organization (e.g. Brion et al., 2005; Cao et al., 2009; Jansen et al., 2006; Subramanian & Youndt, 2005). However, the results are often provocative and fragmented in terms of type of antecedents and nature of the relationship. For example, almost one third of the studies identify formalization as antecedent of
ambidexterity. Though, the direction of the relationship differs between these studies. More than half of the studies identify a positive relation between formalization and exploration (e.g. Brion et al., 2005; Cabello-Medina, Carmona-Lavado & Cuevas-Rodríguez, 2011; Wei, Yi & Yuan, 2011), while one third found a negative relationship (e.g. Farjoun, 2010; Jansen et al., 2006; Raisch & Birkinshaw, 2008). There are also studies that found no significant relationship (Damanpour, 1991; Grover & Goslar, 1993).Based on extensive literature search, described in section “data collection and methodology”, this study identifies 47 different independent variables. These are organized into sixteen conceptual constructs to make the reviewed antecedents more transparent and interpretable (see appendix 3).The judgment criteria for combining independent variables are based on the definition and measurement of the independent variable. Figure 1 illustrates the sixteen conceptual constructs. The reason for conducting this study is because a wide disconformity prevails among scientific articles regarding ambidexterity’sorganizational antecedents. Therefore, no suggestions concerning expected relationshipsare presented in figure 1.
4
Figure 1. Conceptual model organizational antecedents of ambidexterity
The different views concerning the relationship between the organizational antecedents and ambidexterity in academic literature are reviewed below.
Centralization involves the concentration of the decision making authority in an organization (e.g. Cardinal, 2001; Jansen et al., 2006; Pierce & Delbecq, 1977). Increased control enhances the implementation of innovations but restricts the generation of new ideas (e.g. Boumgarden, Nickerson & Zenger, 2012; Cardinal, 2001). A contradictory view could be that radical innovations will originate from leadership to make decisions and allocate resources needed for the innovation.
Connectedness (cultural) reflects the common values, vision and language among employees which enhances exploitative innovations due to routinized and shared views and activities (Beckman, 2006). Structural differentiation Centralization Structural control Structural complexity Strategic leadership Risk Resources Connectedness (cultural) Connectedness
(structural) Firm age
Firm size Formalization Learning Managerial tenure Market orientation Output control Ambidexterity
Connectedness (structural) refers to the amount of direct contact between employees in an organization (e.g. Beckman, 2006; Jansen et al., 2006; Walker, 2008). This contact increases accessibility to knowledge exchanges within an organization and in turn combines this new knowledge into new ideas (Beckman, 2006; Chang et al., 2011; Damanpour, 1991; Damanpour & Schneider, 2006; Walker, 2008). Structural connectedness also stimulates exploitative innovations because it enhances strong norms, trusts and collective behavioural beliefs which in turn assist organizations to improve existing products (Chang et al., 2011; Jansen et al., 2006; Koberg et al., 2003). On the other hand, structural connectedness limits the ability to develop radical innovations due to concern for power and prestige. Employees may disclose information from each other in order to preserve one’s power, to conserve scarce resources and keep control (Germain, 1996).
Koberg et al. (2003) are the only ones that cite firm age as an organizational antecedent of ambidexterity. As firms age, structural rigidity, routinized norms and behaviours constrain organizations to develop radical innovations. A controverting approach could be that these firms are more experienced and possess more market information and could therefore quicker adapt to changes in their industry.
The previous argument concerning the relationship between firm age and ambidexterity is also stated for firm size by Koberg et al. (2003). It is argued that large organizations are more formalized, standardized, and inertia is more common (Damanpour, 1996). An opposing view is that large organizations have access to scarce resources, specialists, and have more control over their external environment which makes the ability to develop explorative innovations higher (Germain, 1996; Walker, 2008).
Formalization refers to the degree rules and procedures are formalized and written down (Jansen et al., 2006). This fosters routines and strict controls which in turn enhances exploitative innovations and inhibits experimentation, creativity and adaptation (Brion et al., 2010; Jansen, Van Den Bosch & Volberda, 2005; Jansen et al., 2006; Sidhu, Commandeur & Volberda, 2007; Walker 2008). In contrast, organizations use strict controls to better codify, replicate and generate knowledge. Strict procedures allow empowerment of employees and thus employees are able to solve problems in novel ways. Hence, rigid rules and procedures in combination with norms and trust make flexibility and creativity possible (Farjoun, 2010; Ireland & Webb, 2007).
Learning refers to collection of information to generate new ideas. Hence, learning fosters exploratory innovation due to adaptability to changing market demands (Wei et al., 2011; Morgan & Berthon, 2008). Another line of reasoning is that although learning increases information diversity, information search is often focused on current job activities due to specialization in an organization. This generates information regarding exploitative innovation (Wei et al., 2011).
Managerial tenure is associated with structural and cultural inertia which is the resistance to change structures, routines, norms, and values (Tushman & O’Reilly, 1996). This resistance is often based on previous organizational success. Managers with a short tenure are less committed to the status quo and are therefore more likely to discover new innovations (Koberg et al., 2003). A
controverting approach could be that managers with a long tenure have more experience and therefore quicker explore the innovations and changes suitable for the organization.
When an organization acquires information about its environment to understand customers’ needs and behaviour, this is called market orientation (Li et al., 2008; Morgan & Berthon, 2008). According to Morgan and Berthon (2008) market orientation focuses more on currentcustomers’ needs and therefore fails to discover radical innovations. On the other handLi et al. (2008) state that market orientation consists of reactive and proactive market orientation. The latter positively influences radical innovations due to focusing on latent customers’ needs.
Output control refers to control systems that are used to monitor outcomes. These systems use tangible and exploitative activities and thus enhance incremental innovations (Cardinal, 2001; Ettlie, Bridges & O’Keefe, 1984; McCarthy & Gordon, 2011).
Resourcesare needed in order to maintain the development of innovations (Cardinal, 2001; Ettlie et al., 1984; Germain, 1996). However, too many resources may make employees less creative in allocating resources which lead to exploitative innovations (Wei et al., 2011).
Risks need to be taken in order to explore new innovations. Hence, risk is associated with explorative innovations. However, decision risk capability is needed in an organization in order to resolve the tensions around exploitative and explorative innovations. This capability supports managers in evaluating risks constantly when making strategic choices between exploitative and explorative innovations (Brion et al., 2010; Chandrasekaran et al., 2012).
Strategic leadership involves giving purpose to the organization and making decisions to pursue this purpose along with the evolution of the organization (e.g. Chandrasekaran et al. 2012; Chang & Hughes, 2012; Damanpour, 1991; Lin & McDonough, 2011). These decisions also involve innovations. Hence, strategic leadership mediates between the tensions resulting from explorative and exploitative decisions. Damanpour (1991) argues that strategic leaders’ favourable attitude towards change enhances explorative innovations and thus strategic leadership is related to the level of risk.
Structural complexity refers to functional differentiation and role specialization. This means respectively the number of units and the amount of specialists in an organization (Damanpour, 1996). Specialization in different business units increases the depth of knowledge and in turn the development of more radical innovations (Damanpour, 1996; Ettlie et al., 1984). A controverting approach could be that specialization leads to single focus on current job activities therefore hampers the exploration of new ideas.
Structural controlcan interfere with the development of new ideas and hampers employees to perform nonroutine activities to explore new opportunities (Cardinal, 2001).
Finally, structural differentiation refers to the physical separation in the organization. Hence, structural differentiation is the subdivision of organizational tasks into different units. However, for an ambidextrous organization this physical separation is difficult due to coexistence of exploitation and exploration. Thus, structural differentiation in an ambidextrous organization entails creating non-spatial differences, like distinctive metrics and processes (Chandraserkaran et al., 2012). Thus, structural differentiation is supportive to an ambidextrous strategy to maintain multiple competencies
(e.g. Boumgarden et al., 2012; Chandrasekaran et al., 2012; Jansen et al. 2009; Raisch, Birkinshaw, Probst & Tushman, 2009).
Data collection and Methodology
The goal of any science is to develop understanding, theories and cumulative knowledge (Hunter and Schmidt, 2004). The aim of this study is to integrate the findings of previous research and reveal patterns of relationships. Hence, this study is conducted due to the discrepancy in the academic literature that is caused by methodological differences between studies. This can be solved by performing a meta-analysis (Hunter and Schmidt, 2004). A meta-analysis will provide a thorough review of current academic literature in the field of organizational antecedents of ambidexterity. According to Hunter and Schmidt (2004) there is no need for additional empirical data, but for making sense of research literature. The meta-analysis of this study will be based on researches that explicitly focus on ambidexterity at firm-level to scrutinize how ambidexterity should be managed in organizations. There are two types of meta-analyses. First, a meta-analysis can be conducted by focusing on a relationship between two variables or a change in one variable across different groups of respondents. Second, a meta-analysis that scrutinizes a large number of meta-factors related to one main concept (e.g. Song et al., 2008). This study performs the second type, because current literature shows discord upon the organizational antecedents (meta-factors) of ambidexterity. Studies were collected that explicitly examined these antecedents.
Theory of data is the basis for conducting a meta-analysis. Understanding this theory of data is needed in order to recognize how data is affected by factors. In turn this is needed to be able to correct their effects. These factors may include sampling error, measurement error, biased sampling, dichotomizations, data errors and other causal factors that cause deficiencies in data. A meta-analysis would be most easily to conduct if these factors were homogeneous across studies, because calculations would include simple averages and simple variances (Hunter and Schmidt, 2004). However, this is not the case. Therefore it is important to select studies as input for this analysis and to follow a protocol to acquire relevant results.
Articles Selection
Current studies were used as input for a meta-analysis. Therefore, data collection consists of systematically reviewing articles in the ambidexterity field. Primary studies described the concept of ambidexterity as ambivalent. Hence, a global description of ambidexterity was taken to select articles for the meta-analysis. As described before, this global description includes the ability of an organization to simultaneously exploit paradoxical strategies regarding innovations. Examples of these paradoxical strategies are exploration and exploitation, initiation and implementation, and incremental and radical innovation. These tensions are derived from literature on ambidexterity. Another important selection criterion was the research level of the articles. The article should examine ambidexterity on
firm-level because this study provides insight on how organizations should manage ambidexterity. The articles were searched for using the Business Source Premier (BSP) with peer reviewed journals dating back to 1965. This precludes missing articles in the field of ambidexterity, because the concept is founded in 1976 (Duncan). The final selection criterion was the inclusion of a correlation matrix in the article, because they serve as main input for a meta-analysis. It should be noted that tensions should be measured as separate dependent variables in order to be able to draw relevant conclusion from the meta-analysis about the concept of ambidexterity. To represent the tensions of ambidexterity, the following keywords were used: ambidexterity, exploration with exploitation, incremental with radical,
initiation with implementation, evolutionary with revolutionary, transformational with modular,
transformational with modularity, transformational with incremental, discontinuous with continuous,
breakthrough with derivative, disruptive with sustaining and framebreaking with incremental. The articles were subjected to the selection criterion of firm-levelresearch, a significant reflection of
ambidexterity, and the publication of a correlation matrix. No limitations like quality of the article were made in order to conduct a proper meta-analysis as described by Hunter and Schmidt (2004). Appendix 1 shows the selected articles per keyword.
After articles were collected from BSP, the selected articles were cross-referenced and resulting articles were subjugated to the selection criteria. This process was repeated until no new articles were found. In total, 62 articles were collected that met the selection criteria of representing the correct level of analysis and a significant reflection of ambidexterity. 16 of the 62 articles included a correlation matrix that could be used as input for the meta-analysis. It should be noted that every article and every database was only used once. This means that a random choice was made between articles that cover the same database. Next, the studies were analyzed through identifying the concepts, independent variables, dependent variables and moderators. This is outlined in appendix 2 where an overview is given of the articles used as input for the meta-analysis by methodological characteristics. The independent variables were combined and labeled to make the data more transparent and interpretable. The judgmentcriteria for combining independent variables were based on the definition and measurement of the independent variable. Additionally, dependent variables were also combined based on the definition and notation in the correlation matrix. The latter is most important in order to be able to conduct a proper meta-analysis that meets the global description of ambidexterity. This means that the independent variable should be tested separately on each paradoxical tension in order to examine the relationship. This resulted in two dependent variables, namely exploration and exploitation. Appendix 3 shows the aggregation of the independent variables into labels. Appendix 4 lists the journals from which the 16 articles originate.
Protocol for Meta-analysis
This paper used the protocol of Song et al. (2008) which is again based on Hunter and Schmidt’s (1990) protocol. These protocols allow making comparisons between different studies. To be able to make this comparison it is important that the relation between two variables is independent of other variables (Hunter and Schmidt, 1990; Song et al., 2008). Therefore, Pearson correlations were used
as input for the meta-analysis. When more than one independent variable was included in the same meta-factor by the same study it was decided to average or take the most representative one of the variables and Cronbach’s alphas. Furthermore these protocols make use of random effect models which is appropriate for this study given the global description of ambidexterity. Random effect models take the possibility of varying parameters between studies into account in contrary of fixed effect models which assume that the same correlation is equal in all studies of the meta-analysis.
As mentioned before, Hunter and Schmidt (2004) state that no single study can provide an optimal basis for scientific conclusions, because every study contains methodological limitations. They call this the myth of the perfect study. Therefore, meta-factors need to be corrected for artifacts. This is the next step in the protocol, following Hunter and Schmidt (1990). These artifacts are dichotomization, sampling error and measurement error. A description and calculation of these corrections is given below.
First, meta-factors were corrected for dichotomization which is the point biserial correlation for the dichotomized variable will be less than the real correlation. This reduction is set on 0.8, because dichotomizationreduces the real correlation by at least 0.8 (Hunter and Schmidt, 1990, 2004; Song et al., 2008). This results in the following formula:
Where ad is the correction for dichotomization, this is 0.8 when the variable is dichotomized and 1 if this is not the case; and is the observed correlation of the primary study i.
Second, sampling error affects the correlation additively and nonsystematically (Hunter and Schmidt, 2004). The actual correlation was weighted by the sample size. Thus, the correction for sampling error is as follows:
Where Ni is the sample size of the primary studyi.
Third, error of measurement has a systematic multiplicative effect on the correlation (Hunter and Schmidt, 2004). The actual correlation was divided by the product of the square root of the reliability of the independent variables and the square root of the reliability of the dependent variables. Hence, the real population correlation is as follows:
Where is the multiple reliability correction factor; is the average of the square roots of reliabilities of independent variable; and is the average of the square roots of reliabilities of
dependent variables. Cronbach’s alphas were used for the reliabilities. However, reliabilities are not always publicized in the articles, in that case averages of reliabilities were taken.
The next step in the protocol is to determine whether an independent variable is an antecedent of ambidexterity. Variance-based tests were used because other tests, like the chi-square test, are biased due to uncorrected artifacts (Hunter and Schmidt, 1990, 2004; Song et al., 2008). The variance based test calculates the total variance (Vartotal) through the variance due to artifacts (Varartif), variance due to sampling error (Vars.e), and the real variance due to heterogeneity (Varreal).
This gives the following formula:
The total variance can also be calculated using the formula:
Where is the observed correlation of the primary study i; Ni is the sample size of the primary study ᵢ ; and is the weighted average of the observed correlations of the primary studies. The latter can be calculated by weighting the observed correlation by the sample size. This results in the following formula:
Second, variance due to artifacts like dichotomization and measurement errors can be calculated by dividing the variances of the reliability correction factor by the average of the square roots of reliabilities (for both independent and dependent variables). This is multiplied by the squared weighted average of the actual correlations. Thus the formula for variance due to artifacts is:
Where is the variance of the square roots of the reliabilities of the independent variable; is the variance of the square roots of the reliabilities of the dependent variable; is the average of the square roots of reliabilities of independent variables; and is the average of the square roots of reliabilities of dependent variables.
Third, the variance due to sampling error was calculated by using the following formula:
Where is the weighted average of the observed correlations of the primary studies; is the average samples size of the primary studies; di is the ith study with a dichotomized variable; and ad is the correction for dichotomization; this is 0.8 when the variable is dichotomized and 1 if this is not the case. Only the first part of the formula is relevant, because no dichotomization was observed.
Finally, the real variance of the population correlation can be calculated by subtracting the variance due to artifacts and variance due to sampling error from the total variance. This results in the following formula:
The independent variable is homogeneous if the real variance is no more than 25 percent of the total variance (Hunter and Schmidt, 1990, 2004; Song et al., 2008). Otherwise, the independent variable is heterogeneous and a moderator analysis is required.
Two significance tests were conducted for the homogeneous variables in order to determine whether the observed value differs enough from the actual value. First, the confidence interval is 95%. The confidence interval must not include zero in order to determine the nature of the relationship between the meta-factor and the dependent variable.The following formula was applied:
Second, the p-value for the real correlation was calculated to estimate the degree of significance. The upper boundary of the confidence interval was calculated by adding the population correlation by the 95% confidence interval of the real population correlation. The same calculation was applied to compute the lower boundary only now the population correlation was subtracted.
Furthermore the explained variance was computed to determine whether the independent variable is explained by moderators. If the explained variance was lower than 75% it means that the relation between the dependent variable and independent variable cannot be explained solely by the independent variable itself. Hence, a moderator analysis is required.
Thus, moderator analysis was conducted for heterogeneous meta-factors. In order to conduct this analysis subgroups were made based on methodological characteristics, outlined in appendix 2. For each subgroup a meta-analysis was conducted to find homogenous meta-factors.
Finally, a file drawer analysis was executed to assess publication bias. This test estimates the number of unlocated studies averaging null results that are required to make the total significance of a meta-factor exceed the p-value of .05, called Xs (Hunter & Schmidt, 2004; Rosenthal, 1979; Song et al., 2008). To calculate the Xs, the p-value for each of the effect sizes is converted into its corresponding Z-value using ordinary normal curve tables. The test is directional (one-tailed). The formula is as follows:
Solving for Xs,
Where is the number of studies that build a meta-factor; and is the average Z-value from independent studies. The higher Xs, the more reliable the p-value is. When Xs is 0 then the meta-factor is already insignificant based on the p-value.
Analysis and Results
Organizational Antecedents of Ambidexterity
The meta-analysis showed sixteen meta-factors related to the ambidexterity concept. Eliminating the meta-factors that are represented by a single study, resulted in a final of eleven meta-factors.Hence, cultural connectedness, managerial tenure, risk, strategic leadership and structural differentiation were not included in the meta-analysis. The definition of each meta-factor is given in table 1.
Table 2 (a and b) shows the results of the meta-analysis. A separation between the two different dependent variables exploration and exploitation was made to meet the global description of ambidexterity. This means that in order to show the effect of the independent variables on the paradoxical tensions of ambidexterity, the relationships between the meta-factors and the dependent variable are illustrated per dependent variable. Table 2a shows the effect of the meta-factors on exploitative innovations and table 2b on exploratory innovations.
Table 1. Definitions of 11 meta-factors
Meta-factors Definitions Selected reference
Centralization The locus of authority and decision making and refers to the extent to which decision making is concentrated in an organization
Jansen, Van Den Bosch and Volberda (2006)
Connectedness (structural) The extent to which employees were networked to various hierarchical levels in their organizational unit
Jansen, Van Den Bosch and Volberda (2006)
Firm age The period an organization exists in the market
Koberg, Detienne and Heppard (2003)
Firm size The magnitude of the
organization
Koberg, Detienne and Heppard (2003)
Formalization The degree to which rules, procedures, instructions, and communications are formalized or written down
Jansen, Van Den Bosch and Volberda (2006)
Learning The cyclic process of the generation of new distinctions and ideas, the distribution and interpretation of these ideas, and organization’s risk-taking action
Morgan and Berthon (2008)
Market orientation Market information acquisition, distribution, interpretation and response by the organization.
Morgan and Berthon (2008)
Output control Desired results; measures how well output aligns with the set standards, and provide respective rewards and punishment for success and failure in goal attainment
Cardinal (2001)
Resources* Representation of an
organization’s resources and potential
Damanpour (1991)
Structural complexity Functional and occupational differentiation
Ettlie, Bridges and O’Keefe (1984)
Structural control Regulates activities and behaviours and is most often implemented in the form of rules and procedures
Cardinal (2001)
* This factor is denoted as technical resources in Damanpour (1991)
The tables present total N, the sum of the sample sizes; K, the number of correlations that build a meta-factor; and ρ, the real population correlation. The confidence interval of the real population correlation is set on 95% and the explained variance is 75%. The confidence interval should not include zero and the explained variance needs to be above 75% in order to consider the meta-factor an antecedent of the dependent variable. As stated before, the explained variance is based on the real variance. However, the real variance can be negative as a result of sampling error. When this occurs, it is recommended to substitute zero for the negative variance (Cronbach, Gleser, Nanda & Rajaratnam, 1972; Brennan, 1983; Hunter & Schmidt, 2004). Xs is the critical number of null-results studies.
Table 2a. Results of the Meta-analysis of exploitative innovations¹ Meta-factor Total N K ρ 95% Confidence Interval Explained Variance (%)² Moderators Xs Centralization 579 4 -0.10 (-0.31,0.11) 38 Yes 0 Connectedness (structural) 1008 6 0.17*** (-0.18,0.51) 16 Yes 196
Firm Age 945 5 0.08** (-0.09,0.25) 41 Yes 45
Firm Size 8998 14 -0.17 (-1.00,0.68) 1 Yes 0 Formalization 825 5 0.32*** (0.16,0.48) 44 Yes 136 Learning 373 2 0.40*** (0.12,0.68) 20 Yes 20 Market Orientation 387 2 0.35*** (0.30,0.39) 91 20 Output Control 113 2 0.19** 100³ 6 Resources 5875 8 -0.23 (-0.81,0.35) 2 Yes 0 Structural Complexity 3389 3 -0.34 (-0.60,-0.08) 7 Yes 0
Structural Control 113 2 0.14* (-0.15,0.43) 44 Yes 2 ¹ For all p-values, one-tailed test statistic; direction depends on the sign of ρ.
² Explained variance lower than 75% means that the meta-factor has moderators.
³ Negative real variance denoted as zero. * p< 0.05
** p< 0.01 *** p< 0.001
The results in table 2a reveal that market orientation (ρ= 0.35, p<.001) and output control (ρ= 0.19, p<.01) are antecedents of exploitation. Table 2b shows that learning (ρ= 0.40, p<.001) and market orientation (ρ= 0.35, p<.001) are antecedents of exploration. This means that these meta-factors are homogeneous, positive significant and directly correlated to exploration or exploitation without the involvement of moderators. Additionally, the results in table 2b suggest that firm age and structural control are insignificant to exploration.
Table 2b. Results of the Meta-analysis of exploratory innovations¹ Meta-factor Total N K ρ 95% Confidence Interval Explained Variance (%)² Moderators Xs Centralization 579 4 -0.47 (-1.00,0.21) 5 Yes 0 Connectedness (structural) 1008 6 0.16** (-0.07,0.38) 31 Yes 66 Firm Age 945 5 -0.05 100³ 0
Firm Size 8998 14 -0.28 (-1.00,0.76) 1 Yes 0 Formalization 825 5 0.12* (-0.19,0.43) 19 Yes 20
Learning 373 2 0.40*** 100³ 20
Market Orientation
387 2 0.35*** 100³ 20
Output Control 113 2 0.43*** (0.21,0.65) 53 Yes 20 Resources 5875 8 0.41*** (0.34,0.48) 58 Yes 352 Structural
Complexity
3389 3 -0.26 (-.50,-0.03) 6 Yes 0
Structural Control 113 2 0.12 100³ 0
¹ For all p-values, one-tailed test statistic; direction depends on the sign of ρ.
² Explained variance lower than 75% means that the meta-factor has moderators.
³ Negative real variance denoted as zero. * p< 0.05
** p< 0.01 *** p< 0.001
Moderators
Table 2a reveals that nine of the eleven meta-factors had heterogeneous correlations. In table 2b this countsfor seven of the eleven meta-factors. Hence, a moderator analysis was conducted in order to find homogeneous meta-factors. A separate meta-analysis was conducted for each subgroup of methodological characteristic (appendix 2). Table 3 (a and b) outlines the results of the moderator analysis. The table presents ρ, the real population correlation; total N, the sum of the sample sizes; K, the number of correlations that build a meta-factor; confidence interval of the real population correlation, which is 95%; explained variance must be more than 75% in order to be a homogeneous meta-factor; and Xs, the critical number of null-results studies. The meta-factor should have at least two subgroups with each a minimum of two studies. The meta-factor is not included in the table when confidence intervals overlap between the two subgroups.
Table 3a. Suggested Moderators for Exploitation¹
Meta-factor Moderator ρ Total
N K 95% Confidence Interval Explained Variance (%)² Xs Formalization 0.32*** 825 5 (0.16,0.48) 44 136 Sample type CEOs 0.48*** 269 2 100³ 48 BU heads 0.50*** 273 3 100³ 21 Resources -0.23 5875 8 (-0.81,0.35) 2 0 Country U.S. -0.001*** 1374 5 100³ 136 Asia 0.024 4501 3 (-0.72,-0.12) 3 0
¹ For all p-values, one-tailed test statistic; direction depends on the sign of ρ.
² Explained variance lower than 75% means that the meta-factor has moderators.
³ Negative real variance denoted as zero.
* p< 0.05 ** p< 0.01 *** p< 0.001
Table 3b. Suggested Moderators for Exploration¹
Meta-factor Moderator ρ Total
N K 95% Confidence Interval Explained Variance (%)² Xs Formalization -0.28 825 5 (-0.19,0.43) 19 20 Sample type CEOs 0.39 269 2 100³ 21 BU heads -0.009*** 273 3 (-0.15.0.13) 51 0 ¹ For all p-values, one-tailed test statistic; direction depends on the sign of ρ.
² Explained variance lower than 75% means that the meta-factor has moderators.
³ Negative real variance denoted as zero. * p< 0.05
** p< 0.01 *** p< 0.001
Table 3a shows that two of the nine heterogeneous meta-factors related to exploitation had moderator subgroups, namely formalization and resources. The relationship between formalization and exploitation depends on the sample type. Formalization has a slightly stronger effect on exploitation when business unit heads (BU heads) were used as a sample compared to samples with Chief Executive Officers (CEOs) as respondents. The relationship between resources and exploitation depends on the country that was selected for the sample. Resources were negatively associated with exploitation for firms in the U.S. However, for firms in Asia, resources were heterogeneously related to exploitation. No moderators were found for the remaining seven heterogeneous meta-factors related
to exploitation: centralization, structural connectedness, firm age, firm size, learning, structural complexity, and structural control. Of these seven meta-factors, learning, structural complexity and structural control haveless than two studies for each subgroup. Additionally, centralization has only one subgroup with more than two studies, when looking at difference in measurement. For both cases, further research is required to judge the validity of these potential moderators. Furthermore, structural connectedness, firm age and firm size showed no methodological moderators.
Table 3b shows that only formalization of the sevenheterogeneous meta-factors related to exploration had moderator subgroups. This relationship is similar to the relationship between formalization and exploitation. However, formalization was insignificantly associated with exploration when CEOs were selected as sample. For samples with BU heads, formalization was heterogeneously related to exploration. No moderators were found for the remaining six heterogeneous meta-factors related to exploration: centralization, structural connectedness, firm size, output control, resources and structural complexity. Similar to exploitation, structural complexity has less than two studies for each subgroup and centralization has only one subgroup with more than two studies regarding difference in measurement. As stated before, these cases give directions for future research. Finally, structural connectedness and firm size showed no methodological moderators.
Conclusions
To summarize the findings of the conducted meta-analysis, a distinction is made between insignificant and significant homogeneous meta-factors, and the heterogeneous meta-factors with and without moderators. Again, a separation between the organizational antecedents of exploitation and exploration is made to keep overview.
The results show that only two of the eleven meta-factors are homogeneous and significant related to exploitation and exploration. Of these eleven meta-factors, two were homogeneous but not significant to exploration. Two meta-factors related to exploitation are moderated. One meta-factor’s sample should not originate from the U.S. and another meta-factor should select business units heads as respondents. One meta-factor is an antecedent of exploration but the relationship is insignificant for a sample consisting of CEOs. After conducting a methodological moderator analysis, still sevenmeta-factors remain heterogeneous related to exploitation and six meta-factors remain heterogeneous related to exploration. For these factors further research is required.
Implications
Looking at the antecedents of exploitation, it can be concluded that formalization, market orientation and output control enhance exploitative innovations. However, formalization is slightly stronger related to exploitation for business unit’s heads. Evidently, exploitative innovations are enhanced when business unit heads pursue a formalized structure in the organization. Additionally, organizations
that control their output positively affect their exploitative innovations due to the focus on efficiency of their operations.
The analysis also found that resourcesare detrimental for exploitative innovations. The study showed that resources have a negative influence on exploitative innovations for firms in the U.S. However, no positive moderating effect of other countries on this relationship is found.Apparently, firms in the U.S. allocate their resources rather to radical innovations although no positive effect has been found between country and resources associated with exploration.
The amount of meta-factors of exploitation that are heterogeneous is significantly high, namely nine out of eleven meta-factors showed a heterogeneous relationship with exploitative innovations. It can be concluded that exploitative innovations are too heterogeneous to examine their antecedents. The moderator analysis showed no methodological moderators for seven of nineheterogeneous meta-factors. This might indicate that other moderators exists, which are not reported in current academic literature (e.g. leadership style and position of the organization in the market).
The examination of the relationship between the meta-factors and exploration resulted in two antecedents of exploratory innovations: learning and market orientation. Hence, collecting market information seems to enhance radical innovations. Both firm age and structural control were insignificant to exploration at the 0.05 level. Apparently, structural rigidity due to structural control or firm’s age does not hamper the development of radical innovations. This is also shown in the relationship between formalization and exploration when CEOs are selected as sample. However, no significant effect of other sample types on this relationship is found. Evidently, CEOs do not impede therelationship between formalization and explorative innovations.
A smaller amount of meta-factors is heterogeneous related to exploration compared to exploitation, namely seven. However, this number is still more than half of the identified antecedents of exploration. Hence, the same conclusion of exploitative innovations being too heterogeneous to examine can be made for exploratory innovations. Again, other moderators might exist that are not yet reported by published researches.
To comply with the concept of ambidexterity, both outcomes of exploitation’s and exploration’s antecedents are compared. Although a significant amount of research is performed in this field, the results are not conclusive. The results show that two of the eleven meta-factors are both related to exploitation as well as exploration of which one is moderated. Market orientation enhances incremental and radical innovations. Hence, organizations need to understand customers’ demands and behaviours in order to be ambidextrous. It should be noted that this finding might have been caused by differences in defining market orientation, since market orientation is often divided into reactive market orientation and proactive market orientation. However, not all studies included in this meta-analysis made this distinction and described market orientation solely as the ability of organizations of understanding only customers’ current needs. A moderator analysis of meta-factor measurement could not be conducted due to insufficient amount of studies to give reliable results.
Another meta-factor that effects both exploitation and exploration is formalization. BU heads and CEOs invigorate the positive relationship between formalization and exploitation. This relationship is slightly stronger for BU heads. However, formalization is heterogeneously and insignificantly related to exploration when respectively BU heads and CEOs are selected as sample. Evidently, BU heads and CEOs that employ a formalized structure foster incremental innovations, possibly due to strict control and routines. However, CEOs do not hamper the development of radical innovations when employing a formalized structure. Organizations may need to consider formalizing their activities to be ambidextrous. Nevertheless, future research should investigate alternative sample types and their influence on the relationship between formalization and ambidexterity.
This study identifies four heterogeneous meta-factors for both exploration and exploitation. Hence, centralization, structural connectedness, firm size and structural complexity are heterogeneous related to ambidexterity. No methodological moderators were found and one might state that other unpublished moderators might exist. Furthermore, firm age and structural control are insignificant associated with exploration. These meta-factors are both heterogeneous antecedents of exploitation which might suggest that the heterogeneous relation with exploitation also applies to ambidexterity. However, future research should investigate the unreported moderators in current literature that affect this relationship.
Limitations
The use of a database in this study yields only studies that have been published. Hence, it is less likely to yield a representative sample due to publication bias. However, this is precluded in this study by the use of a file drawer analysis. Hunter and Schmidt (2004) argue that the file draweranalysisadopts a fixed-effect model. Hence, this analysis is based on the assumption that the population effect size of every study is homogeneous. The file drawer analysis relies on combined statistics ofk studies. However, this study applies a random effect model that assumes that population parameters vary between studies and thus the number of missing studies Xs is much smaller than the number originated form the file drawer analysis (Hunter & Schmidt, 2004; Iyengar & Greenhouse, 1988). Another point of argue concerning this analysis is that in many research fields it is unlikely to have that many ‘lost’ studies as indicated by the file drawer analysis, like it would take 352 missing studies for resources averaging zero effect size to reduce the observed correlation to a p-value of 5%. Furthermore, Pearson correlations are used to examine a linear relationship between the variables. However, when the correlation was zero, the possibility exists that other types of relationships between the variables are present. Finally, the amount of studies included in this meta-analysis was 16 which exceed the requirement of a minimum sample size of two studies to be able to conduct a meta-analysis (Hunter & Schmidt, 2004). However, in order to increase the reliability of the results, a larger sample size is preferable.
Recommendations for Future Research
A meta-analysis does not only correct studies to provide an integrative view on a certain topic, but also gives recommendations for further research. Hence, this study provides directions for what kind of primary studies are needed next in the ambidexterity domain.
First, by identifying the statistically significant antecedents of ambidexterity, the results show that there is no need for future research to replicate market orientation as significant determinant of ambidexterity. Therefore, market orientation may be considered as a control variable. However, empirical research should consider the measurement of this independent variable and its effect on the tensions underlying ambidexterity, such as reactive and proactive market orientation. Additionally, this study only selected studies that measured the independent variables on both exploitation and exploration. This means that studies are eliminated with ambidexterity as dependent variable. However, future research should include these studies when examining market orientation in relation to ambidexterity, because market orientation has a similar relation to both exploitation and exploration.
Second, this study examined the role of moderators on organizational antecedents of ambidexterity, but a significantly high number of heterogeneous antecedents remained without result. Therefore, future research may include the search for moderators that are not yet published in current academic literature.
Third, the study’s meta-analysis suggests that learning and output control are positively significant related to one part of the paradoxical tension of ambidexterity. A heterogeneous moderated relationship is recognized for the other part of the tension. However, no significant moderators were identified. Hence, it is advisable to conduct more moderator research on learning and output control in relation to exploration and/or exploitation.
Fourth, this study’s framework suggests that BU headsand CEOs that employ standardized and formalized rules and procedures are better able to achieve an ambidextrous strategy. However, in the meta-analysis no evidence was found that BU heads influence the relationship between formalization and exploration whereas CEOs do not impede this relationship. Therefore, the suggestion is given here to further investigate this relationship. Furthermore, it is suggested to experiment with different sample types in future ambidexterity research to investigate the influence of different organizational members on the effect of routines and strict control to achieve an ambidextrous organization.
Finally, this study illustrated the organizational antecedents of ambidexterity identified in current academic research. In this study only antecedents represented by more than one study were included. Elimination of antecedents also occurred when the antecedent did not exceed the minimum of two studies per subgroup for the implementation of a moderator analysis. To capture the concept of a meta-analysis to thoroughly review all current academic literature in a certain field, it is suggested that future studies examine the eliminated antecedents.
List of References
Atuahene-Gima, K. 2005.Solving the capability-rigidity paradox in new product innovation.Journal of Marketing, 69(4): 61-83.
Beckman, C.M. 2006. The influence of founding team company affiliations on firm behavior.
Academy of Management Journal, 49(4): 741-758.
Boumgarden, P., Nickerson, J. & Zenger, T.R. 2012. Sailing into the wind: Exploring the
relationships among ambidexterity, vacillation, and organizational performance. Strategic Management Journal, 33: 587-610.
Brennan, R. L. 1983. Elements of Generalizability Theory.Iowa City: ACTPublications.
Brion, S., Mothe, C. & Sabatier, M. 2010. The Impact of Organisational Context and Competences on Innovation Ambidexterity.International Journal of Innovation Management, 14(2): 151-178. Cabello-Medina, C., Carmona-Lavado, A. & Cuevas-Rodrigues, G. 2011.Organisation of R&D
departments as determinant of innovation: testing alternative explanations. Technology Analysis & Strategic Management, 23(4): 383-400.
Cardinal, L.B. 2001. Technological Innovation in the Pharmaceutical Industry: The Use of
Organizational Control in Managing Research and Development.Organization Science, 12(1): 19-36.
Chandrasekaran, A., Linderman, K. & Schroeder, R., 2012. Antecedents to Ambidexterity Competency in High Technology Organizations. Journal of Operations Management, 30(1/2): 134-151. Chang, Y., Hughes, M. &Hotho, S. 2011. Internal and External Antecedents of SMEs’ Innovation
Ambidexterity Outcomes.Management Decision, 49(10): 1658-1676.
Chang, Y. & Hughes, M. 2012.Drivers of Innovation Ambidexterity in Small- to Medium-sized Firms.
European Management Journal, 30(1): 1-17.
Cronbach, L. J., Gleser, G. C., Nanda, H., &Rajaratnam, N. 1972.The Dependability of Behavioral Measurements: Theory of Generalizability for Scores and Profiles. New York: Wiley. Damanpour, F. 1991. Organizational Innovation: A Meta-analysis of Effects of Determinants and
Moderators.Academy of Management Journal, 34(3): 555-590.
Damanpour, F. 1996. Organizational Complexity and Innovation: Developing and Testing Multiple Contingency Models.Management Science, 42(5): 693-716.
Damanpour, F. & Schneider, M. 2006. Phases of the Adoption of Innovation in Organizations: Effects of Environment, Organization and Top Managers. British Journal of Management, 17(3): 215-236.
Duncan, R.B. (1976). The ambidextrous organization: Designing dual structures for innovation, in R. Kilman& L. Pondy (eds.)The Management of Organizational Design. New York: North Holland: 167-188.
Ettlie, J.E., Bridges, W.P. & O’Keefe, R.D. 1984. Organization Strategy and Structural Differences for Radical versus Incremental Innovation.Management Science, 30(6): 682-695.
Review, 35(1): 202-225.
Germain, R. 1996. The Role of Context and Structure in Radical and Incremental Logistics Innovation Adoption.Journal of Business Research, 35(2): 117-127.
Gibson, C.B. &Birskinshaw, J. 2004. The Antecedents, Consequences, and Mediating Role of Organizational Ambidexterity.Academy of Management Journal, 47(2): 209-226.
Greve, H.R. 2007. Exploration and Exploitation in Product Innovation.Industrial & Corporate Change, 16(5): 945-975.
Grover, V. & Goslar, M.D. 1993. The Initiation, Adoption, and Implementation of Telecommunications Technologies in U.S. Organizations.Journal of Management Information Systems, 10(1): 141-163.
Gupta, A.K., Smith, K.G. &Shalley, C.F. 2006.The interplay between exploration and exploitation.
Academy of Management Journal, 49(4): 693-706.
He Z., & Wong, P. 2004. Exploration vs. exploitation: An empirical test of the ambidexterity hypothesis. Organization Science, 15(4): 481-494
Hunter, J. E., & Schmidt, F. L. 1990. Methods of Meta-Analysis: Correcting Error and Bias in Research Findings (1sted.). Newbury Park, CA: Sage Publications Inc.
Hunter, J.E. & Schmidt, F.L. 2004. Methods of Meta-Analysis: Correcting Error and Bias in Research Findings (2nded.). Newbury Park, California: Sage Publications Inc.
Im, G. &Rai, A. 2008. Knowledge sharing ambidexterity in long-term interorganizational relationships. Management Science, 54(7): 1281-1296.
Ireland, D. & Webb, J.W. 2007. Strategic entrepreneurship: Creating competitive advantage through streams of innovation. Business Horizons, 50(1): 49-59.
Iyengar, S., & Greenhouse, J. 1988.Selection models and the file drawerproblem.Statistical Science,
3:109–135.
Jansen, J.J.P., Bosch, F.A.J. van den & Volberda, H.W. 2005. Exploratory Innovation, Exploitative Innovation, and Ambidexterity: The Impact of Environmental and Organizational Antecedents.
Schmalenbach Business Review, 57(4): 351-363.
Jansen, J.J.P., Bosch, F.A.J. van den & Volberda, H.W. 2006. Exploratory Innovation, Exploitative Innovation, and Performance: Effects of Organizational Antecedents and Environmental Moderators. Management Science, 52(11): 1661-1674.
Jansen, J.J.P., Tempelaar, M.P., Bosch, F.A.J. van den & Volberda, H.W. 2009. Structural
Differentiation and Ambidexterity: The Mediating Role of Integration Mechanisms.
Organization Science, 20(4): 797-811.
Koberg, C.S., Detienne, D.R. &Heppard, K.A. 2003. An empirical test of environmental,
organizational, and process factors affecting incremental and radical innovation. Journal of High Technology Management Research, 14(1): 21-45.
Leonard-Barton, D. 1992. Core capabilities and core rigidities: A paradox in managing new Product Development.Strategic Management Journal, 13 (Summer): 111–25.
Levinthal, D. A. and March, J. G. 1993. The myopia of learning.Strategic Management Journal, 14(Special Issue): 95–112.
Li, C., Lin, C. & Chu, C. 2008.The Nature of Market Orientation and the Ambidexterity of Innovations.
Management Decision, 46(7): 1002-1026.
Lin, H. & McDonough, E.F. 2011. Investigating the Role of Leadership and Organizational Culture in Fostering Innovation Ambidexterity.IEEE Transactions on Engineering Management, 58(3): 497-509.
March, J.G. 1991. Exploration and exploitation in organizational learning.Organization Science, 2(1): 71-87.
Matta, V., Koonce, D. &Jeyaraj, A. 2012. Initiation, Experimentation, Implementation of innovations: The case for Radio Frequency Identification Systems.International Journal of Information Management, 32(2): 164-174.
McCarthy, I.P. & Gordon, B.R. 2011. Achieving contextual ambidexterity in R&D organizations: A management control system approach. R&D Management, 41(3): 240-258.
Menguc, B. &Auh, S. 2010. Development and Return on Execution of Product Innovation Capabilities: The Role of Organizational Structure. Industrial Marketing Management, 39(5): 820-831.
Morgan, R.E. &Berthon, P. 2008. Market Orientation, Generative Learning, Innovation Strategy and Business Performance Inter-Relationships in Bioscience Firms.Journal of Management Studies, 45(8): 1329-1353.
Pierce, J.L. & Delbecq, A.L. 1977. Organization structure, individual attitudes and innovation.
Academy of Management Review, 2(1): 27-37.
Quintana-Garcia, C. & Benavides-Velasco, C.A. 2007. Innovative Competence, Exploration and Exploitation: The Influence of Technological Diversification. Research Policy, 37(3): 492-507. Raisch, S. &Birkinshaw, J. 2008. Organizational ambidexterity: Antecedents, outcomes, and
moderators. Journal of Management, 34(3): 375-409.
Raisch, S., Birkinshaw, J.,Probst, G. &Tushman, M.L. 2009. Organizational ambidexterity:
Balancing exploitation and exploration for sustained performance.Organization Science, 20 (4): 685-695.
Rosenthal, R. 1979. The “file drawer problem” and tolerance for null results. Psychological Bulletin,
86: 638–641.
Sidhu, J.S., Commandeur, H.R. &Volberda, H.W. 2007. The multifaceted nature of exploration and exploitation: Value of supply, demand, and spatial search for innovation. Organization Science, 18(1): 20-38.
Song, M., Podoynitsyna, K., Van der Bij, H. & Halman, J.I.M. 2008. Success factors in new ventures: A meta-analysis. Product Innovation Management, 25: 7-27.
Subramaniam, M. &Youndt, M.A. 2005. The Influence of Intellectual Capital on the Types of Innovative Capabilities.Academy of Management Journal, 48(3): 450-463.
revolutionary change. California Management Review, 38(4): 8-30.
Vanhaverbeke, W., Gilsing, V., Beerkens, B. & Duysters, G. 2009. The Role of Alliance Network Redundancy in the Creation of Core and Non-core Technologies.Journal of Management Studies, 46(2): 215-244.
Visser, M. de, Weerd-Nederhof, P. de, Faems, D., Song, M., Looy, B. van & Visscher, K. 2010. Structural ambidexterity in NPD processes: A firm-level assessment of the impact of differentiated structures on innovation performance. Technovation, 30(5/6): 291-299. Walker, R.M. 2008. An Empirical Evaluation of Innovation Types and Organizational and
Environmental Characteristics: Towards a Configuration Framework. Journal of Public Administration Research & Theory, 18(4): 591-615.
Wei, Z., Yi, Y. & Yuan, C. 2011. Bottom-up Learning, Organizational Formalization, and Ambidextrous Innovation.Journal of Organizational Change Management, 24(3): 314-329.
Appendix 1.Amount of selected articles per criteria
Key words Results peer
reviewed
Results firm-level and reflection ambidexterity
Results correlation matrix*
Ambidexterity 104 24 3
Exploration & exploitation 476 27 10
Incremental & radical 318 4 3
Initiation & implementation 225 3 0 Evolutionary & revolutionary 173 2 0
Transformational & modular 1 0 0
Transformation & modularity 2 0 0
Transformational & incremental 26 0 0
Discontinuous & continuous 400 2 0
Breakthrough & derivative 31 0 0
Disruptive & sustaining 25 0 0
Framebreaking& incremental 30 0 0
Total 1,811 62 16
