European Management Journal 37 (2019) 589e600
Contents lists available at ScienceDirect
European Management Journal journal homepage: www.elsevier.com/locate/emj
Top Managers' cognition facilitates organisational ambidexterity: The mediating role of cognitive processes Rafael Wilms a, *, Lothar A. Winnen b, Ralf Lanwehr a a b
South Westphalia University of Applied Sciences, Germany TH Mittelhessen University of Applied Sciences, Germany
a r t i c l e i n f o
a b s t r a c t
Article history: Received 15 September 2016 Received in revised form 26 January 2019 Accepted 28 March 2019 Available online 6 April 2019
It is empirically founded and tested that top management team group processes facilitate ambidexterity, whereas the role of top managers' cognition has remained empirically largely neglected. This study contributes to the literature of top managers' cognition by theoretically developing the relationship between cognitive frames and organisational ambidexterity, and the mediating process of cognitive differentiation and integration. The study empirically tests the proposed model in cross-sectional design by employing a sample of 101 top managers, using partial least squares structural equation modelling. The study shows that top managers' paradoxical frames encourage the engagement in cognitive differentiation and integration, whereas top managers' paradoxical frames are not significantly related to organisational ambidexterity. Solely appreciating exploitation and exploration does not foster ambidexterity, but is an antecedent for top managers' cognitive differentiation and integration. Top managers' cognitive differentiation and integration mediate the relationship between top managers' paradoxical frames and organisational ambidexterity. Additionally, top managers' cognitive integration moderates the effect between top managers' cognitive differentiation and organisational ambidexterity, which stresses the pivotal role of top managers' cognitive integration. © 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Keywords: Organisational Ambidexterity Cognitive Frames Integrative Complexity Top Manager
1. Introduction Recent research has revealed that ambidextrous organisations which are “capable of exploiting existing competencies as well as exploring new opportunities with equal dexterity” (Lubatkin, Simsek, Ling, & Veiga, 2006, p. 647) outperform others (He & Wong, 2004; Junni, Sarala, Taras, & Tarba, 2013). The reason is that these firms focus on the here and now and on the future either at the same time or over time. Exploitation facilitates incremental service or product developments and creates predictable revenues. Even with constant incremental innovations, these services or products might become outdated and obsolete in the long term. In contrast, exploration fosters radical innovations and therefore might create new opportunities to generate income (March, 1991). Unfortunately, exploration has unpredictable and uncertain results. Too much exploration threatens the financial viability of the firm. Ambidextrous organisations thereby engage in mutually reinforcing activities and balance the trade-offs between short-term revenues generated through current markets and existing products, and
* Corresponding author. Lindenstraße 53, 59872, Meschede, Germany. E-mail address:
[email protected] (R. Wilms).
long-term income generated through new opportunities, markets and products. These organisations are especially successful in dynamic or highly volatile markets (Jansen, van den Bosch, & Volberda, 2006; Junni et al., 2013) in which product life cycles are tremendously short. Nonetheless, executing exploitative and explorative activities either at the same time or over time creates tensions throughout the organisation (Boumgarden, Nickerson, & Zenger, 2012), since each action requires different organisational designs (Raisch & Birkinshaw, 2008; Tushman & O'Reilly, 1996; Winnen & Wilms, 2014) as well as different types of management and leadership (Benner & Tushman, 2003; Gibson & Birkinshaw, 2004; Jansen, Vera, & Crossan, 2009; Rosing, Frese, & Bausch, 2011; Zacher & Wilden, 2014). This means that top managers not only need to fully understand and appreciate each activity, but also to embrace their connection and manage their integration in a valueenhancing way. Many different studies stress that the top management of a firm has a huge influence on corporate strategy, and thus on the degree of ambidexterity (Bromiley & Rau, 2016; Eggers & Kaplan, 2013; Lubatkin et al., 2006; Smith, 2014; Smith & Tushman, 2005). In their review of ten years of top management research, Bromiley and Rau (2016, p. 174) state “while there has been an explosion of
https://doi.org/10.1016/j.emj.2019.03.006 0263-2373/© 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
590
R. Wilms et al. / European Management Journal 37 (2019) 589e600
research on the influence of CEO personality and [top management team] social processes on strategy process, much remains to be done in terms of examining CEO and [top management team] cognition”. In line with this, the role of top managers' cognition for organisational ambidexterity has only been theoretically proposed. In particular, it is argued that ambidexterity “is rooted in paradoxical cognition e cognitive frames and processes […] allow […] to effectively embrace, rather than avoid, contradictions” (Smith & Tushman, 2005, p. 533). Only recently, and for the first time, has it been shown that cognitive frames are related to organisational ambidexterity at the strategic business unit level in a Taiwanese sample (Lin & McDonough, 2014). Thus, the relationship between top managers' cognition and organisational ambidexterity is empirically sparsely understood. Furthermore, cognitive frames and cognitive processes could play different roles in creating an ambidextrous organisation. Yet, this distinction has not been empirically addressed. Further theoretical conceptualisation and empirical research is needed for a better understanding of the role of top managers' cognition in fostering organisational ambidexterity. The aim of this study is to address these research gaps. By employing a sample of 101 German top managers, the study makes several contributions to theory and practice. First, it empirically validates the relationship between top managers' cognitive frames, top managers' cognitive processes and organisational ambidexterity (see our theoretical model in Fig. 1). Thereby, it makes an important distinction between cognitive frames and cognitive styles, and investigates their different influences. Secondly, it tests the assumption whether cognitive frames play a role in fostering organisational ambidexterity (Lin & McDonough, 2014). Thirdly, the relationship between top managers' cognitive frames and organisational ambidexterity is mediated by top managers' cognitive processes. Finally, this study extends the understanding of which antecedents foster organisational ambidexterity by focusing on top managers' cognitive frames, and top managers' cognitive styles. In the following, the theoretical lens is elaborated by using the literature of managerial cognition. Building upon this, hypotheses are developed for the relationship between cognitive factors and organisational ambidexterity. After that, the empirical approach is depicted and the proposed hypotheses are tested. The article leads to discussion by stressing implications for theory and practice, followed by limitations and suggestions for future research.
2. Theory and conceptual framework 2.1. Cognitive frames A cognitive frame is a “mental template that individuals impose on an information environment to give it form and meaning” (Walsh, 1995, p. 281) and is composed of cognitive content and cognitive structure (Walsh, 1995). Cognitive content “consists of the things [an agent] knows, assumes and believes” (Finkelstein & Hambrick, 1996, p. 57). Cognitive structure refers to “how the content is arranged, connected or studied in the [agent's] mind” (Finkelstein & Hambrick, 1996, p. 57). 2.2. Paradoxical frames facilitate organisational ambidexterity Cognitive frames act as “cognitive filters that admit certain bits of information into the strategizing process while excluding others” (Daft & Weick, 1984; Porac & Thomas, 2002, p. 178). They provide a perspective on how to approach and interpret an issue (Daft & Weick, 1984; Kahneman, 2011; Kahneman & Tversky, 1979; Lin & McDonough, 2014). Gathering information is susceptible to cognitive biases (Kahneman, 2011). For example, people tend to confirm their assumptions or beliefs, which is called confirmation bias (Nickerson, 1998), or to neglect information that causes psychological discomfort, which is called the ostrich effect (Karlsson, Loewenstein, & Seppi, 2009). Hence, people tend to ignore overly discrepant information (Kiesler & Sproull, 1982), and base their judgements and decisions on prior knowledge (Evans, Barston, & Pollard, 1983; West, Toplak, & Stanovich, 2008). It is tremendously difficult to gain a realistic perception of a strategic situation without being biased from the past. Consequently, individuals who adopt a paradoxical frame “recognise and accept the simultaneous existence of contradictory forces” (Smith & Tushman, 2005, p. 526), whereas individuals who adopt a non-paradoxical frame tend to emphasise only one of these contradictory forces. Transferred to the field of ambidexterity, top managers who adopt a paradoxical frame consider both exploitation and exploration as valuable, whereas top managers who adopt a non-paradoxical frame emphasise either exploitation or exploration. In the 1990s, Polaroid, for example, was not able to develop a digital camera that was successful in the market. In the following, we rely on the in-depth case study from Tripsas and Gavetti (2000)
Fig. 1. Theoretical model.
R. Wilms et al. / European Management Journal 37 (2019) 589e600
as the foundation of our arguments. One of the contributing factors in Polaroid's inability to develop a competitive new product was Polaroid's top management team belief you can only make money with the film and not with hardware (Tripsas & Gavetti, 2000, p. 1155). Accordingly, their current business model of producing instant cameras contradicted new market developments of digital cameras. They were not able to adjust to radical market changes. Even though they had the necessary technological knowledge, they had difficulties to use it for new product development and exploit those new products (Tripsas & Gavetti, 2000, p. 1151). In this case, the overemphasis on exploitation probably hindered Polaroid from successfully exploring new ways to make money in a changing market environment. In other words, the non-paradoxical frame they adopted in the past hindered them from equally appreciating exploitation and exploration. This example stresses that it is vital for top managers to adopt a paradoxical frame and to accept that exploration and exploitation are vital in order to create an ambidextrous organisation. Furthermore, Miron-Spektor and colleagues (Miron-Spektor, Ingram, Keller, Smith, & Lewis, 2018) studied the role of paradoxical frames on in-role performance and innovation on the employee level. Individuals who adopt a paradoxical frame show a better inrole performance (similar to exploitation) and a higher level of innovation (similar to exploration). These results support our notion that accepting paradoxical tensions is necessary to engage in exploitation and exploration simultaneously. As a consequence, we propose: H1. Top managers' paradoxical frames facilitate organisational ambidexterity. 2.3. Cognitive differentiation and integration Different thinking styles are related to strategic decisionmaking. Thinking styles describe a person's preferred mode of processing information. “[I]ndividuals rely on a cognitive style to process the information within their environments that have been structured by […] cognitive frames that they have imposed upon it” (Lin & McDonough, 2014, p. 173). Two cognitive processes have been prominent in paradox research, namely cognitive differentiation and integration (Gotsi, Andriopoulos, Lewis, & Ingram, 2010; Miron-Spektor, Gino, & Argote, 2011; Smith, 2014; Smith & Lewis, 2011), formerly referred to as integrative complexity (Tetlock, Peterson, & Berry, 1993). Cognitive differentiation refers to “the ability to perceive several dimensions in a stimulus array” (Bartunek, Gordon, & Weathersby, 1983, p. 274). Cognitive integration refers to “the development of complex connections among the differentiated characteristics” (Bartunek et al., 1983, p. 274). Cognitive differentiation and integration are highly interrelated, since effective integration depends on effective differentiation (Carroll & Bright, 2010). Thus, cognitive differentiation reflects how carefully each aspect of an issue is analysed, meaning that cognitive differentiation increases the number of attributes within a cognitive frame's content. Cognitive integration, in contrast, reflects how carefully a person builds connections between different characteristics or perspectives of an issue. This means that cognitive integration increases the number of linkages within and among different attributes and affects the structure between and within cognitive frames (Hahn, Preuss, Pinkse, & Figge, 2014). 2.4. Paradoxical frames and cognitive differentiation and integration O’Reilly and Tushman (2008, p. 191) argue that “the crucial task
591
[…] is not the simple organisational structural decision in which the exploratory and exploitative subunits are separated, but the processes by which these units are integrated in a value enhancing way”. Regardless that their argument assumes that organisations separate exploitation and exploration, at the heart of their argument is the emphasis that recognising and appreciating contradictory forces is necessary, but not sufficient. It is rather a matter of how these strategic orientations are synergistically intertwined. Inherently, their argument makes a distinction between cognitively framing a strategic orientation to be important or not, and processing deliberatively all the bits of information of each strategic orientation and the interaction among them, creating synergies. Ergo, distinguishing between cognitive frames and cognitive processes might offer added explanatory value. In line with Miron-Spektor, Efrat-Treister, et al. (2011) and Miron-Spektor, Gino, et al. (2011), we propose that paradoxical frames set the problem scope, meaning that a cognitive (or a paradoxical) frame defines the beginning and the end of a problem. Given a top manager adopts a paradoxical frame, a top manager is likely to propose competing and often contradicting ideas, presumably some of exploitative and some of explorative nature. Consequently, tension arises between these ideas, which promotes discussions about differences and similarities between exploitative and explorative strategies (Smith & Tushman, 2005). Empirically, it has been shown that observing inconsistent stimuli, which is likely to occur in a discussion about competing and even contradicting strategies, fosters integrative complexity (Miron-Spektor, EfratTreister, Rafaeli, & Schwarz-Cohen, 2011). Further, Miron-Spektor, Efrat-Treister, et al. (2011) and Miron-Spektor, Gino, et al. (2011) stressed that participants e primed by a paradoxical frame e showed higher integrative complexity in the following task. Thus, top managers that adopt a paradoxical frame are more likely to engage in the process of cognitive differentiation and integration. Consequently, we propose: H2. Top managers' paradoxical frames facilitate top managers' cognitive differentiation. H3. Top managers' paradoxical frames facilitate top managers' cognitive integration.
2.5. Cognitive differentiation and integration and organisational ambidexterity An ambidextrous strategy requires balancing the interests and needs of exploitation and exploration, either simultaneously or over time, while seeking synergies between them (Raisch, Birkinshaw, Probst, & Tushman, 2009; Smith, Besharov, Wessels, & Chertok, 2012). Cognitive differentiation creates abundant information about exploitation and exploration and renders distinctions between them salient. Cognitive differentiation stresses the needs of different strategies in terms of, for example, tasks, people, culture and formal organisation (Smith & Tushman, 2005) or resource allocation, responsibilities, product design and organisational design (Smith, 2014). Subsequently, each strategy can be improved based on the understanding of how each of the strategies is or becomes successful. Furthermore, distinguishing strategic domains from each other supports understanding and recognising the unique contributions of exploitation and exploration to the overall corporate strategy (Smith et al., 2012). In a simplified example, imagine an electric car manufacturer has a strategic advantage because of their battery technology. They are able to store energy more efficiently than their competitors. Electric cars are their main business. This can be understood as an exploitative strategy,
592
R. Wilms et al. / European Management Journal 37 (2019) 589e600
because the company aims at developing the product and the production process incrementally. They increase production capacity in order to decrease costs and increase profit-margin. At the same time, they produce distributed energy products in another business unit, which are basically products that allow the stationary storage of energy. These products are very useful in connection with photovoltaic systems, especially in remote areas with lots of sunshine. This can be understood as an explorative strategy. The company intends to open up new markets in this way. Cognitive differentiation helps to understand where these businesses differ in detail. Designing a battery for mobile or stationary purposes could differ. For example, abrupt vibrations because of potholes may be a problem for electric car batteries, but obviously not for the distributed energy products. So, top managers recognise by processes of cognitive differentiation that each product is exposed to different demands. Therefore, one could use different materials for both products, for example a cheaper less robust metal for the distributed energy products and a more expensive robust metal for the electric car. The lower-cost material makes the battery more affordable. Since prices matter for the costumers of distributed energy products, the product becomes more successful in the newly entered market. This simplified example demonstrates that cognitive differentiation helps organisations to understand the particular requirements and needs of the exploitative and explorative strategies. Thus, we propose: H4. Top managers' cognitive differentiation is positively associated with organisational ambidexterity. O’Reilly and Tushman (2008) argument points out that the core of organisational ambidexterity is not pursuing an exploitation strategy and an exploration strategy at the same time or over time, but the value-enhancing combination of these strategies. Processes of cognitive integration intertwine exploitation and exploration so that the combination of these strategies creates more value than the sum of each strategy alone (Smith et al., 2012). For example, if the electric car manufacturer from above elaborates that their electric cars and their distributed energy products differ in many ways, but use partly the same components for their construction, they can create synergies by developing and producing these shared components within one business unit. They benefit from economies of scales. In this case, top managers' cognitive integration built a connection between the exploitative and explorative strategy, which creates synergies. Of course, leveraging synergies by using cognitive integration is not limited to economies of scale. Keeping it simple, cognitive integration is a process by which top managers understand where the exploitative strategy can benefit from any kind of infrastructure or asset that belongs to the explorative strategy, and vice versa. Thereby, it facilitates identifying synergies between exploitation and exploration (Smith et al., 2012). Accordingly, we propose: H5. Top managers' cognitive integration is positively associated with organisational ambidexterity. Top managers' cognitive differentiation fosters a deep understanding of each domain of the paradoxical strategy (Hahn et al., 2014; Wong, Ormiston, & Tetlock, 2011), enabling the identification of each domain's needs, strengths, weaknesses as well as opportunities and threats. Top managers' cognitive integration, by contrast, builds connections between each domain of the paradoxical strategy, and creates an understanding of how each strategic domain reinforces the other. Each of these activities can be performed independently of the other. However, only when a top
manager engages in both cognitive differentiation and integration to a high degree, can he or she unfold their full contribution to organisational ambidexterity because the top manager is able to intertwine both strategies with each other. A low degree of engagement in cognitive integration, combined with high degree of engagement in cognitive differentiation is likely to result in exploitative and explorative activities, executed in parallel, but without leveraging the synergies between these activities. In contrast, a low degree of engagement in cognitive differentiation, combined with a high degree of engagement in cognitive integration is likely to result in executing either exploitative or explorative activities, but with a deep understanding of how the activities within the explorative or the exploitative strategy are intertwined. Hence, we propose: H6. Top managers' cognitive integration moderates the relationship between top managers' cognitive differentiation and organisational ambidexterity.
2.6. Paradoxical frames, cognitive differentiation and integration and organisational ambidexterity So far, we have elaborated that top managers' paradoxical frames foster top managers' integrative complexity, and top managers' paradoxical frames and integrative complexity foster organisational ambidexterity. Consequently, we propose that top managers' integrative complexity mediates the relationship between top managers' paradoxical frames and organisational ambidexterity. Adopting a paradoxical frame helps to appreciate exploitation and exploration equally, even though they require contradicting actions (Smith & Tushman, 2005). Accordingly, top managers allocate resources to both activities at the same time or over time (Simsek, Heavey, Veiga, & Souder, 2009). Because of resource constraints neither the exploitative nor the explorative strategy has full access to available resources at a certain time. Given both strategies are executed independently from each other, the outcome of both strategies is basically the sum of its parts. Consequently, paradoxical frames are likely to foster only some degree of organisational ambidexterity. However, paradoxical frames induce cognitive differentiation and integration (see elaboration above; Miron-Spektor, Gino, et al., 2011). Top managers' cognitive differentiation and integration provide the basis to leverage the synergies between the exploitative and the explorative strategy. Through the process of cognitive differentiation, top managers realise how exploitative and explorative activities differ from each other, and understand their particular needs and requirements (Smith, 2014). Through the process of cognitive integration, top managers understand how exploitative and explorative activities can benefit from each other (Smith, 2014). Thus, top managers' cognitive differentiation and integration facilitates an ambidextrous strategy that is more than the sum of its parts because it enables to leverage the synergies between the exploitative and the explorative strategy. As a consequence, organisations, which intertwine their exploitative and explorative strategies can create stronger organisational ambidexterity than those which just seek exploitation and exploration independently form each other. H7. Top managers' cognitive differentiation mediates the relationship between paradoxical frames and organisational ambidexterity. H8. Top managers' cognitive integration mediates the relationship between paradoxical frames and organisational ambidexterity.
R. Wilms et al. / European Management Journal 37 (2019) 589e600
3. Methodology 3.1. Data and procedures The research project was conducted in Germany and aimed at surveying members of top management teams in firms with at least 20 employees in a cross-sectional design. Thus, participants needed to be involved in strategic decisions. Because of the advantages over mail survey, an online survey was developed (Boyer, Olson, Calantone, & Jackson, 2002), using the programme SoSci Survey (Version 2.6; Leiner, 2015). The sample was generated with the German panel provider mo'web research (Düsseldorf), a professional research platform and company that offers access to various expert panels for research and market research issues.1 The panel provider has access to 55,931 potential German participants that can be surveyed in an B2B context. Finally, we received 285 questionnaires. Since the panel provider sent this survey also to people who do not fulfil these inclusion criteria, 184 participants were dropped because the participant reported they were not part of the top management team or not involved in strategic decisions or the number of employees was below 20. These 184 participants only reported the inclusion criteria, and then, the survey ended for them. We ensured that each organisation was only represented by one top executive. None of the respondents who started the questionnaire left the questionnaire incomplete. In the end, 101 participants remained. We did not rely on random sampling with respect to all top managers in Germany. Only top managers that are listed in the panel can be part of our sample. Accordingly, it is impossible for some top managers to be part of our sample. However, random sampling is more important for descriptive studies that aim to describe how a certain attribute is distributed in the population. In studies that aim to test theory, it is more important that participants are relevant for the sample and are actually able to address & Widener, 2018). In similar the phenomena of interest (Spekle vein, the response rate becomes less important, since it is an indicator of response biases, given random sampling is applied (Hiebl & Richter, 2018). Accordingly, non-random sampling does pose a threat to our study. Instead, the quality of information that participants report, and how well they are suited to report about the & Widener, phenomena of interest is crucial for our study (Spekle 2018). 3.2. Quality of data The quality of data is highly influenced by the respondents' ability and motivation to provide an optimal answer (Podsakoff, MacKenzie, & Podsakoff, 2012). According to Kumar, Stern, and Anderson (1993), key informants deliver a better quality of data, compared to multiple respondents, if they are exclusively qualified to provide information about the issue under examination. The following quality criteria were employed to ensure that the participants are knowledgeable top management members, and thus are the ablest to deliver highquality data: (1) job title, (2) management level, (3) involvement in strategic activities, (4) job experience, (5) organisational tenure, and (6) age of the firm. The participants were asked to assign their position in the company by choosing between non-management, line management, middle management, and top management. Only participants on the top management level were retained in the analysis.
1
https://www.mowebresearch.com/en/[last access, 04.02.2017].
593
Secondly, the total amount of time to complete the questionnaire was used as a proxy of respondents' motivation. Respondents who ran through the survey very quickly presumably did not invest a lot of cognitive effort in their responses. All participants who took less than 220 s to finish the questionnaire were defined as speed runners. 15 participants came below the threshold since it cannot be precluded that they were able to answer the questions so quickly, and it would be arbitrarily to exclude them completely. A synthetic control variable was created, referred to as speed runner. Differences between the data are reported. On average, the entire sample had a job experience of 18.32 years (SD ¼ 8.27), an organisational tenure of 11.4 years (SD ¼ 5.64). The mean age of the firms was 38.95 years (SD ¼ 28.19). The median number of employees was 500. Furthermore, 38% of the firms were manufacturing firms, 56% were service firms, and 6% operated in other businesses. Further descriptive statistics are presented in Table 1. With respect to all German sectors, in the current sample, sectors, such as manufacturing, financial and insurance activities, information and communication as well as other services are slightly overrepresented, whereas wholesale, traffic and storage, professional, scientific and technical activities, administrative and support service activities as well as public administration and defense, compulsory social security, education, human health and social work activities are slightly underrepresented (See Table 1; Destatis e Statistisches Bundesamt, 2013). In this sample, the company size is almost equally distributed between firm size between 20 and 100, 101e500, 501e1000 and 1001e5000, stressing that this sample contains smaller as well as larger companies (See Table 1; Destatis e Statistisches Bundesamt, 2013). This limits the representativity of the study. However, the sample includes companies in manufacturing and service businesses, making it highly relevant to both manufacturing and service firms. Finally, we tested non-response bias between early and late respondents. The t-tests revealed no significant difference among items and constructs, which are represented through the average of related items. Additionally, a chi-square test was conducted to analyse differences between sectors, which also revealed no significant difference. 3.3. Measures All questionnaires were administered in German. To ensure content similarity with original scales, a translation/backtranslation procedure was used (Brislin, 1986). Two independent bilingual persons translated all the scales into German. The authors chose the best translation. A third independent bilingual person translated the German items back into English.2 A fourth person confirmed whether the translations were valid. 3.4. Cognitive differentiation and integration (integrative complexity) Integrative complexity comprises cognitive differentiation and integration and was measured by higher-order construct (HOC; Hair, Hult, Ringle, & Sarstedt, 2014). Cognitive differentiation was measured by a 6-item scale. A sample item is “I believe in the value of dissent”. Cognitive integration was measured by a 5-item scale. A sample item is “I highlight connections between seemingly conflicting perspectives raised by people on an issue” (Zhang, Waldman, Han, & Li, 2015). The items were measured on a 6-
2
A full list of items can be obtained on request from the first author.
594
R. Wilms et al. / European Management Journal 37 (2019) 589e600
Table 1 Descriptive statistics. Job Title
Percentage
Industry Affiliation
Percentage
CIO CEO Manager Director Director IT Department leader Authorized officer COO CTO
27% 26% 13% 4% 4% 3% 2% 2% 2%
4% 22% 10% 12% 10% 8% 2% 14% 7%
Managing director Owner and CEO Sales Manager CEO & CIO CFO Consultant HR Manager IT manager Owner Project Manager Senior Head Of Finance and Administration Team leader VP
2% 2% 2% 1% 1% 1% 1% 1% 1% 1% 1%
Agriculture, forestry and fishing Manufacturing, mining and quarrying, other industries Construction Wholesale, traffic and storage Information and communication Financial and insurance activities Real estate activities Professional, scientific and technical activities, administrative and support service activities Public administration and defences, compulsory social security, education, human health and social work activities Other service activities (energy, waste, water, hospitality, art, others) Firm size 20e100 101e500 501e1000 1001e5000 5001e50000 >50000
11% 25% 26% 20% 23% 5% 1%
1% 1%
point Likert scale (1 ¼ strongly disagree to 6 ¼ totally agree).
3.8. Analysis and results
3.5. Paradoxical frame
Before conducting the research project, the researchers discussed the two types of structural equation modelling (SEM), covariance-based SEM and partial least squared (PLS) SEM. Both techniques are complementary to each other. Their usage mainly depends on the research goals, but not on the research field (Hair, Matthews, Matthews & Sarstedt, 2017; Sarstedt, Hair, Ringle, Thiele & Gudgergan, 2016). Nevertheless, PLS SEM is widely applied in Marketing Science (Hair, Sarstedt, Ringle, & Mena, 2012). The authors finally decided to apply PLS SEM for the following reasons: Firstly, the sample size is small and the data is not perfectly normally distributed. PLS SEM does not call for perfectly normally distributed data and can be used for small sample sizes (Hair et al., 2014). Secondly, the objective of explaining how variations in exogenous and mediating variables are related to variations in organisational ambidexterity (prediction, Hair et al., 2017) was rated more important than testing whether this particular model fits the data best (e.g. based on a global fit criterion). PLS SEM minimizes the amount of unexplained variance (Hair et al., 2017). PLS SEM was chosen to analyse the proposed model. Using SmartPLS Version 3.2.3 (Ringle, Wende, & Becker, 2015), the inner weightings are estimated by a path method. The stability of the model is assessed by means of bootstrapping with 5000 resamples as recommended by Hair et al. (2014; 2017).
Paradoxical frame3 was measured by a 3-item scale (Ingram, Lewis, Barton, & Gartner, 2014). A sample item is “It is possible to maintain and develop our core competencies while simultaneously creating new innovations.” Scaling was slightly adjusted: The original scale was measured by a 5-point Likert scale from “strongly disagree/does not make sense (1)” and “strongly agree/makes perfect sense (5)”. In contrast, we used a 4-point Likert scale from “does not make sense (1)” to “makes perfect sense (4)”, distinguishing it from other scale indices. 3.6. Organisational ambidexterity Organisational ambidexterity was measured by an 8-item scale (Jansen, Tempelaar, van den Bosch, & Volberda, 2009), comprising 4 items for exploitation and 4 items for exploration. Organisational ambidexterity was measured by a HOC (Hair et al., 2014). A sample item for exploitation is “We improve our provision's efficiency of products and services”. A sample item for exploration is “Our organisation accepts demands that go beyond existing products and services”. All items of ambidexterity were measured on a 7point Likert scale (1 ¼ strongly disagree to 7 ¼ strongly agree). 3.7. Control variables The study controlled for environmental dynamism which was measured by a 7-point Likert scale (1 ¼ totally disagree to 7 ¼ totally agree). Environmental dynamism enforces organisations to build an ambidextrous orientation (Jansen et al., 2006; Junni et al., 2013). Further, we controlled firm size as well as firm age, CEO or top manager as dummy variable, sector, and speed runner.
3 The original scale is called paradoxical thinking, although the design of the scale fits into the definition of paradoxical frame. Thus, it was used to measure paradoxical frames.
3.9. Sample size considerations In the inner model, three substantive constructs point to organisational ambidexterity. To detect R2 values of 0.10 for a significance level of 5% and a statistical power of 80%, a minimum sample size of 124 is required. To detect R2 values of 0.25 for a significance level of 5% and a statistical power of 80%, a minimum sample size of 59 is required (Cohen, 1992; Hair et al., 2014). Since we expect all the effects between 0.10 and 0.25, a sample size of 101 has sufficient power to test our hypotheses.
R. Wilms et al. / European Management Journal 37 (2019) 589e600
595
Table 2 Assessing the outer model.
Differentiation Integration Paradoxical frames HOC: Organisational Ambidexterity Exploitation Exploration
No. Items
Mean
SD
IL min
IL max
HOC on LOC
CR
CA
AVE
Discr. Valid.
4 4 2 8 4 4
5.088 4.838 3.385 5.490 5.400 5.580
.763 .781 .572 1.100 1.058 1.300
.781 .755 .793 .676 .730 .845
.808 .858 .866 .856 .840 .927
n/a n/a n/a n/a .921 .937
.873 .874 .833 .928 .885 .933
.806 .807 .704 .911 .825 .903
.632 .634 .625 .619 .658 .776
Yes Yes Yes Yes Yes Yes
Note: IL min ¼ item loading minimum; IL max ¼ item loading maximum; LOC on HOC ¼ Path coefficient from higher-order construct to lower-order construct; AVE ¼ average variance extracted; CA ¼ Cronbach's alpha; CR ¼ composite reliability; ME ¼ mean; n/a ¼ not applicable; SD ¼ standard deviation.
3.10. Validity and reliability Three different means are used to analyse the inner model: (1) internal consistency, (2) convergent validity, and (3) discriminant validity (Hair et al., 2014). Internal consistency is assessed using Cronbach's a and composite reliability. All items exceed the threshold of 0.7 for both measures (Nunally & Bernstein, 1994), excepting paradoxical frames, which came below the threshold for Cronbach's a. To assume convergent validity, all items should share more variance with their substantive construct and all substantive constructs should share more variance with their HOC than with error variance, meaning that factor loadings4 should exceed the threshold of 0.708 (0.7082 ¼ 0.5). Hair et al. (2014) recommend only deleting items with lower factor loadings than 0.708 if the removal of an item increases the average variance extracted (AVE) to a value higher than 0.5. Three HOC items are kept even though they are below the threshold. Lower-order constructs as well as HOC fulfil convergent validity conditions (See Table 2 and Table 3). Thus, sufficient convergent validity can be asserted. Sufficient discriminant validity can be assumed if the AVE is higher than 0.5 for all constructs, and if the square root of a construct's AVE is larger than the correlation with any other construct (Fornell & Larcker, 1981; Hair et al., 2014). After dropping 1 item of paradoxical frames and 3 items of integrative complexity, the results were sufficient. Tables 2 and 3 show that each of the constructs meets the AVE criterion and the Fornell-Larcker criterion. Moreover, cross-loadings are examined to further ensure discriminant validity. Each of the items loads strongest on its substantive construct. All criteria to assess the quality of the outer model are met (Hair et al., 2014).
lenient, acquiescent, and consistent with how they think the researcher wants them to respond” (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003, p. 888). Due to the panel design of this study, anonymity and confidentiality is ensured. Further, Podsakoff et al. (2012) recommend eliminating common scale properties which may be a source of common method variance. The response formats are changed across sections. The number of scale points varied from a 4-point to a 7-point Likert scale. Furthermore, different indices were used for example “strongly agree” to “strongly disagree” or “makes no sense” to “makes perfect sense”. Furthermore, a statistical remedy was used to assess the magnitude of common method variance, a common latent factor test (Podsakoff et al., 2003). Following Liang, Saraf, Hu, and Xue (2007), a common method factor was included in the analysis, which was linked to all observed variables. Observed variables were converted into single-indicator constructs, allowing observed variables to be explained by more than one construct. For each of the single-indicator constructs, the incoming paths of the method factor as well as of the substantive construct were examined and compared to each other. “These two path coefficients are equivalent to the observed indicator's loadings on its substantive construct and the method factor” (Liang et al., 2007, p. 87). On average, the substantive constructs explained 81.8% of variance, whereas the method factor accounted for 2.3% of the variance explained. Thus, the ratio of the explained variance by substantive constructs to variance explained by the method factor is about 36 to 1. The method factor only loaded on 1 out of 18 indicators significantly. Due to the small magnitude of explained variance by the method factor and the small number of significant loadings on the method factor, it can be contended that the common method bias is not likely to be a serious threat to this study (See appendix). Additionally, the results do not change if the item that is significantly affected by the method factor is excluded from the analysis.
3.11. Common method bias and same source bias We refer to common method variance and do not distinguish between same source bias and common method bias, since “regardless of whether one considers various rater response styles, item characteristics, and aspects of the measurement context to be ‘method’ factors, they are all sources of systematic measurement error that threaten the validity of a study's findings” (Podsakoff et al., 2012, p. 540). Potentially, the common method bias could threaten the results of this study because it relies on key respondents and on information that was provided at the same time. Several means were used to counter common method variance problems. Firstly, procedural remedies were employed to reduce common method variance. A potential problem occurs when key informants “edit their responses to be more socially desirable,
4 Structural loadings can be interpreted as factor loadings for higher-order constructs (Doll, Xia, & Torkzadeh, 1994).
3.12. Assessing the structural model The model is showcased in Fig. 2, whereas all direct relationships are depicted in Table 2. Due to the path coefficient analysis, hypotheses H2, H3, H4, H5, and H6 are confirmed, whereas H1 showed an insignificant relationship. Four means were used for assessing the structural model: (1) multi-collinearity assessment, (2) R2 value, (3) the effect size f2, and (4) Stone-Geisser's Q2 (Hair et al., 2014). Firstly, the multi-collinearity was assessed using the variance inflation factor (VIF) where values of higher than 5 are problematic (Hair et al., 2014). Within the structural model, the values of the VIFs ranged from 1.130 to 3.697. Thus, multi-collinearity is not likely to be an issue in this study. Secondly, R2 values were examined to assess the predictive accuracy. According to Henseler, Ringle, and Sinkovics (2009, p. 303) “R2 values of 0.67, 0.33, or 0.19 for endogenous latent variables in the inner path model are described as substantial, moderate, or
596
R. Wilms et al. / European Management Journal 37 (2019) 589e600
Table 3 Correlation Table with confidence intervals of the Lower-Order Constructs and Discriminant Validity Assessment by Fronell-Larcker Criterion. Variable
1
1. Exploitation 2. Exploration
.811 .57** [.42, .69] .50** [.33, .63] .66** [.53, .76] .62** [.48, .73] .48** [.31, .61] .01 [-.18, .21] -.13 [-.32, .07] .02 [-.18, .22] .08 [-.12, .27] .22* [.02, .40]
3. Cognitive differentiation 4. Cognitive integration 5. Paradoxical frames 6. Market dynamism 7. Org. age 8. Org. size 9. Position 10. Sector 11. Speed runner
2
3
4
5
6
7
8
9
10
.05 [-.14, .25] -.01 [-.21, .19] -.24* [-.42, .05]
.03 [-.17, .22] .08 [-.11, .28]
-.06 [-.25, .14]
.881 .40** [.23, .56] .43** [.25, .57] .18 [-.01, .37] .69** [.56, .78] -.08 [-.28, .11] -.15 [-.34, .05] -.07 [-.27, .13] .16 [-.04, .34] .23* [.03, .40]
.795 .55** [.40, .68] .36** [.17, .52] .41** [.23, .56] .24* [.04, .41] -.22* [-.40, .03] .06 [-.14, .25] .11 [-.08, .30] .37** [.19, .53]
.796 .58** [.44, .70] .33** [.14, .49] .08 [-.11, .28] -.18 [-.36, .02] -.04 [-.23, .16] .13 [-.07, .31] .20* [.01, .38]
.831 .11 [-.09, .30] .10 [-.10, .29] -.16 [-.35, .03] .08 [-.11, .28] .10 [-.10, .29] .22* [.02, .40]
.842 -.13 [-.32, -.08 [-.27, -.08 [-.27, .12 [-.08, .12 [-.07,
.07] .12] .12] .31] .31]
-.06 [-.25, .18 [-.01, .16 [-.04, .07 [-.13,
.14] .37] .35] .26]
Note. * indicates p < 0.05; ** indicates p < 0.01. Values in square brackets indicate the 95% confidence interval for each correlation. Value on the diagonal is the square root of AVE (bold) for multi-item constructs. HOC: Higher-order construct.
Fig. 2. Results of structural equation modelling. Note: y < 0.10, * < 0.05, ** < 0.01, *** < 0.001. HOC: Higher-order construct, which comprises lower-order constructs. The inner model is displayed within the dashed line. All tests are one-tailed. Only significant control variables are shown in the model.
weak”. As shown in Fig. 2, all explained substantive constructs showed sufficient R2 values. Thirdly, effect size outlines “whether a construct has substantive impact on the endogenous construct” (Hair et al., 2014, p. 177). The effect size f2 of 0.02, 0.15 and 0.35, respectively, represent small, medium, and large effects (Cohen, 1988). Examining the effects sizes (see Table 4), the effects of the independent variables on the dependent variable varied from small to large effects. Fourthly, Stone-Geisser's Q2 was calculated (Geisser, 1974; Stone, 1974), using a blindfolding procedure with an omission distance of 7, to assess the predictive relevance of the structural model. This is the predominant procedure for assessing a structural model for PLS SEM. The test assumes that if all Q2 values for a certain endogenous latent variable are “larger than zero, its explanatory variables provide predictive relevance” (Henseler, Ringle, & Sinkovics, 2009, p. 305). Following the recommendation
of Hair et al. (2014), cross-validated redundancies were used for calculating Q2. All corresponding values were positive, indicating a strong overall predictive power. Furthermore, q2 was calculated. It is a relative measure of predictive relevance and sets Q2included and Q2excluded into relation (see Table 4). q2 can be similarly interpreted as f2. The results are presented in Table 4. Even though PLS SEM does not provide an overall fit index (Hair et al., 2014), each of the evaluation criteria indicated a sufficient model quality. 3.13. Mediation effects Hypotheses H7 and H8 were tested separately. Hair et al. (2014) recommend analysing mediation hypothesis through bootstrapping and avoid the Sobel test (Sobel, 1982). Bootstrapping does not rely on distributional assumptions (Preacher & Hayes, 2008) as
R. Wilms et al. / European Management Journal 37 (2019) 589e600
597
Table 4 Results of the Proposed Structural Equation Modelling (one-tailed tests). Path H1 H2 H3 H4 H5 H6
Paradoxical Frames / Organisational Ambidexterity Paradoxical Frames /Cognitive Differentiation Paradoxical Frames / Cognitive Integration Cognitive Differentiation / Organisational Ambidexterity Cognitive Integration / Organisational Ambidexterity Cognitive Differentiation * Cognitive Integration / Organisational Ambidexteritya
b
p
f2
q2
-.103 .607 .587 .310 .267 .208
.088 .000 .000 .037 .038 .001
.018 .585 .527 .079 .069 .117
<0
.012 .017
C X X X X X
Note: All confirmed hypotheses are marked with an X. a The moderation effect has not been considered for the complete model in the blindfolding procedure.
the Sobel test does. Testing H7, first, the direct effect between paradoxical frames and organisational ambidexterity was calculated (0.115, p ¼ 0.053). Second, the indirect effect was calculated by inserting cognitive differentiation into the model as a mediator. The path coefficients of the relationship from paradoxical frames to cognitive differentiation (0.608, p < 0.000), and from cognitive differentiation to organisational ambidexterity (0.390, p ¼ 0.002) were multiplied, estimating the indirect effect (0.608 0.390 ¼ 0.237). The direct effect decreased to a negative value of 0.037 (p ¼ 0.328), which indicates a suppressor effect. Hair et al. (2014, p. 225) stress that “this kind of situation always represents full mediation”, that is, the result confirms hypothesis H7. Testing H8, firstly, the direct effect between paradoxical frames and organisational ambidexterity was calculated (0.115, p ¼ 0.053). Secondly, the indirect effect was calculated by inserting cognitive integration into the model as a mediator. The path coefficients of the relationship from paradoxical frames to cognitive integration (0.587, p < 0.000), and from cognitive integration to organisational ambidexterity (0.375, p ¼ 0.000) were multiplied, estimating the indirect effect (0.587 0.375 ¼ 0.220). The direct effect decreased to a negative value of 0.051 (p ¼ 0.3258), which indicates a suppressor effect. Hair et al. (2014, p. 225) stress that “this kind of situation always represents full mediation”, that is, the result confirms hypothesis H8.
3.14. Discussion, implications, conclusions, limitations and directions for future research Even though there is a clear understanding that organisations must engage in exploitation and exploration simultaneously, it is only roughly understood how top managers can manage tensions that an ambidextrous strategy entails. Whereas it is empirically founded and tested that social ties and group processes facilitate ambidexterity (Alexiev, Jansen, Van den Bosch, & Volberda, 2010; Heavey, Simsek, & Fox, 2015; Lubatkin et al., 2006), the role of top managers' cognition has remained empirically neglected until recently (Lin & McDonough, 2014; Smith & Tushman, 2005). This study contributes to the literature of top management cognition by theoretically developing the relationship between cognitive frames and integrative complexity and between integrative complexity and organisational ambidexterity. Furthermore, the study empirically tests the proposed model and validates the mediating role of cognitive differentiation and integration between paradoxical frames and organisational ambidexterity. In line with others, the study shows that paradoxical frames encourage individuals to engage in cognitive differentiation and integration (Miron-Spektor, Gino, et al., 2011; Smith & Tushman, 2005), whereas paradoxical frames are not significantly related to organisational ambidexterity. Consequently, solely appreciating exploitation and exploration does not foster ambidexterity, but is an important antecedent for cognitive differentiation and
integration. Thus, constantly communicating the importance of both exploitation and exploration is not sufficient. It rather facilitates the next important step of cognitive differentiation and integration. Top managers' cognitive differentiation and integration directly foster organisational ambidexterity. Cognitive differentiation increases the range of available information in terms of exploitation and exploration which, in turn, facilitates cognitive integration which creates synergies between both strategies (Hahn et al., 2014; Tetlock et al., 1993). Cognitive differentiation and integration each contribute to organisational ambidexterity, even though only when cognitive differentiation and integration are equally emphasised does the full power of integrative complexity unfold. Inherently, O’Reilly and Tushman (2008, p. 191) have already argued that by stating that “the crucial task […] is not the simple organisational structural decision in which the exploratory and exploitative subunits are separated, but the processes by which these units are integrated in a value enhancing way”. Fostering the appreciation of both exploitation and exploration, paradoxical frames encourage top managers to assign financial resources and human resources as well as responsibilities to each of the strategies (Smith, 2014). Pursuing incremental and radical innovation simultaneously requires a deeper understanding of how exploitation and exploration are interrelated. Integrative complexity enables top managers to understand the needs of exploitative and explorative strategies and enables them to build synergies between each of the strategies. Processes of cognitive differentiation and integration enable top executives to recognise, process, and connect information building synergies between exploitation and exploration. Even though cognitive frames structure the information from the environment, only the degree to which the information is differentiated and integrated drives ambidextrous outcomes. This is an important theoretical distinction between recognising contradictory forces and intertwining them synergistically.
3.15. Managerial implications This study makes contributions to managerial practice. First, the study emphasises the importance of paradoxical frames for organisational ambidexterity. The top management should establish that exploitation and exploration are equally important for creating an ambidextrous organisation. Consequently, the top management should set objectives for and allocate resources and responsibilities to both the exploitative and the explorative strategy. Additionally, CEOs should set norms which outline that none of the strategies is more important than the other in the long term. Secondly, top managers should create a deep understanding of the success factors of each of the strategies and set up meetings at which the product design, the organisational design and the allocation of resources to exploitation and exploration are individually discussed. At these meetings, each of the members should
598
R. Wilms et al. / European Management Journal 37 (2019) 589e600
understand the unique contributions, strengths, weaknesses, and needs of both exploitation and exploration, even though he or she might not be responsible for this strategic domain. Additionally, these meetings should result in a consensus about how to improve each strategy. In a second step, a further meeting should be set which aims at elaborating and identifying connections and potential synergies. In this way, the top management members understand how intertwining exploitation and exploration will create an aligned overall corporate strategy. In parallel, organisations could offer training to their manager on how to deal with constant ambiguity. 3.16. Limitations and directions for future research Firstly, the cross-sectional design of the study may limit the causal interpretation of the relationship among the constructs. The results of this study should be taken with caution. Therefore, future research should conduct longitudinal studies to ensure causal effects between the variables. Secondly, even though procedural remedies and statistical remedies were used to minimise or control the common method
Items
IC01_02 IC01_03 IC01_04 IC01_06 IC02_02 IC02_03 IC02_04 IC02_05 A01_01 A01_02 A01_03 A01_04 A02_01 A02_02 A02_03 A02_04 PT01_01 PT01_02
organisational ambidexterity. In addition to that, future studies should take the team member with the highest expression of a characteristic into account. For example, Nijstad, Berger-Selman, and De Dreu (2014) outline that minority dissent is crucial to innovation. This means that a person with a high expression of integrative complexity can substantially influence team processes and their outcomes if this person creates, for instance, minority dissent. Fourthly, paradoxical frames were measured on a 4-point Likert scale. A related problem with low numbers of scale points is limited variance. Therefore, future studies should use scales with more than 5 scale points. Fifth, studies that analyse managerial cognition should use integrative complexity and cognitive frames in combination, since integrative complexity mediates the relationship between paradoxical frames and outcomes. At the same time, further research should investigate how cognitive frames and cognitive processes differently affect strategic formulation and implementation.
Appendix. Common Method Factor
Substantive Construct
Method Factor
Original Sample (O)
P Values
Squared Loading
Original Sample (O)
P Values
Squared Loading
0.809 0.638 0.939 0.790 0.987 0.743 0.683 0.758 0.994 0.312 0.842 1.040 0.731 0.846 0.940 0.998 0.803 0.863 0.818
0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.111 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
0.654 0.407 0.882 0.624 0.974 0.552 0.466 0.575 0.988 0.097 0.709 1.082 0.534 0.716 0.884 0.996 0.645 0.745 0.696
0.006 0.174 0.151 0.013 0.147 0.047 0.112 0.001 0.179 0.468 0.009 0.135 0.145 0.014 0.015 0.135 0.069 0.075 0.009
0.948 0.263 0.342 0.911 0.161 0.750 0.482 0.993 0.204 0.007 0.956 0.061 0.172 0.878 0.820 0.109 0.229 0.264
0.000 0.030 0.023 0.000 0.022 0.002 0.013 0.000 0.032 0.219 0.000 0.018 0.021 0.000 0.000 0.018 0.005 0.006 0.023
bias, the key informant approach is susceptible to it. Different key informants should provide information about the independent variables and the dependent variables or a secondary data source should be used (revenues generated by products older than and younger than three years). These kinds of study design rule out common method variance (Podsakoff et al., 2003, 2012). Thirdly, key informants rated their individual paradoxical frames and their individual integrative complexity. Individual measurements of these constructs might bias the results, whereas we tested whether the answers of CEOs and top management members differed. The aggregated level of the team or the team member with the highest expression of a characteristic may be better predictors of organisational ambidexterity. For example, if a respondent is characterised by low integrative complexity, but a certain number of other team members are or at least one team member is highly integratively complex, the company is likely to be ambidextrous even though the respondent is not integratively complex. Thus, future research should develop scales that measure team integrative complexity and analyse their relationship with
References Alexiev, A. S., Jansen, J. J. P., Van den Bosch, F. A. J., & Volberda, H. W. (2010). Top management team advice seeking and exploratory innovation: The moderating role of TMT heterogeneity. Journal of Management Studies, 47, 1343e1364. Bartunek, J. M., Gordon, J. R., & Weathersby, R. P. (1983). Developing “complicated” understanding in administrators. Academy of Management Review, 8(2), 273e284. https://doi.org/10.5465/AMR.1983.4284737. Benner, M. J., & Tushman, M. L. (2003). Exploitation, exploration and process management: The productivity dilemma revisited. Academy of Management Review, 28(2), 238e256. 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(6), 587e610. https://doi.org/10.1002/ smj.1972. Boyer, K. K., Olson, J. R., Calantone, R. J., & Jackson, E. C. (2002). Print versus electronic surveys: A comparison of two data collection methodologies. Journal of Operations Management, 20(4), 357e373. https://doi.org/10.1016/S02726963(02)00004-9. Brislin, R. W. (1986). Research instruments. In Field Methods in Cross-Cultural Research (pp. 159e162). Bromiley, P., & Rau, D. (2016). Social, behavioral, and cognitive influences on upper echelons during strategy process: A literature review. Journal of Management,
R. Wilms et al. / European Management Journal 37 (2019) 589e600 42(1), 174e202. https://doi.org/10.1177/0149206315617240. Carroll, J., & Bright, A. (2010). Integrative complexity of public beliefs toward wildfire management: Development of a scale. Journal of Applied Social Psychology, 40(2), 344e359. https://doi.org/10.1111/j.1559-1816.2009.00577.x. Cohen, J. (1988). Statistical power analysis: A computer program. Routledge. Cohen, J. (1992). A power primer. Psychological Bulletin, 112(1), 155e159. https://doi. org/10.1037/0033-2909.112.1.155. Daft, R. L., & Weick, K. E. (1984). Toward a model of organizations as interpretation systems. Academy of Management Review, 9(2), 284e295. https://doi.org/10. 5465/AMR.1984.4277657. Destatis e Statistisches Bundesamt. (2013). Unternehmensregister: Betriebe nach ۧenklassen. Betriebsgro Retrieved from https://www.destatis.de/DE/ ZahlenFakten/GesamtwirtschaftUmwelt/UnternehmenHandwerk/ Unternehmensregister/Tabellen/BetriebeBeschaeftigtengroessenklassenWZ08. html. Doll, W. J., Xia, W., & Torkzadeh, G. (1994). A confirmatory factor analysis of the enduser computing satisfaction instrument. MIS Quarterly, 18(4), 453. https://doi. org/10.2307/249524. Eggers, J. P., & Kaplan, S. (2013). Cognition and capabilities: A multi-level perspective. The Academy of Management Annals, 7(1), 295e340. https://doi. org/10.1080/19416520.2013.769318. Evans, J. S. B. T., Barston, J. L., & Pollard, P. (1983). On the conflict between logic and belief in syllogistic reasoning. Memory & Cognition, 11(3), 295e306. https://doi. org/10.3758/BF03196976. Finkelstein, S., & Hambrick, D. C. (1996). Strategic leadership: Top executives and their effects on organizations. South-Western Pub. Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research, 18(1), 39e50. https://doi.org/10.2307/3151312. Geisser, S. (1974). A predictive approach to the random effect model. Biometrika, 61(1), 101e107. https://doi.org/10.1093/biomet/61.1.101. Gibson, C. B., & Birkinshaw, J. (2004). The antecedents, consequences, and meditating role of organizational ambidexterity. Academy of Management Journal, 47(2), 209e226. Gotsi, M., Andriopoulos, C., Lewis, M. W., & Ingram, A. E. (2010). Managing creatives: Paradoxical approaches to identity regulation. Human Relations, 63(6), 781e805. https://doi.org/10.1177/0018726709342929. Hahn, T., Preuss, L., Pinkse, J., & Figge, F. (2014). Cognitive frames in corporate sustainability: Managerial sensemaking with paradoxical and business case frames. Academy of Management Review, 39(4), 463e487. https://doi.org/10. 5465/amr.2012.0341. Hair, J. F., Hult, G. T. M., Ringle, C., & Sarstedt, M. (2014). A primer on partial least squares structural equation modeling (PLS-SEM). Incorporated: SAGE Publications. Hair, J. F., Matthews, L. M., Matthews, R. L., & Sarstedt, M. (2017). PLS-SEM or CBSEM: updated guidelines on which method to use. International Journal of Multivariate Data Analysis, 1, 107. Heavey, C., Simsek, Z., & Fox, B. C. (2015). Managerial social networks and ambidexterity of SMEs: The moderating role of a proactive commitment to innovation. Human Resource Management, 54, 201e221. Henseler, J., Ringle, C. M., & Sinkovics, R. R. (Eds.). (2009). Advances in international marketing. Retrieved from http://www.emeraldinsight.com/10.1108/S14747979(2009)0000020014. He, Z.-L., & Wong, P.-K. (2004). Exploration vs. exploitation: An empirical test of the ambidexterity hypothesis. Organization Science, 15(4), 481e494. https://doi.org/ 10.1287/orsc.1040.0078. Hiebl, M. R. W., & Richter, J. F. (2018). Response rates in management accounting survey research. Journal of Management Accounting Research, 30(2), 59e79. https://doi.org/10.2308/jmar-52073. Ingram, A. E., Lewis, M. W., Barton, S., & Gartner, W. B. (2014). Paradoxes and innovation in family firms: The role of paradoxical thinking. Entrepreneurship: Theory and Practice, 40(1), 161e176. https://doi.org/10.1111/etap.12113. Jansen, J. J. P., Tempelaar, M. P., van den Bosch, F. A. J., & Volberda, H. W. (2009a). Structural differentiation and ambidexterity: The mediating role of integration mechanisms. Organization Science, 20(4), 797e811. https://doi.org/10.1287/orsc. 1080.0415. Jansen, J. J. P., van den Bosch, F. A. J., & Volberda, H. W. (2006). Exploratory innovation, exploitative innovation, and performance: Effects of organizational antecedents and environmental moderators. Management Science, 52(11), 1661e1674. https://doi.org/10.1287/mnsc.1060.0576. Jansen, J. J. P., Vera, D., & Crossan, M. (2009b). Strategic leadership for exploration and exploitation: The moderating role of environmental dynamism. The Leadership Quarterly, 20(1), 5e18. https://doi.org/10.1016/j.leaqua.2008.11.008. Junni, P., Sarala, R. M., Taras, V., & Tarba, S. Y. (2013). Organizational ambidexterity and performance: A meta-analysis. Academy of Management Perspectives, 27(4), 299e312. https://doi.org/10.5465/amp.2012.0015. Kahneman, D. (2011). Thinking fast and thinking slow. New York, NY: Farrar, Strauss and Giroux. Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47(2), 263e292. Karlsson, N., Loewenstein, G., & Seppi, D. (2009). The ostrich effect: Selective attention to information. Journal of Risk and Uncertainty, 38(2), 95e115. https:// doi.org/10.1007/s11166-009-9060-6. Kiesler, S., & Sproull, L. (1982). Managerial response to changing environments: Perspectives on problem sensing from social cognition. Administrative Science
599
Quarterly, 27(4), 548e570. https://doi.org/10.2307/2392530. Kumar, N., Stern, L. W., & Anderson, J. C. (1993). Conducting interorganizational research using key informants. Academy of Management Journal, 36(6), 1633e1651. https://doi.org/10.2307/256824. Leiner, D. J. (2015). SoSci survey (version 2.6. 00-i)[Computer software]. Retrieved October, 20, 2015. Liang, H., Saraf, N., Hu, Q., & Xue, Y. (2007). Assimilation of enterprise systems: The effect of institutional pressures and the mediating role of top management. MIS Quarterly, 59e87. Lin, H.-E., & McDonough, E. F. (2014). Cognitive frames, learning mechanisms, and innovation ambidexterity: Cognitive frames and innovation ambidexterity. Journal of Product Innovation Management, 31, 170e188. https://doi.org/10.1111/ jpim.12199. Lubatkin, M. H., Simsek, Z., Ling, Y., & Veiga, J. F. (2006). Ambidexterity and performance in small-to medium-sized firms: The pivotal role of top management team behavioral integration. Journal of Management, 32(5), 646e672. https:// doi.org/10.1177/0149206306290712. March, J. G. (1991). Exploration and exploitation in organizational learning. Organization Science, 2(1), 71e87. https://doi.org/10.1287/orsc.2.1.71. Miron-Spektor, E., Efrat-Treister, D., Rafaeli, A., & Schwarz-Cohen, O. (2011). Others' anger makes people work harder not smarter: The effect of observing anger and sarcasm on creative and analytic thinking. Journal of Applied Psychology, 96(5), 1065e1075. https://doi.org/10.1037/a0023593. Miron-Spektor, E., Gino, F., & Argote, L. (2011). Paradoxical frames and creative sparks: Enhancing individual creativity through conflict and integration. Organizational Behavior and Human Decision Processes, 116(2), 229e240. https:// doi.org/10.1016/j.obhdp.2011.03.006. Miron-Spektor, E., Ingram, A., Keller, J., Smith, W. K., & Lewis, M. W. (2018). Microfoundations of organizational paradox: The problem is how we think about the problem. Academy of Management Journal, 61(1), 26e45. https://doi. org/10.5465/amj.2016.0594. Nickerson, R. S. (1998). Confirmation bias: A ubiquitous phenomenon in many guises. Review of General Psychology, 2(2), 175e220. https://doi.org/10.1037/ 1089-2680.2.2.175. Nijstad, B. A., Berger-Selman, F., & De Dreu, C. K. W. (2014). Innovation in top management teams: Minority dissent, transformational leadership, and radical innovations. European Journal of Work & Organizational Psychology, 23(2), 310e322. https://doi.org/10.1080/1359432X.2012.734038. Nunally, J., & Bernstein, I. H. (1994). Psychonometric theory. O'Reilly, C. A., & Tushman, M. L. (2008). Ambidexterity as a dynamic capability: Resolving the innovator's dilemma. Research in Organizational Behavior, 28, 185e206. https://doi.org/10.1016/j.riob.2008.06.002. Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, N. P. (2003). Common method biases in behavioral research: A critical review of the literature and recommended remedies. Journal of Applied Psychology, 88(5), 879e903. https:// doi.org/10.1037/0021-9010.88.5.879. Podsakoff, P. M., MacKenzie, S. B., & Podsakoff, N. P. (2012). Sources of method bias in social science research and recommendations on how to control it. Annual Review of Psychology, 63(1), 539e569. https://doi.org/10.1146/annurev-psych120710-100452. Porac, J. F., & Thomas, H. (2002). Managing cognition and strategy: Issues, trends and future directions. In Handbook of Strategy and Management (pp. 165e181). Preacher, K. J., & Hayes, A. F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40(3), 879e891. https://doi.org/10.3758/BRM.40.3.879. Raisch, S., & Birkinshaw, J. (2008). Organizational ambidexterity: Antecedents, outcomes, and moderators. Journal of Management, 34(3), 375e409. https://doi. org/10.1177/0149206308316058. Raisch, S., Birkinshaw, J., Probst, G., & Tushman, M. L. (2009). Organizational ambidexterity: Balancing exploitation and exploration for sustained performance. Organization Science, 20(4), 685e695. https://doi.org/10.1287/orsc.1090. 0428. Ringle, C., Wende, S., & Becker, J. (2015). Smart PLS 3. Boenningstedt: SmartPLS. GmbH. Rosing, K., Frese, M., & Bausch, A. (2011). Explaining the heterogeneity of the leadership-innovation relationship: Ambidextrous leadership. The Leadership Quarterly, 22(5), 956e974. https://doi.org/10.1016/j.leaqua.2011.07.014. Sarstedt, M., Hair, J. F., Ringle, C. M., Thiele, K. O., & Gudergan, S. P. (2016). Estimation issues with PLS and CBSEM: Where the bias lies! Journal of Business Research, 69, 3998e4010. Simsek, Z., Heavey, C., Veiga, J. F., & Souder, D. (2009). A typology for aligning organizational ambidexterity's conceptualizations, antecedents, and outcomes. Journal of Management Studies, 46(5), 864e894. https://doi.org/10.1111/j.14676486.2009.00841.x. Smith, W. K. (2014). Dynamic decision making: A model of senior leaders managing strategic paradoxes. Academy of Management Journal, 57(6), 1592e1623. https:// doi.org/10.5465/amj.2011.0932. Smith, W. K., Besharov, M. L., Wessels, A. K., & Chertok, M. (2012). A paradoxical leadership model for social entrepreneurs: Challenges, leadership skills, and pedagogical tools for managing social and commercial demands. The Academy of Management Learning and Education, 11(3), 463e478. https://doi.org/10.5465/ amle.2011.0021. Smith, W. K., & Lewis, M. W. (2011). Toward a theory of paradox: A dynamic equilibrium model of organizing. Academy of Management Review, 36(2), 381e403.
600
R. Wilms et al. / European Management Journal 37 (2019) 589e600
Smith, W. K., & Tushman, M. L. (2005). Managing strategic contradictions: A top management model for managing innovation streams. Organization Science, 16(5), 522e536. https://doi.org/10.1287/orsc.1050.0134. Sobel, M. E. (1982). Asymptotic confidence intervals for indirect effects in structural equation models. Sociological Methodology, 13, 290e312. , R. F., & Widener, S. K. (2018). Challenging issues in survey research: DisSpekle cussion and suggestions. Journal of Management Accounting Research, 30(2), 3e21. https://doi.org/10.2308/jmar-51860. Stone, M. (1974). Cross-validatory choice and assessment of statistical predictions. Journal of the Royal Statistical Society: Series B, 111e147. Tetlock, P. E., Peterson, R. S., & Berry, J. M. (1993). Flattering and unflattering personality portraits of integratively simple and complex managers. Journal of Personality and Social Psychology, 64(3), 500e511. https://doi.org/10.1037/00223514.64.3.500. Tripsas, M., & Gavetti, G. (2000). Capabilities, cognition, and inertia: Evidence from digital imaging. Strategic Management Journal, 1147e1161. Tushman, M. L., & O'Reilly, C. A. (1996). Ambidextrous organizations: Managing evolutionary. California Management Review, 38(4), 8e30. Walsh, J. P. (1995). Managerial and organizational cognition: Notes from a trip down
memory lane. Organization Science, 6(3), 280e321. https://doi.org/10.1287/orsc. 6.3.280. West, R. F., Toplak, M. E., & Stanovich, K. E. (2008). Heuristics and biases as measures of critical thinking: Associations with cognitive ability and thinking dispositions. Journal of Educational Psychology, 100(4), 930e941. https://doi.org/10. 1037/a0012842. Winnen, L., & Wilms, R. (2014). Intellectual capital architectures and ambidexterity: A dynamic approach. In Proceedings of the the 6th European conference on intellectual capital (pp. 331e340). Bratislava: Academic Conferences Limited. Wong, E. M., Ormiston, M. E., & Tetlock, P. E. (2011). The effects of top management team integrative complexity and decentralized decision making on corporate social performance. Academy of Management Journal, 54(6), 1207e1228. https:// doi.org/10.5465/amj.2008.0762. Zacher, H., & Wilden, R. G. (2014). A daily diary study on ambidextrous leadership and self-reported employee innovation. Journal of Occupational and Organizational Psychology, 87(4), 813e820. https://doi.org/10.1111/joop.12070. Zhang, Y., Waldman, D. A., Han, Y.-L., & Li, X.-B. (2015). Paradoxical leader behaviors in people management: Antecedents and consequences. Academy of Management Journal, 58(2), 538e566. https://doi.org/10.5465/amj.2012.0995.