Abstract
In this study, the authors investigate how network relations affect project performance through creativity. They challenge previous conceptualizations of creativity by proposing the idea that creative outcomes in project industries can be conceptualized as deviation from past projects and deviation from partners’ projects. While previous research has mostly assumed that network relations are beneficial to performance because they increase innovation and creativity, this conceptualization of creativity reveals different empirical regularities. The authors show that network relations are likely to exercise both positive and negative effects on creativity and that creativity is also likely to exercise both positive and negative effects on performance. In addition, networks show both convergent and divergent effects on the two dimensions of creativity. The findings show that network relations activate important trade-offs that organizations must consider in their strategic choices and open up research on the social side of creativity.
‘Here’s to the crazy ones. The misfits. The rebels. The troublemakers. The round pegs in the square holes. The ones who see things differently. They’re not fond of rules. And they have no respect for the status quo. You can quote them, disagree with them, glorify or vilify them. About the only thing you can’t do is ignore them. Because they change things. They invent. They imagine. They heal. They explore. They create.’ (Apple Advertising Campaign, 1997)
Introduction
Perhaps the strongest justification for the need to study networks in organizations derives from the argument that social relations could beneficially contribute to performance (Borgatti and Foster, 2003; Brass et al., 2004). This argument has often been justified by the intermediating effect of creativity. A core tenet of network research is that networks could exercise beneficial effects on performance because they affect the recombination of knowledge that drives the generation of creative outcomes (Baum et al., 2000; Burt, 2004; Ferriani et al., 2009; Powell et al., 1996; Uzzi and Spiro, 2005). The association between network relations and creative outcomes appears to be positive and has gained some empirical confirmation (Ahuja, 2000; Baum at al., 2000; Cattani and Ferriani, 2008; Perry-Smith, 2006; Rowley et al., 2000). Based on such evidence, previous research developed the theoretical position that networks positively affect creative outcomes, which positively affect performance.
Yet, this theoretical position could be challenged when focusing on the concept of creativity largely used by previous research. Creativity has traditionally been conceptualized as a one- dimensional construct defined as concepts, ideas, projects, or products or procedures that are ‘novel’ and ‘useful’ (Amabile, 1988; Oldham and Cummings, 1996; Stein, 1953). Recently, authors called for the expansion of the concept of creativity, recognizing its multidimensional characteristics, whose exploration may reveal novel empirical regularities that would otherwise have been difficult to identify (George, 2007; Unsworth, 2001). In this article, we expand research on the social context of creativity building on a relative view of creativity and on the insight that creative outcomes could be conceptualized as deviations (Mainemelis, 2010). Creative acts can be seen as the opposite of routine acts and entail a differentiation from what is habitual, in relation to the past, and what is conformist, in relation to the social circle a constituency is embedded in (Ford, 1996). Following these premises, we identify two dimensions of creativity: deviation from the past and deviation from partners.
Investigating how these two dimensions of creativity are predicted by networks and predict performance, we disentangle several new effects and expand our understanding beyond the simple theorization of positive relationships. Instead of simply illustrating that networks positively affect creativity, we show in this article that networks can exercise both positive and negative effects on creativity. Instead of simply illustrating that creativity exercises a positive effect on performance, we show that creativity can exercise both positive and negative effects on performance. Showing both the positive and negative effects that link networks, creativity and performance we offer a way to address the empirical controversies over research in networks and creativity (Fleming et al., 2007) and research in networks and performance (Brass et al., 2004; Burt, 2000; Rowley et al., 2000; Shaner and Maznevski, 2011; Zaheer and Soda, 2009). Hence, the exploration of the social context of creativity is opened up, questioning our previous assumptions and fueling new investigations.
Our work is focused on project-based industries, which are assuming increasing importance in management research and reflect a growing tendency among modern firms to rely on project forms of organizing (Bechky, 2006). More specifically, our research focuses on the TV industry (Perretti et al., 2008; Soda et al., 2004, 2008). We studied the entire population of TV movies in Italy for the period 1999–2008. Each TV movie is produced by a TV production team, which is an example of project organization (Bechky, 2006) or ‘hub’ organization (Baker and Faulkner, 1991; Faulkner and Anderson, 1987; Goodman and Goodman, 1976). Each TV production team, especially in Italy, generally belongs to a small, independent and legally constituted organization that is financially responsible for the success or failure of the project. Each team can hence be considered as a particular type of temporary organization formed pooling freelance specialists, who are tied to other teams through co-membership in different projects. Previous research illustrated the distinctive empirical insights that could be derived from studying network effects on creativity in artistic fields such as television, cinema and musicals (Cattani and Ferriani, 2008; Uzzi and Spiro, 2005; Zaheer and Soda, 2009).
‘Think different’: A relative view on creativity
The traditional definition of creativity introduced by Stein (1953) refers to ideas, products, projects, services or technologies that have the characteristic of being novel and useful to a social context. Several authors have endorsed this definition of creativity. Amabile (1988) defines creative outcomes as novel and useful ideas. Oldham and Cummings (1996) define creativity as the generation of products, ideas and procedures that are novel, original, useful and relevant. Scott and Bruce (1994) define creativity as the production of ideas that are both new and create value for an organization.
Yet, recently authors have started challenging this definition of creativity, highlighting some limitations that restrain the possibility of developing a network theory of creativity. First, this conceptualization limits the exploration of performance outcomes of creativity. The attribute ‘useful’ entails dependence on final outcomes. In for-profit organizations ‘useful’ is what contributes to performance and the usefulness attribute of creativity would allow supporting positive relationships while limiting a priori the observation of possible negative effects (George, 2007; Shalley et al., 2004). Second, this conceptualization limits the exploration of the network antecedents of creativity due to a possible endogeneity problem. The choice of forming or maintaining a relationship with a partner firm or group is motivated by the very reason of gaining access to knowledge that is deemed useful and new to the organization (Galaskiewicz, 1985; Gulati, 1995). Hence, organizations could self-select relationships on the basis of the very same criteria that generally describe creativity, creating an endogenous relationship that tends to confirm positive empirical associations. Third, this conceptualization limits the exploration of the ways in which creativity links networks to performance. In fact, using a one-dimensional definition of creativity strongly simplifies a highly complex construct composed of a wide array of different dimensions of creative outcomes worth distinguishing (Unsworth, 2001). Fourth, the attributes of usefulness and novelty depend on the subjective evaluations of different stakeholders and there are often conflicting opinions on what should be considered creative (George, 2007). Fifth, this conceptualization may not be adequate to capture creativity in all domains since in most creative industries and artistic fields the concept of creativity is highly dissociated from a utilitarian perspective and is inspired by the idea of ars gratia artis (art for art’s sake) (Caves, 2000).
There is a possible alternative conceptualization of creativity that considers the lack of one-dimensionality in the construct and does not depend on subjective attributes of usefulness and novelty. Creativity can be intended as a relative concept and as ‘deviation from’ a point of reference. Mainemelis (2010) talks of creative deviance and conceptualizes creativity as a disjunction from an order. Mumford and Gustafson (1988) conceptualize creativity as a departure from the status quo. Ford (1996) argues that creative actions and results could be intended as the opposite of routine actions and results: creativity entails differentiating from what is habitual and conformist. ‘Habitual’ has a point of reference in time and refers to the recursive replication of activities over time. ‘Conformist’ has a point of reference in the social space and refers to the replication of activities among members embedded in a social collective. Building on this view, we here embrace a relative view of creativity as the project’s deviation from past projects and from partners’ projects.
Deviating from the past
A first alternative way of identifying creativity is with respect to the time dimension and conceptualizing it as deviation from the past. Knowledge, practices and ideas held by organizations may reflect their own past experiences or past work. Creativity can thus be portrayed by the extent to which a new project or idea breaks away or diverges from the project or ideas developed by the members of the organization in the past. In poetry or in the arts, artists’ innovations are usually interpreted as fractures or further evolutionary stages in relation to preceding periods. By way of an example, Pablo Picasso’s innovations are often described as experimentations reflected in rapid changes of style with respect to his former periods. Critics generally consider Picasso’s late 1905 paintings as more emotionally detached than those from his previous blue or rose periods. From the First World War (1914–1918) onward, Picasso moved from style to style and built his success and reputation by systematically breaking with his own past. A large body of research has highlighted how difficult it is for actors to de-link from the anchorage of the contents, ideas, formats and products produced in their past experiences (Levitt and March, 1988; March and Simon, 1958; Teece, 1988). Creativity is reached when specific constituencies, organizations, groups or individuals, break from the past. In the TV industry, artistic individuals are often motivated to engage in new projects instead of repeating the same formulas. Certainly not all individuals are motivated by the pursuit of novel approaches and some individuals may be willing to replicate previously successful formulas. Nevertheless, aside from the motives that energize individuals’ behaviors, most artists acknowledge that a main form of creativity in the TV industry lies in proposing contents never developed before. Before beginning our empirical research, we interviewed a well-known TV specialist in order to adopt a definition of creativity that has appropriate validity in the field. He told us: ‘When I think about creativity and new ideas, my usual metaphor is a cage whence I should flee. The cage is a cube deriving from what I have developed in the past.’
Deviating from partners
A second way to identify creativity is with respect to the social space dimension and conceptualizing it as deviation from partners. Previous research argues that creativity can depend on the knowledge pool matured within the organization but also on the knowledge pool residing externally to the organization, which can be accessed through its network ties (Ahuja, 2000; Baum et al., 2000; Fleming et al., 2007; Obstfeld, 2005; Rodan and Galunic, 2004; Rowley et al., 2000; Ruef, 2002; Uzzi and Spiro, 2005). Building on the idea that creativity can relate to deviation from the conformity of the social circle an organization or a group is embedded in (Ford, 1996; George, 2007), we identify another type of creativity by virtue of collaborative ties and define it as deviation from what partners are doing. Again using the example of painters, impressionists all pursued the same artistic ideals, but for one painter to be assessed in terms of ‘his own’ artistic contribution, he had to somehow differentiate himself from the works of others and be acknowledged for his distinctiveness with respect to his group of reference. In different fields, and especially in the artistic field, individuals working in a team may wish to differentiate themselves from their social circle: creative industries value creativity per se, independence from the influences of others and the capacity to propose a differentiated and unique contribution (Caves, 2000). In the TV industry, it is not uncommon for specialists and artistic individuals to attempt to express their own conceptions and to value the possibility of differentiating their creative work from the work of the social circle they are embedded in. Creative industries are highly symbolic and individuals may benefit from presenting their own artistic distinctiveness to their social group and peers. The same TV specialist mentioned to us that creativity is also differentiating one’s work from ‘what partners, friends and colleagues to whom I am connected are currently doing’. Does deviation from partners exercise a similar effect on performance as deviation from the past? The following section of the article offers answers to this question and elaborates a theory that first links networks to deviation from the past and partners and then explores how these two different dimensions of creative outcomes are beneficial or deleterious to performance.
External ties and deviation from the past and from partners
The number of external ties in a team may have an effect on deviation from past projects. The external connections individuals in a team have with other teams in other organizations positively influence the social judgment of those individuals. Individuals tied to many others are judged as prominent, meaning that they are perceived as highly visible and as those who should be followed (Kilduff and Krackhardt, 1994; Zaheer and Soda, 2009). According to Brass (1984), individuals who have many connections are judged to be influential, in the sense that they are perceived to have pull, weight or clout in a social environment. For Wong and Boh (2010), ties are a vehicle of positive referrals that create a favorable reputation as reliable and dependable individuals. Following a rational and functionalist logic, the social judgment of individuals or organizations should directly derive from the established perception of a specific attribute, such as cumulative prior performance, but relational and sociological perspectives have highlighted that the formation of social judgment and reputation is only in part the result of rational explanations and directly depends on social relations (Anderson and Shirako, 2008; Benjamin and Podolny, 1999; Bitektine, 2011; Fombrun and Shanley, 1990; Rao, 1994; Rindova et al., 2005). When there is a favorable social judgment, a perception of prominence or a positive reputation, individuals in organizations are unwilling to promote strategic actions that deviate from what has been done in the past since deviations create ambiguity, destabilize perceptions and can thus compromise a favorable social judgment (Anderson and Shirako, 2008; Bromley, 1993; Deephouse and Carter, 2005; Pfarrer et al., 2010; Phillips and Zuckerman, 2001). External ties may therefore hinder deviation from the past.
There is another mechanism explaining the relationship between external ties and deviation from the past that is related to the conceptualization of external ties as simultaneous participation in different work projects. The more individuals simultaneously work in teams involved in different projects, the less time they have to devote to exploratory thoughts, to search for new information and to use such new information in ideas never developed before (O’Leary et al., 2011). When individuals simultaneously participate in other projects, the group faces difficulties in sharing and integrating information to generate new knowledge and ideas (Wilson et al., 2007). Individuals working in multiple teams will attempt to search for efficiency in processing information and hence prefer utilizing the same knowledge already employed before (O’Leary et al., 2011). As a result, teams are less likely to deviate from the past.
It is important to acknowledge that network ties can also be a vehicle for knowledge sharing between groups belonging to separate organizations and hence facilitate the generation of new insights and a possible deviation from the past. However, knowledge sharing is assumed to lead to benefits in the case of formal alliances (Ahuja, 2000; Baum et al., 2000; Shan et al., 1994). The knowledge sharing benefits of external individual ties for creativity are weaker because ties belonging to diverse cultural and normative environments could be prone to distortions and create noise (Perry-Smith, 2006; Tushman, 1977). Furthermore, network contacts only give the potential for accessing information, but they do not lead to knowledge sharing unless the individuals have motivation to gain specific knowledge (Reinolds et al., 2011). As we mentioned, externally tied individuals in a team are not motivated to deviate from the past and therefore they are unlikely to collect information that enables change. We hence anticipate that:
External ties may also have an effect on deviation from partners’ projects. Network ties can be an important source of imitative behavior among partners (Galaskiewicz and Burt, 1991; Greve, 1996; Soda et al., 2008). Highly tied actors may be imitated by partners because the external ties may be perceived as a signal that the actor is an expert in the field (Zaheer and Soda, 2009). Actors with many ties are perceived by others as highly legitimate, charismatic, expert, authorities in the field and hence those who should be followed (Balkundi et al., 2011; Krackhardt, 1990; Pastor et al., 2002). A highly tied team in an organization may be imitated by partners because imitation is likely to be fostered when organizations are connected to each other through individuals crossing organizational boundaries (Galaskiewicz and Wasserman, 1989). Highly tied actors may also be imitated by partners because, as previously explained, they are likely to obtain a positive social judgment and reputation. Positive social judgments and reputations can be transferred from the highly tied actor to the partners, and partners are motivated to facilitate their transfer since they directly benefit from them (Benjamin and Podolny, 1999; Kilduff and Krackhardt, 1994; Podolny, 1994). As the transfer of social judgments and reputations is enabled when actors are perceived to be similar, partners can engage in imitative behaviors (Haunshild and Miner, 1997; Lieberman and Asaba, 2006). The highly tied team can be imitated by partners so that they can benefit from ‘basking in reflected glory’ (Cialdini et al., 1976). As a consequence of imitative behaviors, teams with many external ties are likely to exhibit lower deviance from partners.
In addition to the effect of imitation by partners, there is another intervening mechanism. More precisely, if the link between two teams is a result of sharing a specialist (i.e. the director or screenwriter) and the specialist is simultaneously engaged in multiple projects, the probability that (s)he will simultaneously use the same artistic approach in the different projects is tangible. In fact, the time pressures and coordination difficulties resulting from participating in multiple teams will make individuals search for efficiency and adopt the same approaches across multiple projects (O’Leary et al., 2011). As a consequence, the teams linked by the co-membership of highly tied specialists will show low content diversity and produce low relative novelty. Accordingly, we hypothesize that:
Internal ties and deviation from the past and from partners
Not only do external ties exercise an effect on deviation from the past and from partners, but previous internal ties may also be relevant. Previous ties reduce risk by fostering the creation of stable and preferential relationships characterized by trust and rich information exchange (Gulati, 1995; Gulati and Gargiulo, 1999; Li and Rowley, 2002). Team stability allows members to learn about each other’s competencies, create familiarity and mutual understanding and help reduce risk and uncertainty (Beckman et al., 2004). However, new members who do not share previous ties can contribute to creative processes by providing a broader range of opportunities and new approaches that refine established knowledge, ideas and organizational routines, rendering them more versatile (Powell et al., 1996). Members who share previous ties take as reference that which they have done together in the past and develop inertial attitudes (Rollag, 2004). According to March (1991) old-timers in a group know more, but are less likely to push the group towards deviating from what has been done in the past since they abide by an established organizational code. Conversely, members who do not share previous internal ties are more prone to foster deviation with respect to the past. Individuals who have never worked together bring enthusiasm and naïveté that pushes the group towards deviating from what has been done in the past (Perretti and Negro, 2006). The generation of new creative ideas can derive from the fresh insights brought by newcomers to the team (Ferriani et al., 2009). Overall, the presence of individuals who do not share previous ties gives the focal team an opportunity to generate new contents with respect to what the individuals in the team have done in the past. Hence, we hypothesize a negative association between the presence of individuals with previous internal ties and the deviation from past projects:
The presence of team members with previous internal ties is also likely to affect deviation from partners’ projects but for a different reason. Teams with individuals with previous internal ties are likely to assimilate more intensely information from their current partners. When individuals have never worked together they lack the common experience that enables them to use information gathered from outside in-group decisions (Ferriani et al., 2009). Individuals who have previously worked together transfer knowledge to each other and trust each other’s sources of knowledge (Gulati and Gargiulo, 1999; Hansen, 1999). When team members know each other, they have gained the trust and reciprocal expectations that enable them to absorb and internally use knowledge from their external contacts, while if team members have never worked together, they do not assimilate their external knowledge for internal group decisions (Oh et al., 2006). According to Ancona and Caldwell (1992), internal processes and external relations co-evolve: when team members do not know each other, internal dissent transfers its effects to restricted and confused relationships with external partners, while when team members know each other they increase the intensity of information exchange and assimilation from external partners. Furthermore, when team members simultaneously work for projects in other teams, they are likely to face coordination difficulties, inefficiencies and disruptions in work schedules leading to a poor integration of knowledge from other projects (O’Leary et al., 2011). Previous common experience could hence be beneficial to integrating information since it facilitates coordination, efficiency and harmonization of work schedules (Goodman and Leyden, 1991; Jehn and Shah, 1997; Okhuysen, 2001). Teams can be motivated to intensely absorb knowledge from partners, because it leads to inter-team trust, better collaboration, efficient coordination and sharing of tacit practices (Hansen, 1999; Hansen et al., 2005). However, intense absorption of information from partners is not likely to make partners dissimilar but, to the contrary, it is likely to facilitate imitative processes and foster similarity with partners (Galaskiewicz and Burt, 1991; Galaskiewicz and Wasserman, 1989; Greve, 1996; Westphal and Zajac, 1997; Westphal et al., 2001), negatively influencing deviation from partners’ projects.
Note that there could also be a counter-argument. Hansen et al. (2005) found that previous experience could generate inward looking biases and other authors argued that the familiarity of members could lead to groupthink and insulation from external influences (Janis, 1972; Katz, 1982). Nevertheless, the effects of team familiarity on exchanges with the external environment seem to be contingent on time: when teams are formed and continuously modified for temporary projects, prior familiarity could facilitate communication exchanges while the deleterious effects of groupthink and insulation start to emerge only if stable teams spend a great deal of time together (Guzzo and Dickson, 1996; Harrison et al., 2003). Given the temporary characteristics of project organizations, this counter-argument is ruled out and we anticipate that:
Partner characteristics and deviation from the past and from partners
We explored the effects of both external and internal ties on deviation from the past and from partners. Another predictive factor that we could consider relates to the characteristics of network partners. Network partners could be heterogeneous and involved in dissimilar projects or homogeneous and involved in similar projects. Access to diverse ideas and contents generates opportunity recognition and allows the focal actor to become a catalyst of new ideas (Rodan and Galunic, 2004). Access to partners with diverse and non-redundant knowledge offers the possibility of recombining such knowledge in innovative ways, resulting in novel insights and creative outputs (Burt, 2004). As Powell at al. (1996) argue, accumulating diversified knowledge from alters provides leverage to access, assimilate and exploit additional ideas. Diversity of information accessed through partners enables opportunities for behaviors never performed in the past, while similarity of information accessed through partners fosters the repetition of similar patterns of behaviors performed previously (Burt, 1992). Holding the number of external connections constant, the variety of team partners gives diversified inputs which increase learning opportunities and consequent possibilities of developing new perspectives (O’Leary et al., 2011). In our context, the connection with alters who hold heterogeneous content and ideas can significantly improve the focal team’s likelihood to recombine them in a more original way. The recombination mechanism has an effect on deviation from past projects as it brings new insights, which help teams break from inertia. We theorized that the presence of individuals with many external ties reduces creativity because of the constraining force of networks. Nevertheless, given the same amount of external ties, heterogeneity of partners facilitates deviation from the past:
The recombination mechanism could also have an effect on deviation from partners. It is reasonable to assume that an idea that results from the combination of similar sources is likely to incorporate analogous perspectives and be similar to its sources. Conversely, an idea that results from the combination of diverse sources is likely to incorporate multifaceted perspectives and hence differ from all of them. Partners are likely to transfer their worldviews and opinions to their contacts (Rice and Aydin, 1991): if all partners consistently and regularly transfer the same opinions to the focal team, we could expect the team’s deviation to be lower. Similar partners also share a common language and increase the incentive for the focal contact to develop and utilize the same language in order to be able to communicate effectively (Reagans and McEvily, 2003). Furthermore, a homogeneous social environment is likely to share common social norms and may thus exercise stronger pressure to conform to others (George, 2007; Woodman et al., 1993). As a consequence, heterogeneity may lead to higher deviation from partners while homogeneity may lead to lower deviation from partners:
Structural holes, partner characteristics and deviation from the past and from partners
Research in social networks showed that not only considering the partners a focal ego is connected to matters but accounting for how those partners are connected to each other also matters. According to Burt (1992), when an individual is connected to partners who are not connected to each other, the individual acts as a broker and spans structural holes. The literature has developed antithetic empirical positions on the question of whether structural holes benefit or hinder creativity (Fleming et al., 2007). Considering the characteristics of partners offers the possibility to address the empirical controversy on the role of structural holes. Does brokerage pay off if partners are homogeneous? Structural holes can trigger beneficial advantages because they allow access to diversified knowledge and control (Burt, 1992, 2000). Diversity in knowledge owned by partners, precisely as we theorized in the previous hypotheses, is likely to have beneficial effects on deviation from partners and from the past. However, if we consider bridging homogeneous partners, the advantage of accessing diversified knowledge is eliminated. The only advantage that remains is control. Nevertheless, the value of control may be contingent on the homogeneity of bridged partners. Control gives power to brokers because they can filter and affect the flow of non-redundant information between partners (Burt, 1992, 1997): if partners already have similar knowledge because they are homogeneous, the power of controlling the flow of information between them inevitably weakens. Hence, when partners are homogeneous the advantages of bridging many structural holes disappear.
Conversely, the condition where fewer structural holes are bridged among homogeneous partners becomes relatively more advantageous. Closed networks, lacking structural holes, affect the recombination of knowledge in ways that have not been tried in the past since they allow developing trust, decreasing opportunistic behaviors and thus facilitate quick and effective information exchange (Fleming et al., 2007). The quality of information exchange is likely to be beneficial to creativity because it facilitates sharing, correcting and improving ideas (Milliken et al., 2003). Breaking from the past and adopting new ideas requires trust so that individuals who lack the knowledge to assess ideas from others remain confident and endorse new directions never followed in the past (Levin and Cross, 2004). In addition, creative outcomes often require tacit and complex knowledge, which could be more easily transferred in a strong and closed network (Hansen, 1999). For these reasons, we hypothesize that, in the case of homogeneous partners, networks with fewer structural holes are relatively better for creativity than networks with more structural holes. Hence, structural holes would be negatively related to deviation from the past:
In all previous hypotheses, we have advanced convergent effects on both deviation from the past and deviation from partners. Differently from the previous cases, structural holes may exercise divergent effects on the two dimensions of creativity. Homogeneous partners are likely to share a common language and can create social norms that push towards social conformity (Reagans and McEvily, 2003). Closed networks lacking structural holes allow the application of shared social norms among partners and enhance pressure for social conformity because the behaviors of each partner can more easily be monitored and controlled by all (Coleman, 1988). In such networks, an actor would have less incentive for deviating from social norms since deviance and inobservance of the social norms are more likely to be sanctioned (Burt, 2000; Coleman, 1988). Hence, it is plausible to assume that an actor embedded in a closed network of individuals performing homogeneous activities will perceive pressure to conform to the group of common partners and execute similar tasks to those of the partners. Diversely, when an actor occupies a position bridging structural holes, the partners cannot exercise any constraining force on behavior due to the lesser capacity to monitor and observe such behavior, leaving the actor in a condition of greater autonomy (Burt, 2000). For this reason, we anticipate that the structure of social relations, differently from the other network variables we observed before, is likely to exercise divergent effects on the two dimensions of creativity and, more specifically, that structural holes will be positively related to deviation from partners’ projects:
Deviation from the past and from partners and performance
Innovative and creative outcomes, intended as novel and useful, are generally assumed to positively explain performance because they facilitate differentiation (Damanpour, 1991; Porter, 1985), although they could also exercise a negative causality as they generate consumer discontinuity (Calantone et al., 2006; McNally et al., 2010). However, deviation from the past is a distinct construct that does not rely on differentiation and discontinuity mechanisms since a project never developed before by the team does not imply that the project is new to the market and to consumers. Deviation from past projects does have a beneficial effect on project performance because of the motivational effect on team members. Getting involved in new projects that have never been developed before empowers the benefits of creativity, creates enthusiasm and excitement that justify greater effort and superior success (Amabile, 1997; Amabile et al., 1996). Engaging in activities never performed before is intellectually challenging for individuals, who respond to the pressure by increasing the effort in performing such activities (Amabile and Gryskiewicz, 1989; Amabile et al., 1996). When individuals develop initiatives for projects never executed before, they gain a sense of personal responsibility, which energizes their motivation for accomplishment (Frese et al., 1996). More specifically, in cultural industries, individuals feel a strong need to express their creative talent (Caves, 2000; Elsbach and Kramer, 2003; Lampel et al., 2000; Perretti and Negro, 2007) and the involvement in projects never developed before can facilitate the expression of unexploited creative potential. The effect on project performance is explained by motivation activated on the individual level as well as on team-level effects. New projects create a positive environment in which team members reinforce each other’s enthusiasm, creating positive team spirit, a feeling of team potency and effectiveness (Campion et al., 1993; Pirola-Merlo and Mann, 2004). According to theories on the social psychology of teams for creativity, when team members perceive they are actively engaged in the search and development of new ideas never implemented before, a favorable team climate emerges, motivating efforts towards the pursuit and the successful accomplishment of the ideas (Anderson and West, 1998; West, 1990).
To be noted is that previous research suggests that too much deviance is deleterious because organizations should balance exploration and exploitation for long-term profitability (Gupta et al., 2006; March, 1991). However, project organizations are different. First, being temporary, they are not concerned with long-term profitability. Second, project forms of organizing are employed in industries characterized by a flexible, unstable and ephemeral nature, where exploration of new possibilities is far more critical than exploiting past resources (Bechky, 2006). Third, project organizations usually do not need to rely on the exploitation of previously developed technologies or have a large resource endowment to exploit. Last, in cultural industries, teams can substantially deviate from the past without necessarily breaking artistic and aesthetic conventions and still exploit the procedural experience matured before (Lampel et al., 2000; Mezias and Mezias, 2000). Many other factors could affect performance, but these considerations make us assert that, ceteris paribus, deviation from the past positively leads to project performance:
The relationship between deviation from partners’ projects and project performance is more complex. The literature argues that collaborations lead to performance through the acquisition of critical resources that would not easily be available to organizations (Baum et al., 2000; Gulati, 1998). If critical resources necessary for high performance are owned by another organization, inter-organizational collaboration could allow accessing such resources and hence improve performance (Dyer and Singh, 1998). Under this logic, it would be reasonable to assume that if an organization partners with organizations performing activities of a different nature, those partners may own critical resources that the organization does not possess and the partnership may thus be more beneficial to performance (Gulati, 1995).
The argument that resource sharing positively explains performance in network collaboration is built on the assumption that companies own most resources needed to conduct their business. However, in project-based industries most resources are not owned by any project organization: project organizations mobilize a large network of relationships with suppliers, contractors and agencies to assemble the external resources necessary to implement the project (Bechky, 2006). In the case of the TV industry, each project team may contract with over a hundred actors among suppliers, contractors, public agencies and distributors, to gain access to the resources needed to implement the TV project. We argue that in cases in which most resources needed to implement the project are external to the partnering organizations, performing similar projects to those of partners could be beneficial to performance. In fact, collaborations may lead to a higher scale and increase negotiation power with suppliers and contractors (Galaskiewicz, 1985; Porter, 1985). If individuals in teams are simultaneously working on homogeneous projects and need to procure similar resources, they can negotiate better conditions with suppliers, contractors and external agents. For example, if a special effects specialist works on three science fiction TV productions, (s)he can contract with the same bundle of suppliers, save time in the search process and negotiate better contractual conditions that would allow staging better productions. In the same way as teams can share the contacts they have to access external resources (Gulati, 1995; Gulati and Gargiulo, 1999), teams performing similar projects can share their connections to mobilize the necessary resources to implement the project.
Furthermore, creativity may be ineffectively implemented in successful projects if the conditions for the implementation of creative ideas are lacking (Amabile et al., 1996). Deviating from partners may be a core condition that decreases the possibility of creative ideas being converted into economic success. When an actor is linked to actors performing tasks that are different from his/her own, an intense flow of knowledge could follow, but such knowledge might not be effectively implemented in the projects and hence creative outcomes might not be translated into success (Damanpour, 1991; Duncan, 1976). Partners with dissimilar content may provide novel information but partners with similar content can positively affect performance since they allow efficiently transferring information and effectively applying it to implement projects (Zaheer and Soda, 2009). Dissimilarity with team partners may decrease overall productivity because it leads to high complexity in managing the novel knowledge acquired from partners (O’Leary et al., 2011). It is possible for teams to have conditions of high creativity and low performance when diversity of knowledge accessed does not allow successful implementation of creative ideas, thus leaving the creative potential unexploited (Ancona and Caldwell, 1992). For the above-mentioned reasons, we hypothesize that:
Methodology
We tested our hypotheses studying TV production organizations. More specifically, our dataset includes made-for-TV movies, mini-series and made-for-TV specials produced and broadcast in Italy during the time period specified. We used two kinds of data, relational and content. We collected relational data over time on TV movie teams from the Internet Movie Database Professional (ImdbPro). This database has been used in previous studies (Sorenson and Waguespack, 2006; Zuckerman et al., 2003) and is widely recognized by industry experts as the most complete database on the movie industry. Furthermore, from Imdb and multiple independent sources we also assembled information on the script synopsis of each TV movie that we analyzed through content analysis (Berelson, 1952; Holsti, 1969).
We used the social network methodology, creating bipartite-affiliation networks for the industry specialists involved in TV production teams. The network is therefore composed of relations of shared membership of specialists. From a focal team’s perspective, the network structure is composed of ties that link a team to other teams by virtue of sharing individual specialists. A project team is linked with past and current projects by having one or more specialists who are or were involved in other projects through which they can gather numerous ideas and information. Since teams are composed of several specialists, including those who perform technical tasks and are not involved in the creation of productions, we selected only those with key-roles instead of all specialists involved in TV movie production (Baker and Faulkner, 1991; Perretti and Negro, 2006). These included directors, screenplay writers, original authors, producers and actors in major starring roles.
The TV production team is our level of analysis. Our dataset contains 385 TV movies produced and broadcast over the period considered in Italy. In order to compute our variables we subdivided the data with 2003 as cut-off year. As a result, our sample is composed of two production groups: 208 TV productions produced in the period 1999–2003, and 177 TV productions produced between the years 2004 and 2008. We selected the 208 productions from 1999 to 2003 to compute past measures and used the 177 productions from 2004 to 2008 to compute our creativity variables, performance and the remaining current independent variables. We subsequently opted for a five-year window of the past. 1 The choice of time window mirrors previous studies (Ahuja, 2000) and is supported by literature on film production teams (Soda et al., 2004, 2008). Moving the five-year window across multiple years enabled us to use the same time span for all productions broadcast in the 2004–2008 period (e.g. for a production broadcast in 2004 we used past network data from 1999–2003). For our entire network, we collected individual data on 2574 specialists and aggregated it at the project/team/organization level, encompassing 385 TV productions.
Econometric approach
To test the antecedents of the two types of creativity and their implications for performance we use a two-stage least squares (2SLS) model with a robust variance estimator to handle potential correlation among error terms in the equations. More precisely, because our dependent variable of performance (audience share) is bounded, we used a Tobit specification of 2SLS (Tobit two-stage least squares, Tobit with endogenous covariates on STATA), which provides more consistent estimates. In order to check the consistency of our econometric approach we used several tests. We began with the Wu–Hausman F-test and the Durbin–Wu–Hausman χ2 tests for endogeneity. These are tests for endogeneity where the null hypothesis states that an ordinary least squares (OLS) estimator of the equation would yield consistent estimates, and thus endogeneity among the regressors would not have deleterious effects on OLS estimates. Moreover, we also checked for the presence of over-identification of our model with the Sargan test, which provides a measure of instrument relevance for all instruments. In our 2SLS model, we treated deviation from the past and deviation from partners as endogenous explanatory variables. The network predictors were treated as instrumental variables and performance was treated as a dependent variable.
Analysis of TV production content
In this article, we adopt an objective measure of creative outcomes derived from content analysis of TV productions. Since creativity is a highly complex and intangible phenomenon, scholars often prefer using subjective assessments of outcomes characterized as novel and creative (Amabile et al., 1996). Objective measures of creativity are not uncommon, especially to observe variance across organizational boundaries (Nerkar and Paruchuri, 2005). Subjective evaluations, however, have the strong advantage of grasping the intangible nature of the construct yet encounter problems when it comes to assessing products or projects for which there are opposite points of view on what constitutes novel and useful work (George, 2007). In creative industries, the definition of creative work is highly controversial (Caves, 2000): almost everyone is convinced of doing work that is somehow creative and has differing opinions on what should be considered as creative. A solution could be to use subjective ratings from external sources, such as critical acclaim (Uzzi and Spiro, 2005). External subjective ratings could enable good measurement but are problematic when measuring creativity as deviation from the past and deviation from partners. Critical acclaim offers an indicator of ‘creative success’ enabling a comparison of success over time but without assessing deviance over content. If we want to reliably capture deviance in creative content, comparing measures within time and within the social space, we need measurement consistency. Objective indicators, though not immune to limitations, allow good and efficient comparisons across time and space, hence offering good operationalization for the constructs of deviation from the past and from partners.
To capture the differences or similarities across our TV productions, we studied their content by using content analysis of the scripts. The aim of the content was to assess TV team expressions of artistic identity and the messages they wished to communicate through their productions. Content analysis can be a powerful tool to determine identity since authors synthesize their artistic identity and the messages they wish to communicate through their writing. Thus, in our research context, the script is a key document for sharing, communicating and understanding the TV team’s meaning and artistic identity. In the script, the movement, actions, expression and dialogues of the characters are also narrated.
Content analysis is a procedure that is widely employed as a replicable and systematic means of collapsing a large amount of text into content categories based on a set of explicit coding rules (Berelson, 1952; Krippendorff, 1980). We adopted an a priori coding of sampled TV productions where the content categories were established prior to the analysis based on the recommendations of industry experts. 2 In a similar sample and in the same cultural and market context, Zaheer and Soda (2009) used the following procedure that we have adopted. They reviewed the TV movie scripts and selected 19 content categories. Thereafter, they validated the list of movie content categories with a panel of six industry experts. The six experts helped to minimize the overlap among the categories and the initial list of 19 categories was reduced to 11 (see Appendix for details of categories and variables). In our research, we used the same list of categories and checked its validity using a sample of 100 scripts from our dataset. Then we checked the validity of content variables with a new panel of six industry experts, verifying that they differed from those who participated in the validity analysis of Zaheer and Soda (2009). The entire panel agreed that the 11 categories captured the identity, content and essence of the productions, providing an accurate and valid measurement of production content. They also confirmed that the categories could provide accurate, valid and reliable operationalization of content deviation. Finally, two independent researchers used the list of categories to code the production content for all TV productions included in our dataset. The inter-rater reliability across variables – Cohen’s kappa (Cohen, 1960) – measuring the degree of rater agreement was very high at .91.
Two-stage least squares analysis (2SLS) with Tobit: First stage variables
Endogenous variables
Deviation from partners
We computed this variable for each TV production team by measuring the dissimilarity of the production content between the focal team and its current alters. We used the 11 content variables mentioned above, assessing the content of TV productions of the focal team and its partners. For each focal team we extracted its ego-network and computed a two-mode matrix Productions by Contents with the n of rows equal to the ego-size (including the ego) and the m of columns equal to the 11 content categories. Thereafter we used the UCINET VI exact matches dissimilarity procedure adopting the measure of dissimilarity as ‘non-matches’ that computes the proportion of cases where xi ≠ yi for all i (Borgatti et al., 2002). The program extracts from the Productions by Contents matrix another matrix, the dissimilarity matrix, which describes the dissimilarities between ego and any partner. Finally, we measured the extent of product content deviation through the dissimilarity index and calculating the average content dissimilarity between the focal team and its partners.
Deviation from the past
We computed this variable for each TV production team by measuring the dissimilarity of the production content between the current focal team and past productions. The procedure to assess the past content deviation is the same as that used for the deviation from current alters. Naturally, instead of the TV productions of current partners, we used the past TV production of each focal team comparing the contents produced concurrently and the contents produced in the past. We built a similar Productions by Contents matrix, extracting through UCINET VI a dissimilarity matrix, and calculated the dissimilarity index with relation to the past and not to the current partners.
Instrumental variables
External ties
This measure calculates the number of external ties of team specialists working in other TV productions weighted by the number of team specialists in each TV production team. We could hence eliminate the effect of team size. We operationalized the measure so that it could have a value equal to or greater than 1. It is equal to 1 when industry specialists working in a TV production team are only working in that specific team, while it is greater when TV production teams share industry specialists.
Previous internal ties
This measure captures the extent to which the focal team is composed of members who know each other or are collaborating for the first time in a TV production. We compute the proportion of specialists who did not have common previous experience with the other teammates.
Network heterogeneity
In order to measure access to partners who held diverse ideas and content we measured the dissimilarity of the production content among alters of each focal TV production (including the ego). For each ego, we assessed the content/idea heterogeneity by adopting the same dissimilarity procedure used for the two dependent variables. We calculated the average dissimilarity among partners and not the dissimilarity of ego with its partners.
Structural brokerage of homogeneous partners
Following our theorization, we wanted to disentangle the potential effect of network brokerage from that of content heterogeneity. The computational effort for this variable was relevant: from each ego TV-team network we extracted all alters with an average content/idea diversity – within the ego – above the average content/idea diversity in the entire population (> .17). Thus, the new ego-network is restricted to alters with low content/idea heterogeneity. Using this ego-network, we computed brokerage for each ego using the reverse of the constraint measure of structural holes (Burt, 1992). Essentially the measure assesses the structural holes spanned by a focal team among similar alters, thus teasing out the effect of diversity from the potential impact of brokerage.
Controls
Since the internal ties variable uses past collaboration among team members, we controlled our models for tie duration, computing a weighted average of the duration of team members’ past relations. We also controlled for the size of the TV production team since team size was considered the simplest predictor of propensity towards creativity and novelty; in fact, new combinations of ideas are enhanced when a number of individuals work together in applying multiple perspectives to a problem (Ruef, 2002). We controlled for specialists’ team past performance since we believe that TV production teams employing high performer specialists could have a consequence on the two realms of deviation. Since our teams are composed of several specialists, including those performing technical tasks, we decided to use only those specialists with key roles. To compute this measure, we added up the performance of past alter teams to calculate the past performance of focal teams. More recent successes were weighted by a decay function based on the age of ties, since the effect of more recent successes is greater than that of older successes. A further control is the deviation from past industry pressures, considering a sort of path-dependency in creativity. This measure captures the extent to which the focal production deviates from the dominant production content in the industry in the past. We calculated the average content dissimilarity between each focal team and all productions produced and broadcast in the past two years. We also controlled for deviation from industry dominant production content, which captures the extent to which the focal production deviates from the dominant production content in the industry. We calculated the average content dissimilarity between each focal team and all productions produced and broadcast in the same year as the focal production. Last, we controlled for time specific effects, including a dummy for each observation year.
Two-stage least squares analysis (2SLS): Second stage variables
Dependent variable: TV movie performance
The movie audience share is the most important economic performance metric. Audience share data are collected in Italy by an independent institution (Auditel). Given the highly skewed nature of audience numbers, we use the natural log of the audience share that watched the TV show as our measure of performance and our dependent variable.
Controls
Several variables were used in the second stage to capture the real effect on performance of our endogenous creativity variables. We considered six TV channels (95% average of the total audience covered), although they do not have the same market power and brand equity in reaching high audience levels. Two major channels (RAI 1 and Canale 5) have historically higher viewership than the other channels. Accordingly, we included in our model a dummy variable for major channel, which is set to 1 when the production was shown on either major channel and 0 otherwise. Moreover, the time slot in which the TV production is broadcast affects the audience share. Since the highest potential viewership occurs in primetime, which in Italy is 8:00–10:30 p.m., we also included a dummy variable to control for primetime (1 when the TV production was broadcast in primetime and 0 otherwise). We also took into account the number of episodes of the TV productions that may reveal something about the balance between creativity and routine dimensions. In order to control for this characteristic we computed a number of episodes dummy equal to 1 when the production is a one-episode TV movie and 0 when more than one. Further, we controlled for time effects by using a series of dummies corresponding to the years of our observation for the dependent variable. Finally, the size of the team is considered a reasonable proxy for the production budget, thus we also controlled for this. The overall model and the hypothesized effects are synthesized in Figure 1.
Illustration of the full model.
Results
We first tested for the appropriateness of treating deviation from the past and deviation from partners as endogenous variables by using the Wu–Hausman F-test and Durbin–Wu–Hausman test. Both tests rejected the null hypothesis that our regressors are exogenous to TV movie performance, suggesting that the use of 2SLS is appropriate to address the issue of endogeneity. Moreover, due to the use of several instrumental variables, we also checked with the Sargan test for potential over-identification in our model. Results indicate that the model is not over-identified. Table 1 reports the descriptive statistics and correlations among our variables and shows that the two endogenous variables, deviation from partners and from the past, are modestly correlated to each other. Likewise, there are significant correlations among independent variables, but such correlations tend to be sufficiently low to exclude any serious risk of multi-collinearity.
Descriptives and correlations.
Table 2 reports the results of the first stage for both deviation from past projects and deviation from partners’ projects. The results support the prediction that the presence of team members with several ties external to the focal project reduces the propensity at the team level to deviate from the past production (External ties β = –.042, p < .001). The same effect is generated on the deviation from current alters’ projects (External ties β = –.172, p < .001). While the negative effect of previous common experiences among team members (Previous internal ties) on deviation from alters’ projects is confirmed and robust (Previous internal ties β = –.036, p < .001), we did not find a significant effect on deviation from the past. The distinction between structural brokering and heterogeneity among alters brings interesting results. By virtue of knowledge and idea recombination, the heterogeneity of partners leads a focal team to deviate from their contents (Partner heterogeneity β = .351, p < .05). The effect of network heterogeneity on deviation from the past is also positive but marginally significant (Partner heterogeneity β = .388, p < .1). Last, bridging structural holes between homogeneous partners seems to exercise divergent effects on the two dimensions of creativity. In fact, while empirical results support that structural holes negatively predict deviation from the past (Structural holes of homogeneous partners β = –.173, p < .01), they also support a positive although modest predictive effect on deviation from partners (Structural holes of homogeneous partners β = .098, p < .1).
Results of the first stage of the Tobit model with endogenous covariates.
Standard errors in parentheses.
p < .10; * p < .05; ** p < .01; *** p < .001.
The results of the second stage are presented in Table 3 and capture the effects of our creativity constructs on performance, which are consistent with our predictions. In fact, while deviation from the past helps teams to break with repetitive formats and contents, with beneficial influences on performance (Deviation from past β = 2.16, p < .01), the effect of deviation from partners on performance is significant and negative (Deviation from partners β = –.98, p < .01).
Results of the second stage of the Tobit Model with endogenous covariates.
Standard errors in parentheses.
p < .10; * p < .05; ** p < .01; *** p < .001.
These models were estimated using the robust variance estimator.
Discussion and conclusion
The present study investigated how two dimensions of creativity, deviation from past and deviation from partners, act as opposing mechanisms that intermediate the relationship between networks and performance. We built on the idea that creativity should not be only conceptualized as a one-dimensional variable (George, 2007; Unsworth, 2001) and elaborated a relative perspective on creativity, intended as deviation from the past and as deviation from partners. Focusing on project organizations in the TV industry, we found positive and negative, convergent and divergent effects on the two dimensions of creativity, opening up research on the social side of creativity. More specifically, we found that external ties and previous internal ties tend to exercise a negative effect on both dimensions of creativity, while network heterogeneity is shown to be positive for deviation from the past and from partners. While those three variables exercise convergent effects on the two dimensions of creativity, bridging structural holes between homogeneous partners exercises divergent effects, positively predicting divergence from partners while negatively predicting divergence from the past. Our results assume valuable importance as we found that the two dimensions of creativity exercise divergent effects on performance: deviation from the past is positively associated with performance and deviation from partners is negatively related to it. Hence, this article highlights a practical trade-off that project organizations may face when they have to strategically decide how to assemble teams for projects. Irrespective of the way teams are assembled in projects, as a function of internal ties, external ties or tie characteristics, there may be contrasting effects on performance. Organizations should hence assess the specificities of the situation and ponder the specific relevance of both creativity factors on performance to understand how teams with particular network characteristics could be advantageous or deleterious to performance. Conversely, bridging structural holes between homogeneous partners increases deviation from partners while decreasing deviation from the past, thus potentially hampering project performance.
Our results somewhat challenge a popular myth. They provide initial evidence that, in project industries, the ‘think different’ approach could be beneficial when related to the past but deleterious when related to partners. So should teams not try to differentiate themselves from their social circle? We do not wish to be assertive in this conclusion. However, we can suggest from our findings that in project industries there is high risk in deviating from partners. Certainly, if team members are particularly talented, they could transform the risk into opportunity and establish a breakthrough. Geniuses ‘think different’ and succeed. Nevertheless, one may need to be a genius to succeed thinking differently. Groups that are not particularly talented and want to be different simply as a result of the possibilities that their networks have enabled could achieve lower economic outcomes. Our results are far from being conclusive but at least, thanks to our conceptualization, start questioning the assumption that creativity is always beneficial to performance, giving rise to controversy that might open up new investigations.
This article offers substantial contributions to research in social networks. First, research in social networks is characterized by some controversy and debate in the empirical evidence related to performance (Brass et al., 2004; Rowley et al., 2000), and networks could activate mechanisms with opposite effects on performance outcomes (Reagans and McEvily, 2003; Reagans et al., 2004). We investigated these mechanisms referring to two dimensions of creativity and have helped disentangle the complex causality that links networks to performance outcomes. Second, our article offers substantial contributions to our understanding of how networks predict creativity. While previous research has mostly supported the positive effects of networks on creative outcomes (Ahuja, 2000; Burt, 2004; Perry-Smith, 2006), we provide evidence for possible deleterious effects stemming from the fact that relationships not only facilitate knowledge sharing, but could also exercise pressure to conform or lead to imitative behaviors. An interesting finding of our article is that the number of external partners constrains creativity, while the variety among external partners enables it. Third, our article contributes to the discussion around the advantages or disadvantages of structural hole positions. Previous evidence has produced contrasting affirmations on the effects of structural holes on creativity (Fleming et al., 2007). We found that, once the effects of access to diverse knowledge are teased out, structural holes could be deleterious to creativity. This finding could revise our previous assumptions on the nature of structural holes and fuel new research in the field. Last, our article provides rich insight on the nested nature of networks. Relationships between organizations are often the result of relationships between individuals belonging to different organizations: investigating multilevel relationships is particularly valuable for network research but there have been few empirical investigations due to difficult operationalization (Brass et al., 2004; Moliterno and Mahony, 2011).
Our data show a possible negative effect of a dimension of creativity. Why would individuals, teams or organizations pursue creative outcomes that negatively predict performance? As a matter of fact, there are several cases in which constituencies may be interested in pursuing creativity even when it could hamper performance. In our case, teams have individuals with high artistic values. Those with external relationships may be motivated to keep their status quo and do not deviate, while those without external relationships may be more motivated to deviate, to manifest their artistic independence, even if it could be deleterious to economic performance. Individuals in groups and in organizations are certainly not only motivated by economic values and may enjoy being different even if it hampers performance. We can also think of scientists or any creative job that has strong intrinsic motivations. We must acknowledge that our results are related to immediate consequences and do not account for long-term consequences. Being different from the social circle may have deleterious effects at the beginning, but in the long term, the difference may establish a new order and drive change on a larger scale. It is not uncommon for creative geniuses to be outcasts at the beginning, facing economic distress, but then they become the point of reference and attain economic wealth.
Some of our findings, more specifically those exploring the relationship between creativity and performance, may be specific to the context of project organizations. Nevertheless, findings on the relationships between networks and creativity as deviations may be extended to larger and more complex organizations. The effect of social relationships on the judgment and reputation of organizations is particularly relevant (Bitektine, 2011; Rindova et al., 2005) and the constraining effect of external ties on deviation could be interestingly explored in other types of organizations. When it comes to larger and more complex organizations, the effect of previous internal ties on deviance could be applied to the study of top management teams. The divergent effects of structural holes on the two dimensions of creativity as deviation may potentially be extended to other organizations, since the mechanisms and effects of structural holes tend to be replicated across levels of analysis (Burt, 1992). As our article often employs individual-level processes to explain intra-organizational relations, multilevel studies that investigate how individuals nested in organizations link such organizations (Moliterno and Mahony, 2011) could interestingly build on our insights.
Our study has some limitations. The first limitation is related to our operationalization of creative outcomes. An objective measure of creative outcomes is limited in capturing the intangible nature of creativity. Content analysis has its limitations as teams could still deviate substantially from the past or from partners even if they perform the same type of TV production. Nevertheless, we believe that our measurement, although limited, captures good variation across teams related to two valuable, albeit not the only, dimensions of creativity. Another limitation relates to the industry we investigated. Particularly in the cultural industry, artistic identity plays a key role in survival and success, and its changes are very unlikely. Results on deviation from the past could mirror this industry particularity. Moreover, our theory is based on the assumptions that innovation and creative processes within teams reflect some internal team characteristics: we do not have specific measures capturing the creative process itself but only its outcome. A further limitation of our data is that we only used co-membership among teams as ties while teams can also be linked through other types of relationships.
Future research could pursue the expansion of the creative outcomes and explore other dimensions that characterize the concept of creativity in different fields. For instance, deviation from other reference points could be considered. Future research could also expand the investigation of predictors of deviation from the past and from partners. Teams and individuals may differ in their motivation to pursue creative or performance outcomes. Some teams may completely disregard creativity and focus on performance. Other teams could disregard performance and engage in creative behaviors even when they potentially hamper performance. This article highlights the need to explore the possible motives or individual differences that could have divergent effects on creativity and performance or moderate the relationship between networks and creativity. Having outlined possible trade-offs triggered by networks and associated with performance, scholars could attempt investigating the contingent situations in which networks mostly exercise positive or negative effects on performance.
To conclude, our article has extended research on networks and creativity offering an additional perspective to understand how networks could trigger mechanisms that are likely to exercise a contrasting final effect on performance. Considering an alternative conceptualization of creativity as deviation from the past and from the social circle of partners allowed us to explore the unique relationships with networks and with performance. We believe that there is still much more that needs to be developed to understand the multidimensional nature of creativity and the complex and divergent effects triggered by networks. Hopefully, this study will stimulate other works in the field and promote the expansion of our knowledge on networks and creativity.
Footnotes
Appendix
Summary of variables to code content
| Variable | Type | Content |
|---|---|---|
| Theme | Categorical | Detective; dramatic; life story; friendship; love; family; Bible themed; religion; sport; fantasy; power/money/career; others |
| Relations | Dummy | 1 for love, friendship, kinship, affiliation, affinity, consanguinity and liaison; 0 otherwise |
| Values | Dummy | 1 for human justice (e.g. story of a crime prosecutors), religion (e.g. stories about the life of saints), freedom and independence, social battles against evil (e.g. citizen or consumers against powerful organizations for environmental protection) or against social prejudice and discrimination; 0 otherwise |
| Pain | Dummy | 1 for stories of disease, suffering, conflicts; 0 otherwise |
| Power and success | Dummy | 1 for power, money, career, social elites; 0 otherwise |
| Profession of characters | Categorical | Dominant professions in which the story is set |
| Characters | Ordinal | Weighted number of protagonists, antagonists, secondary protagonists |
| Ending | Categorical | The nature of the epilogue: happy, ambiguous, unhappy |
| Categorical | Categorical | Context in which the story has been located: Italy, Europe, abroad |
| Time period | Categorical | Time period in which the story is set |
| Schema | Categorical | The conflict schema in the sentimental relationships (Holsti, 1969) |
Acknowledgements
We gratefully acknowledge the financial support of the Claudio Dematté Research Division at SDA Bocconi School of Management. We also thank the Canada Research Chair in Strategic Management in Pluralistic Settings and the Fonds québécois de la recherche sur la nature et la technologie for funding and support.