ناهمگونی سرمایه انسانی، روابط همکاری و الگوهای انتشارات در اتحاد علمی چند رشته ای: یک مطالعه موردی مقایسه ای از دو تیم علمی
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|18445||2004||18 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Research Policy, Volume 33, Issue 4, May 2004, Pages 661–678
In this paper, we compare the publication outcomes of two teams within a multi-university scientific alliance. Scientists in one team share similar scholarly backgrounds and work in a well established paradigm, while scientists in the second team have different backgrounds and work in an emergent discipline. While the alliance has increased the productivity of both teams, this increase was highest for the more heterogeneous team. In addition, while the variety of knowledge concepts employed in their research was initially higher for the heterogeneous team, this gap narrowed over time. We discuss the implications of our research for alliance design.
Science has become increasingly collaborative during the past several decades, and many new organizational forms have emerged to manage collaboration among scientists in productive ways (e.g., Chompalov et al., 2001). In their review of the literature on scientific collaboration, Katz and Martin (1997) argued that collaboration has been spurred by changing patterns of research funding, the professionalization of scientific personnel, the need to pool resources to address increasingly complex and expensive research questions, progressively more specialized scientific disciplines, new communication technologies, and the desire of researchers to enhance their professional visibility and productivity. While empirical research does indeed suggest that scientific collaboration has many desirable outcomes (e.g., Katz and Martin, 1997), it is also clear that collaborative work is difficult, expensive in both time and money, and entails non-trivial problems of coordination and communication among sometimes diverse scientists that can undercut even the best of intentions (e.g., Williams and O’Reilly, 1998 and Reagans and Zuckerman, 2001). In short, collaboration is very much a “double-edged sword,” (Milliken and Martins, 1996), and this double-edge becomes increasingly sharp as firms, universities, and governments spend billions of dollars each year to fund large scale interdisciplinary projects to expand the frontiers of knowledge. It thus becomes desirable, from both a policy and theoretical standpoint, to understand the dynamics of collaborative forms of scientific work and the key tradeoffs that such work carries with it. One form of cooperative endeavor that has become increasingly important in scientific research and development is the interorganizational alliance (e.g., Gulati et al., 2000 and Powell, 1990). Alliances are voluntary arrangements between two or more organizations involving “exchange, sharing, or co-development of products, technologies, or services” (Gulati, 1998). Strategic alliances can be formed at many different organizational levels, and at many different positions along an organization’s value chain. One common use of alliances, however, is to connect the research and/or development functions of two or more organizations in an attempt to capture the benefits of combining the scientific and technological assets of the alliance partners (e.g., Powell et al., 1996 and Hagedoorn, 1993). The most important assets in this regard are the stocks of specialized knowledge possessed by each partner organization. The motivation behind most alliances is to create the conditions for organization-specific knowledge to be transferred across organizations and combined in ways that lead to varied insights that would not be possible if each organization were pursuing research and development activities on its own. While research alliances and partnerships are perhaps most prevalent among private sector firms pursuing joint R&D activities (e.g., Hagedoorn et al., 2000 and Powell, 1990), one important outcropping of the alliance movement over the past two decades has been the formation of scientific alliances among universities intent on sharing and recombining the knowledge of their faculty and research scientists to advance disciplinary and multidisciplinary scientific objectives. Different forms of inter-university collaboration exist in different countries (e.g., Ballesteros and Rico, 2001, Wen and Kobayashi, 2001 and Okubo and Sjoberg, 2000), but in the US, university alliances have been spurred, in part, by the availability of government funds from the National Science Foundation (NSF) and other public agencies that have been earmarked for collaborative inter-university research. Good examples of these collaborative endeavors are NSF’s Science and Technology Centers focusing on topics such as nanotechnology, adaptive optics, and behavioral neuroscience. Other forms of inter-university collaboration are evident in the NSF sponsored Engineering Research Centers on systems engineering, optoelectronics, and advanced electronic materials processing (e.g., Feller et al., 2002). Each of these endeavors represents a multimillion dollar program of collaboration, usually organized around one or two lead universities, and is focused on the transfer and combination of specialized domain knowledge across university boundaries. As in other types of alliances, the goal of inter-university collaboration is to spur new insights in the domains covered by a particular alliance by bringing together researchers at different universities who otherwise would not, or could not, collaborate in their research. In this paper, we explore the dynamics of collaborative work in one large government funded scientific alliance incorporating researchers from 35 US universities and government laboratories. The stated purpose of the alliance is to create new scientific knowledge by deploying complex computational modeling and visualization techniques in a variety of disciplines through the use of high performance computing architectures and networks. While the activities of this alliance are quite varied and distributed, the focus of our study is two teams of scientists who were explicitly recruited to develop, explore, and promote high performance computing in their respective disciplines. These research teams vary in the density of their intellectual and social networks. Scientists in one team have longstanding relationships that existed prior to their alliance affiliation; they share similar scholarly backgrounds, and they work in a traditional area of research that is characterized by a well established disciplinary paradigm. Scientists in the second team have much more varied intellectual histories; they come from different scholarly backgrounds, and they work in a new and emerging discipline that is only now becoming established as a coherent body of knowledge. These differences in human capital make the two teams an ideal venue for exploring some of the dilemmas and tradeoffs of collaborative work in scientific alliances. In particular, the teams constitute naturally occurring comparative cases (e.g., Eisenhardt, 1989 and Yin, 1994) that can be used to assess how the mix of human capital influences the trajectory of alliance-based knowledge production over time. To this end, we will first develop the rationale for our study by reviewing arguments and evidence pointing to the inherent difficulty of knowledge production in interorganizational alliances. We will suggest that the complexity of joint knowledge production creates a set of countervailing social and cognitive forces that influence the amount and quality of the knowledge that is generated by alliance partners. These countervailing forces are intimately bound up with the characteristics of human capital in knowledge production teams. Following our development of this conjecture, we will explore the implications of our argument by empirically examining the publication histories of the two alliance teams that are the focus of our investigation. We will show how each team has a particular publication pattern, and we will suggest that these patterns are related to the intellectual backgrounds of team members. We will follow our analysis with a discussion of the implications of our research for alliance design and evaluation.
نتیجه گیری انگلیسی
Research alliances among universities, firms, and/or government laboratories have evolved as one interorganizational mechanism to combine human and technological capabilities in the service of scientific achievement. As such, alliances must come to grips with the countervailing forces that both encourage and discourage the movement of knowledge across organizational boundaries during joint knowledge production. Past research suggests that alliances are difficult to manage, and that the mobility of knowledge cannot be taken for granted (e.g., Gulati et al., 2000 and Fischer et al., 2002). Joint knowledge production seems to be facilitated when knowledge is codified into transferable representations, when the partners are experts in the relevant knowledge domains, and when the partners have a history of repeated interactions and intellectual relationships (e.g., Simonin, 1999, Lane and Lubatkin, 1998 and Khanna et al., 1998). Yet, these conditions seem to be the same conditions that homogenize the knowledge that is available to alliance partners in the course of their collaboration, perhaps limiting the scope of alliance accomplishments as well (e.g., Uzzi, 1996). If so, it appears that research alliances, as mechanisms for joint knowledge production, may be subject to a tradeoff between the amount and the variety of knowledge that is shared and combined by alliance partners. It was the possibility of this tradeoff, and the desire to understand how alliance partners manage the countervailing pressures involved, that motivated our explorations of Eco and Astro. It is tempting to portray the tradeoff between the amount and variety of knowledge stocks in research partnerships quite starkly. However, in unpacking the similarities and differences between Eco and Astro collaborations, our study suggests that the intellectual composition of Alliance partnerships was intertwined with Alliance outputs in subtle and complex ways. Specifically, the Alliance seems to have been beneficial for both teams of scientists, but for different reasons that parallel their unique configurations of human capital. For Astro scientists, the Alliance has been a funding mechanism to continue their collaboration in computational astrophysics. Ensconced in a single discipline with a strong theoretical paradigm, and already productive and routinized in their prior collaborations, Astro team members have increased their publication output by 22%, their journal variety by 21%, and the number of co-authored publications by 75% during the period of the Alliance’s operation. However, the conceptual variety of their publications, as measured by the content of their publication titles, has remained constant across the 12 year period covered by our data. Although Astro scientists have evolved over time in their topical foci within computational astrophysics, their published output does not reveal significant changes in the degree to which computational modeling per se has been emphasized in their work. Across the 12 year period, the key words “simulation” and “model” have been present in 8–9% of Astro publication titles. Thus, Astro scientists are adding to their already considerable research productivity while working within the Alliance, but the marginal increase in output appears to be an increment to “business-as-usual” as opposed to a fundamentally new direction in their work. For Eco scientists, on the other hand, the Alliance is a scholarly bet that combining their varied disciplinary expertise to produce large scale computational simulations of ecosystems will lead to new combinative insights and enhanced publication opportunities. Having only loosely collaborated on a pairwise basis prior to the Alliance, Eco scientists have faced the problem of developing new collaborative routines that allow them to work distributively on Alliance projects. Despite their self-assessments to the contrary, our data suggest that they have been successful in this respect. During the period of Alliance operation, Eco team members have increased their published output by 85%, their journal variety by 47%, and the number of co-authored publications by 66% when compared to pre-Alliance levels. Perhaps most significantly, however, concomitant with this substantial increase in research productivity and journal variety has come an increased focus of Eco publications around Alliance-related topical objectives. The frequency of both “simulation” and “model” in the titles of Eco publications has doubled across the 12 year period of our study. In short, Eco scientists seem to be using their increased productivity under the Alliance to embrace the new domain of computational modeling, and to disseminate the results of this new conceptual thrust via a broader variety of journal outlets. Given their overlapping intellectual relationships, it is not surprising that Astro scientists have been able to maintain, and even enhance, the momentum of their work under the auspices of the Alliance. In the case of Eco, however, it might be expected that the diversity of the team would have been hard to manage, and that the projects funded by the Alliance would have floundered. While we encountered subjective indications that the Eco team has struggled to find an effective way to coordinate their collective activities, the fact that over time the team has been able to simultaneously increase their journal productivity, variety, and focus on Alliance-related topics suggests that Eco scientists have learned to manage their diversity and align their work with Alliance objectives. This finding lends important support to the increasing use of interorganizational alliances as a mechanism for promoting collaborative scientific research. It suggests that alliance collaboration can be productive even when the alliance partners are intellectually diverse and working in areas that have yet to form strong disciplinary paradigms. Indeed, the conceptual change toward Alliance objectives evidenced by Eco scientists during 1996–2001 provides prima facie evidence that discontinuous knowledge productions can be achieved under the auspices of university scientific partnerships. Much more research is necessary to understand the conditions under which different configurations of human capital can be effectively managed within an alliance to enhance the amount and/or variety of knowledge productions. Both Astro and Eco can be considered alliance success stories, and it would be useful to investigate instances of alliance failures to enrich our understanding of how human capital diversity influences alliance performance over a range of outcomes. Moreover, even though we have argued that Astro and Eco represent two extremes along a continuum from disciplinary homogeneity to heterogeneity, more empirical research must be done to parameterize human capital in a way that makes judgments of homogeneity and heterogeneity less ad hoc. In a recent unpublished paper, for example, Sampson (2001) reported data taken from 464 R&D alliances in the telecommunications industry suggesting that alliance innovation, as measured by patent counts, was most facilitated by moderate levels of alliance knowledge diversity, as measured by overlap in the partners’ previous patent domains. While patent data would not be particularly useful in the context of many university research alliances, Sampson’s study does call attention to the need for standardized definitions and measurements of human capital diversity. It could certainly be the case that the human capital differences existing between Astro and Eco were quite minor when compared to the entire range of possible knowledge combinations that could have been included within the Alliance’s boundaries. It could also be the case that different combinations of disciplinary paradigm development and prior collaboration could result in fundamentally different collaborative activity within alliances. We examined the two Alliance teams that appeared ex ante to have the most contrasting human capital configurations, but studying other combinations of human capital within alliances would be informative. It would also be informative to measure not only the volume and conceptual variety of alliance outputs, but also the scientific novelty and impact of such outputs as well. Our measure of publication title content is only one of a number of different indicators of scientific knowledge production, and an imperfect one at that. Given that the Astro and Eco teams were formed only 5 years ago, it is perhaps premature to inquire whether their respective publications have had differential impacts on their respective disciplines due to the novelty and theoretical importance of data contained within them. Indeed, measuring the novelty and impact of collaborative work spanning multiple disciplines in varied stages of paradigmatic development is likely to prove difficult in any case. Nevertheless, future research is needed to address these complexities in greater detail. Despite these methodological ambiguities, the outputs from Astro and Eco vary in systematic ways that parallel their demographic differences, and the distinctive successes of each team probably should not be dismissed as spurious artifacts. It is thus important to draw out some of the implications of our results for the organization and management of scientific alliances in general. When combined with previous research, our results suggest that three characteristics of scientific alliances might be particularly important in ensuring that an alliance’s knowledge production objectives are met. First, both Astro and Eco scientists have decomposed their Alliance projects into semi-independent modules that can be completed in distributed fashion by team members working in disparate locations. It has long been known that decomposition is one strategy for managing complex problems (e.g., Garud et al., 2002), and decomposition is encouraged in the Alliance given that scientists are dispersed among several different universities. Both teams apparently have recognized that complex collaborative work in distributed environments does not always bring with it the need for close interactions among research partners, and that, indeed, close interactions might actually inhibit the process of joint knowledge production by increasing the costs of coordination. Although team-wide coordinative practices (e.g., team meetings, etc.) seem to be more fully developed in Astro, in both teams such practices are only intermittent. Moreover, neither group of scientists exhibits high levels of collaboration on joint publications, since co-authored publications among team members have represented only a small fraction of their collective outputs. Paradoxically, then, the success of Astro and Eco suggests that collaborative work in distributed alliances might be more successful when loosely, rather than tightly, coupled work relationships exist among the parties involved. This loose coupling of research activity might be especially important for alliance teams having more diverse combinations of human capital. Second, the success of Astro and Eco supports previous research suggesting that joint knowledge production is facilitated when alliances are organized and governed as separate joint equity entities (e.g., Oxley, 1997 and Pisano, 1989). Collaboration on Alliance projects is not a result of direct bilateral or multilateral agreements among the scientists themselves. Astro and Eco are embedded in a centralized governance structure in which various administrative units and advisory committees exert top-down strategic and funding oversight. This independent administrative structure is very visible, includes representatives from partner universities, and is active in organizing websites, showcase events, planning meetings, and team presentations on a regular basis. The existence of a strong central administration has meant that teams are not encumbered with many of the overhead costs of funded research and are free to pursue their research activities within the boundaries of their Alliance goals. In some sense, then, Astro and Eco are situated in a tight Alliance system of centralized governance that has allowed each team to stay loosely organized internally. This tight-loose design could be an especially effective governance strategy in scientific alliances distributed across disparate universities. Finally, a particularly salient characteristic of the Alliance is its stock of centralized high performance computing assets. On more than one occasion, Alliance personnel have described these assets as the “glue” that both justifies the Alliance’s existence and helps to maintain the Alliance’s coherence in its mission. But our results suggest that these assets are more than simply information processing devices. They constitute a representational system that acts as a “boundary object” (e.g., Fujimura, 1992 and Star and Griesemer, 1989) linking disparate disciplines together by forcing team members to translate their ideas into common computer codes and standardized inputs for existing software applications. While such translations have been a matter of course for Astro scientists given their disciplinary focus and prior computational models, for Eco scientists the need to orient around a set of common computing resources has been a new work requirement, and has meant that team members cannot diverge too far from each other in their loosely coupled research efforts. The existence of a common representational task environment would seem to facilitate knowledge transfer and joint knowledge production within teams of disparate and distributed scientists. The possible joint effect of these three contextual influences on how human capital configurations within Eco and Astro played out over time reinforces recent arguments in the management literature that team demography is only a partial explanation for team performance (e.g., Ely and Thomas, 2001 and Martins et al., in press). Research suggests that it is important to understand how both team and organizational practices shape the effects of member demographics on team performance, and that few one-to-one relationships exist between particular team characteristics and subsequent team outcomes. Our data indicate that the Alliance has been a successful endeavor for both Eco and Astro, and that both groups of scientists have benefited in different ways from their participation in it. Joint equity governance, strong top down administrative expectations and support, modularized division of labor, and a pool of common computational resources seem to have allowed each team to channel their configuration of human capital toward productive, albeit different, ends. Our data thus support the growing practice of using interdisciplinary alliances to advance science and innovation, with the qualification that different results might have been obtained in other alliance contexts characterized by other administrative and task parameters. Future research must be conducted to explore more fully how alliance contexts interact with configurations of human capital to shape alliance performance.