COLLABORATION

Systematic research on the nature and structure of collaborative enterprises has been conducted for quite some time across disciplines. Sociologists and economists have done the most extensive work on the characteristics of cooperation, mostly involving business and financial collaborations, but also investigating the features of academic coauthorship, focusing, not surprisingly, on their own disciplinary norms (Ruef Reference Ruefforthcoming). In an effort that mimics that of the APSA Working Group on Collaboration, Hunter and Leahey (Reference Hunter and Leahey2008) undertook an analysis of the major journals in sociology between 1935 and 2005. They find that people doing quantitative work are much more likely to collaborate than those involved in non-quantitative work and suggest that this is because such work is more likely to require data-collection efforts and numerous analytical tools.

Scholarship exploring the nature of collaborative networks in biology, physics, and mathematics demonstrates that while the rate of publication has increased over the last 50 years in each of these fields, the rate of collaboration has spiked more considerably (Newman Reference Newman2004). Not surprisingly, given the multidimensional nature of the discipline, biologists collaborate much more frequently than mathematicians; about 66% of mathematics papers have one author, while only 21% of biological ones are written by individuals working alone. Moreover, the average number of collaborators a biologist has is more than four times greater than the average number a mathematician includes.

With regard to prevalence of collaboration, political science does not exist at either extreme. At one end of the spectrum, The American Historical Review did not publish one coauthored piece in all of 2007 (Jaschik Reference Jaschik2007), while the vast majority of articles in most science and interdisciplinary journals were coauthored. Indeed, as noted above, the leading interdisciplinary journals, such as Nature, Science, and Proceedings of the National Academy of Sciences, with impact factors some 10 times or more higher than their social science counterparts, almost never publish single-author pieces. This may be partly due to the higher article-citation rates associated with multiple-author projects (Lawani Reference Lawani1986), attributed to a combination of factors, including access to larger social networks, increased visibility, greater scope, and enhanced quality of research (Baldi Reference Baldi1998; Goldfinch, Dale, and DeRouen Reference Goldfinch, Dale and DeRouen2003; Katz and Hicks Reference Katz and Hicks1997; Martin Reference Martin1996; Phelan Reference Phelan1999).

For the 2006 “Report of APSA Working Group on Collaboration,” Chandra et al. surveyed 10 major journals across subfields from 1956–2005. They find that in the earliest period, from 1956–1965, less than 10% of articles were coauthored. By 1996–2005, about 40% of articles in the discipline involved some degree of collaborative work. The vast majority of these pieces combined two or three authors, mostly all within the same discipline, if not the same subfield. Collaboration increased most significantly in the subfield of American politics, and least in political theory. The working group also notes an increase over time in collaborations involving more than two people. In examining papers presented at the APSA annual meeting, Chandra et al. find that collaborative papers most often involved participants of asymmetric rank (e.g., a faculty member working with a student). Interestingly, collaboration patterns also appear to differ by sex; less than 10% of collaborations involved women and the majority were composed of all male groups, with mixed gender groups falling in between.

We conducted a brief examination of articles in the American Political Science Review (APSR), to compare rates and styles of collaboration, using the years 1970, 1980, 1990, 2000, and 2008 as benchmarks. We picked this journal because historically it represented the outlet most likely to increase visibility and improve professional standing for its authors within the context of research-oriented departments. Striking patterns emerge by both number and sex. Every research article for each of these years was tabulated (see table 1).

Table 1.

Authorship in the American Political Science Review

Two patterns emerge. Collaboration increased, and the biggest change is that men working in groups of two increased over time. Women, however, have only increased their presence through enhanced mixed-sex collaborative efforts, not through solitary work or same-sex groups. Indeed, the number of women in the discipline has increased over time, but their single-authored publications in the APSR have not substantially increased over the course of the last 40 years (Young Reference Young1995).

More central to our theme is the emergence of large research teams. Interdisciplinary work has only recently begun to emerge in political science, and it is too early to make any historically based empirical assessment, or properly gauge its long-term potential influence on the discipline. However, this latest development may be akin to what was witnessed in the life sciences decades ago and only time will tell if the trickle becomes a trend. Yet, long-term developments regarding large-scale collaboration in the life sciences may be able to offer both instructive and cautionary guidance for our field. Their experience with developing established consensual norms around large interdisciplinary collaborations may provide a useful guide for political scientists in establishing predictable partnerships and clear roles and rules for developing graduate training, evaluating hiring, promotion, journal policy, and determining other professional accolades.

Increasing interdisciplinary work may require more complex forms of collaboration across fields as research projects become larger, more sophisticated, and more demanding in nature. Indeed, certain political science departmentsFootnote ⁴ have started to develop integrated laboratory models more traditionally associated with the life science research laboratories in order to obviate this obstacle, while life science institutes that have historically only included scholars more traditionally associated with the so-called hard sciences, such as geneticists or biologists, have begun to hire social scientists; indeed there has been a substantial growth of interdisciplinary institutes to tackle complex social and biological problems in the last several decades in order to address this need.Footnote ⁵

WHY ENGAGE IN INTERDISCIPLINARY COLLABORATION?

In his theory regarding the structure of science, sociologist Robert Merton (Reference Merton1973) proclaims that the goal of science should be to expand genuine knowledge, emphasizing originality as the factor that advances scientific knowledge most clearly. Not only does collaborative work contribute to originality, but successful teamwork also enhances the development of long-term vision, skills, breadth, and depth among and between researchers of different expertise (Amabile et al. Reference Amabile, Patterson, Mueller, Wojcik, Odomirok, Marsh and Kramer2001; Hagedoorn, Link, and Vonortas Reference Hagedoorn, Link and Vonortas2000; Hamel, Doz, and Prahalad Reference Hamel, Doz and Prahalad1989; Katz and Martin Reference Katz and Martin1997). Those who collaborate have more opportunities to grow intellectually as opposed to those who do not, because collaborators benefit from the inevitable cross-fertilization of research programs, direct and indirect training from affiliates, and the transfer of skills and abilities from others who are experts in areas outside their own, whether through specific training, social networks, graduate student exchange, or direct exposure to new ideas (see Crane Reference Crane1972; Price and Beaver Reference Price and Beaver1966).

There is strong support for the notion that collaboration is critically important for creating better work products. Academics is an enterprise that is built on the accumulation of knowledge and, under this guise, sole-author work can only accomplish as much as a research team to the extent that individuals are able to accumulate as much knowledge over time as those who work in groups. To be clear, sole authors can produce the highest-quality work, but it may be harder for them to do as much as quickly as teams. Combining depth and breadth of skills across a number of scholars can facilitate a better, faster work product than any single author attempting to learn every aspect of each idea examined, or every method required at every stage of development and analysis for complex topics. This view is not new and such intellectual interdependence remains commonplace in certain areas within political science, although seldom is it explicitly acknowledged or understood as such. For example, whether using the ANES, congressional records, voting returns for data analysis, large number quantitative data sets such as COW or MID, or translated ancient archival texts, most authors using such data sets did not develop the questionnaires or collect the original data themselves, much less interview or assess the respondents, code the data, train the interviewers, translate books, build the statistical packages, and so forth. Rather, people who downloaded the data, picked up a translation, or used data made available by a large research grant written by someone else already benefited from their colleagues' prior efforts. However, current disciplinary norms do not cast previously collected or manipulated data, or time spent obtaining funding, as constituting collaboration, or justifying coauthorship.

This does not mean that the value of individual effort loses merit. Rather, it should go without saying that the probability of creating and producing great work depends not only on one's intellectual forerunners, but also in part on one's current peers. It is noteworthy that a small number of authors have a very large number of collaborators, which means that a great deal of work passes through the hands of a small number of the best-connected scientists, or those with high so-called betweenness scores (Newman Reference Newman2004). While this phenomenon means that a few powerbrokers often inadvertently control the flow of information across a large number of individuals within any given field, such highly influential actors do not appear to preferentially work with each other. These individuals wield a disproportionate influence on their areas, due to some combination of their greater productivity and the critical nature of their particular social and intellectual abilities in bringing diverse experts together in pursuit of a common research goal. Often, such individuals function as crucial “translators,” helping to make the implicit norms, assumptions, and expectations of each field transparent to team members from other disciplines. This ability to build bridges often requires deep and sustained exposure to more than one discipline at some point in the person's career, which is why cross-disciplinary postdoctoral fellowships often provide such unique and critical training opportunities. These networking skills do not seem to be easily transferred or replaced across random individuals. For instance, Newman (Reference Newman2004, 5204) finds that “collaboration networks are highly susceptible to the removal of the individuals with the highest betweenness score. One need only remove a few such individuals from the network, it turns out, to break the connection between the others and fracture the network into disconnected parts.” Thus, collaboration in no way negates the critical role of the individual scholar.

THEORIES AND MODELS OF COLLABORATION

Various theories and models regarding collaboration exist. In work specifically examining the nature of collaboration, Leahey and Reikowsky (Reference Leahey and Reikowsky2008) find three distinct styles. The most common they call the reinforcing specialist model, whereby individuals work with others who come from within their own area of specialization (Barnett, Ault, and Kaserman Reference Barnett, Ault and Kaserman1988). The rationale for this arrangement rests on the desire for equally skilled researchers to reduce their research efforts without reducing individually measured research output. The second involves complementary skills, whereby people with non-overlapping skills join together for mutual benefit. The third is spurred by those who migrate into new intellectual territory and begin working with those who have skills in other areas. One can imagine this last style being particularly beneficial when scholars wish to work in interdisciplinary groups focused around topics of common substantive interest, but cannot invest all the time necessary to learn a particularly challenging new methodology or technology in order to work on a particular problem. A good example might be scholars who wish to do studies involving brain-imaging technology to answer their substantive question, but do not have the time or desire to pick up another graduate degree or certification in neuroscience. Learning the technology or performing the analyses is a major multi-year investment. Rather, such scholars benefit by collaborating with those already trained in the technology and methodology.

Broad theories are useful for categorization, but the functional specifics of how people actually work together are often overlooked. The number of approaches to the practical aspects of collaboration is immeasurable; there are probably as many styles as there are groups of collaborators. In a series of open-forum discussions with political scientists we conducted over the last year, two work arrangements emerge as the most common. The first approach rests on a more contractual division of labor and specialization design, whereby individuals know what they bring to the collaborative enterprise. When a project needs one specific skill, it often seems useful and effective to add a collaborator for this purpose. The growth of new methodologies, computations skills, and research areas has made efficiency gains from a more pronounced division of labor (McDowell and Melvin Reference McDowell and Melvin1983). Certainly, the increasingly complex biological, economic, political, and social world has made it almost a technical necessity to specialize (Barnett, Ault, and Kaserman Reference Barnett, Ault and Kaserman1988). In this style of collaboration, the initial interaction revolves almost exclusively around task-related efforts, but the collaborations also often involve a team of individuals coming together to work on a distinct project of mutual interest.

A second work style relies on a more partnership-oriented model, whereby each person is expected to give 100% to the team effort. This style appears more common in groups within the context of a mutually enduring, long-term, shared research agenda. Partnerships with mutual expectations of a long shadow of the future are more likely to overcome problems related to division of labor or order of authorship because many opportunities for balance exist and increasingly emerge over time and across diverse domains of action, from publishing to giving presentations, and so on. In addition, such partners often derive value from sharing a variety of tasks and tackling difficult problems together over time. This finding is consistent with the literature. In approaching multiparty alliances as a social dilemma, Zeng and Chen (Reference Zeng and Chen2003) argue that collaboration can fail if individuals perceive that social defection generates higher individual payoffs. They note that by recognizing the important mutual benefits of cooperation, such challenges can be overcome. Over many repeated successful interchanges, trust can develop within the context of such collaborations if team members are lucky. And, as Adler (Reference Adler2001, 215) importantly notes, “economic and organizational theory have shown that, relative to trust, price and authority are relatively ineffective means of dealing with knowledge-based assets,” adding that “the effect of growing knowledge intensity may indeed be a trend toward a greater reliance on trust.”

THE NATURE OF COLLABORATION AND CHOOSING COAUTHORS

The legitimate reasons an individual might want to collaborate can differ, but we believe that collaboration should not be sought primarily as a vehicle for social or professional networking. Similarly, it may not be a good idea to collaborate with someone just because you are friends. While it is common for friendships to develop over the course of collaborations, it is equally possible for friends to become frustrated at one another as a result of disagreements over work. Rather, we take the view that collaboration is best pursued for the primary goal of achieving a superior work product, whether for the long or short term, than might be accomplished by working alone. In addition, we argue that collaboration should not be pursued as a mechanism to get more credit for less work. While it is true that collaboration may reduce the absolute amount of time it takes to complete a particular project, it does not reduce one individual's time on that project. Nor does it necessarily increase one individual's overall production. Most often, one will spend the same amount of time per article (McDowell and Smith Reference McDowell and Smith1992), but be able to produce an article of greater scope and quality through collaborative efforts.

While assessments of quality in academics often remain notoriously difficult and subjective in nature, the additional ephemeral advantage offered by the collaborative enterprise revolves around establishing and maintaining collegial relationships. These relationships often make the work better, potentially more engaging and more enjoyable, as long as the collaborations themselves are properly managed and negotiated. At the most basic level, it can help to talk a challenging intellectual problem through with someone who is similarly knowledgeable and invested. In collaborative groups with complementary skills, each person can contribute their areas of expertise toward a common product of mutual interest. The problem, of course, is that not all collaborations are born equal, and finding the best fit is not always easy or simple. Here a twist on the analogy presented in the tale of the three bears may be most apt: some require too much, others offer too little, and very few feel just right. But the effort invested in the “just right” productive collaboration can prove well worth the effort.

THE POTENTIAL EFFECTS OF INCREASED MULTIDISCIPLINARY COLLABORATION FOR THE DISCIPLINE

If increased collaboration continues at current rates, particularly in the area of large interdisciplinary teams as has happened in the life sciences, and there is some early evidence of this beginning to occur in political science, such changes will eventually exert a significant impact on departmental decisions related to hiring, tenure and promotion, and the journal review process. Questions each department and the discipline may need to consider revolve around what matters most in these evaluations. Is the overall quality of scholarship or journal reputation more important than the relative ease of assigning credit for work? How does the field consider the high-impact interdisciplinary journals? These are not easy questions. Quality can be very difficult to assess, especially in new or highly specialized areas of research where few people are familiar enough to evaluate others. Certainly no one would argue that lower-quality work at the cost of ease of credit is best, but the practical considerations associated with attributing credit properly remain necessary for promotion, salary increases, and so on. Determining author contributions can prove difficult to the extent that norms of author order are not well instantiated in the discipline, and interdisciplinary journals have historically not been part of the normal disciplinary discourse.

Political science currently resides only slightly on the side of the spectrum favoring solo work, as it also values collaborative work. In many ways, the discipline already considers the quality of coauthored work as seriously as single-authored work in making decisions about professional advancement. Yet such a move obviously encompasses more than just a cultural shift, but also normative change, and such future moves are not without professional and administrative consequences beyond the level of individual departments. It seems likely that a potential shift in reorientation of credit from single- and two-authored works to multiple-authored works may pose a central challenge to the established norms of the discipline.

In addition, if political science follows the patterns we have enumerated in other fields, more coauthored work might also lead to greater overall journal submissions. Such an increase would put greater pressure on an already overtaxed review process. This raises broad and systematic challenges for journal editors and the discipline at large. Editors are already having difficulty securing adequate numbers of qualified referees in a reasonable time period, and the challenge of obtaining a sufficient number of qualified and responsible reviewers constitutes a serious concern for many in the discipline. Without making procedural or staffing changes, increased collaboration and productivity will most likely increase reviewer demands and generate longer wait times for initial responses at prominent journals.

Other fields have attempted to address these concerns in various ways. Interdisciplinary science journals such as Science and Nature publish weekly, and these articles are dramatically shorter than the average article in political science. Some journals have large professional editorials staffs. Reviewers are often given a maximum of 10 days to review an article. Rejection without review is commonplace. Indeed, a large percentage of articles submitted to high-impact journals are rejected prior to review, making reviewer tasks much less burdensome. Editorial staffs give an almost immediate decision, and are not required to give a reason. Thus, the common interpretation around rejection regards fit and not quality. Such rejection policies save overburdened reviewers from having to undertake frivolous reviews, making this process crucial for the high-submission, higher-impact interdisciplinary journals. However, in many ways, such a process runs counter to historical norms in political science that strive to review every manuscript and value the feedback and teaching that comes from such readings. One way in which both traditional norms and more efficient processes might meet is a voluntary editorial reject-without-review process, whereby authors indicate if they are willing to have their manuscripts sent for review or summarily dismissed if the editor is not interested in the article. Such a process has clear advantages for both sides. Authors know their fate relatively quickly, thus allowing them to submit their work to another journal much sooner, yet they can choose to opt out in order to receive feedback on work that may ultimately be rejected for publication. Another tool common in the life sciences is the pre-submission inquiry. In this process, authors send a short abstract to the journal editor to inquire whether the project itself is of interest to the journal before submitting the entire manuscript. The editorial staff replies with a simple encouragement or denial of interest in submission.

However, these benefits do not come without a price. Editorial staffs have to take on additional duties, and often publication costs are pushed onto the authors. In Proceedings of the National Academy of Sciences for example, authors have to pay a significant amount of money for articles running longer than six pages ($750 per page), although Web appendices of virtually unlimited lengths are allowed for free. In addition, many economics and sociology journals charge a submission fee (around $25) to offset administrative costs.

Not all processes developed to address increased numbers of journal submissions require intervention at the journal level. Working models differ greatly by research group and field, but it is not uncommon that approval from a research institute or lab director or some other departmental appointee, such as an advisor, be required prior to journal submission. Such a norm reduces the number of frivolous journal submissions and increases manuscript quality. Senior personnel widely support such policies because they remain invested in increasing the quality of submissions from their departments or research teams, and understand that any manuscript submitted from their labs or departments affects the professional reputation and image of their programs. This image has a direct influence on the placement of graduate students, future recruitment, and even prospects for funding future work. Indeed, such status constitutes a significant form of currency in their own fields and disciplines.

As times change and scholarship becomes more integrated and complex, the discipline has already begun to establish a consensus around more complex criteria for professional evaluation. Indeed, it is already true that sheer number of publications, or citations alone, does not constitute a sufficient assessment of the quality of an individual's contribution to collective work. A thorough review of any given scholar's overall record tells a more coherent story regarding the depth and breadth of the total contribution to the field. The administrative process associated with moving to an even broader assessment of quality for purposes of hiring, promotion, and merit increases will no doubt prove harder to navigate institutionally, and certain changes may take some time. However, the issues posed by increased collaboration of diverse types are already being openly discussed among editorial boards and executive committees.

Relevant to these discussions is the emergence of a recent trend, whereby some of the most profound, innovative, and groundbreaking work involving political topics, conducted by political scientists utilizing interdisciplinary tools, has increasingly been presented in journals outside of the discipline (e.g., see Oxley et al. Reference Oxley, Smith, Alford, Hibbing, Miller, Scalora, Hatemi and Hibbing2008; Fowler and Schreiber Reference Fowler and Schreiber2008; McDermott et al. Reference McDermott, Tingley, Cowden, Frazzetto and Johnson2009; Spezio et al. Reference Spezio, Rangel, Alvarez, O'Doherty, Mattes, Todorov, Kim and Adolphs2008). If this trend continues, such work will advance without the participation of its most relevant contributors. The dual forces pressuring the journal review process and the challenges associated with allocating appropriate institutional and departmental credit for interdisciplinary work provide powerful conflicting reinforcement. Whether an impetus for organizational and normative shifts occurs, there is no doubt that cutting-edge debates involving political topics are now commonplace outside our own field. Whether in neuroscience, genetics, psychology, endocrinology, physiology, anthropology, primatology, or other fields, inherently political topics including affiliation, cooperation, punishment, ideology, conflict, coalition formation, framing, status, and so on have become the focus of primary research interests across a wide variety of disciplines outside political science (e.g., see Adolphs Reference Adolphs2003; Eaves and Hatemi Reference Eaves and Hatemi2008; Inbar, Pizarro, and Bloom Reference Inbar, Pizarro and Bloom2009; Rozin, Haidt, and Fincher Reference Rozin, Haidt and Fincher2009; Sanfey et al. Reference Sanfey, Rilling, Aronson, Nystrom and Cohen2003; Shergill et al. Reference Shergill, Bays, Frith and Wolpert2003; Westen et al. Reference Westen, Blagov, Harenski, Kilts and Hamann2006; Wrangham Reference Wrangham1999). This movement can represent a boon to the field of political science because it demonstrates the import of and widespread interest in the topics we study. But if political research begins to progress in large part outside the field without the expertise of its primary membership, both the larger academic community and the political science community will suffer.