Supported by the National Science Foundation and the U.S. Department of Energy, the study shows that the number of collaborators is the strongest predictor of a scientist’s productivity -- as measured by books and scholarly papers published.
“For many years, people have been trying to encourage collaboration, but we haven’t had much research that actually demonstrates a beneficial effect on productivity,” said Barry Bozeman, Regents Professor of Public Policy at Georgia Tech and lead author of the study. “Since developing and maintaining collaborations requires time, there is always a question about whether the benefits of collaboration outweigh the costs. The work we’ve done suggests that the benefits of collaboration are great, and that collaboration is one of the best predictors of publishing productivity."
Bozeman and doctoral student Sooho Lee based their conclusions on surveys returned by 437 academic scientists and engineers working at major research centers in the United States. They also used curriculum vitae (CV) provided by the same set of scientists and engineers to help obtain measures of collaboration and productivity.
The study relates the number of books and refereed journal articles published by each of the respondents over a five-year period to the number of collaborators, considering not only the total number of books and papers, but also a “fractional count” in which each publication was assigned a score based on the number of authors. Bozeman and Lee also looked at other factors related to publishing productivity, including scientists’ rank, age, gender, collaboration strategies and job satisfaction.
“The impacts of collaboration on publishing productivity are substantial,” said Bozeman. “Collaboration plays a larger role in scientific publishing productivity than many other significant factors.”
The study, “The Impact of Research Collaboration on Scientific Productivity,” will be presented February 16 at the annual meeting of the American Association for the Advancement of Science (AAAS) in Denver, CO. It is part of a session, “Life of Science: Scientific Collaboration in Transition.”
Among other findings:
-- The study reports gender differences in early career stages. “There has been concern that women publish at lower rates than men,” Bozeman said. “The dynamics are very complicated, but we found evidence that while those concerns are valid at the beginning of women’s careers, women later publish at about the same rate as men.”
-- Publishing productivity differs by specialty, with chemists having “an especially high productivity rate” and computer scientists publishing less than any other group of scientists or engineers. Electrical engineers are the most active collaborators, while biology and life science researchers and civil engineers rank below the mean in collaboration.
-- Bozeman and Lee developed a “collaboration cosmopolitanism” scale to rank collaborations outside scientists’ own immediate work group, including scientists in other universities or other nations’ universities. They awarded physicists top ranking in this category for a higher rate of international collaboration. “The people who collaborate with others who are not near them usually are tenured,” Bozeman noted. “The most important predictor for this type of collaboration is having large grants, which may be an indicator for people who have advanced careers and good contacts.”
-- The respondents reported an average of 13.7 collaborators per year. On average, 82 percent of collaborations came from within the scientists’ own work groups. Graduate students made up 41.7 percent of collaborators; women scientists accounted for 26.8 percent of collaborators.
-- The study found correlation between career stage and productivity. On a measure of total publications, productivity peaked between the 23rd and 28th year of the respondents’ careers. Considering the “fractional count” of publications smoothed out the peaks, with the most productive years between the 19th and 29th years after receiving a doctoral degree. In both cases, productivity was lowest in the first eight years.
-- The study shows a close association between citizenship and productivity, with scientists who are not native born publishing at higher rates than U.S. citizens.
The study also examined the reasons that led scientists to form collaborative relationships.
“Sometimes people collaborate because others are close by,” Bozeman said. “Sometimes they collaborate because they have complementary skills, because they are trying to gain access to specialized equipment, share grant money – or attach themselves to a colleague’s scientific reputation.”
In a related study referenced in the AAAS paper, Bozeman and Elizabeth Corley – a recent Georgia Tech School of Public Policy PhD graduate, now a Professor at Columbia University – examined another type of collaboration: mentoring. In these relationships, senior scientists collaborate with junior faculty members and students, not to advance their own work, but to help those building new careers.
“Scientists who were mentors tended to be tenured,” Bozeman noted. “People are able to be a little less self-interested if they are tenured. We also found that the people who were most likely to collaborate with junior faculty and students were also more likely to collaborate with women scientists.”
That study has been accepted for publication in the journal Research Policy.
Contact: Barry Bozeman (404-894-0093); E-mail: (barry.bozeman@pubpolicy.gatech.edu).