How do men and women differ in research collaborations? An analysis of the collaborative motives and strategies of academic researchers

Abstract

Do men and women academic faculty vary in their research collaboration patterns and strategies? This straightforward question does not lend itself to a straightforward answer. A great many sex-correlated variables could possibly mitigate the relationship of sex and collaboration. If one finds sex-correlated differences in the number of collaborators, can one infer that there is something intrinsic to men's and women's work strategies and preferences? Or would such differences owe instead to women's and men's different positions in work structures and hierarchies? The focus here is on two sets of research collaboration variables, numbers of collaborators and the collaboration strategies employed. The study uses questionnaire data from the U.S. National Survey of Academic Scientists (n = 1714) and tests several hypotheses about collaboration numbers and strategies. Regression results indicate, counter to the core hypotheses and almost all published literature, that in a properly specified model, one taking into account such factors as tenure, discipline, family status and doctoral cohort, women actually have somewhat more collaborators on average than do men. For both men and women, those with more industrial interactions and those affiliated with university research centers have more collaborators. Men and women differ in their collaborator choice strategies. Men are more likely to be oriented to “instrumental,” and “experience” strategies, while both men and women are motivated by “mentoring” strategies. Regression analyses show that for both men and women, having a coherent collaborator choice strategy predicts the number of collaborators.

Introduction

Do men and women academic faculty vary in their research collaboration patterns and strategies? This straightforward question has received surprisingly little attention (Fox and Faver, 1984, Kyvik and Teigen, 1996, Lee and Bozeman, 2005). It deserves more. Evidence about sex-based differences in collaboration may well impinge on a variety of crucial issues in research and education, including team-building effectiveness but also such secondary effects as educational attainment, representativeness of the scientific workforce, recruitment and retention of scientific and technical human capital, and perhaps even the quality of the research itself.

While it is easy enough to frame the question of possible sex-based differences in research collaboration, obtaining a confidence-inspiring answer to the question proves less simple. Perhaps sex-correlated variables mitigate the effect of sex on collaboration. If one finds sex-correlated differences in, say, the number of collaborators, can one infer that there is something intrinsic to men's and women's work patterns? Or has one simply failed to properly specify the comparative model (Western, 1998)? Abundant research shows that men and women differ in situational factors that could be presumed to relate to collaboration patterns. For example, compared to men, academic women have lower marriage and partnering rates (Probert, 2005) and research shows (Xie and Shauman, 1998, Toutkoushian and Bellas, 1999, Perna, 2001, Lee and Bozeman, 2005) that single academics, both men and women, tend to be less productive in terms of research publication (and perhaps collaborate less). When academics have children at home, women tend to have greater child rearing responsibilities, even when both spouses are academic faculty (Hamovitch and Morgenstern, 1977). Men and women have different degrees of geographic and job mobility (Rosenfeld and Jones, 1987). In some cases situational and structural factors interact, as when negative academic department norms or expectations interfere with family obligations (Fox, 2010) and, in turn, differentially affect women's research.

Structural factors may also factor in significantly in how scientists collaborator. One factor possibly accounting for differences in men's and women's collaboration is academic mentoring. Studies have shown that academic women, both graduate students and faculty, tend to differ from men with respect to mentoring experiences (Dreher and Ash, 1990, Green and Bauer, 1995, Tenenbaum et al., 2001). Academic women are less likely to have mentors and the nature of women's mentoring differs from men's (Sands et al., 1991, Sambunjak et al., 2006, Kiopa et al., 2009). Many academic researchers learn from their mentoring relationships how to collaborate and how to interpret social dynamics of collaborations (Bozeman and Corley, 2004, Mayer et al., 2008). The complexity of sex-based differences in collaboration mounts when one considers the many non-work issues that may tend to impinge on collaboration (Hunter and Leahey, 2010) or that women are more often the “trailing spouse” (Harvey, 1998, Bailey and Cooke, 1998, Shauman and Noonan, 2007).

Developing a properly specified model of differences between men's and women's collaboration provides a number of benefits. In the first place, such a model permits one to focus on the valid, non-spurious sex-based collaboration differences between men and women, providing a better prospect for understanding and enhancing career development and success. Perhaps just as important, however, is developing a fuller understanding of structural factors mitigating men's and women's collaboration. Even if one were to find, with a properly specified model, that women and men do not differ greatly in their collaboration patterns, the mitigating factors (the intervening variables) are themselves of interest. Thus, if one were to find, for example, that age and cohort interactions have stronger effects on collaboration than do differences in the sex of collaborators, then it is perhaps possible to develop policies addressing differences in power and resource dynamics without muddying the water by treating these as inherently related to sex or gender. Likewise, were one to find that, say, women's lesser degree of affiliation with interdisciplinary research centers explained much of the variation between men's and women's collaboration patterns, then such affiliations could be viewed as policy levers related to career success. In short, situational differences, structural, and climate-related contextual issues have great import; a properly specified model is required to sort out these causal differences.

Studies of research collaboration have examined a wide variety of factors and dynamics of collaboration (Shrum et al., 2007), but most studies of individual level (researcher-to-researcher) collaboration examine one of two general constructions of collaboration. Some studies (e.g. Heffner, 1981, Vinkler, 1993, Wagner, 2005, Heinze and Bauer, 2007, Mattsson et al., 2008) focus on discretely measured co-authorships and factors pertaining to those authorships (e.g. author order, number of authors). These studies have proven quite useful in many respects and have the advantage of being amenable to use of standard, unobtrusive data, especially as developed in Web of Science and similar data repositories. The current study follows in the second of two major traditions (e.g. Katz and Martin, 1997, Melin, 2000, Boardman and Corley, 2008) in which researchers are asked to report on their collaborations. While this has the obvious disadvantage of possible instrument bias, it has several advantages (outlined in detail in Bozeman and Corley, 2004). In particular, reports of collaboration have the advantage of recognizing contributions of persons who may not have co-author status but who made important contributions. Similarly, self-reported collaboration diminishes problems related to the “ghost authors” increasingly evident in co-authoring (Rennie et al., 1997, Mowatt et al., 2002, Wager, 2007).

In this paper, we contribute to the literature on research collaboration by studying how situational, structural and climate-related institutional factors contribute to the collaboration patterns and strategies of research university professors in the United States. We examine the determinants of volume of collaborations, as well as how collaboration strategy preferences affect the volume of such collaborations. We find that situational factors related to collaborations tend to have modest effects on the volume of collaborations, while structural effects such as tenure status and discipline have strong positive effects on collaboration volume. Institutional affiliations that tend to improve collaborative climates – such as multidisciplinary science centers – also increase collaborations. We extend understanding of these phenomena by further inquiring how collaboration strategies affect collaboration volume. In all analyses, we evaluate the extent to which men's and women's collaboration volume and strategies may vary, finding both similarities and differences.