استراتژی همکاری دانشمندان ': مفاهیمی برای سرمایه انسانی علمی و فنی

“Scientific and technical human capital” (S&T human capital) has been defined as the sum of researchers’ professional network ties and their technical skills and resources [Int. J. Technol. Manage. 22 (7–8) (2001) 636]. Our study focuses on one particular means by which scientists acquire and deploy S&T human capital, research collaboration. We examine data from 451 scientists and engineers at academic research centers in the United States. The chief focus is on scientists’ collaboration choices and strategies. Since we are particularly interested in S&T human capital, we pay special attention to strategies that involve mentoring graduate students and junior faculty and to collaborating with women. We also examine collaboration “cosmopolitanism,” the extent to which scientists collaborate with those around them (one’s research group, one’s university) as opposed to those more distant in geography or institutional setting (other universities, researchers in industry, researchers in other nations). Our findings indicate that those who pursue a “mentor” collaboration strategy are likely to be tenured; to collaborate with women; and to have a favorable view about industry and research on industrial applications. Regarding the number of reported collaborators, those who have larger grants have more collaborators. With respect to the percentage of female collaborators, we found, not surprisingly, that female scientists have a somewhat higher percentage (36%) of female collaborators, than males have (24%). There are great differences, however, according to rank, with non-tenure track females having 84% of their collaborations with females. Regarding collaboration cosmopolitanism, we find that most researchers are not particularly cosmopolitan in their selection of collaborators—they tend to work with the people in their own work group. More cosmopolitan collaborators tend have large grants. A major policy implication is that there is great variance in the extent to which collaborations seem to enhance or generate S&T human capital. Not all collaborations are equal with respect to their “public goods” implications.

If we think of “scientific and technical human capital” (S&T human capital) as the sum of researchers’ professional network ties and their technical skills and resources, then the question arises “how do scientists acquire and deploy these assets?” One answer, as economists’ studies (e.g. Becker, 1962) of human capital have shown, is formal education. Researchers acquire and impart knowledge through formal education processes, often resulting in credentials that signify scientific assets. Sociologists have shown that informal network ties, such as invisible colleges, can be just as important to the acquisition and transmission of scientific knowledge. Tacit knowledge often plays an important role in S&T human capital, as demonstrated by recent studies (Nelson and Nelson, 2002 and Balconi, 2002). Our study focuses on one particular means by which scientists acquire and deploy S&T human capital, research collaboration. The relation of research collaboration to S&T human capital is a topic considered by at least a few other researchers (e.g. Laudel, 2001 and Glaser, 2001), but it is not a common theme. The literature on scientists’ research collaboration shows us that collaboration choices are governed by a wide variety of factors including inter-institutional structures (Landry and Amara, 1998), formal (Wen and Kobayashi, 2001) and informal research networks, research alliances and covenants (Pisano, 1991), and arrangements for sharing expensive or scarce scientific resources and equipment (Kelves, 1995). Melin (2000, p. 32) notes, “if we move from macro to micro, we see that intertwined with these structural circumstances there are other, more individual reasons for collaboration.” Our study focuses on the “individual reasons,” particularly strategies researchers pursue in their collaboration choices. We certainly do not discount the significance of external environmental constraints and institutions, but we maintain that many of the factors governing individual scientists’ collaboration choices remain very much within the control of the individual, especially when the researcher works in an academic institution. Much previous research on collaboration focuses on co-authorship. A co-author concept of collaboration has many advantages. Katz and Martin (1997) point out four key advantages of using co-authorship as a measure of collaboration including its verifiability, stability over time, data availability and ease of measurement. But they also note that co-authorship is no more than a partial indicator of collaboration. Our study foregoes the advantages of co-author approaches in favor of a broader conception of collaboration, one that seems to us more appropriate to the study of motives and strategy. Using questionnaire data, we employ a self-reported concept of collaboration, permitting the respondent to determine what is and is not “collaboration.” While a focus on a strategy-based, self-reported concept of collaboration presents its own problems, chiefly a lack of operational precision, it avoids some of the problems of a publication-based measure of collaboration. For instance, in an early case study to investigate collaboration, Hagstrom (1965) found evidence that some publications listed authors for purely social reasons. Stokes and Hartley (1989) showed that sometimes a researcher may be listed as a co-author, simply by virtue of providing material or performing a routine assay. At the other extreme, an individual may provide a key idea for research but, for any of a variety of reasons, not be included as a co-author. La Follette (1992) showed that the practice of making colleagues “honorary co-authors” has become quite common. Our study examines data from 451 scientists and engineers1 at academic research centers in the United States, data from the spectrum of collaborators, ranging from post-doctoral researchers to full professors and research directors. While the respondents to our mailed questionnaire are from a wide variety of universities and from different research fields, all of them work in multidisciplinary settings with a strong propensity toward collaboration. In many of these centers, an avowed objective is to provide quality training and to enhance the research capacities of the persons affiliated with the centers. In the next section of this paper (Section 2), we define the concept of scientific and technical human capital and discuss how this definition is different from past research on human capital and social capital models. We also discuss in Section 2 the concept of research collaboration and how scientific collaboration can play a critical role in developing scientific and technical human capital. In Section 3 of the paper, we describe the data collection methods that were used to complete the analyses that are presented in the paper. The next section of the paper introduces a conceptual model of how research collaboration is related to the development of scientific and technical human capital. In addition, we present and discuss four research hypotheses that operationalize (in an empirical way that can be tested with the data available) the relationship between scientific and technical human capital and research collaboration patterns. In 5 and 6 of the paper, we present the results of our statistical analyses and discuss how the findings relate to the four, previously discussed, research hypotheses. In the last section of the paper (Section 6), we draw some general conclusions for the research project and present several implications for science and technology policy.

Our interest in scientific collaboration is largely an instrumental one. We wish to understand the ways in which collaboration affects scientists’ and engineers’ S&T human capital. From previous studies we assume that collaboration often has salutary effects with respect to socialization, training, transmission of know-how and just as important, the ability to develop the network ties and contacts so critical to scientists’ and engineers’ career success. While we were not able to examine all the elements and relationships of the Life Cycle Model of Collaboration and S&T Human Capital (Fig. 1), we were able to shed some light on the determinants of collaboration and infer implications for S&T human capital. Before extending the discussion beyond our findings, we summarize the most important findings, including their implications for the hypotheses we developed and for our conceptual model. An especially important finding pertains to the delineation of collaboration strategies. The dimensional properties of collaboration variables were easily interpretable in terms of factor-based strategies. Among the “Taskmaster,” “Nationalist,” “Follower,” “Buddy,” “Tactician,” and Mentor” strategies, our chief interest is in the latter, the one we assume has particularly important implications for S&T human capital. Regression results sustained many elements of our hypothesis about the Mentor strategies. Those pursuing a Mentor strategy are likely to: 1. be tenured; 2. more likely to work with graduate students and junior faculty (as professed by the strategy); 3. more likely to collaborate with women; and 4. have a favorable view about industry and research on industrial applications. We feel these findings are rife with policy implications. It is particularly important that mentors are more likely to collaborate with women. If our assumptions, and previous researchers’ (e.g. Fox, 1991 and Cameron, 1978) are correct, about the barriers women scientists face in linking to the social networks that transmit S&T human capital, identifying individuals who employ a Mentor strategy to collaboration could be a useful means of enhancing scientific effectiveness and productivity. Similarly, the finding that a Mentor strategy is associated with a favorable orientation to industry work has important implications for cooperative research. An examination of our original data showed that those tenured professors who have actually worked in industry at some point in their careers are more likely to have a Mentor strategy for collaboration. We were somewhat surprised to find that our research Grants variable was not significantly associated with the Mentor strategy. This finding is confounded, however, by the fact that nearly 90% of the respondents have grants; it is quite possible that having a grant is more important to a Mentor strategy than the size of one’s grants (which our variable measured). Regarding the number of reported collaborators, our hypothesis was sustained only in part. Those who have larger grants have more collaborators (though the relationship is not entirely linear). Those with no current grants (11 collaborators) are well below the mean (14). Tenure and gender seem not to have strong independent effects on number of collaborators (as reported in the regression analysis), though the descriptive data show that females have somewhat fewer collaborators (12) than males (14) and that tenured faculty (14.5) are slightly above the mean (14). Considering the percentage of female collaborators, we found, not surprisingly, that female scientists have a somewhat higher percentage (36%) of female collaborators, than males have (24%). There are great differences, however, according to rank, with non-tenure track females having 84% of their collaborations with females. By contrast, tenured females collaborate with only 34% females. One especially interesting finding, not easily explained, is that tenure-track (but untenured) females have collaboration patterns almost identical to tenured males. Our findings concerning collaboration cosmopolitanism may have implications for policy-makers seeking to stimulate cross-institutional collaboration. An important finding is that most researchers are not particularly cosmopolitan in their selection of collaborators—they tend to work with the people in their own work group. Supporting our hypothesis, more cosmopolitan collaborators tend have large grants. The other variables in the model, tenure and gender, have only trace effects, not statistically significant. Returning to our conceptual model, our results suggest that the hypothesized determinants of S&T human capital endowment, gender, grants, and tenure, affect the two collaboration strategies that have most obvious implications for S&T human capital, Mentor and Tactician (which is, in a sense, the “self-interested” strategy). Taking gender as a precursor variable seems to make sense in terms of the results. Women do, indeed, have different experiences with collaboration and S&T human capital and often these experiences are unfavorable compared to men. From a policy standpoint, the effect of grants is particularly important. The results are somewhat complicated by the fact that almost 90% of the respondents have grants, but it does appear that grants have significant implications for shaping S&T human capital as exhibited in the relationships to cosmopolitan collaboration and number of collaborators. In our judgment, the results are sufficiently encouraging to justify additional inquiry along the same lines. In particular we would hope to see research delving more deeply into the relationship of grants to collaboration and S&T human capital. Are there threshold sizes that have effects? What are the different implications of traditional investigator-led grants versus centers grants and cooperative agreements? We are particularly anxious to follow up on a theme just hinted at here—the salutary effects of industry work and collaboration on S&T human capital. There is a need for a better model of the impacts of gender on collaboration and S&T human capital. What are the external (e.g. family circumstance) factors that affect collaboration and do these have implications for S&T human capital? How do male mentors differ from females? How do mentors of women different than the mentors of men? The answers to such questions seem to bear mightily on S&T human capital and ultimately, on the social health, well-being and productivity of scientists. We would particularly like to know more about the predictors of successful collaboration, especially as success pertains to S&T human capital (e.g. new network ties, increased know-how and tacit knowledge, experience in acquiring and managing resources). We expect that the nature of external network ties is quite different with respect to industry and science application networks than in traditional scientific networks and related, that different collaboration strategies may be effective. But this is not a question amenable to valid answer via mailed questionnaire. The public policy relevance of our study is broad and diffuse, rather than narrow and specific. The chief implication is that it is important to ensure that collaborations generate S&T human capital. Under the procedures of some government agencies, research proposals get “points” for including graduate students and female and minority scientists and engineers. But it is important to ensure that this inclusion is neither window-dressing nor exploitive. The inclusion of early career and underrepresented scientists in funded projects does not insure that they will have collaboration opportunities and it does not ensure that the collaboration opportunities afforded will help them significantly to enhance their S&T human capital.