Disciplinary Diversity - Theory and Praxis

Assessing the Effectiveness of Various Funding Strategies to Foster Disciplinary Diversity in Research Teams

Diverse funding strategies aimed at fostering disciplinary diversity on research teams have been established by a wide range of national and international funders (e.g., HFSPO, US-NSF, NSERC, European Commission). Through the recombination of knowledge pieces outside the context in which they emerged, such strategies aim to uncover innovative solutions to the increasingly complex challenges faced by 21st century societies.

The Human Frontier Science Program Organization (HFSPO) was one of the early adopters of such strategies, and the Human Frontier Science Program (HFSP) currently integrates multiple strategies in support of disciplinary diversity. Science-Metrix recently completed a review of HFSP for the 2009–2018 funding period. The review was overseen by HFSPO’s Independent Scientific Review Committee, which provided methodological advice and approved all project deliverables.

In performing the HFSP review, Science-Metrix approached disciplinary diversity through two lenses: multidisciplinarity and interdisciplinarity. The term multidisciplinarity is used to refer to research teams that integrate members from various disciplines, while the term interdisciplinarity is specifically used to refer to the integration of expertise and approaches from different disciplines into a research paper. In the former case, the Rao-Stirling index was used to quantify diversity in the disciplinary background of a paper’s authors, whereas in the latter case, it was used to quantify diversity in the scientific subfields of a paper’s references. These two dimensions may be independent: a multidisciplinary team may yet produce quite monodisciplinary research, while a monodisciplinary team may yet produce highly interdisciplinary research.

This dual view on disciplinary diversity was in part motivated by the composition of HFSP’s funding portfolio. For example, the Cross-Disciplinary Fellowships (CDFs) fund postdocs from outside the life sciences to conduct biology-related research abroad for the first time. Such a funding scheme, while it will generate some disciplinary diversity within research teams (i.e., a postdoctoral fellow from outside the life sciences [e.g., in physics, chemistry] working with a supervisor in the life sciences), appears quite suitable to also ensure strong knowledge integration across disciplines (i.e., a postdoctoral fellow from outside the life sciences [e.g., in physics, chemistry] working on a problem within the life sciences). HFSP also includes the Research Grants (RGs) scheme, which funds multidisciplinary teams to conduct research that could not have been undertaken by individual labs. While this funding scheme is likely to ensure disciplinary diversity within research teams (i.e., multidisciplinarity), it might not translate into a level of knowledge integration that is as strong as for the CDFs (i.e., interdisciplinarity).

This presentation will compare the multidisciplinarity and interdisciplinarity of the research papers resulting from different HFSP funding schemes to provide empirical evidence on the relative effectiveness of different strategies aimed at promoting disciplinary diversity in research. A comparison between career stage will also be presented to test the hypothesis that early-career researchers might be bolder in taking novel approaches to research, whereas established researchers may be playing a role in stimulating that daring among young colleagues.

Measuring the Division of Labor in Interdisciplinary Science

Team science is the new norm and interdisciplinary collaborations have become more common. How labor is divided on scientific teams—who does what, in what proportion, and for what credit—varies across projects. This paper investigates common patterns of division of labor on disciplinary and multidisciplinary research teams using data from the contributor sections of 11,743 articles published in PNAS in a five year period (2013-2017; 70% of all PNAS articles published in that period).

Journal article contributor sections report what type of labor each author performed—they identify, for instance, who designed experiments, performed experiments, analyzed data, or wrote the paper. Contributor sections have become increasingly common in science and medical journals as a way of designating credit and discouraging ghost and gift authorship.

Larivière et al. (2016) analyze contributor sections for PLOS articles, introducing a metric of evenness for how evenly labor is distributed between authors of papers in different disciplines. Our study extends those findings by tagging author contributions by their individual disciplinary affiliation rather than discipline of the paper, and by using the link between authors’ disciplinary identities and types of labor performed to study common patterns of interdisciplinary division of labor in research teams.

First we refine the evenness measure and examine its implications. On our analysis, a team with a more even division of labor is one that has a higher percentage of authors contributing to several tasks. In contrast, a team with a “parceled” division of labor is one where contributors have distinct tasks, as on an assembly line. The evenness of the division of labor can be reported as the average of the number of tasks of its contributing authors. We demonstrate a significant correlation between team size and average evenness, and we report on both team size and average evenness according to article subject category. Applied mathematics and computer science, for instance, have smaller team sizes and greater average evenness, while genetics and immunology have larger team sizes and a more parceled division of labor. Our analysis also compares the evenness scores for papers in different subject categories when team size is taken into account.

We then use novel automated processes to study the division of labor in multidisciplinary teams. From our initial dataset, we hone in on two subsets: 1) those with teams including both biologists and engineers, and 2) those with teams including both social scientists and non-social scientists. Controlling for team size (see above), we draw out implications from our data for the understanding of division of labor in teams that fit these interdisciplinary profiles. Existing metrics of interdisciplinary integration (e.g. Porter & Rafols 2009) report the degree to which a corpus of articles in a research area cite outside their research area and how far away they cite. Our procedures shows how linking affiliation data with contributor data can supplement these metrics to provide a picture of common patterns of division of labor in teams meeting various interdisciplinary profiles.

Do Interdisciplinary Researchers Tend to Join Teams?

Interdisciplinary researchers can be reasonably characterized as conducting work that spans the boundaries of traditional disciplines. It is an open question, though, whether interdisciplinarians themselves are prone to span boundaries interpersonally by participating in teams. While substantial bibliometric work has been completed on the pathways of interdisciplinary research products (as reviewed by Kniffin and Hanks, 2017), we will focus on the people conducting research to explore the degree to which interdisciplinarians are more or less likely to participate in teamwork.

Teamwork is interesting to examine in relation to whether someone conducts interdisciplinary research since there is evidence of countervailing trends. In our own research, for example, we have found that STEM doctoral graduates who participate in teams tend to earn significantly higher salaries (Kniffin and Hanks, 2018) while our analysis of near-term outcomes for all doctoral graduates in the United States from 2010 suggests that interdisciplinarians are penalized in the labor market for at least the year after earning the PhD.

In order to address our focal question, we looked at the 2006 National Science Foundation’s (NSF) Survey of Doctorate Recipients (SDR) – a longitudinal panel that is matched with responses to the NSF’s Survey of Earned Doctorates (SED) – since the 2006 wave uniquely asked respondents to indicate the degree to which they participated in teamwork with others. We find (with N = 2,748) that people who conducted interdisciplinary dissertations were – controlling for discipline and demographic variables as well as employment sector (industry, education, and government) – significantly more likely to be part of teams with (a) others who are part of different organizations in the United States or (b) others who are part of an institution outside of the United States. In contrast, interdisciplinary dissertators were not significantly more likely to be part of teams on-site or across units within their own organizations.

Our research provides important baselines for future studies that will examine the question of whether interdisciplinary researchers tend to have – over longer career spans – more varied outcomes than mono-disciplinary researchers. For example, bibliometric research suggests that interdisciplinary research products are more likely to be “sleepy beauties” or “late bloomers” (e.g., Ke et al., 2015) while our research will help inform whether a comparable pattern exists for the individuals who conduct early-career interdisciplinary research.

References

Kniffin, K. M., and Hanks, A. S.  2017.  Antecedents and Near-Term Consequences for Interdisciplinary Dissertators.  Scientometrics, 111: 1225-1250.

Kniffin, K. M., and Hanks, A. S.  2018.  The Tradeoffs of Teamwork among STEM Doctoral Graduates.  In Press, American Psychologist.

Ke, Q., Ferrara, E., Radicchi, F., & Flammini, A. (2015). Defining and identifying Sleeping Beauties in science. Proceedings of the National Academy of Sciences, 112(24), 7426-7431.

Spectacles of Inquiry: Perspective-Taking and Interdisciplinary Values

The talk is organized around discussion of perspective taking as an epistemic activity or practice, one especially important in interdisciplinary communities.  I first distinguish several senses of perspective that have emerged in contemporary scholarship:  disciplinary perspective, perspective as framework within a discipline, and perspective as personal or cultural identification.  I suggest that controversies surrounding the idea that knowledge is perspectival stem in part from a conflation of different senses of perspective, as well as from the assumption that a researcher is hopelessly locked within a given perspective and unable to transcend it.  As a corrective, I offer a view of perspective-taking as an intentional effort to transcend a given epistemic vantage point and to imaginatively occupy a different vantage point, by means of which the researcher can trade one set of epistemic constraints or limitations for another set.  As an intentional act of this kind, problem-solving is valuable in innovative interdisciplinary science communities because it encourages the ability to navigate strategically between viewpoints for in the interests of a given problem-solving goal and in response to the complexity and demands of a phenomenon.  Second, at the level of interaction, perspective-taking can facilitate communication between researchers from different disciplinary backgrounds and enable anticipation of what is needed for effective collaboration.

I will offer an illustration of the importance of perspective taking in interdisciplinary science settings with reference to an analysis of interview transcripts with bioengineering scientists, in the context of a multidisciplinary investigation of cognitive and learning practices in four different research laboratory contexts, on which I collaborated with the ethnographic study’s principal investigators.  One finding from the study was that researchers from divergent science backgrounds (e.g., engineering and biology) or different skill sets (e.g., experimenting and computational modeling) expressed different views about priorities in problem-solving and revealed implicit hierarchies, with some practices viewed as more consistent with a researcher’s view of “good science.”  On the other hand, our team found that researchers also displayed the ability to adopt perspectives different from that with which they started through concerted effort and targeted educational experiences.

After offering these illustrations, I emphasize that a proper understanding of perspective taking is one that establishes it as a practice that functions to enhance knowledge of complex phenomena and facilitate innovative strategies in collaborative problem-solving.  I will end by exploring what is required to cultivate perspective-taking in interdisciplinary settings.

SciTS Presentation: Spectacles of Inquiry: Perspective-Taking and Inerdisciplinary Values

A Team Science Approach to Advance the Understanding of Low Back Pain

INTRODUCTION: Among health disorders contributing to the global burden of disease, low back pain (LBP) is a leading cause of disability. Like other persistent pain conditions, LBP is increasingly recognized as a multifactorial problem involving biological, psychological and social factors. It is difficult to fully appreciate the complexity of LBP because the knowledge necessary to do so spans many areas of expertise. Indeed, perspectives differ between LBP experts and diverse treatments, supported with variable evidence, are offered. This is not surprising, considering theoretical foundations and emphases vary among disciplines, and these experts (e.g., researchers and clinicians) may have different “mental models” of what and how various factors relate to LBP. A novel way to integrate the interdisciplinary knowledge shared among various stakeholders is a team science approach using collaborative modeling. This study used this approach with a multidisciplinary team of experts to build a model to enhance understanding of the LBP problem.

METHODS: Participants who have contributed significant knowledge to the area of LBP (e.g., publications, contributions to societies, etc.), were selectively recruited for this study. Each participant underwent a structured one-on-one interview to construct a fuzzy cognitive map (FCM) (Mental Modeler software, www.mentalmodeler.org) representing his/her understanding of how factors related to LBP interact and affect patient outcomes (pain, disability and quality of life). This process involved nomination of factors contributing to patients’ outcomes and the weighting of the connections (strength of the effect) between these factors. All individual FCMs were then integrated into one meta-model (Gephi software, www.gephi.org). Factors in the meta-model were grouped into 10 categories: Nociceptive Detection & Processing, Behavioral/Lifestyle, Tissue Injury/Pathology, Social/Work/Contextual, Psychological, Comorbidities, Biomechanical, Individual Factors, Outcomes, and Treatment. To determine the importance of each factor expressed in the meta-model, centrality was computed as: Centrality = sum|weight of connections in|+sum|weight of connections out|, where the weight determines the strength of the connection. Based on this definition, centrality is proportional to the number of connections to and from the specific factor in the meta-model, as well as by the weighting of these connections.

RESULTS: From 38 invited experts, 29 (76%) agreed to participate. They represented nine disciplines and eight countries (Basic Science (n=3), Chiropractic (n=4), Orthopedic Surgery (n=2), Physical Medicine & Rehabilitation (n=2), Physical Therapy (n=10), Psychology (n=2), Exercise Science (n=1), Biomechanics (n=3), and Epidemiology (n=2)). The meta-model, that integrated the understanding of LBP of all participants, consisted of 145 factors and 2,233 connections representing interactions among these factors. “Psychology” emerged as the most central category in the meta-model, followed by “Biomechanics” and “Social/Work/Contextual” categories.

DISCUSSION: The integration of 29 FCMs, representing diverse participants’ views of LBP dynamics, resulted in a meta-model that was extremely complex, but still feasible to produce meaningful interpretations. This approach could provide the framework for a larger, community-wide platform for further development and refinement of this meta-model. Such a meta-model could then be used to simulate other “what if” scenarios, to identify gaps in knowledge, and to inform new essential research directions to ultimately improve patient care and outcomes for LBP.

SciTS Presentation: A Team Science Approach to Advance the Understanding of Low Back Pain (LBP)