A JSM 2015 Panel Discussion

Dongseok Choi, John E. Kolassa, Mani Lakshminarayanan, Barry D. Nussbaum, A. James O’Malley, Wei Shen, and Kelly H. Zou

During the 2015 Joint Statistical Meetings (JSM) in Seattle, Washington, the American Statistical Association (ASA), through its Committee on Statistical Partnerships among Academe, Industry and Government (SPAIG) and several co-sponsors, highlighted cutting-edge collaborative projects through successful partnerships between academia, industry, and/or government organizations.

Six distinguished statisticians, herein the “panel,” from various professional sectors offered valuable thoughts and advice on how statisticians may form vital collaborations across sectors.

This session had a wide appeal, given the increasing focus on inter-disciplinary research and the emergence of complex and high dimensional big data as such challenges are common across all sectors. The discussion was extremely valuable to statisticians across diverse areas of statistical practice, especially to those in smaller institutions or organizations for whom collaborations may not have existed when they joined.

According to a Cisco blog piece written by Dominguez (2011), the Chairman of Google, Eric Schmidt, once said, “When you say collaboration, the average 45-year-old thinks they know what you’re talking about: teams sitting down, having a nice conversation with nice objectives and a nice attitude.”

Schrage (1990) wrote in his book, No More Teams!: Mastering the Dynamics of Creative Collaboration!, defined the following: “Collaboration describes a process of value creation that our traditional structures of communication and teamwork can’t achieve.”

The Oxford English Dictionary (2015) says that collaboration is “The action of working with someone to produce or create something,” while Merriam-Webster (2015) defines it as “to work with another person or group in order to achieve or do something.”

In the statistical field and related professions, sharing of ideas from different organizations frequently leads to exchange visits, support for graduate students, consulting jobs, grant award for faculty, and continuing education opportunities for statisticians outside of academe. Through these activities statistical problems from outside academe become case studies in the classroom and dissertation topics. The intellectual exchange is a key component of partnerships. Below are a few thoughts on the collaborative processes.

Definitions and Views

There are several fruitful examples based on collaboration by different career sectors. For example, an academic statistician may collaborate with colleagues within the university, through formal statistical support structures created by the institution, or through more ad hoc arrangements. Collaborators may be researchers from other academic, industrial, or governmental institutions that reach out for collaboration.

Government may not seek collaboration, per se. However they do look for innovative, creative ways to do things. Many of these could involve internships or cooperative agreements with universities, for example, through requests for applications in terms of grant-funding.

Formation and Process

The demands for statistical collaborations have been steadily increasing during the last decade or two in particular with “omics” (e.g., genomics, proteomics or metabolomics) data, and there will be likely even stronger demands with big data. However, finding a right collaborator or forming a partnership may not be easy even with today’s internet or simple notification service. In our experience, a collaborator has been traditionally sought by personal connections or at professional conferences. Interestingly, a recent trend of open competitions for data analysis, e.g., the Netflix prize or Kaggle competitions, may become a new way of finding consultant or collaboration in the future.

One might use the terms collaboration and consultation to refer to work with other researchers whose area of expertise is not statistics. Consultation connotes a short-term arrangement in which the primary intellectual contribution of the statistician is the selection and application of the proper known statistical designs and techniques within the boundaries of the problem specified by the soliciting party; collaboration connotes a longer-term relationship in which statistical ideas may be involved from the problem identification to the implementation phase of a problem with new statistical methodology possibly being developed to solve novel statistical problems in the field to which it is applied.

Sometimes, a consultation may evolve into a long-term collaboration.

How to Measure Success

Necessity is the mother of invention. Here, a problem that really needs a solution is the main element leading to a collaborative success. The criteria for successful collaboration can be different for each party. A statistician, who is a faculty member at a university, may need to demonstrate academic productivity and scientific innovation through significant contributions to human knowledge and theoretical development. This contrition is generally made through publications in peer-reviewed journals and conference proceedings. Industry collaborators may be driven more by incentives such as new inventions and innovation through appropriate application of theory developed by academia, as well as business values added, and thus they may hold their collaborating partners to tight deadlines or expenditures in order to meet budgets.

For example, Rutgers, The State University of New Jersey and Pfizer have received a recent SPAIG Award. Pfizer generously supported statisticians from Rutgers through sponsorship of joint workshops and seminars, through collaboration with Rutgers faculty, with funds for Rutgers students, and with internship opportunities for Rutgers doctoral and Master’s level students.

Another SPAIG Award has recognized the strong and enduring relationship between Baylor University and Eli Lilly, which has brought methodological enhancements to both clinical and non-clinical pharmaceutical applications. Successful collaboration has paved the way for partnerships between Eli Lilly and other universities. The number of graduate students and faculty supported by Eli Lilly, as well as the hiring of Baylor graduate students, indicated that the partnership is highly valued by both entities.

One obstacle to forming collaborations across domains is that academe, industry, and government entities may have very different values in terms of measuring success. Academics may be less inclined to partake in important collaborations if they do not involve cutting-edge techniques, and they may shy away from collaborations where confidentiality may prohibit publication. Government or regulators may be resistant to adopting new statistical methods until they have been thoroughly vetted firstly in the literature and secondly through their own filtration process.

Yet despite these challenges, it is possible to form collaborations across these sectors if one is willing to understand the position(s) of the other parties and compromise in realistic ways. In the academic world, the tightening of research funders has led to a widening of criteria for promotion in many institutions and increased recognition of the work products that can arise from collaborations with industry and government. Even unsuccessful collaboration could have imbedded the seeds for an alternative approach, potentially resulting in successful collaboration in the future.

Collaborations

The ASA recognizes that the elements of statistical analysis can best be applied in conjunction with subject matter experts. Typically these may be found in other organizations….hence the recognition of outstanding collaborations.

In the era of the big tent of statistics, there will be more innovative ideas and cutting-edge products as a result of successful collaborations. The mission of SPAIG is to identify, lead, and promote initiatives that foster partnerships between academe and business, industry, & government (B/I/G). The SPAIG aims to help ASA member through the following activities:

• Establish a partnership award recognizing successful partnership efforts between academe and B/I/G
• Promote periodic JSM sessions that focus on partnership case studies and other relevant partnering
• Recommend and support periodic salary surveys of B/I/G statisticians
• Maintain a SPAIG web site to communicate partnership activities, progress on SPAIG initiatives, the partnership award program, and other SPAIG information

The annual SPAIG award recognized successful collaborations. It is currently on hiatus, pending a revision of the requirements, which could provide incentives to new collaborations and recognize individuals, not just their organization.

Reflection and Remarks

Statistical societies, such as the ASA, provide the most important encouragements to collaboration in the form of the JSM conferences, as well as other statistical activities, which allow for the discussion of existing collaborations and the dissemination of the products of such collaborations. Statistical societies encourage collaboration by sponsoring journals in which the fruits of our labors may be displayed.

Notably, the ASA maintains a registry of consultants that can be manually searched by area of specialty. This directory provides a wonderful service to individuals and organizations seeking statistical help and has result in a substantial number of consulting and collaborative relationships between statisticians and collaborators in a multitude of fields.

In the future, it is possible to imagine the registry linking to the ASA’s accreditation process, i.e., PStat® or GStat®, such that a statistician gains credit for validated work they perform in collaborations that manifested from the information provided on the site.

We thank our predecessors in the field who made these opportunities possible. Statistical societies might further this fostering of collaboration by offering registries of people seeking statistical collaborators to complement the registry of statistician contacts and their areas of expertise. The ASA and other statistical societies may consider maintaining a list of past, ongoing, and current collaborations and the work products. This effort may provide further illustration of the benefits on working with a statistician to subject-matter experts by encouraging some of them to seek statistical help when they might not of otherwise.

For more about the panelists and further reading about this topic, please see the Committee on Statistical Partnerships among Academe, Industry and Government website.

EDITOR’S NOTE: The views expressed in this article may not necessarily reflect those of the authors’ employers.