Closing CSCAR is a short-sighted decision

On August 13, the Office of the Vice President for Research announced that the Consulting for Statistics, Computing and Analytics Research unit will be permanently closed. OVPR provided only vague justification for this move, claiming it would make them more effective in meeting a variety of needs on campus.

CSCAR is a University of Michigan service unit that provides free statistical consulting to students, staff and faculty, enabling them to pursue research questions beyond their advisers’ or collaborators’ expertise while ensuring methodological rigor. CSCAR also provides practical workshops on a variety of statistical software and methods for nominal fees. The faculty, staff and graduate students who work there are experts in statistics, data science and survey methods. Closing CSCAR is a disheartening, short-sighted decision. 

Access to this unique resource has been a valuable tool in recruiting faculty candidates to the University particularly given the current uncertainty of federal research funding. With funding becoming increasingly competitive, eliminating such a resource is the very definition of “penny wise and pound foolish.” It remains unclear how OVPR intends to meet the needs of faculty, staff and students in CSCAR’s absence.

CSCAR also relieves pressure on faculty: Professors can direct their students to CSCAR for help with methods or programming languages outside their own expertise, allowing research to move forward more effectively. Closing CSCAR will shift this burden directly onto faculty — many of whom cannot realistically provide the same breadth of expertise — thereby weakening both training and research across the University.

Each year, CSCAR provides critical support during thousands of consulting appointments with students, faculty and staff from multiple schools and colleges across the University. CSCAR helps faculty strengthen grant applications by applying advanced methods, assists students and faculty in learning and implementing new techniques and improves the quality of publications through rigorous analyses and guidance during peer review. It is the place where researchers who lack specialized methodological expertise can access such expertise ensuring that cutting-edge methods are put into practice rather than remaining out of reach.

Graduate students in statistics, biostatistics and data science have also been central to CSCAR’s work, gaining hands-on training and support that has launched hundreds of successful careers. Eliminating CSCAR means losing not only a vital service for researchers, but also a proven pipeline for training the next generation of quantitative experts.

OVPR claims that there are other resources on campus that can effectively replace CSCAR, but these units are not designed, staffed or funded to absorb CSCAR’s volume or breadth of work. For example, the Biostatistics Department’s Statcom group exists to provide limited, pro bono consulting for community nonprofits — not to serve hundreds of campus researchers. Suggesting otherwise will quickly overwhelm that resource and undermine its mission. The Michigan Institute for Clinical and Health Research is focused on providing support for research involving translational science in health-related fields, which, while very valuable for the campus, is far too narrow of a focus for the wide range of researchers that CSCAR assists. In short, no patchwork of existing units can replicate CSCAR’s campus-wide role.

Client feedback has been overwhelmingly positive throughout the years, but, of course, CSCAR is not perfect. At times, clients have long wait times for appointments due to high volume. CSCAR staff and students learn how to provide statistical consultation on the job, and mistakes occasionally happen. Mismatches between clients and staff expertise also occur. Despite these issues, feedback was regularly solicited and CSCAR was constantly committed to improvement. While recent years have seen yet another evolution of CSCAR to focus more on machine learning and artificial intelligence tools, not every CSCAR client has an advanced technical skillset. However, the overall quality of services has been stellar, and appointments and workshops have continued to be in extremely high demand.

CSCAR has been an integral part of this campus for nearly 80 years, since its foundation at the Statistical Research Lab in 1946. It is virtually without parallel at other peer institutions, and there are no comparable services at the University.

OVPR should reconsider the decision to shut down CSCAR because the assistance it provides for research and teaching are enormously helpful to the success of the University, and there are not equivalent resources to replace its loss. Little money is being saved with this decision, the quality of the quantitative research being conducted at Michigan will be substantially affected and the proposed alternatives are vague and insufficient.   

Rachel Best is an associate professor and director of graduate studies in the Department of Sociology.

Michael Elliott is a professor of biostatistics and research professor of survey methodology at the Institute for Social Research.

John Kubale is a research assistant professor at the Institute for Social Research.

Robert Manduca is an assistant professor in the Department of Sociology.

Abram Wagner is an assistant professor of epidemiology and global public health in the Department of Epidemiology.

The post Closing CSCAR is a short-sighted decision appeared first on The Michigan Daily.