Coleman Harris

Inspired by a high-school teacher who taught the beauty of mathematics, James L. Rosenberger chose to study mathematics at Eastern Mennonite University. In the final year of his undergraduate degree, he was introduced to statistics. He recalls he particularly enjoyed this course in statistics and FORTRAN programming that, combined with his math degree, landed him his first statistics job as a data assistant for a cardiovascular research group at New York University.

While learning to curate data and perform statistical analysis, Rosenberger earned his master’s from the Polytechnic University of New York. Subsequently, his doctoral work at Cornell University cemented his interest in applied statistics, as he had the opportunity to collaborate and consult with researchers in both agriculture and medicine. Rosenberger took a tenure-track position in the statistics department at Penn State, where one of his major responsibilities was to lead the recently formed Statistical Consulting Center, following his PhD. He then took a brief leave of absence to pursue a postdoctoral position at Harvard University working on robust statistics and scientific reporting with esteemed statistician Fred Mosteller—another opportunity to collaborate across disciplines.

Rosenberger turned the theoretical norms of his statistics department on its head and was awarded tenure with a body of work concentrated on applied statistics upon his return to Penn State. He then focused on shaping the philosophy of the consulting center around three goals: training students to deliver sound statistical advice to researchers; improving the research quality of clients using the center; and finding stimulating research opportunities requiring new statistical methodology.

A researcher with a propensity for side trips, Rosenberger took two years to help develop a college of agriculture in Zimbabwe in the model of an American land-grant university, which focused on useful research for the local economy and agricultural sector, particularly for the benefit of the region. Returning to the United States, he spent 15 years as the department head at Penn State, seeking to build a broad department—from theoretically rigorous statisticians to strong applied statisticians. He then spent two years on sabbatical at the National Science Foundation as a statistics program officer, enjoying the exposure to various disciplines using statistical methodology to improve their respective scientific fields.

When he returned to Penn State, Rosenberger created a bioinformatics consulting center to embrace the rapidly changing field, one that includes constant statistical challenges associated with large data dimensionality and reproducible results. Stepping down as department head, he served as a local councilman for the Borough of State College Pennsylvania Council. He cites the experience as an enjoyable venue to flex his leadership and collaboration skills, teaching him much about the value statisticians can have in consulting relationships.

Since 2017, Rosenberger has served as the director of the National Institute of Statistical Sciences (NISS)—a national non-profit organization that brings together the academic, government, and industry sectors around statistical-and data science issues. Among its diverse interests, NISS focuses on delivering sound research related to public policy and statistical issues in the rapidly growing field of data science. Rosenberger sums up this vision best: “As statisticians, we need to recognize that we are not the whole field of data science, but that statistics is a vital component of what needs to be done.”

James Rosenberger on Consulting

Would you consider yourself a statistical consultant? If not, what would you call yourself?

Well, I have used the term statistical consultant most of my career. I would, however, recommend for most academics involved in statistical consulting to actually think of themselves as a collaborator. It’s a term that implies a broader role, a more equal status with the project leader. I think over the years, that’s become quite clear to me, and it resolves a lot of the stressful issues statistical consultants face getting adequate credit for our contributions.

What have been some of the most rewarding collaborations throughout your career?

It’s more of a bundle of applications I found rewarding, and that had to do with my work with agricultural scientists. The reward for me was understanding their reasons, challenging appropriately, and then applying the best methodology or experimental designs that fit their situation [so they could] do their work more efficiently. Another area I worked in for quite a while was transportation research. And again, in that setting, I was often the statistician who understood the research and, by learning enough of a particular discipline, you become a really vital member of the team.

Even before coming to Penn State, I worked in a medical center where my role was to practice the art and science of statistics when things were not as rigorously designed as modern science dictates.

What advice would you give someone who wants to have a career as a statistical consultant?

One bit of advice would be to pick good scientists to work with—those who bring the best of their side of the collaborative role to the process. I would also say pick an area you know enough about that you can communicate with the folks you’re collaborating with. And last, pick areas you’re genuinely excited about, so it’s not a drudgery, but you have a passion for really getting into the heart of the science you’re collaborating on. I think that goes a long way toward successful collaboration.

What makes a good statistical consultant?

One important thing is the ability to communicate well and to be interested in what the collaborator is doing. So, in general, the ability to communicate both orally and in writing, so you can express yourself and be understood—those communication skills are very valuable.

What in your training best prepared you to be a good, well-rounded statistical consultant?

I understand a lot about broad scientific questions and how things work. And partly, that’s a matter of growing up in the small business environment, where I understood the financial side, the transportation, logistics—a whole broad range of things one learns if they grow up on a farm or [around] a small business. So, I think that was part of why I could communicate with a broad spectrum of scientists, just my ability to understand what they were trying to do and interact and focus my contribution on the statistical issues that needed to be solved.

What is the most important career advice you have received?

Pursue the things that bring you joy. You can’t be successful doing something that is drudgery to you. Ask yourself periodically what part of your life or part of your job is most rewarding, and then pursue that aspect. That’s why when I stepped out as department head, I jumped into the political arena. I enjoyed the interaction with a wide range of people, and I think I was fairly good at resolving conflict rather than contributing to it.

How did your work in academia prepare you for your government roles, both with the NSF and NISS?

I think my general understanding of science and how things work provided a great background for being an effective statistical consultant. I was usually a quick study and able to understand the needs of a scientist asking for advice or wanting to collaborate on a research project. I think the ability to listen and understand the perspective of your collaborator is very important. Then, bringing my understanding of statistics and statistical tools—including design of experiments and general linear and nonlinear modeling—allowed me to be successful working with scientists in a variety of fields.

How does a statistical consultant role differ for someone who is in academia compared to someone working in government? Was there anything new you had to learn to succeed in the government roles?

During my time at the NSF, I was exposed to a broad range of research proposals and had to ensure the review process was fair to innovative and even unconventional ideas. What I enjoyed most was working on cross-disciplinary initiatives that linked the mathematical sciences—including statistics—with various disciplines that needed good statistical methodology to make valid and reproducible conclusions.

From a long-term perspective, my most challenging role was as chair of the Mathematical Sciences Working Group, which was directed to develop an initiative to support NSF proposals to increase the budget. We built our case on the important role the mathematical sciences could play in improving the quality of the sciences broadly. The impact of this strategy was seen by a doubling of the division’s budget over the next five years.