A national expert in biostatistics and policy evaluation, Elizabeth A. Stuart, PhD, is a Bloomberg Professor of American Health at Johns Hopkins University. She is also Associate Dean for Education and a Professor in the Department of Mental Health, with joint appointments in the departments of Biostatistics and Health Policy and Management. Dr. Stuart’s primary research interests are in the development and use of methodology to better design and analyze the causal effects of public health interventions.
In this interview, Dr. Stuart offers a unique perspective on how funders can best support the role of statisticians in biomedical and public health research.
How would you describe your research to a layperson?
I develop statistical methods and study designs to help evaluate the effects of policies and programs across public policy and public health. I am particularly interested in methods that help us understand how well programs and policies will work in particular target populations of interest, and in thinking about the trade-offs between different study designs.
Describe your experience with funders (private, NIH, others). What do funders do right? What could they improve? If you could restructure how scientific funders (NIH/NSF/etc.) are organized or how they operate, what would you change?
I have received funding from a variety of funders – NIH, IES, PCORI, WT Grant Foundation, NSF.
There are a few things I particular appreciate in funders:
- Recognition that studies sometimes need to change, and that projects and aims evolve in ways that still can lead to excellent science even if it wasn’t exactly what was originally proposed,
- Operation as a partner in the work, involved to, for example, make connections with related work and projects, but also letting the science proceed without too much oversight and bureaucracy, and
- Proposal procedures that are not too burdensome, especially for relatively small amounts of money – very few proposals should require more than 15 pages, and many can be < 5!
One thing I would say that the National Institute of Mental Health does very well is to have multidisciplinary grant review panels. And for NIH in general there is a real appreciation for the value in making sure grant teams have a range of experiences and expertise covered – this really helps facilitate a culture of interdisciplinary work and not working in silos.
For example, on the mental health services review panel I chaired for a few years it was basically impossible for a grant to get funded if it didn’t have a collaborating statistician as well the right mix of clinical or substantive experts. And any projects that used qualitative or mixed methods had to have experts in those areas. This really helped prevent people from staying in their own little “lane” and not learning from those around them with highly relevant expertise.
Surveys show that scientists say they spend upwards of 44% of their time on proposals, reports, IRBs, budgets–that is, administrative and regulatory requirements. Is that consistent with your experience? Is there anything that could be streamlined?
I personally don’t spent 44% of my time on that but that is in part because I have excellent grants management help at Johns Hopkins who do all of the heavy lifting with respect to budgets and budgeting, and the work I do is generally exempt from full IRB review.
I very much appreciate the need for streamlining, though; on some projects it feels like I spend almost as much time applying for the funds, filling out progress reports, and checking in with the funder than I do doing the actual work. For example, in almost all cases, annual reports should be sufficient, without need for anything more frequent than that.
If you had no constraints in terms of funding or the need to publish, is there anything that would be different about your research?
I am pretty happy with the mix of work I do now! I am lucky to be able to do a combination of statistical methods work and applied work. That said, given that my statistical methods work does sometimes just require my own time and maybe a student, it does sometimes feel like a lot of work to have to go get grant funded to do it! (This is unlike other projects where people need funding to implement programs, travel, compensation of staff, etc.).
The other thing I would like to do is spend more time on scientific dissemination and statistical practice activities, which are hard to get funded through grants, but my soft money position makes it hard to spend substantial time on them without salary coverage. Similarly, I have thought about writing a book on causal inference but it is hard to find time for that given the need to stay funded on grants.
If you could change the organization or management of universities, what reforms would you recommend?
I am always happy when people get out of their silos and work across areas. There are two ways I think we do that well at Johns Hopkins. First, PhD students in public health have to have an oral exam committee (and their final defense committee) made up of faculty from at least 3 departments in the School. This forces the students – and faculty – to learn to talk across fields and areas, and we have found some wonderful collaborations through that. Similarly, the PhD curricula for students in each program cut across Departments – they aren’t just taking classes within their own Department but rather from across the School and even the University in some cases.
Second, the University has an internal funding award – called Discovery Awards – that have a nice and short application (< 5 pages) and are designed to facilitate cross-Division collaboration (e.g., public health and education, or medicine and engineering), between faculty who have not previously worked together. These structures and incentives can really help get people talking across fields.
What’s a paper in your field in the past 1-5 years that you wish you could have published?
Great question! This is somewhat cheating because it’s more like 10 years old but I just love this paper by Jose Zubizarreta, Magda Cerda, and Paul Rosenbaum.
What I love about it is that it has some very sophisticated statistical concepts in it – core foundational ideas about study design and limiting bias due to unobserved confounders – but it is written in an incredibly accessible way, and with a very compelling applied example.
Given the goal of improving the practice and funding of science, is there anything else I should have asked you?
I can’t think of anything!