A major overhaul of how we educate graduate students in biomedical research is long overdue.
Science has changed dramatically over the past three decades. The amount of information available about biological systems has grown exponentially. New methods allow us to examine the inner workings of cells with unprecedented resolution and to generate expansive datasets describing the expression of every mRNA or metabolite in a system. Biomedical research is becoming increasingly interdisciplinary and collaborative, and the questions we seek to answer are more and more complex. Finally, as the scientific enterprise has expanded, Ph.D.s have pursued increasingly diverse careers in the research and development, education and related sectors.
Despite these major changes, we educate Ph.D. students in biomedical research in essentially the same way as we did 25 or more years ago. As Alan Leshner put it in a recent editorial 
in Science magazine, “It is time for the scientific and education communities to take a more fundamental look at how graduate education in science is structured and consider, given the current environment, whether a major reconfiguration of the entire system is needed.”
Problems related to the reproducibility and rigor of scientific studies are likely driven in part by the inadequacies of an outdated system for educating our trainees. When nearly any student can sequence hundreds of millions of bases of DNA in a few days, does it make sense that all of our students are not given a significant amount of training in quantitative and computational analyses? And as we delve into more complex biological systems, shouldn’t students be receiving in-depth training in rigorous experimental design and data interpretation before they embark on their thesis work?
We have begun efforts to catalyze this much-needed change in graduate education. We and other NIH institutes and centers are supporting the development of training modules that can be used to enhance students’ abilities to conduct rigorous and reproducible research. We also recently issued administrative supplements to some of our predoctoral T32 training grants to enable the development of new curricular components in areas related to the conduct of rigorous and reproducible research or to build skills needed for a variety of different scientific careers. We plan to reissue modified versions of both of these funding opportunities one more time.
But this is just the beginning of what we intend to be a major effort to promote the reworking and revitalization of biomedical research education and training. As always, a key part of this effort will be facilitating a broad discussion within the community about challenges, opportunities and strategies for moving forward. These conversations, which have already begun (e.g., Future of Bioscience Graduate and Professional Training, 2015 ), should build upon previous calls for educational reform at both the graduate and undergraduate levels (e.g., Reshaping the Graduate Education of Scientists and Engineers, 1995 ; BIO2010: Transforming Undergraduate Education for Future Research Biologists, 2003 ; Vision and Change in Undergraduate Biology Education; Advancing Graduate Education in the Chemical Sciences, 2013).
There are many important issues we need to address, but a few essential ones are:
- Enabling institutions to do experiments to identify optimal models for training.
- Focusing on the development of the skills needed to be an outstanding scientist (see Figure 1 above). These skills range from the hard (e.g., quantitative and computational) to the operational (e.g., experimental design and interpretation of data) to the soft (e.g., communication and teamwork).
- Using evidence-based approaches to education.
- Incorporating active learning strategies into curricula.
- Optimizing the size of graduate programs and the research environments in which students work.
- Emphasizing mentoring throughout training, including through the use of structured exercises such as individual development plans.
- Facilitating student career development and ensuring that students find the right career paths during graduate school.
- Identifying and emphasizing best practices for mentors as they design and implement new education and training models.
- Creating and sustaining a diverse scientific workforce.
- Tracking student outcomes and evaluating program results.
- Making the aggregate information from these outcomes analyses easily available to current and prospective students.
- Creating a culture that continually optimizes training strategies to meet the evolving needs of the scientific enterprise.
We look forward to hearing your thoughts on how we can ensure that we are giving the scientists of tomorrow the skills and knowledge they need to push biomedical research forward as efficiently and effectively as possible.
Very nice summary of some festering problems with graduate education, and valuable steps toward the cultural change required for widespread implementation.
Some of these general issues with graduate education even made it to the op-ed page in the Los Angeles Times this morning (“Preparing PhDs for the real World”). This problem has been discussed for a long time, but the necessary changes have been thwarted by incentives of PIs to continue using graduate students as the primary workforce that drives their research accomplishments.
Excellent piece, very scholarly written , enjoyed reading. Thanks for sharing
Thank you very much for this wonderful article that includes very important information about the range of skills necessary to become a great scientist. I believe it is critical for us to realize that people of different backgrounds might bring different foundations of hard and soft skills that require alternative approaches to help them become more successful scientists. For example, aspiring deaf/hard-of-hearing scientists are likely to be more successful if they are provided with supplemental academic supports to address some of their foundational hard skills weaknesses (English writing skills) in addition to deaf-specific mentoring to address some significant soft skills weaknesses that frequently appear in this group of aspiring scientists.
While all the bulleted points are valid and are deserving of consideration, the elephant in the room to me is that we now expect individual faculty to support biomedical students financially. Few universities support graduate student stipends beyond the first year. This means that faculty have to support them on their grants, including R01’s and other “non-training” sources of funds. This results in far too many faculty having to push students for data generation rather than taking the time to properly train the student in problem solving and expanding their scope of expertise. I’ve seen many faculty deny students opportunities to attend conferences, give seminars, etc., not even apply for fellowships because it takes too much time away from the bench. This has also led to bare minimal coursework and so our students do not have the breadth of instruction they need despite the ever-expanding knowledge base needed for collaborative and interdisciplinary research. Some of my colleagues have argued that the cream will rise to the top and the competitive students will succeed, but is that an efficient use of our limited time and money? I think not.
This is a very important problem indeed. As computational analysis has become almost integral to any biomedical research effort, one key aspect of the problem is that related with the difficulty of delivering computational analyses methodologies in a way that is reproducible and that encapsulate the entire content of a project’s analysis from data acquisition to publication. Jill Mesirov and I have been trying to address this problem for years and more recently we submitted a U24 grant to developed an environment to do this in the form of electronic notebooks (NIH U24CA194107). There also are several efforts by others across the community in the same direction but a missing piece in the game is having an effective incentive for students and researchers to actually use those environments e.g. every time that a paper is published. This is part because it takes a lot of additional work to make a computational analysis complete, clean, well documented and reproducible to others (besides the person who did it originally).
My specific suggestion to address this problem is to provide some sort of administrative supplements, maybe attached to standard grants, or made specific for students and maybe postdocs and researchers too, that supports those cases where people go that extra mile and provide a complete and reproducible version of an analysis as part of the publication of a paper. A variant of this can perhaps target academic courses of seminars where again, incentives are provided to students, teachers or professors that provide an entire course embedded in a computational environment where people can easily reproduce the results, learn from them, and use them as an effective starting point or building blocks for new projects, etc.
The benefit of this is not only that those specific analysis are made publicly available and not lost, but also that people can learn to think with a higher degree of scientific rigor and with the understating that any real scientific methodology has to be one that others, besides the originator or original author, can understand, reproduce and apply easily and effectively.
We have addressed “Incorporating active learning strategies into curricula,” in the Rutgers-Newark doctoral program in Psychology, as follows.
Three years ago, we made a significant change in the format of our qualifying exam, a change that was initially met with resistance by our faculty, but which is now unanimously and enthusiastically supported by our faculty and students. The previous format utilized comprehensive exams that basically repeated final exam questions from the courses the students had taken during the previous two years. In preparing for the exams, the students felt that they were back-tracking, and the faculty found reading the students’ exams tedious. Our new format requires that our students write a grant application using the NIH guidelines. Now the students feel that the process accelerates their progress toward their degree because they develop a literature review that they can apply to their dissertation, they are challenged to identify a gap in the literature, formulate hypotheses and experiments, anticipate results, and provide potential interpretations, and implications. This can all be applied to their doctoral dissertation. We faculty get a much clearer idea, than before, of each of our student’s scientific thinking, creativity, and ability to communicate. Our students benefit from feedback from diverse faculty perspectives. This format familiarizes the students with the requirements of a grant application; it helps to demystify the process and alleviate their apprehensions, and advances their career trajectory.
Thanks for wonderful article highlighting the graduate education as key priority for better, modern research workforce. With changes via common core in K-12 with more energy on pre-college curriculum and linking {undergraduate] advancement will certainly prepare better flow of trained, talented students ready for final transformation into graduate program.
Thank you for a most informative and provocative commentary on the current state of graduate education. Overlooked in the article itself, and the astute responses, is the inexorable pressure on graduate faculty to justify their existence through acquisition of extramural funding. The effort required to achieve this goal erodes focus on faculty’s responsibility to train the next generation of biomedical scientists. Department chairs and College Deans, whose performance is necessarily tied at least in part to success of their faculty in the race to funding independence, need to realize that support of graduate student training in the form of teaching effort compensation (graduate students after all are critical elements of funding success) is essential if all parties (Deans, Chairs, faculty and students) are to be recognized as sine qua non for fulfillment of the institution’s mission.
Thank you very much for this summary of potential goals and of some of the key challenges faced in reforming graduate education. My department has been part of two major national studies, the Carnegie Initiative on the Doctorate and the ACS study on Advancing Graduate Education in the Chemical Sciences. With the full support of our administration and faculty, we launched an entirely new graduate curriculum this academic year that addresses most of the recommendation in the first (of four) conclusion of the ACS report that you cite. Adding the soft skills you mentioned to the curriculum can in fact be accomplished without sacrificing mastery of the subject or increasing the time to degree. Further details concerning the curriculum can be found at ChemGradEd.com (including a seminar that was delivered at the NSF Division of Graduate Education in December). I agree with David Pollock, a real challenge going forward is the reform of the financial model for supporting the education of graduate students. In fact, this is the second conclusion of the ACS report.