Present: Faculty: Karen Bandeen-Roche; Ron Brookmeyer; Brian Caffo; Ciprian Crainiceanu; Marie Diener-West; Francesca Dominici; Constantine Frangakis; Rafael Irizarry; Elizabeth Johnson; Kung-Yee Liang; Tom Louis; John McGready; Lucy Meoni; Roger Peng; Luu Pham; Chuck Rohde; Ingo Ruczinski; Dan Scharfstein; Carol Thompson; Rick Thompson; Jim Tonascia; Mei-Cheng Wang; Scott Zeger; Staff: Mary Joy Argo; Cindy Hockett.

Welcome and Overview

Scott Zeger welcomed everyone and congratulated all of us on another successful year.  Planning for our self-study provides us with the opportunity to reflect on where we've come from as a department and where we're going.

By way of background, Scott explained that over the last decade, science has undergone changes and, as a result, so has our field.  In 1997, our computer speed was much slower, the School's email system was only about a year old, there was no Google.  Today, we operate in a high-speed computing environment, Google has conducted at least 90 million searches, and there now exist large, publicly-available datasets and databases that are a direct result of explosive growth in the fields of genome sequencing, imaging, and biomarkers.  Our department has been impacted by these changes in many ways but particularly by the fact that we are now a larger department (almost 50% more faculty as ten years ago) and a more diverse department (both in terms of nationality and in terms of research).  Our department now has a number of working groups targeting specific scientific areas -- for example, genetics/genomics, environmental health sciences, imaging, causal inference.  As well, the nature of our research has become more diverse. Ten years ago, faculty seldom depended upon infrastructures to carry out their research programs; now, several do.  Similarly, the nature of our published research has changed.  We now publish in a much larger range of journals, some of which have very short time scales.

Being bigger and more diverse raises challenges, some of which were made apparent in our recent faculty search.  Through discussions about these challenges, three guiding principles seem to have emerged:

1. All of us -- faculty, students, and staff -- want to be part of one of the best departments of biostatistics.

2. Faculty want individual respect and equitable treatment wherever they work along the spectrum of research.

3. Faculty want to keep communication channels open so that differences get aired and resolved.

There was agreement that we have sufficient common ground to address challenges and resolve differences of opinion.

Break-Out Group Reports on Preliminary Answers to the Following Core Questions about the Department's Future

1. What are the primary missions of the department and what are the principles we should follow to remain among the best? (All participants)
   
   It was suggested that the words "novel" and "ultimately" be removed from the mission statement.
   
   There was agreement that even though we cannot fill the demand for all biostatistical research support, the Center can support shorter-term consultations.  A core issue seemed to be whether we should be driven by the amount of research we do or by the amount of education we do.  Which is the driving force?  The group consensus is that education should be the driving force, since our department is the home of statistical education for the School of Public Health.  There was enthusiasm for communicating in the report our need to be selective about which substantive areas we interact with.  We cannot support or compete for every single grant or research opportunity that comes our way.  While focusing on education and selected research problems, we must also work to remain central to the School and University.
    

2. Is our PhD program properly structured and implemented, or do we need to make further changes so that our graduates are best prepared to make major contributions to biostatistics and health throughout their careers? (Facilitator: Liang; Participants: Argo, Crainiceanu, Frangakis, Irizarry, Wang)
   
   Kung-Yee Liang reported that this group felt unanimously that it was important to maintain our PhD program as our lead program.  While there was also agreement that our graduates make worthwhile contributions, there was also the recognition that the nature of those contributions varies depending on their type of position.  Academic biostatisticians may have many more publications than biostatisticians working in industry.  Some of our graduates may make methodologic advances, while others contribute to the work of clinicians, which in turn helps advance science in general.   This group also praised Dan Scharfstein for his leadership of the graduate program over the past several years.
   
   Some suggested improvements include:
   
   -- Make students more independent earlier in their research.
   
   -- Set a solid expectation that all students publish their thesis research (perhaps have this explicitly stated as a mission of the program?).
   
   -- Pre-screen applicants who have a high likelihood of becoming independent researchers.  This could be done by looking at past admissions 
      data for predictors, considering incentives to bolster recruitment, and making our program standards more rigorous.
   
   -- Identify training grant-like funding sources for non-US students.  (A related topic of discussion concerned whether or not faculty should
       shift some of their focus from substantive grants to methodologic ones. Would this free up more faculty time for student advising?)
   
   -- Enlarge the geographic diversity of our applicant pool.  Concentrate our Sommer Scholar nominees on foreign applicants.  Set aside more
       general funds support for foreign students.
    

3. What is the potential for master's-level education in biostatistics and bioinformatics for the next 5 years?  Should the department continue to focus on doctoral education or expand its master's programs?  If the latter, in what direction? (Facilitator: Brookmeyer; Participants: Ruczinski,  Scharfstein)
   
   Ron Brookmeyer reported that this group saw both strengths and challenges in our existing master's programs.  One was the need to clearly differentiate the ScM and MHS (biostatistics) degrees.  Given our recent decision to try to shorten the time for the ScM degree, how will our one-year stand-alone MHS be different (other than that a thesis is not required)?  Should we discontinue the stand-alone MHS, since applications and enrollments in that program have been decreasing over the years? 
   
   On the ScM side of things, students in that program have expressed a desire for more educational integration with PhD students, as well as some dissatisfaction with their quality of life (ie, no designated office space).  Many faculty who advise ScM students are unclear about expectations for a master's thesis, while other faculty are uninvolved in master's advising.  It has been suggested that a formal internship requirement for our ScM program might prove beneficial for post-graduation job prospects, but there was recognition that such a program would need dedicated faculty oversight. There was agreement among the group that all faculty should play a role in our ScM program and that, in order to strengthen the program, we need to clarify what kind of master's-level statisticians we want to produce.
   
   Another concern is the state of our MHS program in bioinformatics, which, for reasons as yet unclear, has not grown as expected.  Given Karl Broman's upcoming departure from Hopkins, we will need to identify a replacement for him as a co-director of the bioinformatics program.  We may also need to look at our competitors programs in bioinformatics to see how ours differs. 
   
   Our concurrent MHS program in biostatistics (open to doctoral students from other divisions of the school and university) seems to be one of our department's great successes and has a positive effect on our relationship to other departments; we should continue to grow and support this program.  Some past discontent among students in this program may have been alleviated by our recent decision to assign concurrent MHS students a Biostatistics advisor.  There was speculation that perhaps we could partner with relevant Homewood undergraduate programs to develop a five-year BA/MHS that be similar in content to the concurrent MHS.
    

4. Do our current educational offerings for health professionals and scientists optimize the opportunity for learning by students in our school, university and beyond?  What are the best opportunities and strategies to improve the teaching of biostatistics and to expand our influence on public health through biostatistics education? (Facilitator: Diener-West; Participants: Caffo, Johnson, McGready, R. Thompson)
   
   Marie Diener-West reported that this group recognized the department's need to optimize statistical learning for public health professionals (both for courses of ours that they take and for their doctoral research).  Of concern is how well we integrate with other departments' introductory sequences -- Epidemiology, in particular?  Among the group was a suspicion that PhD and some ScM students from other departments are not having their statistical educational needs met; there was consensus that our 651-4 sequence should be taken in the first year by students from other departments, and then offer a second-year sequence that could cover advanced statistical methods for public health practitioners.  This new sequence could consist of 3-4 stand-alone terms; topics covered could include causal inference/study design, hierarchical data, multilevel data, and prediction and evaluation.  For non-Biostatistics students, this new sequence could replace our current LDA and multilevel models courses.  In the 3rd year, we might offer a doctoral seminar/data analysis workshop where students would bring their own thesis proposals and data.
   
   Some other ideas for improved statistical education for public health professionals included more onsite short courses, offsite short courses (ie, at a hotel near an airport), a data management short course, the hiring of a marketer who could sell our current collection of educational modules to industry and government agencies, and a collaboration with other institutions to put together a "wiki" book of basic statistical concepts.  The group also felt the need to have the department better communicate the importance of teaching assistants by emphasizing TA training, encouraging our students to develop a teaching portfolio, and by making teaching an integral part of their educational evaluation.
    

5. How should the department's size be determined in the future?  If we self-regulate our size below the demand, how should we organize our relationships to the growing numbers of research biostatisticians within Public Health, Medicine, and the University? (Facilitator: Bandeen-Roche; Participants: Dominici, Irizarry, Meoni, Rohde)
   
   In her overview of this group's deliberations, Karen Bandeen-Roche noted that no one thought the department should be a lot larger than its current size and that, as discussed earlier, education should be the driving force.  Some guiding principles emerged for the department's size were: a "committee of the whole" structure (ie, approximately 20-25 tenure-track faculty), a collegial culture, no large separate centers, and "right-sizing" to keep funding balanced.  In light of these principles, one challenge for the future might be how to address "spin-off"-type centers. 
   
   This group noted that some positive outcomes of our relatively small size were disciplinary excellence, leadership in important science, and a collegial environment. 
   
   In a discussion of the question "What is size?" this group felt that our size would comprise faculty, students, postdocs, and staff and that there should be the right mix between groups.  Size should be determined by our department's educational mission, which leads to the question of who we should be hiring.  We have a long-held tradition of hiring only the "best" candidates, but in light of sometimes-shifting educational and research needs, what does "best" mean"?
   
   There was also consideration of how we should related to statisticians outside of our department.  Given that we do not want to become a significantly-larger entity, there are several possible models we could follow.  The first is to continue with our practice of reaching out to other Hopkins statisticians via joint and adjunct appointments.  There was agreement that we should have uniform treatment of our joint appointments, rather than having special rules for certain joint appointees.  Another option is to pursue a "hub and spoke" model, similar to what we have with Karen Bandeen-Roche and the gerontology group/Center on Aging, where Karen is the liaison between our department and Aging statisticians.  A third way is to continue with our various working groups (ie, environmental health).  There was consensus that both Options 2 and 3 could benefit from more interaction with the departmental faculty at large.
   
   The question arose of whether and how best to allocate resources to the various working groups.  Some see discrepancies of treatment between the "hub and spoke" and working group models.  There was also agreement that the Genome Cafe no longer be associated with a particular research group but be open to all (and possibly have a name change). Finally, this group considered the larger issue of how to ensure that working groups do not get in the way of collegiality.  
    

6. What are the emerging research opportunities for the near term?  What role should the department play in facilitating faculty research -- for example, working groups or large research grants that may require investment in infrastructure?  What is the ideal role for the Center in supporting faculty research? (Facilitator: Louis: Participants: Hockett, Peng, Pham, C. Thompson, Tonascia)
   
   Tom Louis reported in writing that this group thought the guiding principle for recruitments should be to hire the best applicant so long as there is a reasonable fit with current or anticipated educational and research activities.  Hiring decisions should continue to be made in by the Department’s Committee of the Whole.
   
   The group discussed particular research opportunities are detailed in their written report.
   
   All infrastructure investment decisions and all but minor support of working groups and individual faculty decisions should be made through shared governance by the departmental Committee of the Whole. The department may initiate a seed grants program with proposals reviewed by a smaller group, but this group should report to the full faculty.
   
   This group felt that the Center has several current and potential roles, but is largely a shorter-term service group.  Details are in their written report.
   
   There was agreement that the Center plays an important role in facilitating faculty collaborations and there is the potential to do more.  For example, the Center handles some consulting requests that the faculty cannot handle. The Center is also useful for making short-term consulting arrangements (ie, having a student conduct analyses for a National Academy Panel and not having the faculty member engage in an individual consulting arrangement).  The Center facilitates short courses, scouts around for interested parties, coordinates arrangements and helps with preparing materials.  It has the potential to educate our graduate students on the technical, interpersonal, political and financial dimensions of consulting.