NIH-based training grants are becoming increasingly competitive, and thus it is critical that students appointed to them make a long-term commitment to carrying out research in the relevant subject area. All training grants have requirements which must be fulfilled to justify the financial support of the student (tuition, fees, and stipend).
For the environmental statistics training grant, these include the following:
Coursework: The doctoral program in Biostatistics at HSPH combines first class training in statistical theory and methods with training in various substantive areas of public health, computing, consulting and teaching. To receive his or her degree, a student must earn a total of at least 70 course credits, and pass both a written qualifying exam (offered in the winter session between the 3rd and 4th semesters of coursework) and an oral examination before beginning dissertation research. The Department of Biostatistics explicitly specifies all of these requirements for the doctoral program in the Department’s Graduate Student Handbook, and makes this document broadly available on the Department’s website. The required and recommended coursework is designed to develop competency in several important areas, as defined below.
- Statistical Theory and Methods: Students complete at least 44 credits of coursework in advanced biostatistical theory and methods. Required courses include a semester of statistical methods (BST 232), probability 1 (BST 230), and statistical inference (BST 231). Students may choose from a wide range of upper-level courses including study design, survival analysis, multivariate analysis, Bayesian methods, advanced regression and statistical learning, advanced probability theory, statistical genetics and various special topics courses. Advanced courses from MIT and other departments at Harvard can count towards this requirement once approved by the Department’s Director of Graduate Studies.
- Statistical Genetics and Genomics: Although not required, students have the option to take several advanced courses in the area of statistical genetics. Current offerings include “Fundamental Concepts in Gene Mapping” (BST 227), “Introductory Genomics & Bioinformatics for Heath Research” (BST 280), and “Genomic Data Analysis” (BST 281), all aimed at both Masters and PhD level students. “Advanced Statistical Genetics” (BST247) and “Advanced Computational Biology” (BST 290) are PhD level courses. Courses are also available from the Department of Epidemiology, such as “Principles of Genetic Epidemiology”, (EPI 507) and “Analysis of Genetic Association Studies” (EPI 293).
- Computation and Software Package Development: The Department of Biostatistics requires instruction in computer science in its doctoral core, with all students required to take “Introduction to Data Structures and Algorithms” (BST 234) in the second semester of the program. Trainees also have access to a wide range of courses focusing on high-performance research computing. In addition to working with software engineer Naeem Khoshnevis (see Section B.2), formal training in R software package development is available in “Big Data Computing in R” (BST 262). In addition, trainees often take courses in the Computer Science department, such as “Advanced Machine Learning” (CS 281) or “Visualization” (CS 171).
- Data Science: The Department of Biostatistics’ doctoral course “Advanced Regression and Statistical Learning” (BST 235) covers the theory and application of statistical and machine learning methods, and is taken by almost all of our doctoral students. Additional offerings include “Data Science 2” (BST 261), which covers deep learning methods, and “Introduction to Social and Biological Networks” (BST 267).
- Epidemiology: Trainees are expected to be familiar with the fundamental principles and practices of epidemiology, including the design and analysis of cohort, cross–sectional, and case–control studies. All students are required to take the introductory course EPI 201, and are encouraged to take additional advanced courses.
- Responsible Conduct of Research: First year doctoral students and postdoctoral fellows in our Department are required to take a 1.25 credit HCSPH course entitled “Responsible Conduct of Research” (HPM 548), taught by Dr. Delia Wolf, Senior Lecturer, Health Policy and Management, and Associate Dean, Regulatory Affairs and Research Compliance. Further, all trainees (pre-doctoral and postdoctoral are required to take a number of Biomedical modules from the CITI (Collaborative Institutional Training Initiative) Program Human Subjects Research Catalog. See Section 3, “Plan for Instruction in the Responsible Conduct of Research” for details.
- Scientific Collaboration: The Department of Biostatistics requires doctoral students to satisfy a 2.5 credit requirement of formal statistical consulting and collaboration. The course helps trainees develop skills in effective consultation and collaboration with non-statistician investigators in the planning and analysis of studies, including research ethics and IRB requirements. This course is currently offered with faculty instruction through the graduate student-run Biostatistics consulting lab, founded by two T32 former trainees (Emily Slade, Katrina Devick) and recently directed by another (Christina Howe).
- Principles, Statistical and Computational Tools for Reproducible Science: The Department requires all doctoral students to take a 2.5 credit (half-semester course) “Reproducible Data Science” (BST 270). The course is offered in a “flipped” format, in which students watch pre-recorded lectures outside of class and then attend hands-on coding in-person sessions with the faculty instructor. The pre-recorded material was developed by the Department of Biostatistics and is offered as an online course “Principles and Statistical and Computational Tools for Reproducible Science” on Harvard’s online learning platform, HarvardX. See Section 4 for details. Also, as part of the dissertation proposal oral exam, plans for reproducibility is an area within the scope of examination of the student’s proposal.
- Causal Analysis Methods: The Departments of Biostatistics and Epidemiology at HCSPH have a series of courses in this area. Offerings include “Advanced Epidemiological Methods” (EPI 207), “Models for Causal Inference” (EPI 209), and “Theory and Methods for Causality 1” (BST 256) and “Theory and Methods for Causality 2” (BST 257), which together cover state-of-the-art methods for drawing causal conclusions from experimental and observational studies and establishes their theoretical properties. The course “Methods for Mediation and Interaction” (ID 542) contains vital material for mediation analyses of health effects of one or more environmental exposures.
- Cognate Requirement. The Department requires students to explore in some depth a selected cognate field, a non-quantitative field outside of biostatistics or statistics. Students must complete 8-10 credits of ordinally graded courses of study in the cognate field. Students supported by this training program are required to choose a minor of relevance to environmental health. The School offers a broad array of suitable courses. Some particularly popular past course choices for our students have been: “Introduction to Environmental Health” (EH 201), “Human Physiology” (EH205), “Cardiovascular Epidemiology (EPI 223), “Epidemiology of Environmental and Occupational Health Regulations” (EH 236), and “Environmental Epigenetics” (EH 298), among others.
Independent Research Starting in Second Semester: Pre-doctoral trainees are required to sign up for at least two research credits per semester beginning in their second semester (Spring) in the program. They work with a faculty member on an applied or methodological research project. Trainees are teamed with one primary (Biostatistics) and one secondary (Environmental Health, Epidemiology, SBS) preceptor. Each Winter, the Program Co-Directors meet with each first-year trainee to discuss a list of available projects and faculty mentors. Students then continue independent research during a summer project after their first year in the program. For example, summer projects in recent years include research related to climate exposures and preeclampsia and gestational diabetes in Project Viva (Christine Howe with Drs. Brent Coull and Diane Gold), PFAS exposures and women’s pregnancy outcomes in Project Viva (Sharon Caslin with Dr. Briana Stephenson and Emily Oken), and the impact of COVID-19 lockdowns on ambient air pollution levels (Kevin Chen with Drs. Rachel Nethery and Lucas Henneman). In year 2, each pre-doctoral trainee is required to present their summer project work to fellow students, postdoctoral fellows, and Department faculty at a one-day Departmental symposium.
Dissertation Work: After passing their written qualifying examinations in August of their first summer in the program, students are expected to identify an area of interest and begin defining their desired dissertation topic. Typically, students do this by talking to various faculty about their research and attending departmental seminars and working groups. Almost always, a student’s dissertation research will evolve out of their project from the first summer or something closely related, and environmental biostatistics trainees supported by this training program generally (but not always) choose one of the grant’s primary preceptors as their dissertation advisor. Once decided on a dissertation advisor and a suitable topic, the student must form a thesis research committee (typically made up of both Biostatistics and secondary preceptors), write a dissertation proposal in the form of an F31 grant proposal on the chosen topic, and defend this proposal to the committee. Once the student passes the oral exam, the student conducts independent research with their primary thesis advisor and thesis committee. The student must hold formal meetings with the committee once a semester, and progress reports must be submitted to the Department’s Academic Standing Committee on this same cadence.
Environmental Biostatistics Seminars: Environmental Statistics trainees (both pre- and postdoctoral) are required to attend environmental biostatistics seminars. The “Environmental Statistics Seminar Series” focuses on cutting-edge quantitative issues in the EHS and meets approximately once a month. Speakers in the series include faculty, students and postdoctoral fellows from Biostatistics, Epidemiology and Environmental Health, and prominent experts from outside the University. The speakers and topics for the seminar series are available on the Department’s website, as well as included in the Department’s weekly newsletter that is distributed by email each Monday afternoon. The NSAPH consortium also have regular meetings that many trainees also attend. Trainees have the opportunity to present in EHS seminar series as well, such as the Superfund Trainee Seminar Series, NIEHS chalk talks, and the Project Viva ECHO seminar series. NEW Given such a rich research environment with an assortment of series in which to engage, at the beginning of each year, Program Faculty will collate web pages and listservs of all of these activities and share with all trainees, and discuss which series might be of most interest to each trainee.
Departmental Colloquia are usually presented by well-known statisticians visiting from other institutions and provide the students with opportunities to learn about current research in many different areas. Many of the colloquia have relevance to the environmental health sciences.
Attendance at Professional Meetings: We strongly encourage all trainees to attend at least one professional conference each year. Popular choices include the annual meetings of the Eastern North American Region of the International Biometrics Society (ENAR); Women in Data Science (WiDS); the Joint Statistical Meetings (JSM); and the International Society of Environmental Epidemiology (ISEE). Further, we encourage both pre- and postdoctoral trainees to take the lead in organizing an invited or topic-contributed session. Trainees also have the opportunity to participate in interdisciplinary workshops held by the NIEHS, U.S. EPA, Health Effects Institute (HEI), or other environmental research agencies on quantitative issues in the EHS (e.g. the NIEHS PRIME workshops). Other opportunities include meetings of the scientific advisory committees associated with the HCSPH-based research centers and this T32 program. We will continue to make attendance at these smaller venues a point of emphasis, as this provides invaluable networking opportunities for trainees.
Grant Writing: Grant writing is built into a trainee’s program at several stages. The oral exam is required to be written in the form of the scientific portion of an F31 application. While not required, students with particularly strong proposals are encouraged by their committees to formally submit this proposal for F31 funding to NIEHS. NEW: In this case, in addition to feedback from the student’s thesis committee, the student has the opportunity to present the aims of the proposal for feedback at one of the Harvard NIEHS P30 Center’s Specific Aims Review Sessions, which are led by experienced grant writers Diane Gold (preceptor in this program) and Frank Speizer, the founding Principal Investigator of the Nurses’ Health Study (see letter of support from Center Director Marc Weisskopf). Third, predoctoral trainees have many opportunities in the third year to assist a primary preceptor on an NIH interdisciplinary research proposal. There is no shortage of proposals going in at any one time, with each primary preceptor typically collaborating on at least 3-5 interdisciplinary proposals at any given time. This research activity was initiated at the beginning of the last funding cycle as a result of suggestions by multiple alumni in our anonymous surveys (see Section B.4.a). Former students noted multidisciplinary grant writing is a key skill to have upon graduation, as young investigators are expected to start these activities upon acceptance of a new academic research position.
Annual Workshop on Environmental Statistics: We will hold an annual day-long-retreat where students and faculty connected with the Environmental Statistics Program come together in an off-campus location. The workshop will consist of work-in-progress presentations made by pre- and postdoctoral trainees as well as keynote talks presented by either an internationally-renowned environmental biostatistician or a faculty member from HCSPH. In the past, this format is effective in generating interest in and commitment to environmental health issues among the trainees, as the environmental presentations often entice one or more of our younger trainees just embarking on their dissertation research to work on a problem described in one of these talks.
Focused Multidisciplinary Working Groups: On a number of occasions, the Environmental Statistics Program has formed small, focused working groups comprised of students, fellows, research scientists, and faculty, with the goal of learning together about an emerging topic. Typically these groups are initiated and organized by preceptors jointly with T32 trainees, and topics are selected according to their interests in any given year. During the current funding period, in 2019 Dr. Rajarshi Mukherjee jointly with T32 trainee Aaron Fisher organized a reading group on deep learning methods, and in 2020 Dr. Rachel Nethery jointly with T32 trainee Kevin Josey organized a reading group on causal inference and machine learning (see Section 6.1 in the Progress Report for specific topics covered in each).
Symposia: Every year or two, Environmental Statistics faculty organize one- to two-day symposia on a particular topic that requires longer discussion than that provided by an hour-long seminar.
Progress Report: Trainees will be asked to provide an annual progress report including details on how their training relates to environmental statistics, for the annual NIH progress reports.