Applicants should submit a cover letter, CV, and contact information for three references to Walter Dempsey (wdem@umich.edu) with the subject line "Dempsey Lab Potential Postdoc Position".

The Department of Biostatistics at the University of Michigan School of Public Health is seeking a highly motivated individual with an excellent academic track record for a postdoctoral position. The department is rated as one of the nation's top biostatistics programs. This position will support the design and analysis of one of the first hybrid experimental designs in suicide prevention.  Beyond supporting the study, the position will focus on developing statistical methodology for hybrid experimental designs (HEDs).  Specifically, the position will develop causal inference techniques and reinforcement learning algorithms to be applied to data from these studies. The postdoctoral fellowship is meant to cater to the individual's long-term goals.   There will also be potential opportunities for working with junior trainees, management and direction of working groups and classroom teaching opportunities as they arise.

The postdoc will work with a team of faculty collaborators, including Dr. Walter Dempsey from the University of Michigan, and Drs. Rebecca Fortgang and Matthew Nock from Harvard University. 

This work will involve projects that include a combination of the following:

1. Developing and applying novel causal inference methods to inform the construction of multimodality adaptive interventions (MADIs). MADIs integrate human-delivered and digital components that are adapted at multiple timescales. This research aims to answer scientific questions about how to best combine these components when they are adapted at different rates, such as determining if the effect of a daily digital message changes when a participant also receives human coaching.
2. Building and refining experimental designs for sequential decision-making in mobile health, with a specific focus on hybrid experimental designs (HEDs). A HED integrates a Sequential, Multiple-Assignment, Randomized Trial (SMART) with a Microrandomized Trial (MRT), allowing for sequential random assignments at both slow (e.g., monthly) and fast (e.g., daily) timescales. The research will focus on developing methods that can analyze data from HEDs to inform the joint sequencing and adaptation of intervention components.
3. Applying reinforcement learning methods to develop effective and scalable psychological interventions. This research will leverage data from HEDs to understand the interplay between intervention components and how they influence proximal (short-term) and distal (long-term) outcomes.

The postdoctoral researcher will be mentored by Walter Dempsey and might also work with other faculty collaborators within the Department of Biostatistics, Statistics, Computer Science and Engineering (CSE) or the Institute of Social Research. The Department of Biostatistics at the University of Michigan is a leader in developing analytical methods to turn data into knowledge. Faculty and students conduct cutting-edge biostatistical research with over $50M in funded research annually. They are involved in a wide range of collaborative research activities with faculty across the University of Michigan campus, including the Schools of Public Health, Medicine, Nursing, and Dentistry, and the Institute for Social Research.

Duties and Responsibilities: support design, monitoring, and analysis of a hybrid experimental design (HED) for suicide prevention, develop new methods for causal inference/sequential decision making and its interface with suicide prevention research; develop software; and potentially collaborate on applied projects in Biomedical, Public Health and Social Science Faculty.  

Candidates should have a doctoral degree in biostatistics, statistics, computer science or a related field. Strong computational skills in R/Python and expertise in one or more of the areas of reinforcement learning and/or causal inference are desired but not required. Start date and term are negotiable.   An individualized development plan, mentoring, and travel support for at least two conferences per year will be provided as well as competitive salary and benefits.

Positions that are eligible for hybrid or mobile/remote work mode are at the discretion of the hiring department. Work agreements are reviewed annually at a minimum and are subject to change at any time, and for any reason, throughout the course of employment. Learn more about the work modes.

Job openings are posted for a minimum of seven calendar days.  The review and selection process may begin as early as the eighth day after posting. This opening may be removed from posting boards and filled anytime after the minimum posting period has ended.

The University of Michigan is an equal employment opportunity employer.