Member Spotlight Sherri Rose
As a third-year PhD student in biostatistics at UC Berkeley, this is a particularly exciting time for me. My dissertation research with advisor Mark van der Laan focuses on causal inference methodology for biased sampling designs, specifically case-control studies. Targeted maximum likelihood estimation (TMLE) causal inference methodology has been actively developed in the van der Laan group for several years. It is a two-step procedure in which one first obtains an estimate of the data-generating distribution and then updates the initial fit in a step targeted toward making an optimal bias-variance trade-off for the parameter of interest, instead of the overall density.
The focus of my work has been case-control weighted TMLE, which uses the prevalence probability in case-control weights to estimate causal parameters not previously available for case-control studies. The method also can be adapted for specific types of case-control study designs, such as individually matched, frequency matched, nested, and incidence-density. We are now applying case-control weighted TMLE to interesting case-control data sets of all types in collaboration with the University of Pennsylvania, Fred Hutchinson Cancer Research Center, Stanford University Medical Center, Universite Paris Descarte, and the Dana-Farber Cancer Institute, among others.
The managing editor of Amstat News is searching for ASA members who are willing to put themselves in the spotlight and write a brief article about their life, to be published in an upcoming issue.
The article should be 1,000 or fewer words and contain professional and personal information. Please include a photo or two of yourself and email it to Amstat News Managing Editor Megan Murphy at firstname.lastname@example.org.
My journey into statistics did not, of course, begin here. I earned my master’s in biostatistics at Berkeley in 2007 and my bachelor’s in statistics from The George Washington University (GW) in 2005. As an undergraduate, I was constantly questioning whether to apply to medical school or pursue a PhD. Attending the Summer Institute for Training in Biostatistics (SIBS) at Boston University in 2004 solidified my thinking that a PhD was a better fit, given my interest in medical research.
SIBS is designed to expose undergraduate students interested in mathematics and statistics to the field of biostatistics and is also offered at North Carolina State University and the University of Wisconsin, Madison. I highly recommend this program to undergraduates considering a career in biostatistics. It is an excellent way to learn about the variety of careers and opportunities for biostatisticians while receiving hands-on training. SIBS was also the first experience in which I spent any concentrated time with other women who loved mathematics. Almost all of my classmates in the upper-division mathematics and statistics courses at GW were men. At SIBS, at least half of the attendees were women, and several of the professors were women as well.
After attending SIBS, I began researching and applying to graduate programs in biostatistics. Selecting a program is obviously a very personal choice and should be based on the factors that are most important to you. I visited several schools, ultimately choosing Berkeley because of the faculty and their research. (I should note I picked Berkeley before I knew whether they were going to pick me.) I couldn’t have made a better decision. My educational and research experiences at Berkeley have exceeded all my expectations. Not only are the faculty in my department at the very cutting edge of biostatistics methodology, but the opportunities for collaboration are endless. I’ve been so happy here that I decided to get actively involved in student recruitment for our department in my roles as the president of UC Berkeley’s Biostatistics Graduate Student Association and as a student ambassador for the School of Public Health. If you’re interested in applying, I highly recommend visiting campus and meeting with the biostatistics faculty at the pre-application information session, usually held in October of each year.
An outstanding education isn’t the only thing I’ve found on Berkeley’s campus. I met my boyfriend, Burke, who is a systems administrator, in the biostatistics computer lab in 2005. It seems only fitting for a biostatistician to find her partner amidst the hum of a dozen servers, right? He never fails to bring the humor to my graduate student experience. Burke was also the first to predict that I would work with van der Laan for my doctoral research. A graduate of the University of Nebraska-Lincoln, Burke’s been at Berkeley for more than 10 years.
Returning to the present, I just attended the Joint Statistical Meetings in Washington, DC, where I was fortunate to present my work on TMLE for nested case-control studies, chair a session on missing data, and receive a graduate student award from the ASA Section on Statistics in Epidemiology. It’s amazing that graduate students are afforded such opportunities at the beginning of their careers. Seven of van der Laan’s other doctoral students also presented their work at JSM, and van der Laan gave a talk in an invited session. Thus, it definitely felt like a family affair. It was also my first time at JSM, and I was particularly fond of the “first-timer” ribbon attached to my badge.
After my positive JSM experience, I sit here looking forward to the coming year as we dive further into collaboration and application of our methodology. For those interested in learning more about TMLE methods for both observational and clinical trial data, van der Laan will be publishing an open-access compilation of papers on TMLE this month. The collection is titled “Readings in Targeted Maximum Likelihood Estimation, First Edition” and contains more than 15 papers, most of which have also been published in peer-reviewed journals.