I have now been at my postdoctoral appointment for about four months. The postdoc is a bit of a weird position: you’re not a graduate student anymore, but you’re not quite faculty yet. Here are some things I’ve learned in the past few months in how a postdoc differs from a PhD:
1. Your time is limited. This is a job whose main purpose is to help you get your next job (ideally a tenure-track position if you are going the academic route). Even if you aren’t interested in academia ultimately, a postdoc is still a fixed term gig with a clear end date (typically two to three years). Consequently, you do have to become somewhat deft at managing your time and try to make the most of things in a short time. If your hope is to transition to an industry career, then you also have to budget your time accordingly to develop the necessary skills for industry.
2. You get very little supervision as a postdoc. It is possible that there are a few supervisors/PI’s who closely manage postdocs, but in general, I think most PI’s tend to afford greater independence to their postdocs. I typically meet with my PI’s once a week to give them updates on my progress, but otherwise, I am mostly left to my own devices. During my PhD, my PhD advisor would steer me in the right direction, suggesting papers to read and ideas to try. Towards the end of my doctorate, I was initiating the ideas and write-ups of papers, but my advisor was there to guide me throughout the process.
Nowadays, I work in the same general area as my PI’s, but I have to find the open problems myself and see where improvements can be made to the existing literature. I have a lot more control over the direction of my own projects.
3. You have to get used to learning new things “on the fly.” When I first began PhD research, my instinct was to read as much of the literature as I could on my potential thesis topic. This seemed to serve me quite well, as I didn’t really know anything and I had to become used to reading academic papers in my discipline. Since the date of my PhD completion was pretty “open-ended” (anywhere from 2-4 years after passing my qualifying exams), I had a lot more time to read “everything” I could get my hands on.
However, nowadays, I have to learn new things fairly quickly (in part because of my postdoc time is limited before I have to go back on the job market). My PhD research focused mainly on theory for linear regression models and multiple testing. However, once I transitioned into my postdoc, I learned that many public health practitioners and economists prefer to use more flexible nonparametric and semiparametric methods (such as random forests or additive models), because they can more flexibly capture nonlinear relationships in longitudinal/panel data.
Seeing as I had not previously studied much about nonparametric or semiparametric methods, I had to catch up to speed. Fortunately, reading dense technical papers becomes easier with more practice and familiarity. But in order to ensure that I could become productive in a short amount of time, I had to be much more “strategic” and selective in what I was reading. Nowadays, I mainly read review papers to catch up on current work and relevant book chapters and class notes/lecture slides to learn the necessary background. I closely read a few papers that seem to be most related to my projects. I skim a lot more papers, but if they do not appear to be super relevant, then I don’t reread them (or I only reread the parts that are closely related to my work).
The same applies to learning new software tools. Another part of my research is to develop tools for data mining electronic health records (EHR). This required that I learn a little bit of SQL, something I had never used before. With the help of Google and a few good examples, I was able to learn what I needed. I am by no means an expert on SQL, but I learned what I had to in order to get work done.
4. Research gets done a lot more efficiently when done in collaboration with others, and it’s up to you to find collaborators. One thing that has helped me a great deal in transitioning to a new research area is to actively search for collaborators working on similar things. This includes PhD students, other postdocs, and professors.
I have one current collaboration with faculty in the Biostatistics, Epidemiology, and Informatics Department working on reducing bias in parameter estimates due to phenotype misclassification in electronic health records (EHR). I am currently working mainly on the computational aspect, i.e. using the R statistical software package to come up with a way to improve the runtime of a messy multi-dimensional numerical integration. Meanwhile, one of my other collaborators refines the model specification.
I have also recently come up with an idea for a scalable nonparametric model, so I reached out to one person who had experience with computational algorithms that could implement this model, and I reached out to another who had been working on similar kinds of statistical models. Together, the three of us have been able to put our heads together and expedite the research process.
I highly recommend that postdocs look for collaborators who can help you with things like implementation, refinement of your ideas, etc. Most likely, others will be able to point out potential shortcomings with your ideas and suggest areas for improvements, and you can learn a lot just from these discussions. You can also delegate different responsibilities for the project, so that you get work done faster.
All in all, I have to say that I have learned a great deal in my four months as a postdoc so far. Now that I am a lot more comfortable working on my new research topics and now that a few of my projects finally seem to be taking off, I am hoping to have some papers completed and submitted in early 2019. Next year, I am also hoping to be a co-author on a grant proposal, so I can learn about the process of securing “soft money” for my future research. If you are interested in my research, please keep a lookout for new manuscripts next year.