Some thoughts on doing a postdoc

Since I have a new personal webpage, I have decided to start occasionally blogging again. I had several blog entries that I had written on my old University of Florida personal website that I have also moved onto my new site (you may read them here).

In a previous post, I discussed the process of applying to postdocs and some perspectives on pursuing academia vs. industry. I am still in the process of learning a lot of things and navigating my new postdoctoral appointment, so I hope to have more updates in the future about the postdoc experience. But I hope that in this post, I will shed some light on what doing a postdoc entails and my thoughts about doing a postdoc.

Why do a postdoc?

The academic job market is very competitive in most science fields. In the fields of statistics and biostatistics, in particular, it seems to be quite difficult to land a tenure-track job right away at an R1 in the United States, unless one has at least one first-author paper (in press or in revision for) one of the top journals (in statistics, those would be: Annals of Statistics, Biometrika, Journal of the Royal Statistical Society: Series B, and Journal of the American Statistical Association. Biostatistics includes the aforementioned, but papers in venues such as Biometrics, Bioinformatics, and Nature Methods are also highly prized). Publications in top journals may be less critical for liberal arts colleges that place more emphasis on teaching (although research is still an important component at these schools and postdoc training may still be preferred for a lot of teaching schools). But in general, due to the competitiveness, postdoctoral training is becoming almost a requirement for an Assistant Professorship in a lot of fields.

But even if one has a publication or two in top journals, it may still be worthwhile to do a postdoc, because it gives you an opportunity to learn a new area and develop greater independence as a researcher. Most PhD students write their thesis in an area that is of particular interest to their advisor. A postdoc gives you a way to branch out from that and develop a reputation that is independent of your advisor.

Moreover, if you are interested in an academic career, then a postdoc without any teaching requirements gives you an opportunity to focus solely on your research. As those of us who have taught college courses before know, teaching is a huge time commitment, and a postdoc can give you more time to mature as a researcher before having to take on teaching responsibilities.

A postdoc can also serve as a great way to increase your (first-author) publication count, lay the groundwork for your future research agenda, and  make you even more competitive in the job market (for either academia or industry). For example, a hiring committee may like to see that you have other collaborations on your CV besides only publications that were written with your PhD advisor. Assuming that the postdoc goes well, you could have more and better employment options by the end of it.

How is a postdoc different from being a graduate student?

Well, for one,  you get paid much more. In many ways, it is similar to being a grad student. I typically meet with  my mentors (the Principal Investigators whom I work under) once a week, just as I did with my PhD thesis advisor. The rest of the time, I get to spend it however I want (though it’s certainly good to be productive during that time!).

I suppose one difference is that there is slightly more independence as a postdoc. In the early stages of research as a PhD student, one typically gets some guidance from their PhD advisor, and a great deal of time is spent working with your PhD advisor to try to narrow the scope of the PhD project and find a good, “manageable” topic. As a postdoc, it is generally expected that you are trying to become a more independent researcher. Accordingly, most PIs do not micromanage their postdocs, and allow you much more freedom to find the topics that you will pursue.

Additionally, postdocs can be involved in things like grant writing, giving public seminars, etc. that they may not have been able to do as PhD students, but which are part of the responsibilities of full-time faculty.

Should I do a postdoc in the same area as my PhD research?

In some fields like pure mathematics, it may be advisable to stay in the same general area as your PhD research (same general sub-discipline, but working on different problems within it). But in general, I would encourage graduating PhD students to actively step out of their comfort zone and to do a postdoc in something different from their PhD research. First, this gives you the chance to generate new research ideas totally independently from your PhD advisor and collaborate with other scientists. And second, it furthers your training in scientific and technical skills that you may not have previously developed.

For my postdoc, I am working on scalable algorithms and distributed algorithms. I am also working on new methodology to handle complex data with known dependencies, interactions, and correlations (e.g. longitudinal and spatial data), where computing the full likelihood function of the data is typically infeasible. I have no prior experience with distributed computing, having written my PhD thesis on statistical theory. Moreover, my PhD thesis dealt only with models where the data (Xi, yi), i = 1, …, n, were assumed to be independent with identically distributed error or noise. So I am learning a lot of new things for the first time as part of my postdoc that I never learned in my doctoral research.

What does it take to get a good postdoc?

There is much more detail in my post about applying to postdocs. But the most important things for getting a good postdoc (and for getting a tenure-track faculty job after that) are: 1) strong letters of recommendation, and 2) a solid or promising publication record. The prestige of the PhD granting department is typically of secondary importance (in my field anyway). If the PhD advisor is very famous and writes a very strong letter, then sometimes it is not necessary to have publications accepted or in press to land a good postdoc (but at the very least, one should have a paper or two submitted or under review).

By the time I was finishing up my PhD, I had one publication in the Journal of Multivariate Analysis, which is considered a top 6 journal in mathematical/theoretical statistics (Source). I also had two other papers that were/are under review. I think I did pretty well in the postdoc job market (and was even shortlisted for one tenure-track faculty job), so a good rule of thumb for getting a good postdoc (in statistics/biostatistics) is probably this: one publication accepted/published in a respectable outlet, and one or two more manuscripts submitted or about to be submitted.

Of course, the most competitive candidates will have several papers that have been accepted or published in top journals by the time they have graduated, and these PhD graduates will have the best chance of nabbing very prestigious postdocs at places like Stanford, Harvard, or UC Berkeley. But I did pretty okay with my record.

With an eye towards eventually getting a tenure-track faculty job, I am hoping to use my postdoc years to learn a new field and new skills (such as grant writing), to hone my current skills (especially in programming), and further expand my first-author publication record. I will update this blog periodically on my progress.



Leave Comment

Your email address will not be published. Required fields are marked *