Making the most of the postdoc: 15 months in

As displayed on my main homepage, I am on the tenure-track market for the 2019-2020 academic year. I began my postdoc in August 2018, which means that I have been at my current position for one year and three months. I have written several blog posts in the past about being a postdoc, if you are interested:

I am pleased to state that my time as a postdoc has been quite productive (although very stressful at times). I have been able to successfully transition into new research areas that were different from my PhD dissertation research. Prior to beginning my postdoc, I had never worked with nonparametric/semiparametric models, statistical models for longitudinal and functional data, meta-analysis, or analysis of electronic health records (my current main research areas). So my time as a postdoc has also allowed me to mature greatly as a researcher and gain expertise in many new areas.

I also believe that my profile has become much more competitive for tenure-track faculty positions as a result of doing a postdoc. When I was applying to postdocs in the fall of 2017 and early spring of 2018, I had only one publication. The rest of my papers were either still under review (not even under invited revision) or still in preparation. 15 months after starting my postdoc, I have four publications, two papers under invited revision, and several manuscripts in preparation. So as you can see, my profile became much stronger in just over one year of postdoc’ing.

In order to accomplish this, it required being strategic about my time. Below are some tips I have about making the most of your limited time as a postdoc.

1. Try to wrap up the papers from your PhD as soon as possible.

Academic publishing can take a long time. After you account for rejections, reject-and-resubmits, and potentially several rounds of revision, the process from submission to publication can take well over a year. Fortunately, if you have finished several unpublished projects as part of your PhD research and have written them up for your dissertation, you will have already done most of the (preliminary) work for these papers. So these should hopefully not take too long to write up into their own manuscripts and submit them. Since you need quality publications on your CV by the time you go on the tenure-track job market, the papers that were submitted around the time of your PhD graduation offer the best insurance for getting at least a few in press in time for faculty job applications.

If you are like me, several of the papers from your PhD work will get flat-out rejected after a few months. I would suggest taking a week off to mourn/reflect/etc., have a chat with your PhD advisor about next steps, and then rework the manuscript as quickly as possible to address the criticisms in the reports by the editors and reviewers. See also my post “On rejection in academia” for more about handling academic rejection.

Likewise, if you are fortunate enough to get a request for a revision, try to get on it as fast as you can. In my field, journals typically allow 6-12 months for a major revision/resubmission and 3 months for a minor revision. I submitted major revisions within three months and minor revisions within one month. I still had to do work for my postdoc, but I spent at least a few hours each weekday to focus only on revisions. By September of this year, all the remaining chapters from my PhD dissertation had been accepted.

2. Collaborate with others in order to get work done faster.

I talked a lot about this in my post “Postdoc: Four months in”, but it is definitely worth reiterating. You will be able to get work done much more quickly if you seek out collaborations and leverage the expertise of other people.

One of my postdoc papers, currently under invited revision for Journal of the American Statistical Association, could not have been completed in such a short period of time had I not collaborated with my two co-first authors, who had previous experience working on efficient optimization algorithms and methods to detect pairwise (nonlinear) interactions in high-dimensional models. I was able to learn so much from them. By combining all of our skills together and collaborating on writing the actual manuscript, we were able to complete this paper and submit it to a top journal in our field in just a matter of four months.

In addition, it is a good idea to seek out collaborations with people at other schools, not just your current workplace. This not only helps to extend your network (see point #4 below for more), but it can help make your job application stand out. The ability to collaborate with other scientists (as witnessed by having a few second/third-author and middle-author collaborative publications) is prized highly in many science fields. First-author papers demonstrate your independence and ability to spearhead a project, so they are certainly the most important thing on your CV. But being seen as a “team player” through other collaborations is also a major plus.

3. Choose your postdoc projects carefully.

As mentioned in my post “Postdoc: Four months in”, you have a lot more freedom as a postdoc to direct your own research projects. This can be a bit overwhelming, but necessary to prepare you for the research independence needed to succeed as a faculty at a major research university. As I discussed in my post “On rejection in academia”, a major criterion for publication in top journals is novelty and the potential to have a wide impact on the field and springboard further developments.

So the choice of topic matters quite a bit. For my postdoc, I actively sought out: (a) “hot areas” for current research, and (b) areas where the literature was not nearly as developed.

Regarding point (a), there is currently a great deal of interest in “big data” and algorithms that can handle big data in a computationally efficient, scalable manner. Left unmodified, traditional tools for Bayesian computation such as MCMC can be ill-suited for such high-dimensional data. So I looked to see if I could develop faster alternatives to MCMC, or alternatively, modify existing MCMC algorithms so that they can run faster on such data.

Regarding point (b), I looked to see if there were any statistical problems that were not adequately addressed by Bayesian statisticians, either methodologically or theoretically. There are some statistical models that have been studied quite extensively, such as normal linear regression and logistic regression. As a result, it can be rather difficult to publish on these topics in top journals. It is not impossible to publish on such topics in the top journals, but it has to be really, really good in order to pass top journals’ criteria for “novelty.”

For me, I tend to go with the path of least resistance, which is to work on areas where there has not been as much work done. Consequently, I made a foray into varying coefficient models for my second post-doc paper, where the work within the Bayesian framework is much more sparse than it is for frequentist counterparts. This paper is also under invited revision for a top journal in my field.

4. Network with the top people in your field.

If you did not get your PhD from a “brand” name school (like an Ivy League school, Stanford, MIT, etc.), I would not worry about it too much. There isn’t anything you can do to change that. At least in my field — and in many STEM fields, it does not seem as though faculty search committees care that much where you got your PhD, provided that it is from a place at least somewhat reputable (i.e. not a completely unknown, unranked school) and provided that your CV demonstrates consistent research productivity and publication in quality journals/conferences.

That said, the chief advantage of being a PhD student and/or doing a postdoc at a “top” school is the opportunity to work closely with the most renowned, famous professors. Top places have a lot more renowned scholars. Proximity to them helps a lot, and it was part of the reason why I chose to do my postdoc at UPenn. If you have a famous professor who is intimately familiar with your work and who can vouch for you in your academic job applications, it can elevate your reputation and your job applications quite a bit. If you are not currently at a “top” school, it shouldn’t be a big deal, but I would highly recommend reaching out to the top people in your field and making sure that they become familiar with your work. Attending conferences is also an excellent way to network with the top people in your field and get your name on their radar if you are not able to meet or talk with them regularly.

That is all I have for now. As I am presently in the midst of a job hunt, I do not have much free time at all. However, once my job hunt is over, I will be making a few more blog posts about my experience being on the tenure-track job market.