I submitted the final draft of my PhD dissertation and had the oral defense for my PhD thesis on Thursday, May 3, 2018. In August of 2018, I joined the University of Pennsylvania as a postdoctoral associate. In this blog post, I will discuss something that many senior PhD students must wrestle with: whether to choose academia or industry. My previous blog entries focused on several aspects of grad school life:
- How I finished a PhD in under four years
- Graduate school survival tips (including publishing, choosing an advisor and a good thesis topic, etc.)
In this blog entry, I will share some of my thoughts about pursuing an academic career as a graduating PhD student.
The Big Decision: Industry vs. Academia
First, all PhD students should know that that they absolutely do have agency when it comes to their career choices and that they are not obligated to do what their advisor or department “expects” of them. Most PhD advisors are aware that the job market for tenure track positions is very tight and that ultimately, many of their students will need to find employment outside of the academe. So if you are a PhD student, do not feel as though you would be “disappointing” your advisor if you do not want to pursue an academic career. It is your life, and a good supervisor will be supportive no matter what.
There are many pros with industry. You certainly cannot beat the salary, the location flexibility, or the work-life balance for industry. A few distinguished professors do make more than their industry counterparts, but in general, industry will pay more. In addition, there is a lot of interesting work for PhDs in industry, especially if you work as a research scientist. If you are in a STEM field, you will often need to teach yourself new software tools, technologies, and current methodology and algorithms, so the industrial work environment can certainly be intellectually stimulating. For the most part, as long as you put in eight hours a day of honest work, you do not need to work on evenings or weekends. Finally, because there are more concrete deadlines, “milestones,” and deliverables, industry work tends to have a more immediate and direct impact to the public than academic research.
In academia, the salary tends to be lower than in industry. But the salary is certainly still more than adequate to support a comfortable lifestyle and a family. The main benefits of academia are twofold: 1) the freedom and independence, and 2) the ability to think deeply about problems that interest you. As a tenure-track or tenured professor, you can basically do anything you want with your time as long as you fulfill a few hours of teaching per week and attend required departmental and committee meetings. The rest of your time is yours to spend however you want, and during summers, you do not have any teaching responsibilities or mandatory meetings. Granted, you do need to meet certain requirements in order to receive tenure, but how you spend your time working to reach those milestones is mostly up to you.
Apart from the freedom in academia, you can really explore your chosen field in depth and branch off into totally new areas if you’d like. You don’t have a boss (in the traditional sense of the word) to tell you what you “need” to work on. As a PhD student, I found that I was interested in the theoretical foundations of many statistical methods that have gained popularity in recent years (e.g. classifiers, ensemble methods, etc.). I wanted to explore “why” some methods from statistics and machine learning work so well, what their large-sample properties are, and when/why they may be deficient. There are a few industry jobs that are focused on basic science (like Microsoft Research or IBM Research), but for the most part, there is a greater emphasis on producing/implementing software that meets certain functional specifications in industry, rather than deeply understanding the underlying science behind it.
In addition to the above, I found that I do enjoy teaching and writing papers. I felt immensely fulfilled when students experienced the “A-ha! I finally get it!” moment after I explained the material in-depth to them, and I loved to watch my ideas take form on paper when I wrote my manuscripts. If you absolutely loathe teaching or writing papers, then academia is probably not the right career path for you. But I have found it to be an enjoyable process, even if it is sometimes tedious.
Applying to Academic Jobs
Because I enjoy the freedom and consider myself to be 100% self-motivated, I decided to pursue academia. I applied to primarily postdocs, but I also applied to a small number of tenure-track positions. The main materials I needed to submit were: a curriculum vitae, a cover letter, a research statement (i.e. a document which described my past research and my future research agenda), a teaching statement (for the positions that had a teaching requirement), a representative manuscript of my work, and recommendation letters from my PhD advisor and two members of my PhD committee. You may take a look at my research and teaching statements below.
- My research statement for academic job applications
- My teaching statement for academic job applications
In my field, it is currently quite difficult to get a tenure-track faculty position without doing a postdoc, unless you have several publications in top journals as a PhD student and have demonstrable evidence (from recommendation letters, research statement, and your CV) that you can parlay your thesis research into a productive research career for the pre-tenure years. Only a tiny minority of PhD students in Statistics/Biostatistics will have several publications in top journals and be able to milk their thesis topic for years by the time that they graduate, so most PhD students who are interested in academic careers will need to do postdocs. While I had learned a lot from conducting PhD research, I felt that I needed to develop a wider skill set and gain greater independence from my PhD advisor. And a postdoc would give me the opportunity to do this.
Some PhD students who are strongly considering academia are hesitant about doing a postdoc, but I would say that if you really want to pursue an academic job, then there is no harm in spending a few more years building up your CV in a postdoc position. As an analogy, people who are passionate about becoming medical doctors must do at least four years of residency after med school before they can obtain an unrestricted license to practice medicine — and many MDs choose to do a fellowship or two in a certain specialty after they have completed the residency requirement. Granted, medical doctors are more-or-less guaranteed employment after they have completed close to a decade (or more) of training, which is not a guarantee in academia. But I mainly wish to convey that if you are strongly passionate about a career path (e.g. academia), there is often no gain in reaching your goal without some sacrifice.
Choosing the Right Postdoc
I received three postdoc offers: one that was very theoretical work, one that was a mix of applied work and theoretical work, and one that was essentially all applied. None of these postdocs were in areas related to my PhD thesis research, but I viewed this as a great opportunity to learn something totally different from what I had been working on for almost two years. My PhD thesis work was primarily theoretical. I introduced new prior distributions for high-dimensional inference in the Bayesian framework and studied various theoretical properties of my priors (including minimaxity, posterior convergence rates, oracle properties, and consistency). While I developed solid training as a theoretician in my PhD studies, I wanted to learn more from a computational/applied perspective as well.
In the end, I chose the postdoc that was a mixture of applied and theoretical work. On the applied side, the position is aimed at developing novel statistical methods and algorithms for analyzing and mining electronic health records (EHR). The position has a bit of a computer science component to it, in that I am learning new tools from CS such as distributed algorithms and other scalable algorithms. In addition to the computational and applied side, the group I joined at Penn also publishes papers on statistical theory for the methods that are developed. The main motivation for this work is still principally its application to EHR, but I also have freedom to explore the theoretical foundations of the methods that I will be working on.
This position will last for two to three years. In my last year, I will apply to tenure-track academic jobs.I am very much looking forward to continuing my training as a scientist and scholar and working towards my goal of becoming an Assistant Professor.