Misconceptions About Getting a PhD (in biostatistics)

Misconception #1: If you enjoy your field and want to study it at a deeper level, you should get a PhD.

It’s not exactly that this isn’t true, but I think it’s not an accurate representation of the purpose of your PhD program. I understand why it’s a common misconception for college students to have though, because it’s easy to assume that the PhD is an extension of the college experience.

For the first 1~2 years of the PhD, it is quite similar to college because your focus is on taking classes and passing exams.¹ But once you’re done with the required classes and start the research phase, you become a research assistant and the experience changes. The goal of scientific research isn’t about you studying or learning one particular subject – it’s about producing new ideas. This experience is a lot less structured and a lot more uncertain than studying and taking classes, where your progress is much more linear and predictable.

So I hear a lot of students say something like they want to get a PhD in order to learn the subject at a deeper level. The problem with that statement is that that’s not really the goal of a PhD program, even though you’ll likely learn many things along the way. The overarching goal isn’t to teach you knowledge, but to train you to become a researcher, and maybe that’s not something you would enjoy even if you like the subject and you like to learn.

That’s one reason I’m somewhat more circumspect nowadays about recommending a PhD to people than I was in years past (e.g. in this post). Because even though a PhD program has the illusion of being the next natural step in your education, it’s really quite a different thing than what you’ve previously experienced as a student.

Misconception #2: You should only get a PhD if you already know what you want to work on.

My understanding is that this is true for many other fields (and/or in other countries), but it’s not true for biostatistics. The admissions process is usually not advisor-specific and a lot of students will explore and consider different research areas during their first few years. The department at JHU also encourages students to explore different areas in their first year.

Most students will decide on a research area/advisor sometime in their second year, though it’s possible to switch advisors (or add co-advisors) even into the third year. The practicalities of doing so are more about how much time it takes to do enough research for a thesis (so if you switch too late in the program, you may not be able to graduate in 5 years), rather than about you being forced to stay in a research area you had committed to earlier on.

Misconception #3: A PhD degree is about studying an esoteric and narrow topic with no practical relevance to the public.

Again, maybe this is true for some fields, but I don’t think it’s the case for biostatistics. As an aside, I do believe in knowledge for knowledge’s sake,² so I’m not dunking on esoteric research at all and I don’t agree with the idea that studying “impractical” things are an inherently bad thing.

However, I feel like this is a very common criticism of higher education. If it’s something you want to avoid, I can say that it really doesn’t apply to the field of biostatistics.

First, I would say that for scientific research in general, we’re always thinking about why what we’re doing matters. In fact, it’s literally written into research papers: your goal is to clearly show why anyone would care about your results. In order to get published or to get funding, you need to convince other scientists that what you’re doing is useful. So while your research may seem very nuanced, niche, or incomprehensible to someone outside the field, the way science works is that all of these small pieces of research build up to a bigger picture.

Second, biostatistics is a special field because it’s grounded in theory, but the research is applied. Research in this field tends to be highly collaborative with folks from other departments like biology and medicine, and we get to work on all the fun (in my opinion) statistics stuff. My PhD advisor in particular has very broad interests in cancer research. With him and his collaborators, I worked on developing ML models to detect cancer from DNA fragments in blood tests, modeling the economic benefits of early detection, estimating the health risks of chronic exposure to radiation, etc. All of these have very obvious applications to the lives of the general public. Similarly, much of the research I see going on in the department also has intuitive implications for public health, medicine, or biology.

Misconception #4: Doing research is about making tiny contributions that are not meaningful.

I understand where this perspective is coming from and I think it’s both true and false. It’s true that most papers, and most PhD research, don’t have a major impact and just make a relatively “small” contribution to the field. But maybe that’s a good thing. Consider the following: if every paper resulted in a paradigm shift, the field would be very shaky and consequently hard to build knowledge from.

But it’s not true in the way I often see get expressed in the data science community. There seems to be this notion that all stats/ML research in academia is about making very tiny and inconsequential improvements in model performance. At least in my limited experience in biostatistics, this is not true at all. As I discussed earlier, you want to show in your research papers that there’s something the reader should care about, something meaningful, so ideally, you actually want to show that there are important improvements with your method.³ Practically speaking, if your fancy new method doesn’t offer a substantial benefit (better performance, faster runtime or just being easier to use), it’s pretty much guaranteed that no one is going to put in the effort to use it. In other words, even though academic research isn’t tied to clear company objectives like sales or customer growth, that doesn’t mean that academics waste their time on inconsequential things either.

Conclusion and further reading

As I discussed in my last post, my perspective is only one out of many. To conclude, below is a list of some posts on the PhD experience written by other people that I’ve collected over the years. I agree with much of what they have to say, though not necessarily everything, but I think it could still be helpful to read some different viewpoints.

• Why PhD experiences are so variable and what you can do about it, Arvind Narayanan (cs professor)
• Applying to Graduate School in Statistics and Biostatistics, Simon Couch (biostats PhD student)
• Advice for Graduate Students, Frank Vahid (cs professor)
• Modest Advice for New Graduate Students, Dorsa Amir (psychology postdoc)
• Advice For Graduate Students in Statistics, J. Michael Steele (stats professor)