Editor’s Note: This blog post was originally published on March 15, 2021. A version is republished here with permission.
Simon Couch is a statistics student and developer of software packages for statistical modeling. He recently completed his BA in mathematics-statistics (minor in sociology) from Reed College and will be joining the Johns Hopkins Bloomberg School of Public Health in the fall of 2021 to pursue his PhD in biostatistics as an NSF Graduate Research Fellow. Couch co-authors and maintains R packages such as broom, infer, and stacks and is a former RStudio intern. His research interests are at the intersection of statistics, software, and sociology, and he regularly blogs about his work.
Since last fall, I’ve been going through the process of applying to graduate school in (bio)statistics. I found I was only able to learn about some parts of the process through office hours, personal meetings, and Twitter DMs (direct messaging), and I thought it would be worth publicly compiling lessons learned. I’m far from an expert about how this all works and can only speak to my personal experience.
A few things to note that influenced my personal experience: I’m a cis white man with US citizenship who is an alumnus (soon-to-be) of a private US liberal arts college majoring in math with a concentration in statistics. I ultimately decided to apply to PhD programs in biostatistics in the US during fall 2020.
When I started my undergrad, I didn’t know what a PhD was and had little—if any—sense for what graduate school looked like. However, by the time I was starting to think about writing my applications, I had learned a good bit more about what graduate school was. My test scores and GPA were quite unimpressive, but I’d been lucky enough to gain experience in statistical research and software development and had strong recommendation letters.
Many of these answered questions speak to PhD programs more so than MA/MS programs, and some apply more to biostatistics than statistics. I don’t have a good understanding of how many of these answers apply to schools outside the US, and many of these answers depend on my lived experience in some other way. I’ll try to specify when I understand that to be the case.
I’ve tried to be as forthcoming as possible while writing this, as I’m not sure it helps anyone to keep so much of this information behind closed doors. I apologize if I’m unnecessarily frank.
What Is Graduate School in (Bio)statistics?
This was the most difficult question for me to answer and the question that resulted in the most people looking at me like I had just grown a second head when I asked it.
A reality of graduate school: Many who attend(ed) graduate school grew up around a lot of people who attended graduate school. That does not mean you need to have grown up around a lot of people who attended graduate school to go (or so I’m told—we’ll see). In many important ways, though, it doesn’t look like your time as an undergraduate.
“Graduate school,” at least in (bio)statistics, generally refers to master’s (MA/MS) and doctoral (PhD) programs. Typically, you enter graduate school after earning a bachelor’s (BA/BS) degree, whether that’s directly after or following a few years of work experience. You can also apply to PhD programs after earning an MA/MS. I’ll speak more to this in a bit, but MA/MS programs typically take one to two years and present somewhat more like undergraduate programs. PhD programs take longer—four to seven years—and look (and pay) a bit more like a job.
Should I Go?
I recommend spending a good amount of time with this question, especially if you’re coming from an institutional setting where going to grad school feels like the “logical next step.” In some ways, it’s not.
The best first step to answering this question is learning a lot about what it means to attend grad school—for your finances, lifestyle, job prospects, and life timeline. I can speak somewhat to how these things could look if you do attend grad school, but how they might look if you don’t is more specific to you.
You will do a lot of the following during your time in grad school:
- Take classes in statistics (and possibly fields of specialization)
- Teach courses in statistics, but maybe also math
- Take part in research, including the following:
- Meet with lots of folks and talk science
- Write math, code, papers
- Attend conferences
- Attend and give talks
- All of the above at once
- Work quite a bit
You will not get rich.
The relative importance of those first three bullets can depend a lot on whether you’re doing an MA/MS or PhD. More on that in a sec.
If most of the above get you pretty stoked, maybe grad school is right for you.
How Much Does It Cost?
This actually wasn’t one of my first questions, but it ought to have been. I assumed since grad school is, you know, school, you probably pay for it like undergrad. Sometimes (maybe often?) not—read on. 🙂
Should I Do an MA/MS or PhD?
Assorted thoughts about how the two are different and alike include the following:
- MA/MS programs are shorter (usually 1–2 years).
- MA/MS programs tend to look a bit more like undergraduate programs in that 1) you usually pay to attend them and 2) the majority of the experience revolves around taking classes. You might also do some research or teach.
- The first year or two of a PhD is mostly focused on coursework. The latter part is generally based on you carrying out research and teaching undergraduate courses. It lasts something like 2–5 years. This research culminates in a dissertation, which is … a big paper, often composed of research papers you published during your time in the program, and some change.
- You get paid to do a PhD. Bonkers. Usually, salaries (“stipends”) range from $20,000–35,000 annually (depending, among other things, on the cost of living where you’re doing your PhD) and cover the cost of tuition. You are required to do some sort of research or teaching along the way. Don’t do a PhD if you will not be financially supported by your department.
- It seems like MS programs tend to offer a wider range of degree titles (e.g., data science or business analytics) tailored to specific career goals. MA programs tend to look more like the first two years of a PhD and are funded more often than MS programs.
- It seems like PhD programs allow you to dive deeper into specific concentrations in your latter years of the program.
- You can apply to PhD programs after graduating with an MA/MS! Some MA programs will offer graduating students admission into their PhD programs.
I’m not sure how well this applies to programs outside of the US.
What’s the Difference Between Statistics and Biostatistics?
Of all of the questions I try to speak to in this blog post, I feel like this answer might be the most unsatisfactory for people who really know “what’s up.” I think many in the statistical community could benefit from speaking and listening earnestly to how we delineate these fields. I’ll list a few of the main tendencies I’ve picked up on over the last year or two. In reality, these characteristics exist more so at the departmental level, rather than the “field” level, and you’ll see a lot of variation in how departments in either field position themselves relative to these traits.
- Biostatistics departments are usually situated in public health schools, while statistics departments tend be situated in schools of arts and sciences with some relation to the mathematics department.
- Many biostatistics departments seem to really value interdisciplinary research with collaborators from elsewhere in the school of public health. Statistics departments seem to be more self-sustaining in generating their research questions.
- Biostatistics seems to focus more on application, while statistics seems to focus more on theory. You will surely take part in both in either kind of program, though.
- Statistics departments seem to look to your math chops (however displayed) in admissions more than biostatistics departments. Biostatistics departments seem to appreciate some non-math backgrounds more than statistics departments might, like software development or fields in public health. You’ll need math chops for either, however.
There are a few “applied statistics” programs out there, as well. They tend to look somewhat more like biostatistics programs (omitting the first bullet point), yet draw from a wide pool of disciplines in their collaborative work.
If any of these distinctions make you feel as if you’re particularly excited about biostatistics or statistics, I’d encourage you to look for programs that exhibit that trait, rather than fall into the biostats/stats bin I mention above. These traits exist on spectra, and the biostats/stats feature here is only moderately predictive.
What’s the Deal with the GRE?
The GRE is like the ACT/SAT of graduate admissions. In comparison to those tests, though, it’s more expensive and even less correlated with success in the program it’s supposed to test your preparation for. Nevertheless, it’s still a part of admissions for many programs, so it’s worth speaking to.
The test will put you out $200 or so, but fee waivers are available. Also, some undergraduate institutions have internal scholarships to help pay for taking the test, so that’s worth a look.
There is all sorts of advice out there for how to do well on the test, and I didn’t do well, so I won’t speak to that. Following are a few stray notes about how the test is situated/regarded, though:
- There’s a “general” test and a “subject” test. The general test feels more like the SAT/ACT and is required by many more programs than the subject test. Generally, biostatistics programs don’t require the subject test. Some statistics programs do.
- The general test is broken up into math, reading, and writing sections. The math and reading sections are graded on a scale from 130–170, and you receive a separate score for each. Apparently, these programs don’t care too much about your reading score. More emphasis is placed on the math score, though. Programs will typically mention some sort of distributional measures about their admitted students’ test scores on their admissions websites. The most competitive programs, if they require the GRE, tend to admit only students with near-perfect math sub-scores—think 166–170. Typical for less-competitive programs seems to be 150s through mid-low 160s.
- I didn’t take the subject test, but my understanding is it’s very hard to achieve a score that will put you ahead in admissions unless you have significant coursework in computation-based, upper-level mathematics courses and/or are willing to put significant time into studying for the test.
- Some graduate schools use the GRE as a “filter.” At these schools, a score below some threshold means the committee may never put eyes on your application. I don’t have a good sense for how common this practice is.
Who’s to say whether grad programs will stick with this decision to omit the GRE as an application requirement once in-person standardized testing is available again? Props to those who do.
To How Many Schools Should I Apply?
There are a few things to think about here. The biggest limiting factor for me was price—it’s about $100 per application. Most schools provide fee waivers, which are a varying degree of 1) financially helpful and 2) a pain to apply for. Generally, you might need to be on a Pell grant to apply for a fee waiver, and the waiver will cover most—but not all—of the application fee. Also, ask professors/mentors/staff at your institution about possible pools of money you may be able to draw from to help cover these fees.
Another thing to keep in mind is how you think about your chances of getting into the schools you apply to. If you feel you have a strong application and the schools you’re applying to aren’t particularly competitive, you might decide to apply to fewer schools than you otherwise would, though I’d caution from leaning on this sort of thinking too heavily. From what I’ve seen, folks’s rate of admission to grad programs has been much less correlated with that schools’ ranking than I expected while applying.
Some sage advice I received that ultimately influenced me to decide to cut back on my number of applications: Don’t apply anywhere you don’t genuinely want to go to. If you’re not feeling excited about living in a city and working in a department for four to seven years (or one to two for a master’s), be earnest with yourself—save your money and don’t apply.
Bottom line: I applied to seven schools and was rejected by most of them. I’ve heard of some folks applying to four or five and some well into the teens. It seems like a typical number is seven to 10.
What Does a Grad School Application Look Like?
A few of the common elements of these applications, binned by how important they seem to be are the following:
Very important
- Solid letters of recommendation
- One or more of research, internship/work, or software development experience or some other “selling point”
Important
- Thoughtful personal/research statement
- Thoughtful diversity statement, if applicable
- Solid grades in key courses
- A lack of a negative internet presence
Good to have
- Solid grades in courses early on in undergrad
- Positive internet presence
Again, the relative importance of each of these will vary quite a bit depending on the program of interest, and I may be flat out wrong in some of my generalizations here.
A solid letter of recommendation is from a professor or research mentor who knows you well and can speak to your specific strengths, ideally at length. Preferably, they have a terminal degree in their field (e.g., a PhD in (bio)stat, math, etc.). Usually, programs will ask for three letters—if possible, at least one of those should be from a (bio)statistics professor or practitioner. If you’re applying to stat programs, one of these probably ought to be from a math professor, ideally your professor for real analysis, if you’ve taken it. If you’re applying to biostat programs, one of these probably should be from a research or internship mentor. You should keep your recommenders in the loop about how your application is coming together and where you’re applying (including the application deadlines for those programs).
There a few things I’m thinking about when I say “research, internship/work, or software development experience or some other ‘selling point’.” For one, having done one of these means you know what you’re getting into beyond coursework. If you’ve gotten a feel for any of these, you’ll have a better sense of what grad school could be like. Also, having done one of these likely means you had a supervisor or collaborator you worked closely with who can write you a strong and specific letter of recommendation. Last, having done one or more of these will help you articulate your “story.”
When I say “story,” I’m mostly thinking about the personal statement. Your personal statement gives you a chance to explain how it is you became interested in grad school and how your previous experiences show you will succeed there. There is a lot of advice out there about how to write a personal statement, and most of the prompts you’ll come across are very similar, so I won’t speak to this too much. Rohan Alexander recently wrote a thoughtful blog post that spends time on what a personal statement ought to look like. He also includes some good notes on how to think about the role of recommenders.
One thing, though—I think some are hesitant to be publicly forthcoming in their workflow on personal statements. The private advice I’ve been given about how much time and effort to put into a personal statement has often been much less than the amount I’ve seen recommended online. So, to be frank—personally, most of my personal statement was generic and sent to every program I applied to. For each program, I wrote a few sentences about why I was specifically interested in it and pushed surrounding sentences from the generic document around as needed so the program-specific statements flowed naturally. I also did find and replace for the program name and type (biostat vs. stat) in a couple places. Before submitting each to the official portal, I read the document in its entirety. I spent a weekend total on writing my personal statements and had my two roommates give a round of edits. This does not include the time spent learning the information about programs I ultimately drew from in writing my statements.
I mention “positive” and “negative” internet presence above. I’m generally thinking about what might come up if I look up your name with a search engine. Positive kinds of presence could be a personal website, LinkedIn, blog, professional Twitter, etc. Negative kinds of presence are the typical social media goofballery you’ve probably been warned to be wary of participating in—just give a thoughtful eye to your privacy settings.
Have more questions? Reach out to Couch and read the rest of his blog.
Leave a Reply