So You Want a (Visualization) Ph.D.?
One of the best ways to get involved in visualization research is to get a Ph.D. In this post, I outline reasons for pursuing a Ph.D., what you should expect from the program, and how to apply. Throughout this post, I try to view these questions through the lens of visualization.
In a way, I’m a poor choice for writing a guide on getting a Ph.D., because I more or less stumbled into pursuing my own Ph.D. in the first place. All through my undergraduate years, I was convinced I would become a game developer, and it was only in the last semester when some freelance work soured that I started thinking seriously about whether I actually wanted to write code for a living. Then, out of the blue, one of my professors sent an email encouraging me to apply to my university’s Ph.D. program. I did, I interviewed, and I was admitted. I accepted without applying to any other universities or even thinking too hard about my future career or research. In other words, me writing this post is more a case of “do as I say, not as I do.”
With that out of the way, let me return to the original question of this post: So you want to get a Ph.D. in visualization? Well, if you’re reading this, congratulations! You have already done a lot more planning and research than I ever did. Here I want to give some practical guidance on how you go about finding a suitable Ph.D. program, how you put together your applications, and how you make your final choice. Please note that this is not a definite “getting into grad school” guide, but instead I focus on aspects of your application specific to visualization research. There are many general guides around; for example, Andrej Karpathy’s guide is a good one.
But let me first take a step back. You may be here because you have already made up your mind, in which case you can skip the next section. However, if you are still unsure, let me first explain why getting a Ph.D. in visualization is a good thing.
Why get a Ph.D.?
Since you’re reading this, I am going to assume you are already sold on visualization as a research area. Perhaps you’ve already built a couple of visualizations (or perhaps not). You may have played around with some visualization tools, such as Tableau, Power BI, or even D3. Maybe you’re even thinking about making visualization your career — -or perhaps it already is! But you’re probably also a little restless. Just like Neo in The Matrix, you’re restless because you know something. What you know you can’t explain but you feel it. Maybe you’ve come up against the limits of what your visualization tool can do, or perhaps you feel that you are solving the same problems over and over, and never get enough time to sit down and think about how to solve those problems once and for all.
Research in visualization is obviously the missing piece, and when you read Jessica Hullman’s “What is visualization research? What should it be?” (if you didn’t, then please read that first and then come back here), you probably found yourself nodding. This is what you want to do. You’re looking for the red pill, then one that will let you get involved in visualization research.
I’m an academic, so in this post, my answer is to get a Ph.D. in visualization. There are many reasons to do so: (1) the Doctor of Philosophy (or equivalent) is the highest degree awarded by universities in most countries, making you by definition an expert on your research topic; (2) you will get to work closely with several established experts who can teach you everything you need to know; and (3) you will get the chance to think hard and deep about a specific problem in order to advance human knowledge. In a way, the Ph.D. is in fact a “driver’s license” in that it certifies that you are now able to conduct independent research.
Oh, and you also get to call yourself “doctor” (although after 12 years, I still find it surprisingly hard to introduce myself so; that must be my Swedish “Law of Jante” talking).
At the same time, getting a Ph.D. is not the answer for all people who want to do visualization research. First of all, most Ph.D. programs require a significant time investment: in Europe, you can often get away with 3 years provided you already have a master’s degree, but in North America, the typical completion time is somewhere between 5 and 6 years. Second, there is a corresponding financial investment involved: most competent Ph.D. students have paid positions, but the salary is likely much lower than a company would pay. And third, while the Ph.D. degree can be seen as a driver’s license to conduct research, it is by no means the only way to learn how to drive the car. Many visualization practitioners don’t have Ph.D.s (although they often have some form of university degree), and still perform full-fledged research.
Perhaps the strongest argument I have to pursue a Ph.D. is that everyone recognizes the degree and you will have to spend less time defending your expertise.
Fair warning: if you want to be a professor at a research university, you will most likely need a Ph.D.
What do I need to get a Ph.D.?
First of all, a Ph.D. requires a fair chunk of uninterrupted time that you can afford to spend on thinking deeply about a specific (and often somewhat academic) problem. As stated above, you should expect to spend at least 3 years, and realistically closer to 5 or 6 years, pursuing this degree. As part of this commitment comes also an implicit financial burden: during these three, five, or even six years, you may be expected to survive on a graduate student’s stipend, which is likely significantly lower than what you would receive as a graphic designer, software engineering, or user researcher at a tech company.
However, it is worth emphasizing that most engineering, informatics, and computer science Ph.D. students do receive a salary during their Ph.D. studies, and their tuition is typically fully covered by the hiring department. We do recognize that taking on a graduate student position can be a significant pay cut compared to industry, so we try to do at least something to offset this cost. In short: even if you get admitted by a top private university, don’t worry about spending tens of thousands of dollars in tuition fees every year. It is typically covered (although you will want to check your offer letter to be sure).
What about part-time Ph.D. studies? To be perfectly honest, I didn’t even conceive of this as an option until I moved to UMD, where there is a fair share of working professionals who are looking to bolster their careers with a Ph.D. while continuing their day job. It is a tricky path to tread, as your time as a student is precious and often required to actually make a dent in the problems you face. Most part-time students I know ended up going full-time for at least part of their Ph.D. studies. However, I am sure it is possible to finish your degree while still holding down a full-time job. Just don’t expect to get much sleep, or to spend a longer time doing it.
The other aspect tied to time is your age. I’ve seen a lot of so-called “common wisdom” on the cutoff date for a Ph.D.; when it makes sense to start your Ph.D. (before 35 seems to be the most common number), or when it is “too late”. These rules of thumb are useless. There is no real cutoff date to begin a Ph.D. program: it all depends on your curiosity and your drive to fulfil your potential. While I am not a great example in this regard either (I went straight from masters to Ph.D. without passing “go”), I know a good many people who spent a couple of years in industry before deciding to come back to school for a Ph.D. In fact, I would argue that this practice adds some much-needed “seasoning” that will only serve as an advantage for your future career.
Here’s a case where my example can actually be useful: I’ve so far worked with two Ph.D. students who are older than me. While I’m sure they were the ones confronted with the biggest challenge — -getting used to a kid who was still wet behind the ears telling them what to do — -I never hesitated in taking them on. Heck, I saw their real-life skills as assets, skills that I could never hope to instil in them myself. In other words, as cheesy as it sounds, don’t let your age stand in the way of your dreams. It’s never too late.
The final question is about whether you need hardcore technical skills to get a visualization Ph.D. The answer is the same as I give for prospective students who want to get a Ph.D. in human-computer interaction: “it depends” or “not necessarily.” There are many flavors of visualization research: some of it is theoretical and requires someone with deep analytical skills, some of it is empirical and requires someone with knowledge of evaluation methodology, some of it is tied to visual communication and requires a keen design sense, and some of it is indeed technical and requires superior software engineering skills. This means that you may not even need to apply to a computer or information science department; design, journalism, architecture, or even art departments might be a good fit for a visualization Ph.D. This can make your life easier, as many graduate programs will require a suitable undergraduate degree to apply.
Also, note the keyword “student” in “Ph.D. student”: you are not supposed to be a fully formed researcher when you start. In other words, if you want to do hardcore technical work, you will be able to acquire those skills as part of your studies. This means that as long as you can find an advisor willing to take you on (more on that below), your past skills, whatever they are, are likely going to serve as opportunities rather than barriers.
How do I find a Ph.D. program in visualization?
While visualization is nominally part of any computer science or informatics department, it doesn’t make sense to blindly apply to universities if there are no faculty members able to advise you. Once again, let my example guide you in what not to do: when I accepted the Ph.D. program admission to the computer science department at Chalmers University, there was no single visualization faculty member on staff. This didn’t stop me, but it did make my life quite difficult since I had to figure out most things on my own. It doesn’t have to be this way.
Instead of such a trial and error approach, I would recommend that you work backwards. Unless you have a specific reason to choose a specific university (such as a geographic one; maybe you can’t relocate), don’t start from the university you want to go to, but start with the faculty member you want to work with. This is where all that idle web surfing experience can come in useful: you need to become an expert in finding faculty members that have research interests that match your own, and the only way to do so is to trawl their websites and read their papers.
Why, you ask? The answer is simple: faculty members tend to be busy teaching classes, writing grants, and advising their current students, and visualization faculty members are no different. We receive many, many emails from prospective students, and, as a rule, we are all looking for some kind of “hook” that will make us sit up straight in our chairs, remove our finger from the delete key, and continue reading. What this hook should be is underspecified, but at least you should have stalked their website enough to know their general research area and projects. In my experience, a surefire way to attract any researcher is to read their papers and provide some intelligent comment on the work (don’t forget the latter; just stating that you read a paper gives you no points). We’re all human, after all, and we like it when intelligent young (or old) people read our writing and think about it.
I will tell you one thing that is not flattering, and will get you nowhere: form letters that are the same for all professors and where you just blindly change the name of the recipient. Detecting form letters is a highly advanced skill that most faculty members tend to perfect in the first months of their appointments. It might seem like a great idea to send a hundred emails in one fell swoop, but trust me, quality far outweighs quantity in this regard. Do some research to identify the individuals that seem to be doing work most relevant to you, and then send out some carefully crafted emails to those individuals. Anything else is a waste of both their time and yours.
How do you find suitable advisors? One way is to look at recent programs for the IEEE VIS, IEEE EuroVis, and IEEE PacificVis conferences; because of their fast turnaround, papers in conferences tend to be representative of the current research of each faculty member. You may want to include the ACM CHI conference, as many visualization researchers (myself included) tend to publish in this conference as well. Studying the conference programs will give you an idea of current trends in the research community, and will highlight the people that are active in the topics that interest you. Of course, attending the conference in person would be even better, but that is often expensive and is not expected.
There are several finer points in picking out potential advisors. One of them is seniority. While it is often tempting to go for the biggest names in your research area, just remember that successful faculty members typically advise many students and have many additional commitments competing for their time. There is something to be said about working with a young and ambitious faculty member who is just starting out on their career and growing his or her research group from scratch. The senior faculty member may have a stellar reputation and a proven track record, but their time and efforts will be limited, whereas the junior faculty member will meet with you almost every day. Concretely speaking, it may be the difference between getting pointed in the right direction and then receiving feedback on your paper, to working shoulder to shoulder with your advisor and getting feedback on your source code, evaluation protocols, and data analysis.
Another point is the lab culture. The lab as a whole is often a reflection of the potential advisor’s style, and you want to make sure that style fits your own. Expand your web search to the website of the research group (if there is one), of other faculty members, and of the students in the group. Perhaps they have picnics, dinners, and other social activities? Perhaps you can even reach out to some of the current students and ask them about their experiences working with the potential advisor?
Finally, if possible, try to find a department with more than one faculty member active in research that is of interest to you. You never know if the chemistry will work out for your advisor, or if they will have funding. Maybe you will even realize that the research topic you picked does not work for you. Having some credible alternatives at the same school will make your life easier if you feel the need to switch direction for your Ph.D.
How do I apply?
Now, having identified some possible advisors (and don’t just pick one; you never know whether you will be admitted and whether they have funding to hire new students), you should reach out to them. In other words, don’t just apply, but send them an email with plenty of time to spare before the application deadline. Attach your CV, outline your background, and provide some of the above-mentioned commentary on their work and why you are interested in it (i.e., the “hook”). If you have a portfolio or website, link to it. Remember, no form letters!
You should also offer to meet with them on Skype or Hangouts, but make it a light request. Ideally, they will rise to the bait and schedule a 30-minute call with you, but they may also be overcommitted and have no time for a personal meeting. This is not a red flag in itself: the end of the semester, when most Ph.D. student applications are due, tend to be as brutal for faculty as for students, if not more so. Regardless, hopefully you will receive a positive response as well as potentially some inside tips on how to apply. At the very least, you have set up what you wanted to do: added some context (and maybe a face) to your name, so that your would-be advisor will be primed when they come across your application during admissions review.
And yes, deadlines. In the U.S., which I am most familiar with, Ph.D. student application deadlines tend to fall in December, often December 15. Be sure to plan ahead for this deadline and don’t wait until the last minute, as you may have to rely on a significant number of external factors to complete it. For example, for many U.S. universities, you may need to have taken the GREs, and most applications will often require two or three (or more!) letters from your references. These letters are important and need to be more than “X worked at Y for Z years.” Writing a good letter takes time. Be sure to contact your references well in advance of the deadline, provide them with an updated CV, outline your goal, and ensure they are enthusiastic about helping (a bland or negative letter is worse than no letter at all). For non-U.S. Ph.D. programs, the deadlines may be different, but the basic tenets are the same.
As for the application itself, this differs from school to school. Common among them is that there typically is a field where you can enter faculty members of interest; be sure to list your would-be advisor in this spot so that they will be notified of your application. You could also send them a quick email once you have applied to remind them. Furthermore, most applications include a “statement of purpose” where you write about your motivations for graduate school. Being specific about your visualization interests here, and potentially including images if the format allows it, is never a bad idea. Take help from any writing resources that are available to you, particularly if English is not your native tongue. Finally, if you have a portfolio or website (never a bad idea), be sure to include it in your application.
Students ask me what faculty members look at first when reviewing a Ph.D. student application. I’m sure it varies from faculty member to faculty member, but I know that one thing that makes me sit up and pay attention is when I see students who have already published one or more peer-reviewed papers, ideally in the research topic they want to work in. I know that this is a high bar — -after all, I already said we are not looking for fully-formed researchers, but prospective students — -but if you do find a way to have some research experience in your undergraduate studies, you could have a leg up. If you don’t, don’t worry: most Ph.D. students I have hired had no prior publications. Again, the strongest signal is whether the student has reached out to me prior to applying so that we’ve already had a chance to talk.
For my own Ph.D. studies, I went with the only game in town. As a rule, we Swedes are not a migratory people, and I was firmly rooted in the city in which I had done my undergraduate studies. The software engineering industry had little appeal to me, whereas the altruistic aspects of a life in the service of science (Science with a big S!) was plenty attractive. So, I took my professor’s advice, and he ended up becoming my Ph.D. advisor.
I marvel sometimes when I look back at this time, my own origin story. There were so many times when I could have picked another path and ended up in a vastly different situation. I think the lesson to be learned here is that the career of an academic is a meandering one, even for a visualization researcher such as myself, who supposedly base our decisions on cold and hard data. I hope that the data I provide in this post will help you make up your mind, and that you will join us on our meandering path of visualization research!