How NOT to get into Graduate school
A list of common errors students make on graduate school applications (in Data Science, but probably elsewhere as well…)
In case you didn’t know, December is grad school application season. Many application deadlines are in January so hopefully you have already started your applications if you plan on applying. Thus, error #1:
1. Thinking all applications are the same and throwing your application together last minute
This is not the most egregious error, but I see lots and lots obvious cut-and-paste errors. Sometimes, I even see another university name in an essay.
There are two issues here, A. Detail orientation and B. Interest level.
There is a (probably apocryphal) story about a band who listed in their concert contract when they were touring that they wanted a jar full of M&Ms with all the blue ones taken out (or something like this) in their dressing room. It was buried way down in the middle of the contract. They didn’t care about M&Ms, what they cared about was locations that were detail oriented enough to read everything and pay attention to everything. If they walked in and saw no blue M&Ms, they knew that they had a solid location who was paying attention. If not, they needed to do their own checking to be sure everything was going to be fine.
Start early. Pay attention to details.
The same goes for applications. Lots of little errors are a signal the either A. you don’t care that much about coming to our school or B. you are not all that detail-oriented. Either is a problem. Reduce careless errors in your packets. That often means getting everything done, then stepping away for a day or two, then coming back. That way you are much more likely to catch mistakes.
Second, we want to know you are actually interested in our school. Demonstrating to us that you see applications as a numbers game and you just need to submit as many as possible to get a few positive responses is not a good strategy. Instead, showing that you have read some faculty bios and are already trying to see how you might fit in at our school. Show that you’ve looked at student profiles or even reached out to current students via social media to learn about the school. For example, if you are female, you may notice that Vanderbilt’s Data Science Masters program is 30–40% female and you are excited about this (yes, that was a bit of a plug…) This shows that you care and are really interested in our school specifically.
2. Submitting bad recommendation letters
I don’t mean bad as in someone says bad things about you. That never happens. What I mean by bad recommendation letters are letters that have virtually no content. I have seen many letters that are a version of “Student X was in my class and got an A.” and that’s about it.
This is a bad letter because it tells us that this recommendation is the best you could do. It shows a general lack of engagement with faculty (or your supervisor at work or coworkers) such that you can’t come up with even three people who can speak to your work.
Who should you ask? You absolutely need to ask people who know you and your work as best as possible. The biggest conundrum here is someone who is not in a quantitative field, say journalism. What do you do if you are a journalism major, have taken a few online courses in programming (MOOCs or other automated learning with no human interaction) and want to get into Data Science?
You are best served by asking people who know you, your skills, and/or your work ethic regardless of what field they are in.
Do not ask your math professor from your freshman year. They won’t remember you and it won’t come off well. You are best off asking people close to you, who know you and your work ethic and can vouch for your other skills. Maybe they can talk about how you are a creative problem solver or a leader or a consensus builder, etc. There are lots of skills you need to succeed, don’t think you have to reach and get a recommendation from someone who knows nothing about you, but is in a field you think is important.
So, to sum up, you are best off asking people who really know you and can speak to your work, your skills, and your commitment. Find other ways to address any weaknesses in your profile. Which is error #3…
3. Ignoring your weaknesses
If you have a weakness in your application, address it directly. The most common one we see is someone who has weak quantitative skills. If you are in the humanities and took just one math class your freshman year and you are applying for a quantitative degree, you need to address this somehow, likely in your personal statement or essay.
Here is some perspective: our goal is not to play “gatekeeper” and eliminate anyone who doesn’t meet our “standards.” We simply want everyone who comes to our program to succeed. Our biggest worry is that someone without the necessary prerequisites will come, struggle a lot, consume a lot of energy (both theirs and ours) and ultimately drop out. That serves nobody.
If you have a clear weakness and do not discuss it, then we typically interpret that as you not being aware of your own weaknesses. That is not a way to get an acceptance letter.
So, if you are addressing the fact that you perhaps just barely meet our quant requirements, you need to explain to us why you think you will succeed in the program. It could be other courses you’ve taken (say, Gilbert Strang’s Linear Algebra course at MIT OpenCourseWare — excellent) or how you loved math as a kid, but just got away from it a really want to get back into it.
We (at Vanderbilt) love folks from diverse backgrounds in our Data Science program (humanities, social sciences, etc.). We are not looking for a class full of computer science and math majors. However, we want you to succeed in our program also. Show us why you think you can do it, even though it may look doubtful on paper.
If you have a clear weakness and do not discuss it, then we typically interpret that as you not being aware of your own weaknesses. That is not a way to get an acceptance letter.
4. Writing an essay that is a narrative version of your application.
Too many students go to grad school because they like school and don’t know what else to do. This is a terrible reason to go to grad school. The way this usually shows up is a personal statement that simply recaps the applicant’s accomplishments.
Too many students go to grad school because they like school and don’t know what else to do. This is a terrible reason to go to grad school.
Why do you want to get into Data Science? Many people tell stories of their love of numbers, or a parent or guardian who stimulated their imagination. These are all great. Better is linking it to your work — many students talk about a first job or internship where they got a taste and they want more. This makes for an excellent essay.
Also, be forward looking. Where do you want to go? You don’t need a detailed plan because nobody knows what the world might look like. Lots of things change and it is not likely your path will look like that, but we want to see that you are want to go somewhere and how a degree in Data Science fits into that plan.
5. Ignoring soft skills
In Data Science, we get a lot of applications from quantitative students. These students often have some combination of math, computer science, statistics or engineering backgrounds. This kind of background is awesome for data science! I have an engineering background myself.
However, having rock star quantitative skills is not sufficient to be successful. You need to be able to communicate what you are doing to others. You need to be able to persuade others that your approach is correct. Simply developing an amazing algorithm is not often enough — you have to successfully tell others why it matters and why they should use it. History is littered with amazing ideas that lost out to second-best ideas that were better communicated.
Having rock star quantitative skills is not sufficient to be successful. You need to be able to communicate what you are doing to others.
If you have solid quantitative skills in your applications (courses, major, test scores, etc.) then you should focus mostly on developing your soft skills. When have you worked on a team? When have you shown leadership skills? How do you handle conflict? How do you handle it when someone else does not do their part?
Data science is a team sport, so tell us how you’ve done that. If the idea of working on a team sounds terrible to you, then you are applying for the wrong degree. Go get your PhD where you can work in a research lab with coauthors who will see your amazing tech abilities and help you communicate them to other like-minded people.
If the idea of working on a team sounds terrible to you, then you are applying for the wrong degree.
6. Having no demonstrated quantitative skills
When I say ‘demonstrated’ I mean rigorously assessed by a human instructor in a course that gives grades, not certificates of completion. The best way to demonstrate quantitative skills is to have taken a course where you received a grade from a human being, either in person or online. Standardized tests (GRE, GMAT) also work well here, though I know they are mostly algorithmic. Is this you? Don’t stop here and despair — read to the bottom.
What I call ‘automated learning’ is second best — automated learning is self-paced online courses where all the grading is algorithmic or peer-based. You get a certificate of completion at the end along with the other thousands of people doing the same thing.
If you are applying for next year, it may be too late to enroll in a class like this. Instead, study up and do your best on the GRE/GMAT. Go ahead and take an automated learning course and then be sure you address your weakness here in your application (as I noted above). Talk about how you loved the class and what most stood out to you.
If you are no longer in school, it is a great idea to actually enroll in some kind of online and/or part-time program in the evenings so you can provide a transcript. I did exactly this before my PhD — I enrolled in a part time Masters of Finance program at the local university. I never finished the degree, but I took a few classes and got grades. This was important because I was going back to school after nine years of working and I wanted to show (to the admission committee and myself) that I could do it again.
Having no demonstrated quant skills is not automatically disqualifying. However, you need to show that you know it is a weakness and are already working on it somehow.
I put this one last because it is not automatically disqualifying. To repeat myself a bit if this is you: do your best and explain why you think you can still succeed even though this is a weakness. And do your best to begin addressing it by taking some kind of part-time class. An automated learning class is better than nothing if that is the only thing you can do right now. That introduction will help you in any program or class you take alter.
Disclaimer
Actually, none of these things will automatically disqualify you from consideration. Neither will avoiding all of these mistakes guarantee you admission. I am simply trying to help you present the best version of yourself so that you get your best possible chance at getting that acceptance letter.
Also, these tips are (somewhat) tailored to Vanderbilt, though I think they likely apply broadly. Some other programs perhaps more strongly emphasize top quantitative skills than we do, and correspondingly worry less about soft skills. Again, check into specific programs to see which one is best for you.
Finally, should you even go to grad school? Check out these two articles to get my take. Applying is a lot of work — don’t waste your time.
