Advice on applying for the NSF fellowship

Introduction

I received the NSF graduate research fellowship in 2008. This page is my advice on ways to make your application stronger. UPDATE: the requirements have changed (for example, there are only 2 essays now), but most of the advice below is still relevant. The first resource is the website: http://www.nsfgrfp.org/.

Every first and second year graduate student in the sciences should apply for the NSF. It's a prestigious and valuable award. The exact eligibility requirements are on the website.

The most important thing to keep in mind as you prepare your application is that the reviewers' criteria is "intellectual merit" and "broader impacts". See the description of these criteria here and the rating sheet the reviewers use here.

Proposed research essay

I spent the most time on the proposed research essay. In two pages, you have to convince the reviewers that they should give you $100,000. Note: you're under no obligation to pursue the research you propose!

My strategy: I broke my proposal into 3 smaller, loosely related subprojects. I started with an overview of the general research area. Then described the problems and how I planned to solve them. Each subproject had at least two or three goals--usually one simple short-term one and one more difficult part. For example, I said I was going to extend the "ssSSA" to include additional slow reaction channels as suggested in a reference, then I was going to develop the theory and algorithms for extending the ssSSA to include additional fast reaction channels and perform numerical simulations to validate the theory.

What? You have no idea what I'm talking about? Neither do the reviewers. But there are a few things to notice: first, I cited a reference to show I'm familiar with the literature (I think I had 5 references total); next, I proposed to do something suggested in the reference which suggests it's a reasonable idea; finally, I talked about extending the method, developing mathematical theory, and performing numerical experiments.

In all subprojects, I mentioned how the work could be extended to more general situations or applied to larger problems.

I also used this essay to explain that the project required expertise in several disciplines and that UCSB was the perfect institution to pursue the project because they are excellent at fostering collaborations. Further, I mentioned that my advisor is an expert in some of the fields and has collaborations with other experts that would be invaluable in this project.

Finally, in the conclusion, I said that I was going to put the algorithms I developed into open source software that will be easy to use by practicing scientists. Also, always mention that important findings will be published in scientific journals and presented at conferences.

Previous research essay

The previous research essay is a little more "touchy-feely". You want to add just a little bit of cheese. And you want to make it sound like you already know how to do research and have developed a lot of important skills.

I started with my undergraduate experience where I mentioned that I was in the honors program and did an individual research project. Remember, make it sound significant even if it isn't, like when I said that "I confirmed the professor's hypothesis about dispersion error"--it was actually pretty trivial but I think I made it sound OK. Then I described my experience at the Mayo Clinic where I worked for two years before grad school. I pointed out that it was very team-oriented (as opposed to my undergrad project). This led to UCSB where I also mentioned the strong reputation for interdisciplinary projects. I mentioned some introductory work I had done with my advisor and even a class project. To make the class project sound more serious, I mentioned that my project partner was continuing to study the same model from the project for his dissertation research. Finally, I mentioned work for a paper that we had recently submitted and gave a publication list. A 2nd year student should submit a paper right before you complete the application--why not? It shows you're making progress.

Personal statement

The personal statement essay is similar to the essay you wrote to get into grad school. This one should have the most cheese. You want to make it sound like you're excited about research--but be careful not to do the "child genious" routine. You don't want too much cheese. The biggest difference between your graduate school admission essay is that you probably want to specifically address the broader impacts criterion in this essay. For example, I described some volunteer work I had done where I spoke to local high school students to encourage them to major in computer science. I worked that in by saying that another reason UCSB is a good fit for me is that it outreach is encouraged. This is also the place to describe any teaching, tutoring or mentoring you've done. I closed by saying that the NSF fellowship would support this important research and outreach.

Reference letters

Like your graduate school letters of recommendation, these letters should be from people who can speak to your ability as a researcher. Ask them if they would be willing to write you a strong letter for this very important fellowship. If they agree, I think it's a good idea to send them some information that will help them write the letter. First, give them this link to the letter writer FAQ. Then, give them a concise description of your tasks and accomplishments to remind them what you did for them. I even suggested to one letter writer I knew really well to mention the phrase "broader impacts".

Transcripts and GRE scores

You have to submit transcripts for all colleges you've attended. GRE scores are not required, but I would submit them unless you think they're really going to hurt you. For example, I scored below the 50th percentile on the computer science subject test so I didn't submit.

Final thoughts

Remember, the reviewers are going to be knowledegable in your broad field, but probably not your particular research area. Also, they're going to be reviewing hundreds of applications--they won't be going over it with a fine-toothed comb. Before I started writing my essays, I made a spreadsheet of all the intellectual merit and broader impacts criteria with particular points that I was going to use to demonstrate those criteria. For each essay I made a spreadsheet of all the points that were mentioned in the essay instructions. Look objectively at your application. Highlight your strengths and manage your weaknesses.