Some things I've learned up to now about the NSF GRFP


Update 10/16/14

Dr. Muller-Parker refuted my previous claims in this post recently. So I've modified the previous posts to reflect what the NSF GRFP has in terms of on-record statistics. I previously said that there is possibly a higher rate of success if you were to exclude those who were rejected without review. Dr. Muller-Parker has corrected the statistic of 2/3rd of submissions to 10% of submissions. Within the 10% who are rejected without review, 75% are due to lack of minimum letters of recommendation. 

I'm finishing up my NSF GRFP grant this week. For me, it's my first external grant I've had to write and also the first grant I've had to write all of (note: incorporated edits from my advisors and good friends). The process itself is seemingly different for everyone I've talked to for help, so here are some points I learned along the way.

3/4ths of NSF GRFP submissions that are rejected without review are because letters of recommendation were turned in LATE.

This stat comes from an NSF GRFP webinar hosted by GRFP Program Director Giselle Muller-Parker and CGS/NSF Dean in Residence Henning Schroeder. That's a pretty hefty amount of applicants being denied.

It was hard to find solid statistics on recent NSF GRFP applicants (edit: this is a good reference for NSF GRFP stats, but still not comprehensive), but this document from the NSF site states an average of 15% of applicants are awarded. The acceptance of 10% - 15% (Dr. M-P estimates closer to 15%) has been consistent for the last handful of years.

Year Number of Applications Awarded GRFs Success rate (%)
1952 2685 569 21.19180633
1962 4977 987 19.83122363
1972 5005 550 10.98901099
1982 2672 500 18.71257485
1992 7723 740 9.581768743
1998 4851 766 15.79055865
1999 4796 900 18.76563803
Average 4672.714286 716 16.40894017

This table is from a publication on GRFP, with a few modifications. The averages at the bottom only account for what is listed in the table -- thus, the 15% average as stated by this publication takes into account more years than these data present. Assuming the field of applicants left are people who submitted everything correctly, it seems that the GRFP isn't impossible to receive if applied for. 2,077 GRFs were awarded last year, suggesting about 13,000 applications were filed. That number may seem pretty different from the ones listed above, but this nifty NSF GRFP presentation speaks to similar numbers and success rates.

The biggest recommendation I would have, based off all of this information, is HAVE EVERYTHING ACCOUNTED FOR. Your two documents (study proposal and personal statement) are the most important ones, but don't overlook other aspects. Transcripts -- upload unofficial ones! Reference letters -- stay on top of them! Email 20, 10, 5 and 1 day(s) before it's due. Then, contact anyone who hasn't contacted you back EVERY DAY until it's submitted.

Don't be eliminated from this because you or your letter writers got lazy.

There is seemingly a lack of GRFP reviewer enthusiasm?

This is much more hearsay than anything, but the murmurs from the NSF GRFP reviewers I know have said that this year has moved all online -- training, panel discussion, etc. I'm not so sure as to how long they've moved some of these things to all online, but the handful of people I know think it is sort of a rip-off since they were able to get a trip out of it for the in-person panel meeting. Now that there is no in-person meeting, the $280/day or whatever it is that you get for reviewing isn't as appealing -- seemingly. I'd take this point with a grain of salt since it really shouldn't have an effect the review process. I'm not exactly too sure what to think of people who willingly choose to be NSF GRFP panelists. I'd assume they are younger reviewers (most of the people I've come into contact with are younger profs), but that's just a hunch based off of my very limited sample.

Start early. Six weeks minimum.

I was informed about it on September 5th. After a discussion with my advisor, she gave me 3 weeks before my first draft was due (which ended up being turned in on October 3rd). Since then, the document has gone through tons of revisions -- with me and Dr. Jordan (my advisor) going through the documents practically every other night since the middle of October.

In contrast, I attended an NSF GRFP meeting on my campus a few days after I submitted my first draft to Dr. Jordan. A few other graduate students discussed how they have barely begun their drafts. Looking back on that moment now -- that would be utterly insane. We just got into the final revisions (trimming down to two pages, mostly) in the last couple days (all documents are due 4 days from this post). If a week was subtracted from this editing process, I really don't know what kind of proposal I'd be submitting. Surely, it couldn't be as crisp as it is now.

Minimum: two weeks to draft, four weeks to edit. With the last week of editing ending on your deadline.

Ideally: 8 to 10 weeks for everything. Four to draft, at least four for major revisions. Maybe an extra week for minor tweaks from other people.

The current NSF GRFP is different from previous versions -- past personal statements and research experience docs are DIFFERENT.

I had a lot of help in terms of previous versions of successfully awarded NSF GRFP applications (specifically from @katiesci and @rmpollet), which helped guide the structuring for my documents. However, the previous format was:

  • Study proposal (2 pages)
  • Personal statement (2 pages)
  • Research experience (2 pages)

The current format is:

  • Study proposal (2 pages)
  • Personal statement (3 pages)

The personal statement is to include research experience and how that will help guide a successful study. I find this format a lot better, since my personal statement involves so much of my research experience. And that sentiment was echoed in a handful of awarded NSF GRFP statements I read. But clearly, four pages to write all of that stuff versus three pages is much different. Flow becomes a secret assassin -- one that wasn't necessarily implicit in the previous format (since the two documents were stand-alone).

The study proposal was the same. The older ones I was given (awarded in the 90s) had no real limitations about adjusting margins or font sizes. My advisor was awarded an NSF GRFP when she was at grad school. Her grant proposal was in 10 point, no-serif font with 0.5" margins. That'd be really nice for me. But that'd also send me to a rejection without review.

Use your previous statements with extreme caution. You don't want to waste your time writing three documents when you really have to write two.

Advisor edits are key.

Two of my three recommendation letters are coming from my current advisors. They have been invaluable in this process. On one hand, my advisor Dr. Jordan was awarded this previously. On the other hand, my other advisor, Dr. Gillam, is a grant-getting machine. Both advisors went with a "hands off" approach for the most part -- all major edits from me came from suggestions rather than actual rewording/restructuring of what I wrote. This was both time effective for them (since they didn't have to spend time actually retooling things) and constructive for me (since I'm a grant noob).

I highly recommend this approach. The only "physical" edits I had came from my old friend Brandon Lowder -- a Berkeley alum with a BA in literature -- who edited for flow and tone. Those edits were well received (I assume, since they weren't re-edited out).

Look for more than just your advisor. Two or three people (preferably one who is a grant-getter) should be helping you out in order to get a better picture of how to say what you want to say most effectively.

Try avoiding a mass of edits.

Something I may switch over to is Google Docs. The ability to edit simultaneously with a chat window in a corner is something that no other free word processor can speak to. It would also help mitigate the editingnado when one sends out a draft to EVERYONE.

Even just sending a draft to two people can sometimes be difficult. I spent two hours one editing session trying to merge both of my advisor's edits together as best as I could.

The majority of edits came from my main advisor. That's a no-brainer, in my opinion. The second most edits have come from my other advisor (Dr. Gillam is second-most interested in my development, first being Dr. Jordan). The other edits have come from Brandon and the citizens of Twitter's science community (most notably, @biochembelle).

The four weeks of editing is both time consuming to actually edit but also time consuming to merge edits carefully. Avoiding the time suck from merging edits would be most advantageous.

You have two (really, three) chances at this.

This grant is specifically intended for entry level graduate students. The most commonly known students are the first- and second-year graduate students. However, graduating undergraduates can also apply for this grant. Now, you might not actually have the cleanest of ideas as an undergraduate, but having funding going into graduate school is a HUGE feather-in-the-cap. Especially if you are in a position to even propose an idea at all. To be frank, my proposal has been sitting with me practically since my last semester of undergraduate work. If I could have been awarded this fellowship, my chances of entering graduate school straight out of undergrad would have been amazing. If not guaranteed.

Your last year of eligibility is your second year of graduate school. More precisely, you cannot have completed more than 12 months of graduate study by August of the year you are applying. Essentially, if you enter graduate school in the Fall, you should have two cracks at this award. If you were one of those genius kids who also applied in your final year of UG, then you have three shots.

So if you get rejected, don't ruin yourself over it.

Alternate solution: NIH F31

Almost conveniently, the NIH F31 is due a month after the NSF GRFP. If you are feeling like you weren't so competitive on the NSF GRFP, it's probably best to re-tool your application to fit the NIH format and language.

If you don't want to grind your NSF GRFP into an NIH F31, you could wait until the GRF awards are announced in late March/early April. Conveniently again, NIH F31 grants are due in early April. So if you are rejected and have been working on an F31 since the beginning of the year, this would be a good chance to submit.

Lastly, NIH F31's can also be submitted during the August cycle -- right when the NSF GRFPs open the application form. If you have been working over the summer on a GRFP proposal, you could in theory also be working on a NIH F31.


2 thoughts on “Some things I've learned up to now about the NSF GRFP”

  1. katiesci says:

    It's worth mentioning that the undergraduate applications are reviewed separately from the graduate student ones.

    1. mmm says:


Leave a Reply