Thread by @azfaust

Let's pretend that there is no pandemic and we can all focus on our work. Let's do a thread on writing NSF grants.

I learned to write NIH grants first, so I came to NSF writing from an entirely NIH perspective. In some ways, that was helpful but I also had to unlearn some bad habits. I'll share my experience - but please feel free to chime in.

There are 2 required sections of an NSF grant: Intellectual Merit and Broader Impacts. Intellectual merit is where you put all the science-y stuff: background, prelim data, research approach.

Broader Impacts is really 2 parts: the potential societal benefits of your research and the practical activities you will engage in to broaden participation in STEM.

For Intellectual Merit, the basic science grant writing applies: explain why what you are doing is important, demonstrate feasibility with preliminary data, describe the planned research approach, describe the statistical analyses, discuss alternative approaches.

One section in IM that you do have to include is results of previous NSF support for either the PI or the co-PIs (oh, did I mention that NSF has co-PIs?). This only applies if you have had NSF funding in the last 5 years or if you currently have NSF funding.

What do you have to include? Award number, title, grant summary (on both IM and BI), publications resulting from that award, other research products (data, models, etc) and if they are publicly available, and if the current proposal is related, how they are elated.

The general structure of the IM section is really up to the writer. I have seen everything from a copy of an NIH grant outline to a list of 5 major objectives with relevant material interspersed. There is no separate innovation section - if you are using a novel technique...

... the burden is on you to compare it to the established approaches. In my experience, there are no bonus points for doing something new if the old way works well to answer the question.

Oh, this brings me to the question. Absolutely have a clear question that you will answer. I've worked on grants across a wide swath of NSF - from Mathematics to Ecology - and a clearly stated question will never fail you.

Hypotheses do not work for every research area - Chemistry to me is always a great example of a field that doesn't necessarily lend itself hypotheses very well - but all research should start with a question.

So if I am starting a new NSF grant and need an outline for my IM section? 1) Background (which should include the question and a discussion of what other groups are doing to answer that question) 2) Aims 3) Preliminary data (that demonstrates that the team can do what they say)

3) Results of the previous NSF support (to demonstrate productivity of the team) 4) Research approach (includes experimental description, analysis, alternative approaches, outcomes)

I leave my impact section to the Broader Impacts. I usually separate BI into 2 sections: societal impacts and educational and outreach activities.

Societal impacts I keep brief - 1-2 paragraphs at most. A short description of how the research will benefit the society. This is mostly fluff (sorry).

Now, educational and outreach activities matter. They won't win you the grant but if they come across as an afterthought, they may lose you the grant.

(Several NSF POs told me that given a tie between proposals, they give funding preference to proposals that have an exciting Broader Impacts section).

A lot of people write in that they will start a new class for grad and undergrad students. That's easy for a lot of people and it's fine. But try to think bigger than that. NSF is really invested in STEM education and outreach to non-university communities.

That can be junior colleges, high-schools, or elementary schools. Outreach to communities through community organizations is also fantastic.

I strongly suggest finding programs that are already in place as opposed to proposing to start something on your own. The reason for that is that university faculty are usually too busy to really dedicate themselves to these sort of endeavors.

They require a lot of planning and time that few faculty have. Reach out to local museums who may already have community programs you can plug into. Another great resource are your education science colleagues. See if any are interested in a collaboration.

Don't skimp on BI - you can absolutely budget for BI activities. So don't shy away from something just because it costs money.

I also have to remind you all that NSF takes broadening STEM participation very seriously. They want to see a track record of training under-represented populations in STEM and they want to see that the BI effort relates to this.

Whatever educational and outreach activity you propose, do explain how it will benefit under-represented populations in STEM.

Finally, include a scientific justification for your educational activities. For example, show evidence from published research that the activities you propose are beneficial to the target community.

In the end, my experience with NSF has taught me that you don't need rigid outlines (looking at you NIH and Dod) to write a good proposal. And that you can separate good from mediocre by reading how they think about problems outside of their immediate area of research.