Funding rates in the US are declining. Even if NSF gets a little boost to its budget, the situation will remain grim in the United States and many places beyond.
For every funded project, there are many other meritorious and important projects that go unfunded. At least in my subfields in the USA, government funding tends to be feast-or-famine: there are haves and the have-nots. With simultaneous strokes of skill and luck, you can get funded. But with a stroke or two in the other direction, you can end up cashless.
So what to do about declining funding rates? Last year, Brian McGill of Dynamic Ecology ran a poll and then evaluated the results. A drop in award sizes, to increase the number of awards, was a popular option. I now want to ask: okay, how much? Where do your preferences lie on the scale?
Going with generalizations from historical data (following the practice of McGill), let’s just say that average award sizes (in biology) are $500,000 over three years, with a funding rate of 7.5%.
If you were the Grand Master of Funding Rates, how would you tweak risk and reward for 3-year projects?
Of course, if funding rates increase and award sizes drop, then behavioral ecology theory suggests that we would see an increase in the number of proposals and a decline in the quality of the proposals. That presumably could be factored into a more complex model. But if we just broadly look at how many people are submitting and how much they get, given the amount of current competition and current funding, what are your preferences?
Using this coarse measure, I am really interested where the sweet spot might be — whether as a community, we want to spend less time writing proposals, and spend more time doing science with less money. This is what I hear the situation is like with NSERC in Canada. Thoughts, insights, reasons you chose a particular option?
This indeed what the NSERC model was for many years – little time spent on, or worry about, securing funding; but small grants. The model has been shifting NSF-ward over the last few years, much to the distress of many of us in the Great White North.
I’m a bit surprised to find I’m at the high end of the spectrum! (250k/15%)
I started from the premise that a postdoc or a grad student costs about $50k a year (mileage varies depending on tuition support, benefits, seniority, &c), so the minimum is $150k for a three-year project with one person dedicated to it … but maybe we shouldn’t be thinking about NSF as the means to add people to the lab?
I agree with you, once you start paying for personnel, you burn through the grant really quickly. My pretty cheap undergrads-in-Costa-Rica-over-the-summer lab squeaked by on 150k over three years.
Great poll!
I agree NSERC has had a very different point on the risk reward curve – a point at which the average colleague I talked to was much happier than in the US. But any comparison with NSERC has to take into account there are other systemic differences including no indirect and no summer salary (almost everybody is a 12 month-appointment). Those things tend to make researchers happier too.
I won’t comment on the poll results to avoid biasing. But I have been in a lot of conversations (about journal accept rates, grant funding rates, etc) where people agree 25%-33% is an optimal acceptance rate to ensure you are competitive and getting only high quality work but also not selecting only a subset of the quality work in a more or less random fashion.. NSERC has traditionally been to one side of this (50%) while NSF has been way to the other side (5-7%).
Big difference in Canada, in addition to 12 month salaries, is that most students get NSERC grants to cover their salaries and tuition. So the “only” thing you need money for is the actual research. In the US, nearly all of the funding is needed for faculty salary, student salary, tuition, and overheads. For me, about 95-99% of my grants pay for things that are obtained from other sources in Canada. (I am a modeler so I need few field supplies.)
In other words, I might need $400,000 to fund a multi-year project in the US, but could do the identical research in Canada with funding of $20,000 (travel, computer, publication costs).
@Trevor:
“most students get NSERC grants to cover their salaries and tuition”
That’s not true at Calgary or any other Canadian university with which I’m familiar. The vast majority of Canadian grad students do not hold one of the major NSERC fellowships, which are extremely competitive. Rather, at Calgary, most graduate students are supported as TAs during the academic year and paid a summer salary from their supervisor’s NSERC grants over the summer.
It is true that systemic differences between the US and Canada, in combination with differences between NSERC and NSF policies, mean that Canadian academics spend their NSERC Discovery Grants on a different mix of stuff than US academics spend their NSF grants on.
That is not true at every Canadian institution–I’m at a well-regarded comprehensive university, and <20% of the grad students in my department have NSERC scholarships. For students without scholarships, we pay ~$13,000/year per student, more if they don’t TA. At that rate, an average NSERC grant in E&E doesn’t go far at all.
I’ve spent time on both sides of the border, and held both NSF and NSERC grants, and I’d say that both systems are difficult, just in different ways. While NSERC works well for some people, it is not the promised land.
I’m not a fan of massive grants for a select few. A while back, I argued against this very thing in the CERC program in Canada (https://labandfield.wordpress.com/2013/05/06/the-value-of-the-canada-excellence-research-chairs-cerc-program/). We also link to a paper (a rebuttal, and a re-rebuttal) that found it was cheaper to give every applicant a flat rate of $30-40k/yr than to run the competition & dole out the funds
I feel like part of the problem with these amounts is institutional overhead. I have heard (with no data supporting it) that institutions are taking an increasingly larger amount from NSF grants over the last decade+ as public funding of institutions has shifted. If overhead amounts were capped by NSF (like they are at some other government agencies), what would our funding landscape look like?
As with Brian’s post on a similar topic, I’m an outlier on this one. Assuming that all we are changing is the award size (that is, that overhead rates or other things aren’t changing, too), $50,000/year is not a lot of money for personnel-intensive research. If I pay a technician $30,000/year, add in 30% for fringe, plus 55ish% overhead, and one technician has taken up more than the budget, even without any money for supplies, travel, etc. In my opinion, this would shift people towards using grad students as the personnel, since they can be supported on teaching assistantships in many departments. I think that would be a bad move. I don’t think anyone thinks we need to be training more PhD students, and I think there’s an extremely important place in science for technicians.
It’s pretty clear that if grant sizes are anywhere below 100K/year, then you can’t really have any full-time staff employed on the project at all. So if award sizes did shrink more than a little, then it would require a big change in the modus operandi for projects conducted at research institutions. I’m not sure if folks choosing lower amounts of funding realize that, but it’s an important thing to point out.
I always think of important cross-tab questions I should have asked well into the poll unfortunately (and it requires switching to Google Poll so its a pain). With your poll, I would have been really curious to see how the answers differed between those who have actually had a grant (i.e. a proxy for understanding overhead, benefits, etc) and those who haven’t. I venture those who have had a grant are not voting $50K or $75K because in the US with overhead, benefits, tuition & insurance add-ons, etc over 3 years that is pretty much 1 month of tech salary, a little travel by car and a few supplies. However, I think $150K-$200K is still realistic but a big change.
All the work I do is done on less than 50K/year. 80-90% of that is grad student summer salary+fringe. Most of the rest is fuel. All cobbled together from grants 5k – 100k each. And Meghan is right: this means no technicians, just myself , undergrads and 1-3 grad students (mostly Masters).
To be honest I see little/no relationship between quality and likelihood of funding among my own failed and successful proposals. Of course, that could just mean that if I like, it is the kiss of death :). I do have a contrary nature.
In followup to Alex’s suggestion: Unless one can prove that a review process adds value, a lottery is more efficient. As a reviewer, I can sort proposals into coarse categories, but do not think it is possible to rank by predicted future impact among a group of very good proposals. When success rates are so low, I doubt there is any value added by panel and reviewer time spent deciding among the top proposals. So: quick cut to eliminate poor proposals, then lottery.
Full disclosure: Despite several invited full proposals over the last few years, I have never had an NSF grant funded. So sour grapes and all.
I think the 100 and 150 grand options are the best. 250 is obviously a nicer amount, but the 15% funding rate is too low. That is the rate with which we who depend upon the NIH are dealing (7-8% for first submission, 15% when revised grants are included). It’s too low.
What I would like to see is a shift to smaller labs at R1 schools. This would create more faculty openings, and would likely encourage better one-on-one student mentoring. I was had a colleague who was so proud that he had 20+ people working in his lab. Most of them were terribly unhappy and felt that they were drowning under his “care.” I was disgusted when he invited a speaker from Japan who had roughly 30 people in his lab and my colleague went on and on about how jealous he was that the speaker had more than he had. The guy was obsessed with collecting personnel and papers, and seemed to have only minimal concern with the success and happiness of the individual student or postdoc. Most of his postdocs could have easily run a better lab themselves, anyway. It was a truly disturbing spectacle. So I say let’s limit labs to 5 or fewer people, make most grants smaller, and give more people a chance to be good independent scientists and teachers.
As mentioned, the question of whether postdocs are included in this sum or whether funding for postdocs would change is important. With overhead included, I cost $100,000 per year. So a $150,000 grant covers about a year (assuming $50,000 for the research, travel, conferences, publishing, etc.). Lowering grant funding is likely to make postdoc positions even shorter than they already are, making postdocs less about building a career and more of a way to get short-term high quality labor.