Prof-like Substance has a really nice summary of the real costs involved in giving someone (e.g. a grad student) a raise. This is in response to a discussion over at DrugMonkey's. Go read both posts.

In his post PlS points readers at the NSF's funding info page - anyone even fantasizing about applying to the NSF should have a look there. In the comments section to PlS's post, lylebot pointed out:

The link to the NSF funding info page is interesting, but I’m not sure I understand where the numbers are coming from. It gives funding rates of 20-40% for programs that I know are much lower.

I added the emphasis.

That's a great point. I "know" from my experience on NSF review panels that the funding rate is much lower than 20%. And yet that's not what the "official" numbers are saying. Where does this discrepancy come from?*

Buggered if I know, but that won't stop me from speculating...

And this is speculation, so take it as such.

So in my particular area the funding rate that everyone "knows" would be much closer to 10-15% than the ~20-30% listed on the NSF's funding info page. That's the funding rate we have in mind when on a panel. It's the funding rate bandied about by reviewers and PO's alike.

One possible source for this discrepancy might be that the ~10-15% funding rate we "know" really just applies to new applications, and not renewals. The funding rate for renewals is certainly higher than that for new applications. I'm not so sure about this explanation though. One would think that the distinction between funding rates would be made clearer and that we'd be talking about two rates rather than one.

Perhaps the discrepancy comes about because the ~10-15% we "know" applies only to the proposals we're reviewing. The "unknown" 5-15% may well include proposals we don't review. Now I'm not suggesting here that the NSF has some secret stock of research proposals they have decided should receive funding without peer-review. There are however "proposals" that aren't peer-reviewed. Here I'm thinking about the various supplements available to NSF-funded investigators (e.g. REU and RET supplements). You do need to submit a "proposal" to get one of these, but it's not reviewed per se. Or rather, it's the PO that reviews it. And I suspect the success rate is very, very high for these.** Such supplements are submitted via FastLane, NSF's online proposal submission and review system. I believe all funding data is also handled through FastLane - certainly supplements I've received are listed on there. Is it possible the NSF's funding info page simply pulls that data automagically from FastLane? If so, up goes the funding rate. Indeed, if supplements are included in the count, my personal funding rate at the NSF is a whopping 82%!

And of course there are grants other than research grants given out by the NSF. Those in the US may well have noticed that some of the kid's programs on PBS are funded in part by the NSF (e.g. Fetch and Cyberchase). One wonders what NSF Directorates fund these "non-research" grants*** and what the funding rates are.

So what's the bottom line here? It's not clear what the research grant funding rate is. The sort of data supplied on the NSF's funding info page is confusing. Dear NSF, can we have it broken down into useful nuggets? Please?

*The NSF is not alone with these funding rate "discrepancies." We all "know" that NIH funding rates are in the toilet. Single digit numbers have been discussed at great length. Yet, at least at NIGMS, the official numbers appear much higher...

** I've always received the REU supplements I've applied for. In fact, I've never heard of someone having one turned down. Of course that's just my experience.

*** Before you're tempted to whine in the comments that NSF should only be funding science research, go read their strategic plan. It is mandated that they do far more than just fund research. If you have a problem with that, and I personally do not, go whine at your congresscitter.