Skip to content
 

Crack Shot

Raghu Parthasarathy writes:

You might find this interesting, an article (and related essay) on the steadily declining percentage of NIH awards going to mid-career scientists and the steadily increasing percentage going to older researchers. The key figure is below. The part that may be of particular interest to you, since you’ve written about age-adjustment in demographic work: does an analysis like this have to account for the changing demographics of the US population (which wasn’t done), or it that irrelevant since there’s no necessarily link between the age distribution of scientists and that of the society they’re drawn from? I have no idea, but I figured you might.

Most of the article is about the National Heart Lung and Blood Institute, one of the NIH institutes, but I would bet that its findings are quite general.

Jeez, what an ugly pair of graphs! Actually, I’ve seen a lot worse. These graphs are actually pretty functional. But uuuuuugly. And what’s with those R-squareds? Anyway, the news seems good to me—the crossover point seems to be happening just about when I turn 55. And I’d sure like some of that National Heart Lung and Blood Institute for Stan.

16 Comments

  1. Jack PQ says:

    R2 of 0.993???? That’s just ridiculous. Doesn’t pass the smell test.

  2. Glen says:

    Are the R2 from linear regression? But the lines connecting the points are not linear, so where do the forecasts come from? Maybe the dashed lines is being buried beneath the solid lines, but then those R2 are too high. Plus why no R2 or forecast for blue lines (I assume because linear regression wasn’t appropriate). I’m confused.

    • Ben Prytherch says:

      The R-squareds for the blue lines look like they could be as low as 0.96, or maybe even 0.95. Not impressive enough to warrant putting on the graph.

      The dashed lines appear to have been made using MS Paint, or something similar. For the green lines, they appear to start at 1998 and go to straight to the 2022 projection. For the red lines, it looks like they start around 2012ish.

  3. A kitty from an alternative bubble of reality says:

    In which way you think the figures are ugly? To my eyes there really isn’t that much that bothers me. The only “my ‘obsession with detail’ type of mindset wouldn’t have let that slip” moment to me comes from the dashed prediction/extrapolation lines that continue under the main lines. As you said, the figures seem really functional, and in that sense I’d find them pretty well executed: there’s quite a bit of information, but at any point I don’t feel like I’d have to spit anything back on plate because I can’t get it down. But maybe I’m talking complete rubbish and my eyes – those treacherous little gelatinous clumps – are feeding me with false information! Mrr!

    • Andrew says:

      Kitty:

      I agree that the graphs are not too bad and they are generally functional. So why do I hate them? A bunch of reasons. First, they’re graphing 3 numbers that add up to 100%. I think it was a poor choice to graph the 3 lines as they did; it would make more sense to plot the lines for “under 40” and “under 55”. For that matter, they could’ve plot lines for under 30, under 40, under 50, etc. No need to just bin the data into 3 age groups. Second, all those ugly symbols and colors and horizontal lines. And those R-squareds! Whassup with that? Just plot the data.

    • Dzhaughn says:

      I find the Z-shape very distracting; the triangles draw the eye backwards along the green diagonal. The projections into the future are dumb, and why omit the projection for the blue line?

  4. Gordon A Fox says:

    Data on a change in the proportion of investigators being funded who are in an age group isn’t really interpretable without accounting for change in the proportion of all investigators (or all grant applicants) who are in the age group. If there’s, say, a 10% increase in the fraction of grants awarded to those 56-70+, how can I understand it without knowing how the composition of investigators as a whole changed over that time period?

  5. Alex says:

    A ternary plot would probably show the trends more nicely. All the data in just the one line.

  6. What about the original question, “… does an analysis like this have to account for the changing demographics of the US population (which wasn’t done), or it that irrelevant since there’s no necessarily link between the age distribution of scientists and that of the society they’re drawn from?” Of course this matters — if average time to do a Ph.D. plus postdoc goes up, it removes people from the 24–40 pool.

    And how about a second question from me: is this based on dollars or on just the number of grants? NIH could change these numbers pretty easily by giving out a pile of $5K grants to young investigators if it’s number of grants, but they’d have to give out tens of thousands of them to balance the $20M center grants that NIH likes to give out to the big players. Presumably the really big grants that are all admin and involve a lot of intricate university negotiations and cross-university negotiations and so go to older and more administrativey inclined types of researchers.

    And then a third question: who actually gets the money? Andrew is PI on a lot of grants, but he gets almost none of that money because he doesn’t buy himself out of teaching and can only supplement his 9-month salary with 2 summer months of salary, which has to be split among a dozen grants. So while he’ll be a PI on a grant, he might only get 5% of the income directly to his own salary and all the rest goes to (a) Columbia for overhead, (b) postdocs, and (c) soft-money research scientists like me (though I’m older than Andrew, so that doesn’t help these numbers, our other research scientists are 20 years younger). For others whose departments don’t suppor their grad students, as in computer science, the money trickles down all the way to grad students.

    I’m not saying this is a good thing or we should expect younger people to be supported by trickle down (because it takes away their independent control, though in practice Andrew and I aren’t exactly strict taskmasters!).

    Finally a fourth question: what does the binning into 24–40, 41-55, and 56+ demographics do to this? Would it have looked the same with 24–35, 36–45, 46–55, 56–65, 66–75? What about 5 year bins? Shouldn’t there be some kind of smoothing to connect this to Jonathan and Andrew’s work on age cohorts for mortality?

    • Thanks — yes, I was, in fact, hoping for a discussion of the data rather than the aesthetics. (Though I agree that they’re not the most elegant graphs.) As for Bob’s questions:

      — the example you give, of average time to do a Ph.D. plus postdoc rising, is one of demographics *within* the NIH awardee population changing, which obviously changes the award distribution statistics; my question was about the general demographics of the population.

      — “NIH could change these numbers pretty easily by giving out a pile of $5K.” If by “pretty easily” you mean “completely changing the structure of research project grants and hoping that no one notices or is annoyed by this,” then sure.

      — “… to give out tens of thousands of them to balance the $20M center grants that NIH likes to give out to the big players.” ?? References ?? Last I checked, about 55% of the NIH budget, or about $17 billion, goes to research project grants, the vast majority of which are single-investigator grants. Most NIH institutes are severely cutting large centers. And even many centers are collections of labs working together, not giant behemoths. (I’m part of a center, for example, that was funded with ~$10m for ~10 labs over 5 years. I’m flattered that I’m a “big player!”)

      So: thoughts?

        • In case anyone is interested in this topic, this just came out, a demographic analysis as requested.

          Why the US science and engineering workforce is aging rapidly

          Abstract:
          The science and engineering workforce has aged rapidly in recent years, both in absolute terms and relative to the workforce as a whole. This is a potential concern if the large number of older scientists crowds out younger scientists, making it difficult for them to establish independent careers. In addition, scientists are believed to be most creative earlier in their careers, so the aging of the workforce may slow the pace of scientific progress. We develop and simulate a demographic model, which shows that a substantial majority of recent aging is a result of the aging of the large baby boom cohort of scientists. However, changes in behavior have also played a significant role, in particular, a decline in the retirement rate of older scientists, induced in part by the elimination of mandatory retirement in universities in 1994. Furthermore, the age distribution of the scientific workforce is still adjusting. Current retirement rates and other determinants of employment in science imply a steady-state mean age 2.3 y higher than the 2008 level of 48.6.

  7. How about a plot of dollar weighted mean age divided by unweighted mean age vs time as a fraction of grant duration

    My dimensionless suggestion

Leave a Reply