whether or not you should pursue a career in science still depends mostly on that thing that is you

I took the plunge a couple of days ago and answered my first question on Quora. Since Brad Voytek won’t shut up about how great Quora is, I figured I should give it a whirl. So far, Brad is not wrong.

The question in question is: “How much do you agree with Johnathan Katz’s advice on (not) choosing science as a career? Or how realistic is it today (the article was written in 1999)?” The Katz piece referred to is here. The gist of it should be familiar to many academics; the argument boils down to the observation that relatively few people who start graduate programs in science actually end up with permanent research positions, and even then, the need to obtain funding often crowds out the time one has to do actual science. Katz’s advice is basically: don’t pursue a career in science. It’s not an optimistic piece.

My answer is, I think, somewhat more optimistic. Here’s the full text:

The real question is what you think it means to be a scientist. Science differs from many other professions in that the typical process of training as a scientist–i.e., getting a Ph.D. in a scientific field from a major research university–doesn’t guarantee you a position among the ranks of the people who are training you. In fact, it doesn’t come close to guaranteeing it; the proportion of PhD graduates in science who go on to obtain tenure-track positions at research-intensive universities is very small–around 10% in most recent estimates. So there is a very real sense in which modern academic science is a bit of a pyramid scheme: there are a relatively small number of people at the top, and a lot of people on the rungs below laboring to get up to the top–most of whom will, by definition, fail to get there.

If you equate a career in science solely with a tenure-track position at a major research university, and are considering the prospect of a Ph.D. in science solely as an investment intended to secure that kind of position, then Katz’s conclusion is difficult to escape. He is, in most respects, correct: in most biomedical, social, and natural science fields, science is now an extremely competitive enterprise. Not everyone makes it through the PhD; of those who do, not everyone makes it into–and then through–one more more postdocs; and of those who do that, relatively few secure tenure-track positions. Then, of those few “lucky” ones, some will fail to get tenure, and many others will find themselves spending much or most of their time writing grants and managing people instead of actually doing science. So from that perspective, Katz is probably right: if what you mean when you say you want to become a scientist is that you want to run your own lab at a major research university, then your odds of achieving that at the outset are probably not very good (though, to be clear, they’re still undoubtedly better than your odds of becoming a successful artist, musician, or professional athlete). Unless you have really, really good reasons to think that you’re particularly brilliant, hard-working, and creative (note: undergraduate grades, casual feedback from family and friends, and your own internal gut sense do not qualify as really, really good reasons), you probably should not pursue a career in science.

But that’s only true given a rather narrow conception where your pursuit of a scientific career is motivated entirely by the end goal rather than by the process, and where failure is anything other than ending up with a permanent tenure-track position. By contrast, if what you’re really after is an environment in which you can pursue interesting questions in a rigorous way, surrounded by brilliant minds who share your interests, and with more freedom than you might find at a typical 9 to 5 job, the dream of being a scientist is certainly still alive, and is worth pursuing. The trivial demonstration of this is that if you’re one of the many people who actuallyenjoy the graduate school environment (yes, they do exist!), it may not even matter to you that much whether or not you have a good shot of getting a tenure-track position when you graduate.

To see this, imagine that you’ve just graduated with an undergraduate degree in science, and someone offers you a choice between two positions for the next six years. One position is (relatively) financially secure, but involves rather boring work of quesitonable utility to society, an inflexible schedule, and colleagues who are mostly only there for a paycheck. The other position has terrible pay, but offers fascinating and potentially important work, a flexible lifestyle, and colleagues who are there because they share your interests and want to do scientific research.

Admittedly, real-world choices are rarely this stark. Many non-academic jobs offer many of the same perceived benefits of academia (e.g., many tech jobs offer excellent working conditions, flexible schedules, and important work). Conversely, many academic environments don’t quite live up to the ideal of a place where you can go to pursue your intellectual passion unfettered by the annoyances of “real” jobs–there’s often just as much in the way of political intrigue, personality dysfunction, and menial due-paying duties. But to a first approximation, this is basically the choice you have when considering whether to go to graduate school in science or pursue some other career: you’re trading financial security and a fixed 40-hour work week against intellectual engagement and a flexible lifestyle. And the point to note is that, even if we completely ignore what happens after the six years of grad school are up, there is clearly a non-negligible segment of the population who would quite happy opt for the second choice–even recognizing full well that at the end of six years they may have to leave and move onto something else, with little to show for their effort. (Of course, in reality we don’t need to ignore what happens after six years, because many PhDs who don’t get tenure-track positions find rewarding careers in other fields–many of them scientific in nature. And, even though it may not be a great economic investment, having a Ph.D. in science is a great thing to be able to put on one’s resume when applying for a very broad range of non-academic positions.)

The bottom line is that whether or not you should pursue a career in science has as much or more to do with your goals and personality as it does with the current environment within or outside of (academic) science. In an ideal world (which is certainly what the 1970s as described by Katz sound like, though I wasn’t around then), it wouldn’t matter: if you had any inkling that you wanted to do science for a living, you would simply go to grad school in science, and everything would probably work itself out. But given real-world constraints, it’s absolutely essentially that you think very carefully about what kind of environment makes you happy and what your expectations and goals for the future are. You have to ask yourself: Am I the kind of person who values intellectual freedom more than financial security? Do I really love the process of actually doing science–not some idealized movie version of it, but the actual messy process–enough to warrant investing a huge amount of my time and energy over the next few years? Can I deal with perpetual uncertainty about my future? And ultimately, would I be okay doing something that I really enjoy for six years if at the end of that time I have to walk away and do something very different?

If the answer to all of these questions is yes–and for many people it is!–then pursuing a career in science is still a very good thing to do (and hey, you can always quit early if you don’t like it–then you’ve lost very little time!). If the answer to any of them is no, then Katz may be right. A prospective career in science may or may not be for you, but at the very least, you should carefully consider alternative prospects. There’s absolutely no shame in going either route; the important thing is just to make an honest decision that takes the facts as they are and not as you wish that they were.

A couple of other thoughts I’ll add belatedly:

  • Calling academia a pyramid scheme is admittedly a bit hyperbolic. It’s true that the personnel structure in academia broadly has the shape of a pyramid, but that’s true of most organizations in most other domains too. Pyramid schemes are typically built on promises and lies that (almost by definition) can’t be realized, and I don’t think many people who enter a Ph.D. program in science can claim with a straight face that they were guaranteed a permanent research position at the end of the road (or that it’s impossible to get such a position). As I suggested in this post, it’s much more likely that everyone involved is simply guilty of minor (self-)deception: faculty don’t go out of their way to tell prospective students what the odds are of actually getting a tenure-track position, and prospective grad students don’t work very hard to find out the painful truth, or to tell faculty what their real intentions are after they graduate. And it may actually be better for everyone that way.
  • Just in case it’s not clear from the above, I’m not in any way condoning the historically low levels of science funding, or the fact that very few science PhDs go on to careers in academic research. I would love for NIH and NSF budgets (or whatever your local agency is) to grow substantially–and for everyone get exactly the kind of job they want, academic or not. But that’s not the world we live in, so we may as well be pragmatic about it and try to identify the conditions under which it does or doesn’t make sense to pursue a career in science right now.
  • I briefly mention this above, but it’s probably worth stressing that there are many jobs outside of academia that still allow one to do scientific research, albeit typically with less freedom (but often for better hours and pay). In particular, the market for data scientists is booming right now, and many of the hires are coming directly from academia. One lesson to take away from this is: if you’re in a science Ph.D. program right now, you should really spend as much time as you can building up your quantitative and technical skills, because they could very well be the difference between a job that involves scientific research and one that doesn’t in the event you leave academia. And those skills will still serve you well in your research career even if you end up staying in academia.

 

on writing: some anecdotal observations, in no particular order

  • Early on in graduate school, I invested in the book “How to Write a Lot“. I enjoyed reading it–mostly because I (mistakenly) enjoyed thinking to myself, “hey, I bet as soon as I finish this book, I’m going to start being super productive!” But I can save you the $9 and tell you there’s really only one take-home point: schedule writing like any other activity, and stick to your schedule no matter what. Though, having said that, I don’t really do that myself. I find I tend to write about 20 hours a week on average. On a very good day, I manage to get a couple of thousand words written, but much more often, I get 200 words written that I then proceed to rewrite furiously and finally trash in frustration. But it all adds up in the long run I guess.
  • Some people are good at writing one thing at a time; they can sit down for a week and crank out a solid draft of a paper without every looking sideways at another project. Personally, unless I have a looming deadline (and I mean a real deadline–more on that below), I find that impossible to do; my general tendency is to work on one writing project for an hour or two, and then switch to something else. Otherwise I pretty much lose my mind. I also find it helps to reward myself–i.e., I’ll work on something I really don’t want to do for an hour, and then play video games for a while switch to writing something more pleasant.
  • I can rarely get any ‘real’ writing (i.e., stuff that leads to publications) done after around 6 pm; late mornings (i.e., right after I wake up) are usually my most productive writing time. And I generally only write for fun (blogging, writing fiction, etc.) after 9 pm. There are exceptions, but by and large that’s my system.
  • I don’t write many drafts. I don’t mean that I never revise papers, because I do–obsessively. But I don’t sit down thinking “I’m going to write a very rough draft, and then I’ll go back and clean up the language.” I sit down thinking “I’m going to write a perfect paper the first time around,” and then I very slowly crank out a draft that’s remarkably far from being perfect. I suspect the former approach is actually the more efficient one, but I can’t bring myself to do it. I hate seeing malformed sentences on the page, even if I know I’m only going to delete them later. It always amazes and impresses me when I get Word documents from collaborators with titles like “AmazingNatureSubmissionVersion18”. I just give my documents all the title “paper_draft”. There might be a V2 or a V3, but there will never, ever be a V18.
  • Papers are not meant to be written linearly. I don’t know anyone who starts with the Introduction, then does the Methods and Results, and then finishes with the Discussion. Personally I don’t even write papers one section at a time. I usually start out by frantically writing down ideas as they pop into my head, and jumping around the document as I think of other things I want to say. I frequently write half a sentence down and then finish it with a bunch of question marks (like so: ???) to indicate I need to come back later and patch it up. Incidentally, this is also why I’m terrified to ever show anyone any of my unfinished paper drafts: an unsuspecting reader would surely come away thinking I suffer from a serious thought disorder. (I suppose they might be right.)
  • Okay, that last point is not entirely true. I don’t write papers completely haphazardly; I do tend to write Methods and Results before Intro and Discussion. I gather that this is a pretty common approach. On the rare occasions when I’ve started writing the Introduction first, I’ve invariably ended up having to completely rewrite it, because it usually turns out the results aren’t actually what I thought they were.
  • My sense is that most academics get more comfortable writing as time goes on. Relatively few grad students have the perseverance to rapidly crank out publication-worthy papers from day 1 (I was definitely not one of them). I don’t think this is just a matter of practice; I suspect part of it is a natural maturation process. People generally get more conscientious as they age; it stands to reason that writing (as an activity most people find unpleasant) should get easier too. I’m better at motivating myself to write papers now, but I’m also much better about doing the dishes and laundry–and I’m pretty sure that’s not because practice makes dishwashing perfect.
  • When I started grad school, I was pretty sure I’d never publish anything, let alone graduate, because I’d never handed in a paper as an undergraduate that wasn’t written at the last minute, whereas in academia, there are virtually no hard deadlines (see below). I’m not sure exactly what changed. I’m still continually surprised every time something I wrote gets published. And I often catch myself telling myself, “hey, self, how the hell did you ever manage to pay attention long enough to write 5,000 words?” And then I reply to myself, “well, self, since you ask, I took a lot of stimulants.”
  • I pace around a lot when I write. A lot. To the point where my labmates–who are all uncommonly nice people–start shooting death glares my way. It’s a heritable tendency, I guess (the pacing, not the death glare attraction); my father also used to pace obsessively. I’m not sure what the biological explanation for it is. My best guess is it’s an arousal-mediated effect: I can think pretty well when I’m around other people, or when I’m in motion, but if I’m sitting at a desk and I don’t already know exactly what I want to say, I can’t get anything done. I generally pace around the lab or house for a while figuring out what I want to say, and then I sit down and write until I’ve forgotten what I want to say, or decide I didn’t really want to say that after all. In practice this usually works out to 10 minutes of pacing for every 5 minutes of writing. I envy people who can just sit down and calmly write for two or three hours without interruption (though I don’t think there are that many of them). At the same time, I’m pretty sure I burn a lot of calories this way.
  • I’ve been pleasantly surprised to discover that I much prefer writing grant proposals to writing papers–to the point where I actually enjoy writing grant proposals. I suspect the main reason for this is that grant proposals have a kind of openness that papers don’t; with a paper, you’re constrained to telling the story the data actually support, whereas a grant proposal is as good as your vision of what’s possible (okay, and plausible). A second part of it is probably the novelty of discovery: once you conduct your analyses, all that’s left is to tell other people what you found, which (to me) isn’t so exciting. I mean, I already think I know what’s going on; what do I care if you know? Whereas when writing a grant, a big part of the appeal for me is that I could actually go out and discover new stuff–just as long as I can convince someone to give me some money first.
  • At a a departmental seminar attended by about 30 people, I once heard a student express concern about an in-progress review article that he and several of the other people at the seminar were collaboratively working on. The concern was that if all of the collaborators couldn’t agree on what was going to go in the paper (and they didn’t seem to be able to at that point), the paper wouldn’t get written in time to make the rapidly approaching deadline dictated by the journal editor. A senior and very brilliant professor responded to the student’s concern by pointing out that this couldn’t possibly be a real problem seeing as in reality there is actually no such thing as a hard writing deadline. This observation didn’t go over so well with some of the other senior professors, who weren’t thrilled that their students were being handed the key to the kingdom of academic procrastination so early in their careers. But it was true, of course: with the major exception of grant proposals (EDIT: and as Garrett points out in the comments below, conference publications in disciplines like Computer Science), most of the things academics write (journal articles, reviews, commentaries, book chapters, etc.) operate on a very flexible schedule. Usually when someone asks you to write something for them, there is some vague mention somewhere of some theoretical deadline, which is typically a date that seems so amazingly far off into the future that you wonder if you’ll even be the same person when it rolls around. And then, much to your surprise, the deadline rolls around and you realize that you must in fact really bea different person, because you don’t seem to have any real desire to work on this thing you signed up for, and instead of writing it, why don’t you just ask the editor for an extension while you go rustle up some motivation. So you send a polite email, and the editor grudgingly says, “well, hmm, okay, you can have another two weeks,” to which you smile and nod sagely, and then, two weeks later, you send another similarly worded but even more obsequious email that starts with the words “so, about that extension…”

    The basic point here is that there’s an interesting dilemma: even though there rarely are any strict writing deadlines, it’s to almost everyone’s benefit to pretend they exist. If I ever find out that the true deadline (insofar as such a thing exists) for the chapter I’m working on right now is 6 months from now and not 3 months ago (which is what they told me), I’ll probably relax and stop working on it for, say, the next 5 and a half months. I sometimes think that the most productive academics are the ones who are just really really good at repeatedly lying to themselves.

  • I’m a big believer in structured procrastination when it comes to writing. I try to always have a really unpleasant but not-so-important task in the background, which then forces me to work on only-slightly-unpleasant-but-often-more-important tasks. Except it often turns out that the unpleasant-but-no-so-important task is actually an unpleasant-but-really-important task after all, and then I wake up in a cold sweat in the middle of the night thinking of all the ways I’ve screwed myself over. No, just kidding. I just bitch about it to my wife for a while and then drown my sorrows in an extra helping of ice cream.
  • I’m really, really, bad at restarting projects I’ve put on the back burner for a while. Right now there are 3 or 4 papers I’ve been working on on-and-off for 3 or 4 years, and every time I pick them up, I write a couple of hundred words and then put them away for a couple of months. I guess what I’m saying is that if you ever have the misfortune of collaborating on a paper with me, you should make sure to nag me several times a week until I get so fed up with you I sit down and write the damn paper. Otherwise it may never see the light of day.
  • I like writing fiction in my spare time. I also occasionally write whiny songs. I’m pretty terrible at both of these things, but I enjoy them, and I’m told (though I don’t believe it for a second) that that’s the important thing.

what Paul Meehl might say about graduate school admissions

Sanjay Srivastava has an excellent post up today discussing the common belief among many academics (or at least psychologists) that graduate school admission interviews aren’t very predictive of actual success, and should be assigned little or no weight when making admissions decisions:

The argument usually goes something like this: “All the evidence from personnel selection studies says that interviews don’t predict anything. We are wasting people’s time and money by interviewing grad students, and we are possibly making our decisions worse by substituting bad information for good.“

I have been hearing more or less that same thing for years, starting when I was grad school myself. In fact, I have heard it often enough that, not being familiar with the literature myself, I accepted what people were saying at face value. But I finally got curious about what the literature actually says, so I looked it up.

I confess that I must have been drinking from the kool-aid spigot, because until I read Sanjay’s post, I’d long believed something very much like this myself, and for much the same reason. I’d never bothered to actually, you know, look at the data myself. Turns out the evidence and the kool-aid are not compatible:

A little Google Scholaring for terms like “employment interviews“ and “incremental validity“ led me to a bunch of meta-analyses that concluded that in fact interviews can and do provide useful information above and beyond other valid sources of information (like cognitive ability tests, work sample tests, conscientiousness, etc.). One of the most heavily cited is a 1998 Psych Bulletin paper by Schmidt and Hunter (link is a pdf; it’s also discussed in this blog post). Another was this paper by Cortina et al, which makes finer distinctions among different kinds of interviews. The meta-analyses generally seem to agree that (a) interviews correlate with job performance assessments and other criterion measures, (b) interviews aren’t as strong predictors as cognitive ability, (c) but they do provide incremental (non-overlapping) information, and (d) in those meta-analyses that make distinctions between different kinds of interviews, structured interviews are better than unstructured interviews.

This seems entirely reasonable, and I agree with Sanjay that it clearly shows that admissions interviews aren’t useless, at least in an actuarial sense. That said, after thinking about it for a while, I’m not sure these findings really address the central question admissions committees care about. When deciding which candidates to admit as students, the relevant question isn’t really what factors predict success in graduate school?, it’s what factors should the admissions committee attend to when making a decision? These may seem like the same thing, but they’re not. And the reason they’re not is that knowing which factors are predictive of success is no guarantee that faculty are actually going to be able to use that information in an appropriate way. Knowing what predicts performance is only half the story, as it were; you also need to know exactly how to weight different factors appropriately in order to generate an optimal prediction.

In practice, humans turn out to be incredibly bad at predicting outcomes based on multiple factors. An enormous literature on mechanical (or actuarial) prediction, which Sanjay mentions in his post, has repeatedly demonstrated that in many domains, human judgments are consistently and often substantially outperformed by simple regression equations. There are several reasons for this gap, but one of the biggest ones is that people are just shitty at quantitatively integrating multiple continuous variables. When you visit a car dealership, you may very well be aware that your long-term satisfaction with any purchase is likely to depend on some combination of horsepower, handling, gas mileage, seating comfort, number of cupholders, and so on. But the odds that you’ll actually be able to combine that information in an optimal way are essentially nil. Our brains are simply not designed to work that way; you can’t internally compute the value you’d get out of a car using an equation like 1.03*cupholders + 0.021*horsepower + 0.3*mileage. Some of us try to do it that way–e.g., by making very long pro and con lists detailing all the relevant factors we can possibly think of–but it tends not to work out very well (e.g., you total up the numbers and realize, hey, that’s not the answer I wanted! And then you go buy that antique ’68 Cadillac you had your eye on the whole time you were pretending to count cupholders in the Nissan Maxima).

Admissions committees face much the same problem. The trouble lies not so much in determining which factors predict graduate school success (or, for that matter, many other outcomes we care about in daily life), but in determining how to best combine them. Knowing that interview performance incrementally improves predictions is only useful if you can actually trust decision-makers to weight that variable very lightly relative to other more meaningful predictors like GREs and GPAs. And that’s a difficult proposition, because I suspect that admissions discussions rarely go like this:

Faculty Member 1: I think we should accept Candidate X. Her GREs are off the chart, great GPA, already has two publications.
Faculty Member 2: I didn’t like X at all. She didn’t seem very excited to be here.
FM1: Well, that doesn’t matter so much. Unless you really got a strong feeling that she wouldn’t stick it out in the program, it probably won’t make much of a difference, performance-wise.
FM2: Okay, fine, we’ll accept her.

And more often go like this:

FM1: Let’s take Candidate X. Her GREs are off the chart, great GPA, already has two publications.
FM2: I didn’t like X at all. She didn’t seem very excited to be here.
FM1: Oh, you thought so too? That’s kind of how I felt too, but I didn’t want to say anything.
FM2: Okay, we won’t accept X. We have plenty of other good candidates with numbers that are nearly as good and who seemed more pleasant.

Admittedly, I don’t have any direct evidence to back up this conjecture. Except that I think it would be pretty remarkable if academic faculty departed from experts in pretty much every other domain that’s been tested (clinical practice, medical diagnosis, criminal recidivism, etc.) and were actually able to do as well (or even close to as well) as a simple regression equation. For what it’s worth, in many of the studies of mechanical prediction, the human experts are explicitly given all of the information passed to the prediction equation, and still do relatively poorly. In other words, you can hand a clinical psychologist a folder full of quantitative information about a patient, tell them to weight it however they want, and even the best clinicians are still going to be outperformed by a mechanical prediction (if you doubt this to be true, I second Sanjay in directing you to Paul Meehl’s seminal body of work–truly some of the most important and elegant work ever done in psychology, and if you haven’t read it, you’re missing out). And in some sense, faculty members aren’t really even experts about admissions, since they only do it once a year. So I’m pretty skeptical that admissions committees actually manage to weight their firsthand personal experience with candidates appropriately when making their final decisions. It seems much more likely that any personality impressions they come away with will just tend to drown out prior assessments based on (relatively) objective data.

That all said, I couldn’t agree more with Sanjay’s ultimate conclusion, so I’ll just end with this quote:

That, of course, is a testable question. So if you are an evidence-based curmudgeon, you should probably want some relevant data. I was not able to find any studies that specifically addressed the importance of rapport and interest-matching as predictors of later performance in a doctoral program. (Indeed, validity studies of graduate admissions are few and far between, and the ones I could find were mostly for medical school and MBA programs, which are very different from research-oriented Ph.D. programs.) It would be worth doing such studies, but not easy.

Oh, except that I do want to add that I really like the phrase “evidence-based curmudgeon“, and I’m totally stealing it.