Inside Higher Ed reports on a new study of Ph.D. completion rates by discipline. The original data are available as PowerPoint files that I haven't looked at, but IHE provides a summary in tabular form. Because everything looks more scientific as a graph, I cranked them into Excel and after the requisite ten minutes spent undoing all of Excel's horrid default graph options, ended up with this:
There's really not a whole lot surprising here: Scientists and engineers finish their degrees faster (in a bit more than six years), and humanists take much longer (just under 50% have finished in ten years). Humanities students are also more likely to end up with sizeable grad school debt-- 38% have more than $35,000 in loans, compared to 20% for math and physical science, 18% for life sciences, and only 12% for engineering. Moral: don't go to grad school in the humanities.
One thing that was interesting to me is that the science and engineering curves show a clear saturation-- after about eight years, they don't increase much. A bit over 60% of the people entering engineering or life science Ph.D.'s finish in ten years, while it's about 55% for math and physical science students. It's not clear from the story whether those people ever finish-- I'm betting that very few of them are on the Brian May thirty-year plan. Most of those people are probably leaving with terminal Master's degrees.
35-45% attrition may seem like a lot, but then, it's not actually that much different from overall college graduation rates, which are at about 57% in six years. I suspect the higher numbers for engineers reflect the fact that they're more likely to work for a few years before going for the doctorate (similarly, the lower debt numbers probably reflect the fact that engineers are more likely to have accumulated savings, or employers who will sponsor them through the degree program).
Anyway, as I said, there's not much new here, but it's nice to see the conventional wisdom confirmed.
- Log in to post comments
Make your own format the new default in Excel! It's easy and it's the best thing ever--you can save other graph types too and choose them instead of the default!
"Humanities students are also more likely to end up with sizeable grad school debt"
And have a more difficult time paying it off due to low wages.
Presumably humanities students are more likely to be TAing right through their PhD, which will slow them down somewhat, too.
That's maybe one of the most scary graphs I've seen in a long time. Over here, if you don't submit within four years you need special permission to submit at all!
(http://www.scieng.ed.ac.uk/Postgraduate/PhD/Howtosubmit.htm)
Steve has brought up an interesting point. The time constraints on PhD's are more restrictive in Australia as well. Perhaps due to the amount of funding available to pay for research assistants and teaching assistants? Has this subject been covered by anyone before? (For example, are restrictive time constraints "better"). Or is it irrelevant.
That's maybe one of the most scary graphs I've seen in a long time. Over here, if you don't submit within four years you need special permission to submit at all!
There are some schools in the US that have a similar requirement-- I want to say Princeton, but I could be wrong about that-- but it's not that common.
I think it's mostly due to differences in the overall educational system. Undergrad degrees in Europe (at least some parts of Europe) are closer to a Master's degree in the US, which means that students entering grad school are starting in a better position to finish quickly.
I doubt it makes much practical difference whether students are working with a tight limit or not. What counts as "enough" to get a Ph.D. is pretty nebulous, anyway, and you could easily redefine it to a level that is possible to reach in four years rather than six.
Humanities grad students end up with sizeable debt because there simply isn't the kind of funding available for scholarships and financial aid (other than loans) that there is for the sciences. They take longer to finish because not only must they TA throughout their studies, they often have to work a part-time job at the same time. And so forth. Financial considerations were another reason I left grad school in the humanities.... It would, indeed, have taken me a good 10 years to finish!
Brett@5: I guess it depends on how fast the field as a whole is moving. You may have seen the cartoon of the two lab-coated scientists walking past a tramp, and one saying to the other ~"Poor Smithers. He was doing fine until he got a term behind with the literature"~....
Are they counting from the start of a master's degree, or the start of the actual PhD program? In my field (music theory), it is more likely to get a MA at one school and then move to another for the PhD, rather than doing a joint MA/PhD program. Let's see, from the start of my MA it took seven years to complete, and five years from the start of the PhD. That includes three years of teaching full time, and TAing for the other four/two years. I was one of the fastest ones in my program, graduating at the same time as some who had been working on the degree for ten or more years. One person beat me, because he did not TA for the first two years of his program (he had a rare research fellowship).
There are some schools in the US that have a similar requirement-- I want to say Princeton, but I could be wrong about that-- but it's not that common.
Not true in physics, but somewhat true in astronomy.
Princeton's physics program attempts to get people done in 5 years. Most people are done by the 6th, and it's really not that big of a deal. After that, you're sort of on your own, but IIRC you can complete your thesis and submit/defend up to about 10 years after that without jumping through a lot of hoops. You just won't have any sort of financial aid or student status.
In astronomy, it's 4 years for most people, because the department really wants people to finish in 4. A quote from a page on the Princeton astro site: "Under exceptional circumstances the student may request financial assistance and extension of candidacy for the fifth year."
Asad
Caltech requires special signatures (and chat? I can't remember) from the Dean every year after your fourth if you haven't passed Candidacy, and every year after your fifth if you haven't graduated. It's not a real threat, though.
I'd like to see the science curves at top-twenty institutions. My intuition is that the long tail is mostly due to lower-tier schools.
Note that physical science includes chemistry (about twice as many BS grads compared to physics, not sure about grad students) and geosciences, so those data do not necessarily reflect the attrition in physics.
The AIP data suggest it is 45 to 50% depending on what you put in the denominator. There were 1100 PhD's granted in 2004, and about 2500 first year students entered graduate school each year in the late 90s, about 6 to 8 years earlier. That gives 44%, but only 87% of the 2003 first-year students declared a PhD as a goal. If you keep only that group in the denominator, the fraction rises to just over 50%. The truth is probably somewhere in between because some "undecided" students (and maybe some of the "MS only" or "no degree" students) will go on to a PhD. Students coming in with an MS will also mess up the ratio, but half looks about right.
PS -
My series of articles (I project it will run to four) on jobs in physics is making progress, but part 2 (demand for faculty) is still a working draft on the blog as I sort out how tables turn out within blogspot. I hope to wrap it up later today and move on to some discussion of job skills and preparing for the t-t while in grad school or on a post doc.
"NL", there is no long tail. What you see is a curve that is asymptotic to the completion percentage of all first year students (or maybe all students entering a PhD program) in each of those areas.
It is true that mean-time-to-PhD varies between quartiles in the National Academy data I recall seeing sometime in the past, but not by much more than a year or two. It also varies a lot within a quartile, and will vary by sub field. A HEP experimental student can sometimes wait a long time just to get data to analyze, while an organic chemistry student just has to carry out a project in the lab.
Should have looked before my previous posting. The link to the NAS table for physics is
http://books.nap.edu/html/researchdoc/appendix_l.html
You want table L-7. Note the copyright information. This is the 1995 study; they are probably working on the next one now. You can also look at other fields, if you wish.
CalTech is listed as 6.7 years to degree, which is more than the University of Miss at 6.4. Since Ole Miss is about as close to the bottom of the bottom quartile as CalTech is to the top of the top, there are no simple arguments. But it is true that the bottom quartile, on average, takes 1.5 years longer than the top quartile. However, with only 1/8 as many PhDs as the top quartile, they don't bias things very much.
I don't recall Princeton's molecular biology department saying anything about a 4 year limit when I visited; their average time to completion is around 5.5 years or so.