Friday, July 24, 2009


Today I was perusing my department's data on graduate student years-to-degree for students who graduated in the past 5 years. The dataset included both MS and PhD students, and classified PhD students according to whether they arrived with an MS or an undergraduate degree.

There were four things that interested me about this dataset:

1. There was no correlation between years-to-degree and whether a student had an MS (from another institution) or a BS/BA degree at the start of the PhD program.

2. Most students needed more time than was covered by guaranteed financial support, but not much more (many needed ~ 1 year more).

3. When a colleague and I looked at the dataset and applied our qualitative and totally subjective evaluations as to which students were our most 'successful' (as in smart and hard-working, got awards/fellowships in grad school, published, got jobs after graduation etc.), we saw no trend in years-to-degree. Some 'successful' students zipped through the PhD program, finishing in 3-4 years; others took significantly longer.

4. Years-to-degree did not vary as a function of specific advisers. An individual adviser might have students who finished in 3-4 years and students who finished in >7 years.

Years-to-degree depends on a lot of factors, including (in no particular order):
  • the nature of the project;
  • the student's funding situation (TA, RA, no funding, continuous/discontinuous funding);
  • the student's extent of research experience prior to starting the graduate program;
  • the methods involved and access to necessary facilities or other support;
  • whether there are any unforeseen setbacks;
  • the student's work habits;
  • the student's health (physical and mental);
  • the presence or absence of personal crises or other significant events (weddings..), including those involving family, friends, or pets;
  • the work environment (research group or department dynamics);
  • the adviser's accessibility, level of interaction, sanity level;
  • the amount of additional non-research responsibilities required (e.g., departmental service)
  • number of degree requirements (courses, exams);
  • whether preliminary exams are passed on the first try;
  • expectations regarding number of published/submitted papers before graduation;
  • presence/absence of a post-graduation job offer (and amount of time spent job hunting while a student);
  • whether the student has an iPhone.
and likely many more that I have not listed. It is no wonder that years-to-degree varies so much and may not correlate with any obvious factor.


mukuge said...

hah! I like your last bullet point. It just serves to prove my obsevations that pretty little gadgets meant to help one go through life often becomes an Achilles' heel.

And yeah, iPhone interface programs hog up a computer's RAM unnecessarily.

El Charro said...

Years-to-degree did not vary as a function of specific advisers. An individual adviser might have students who finished in 3-4 years and students who finished in >7 years

Is there a time correlation on this? For example, when the advisers were assistant prof, did the students (on average) take more or less to graduate then when the advisers were full profs?

Alyssa said...

whether the student has an iPhone.

LOL! Yes, it's amazing the range of years-to-degree there is. I personally think one of the major factors is the want of the student to leave! Some people feel comfortable staying somewhere forever - some others (myself included) have a set time in their head they want to be at that institution, and will do everything they can to leave when they planned too!

qaz said...

I've always wondered at this years-to-degree obsession. (Is my department the only one that does this?) Graduate school is often one of the few places that there is room to maneuver intellectually. But there seems to be this "must-keep-them-on-track" obsession. Obviously, we don't want 13th years. But are 6th and 7th years really so awful?

Is this all just because people take a long time to get to faculty and their first R01? But I thought that occurred primarily because of long post-docs, not because of long graduate studies.

FSP - when you say "jobs after graduation" are you measuring immediately after or 10 years later?

Anonymous said...

I wonder if the current economic crisis will affect the years-to-degree of current senior graduate students. I know one student in my lab who will be ready to defend in the fall but really doesn't want to -- she worries that she won't find a job and is afraid to lose her subsidized student housing without one. She has a husband whose job in finance is no longer secure, as well as a young child, which only makes the situation harder.

Angela said...

Was having an iPhone a positive or negative? For me, it would be a motivating factor as it provides me a means to a) check email when not in lab/near a PC and b) listen to music. Though, thinking of my husband specifically, an iPhone could also be a major distraction from work.

Unknown said...

Thank you for that list. I think I'm going to print it out and hang it on the lab bulletin board, so I can just point to it whenever I start to hear that old line about "Student So-and-So graduated in 3.5 years..." Yeah, so? Do I look like Student So-and-So?

It is just awesome that some PIs like yourself realize all of these other factors which can influence years-to-degree. Wait...iPhone? What?

plam said...

Thing 1 is consistent with my experience (that MSc diploma in my closet didn't help me graduate faster) and Thing 2 I've heard before; it seems to make sense. Thing 4 totally surprises me; most of my advisor's students graduated in about 6 years, and I thought that years-to-degree seemed correlated to one's research area. In computer science, systems students always seem to take longer, while theory students always graduate faster but do postdocs afterwards. Perhaps CS is more fragmented as a discipline than others.

I agree that many other factors affect time-to-degree, though.

Grad Student Hoping to Justify Buying an iPhone said...

Does having an iPhone mean shorter years-to-degree or longer?

Tom said...

I'd suggest doing ordination analysis to see if several factors may be playing a role in conjunction. Your data might work with principle components analysis.

Anonymous said...

Can you please elaborate on whether the correlations you observed were positive or negative for each of these items?

davidjhemmer said...

I definitely see this in mathematics, students who arrive with a masters from another institution do not do any better or graduate any quicker. They still have to pass our qualifying exams and jump through all our hoops. Going from undergrad to a different master's institution to a 3rd institution for the Ph.D. is generally a waste of time.

Anonymous said...

So, does having an iPhone help or hurt? One of the reasons I had been thinking about getting one was for productivity tools from the App Store, e.g. tasks lists and things. Also, it seems like it would make one more responsive to email, which isn't *completely* orthogonal to productivity. ;-)

Kevin said...

Doing any multifactor analysis on such a tiny dataset would be a waste of effort. If there isn't a correlation that jumps out at you in a quick look over the data, there isn't enough data in one university's graduates to find a more subtle relationship.

Some conjectures I have based on my experience:

1) students finish within a year when the funding runs out. Thus guaranteeing 5 years funding leads to 6-year completion. When a PI loses funding, their students all quickly finish and leave (or leave without finishing).

2) Student research ability is not highly correlated with time to degree, and only modestly correlated with completing a PhD. Many of the best researchers do not finish PhDs---sometimes from writer's block and sometimes from being seduced into yet another project.

3) It is hard to get grad students to leave a congenial department with good research going on. It is relatively easy to get them to leave a hostile department with only boring research. So a short time-to-degree usually indicates an inferior department.

On another point, the current economy and stimulus funding has resulted in a huge number of new postdoc positions, but fewer permanent positions. A lot of nearly-finished grad students are hurrying to get one of the postdoc positions. We'll be seeing real problems on the job market in 2-3 years, when the stimulus funds are gone and there are a huge number of postdocs unable to find permanent positions.

Female Science Professor said...

Whether some items in the list have a positive or negative effect on years-to-degree depends on the student and the overall situation. This includes the last item, which was sort of a joke, but only sort of.

Curt F. said...

For particular departments or individuals, it might well be that years-to-degree does not correlate with any obvious factor.

But I don't think FSP's department's recent data give a very revealing picture of the time to complete Ph.D. degrees in the US.

First, I would like to ask FSP how the average time-to-degree has changed over time. Does your department have data for 1970 or 1990?

Second, why does time-to-degree vary so strongly with the field of academic study? I am not so naive to believe that fundamental differences in the modes of inquiry between fields accounts for the difference. That argument would hold more weight if Ph.D. completion was decided inter-departmentally instead of intra-departmentally. In that case, perhaps a case could be made that there is a single, field-independent standard for what constitutes a Ph.D. and that it just takes longer to get to that point in some fields than others. But when biology professor decide how much biology is enough for a Ph.D., I don't think there is any such incentive, and instead, I think the time to degree completion is more likely a function of labor supply and demand for both Ph.D. graduates and graduate students.

Third, isn't it a fault of the system if students who are later more successful in their careers, on average, don't finish their Ph.D.'s sooner? Let me explain this idea.

PIs, departments, and fellowship program administrators tend to follow a quasi-socialist payroll structure - all TAs, RAs, or fellows get paid the same amount. There are no merit-based raises.

I think we can all assume that *on average*, people who are later successful in their careers would also tend to (try to) be more effective in grad school. But if there's no correlation between time-to-degree and effectiveness, isn't the system effectively disincentivizing effectiveness?

A Ph.D. is a significant opportunity cost for most STEM B.S. or M.S. degree holders. If they can finish sooner, it costs them less, after all.

Lastly, what about the effect of writing blog comments on time-to-degree?

Anonymous said...

Dear qaz (7:21 am),

Are you crazy!?! There are so many reasons to obsess about years to degree, I'm not even sure where to start! But, I'll make a quick list:

1. Money. Grad students are paid very low salaries (say $25k). Many Ph.D.s who enter industry get starting salaries of $75-125k. Doing the math, it costs the student about $50-100k per year their graduation is delayed. That's a LOT of money! Not to mention terrible "benefits." I get 2 weeks *unpaid* vacation per year as a PhD student (no paid vacation, period). And terrible health insurance. And no retirement benefits.

2. Sanity. As a student approaching my 8th year, I can describe the pressure I feel to be finished. Every day, it crosses my mind that if anything goes wrong (serious illness for myself or family, serious fallout with my advisor, or just seriously being fed up with working on the same thing for too long), I will not get a Ph.D. I will have worked my tail off for 8 years for very low pay for no degree. Projects and Ph.D.s need a reasonable time to finish, because of the instability of life.

3. Life. Many people put their lives on hold pending graduation. (See FSP's post on maternity leave for PhD's and post docs, and the ensuing comments.) As my own personal example, my husband had to take a job 1000 miles away last October. I am stuck where I am until my advisor allows me to graduate. Or until I get fed up and leave without the degree. After 8 years.

4. Conflict of interest. It is a major conflict of interest for advisors to let cheap-labor students graduate just when the student is at their most productive point (5th, 6th, 7th year). Thus, to prevent student abuse, someone needs to be watching out for advisors keeping students for too long.

Grad student who bought an iPhone yesterday said...

Curt F: In Ecology, it's hard to get a good research project done with less than 3 seasons of data, and sometimes your first season (post classwork) is a dud. Takes a while!

Also, to 'grad student hoping to justify buying an iPhone soon': I hope it makes it shorter! Becuase I'll be much more productive! Though, since my purchase I've been obsessed with downloading apps. Hopefully this trend will diminish over time.

CurlyO said...

Wow. Over here in Europe (at least in my field and country, but I do not think it varies a lot) the average (!) time for a PhD is about three three and half a years. Four years is still well accepted. But if it would take you five years to finish your PhD people may start to wonder whether there is something seriously wrong with you, your advisor, or your karma!

I'm not sure whether I could motivate myself to go through a PhD that is supposed to take eight years...
I wonder whether the proportion of MSc holders that also start a PhD is different in Europe compared to the States?

Kevin said...

"It is a major conflict of interest for advisors to let cheap-labor students graduate just when the student is at their most productive point (5th, 6th, 7th year). Thus, to prevent student abuse, someone needs to be watching out for advisers keeping students for too long."

Yes. This is responsibility of the grad director and of the thesis committee. It is also why grad students should meet with their thesis committee and grad director on an annual basis (or more frequently).

One of the main tasks of the committee at advancement to candidacy is to get the adviser, student, and committee to agree on how much research will constitute finishing the thesis, to prevent both this abuse and the more common problem of grad students wanting to add "just one more" result to their thesis research.

Daniel said...

From the comments:

what's a STEM B.S. or M.S??

S for Science?
T for Technology??
E for Engineering???

Anonymous said...

it's true that my European PhD only lasted 3 years, but: my undergrad was 5 years coursework plus one year research, and most people stay on the same project. I switched, but within the same group.
Anyhow, the total years from high school to PhD were the same than here. The brits have it a bit easier, but all the ones who stayed in academics I know had either two postdocs, or one extremely long one.

Alex said...

I wonder if part of the reason for the apparent lack of relationship between student success and time to degree is that fast completion of a Ph.D. could indicate a student who does good work quickly OR a hands-on advisor who steered the project very smoothly but didn't let the student really develop something on his or her own. Conversely, slow completion could indicate a weak student who struggled in a bad way, or a student who was left alone to struggle in a good way and learn how to come up with questions and solutions on his or her own.

I took a fairly long time, but I was left alone to find my own solutions to problems, and ultimately come up with my own project when nothing else was working. I think it served me well in becoming an independent research. (My postdoc advisor would say that I learned how to be a bit TOO independent.) I wound up in an academic job, which is pretty good considering the odds, but not at a top research university, so I would certainly not be considered a "star" alum.

Doctor Pion said...

STEM = Science, Technology, Engineering, Mathematics.

Shall I assume that "got jobs after graduation" means ACADEMIC jobs? Do you track the ones who do not go on to any form of academia (about 2/3 of all physicists) or just the ones that do? Do you know where all of them are 5 years later?

This is not a minor point, since knowing what skills get used on the job might help your students in the future.

Chad at Uncertain Principles has been posting examples from physics for over a week. They are all tagged PNAS.

Ms.PhD said...

isn't the system effectively disincentivizing effectiveness?

YES. IT IS. !!!!!!!!!!

Also, FSP, I think if you look more closely, you'll see that some PIs do have a record of *always* keeping students AT LEAST 5-7 years (at least where I've worked, no one gets out sooner unless they work for certain PIs).

In this analysis, did you include some normalization for how many students the PI had, total? Because it's kind of apples and oranges to compare a PI who has only had 1 or 2 students ever, to someone who has had several groups of 8 students in the lab at a time.

In fact, I wonder if you analyzed whether lab size correlated at all with student progression?

I also agree with the person who said you didn't discuss whether it was a factor, at what stage the PI was in their career at the time when the student joined. I think you might see some trends here, if you had a large enough data set.

Obviously the iPhone thing is a joke. Your dataset can't include many students who had iPhones. The service plans are too expensive on a grad student salary (they're barely affordable on a postdoc salary!), for one thing, and for another, they haven't been around that long.