Igotnothin

I think TakeruK's advice is big-picture, on PhD training in general. I have a few comments regarding stats/biostats PhD programs in particular: 1. Coursework is not a small component of PhD programs. You will learn a ton about statistical theory, proving that a certain estimator has good properties, etc. 2. I'd say less than one-third of stats/biostats PhDs pursue academic research after graduating. Many go into industry (biotech, pharmaceuticals, start-ups, tech), government (NIH, CDC), applied work (e.g. as an epidemiologist/data analyst), consulting, etc. 3. You mention that you like "using math to solve problems, not for the sake of the subject itself." This philosophy is perfectly compatible with biostatistics PhD and I'd say statistics PhD for the most part. Math PhD programs seem to take pride in the study of math whether it's useful or not. In biostatistics, virtually every paper relates to a real problem that arises in clinical or epidemiological studies. From what you've written I think stats/biostats PhD could be a nice fit.

I feel for anyone who spends ~6 years getting a PhD then ends up having to work ridiculously hard for low pay as a postdoc for 5, 10 years, even longer. Particularly if TT positions were way more common when you went into grad school than when you got out. But in general I think the problem is that applicants aren't looking at the employment numbers. They're excited and flattered to get in to a PhD program, and they enroll without really considering job prospects. Not as much as they should, anyway. If you pursue a degree that isn't in high demand, you can only blame "the system" so much when you graduate and can't find a job.

It's just one school deciding to balance their budget differently - got some extra money and put it into arts and humanities. Sounds like it's mainly a strategic move, as opposed to a conscious effort to improve the lives of PhD students in those fields.

Interesting but I don't see how this is any different than when any other school raises their stipends every few years. You have some extra cash and want to compete with Stanford, so you go up to $29k. Nothing groundbreaking in my opinion.

Lots of interesting discussion here. My take on some of the issues raised: 1. I put a ton of weight on stipend when deciding on a PhD program. I'm very happy at a mid-ranked department that pays well. I'm working with an awesome adviser and maximizing my annual Roth IRA contributions. 2. It'd be nice to raise stipends from $25k to $50k to accommodate folks with kids, debt, etc. But I doubt that is in the realm of possibility at 99.9% of schools. 3. If stipends did go from $25k to $50k, competition would get fierce due to higher demand and presumably fewer spots. I might not be in a PhD program if stipends were that high, because acceptance rates would be so low. 4. I think there is some merit to victorydance's much-hated comment about family planning. If I see $25k a year in my 5-year future, I'm not going to start a family. On the other hand, if I'm set on starting a family, I'm not going for $25k a year. In that sense the low pay probably does keep certain folks out of academia. 5. I don't feel bad for PhD students and their small stipends. In fact I take exception to students complaining about being poor, and even taking advantage of low-income housing meant for those who really need it. At risk of subjecting myself to downvotes, let's be honest here: we as a group (PhD students) are mostly from high SES backgrounds, and we're heading for the high income brackets after getting our degrees. We're doing all right.

The idea that you're not going to be taken seriously as a researcher and will be less likely to get grants because you don't have a rehearsed 30-second elevator pitch is crazy. How on earth would this affect your chances of getting grant money? If you're a normal person, you can briefly tell people what you do when they ask.

The person you were talking to really liked it?

The moment you decide to prepare a 30-second "elevator speech" to tell non-academics what you do, you're already overthinking it. If someone asks you what you do, tell them. Don't recite a 30-second summary that you wrote up and rehearsed.

I think many of the suggestions on this thread will actually alienate non-academics. Bringing up an academic issue you read about in The New Yorker... Giving a 30-second "dumbed down" summary of your research that insults their intelligence... Humblebragging about where you go to school... I met a Hopkins med student in DC one time, and he said he was in med school. Someone asked him where, and he said "in Baltimore." At that point you're going so far out of your way to NOT sound braggy that it ends up sounding more braggy than it would have if you just said "I'm a med student at Johns Hopkins." Because we know you're not saying Hopkins because of the impressed reaction you know you're going to get. Which means you are overly aware that you're in a prestigious academic program.

@trigga congrats on your acceptances. Tough call indeed. Your thought process seems really clear... My $.02 is that it might be worth taking the extra debt for a Harvard Master's n biostats. Having an MS or MPH from the #1 biostats department in the country is something that I think will give you a real boost in terms of job prospects.

The Master's degree might be a good investment even if you have to take loans. Tuition might not be too bad at public schools like Michigan, UNC, Minnesota, etc. Not really sure though. Michigan has one of the best Master's programs in biostats and they have quite a few funding opportunities, e.g. work as a graduate instructor or research assistant. Both of those apparently pay full tuition and a stipend in exchange for 20 hours per week. Again I'm not really familiar with Michigan's program but it might be worth a shot. Here's the page which has info on funding: http://www.sph.umich.edu/biostat/programs/masters.html

I think I agree with hausinthehouse. A biostats MS/MPH/MSPH at a top 10 program would be a really good choice. Gives you a chance to develop strong math/stats skills and if you perform well it will help you a lot towards a PhD admission. Even if it doesn't help you get in to a PhD program, a Master's in biostats is a very useful degree that could allow you to get many good jobs in industry. Best of luck whatever you decide.

I agree with the others. I think it will be difficult to convince a top-10 biostat admissions committee that the math course is equivalent to a calc sequence and linear algebra. Even then you would be at a disadvantage vs. applicants that took real analysis (which is not required, but is a plus). I had a relatively weak math background when I applied to biostats PhD and most folks recommend I apply to epi programs instead. It really bothered me because I was more interested in biostats and the job opportunities are better. I'm glad I went for biostats (got in to a 5-10 ranked program) but looking back epi PhD is also a great option, and it was more compatible with my math background at the time. I'm sure you've already considered it, but just wanted to mention it. In my opinion if you really want to do a biostats PhD you will probably have to take one or several math courses first. Maybe Calc III and linear algebra. Might be able to do them online. If you get those math courses in, I think you could potentially be competitive for a program outside the top 5, considering your public health experience and hopefully one or two papers (and hopefully a good undergrad GPA). Might end up having schools admit you to their MS program instead, that happens pretty often.

Well it is an interesting issue and I admit I can see it possibly going either way. I definitely like the idea of putting in your 8 hours and then bouncing and not doing any work in the evenings or on the weekends. But it's hard to believe that would make me more productive. If I work for 2 or 3 hours on a Saturday afternoon, I usually make some good progress. I can't remember a time I put in 3 hours on a Saturday and made so many mistakes that I actually regressed on my work.

In the "Conclusions" section of that paper: The literature on scheduled overtime was found to be very sparse; dated to the late 1960s and earlier; based on small sample sizes; and largely developed from questionable or unknown sources. Although there appears to be a number of data sources, this is an illusion because many of the articles and publications quote other sources while providing no new data or insight. Where the data source is known, other pertinent information, such as the environmental and site conditions, quality of management and supervision, and labor situation, is unknown. The various graphs and data that have been published are inherently unreliable, except perhaps to suggest an upper bound on the losses of efficiency that might be expected. The literature offers no guidance as to what circumstances may lead to losses of efficiency. With respect to the loss of efficiency as a function of the number of hours per day and the number of days per week, the literature provides strange and largely unbelievable results.

Do you have a reference for a peer-reviewed experimental study that used randomization and blinding to look at this issue (and is written in English)? Your link is to a summary of 100-year old studies written in German.

Highly doubt this is true. Randomize 100 workers to work 60 hours for a week, and another 100 to work 40 hours, and compare any measure of performance. Pretty sure the 60 hours/week group gets more done. Not that I'd want to work that much.

Congrats and good luck!

Ah okay missed that part.

I think even if you give GW the benefit of the doubt that funding with the 35 hour/week job will go through, UMD is still the better option. I don't see how you can work 35 hours a week while in a full-time PhD program.

There are a lot of considerations and I think you've done a great job outlining them. In my opinion UMD is the right choice. Like you wrote, it's not ideal to be married to a 35 hour/week job to finance your funding. You would be extremely busy when you consider classes, homework, and 35 hours per week of work. On the other hand, your funding at UMD is guaranteed with few strings attached. Much more flexible and in my opinion more likely that you will be happy there. Best of luck! (By the way I am in a biostatistics PhD at a mid-ranked program)

It's going to be awkward but I think it's possible. Path of least resistance in my opinion is to portray it as "I am now really interested in this topic (e.g. machine learning) and schools X, Y, and Z specialize in it." You have to figure out what the topic is.... And it's a bit of a fib... But it gives you an explanation. Actually to be honest I'm not sure I recommend this. But I'll post it anyway as an idea.

MS thesis acceptance rates are near 100% at most programs I'm familiar with. That's what I mean by "basically guaranteed."

Peer-reviewed paper's always better in my opinion. When you start a MS program with a thesis component you're basically guaranteed to have your thesis "accepted" by your committee. It's way more impressive to submit a journal article that 2 or 3 experts in the field deem an important contribution to the field.

I think I've heard of this journal. Wouldn't recommend submitting to it though. Better to submit null results to a journal in your field than a generic "negative results journal." I also wanted to point out a few statistical issues that are relevant. First, note that you really mean "fail to reject the null" rather than "prove the null." There is a class of study designs known as equivalence testing or non-inferiority testing which allow you to prove that two or more groups have equal means, but I doubt this is what we're dealing with here. Second, whenever you do a study and find no association you need to talk about statistical power. Ideally the study was powered to detect some pre-specified effect with 80% or 90% probability. But if it wasn't designed carefully or if it was a secondary data analysis then it could be drastically underpowered. That would make your result less interesting and harder to publish because you just didn't have the sample size to answer the question adequately. Definitely a good idea to look into this so you can defend your findings when reviewers or committee members bring it up.