MS Options for Biostatistics PhD

[efh0888]

Thoughts?

1. Statistics - focus on data mining but includes a stat theory sequence (at the level of Casella and Berger I believe)
2. Biostatistics - seems light in terms of theory but gives exposure to public health
3. Math - not applied, mostly theory but includes some computational classes, and has a concentration in where you take a sequence measure theoretic probability and mathematical statistics

 

 

[Residuals]

Math might be the best choice for a straight-up statistics department but I have heard that biostats Ph.D. programs don't see that much math as necessary (namely, beyond the UG sequences of linear algebra, analysis, multivariate calc, etc.). I think, however, an MS in biostats is a swing to the other direction of being not a sufficient foundation. I, personally, think an MS in stats is the sweet spot choice.

[efh0888]

@arima thanks, you make some good points. One thing I forgot to mention – of the ones I've been looking at, i.e. in-state options for financial reasons, the math degree allows for a thesis whereas the other 2 do not. I have a full-time job I love, and again for financial considerations, would almost surely do an MS part-time making it difficult to get research experience otherwise. How important do you think research experience is for biostat phd admissions? Is that a major factor to consider between the degrees?

Also, what about factoring in schools like FSU and NC State where biostat is part of the stat department? Would they look down on a stat MS?

FWIW in undergrad took linear algebra, calc sequence, diff eq, an intro to proofs type of course, stat theory 1 & 2, and a lot of stat methods courses. Also took grad level stochastic processes and advanced calc while in my first master's. Advanced calc was only 1 semester and was really tough but imagine it was light compared to a real analysis course. So I think I've got the chops for either math or stats. 

Also, not sure if it would matter for the choice here but my goal after the PhD would be industry.

Edited May 28, 2016 by efh0888

[Residuals]

I am not convinced that you will find many biostats PhD students with an MS in pure math. Actually, even with what I said of swinging too much to the left, I think a MS in biostats is better than pure math particularly at dedicated biostats PhD programs. Look at the backgrounds at top-ranked Hopkins:

http://www.jhsph.edu/departments/biostatistics/directory/students/phd.html

Mostly, UG in math or stats, then a smattering of MS biostats, MS stats, and MS applied math (vs. pure math). I think MS in pure math is overkill for biostats. I think also there is a subtle risk that an applicant with MS pure math to biostats might be viewed as a person who is applying also to stats departments and his/her application to biostats is a back up choice.

Edited May 28, 2016 by arima

[efh0888]

That's encouraging actually. I took a look at some student profiles and you're absolutely right.

In fact what draws me towards biostats is the focus on applications and methods over theory (while still taking a rigorous approach), so I'd much rather go for an MS in stats or biostats instead of pure math. Additionally, I'm sure there are more opportunities to work with truly big data sets and to get exposure to areas like machine learning.

Edited May 28, 2016 by efh0888

[Statapplicant2017]

On 5/27/2016 at 7:03 PM, arima said:

Math might be the best choice for a straight-up statistics department but I have heard that biostats Ph.D. programs don't see that much math as necessary (namely, beyond the UG sequences of linear algebra, analysis, multivariate calc, etc.). I think, however, an MS in biostats is a swing to the other direction of being not a sufficient foundation. I, personally, think an MS in stats is the sweet spot choice.

Alternatively for someone interested in doing applied statistics work but wanting to leave the door open to return to a PhD in the future would an MS in stats be the sweet spot between the 3? (Stats,Biostats,Applied Math)

[compscian]

I'll add a different perspective: it doesn't really matter what your MS is in. Any good PhD program worth their salt will value relevant research experience more than coursework. If your ultimate goal is a PhD, I'd suggest you pick a program that is very light on requirements and will give credits for research work, even if thesis option is unavailable.

Personally, I did an undergrad in MechE/AMath and wanted to do a PhD in ML/data science. I had little coursework apart from the standard math sequence: CS (zero outside ML/AI); EE (only optimization and signal processing; zero in circuits); Statistics (no dept in my univ). I spent a year with an EECS professor doing research and published a paper, and didn't have trouble getting admitted to all three departments above. Coursework doesn't matter.

[efh0888]

Bump...

I've been looking into this again more recently.  Rob Weiss at UCLA suggests focusing on math (https://faculty.biostat.ucla.edu/robweiss/preparation) which is what I'll probably do.  But like @compscian said, I agree that gaining research skills is ultimately more important than coursework.  I have a family so really need to keep working full-time as long as I can, so I don't have many options to get both the math and the research.  So I've found a synchronous/online MS program in Mathematical Sciences housed in a combined Math/Stats department that actually looks very rigorous.  It is light on course requirements, only requiring 2 core classes in matrix theory and mathematical statistics, and the rest is pretty much open (just needs advisor approval).  They offer a bunch of stuff online on top of the core including real analysis, complex analysis, topology, numerical analysis, advanced probability and inference, statistical modeling,  regression, DOE, etc.  The best part is they also offer a 2-semester research course as well as a thesis option.  Now for the big downside -- it's from a no-name state school.  

What are your thoughts on how much the prestige of the school would matter in admissions?  For my situation, I'm hoping that adding the graduate math coursework plus relevant research experience from any school will put me in a solid position.  But hey, I figure an "ensemble" of diverse opinions will perform better on average than just one, right guys?  

[efh0888]

Also love hearing from the experts, so if @cyberwulf and/or @biostat_prof would be willing to chime in that would be wonderful.