Undergrad torn about what path to take...Needs Advice!

[offcenter24]

Hey guys,

I just finished my freshman year at a university ranked around 25 by US News. I have conditional acceptance to my university's medical school as long as I maintain a certain GPA, but I've begun to seriously doubt if being a physician is a right fit for me, since the human body isn't particularly fascinating to me and I'm not exactly Mother Teresa when it comes to caring about curing the sick. My (very Tiger-Mom-like) parents have pushed me to be a doctor my whole life, and I just sort of went along with it until recently.

I have a gut feeling that my true interests and strengths lie in math, statistics in particular (judging mainly from my recreational interest in reading about machine learning and data competitions like the Netflix Prize), and as such I've been looking to graduate programs in statistics as well. Unfortunately, I'm not really sure how "cut out" for getting a statistics Ph.D (I don't think I'd be satisfied with just an M.S.) I am. The profiles on mathematicsgre.com can be pretty intimidating what with their Putnam achievements, REU's, graduate courses, etc. As of now my math background grandly consists of Calc 2 and some competition math in high school (took the AMC and AIME)

All that aside, I guess my main question is this: What can I do to quickly figure out if I have the interest and aptitude to pursue a Ph.D in stats? i.e. join my school's Putnam team, take a real analysis/proof course, go sit on top of a mountain and think about it for ten days, etc....

Any advice from you older, wiser, more enlightened math/stats folks would be greatly appreciated!! Thanks in advance.

[Quantum Buckyball]

Why not just become a MD/PhD? Finish your MD first and then apply to the PhD program afterward, or just do the joint program. I'm gonna finish my PhD first before I apply to medical schools, I like primary care and won't mind being a family physician

Edited May 10, 2012 by SPDFG13579

[jmcq]

Well I never competed in the Putnam and only took a couple of graduate courses (one in Economics and one in Statistics) as an undergraduate. I did have a very strong GPA and strong letters of recommendation. I went to a solid undergraduate state school (not ivy league or anything) and I managed to get fully funded offers to 2 "top 10" PhD programs for Statistics. So you don't have to be a crazy math genius to get in. This might not be the case for a pure math PhD at a top school but I can't really speak to that.

As a Freshman you have plenty of time. My Freshman year I didn't take any math courses at all (I thought I was going to study linguistics).

The courses that were the biggest indicators of aptitude in Statistics at a graduate level I think were: 1) a proof based course on Real analysis and 2) an upper undergrad/grad course on mathematical statistics (covering things through Cramer-Rao bounds, Maximum likelhiood estimators etc). If you feel comfortable in those courses then I think you'll be fine. As a general rule, the more math you take the easier time you'll have. It's easy to pick up the statistics, it's harder to pick up the math. Although taking courses on Abstract Algebra and Topology while helping you with mathematics, probably won't be very useful as a Stats PhD. Focus on Real Analysis, Calculus, and Linear Algebra.

As far as interest, I mean if you're already looking at things like the Netflix competition and think that's cool, that probably says enough (Machine learning is also my area of interest). Read some papers in the area (don't expect to get the math but the intro and conclusions usually gives you a general idea of what they are trying to do). Other than that take some statistics classes and if you find yourself thinking and talking about statistics even when you don't have to I think that's a great indicator.

Feel free to PM me if you have any specific questions that I might be able to help with.

[wine in coffee cups]

Several points:

• First, forget about Putnam, AMC/AIME, etc. I realize that's how math ability is measured before college and something competitive types get fixated on, but those exams test the narrow skill of getting to clever insights in a very short period of time. That's it. Not that stat departments wouldn't look warmly on high Putnam scorers, but the type of thinking and abilities involved there are not particularly relevant to statistics research (or even the ability to succeed in upper-level math classes) and won't give you much indication if you'd do well the field or not.
  
• Most of the profiles on mathematicsgre.com are for pure or applied math. I think there were maybe a dozen-odd statistics profiles up this year and I don't remember those being overwhelmingly Putnam-y, REU-y, or grad school course-y. Admissions are not as competitive as pure math admissions and stats programs seem more welcoming of different kinds of experiences.
  
• Best way for you to figure out if you actually statistics and have the aptitude is to take more statistics, math, and CS courses and get A's. Be sure to take a calculus-based probability class and a theoretical undergraduate statistics course. Hit up multivariable calculus and linear algebra on the way. If you aren't underwater and like these classes, that's a good sign. If you still aren't underwater by the time you get to proof-based analysis courses, that's a great sign.
  
• Consider changing to a math or statistics major. (It's not like you're trying to switch to medieval gender studies in art or whatever, so I would think tiger parents could be down with math.) Not sure about your conditional MD admit arrangement, but typically med schools don't require a particular major anyway. If your parents won't let you close the med school door just yet, I suggest majoring in math and taking just the bare minimum bio/chem you need to satisfy the med school's entry requirements.
  
• Programming is a big part of day-to-day work for statisticians, so take an intro to programming class (and more advanced CS as it interests you). You should also mess around with R on your own. Eventually you'll need to get comfortable with control structures/logic/looping, debugging algorithms, processing raw data, etc. Not the most fun part of statistics, but if you really struggle to understand these concepts or despise coding with ever fiber of your being, not a great sign. Also, note that professors in all kinds of departments sometimes take on undergrad students to help clean and analyze data. Barring finding a statistics research project or REU, look out for those opportunities as they will be good experiences whether you apply to grad school right away or look for a job first.
  
• You're already doing some of this by reading about ML on your own, but try following blogs related to statistics, probability, computing, visualization, or areas of application that fascinate you and see if those sustain your interest.
  

[cyberwulf]

Lots of good advice on this thread already. OP, if you maintain some interest in medical science but think you're more quantitatively oriented, you might also want to look into programs in biostatistics. The top places offer similar theoretical foundation to good stat departments, but the admissions requirements tend to be a bit lighter on the math (Calc 1-3 and Linear Algebra are usual pre-requisites).

[TakeruK]

You still have a lot of time to decide, even though it might not feel like it.

If mathematicsgre.com is anything like physicsgre.com (and I think it is!), then the people who post their profiles there are NOT a fair sample of graduate school applicants. It's mostly the high ranking students that know about and choose to post -- some people are intimidated by the scary profiles and choose to only read the threads and not post. If I had thought everyone who posted on physicsgre.com was a representative sample of applicants, I'd never apply anywhere!

Secondly, I'm not sure how your med school works, but usually you can get into med school with any undergraduate degree, as long as you take some list of pre-requisites. At my undergrad, this list was not so extensive that you could not pursue your own major as well -- you would just have to use most (if not all) of your electives on med school related courses. Plenty of my friends finished their BSc in Physics and then applied to medical school. It might be a bit more work but you can probably still apply to BOTH MDs and PhDs programs in your 4th year if you still haven't decided yet. But basically, I'm saying that you keep pursuing both paths for now and see how you feel in a couple of years. I know that I didn't really realise that I wanted to do grad schol until my 3rd year -- at this point, I had to take a couple of extra courses to catch up to the "grad school track" but if I didn't keep my options open in earlier years, it could have meant a lot of extra work.

Finally, I really really want to say, don't give in to family pressures when pursuing your education!! I had felt some pressure from my own parents (they didn't understand why I would want to do a PhD, and especially not in astronomy, where there is no real job market other than academia -- they were really pushing me towards medical physics for awhile) but by the time I actually graduated, they were very understanding and I'm grateful for that. So, keeping both PhD and med school options open might give your parents some time to realise that you are your own person with your own interests and they might back off on the med school pressure.

I don't think there is any way for you to immediately know what your PhD interests, if any, would be, and you have many more years of college to go through. I think keeping as many options open as possible will give you more choices down the road. But it's also not like you have to decide your future forever either -- even if you change your mind later, you might still be able to switch paths (it might cost you a year of coursework, but it could be worth it when you finally know what your passion is).