Statistics PhD / MS: Profile Evaluation (Please)

[aridneptune]

Hello all - I'm new to this forum. I'm looking to apply to Statistics MS or PhD programs for fall 2014 (prefer PhD but will apply for & do MS as a stepping-stone if necessary). I'd like some candid feedback on my chances. To preface, my main weaknesses are (1) I've been out of school for 3 years; (2) limited statistics coursework (at least directly). Anyway, here we go:
 

Undergrad: Majored in mathematics and economics (double major). 3.99 GPA at UNC. Math GPA 4.00.

 

Undergrad coursework: Probability theory, real analysis, topology, calculus series, linear algebra, differential equations, econometrics, economic statistics (e.g. applied statistics).

 

GRE: Taking it at the end of September. I'd think ~162-167 quantitative, 160-165 verbal.

 

Research Experience: Worked on several projects in economics and political science. Research intern for State Department.

 

Recommendations: I expect strong recommendations from a math professor and several economics professors. I'm not sure how well-known they are.

 

Professional Experience: I've worked on Wall Street (trader) for three years after graduation (graduated in 2010).

 

Goal: MS or PhD (preferable) at Cornell, UNC, Berkley, Michigan, University of Washington. I'm trying to keep the schools in a reasonable range (e.g. I don't think I can crack the very top schools).

 

Short Statement: I originally majored in math as a means to pursue a PhD in Economics, but followed the siren song of money and went into the financial industry after graduation. Long story short: it was a mistake. I've discovered (the long and painful way) that money is not as central to my personal utility function as I thought. Lately I have become disillusioned with Economics and want to pursue an academic career in statistics. I hugely enjoyed my undergrad courses in econometrics and probability, and am also very much enjoying the Johns Hopkins mathematical statistics online course series. I'm attracted to data analysis and feel that statistics is both interesting and useful in a world in which data are increasingly ubiquitous.

 

Now, how good might my chances be for an MS (or PhD) at these institutions? Will my 3-year hiatus from academia hurt me? If I'm being honest with myself my top choice is UNC: Chapel Hill is a phenomenal town and I'd love to return there for a spell. I'm also considering applying to the newly-minted crop of MS Analytics programs (namely NCSU). Any thoughts on those?

 

I really appreciate any input anyone might have. Thanks in advance.

[Igotnothin]

Man, we have some well-qualified applicants on these forums. I'm in biostatistics, but from what I can tell your profile looks great and you will have no problem getting into a PhD program in statistics.

[DMX]

I think the above poster is a little bit too optimistic. I had a similar profile to you (similar major, similar GPA, slightly stronger school, worked at a bank for 3 years, had a Stats Master's too) but failed to crack the top 15 PhD stats programs.

 

I think you have a small shot at UNC and (maybe) Michigan PhD programs.

 

For Master's though, I think you will be competitive at virtually all stats programs.

Edited August 14, 2013 by DMX

[timmmythetooth]

You may be underestimating the strength of Berkeley and Washington if you don't consider them top schools; check out the USNEWS rankings.  I think they are decently representative of statisticians' views, though rankings only go so far.  However, your profile looks pretty strong to me.  Statistics programs often don't have many prerequisite Statistics courses, and your strong math background is more important.  The only omission that may be important is the lack of a formal Mathematical Statistics course, but I wouldn't worry too much about that, especially since you're studying it on your own anyway.

 

Most programs do want you to have some knowledge of programming, though (the more the better), and I don't see any mention of that.

 

The hiatus won't necessarily hurt you, I think.  PhD programs are very much like a tough job with low pay, and the increased maturity one can gain from working in industry for a bit can be a plus.  However, it is very important that you frame your motivations properly in the statement of purpose.  You don't want to come off as simply bored with your career, and make sure you highlight a desire to develop new methods and do research, rather than practice data analysis on its own.  Also, statements about how statistics is increasingly relevant in a world with increasingly ubiquitous data are about as ubiquitous as the data are.  I would guess that admissions committees read something similar in many of the SOP's they read, so I'd recommend avoiding saying such things directly (even though it's a totally valid statement).

 

And finally, that was all geared towards PhD admissions, but if you were instead interested in MA programs I think you would be quite competitive everywhere.

Edited August 14, 2013 by timmmythetooth

[aridneptune]

Thanks very much for the advice. My programming experience is indeed limited (basically STATA and SAS for empirical research). I meant Berkley / Washington MS programs above (excuse the confusion there). I also appreciate the SOP advice (and the pun) - do you think that the 'SOP-is-of-limited-importance-in-statistics' advice in the sticky on these forums is relevant?

 

Perhaps the best course of action would be to get an MS, prove my bona fides, and then move onto a PhD? I'm of mixed feelings about this. On the one hand, it's a greater financial burden and two years of opportunity cost. On the other, it's a shorter degree (I have the option to go terminal), is perhaps more marketable in the private sector and will give me time to get up to snuff. What do you guys & gals think?

[DMX]

Thanks very much for the advice. My programming experience is indeed limited (basically STATA and SAS for empirical research). I meant Berkley / Washington MS programs above (excuse the confusion there). I also appreciate the SOP advice (and the pun) - do you think that the 'SOP-is-of-limited-importance-in-statistics' advice in the sticky on these forums is relevant?

 

Perhaps the best course of action would be to get an MS, prove my bona fides, and then move onto a PhD? I'm of mixed feelings about this. On the one hand, it's a greater financial burden and two years of opportunity cost. On the other, it's a shorter degree (I have the option to go terminal), is perhaps more marketable in the private sector and will give me time to get up to snuff. What do you guys & gals think?

 

I think comleting an MS is a sound strategy (hopefully you have some money saved up from your employment? If you're in NYC there are decent part-time options at Columbia and NYU--hit me up if you're interested).

 

Also, think about what your end goals are. If you want to do develop new statistical methodologies and stay in academia, PhD is probably right. If you want to be a practioner and work in industry (which is the sense I get from your short statement), an MS may be just fine.

 

You can use the MS degree to see where you lie on this spectrum of applied vs theory--take some high level theory classes (i.e. beyond your usual probability/inference classes), and also some practical classes (most MS programs will have a 'Data Analysis' type course. Take them seriously).

Edited August 14, 2013 by DMX

[Igotnothin]

I think the above poster is a little bit too optimistic. I had a similar profile to you (similar major, similar GPA, slightly stronger school, worked at a bank for 3 years, had a Stats Master's too) but failed to crack the top 15 PhD stats programs.

 

I think you have a small shot at UNC and (maybe) Michigan PhD programs.

 

For Master's though, I think you will be competitive at virtually all stats programs.

 

Interesting. Are you saying you got accepted to programs, just not in the top 15?

 

I strongly recommend not going for a master's first. There are faster and cheaper ways of getting into PhD programs. Unless you're dead set on getting in to a top-tier program, save your $50-100k and 2 years and apply for a PhD now. Just make sure you apply widely and put together a solid application. Good luck.

[cyberwulf]

I think the above poster is a little bit too optimistic. I had a similar profile to you (similar major, similar GPA, slightly stronger school, worked at a bank for 3 years, had a Stats Master's too) but failed to crack the top 15 PhD stats programs.

 

I think you have a small shot at UNC and (maybe) Michigan PhD programs.

 

For Master's though, I think you will be competitive at virtually all stats programs.

 

DMX, your results last year were pretty surprising to me and not representative of what happens to the typical student with a record like yours (and the OP's). Have you contacted departments to get a sense of why you didn't get in? The only plausible explanation (assuming you're not exaggerating your academic credentials) seems to be a negative letter or something similarly disastrous (e.g., awful TOEFL score if you're a foreign student).

 

A 4.0 math/econ double major from one of the best state universities in the country has the pedigree to be competitive ("in the discussion") for admission at every PhD program in the country. Whether or not such a student cracks the top few departments will depend mostly on the strength of their letters of recommendation. 

 

So, OP, I think that Cornell/UNC/Berkeley/Michigan/UW is a good start to a list of places to apply for *PhD* admission! You might want to add some more "local" places like NC State and Duke. Your "safeties" should probably be the Masters programs at these schools, where you're essentially guaranteed to be admitted.

[wine in coffee cups]

Thanks very much for the advice. My programming experience is indeed limited (basically STATA and SAS for empirical research). I meant Berkley / Washington MS programs above (excuse the confusion there). I also appreciate the SOP advice (and the pun) - do you think that the 'SOP-is-of-limited-importance-in-statistics' advice in the sticky on these forums is relevant?

 

Perhaps the best course of action would be to get an MS, prove my bona fides, and then move onto a PhD? I'm of mixed feelings about this. On the one hand, it's a greater financial burden and two years of opportunity cost. On the other, it's a shorter degree (I have the option to go terminal), is perhaps more marketable in the private sector and will give me time to get up to snuff. What do you guys & gals think?

nononononono go right for the PhD. You don't have to prove yourself in an MS first with your credentials. A US citizen (right?) who studied math with a pretty much perfect GPA from UNC should be competitive at top PhD programs. (I would gently point out that DMX above is an international applicant and so his/her experience might be better thought of as a lower bound on how a similarly credentialed American might do -- acceptance rates are way higher for citizens/PRs.)

 

I respectfully disagree with cyberwulf's advice to not worry much about the SOP in the stickied topic, at least wrt to the most selective statistics departments (like the ones you are interested in). Granted, I toooootally believe him in the context of most biostat applications, where some very good programs like UNC and UMN still accept so someone with good recs, high grades, and enough math is almost always going to be to be on the right side of their dividing line.

 

But relative to most biostat programs or lesser known statistics departments, the top statistics departments have a much larger volume of applicants, and I have to think that a much larger proportion of them have strong math and statistics backgrounds (as opposed to maybe more coming from a bio-related track and meeting math requirements minimally). So though I think you have a decent shot at top stats departments based on sheer numbers alone, nonetheless consider that many of the best colleges/universities in the US are going to have one or more math majors with very high GPAs applying when you are, and they're more or less each applying to a subset of the same ~20 statistics programs plus Harvard, UW, JHU biostat. That is, functionally there are several dozen aridneptune doppelgangers out there (plus other strong applicants from around the world) and a large proportion of them are competing for spots with you at each place you apply. Any program that doesn't accept a majority of such students is going to be iffy for you.

 

Thus I think if you want to improve your chances at your top choices, you should do whatever you can to state your case about being qualified and a good fit within each department. This goes triply extra for any smaller programs you might end up applying to with fewer areas of active research (e.g. Harvard, Yale, Northwestern). Also, remember that the SOP is the only piece of your application you have complete control of at this point: your transcript and test scores are what they are, your recommenders will say whatever they want. (Though they might be more likely to say things you want them to say and reinforce your positive characteristics if they can look at your SOP draft when they write it!) So whatever your perceived weaknesses are, you have the opportunity to frame them however you like and draw attention to your strengths. Let's also not forget that it's pretty stupid to pay $100 per school in application fees and GRE score sending and not give every piece of your application your all.

 

So yeah, I say do worry about the SOP, really put your back into it. It paid off for me -- I was told as much by several of the schools I got into. Try not to have your app come off too strongly as "why I don't want want to work in finance anymore". Work on conveying the "why I do want to study statistics" positive message in a personal and authentic way. Let there be no doubts that you should be in a statistics department instead of an economics department or a PhD track in a business school. Being out of school shouldn't hurt you, but you need to make them believe you belong there. For starters, you'll want to mention your self-study in statistics and write about any interesting statistics-related projects you've worked on (as a student or at your job) that led you to want to do research in statistics.

 

Good luck etc.

[DMX]

DMX, your results last year were pretty surprising to me and not representative of what happens to the typical student with a record like yours (and the OP's). Have you contacted departments to get a sense of why you didn't get in? The only plausible explanation (assuming you're not exaggerating your academic credentials) seems to be a negative letter or something similarly disastrous (e.g., awful TOEFL score if you're a foreign student).

 

A 4.0 math/econ double major from one of the best state universities in the country has the pedigree to be competitive ("in the discussion") for admission at every PhD program in the country. Whether or not such a student cracks the top few departments will depend mostly on the strength of their letters of recommendation. 

 

So, OP, I think that Cornell/UNC/Berkeley/Michigan/UW is a good start to a list of places to apply for *PhD* admission! You might want to add some more "local" places like NC State and Duke. Your "safeties" should probably be the Masters programs at these schools, where you're essentially guaranteed to be admitted.

 

Interesting... I think being an international certainly hurt me somewhat, and my letters were your run-of-the-mill "did well in class" type letters. I was able to talk to a couple of people from programs from which I was rejected, and from their description it seemed like the bar was far higher than I had initially thought (especially at Stats programs).

Edited August 15, 2013 by DMX

[cyberwulf]

 

I respectfully disagree with cyberwulf's advice to not worry much about the SOP in the stickied topic, at least wrt to the most selective statistics departments (like the ones you are interested in). Granted, I toooootally believe him in the context of most biostat applications, where some very good programs like UNC and UMN still accept so someone with good recs, high grades, and enough math is almost always going to be to be on the right side of their dividing line.

 

But relative to most biostat programs or lesser known statistics departments, the top statistics departments have a much larger volume of applicants, and I have to think that a much larger proportion of them have strong math and statistics backgrounds (as opposed to maybe more coming from a bio-related track and meeting math requirements minimally). So though I think you have a decent shot at top stats departments based on sheer numbers alone, nonetheless consider that many of the best colleges/universities in the US are going to have one or more math majors with very high GPAs applying when you are, and they're more or less each applying to a subset of the same ~20 statistics programs plus Harvard, UW, JHU biostat. That is, functionally there are several dozen aridneptune doppelgangers out there (plus other strong applicants from around the world) and a large proportion of them are competing for spots with you at each place you apply. Any program that doesn't accept a majority of such students is going to be iffy for you.

 

 

I'm going to quibble with your numbers here: If there are indeed "dozens" of applicants with perfect GPAs from top 50 U.S. schools applying to stats departments, then it would presumably follow that these folks would fill up the top 5-6 departments making it virtually impossible to get in without a perfect academic record. This just isn't what happens, so the number of domestic applicants with records similar to the OP's must be lower, perhaps 10-15 people. Many (probably a majority) of the best math majors at elite schools are applying to *math* PhD programs, with the fraction increasing with school prestige (e.g., a near-4.0 in math from Princeton is probably being strongly encouraged to do a PhD in math because they will have their pick of where to study, whereas a math student with similar GPA from University of Virginia might view stats as a more viable option).  Some students presumably go the econ route. So stat only sees a fraction of the top students, perhaps 20-30%. 

 

I can believe that there might be "dozens" (perhaps 25-40) of applicants to stats programs with near-perfect GPAs, but the majority of these will be from schools much less prestigious than UNC.

 

 

Thus I think if you want to improve your chances at your top choices, you should do whatever you can to state your case about being qualified and a good fit within each department. This goes triply extra for any smaller programs you might end up applying to with fewer areas of active research (e.g. Harvard, Yale, Northwestern). Also, remember that the SOP is the only piece of your application you have complete control of at this point: your transcript and test scores are what they are, your recommenders will say whatever they want. (Though they might be more likely to say things you want them to say and reinforce your positive characteristics if they can look at your SOP draft when they write it!) So whatever your perceived weaknesses are, you have the opportunity to frame them however you like and draw attention to your strengths. Let's also not forget that it's pretty stupid to pay $100 per school in application fees and GRE score sending and not give every piece of your application your all.

 

So yeah, I say do worry about the SOP, really put your back into it. It paid off for me -- I was told as much by several of the schools I got into. 

 

Two things:

 

1) For the OP, I think the SOP actually may play a slightly bigger role than for the typical applicant because of the time spent away from school. It is helpful in such cases to explain what has motivated your change in career trajectory. It may also be an opportunity to say what skills you gained during your time in the "real world" (e.g., applied statistics). My general point in the sticky was that most applicants straight out of undergrad have nothing interesting/useful to say because they have little exposure to statistical research, so they shouldn't worry too much about what they write. And even for the OP, I still contend that the statement will play a minor role, at best, in determining where you get in. Letters will be much more important.

 

2) Faculty love to think (and say!) that "all our applicants are uniformly superb, so the only way to distinguish between them is the personal statement." But that is just chest-thumping fiction; GPAs, test scores, and letters of recommendation still allow a ranking of candidates that most would agree on. wicc, I don't doubt that faculty you talked to remembered and liked your SOP, but I question whether it really made the difference between acceptance and rejection or whether it merely made your application stand out among the group that was going to be accepted anyway. 

[wine in coffee cups]

cyberwulf, I'm not sure I would hold up a perfect GPA as the standard for a stats applicant to be considered uber-competitive and in the peer group with aridneptune, but let's say something like >3.9 within math (with lots of courses), and not much lower overall from the kinds of schools you've heard of. Without weird weightings, that essentially means over 3/4 of math grades are A's and the remainder are A-'s, which sounds like a reasonable peer group to me. Given that, I'm quibbling with your quibble (instead of working on my research ) and think you're underestimating how many stats applicants are out there with comparable profiles.

 

I had a job some years back analyzing data for a consortium of 30-odd highly ranked private colleges, more or less the same ones that are vastly overrepresented among baccalaureate institutions in PhD programs. One of the fun facts I learned was that at pretty much all of these schools, math majors disproportionately had very high GPAs. My undergrad is a typical example: over one-third of math majors in my year graduated in the top 5% of students (>3.9 overall). Nearly all the rest were still in the top 25% of students (>3.7 overall, almost certainly higher in major). When I used to do resume review and interviewing in my old job, too, we looked for econ or math majors with high grades in statistics/econometrics classes mostly from fancy schmancy schools like Dartmouth and Wesleyan. There, too, I encountered a surprising number of students with very strong math backgrounds and grades. 3.9s in math, apparently not rare at all and not limited to those with math or econ PhD aspirations! Not certain why, but my observation is that a lot of students won't touch non-required math unless they've already gotten a lot of A's in the lower level courses. By this selection mechanism, then students who continue taking mostly math (and related classes in CS, econ, physics, or logic) will continue to get mostly A's and A-'s, especially since upper-level classes rarely curve.

 

So sure, the best math majors mostly go to pure math PhDs, I wouldn't dispute that. But because of the high concentration of near-perfect grades among math majors, at most good schools there will still be at least one math-ish major (current or someone who graduated in the past few years) who meets the high grade standards in my first paragraph and is applying to statistics or biostatistics programs. In the aggregated stats applicant pool, count on at minimum a dozen strong applicants from LACs like Pomona, Swarthmore, Amherst, Carleton, Williams, a bunch more -- even little ol' St. Olaf is a powerhouse. I think I'm underestimating the LAC contingent actually. Then count on a couple dozen more applicants from the mega name-brand research universities (Stanford's programs alone must send out a half dozen people at a time and I think Chicago too). Throw in another couple dozen apps from other respected schools like UNC. Basically, name all the good private and public schools you can, and imagine that there's on average one current senior or recent grad with strong credentials applying to stats departments (because, again, high grades are not at all unusual for math majors). And surely most of the recommendations for this crowd are quite good, especially for the substantial portion of these applicants coming from small schools who get detailed personalized recs or those with access to bigger names?

 

Anyway, a couple dozen here and there ballparks to something like 60 applicants in the pool who are going to be at the top of the piles everywhere they apply, so I think a fair number more than what you have in mind. It would be a big pain, but you could look up profiles of first-year grad students at the top 15 or so statistics departments, and I bet you would find enough students who majored in something math-ish graduating with high honors (proxy for high GPA) at a well-reputed undergrad to substantiate my claim. In my cohort alone there are at least 6 of us who would meet that description, have to imagine the scene is similar elsewhere.

[aridneptune]

Well, I'm glad to have sparked such an interesting discussion. I'd like to give a few personal notes and then my response to what wine in coffee cups noted above.

 

So first, personal stuff. I am indeed a US citizen. And I'd like to suggest that my LORs should be quite strong (albeit I have limited experience with the grad school application process). I still talk to and work with several professors from my school - I see them in person 4-5 times / year, have a regular email conversation going, and give 'how-to-make-it-in-the-world-of-work' talks in their classes during recruiting season. Incidentally, on that last point, I work for a major bank (in case that makes any difference).

 

Now, as to SOP: man writing that thing is difficult! I'd forgotten how painful writing college essays was. I mean, how does one put one's life goals on paper in an attractive way for external consumption? Well, anyway, I'm doing the best I can - and I'll post the results when I get it written (probably not for a couple of weeks as I have several midsummer weddings / trips coming up). I want to clarify that I am not simply bored with my job. I took math and statistics in preparation for an econ PhD but came to love the fields on their own merits. I want to study statistics because I both enjoy and want to learn more about the field. I took my job out of undergrad because I was (1) overweighting money in my personal utility function and; (2) overly nervous of getting gainful employment after the recession (this was 2009). Working on Wall Street has been useful in that I now understand the value of education and academic life in a way I really didn't before. Well, that and I can work 15-hour days and saved some money.

 

Is it possible to apply to both a school's PhD and MS programs in the same year? cyberwulf, I very much appreciate the advice and plan to apply to several PhD programs and several MS programs as safeties. It goes without saying that I'll put everything I can into each application. I'm not sure where you guys are getting $100k for a 2-year MS degree, but even if it is that much I can do it without debt.

 

Finally, a note on wine in coffee cups' comment. I'm sure you've seen many more students than I have, but my experience is rather different from yours. At UNC, very few students had math GPAs above 3.5 (and I do not say this to toot my own horn or anything, just a note). Calculus / linear algebra / diff EQ classes were maybe 40-50% CS or Chemistry majors who weren't interested in the topic and did the minimum possible work. Any upper-level math class (topology or analysis, for instance) had about 15% A, 30% B, 45% C, 10% D-F (in my experience). Of course much depended on the course and professor. Nonetheless, I'd say at most 15-20% of math majors had an A / A- average. And the graduating class was only ~50 students or so in 2010. Just my two cents.

Edited August 16, 2013 by aridneptune

[cyberwulf]

There's a distinction between someone with a 4.0 and someone with a 3.9, a distinction which is particularly important for top graduate programs: the former student hasn't yet met a course they couldn't ace. Schools are trying to ascertain applicants' upside/potential, and a 4.0 (vs. a 3.9) is a really good indicator that you have a lot of it.

 

I'm genuinely surprised to hear about the GPAs of math majors in the data you analyzed. Regardless, I still think you're overestimating the number of elite applicants; at least in biostat, I can tell you that there are no more than 5-10 with the type of profile you're describing applying in any given year. At our (fairly well-ranked) institution, we probably average 3-5 applicants from the top 20 liberal arts colleges TOTAL. I doubt that the applicant pool in stat is 5-10 times deeper than biostat.

[Shostakovich]

There's a distinction between someone with a 4.0 and someone with a 3.9, a distinction which is particularly important for top graduate programs: the former student hasn't yet met a course they couldn't ace. Schools are trying to ascertain applicants' upside/potential, and a 4.0 (vs. a 3.9) is a really good indicator that you have a lot of it.

 

I'm genuinely surprised to hear about the GPAs of math majors in the data you analyzed. Regardless, I still think you're overestimating the number of elite applicants; at least in biostat, I can tell you that there are no more than 5-10 with the type of profile you're describing applying in any given year. At our (fairly well-ranked) institution, we probably average 3-5 applicants from the top 20 liberal arts colleges TOTAL. I doubt that the applicant pool in stat is 5-10 times deeper than biostat.

 

I think it's possible that the state school system uses a more stringent system of grading for their science courses, or that the kids in the system tend to not do as much work. Probably would say both based on personal experience.

 

I remember majoring in math at a UC system it was pretty standard for the classes to be curved to a C+ or B-. For my brother's graduating class this year (also at a UC campus) he was telling me GPA's in the high 3.6's were already eligible for Cum Laude, which is something like top 10-15%.

[DMX]

I'm genuinely surprised to hear about the GPAs of math majors in the data you analyzed. Regardless, I still think you're overestimating the number of elite applicants; at least in biostat, I can tell you that there are no more than 5-10 with the type of profile you're describing applying in any given year. At our (fairly well-ranked) institution, we probably average 3-5 applicants from the top 20 liberal arts colleges TOTAL. I doubt that the applicant pool in stat is 5-10 times deeper than biostat.

Is this conditioned on domestic applicants?

 

My experience has been more in line with what wine-in-coffee-cups has said. Most of my math-major friends (who realized early on that they didn't want to go for a math PhD) had 3.9+, because:

1) they took only the pre-requisites

2) math classes had a very generous curve

 

And for the semi-advanced math courses (upper level linear algebra, intro number theory, intro topology etc.), the classes were usually curved to B+/A-. 

 

This was at a private institution--perhaps there is not as much grade inflation at public schools.

[dasgut]

I'd GENERALLY agree with the tendency toward high GPAs for math majors. In my MS class we had about 10. Only 2 of use graduated. We had one Phd who graduated with just an MS (he went to a different PhD program). Of the initial 10, eight failed out. The other two of us had 3.8 and 3.97 respectively. The PhD who just took the MS also had a 3.97.

[cyberwulf]

MS and undergrad GPA mean totally different things. A 3.7 MS student is usually about average, so most decent schools look for a 3.9+ Masters GPA, particularly if the program isn't well known.

[biostat_prof]

I'm going to second cyberwulf's comments and say that if anything, he is being conservative. I worked with one student a year or two ago with comparable numbers to the OP who literally ran the table when she applied to grad school: She applied to every single top-ranked stat/biostat department and was accepted at all of them. Granted, she had some very impressive research experience and absolutely ridiculous recommendation letters (and being female also helps at stat departments), but I think you are competitive everywhere if your recommendations are strong. Don't be scared by DMX's experience; the competition is far, far more intense for foreign students. I would just apply to all of the top-ranked stat departments (and maybe biostat as well if that interests you at all). You should definitely get in somewhere. I don't know that you even need to bother applying to MS programs unless you want the option of reapplying in two years if you're not accepted at a top-ranked school. I can't imagine you get rejected everywhere unless one of your recommenders says that you are a psychopathic felon or something equally disastrous.

 

Also, check your PM.

Edited August 18, 2013 by biostat_prof