Stat Assistant Professor

I'm not really sure. I haven't seen any alumni of my department go to work at a government research lab... I imagine if you have several summers of interning in a research division of a national lab or a major corporation, that it would be beneficial. Most who go the industry or government route from my department go to large tech companies, big pharma, financial institutions, or government agencies (FDA, Freddie Mac, etc.). And I have only seen these alumni complete a single internship the summer before they graduated (and some didn't even do any internships at all).

Most of the students at my institution who go into industry only do one internship the summer before they intend to graduate (and this internship was the ONLY non-academic work experience many of them had). The summers before that, they either traveled and/or did research. I think that if you are a PhD student in Statistics, the most recent internship you have is the most essential one, because companies will typically make job offers for their summer interns for the following year based on their performance. Other students also do one internship but find that they prefer the academic lifestyle, so they go on to do postdocs instead (or in some cases, they get a TT position if they have several strong publications). I recommend you spend your first few summers preparing for qualifying exams and trying to get research done. This will give you more perspective on what you really want and also give you more time to get work done (as an academic, I can say that the summer truly is the best time to get work done). Then if you are still undecided, you can do an internship the summer before you plan to graduate so you can compare and contrast. * ETA: An exception might be if you really want to work in a research division of a major company (like Microsoft Research or IBM Research). Then having several summers of industry research internships would be helpful. But otherwise, I haven't seen that it is necessary to have more than one to get a good job with a Statistics PhD.

It is impossible to tell your chances from your post. Undergrad economics can be done either with rigorous mathematics or it can be done by taking less math-heavy classes. How did you perform in your undergrad math classes? Did you take any math beyond linear algebra and Calculus III, etc (I assume you took those at the minimum since you were admitted to a Masters in Statistics program)? Assuming that your undergraduate grades were good, it may also depend on whether you think you can get quality letters of recommendation from your online instructors. Additionally, you will need to take one semester of real analysis, if you haven't already (and ideally a few advanced math classes beyond Calc III/Lin Alg). Since you say you are working part time while you study, would it be possible for you to take some undergraduate math classes at a local university (including, most crucially, real analysis)?

Almost all programs have a written exam covering Masters level statistics theory (at the level of Casella and Berger) and possibly applied statistics. If you do not get a PhD-level pas on this exam within two or three attempts, usually you'll be asked to finish with a Masters but not continue on in the PhD program (at this point, some students will also, on their own volition, elect not to continue on to the PhD). At some schools, those who DO continue on the PhD track have to then take a second written exam (or exams) based on the PhD-level theory classes. At some schools, there are tests in different subject areas (e.g. advanced statistical theory, probability theory, applied statistics, or whatever) and you pick two of the three and need to pass those. This was how it was for the PhD in Statistics at both the department where I got my Masters and my current department (i.e. two sets of exams: a "Basic" exam and an "Advanced" exam). The "Basic" exam is (in theory) meant to undo any admissions mistakes, but is obviously also subjective.

Even in Statistics, the Math Subject GRE is not typically required, except at a few programs. And some very good programs don't even consider it at all (for example, Duke does not consider it at all). A good score is helpful if you have a lighter math background, but I wouldn't worry about it too much if your math grades are good (unless you plan to apply to one of the few programs that requires it, e.g. Stanford or UChicago). Plus, many students have taken it and scored very well on it but still get rejected for whatever reason (it's not uncommon for many students from China, for example, to score in the 90+ percentile on the math subject GRE but still get denied admission to top programs in statistics). If you apply to Statistics programs but are primarily interested in biostat, it may be helpful to apply to a few programs where you can do a Biostatistics concentration (e.g. University of Wisconsin Statistics, Rutgers Statistics, or NCSU Statistics -- in NCSU Stat Dept, they have a statistical genetics group). Otherwise, the options to focus on that might be limited in a Statistics department.

I attend University of Florida (graduating in August though). And for what it is worth, the UF Statistics Dept seems to be a bit more "open-minded" in the applicants that it admits (as in, we have admitted students who attended very little-known undergrad institutions, students who didn't major in math or stat as undergrads but then later switched, etc.). My undergrad major was Economics (my Masters is in Applied Math though), and there is even one alumnus who's currently an Associate Prof of Statistics at an R1 who majored in Journalism for his undergrad (but then he got a Masters in Math, so I think that helped). As for the alumnus that I mentioned, he is doing research on Bayesian nonparametrics, and I believe that he approaches it from both applied and theoretical angles (having several publications in top theoretical and top applied journals). I think our program does a fine job preparing graduates for their future careers, even if it is not ranked in the tip-top tier. In academia, we place grads in prestigious postdocs (Duke, Stanford, Carnegie Mellon, UPenn, etc.) who have gone on to get Assistant Professorships, or occasionally, alumni get TT jobs right after graduation (I suppose at a "lower" ranked school, it's tougher to pull this off than if you graduate from say, Stanford or Berkeley). But for the most part, anyone who wants an academic job can get one, even if most of us have to spend some time postdoc'ing after we finish the PhD.

Yes, I would apply to a wider range of schools and target ones that are more aligned with your personal interest. It may also help to look at programs that have a reputation for accepting students from more varied backgrounds (you mentioned your background in finance/accounting). Actually, one of the Statistics PhD alumni from my current program had a Bachelor's degree in Finance (only had a minor in Statistics), but switched to Stats for his PhD. He is now an Assistant Professor in Statistics at an R1, doing very well. Some of the top 10 programs, on the other hand, may be more likely to accept students from the most prestigious undergrad schools who have the strongest mathematical backgrounds. You stated that you are interested in statistical learning. This is quite broad, so again, you should target programs that are most aligned with your interests. For example, Yale has some very strong faculty working on statistical learning (Barron, Lafferty, Zhou), but it is heavily theoretical. So they may be looking to recruit students who are particularly interested in theory. If you want to do less theory, then perhaps look at programs in statistics and biostatistics that also have research in the area of machine learning but from a more applied/computational angle.

You have a 3.9 GPA, excellent GRE scores, and majors in math and stat from an Ivy League school. Assuming you can secure decent to strong rec letters, you will probably get admitted to most of the schools you listed. Any reason you would not want to try applying to Stanford, UChicago, Berkeley, Harvard, Duke, UPenn Wharton, etc.? You can probably get into a few of those too.

Thanks for the clarification. It seems like your application was very strong. It's just hard for most everyone (especially international students) to get admitted into top stat PhD programs. There may also be a bit of "elitism" at top schools, for better or for worse. For example, there was an American (*NOT* international) student in the combined Bachelors/Masters in Statistics program at my school who had a 4.0 GPA, took basically all pure math and statistics PhD level courses in his last two years, did legitimate research on measure theoretic probability theory in the mathematics department, and was STILL rejected from most of the top stat PhD programs. Now, my school is by no means bad (it is in the top 10 public universities, according to USNWR), but it does seem like the more prestigious your undergrad pedigree (Ivy League, MIT, Stanford, etc.), the better your chances are for being admitted to a prestigious PhD program if you were a top performing student. As for econometrics and biostat, I would say to take some time to reflect on what you really want to do. The culture of a top stat department in theory would probably differ quite a bit from the culture of a more applied statistics dept, an econ dept, or a biostat dept that is focused heavily on say, genomics (all will train you in theory and have some research in theory, but the *predominant* research focus will likely have a specific "nature" to it). If your interests lie more in applied machine learning and computational statistics, I would target departments that are especially strong in those areas. Re: work experience. It is unlikely to sway your application either way. You can of course mention in your SOP that it helped to shape your interests and passions (for instance, in my statement of purpose, I mentioned that working on Monte Carlo simulations and doing statistical analysis on analog signals while I was working as an engineer led me to want to pursue statistics). But playing it up *too* much as if it gives you solid preparation for a grad program is probably unrealistic and would be viewed negatively.

Did the programs you applied to request only three recommendation letters? Sending four might have been seen as excessive, and it's really best to only send three of them (the strongest ones you think you can muster) if you reapply to programs in the future. How was your statement of purpose? I know that having an amazing statement of purpose isn't the most important thing for admissions, but a sloppy SOP or one that contains "red" flags can also hurt an application (e.g. one where you indicate that you have *no* intention of even considering an academic career, one that comes across as overconfident or one where you try to make it seem as though you have more expertise than you really do, etc., etc.). Based on your test scores and grades, I cannot imagine that you were denied admission because of your raw numbers, which are very good. I would try to work on strengthening other aspects of your application and make sure to read the applications instructions carefully and follow them to a T.

In statistics (and I think biostatistics too), the PhD thesis is usually two or three of your papers stapled together, along with a literature review and a conclusion/future work section. Your PhD advisor will help guide you along the way, so you won't be left completely to yourself -- they will help you find a "doable" open problem for your first project, and help you through the entire process of: manuscript preparation (you'll probably go through multiple edits *before* even submitting to a journal), the initial submission, revise and resubmit process, the point-by-point response to reviewers, etc. Your advisor will know what is considered publishable quality and also have some idea of which journals are the most appropriate venues to submit your work to. Most PhD students have no idea what open problems are out there when they first start, so the bulk of the first semester after passing written qualifying exams will probably be spent just reading papers and books, teaching yourself a new area, and doing small exploratory projects (e.g. running simulations, reproducing results from a "seminal" paper, or something like that). A lot of stat/biostat PhD students will also have one or a couple of third/fourth author publications from helping out other people on projects (i.e. they'll be listed as a co-author for writing some R code, performing some data analysis, or otherwise making some small contribution). However, by the time you finish, you should ideally have at least two FIRST author papers where most of the work was your own (the idea, the algorithm(s), the theory and proofs if any). At the end of the PhD, you ideally *should* know more about your topic than your advisor (it's your research after all).

Depends when the application deadlines are. If you think you can get strong letters from Masters instructor(s) by the time that applications are due, you can still apply for PhD programs (or alternatively, if you can secure good recommendation letters from your undergrad professors). Don't worry about research experience for PhD admissions. A small minority of applicants have noteworthy research, but it is not very common either (REUs and Masters projects don't usually have any resemblance to PhD research).

It is a smaller, more theoretical department, but I think it has some very strong faculty... for example, Harrison Zhou, Andrew R. Barron (big name in information theory), and they recently got John Lafferty too (a huge name in statistical learning). I think it is an excellent department.

A few other things to keep in mind: when you're still taking courses, it can be easy to get discouraged if there are other students who seem to be "faster" or more clever when it comes to taking tests. Specifically in the field of statistics, it can be daunting for some domestic students, because the international students have (in general) been exposed to more mathematics. However, it is important to do your best and not to focus much on other people -- what's most important that YOU understand what's going on in your classes. Moreover, acing classes and exams is *not* what earning the PhD is about. Once you get to the research phase, most people will be on even footing at that point -- since just about everyone needs to teach themselves a brand new area. And there are quite a few students who were acing all their classes but who struggle through research, because they had difficulty adjusting from the mindset of student to researcher (the latter of which is requires a completely different mindset from being a top student: the latter is necessarily about making mistakes, muddling through things that don't seem to make any sense at first, being stuck for WEEKS at a time, abandoning unpromising avenues, etc.). Even if you succeed at completing your PhD, it can still be intimidating because there are established researchers who seem to be churning out papers nonstop in top journals, etc. But just keep in mind that these are people who have been doing what they have been doing for many years, *and* despite all of it, even they still get a lot of their papers rejected. Plus, it can take years to reach their level of skill and expertise. But if you keep pushing yourself, you can eventually build your skills and become like them (if you aspire to that) with more experience.

You have a very strong profile for just about any Masters program, but your math background is a bit sparser than other applicants for PhD programs (in Statistics). Taking an advanced proof-based linear algebra class and a real analysis class and having good grades in these on your transcript should boost your application. As for publications, a lot of international applicants will have one (most likely not in in a venue like JASA, Annals, JRSS, etc., but sometimes in a business or an econometrics journal or a less prestigious stat journal), but I also don't think it is required to get admission to a reputable school. Admissions are very tough though, so I would apply widely.

I went through two Windows laptops as a grad student and found that it was sufficient. If you want to do developing on your own free time that is unrelated to your research, then *maybe* it helps to have a Linux machine. But if you need to do any sort of high-performance computing for your research (e.g. running simulations that might take awhile to finish), you will most likely not do it on your personal machine anyway, but instead, on your school's HPC cluster that has tens of thousands of CPU cores. And to access that from your personal computer, you just need an SSH client (any machine will do).

That is quite surprising to me that you didn't get in *anywhere.* I agree that lower ranked doesn't imply easier to get into... and I am not sure I agree with some of the "lower" rankings either (e.g. I think very highly of Yale and Rutgers Statistics depts personally, but that doesn't seem to be reflected in the rankings?). Then again, it seems to be very, very competitive for non-domestic students right now (it's still competitive for domestic but much less so -- it seems as though the strongest math students in the USA tend to want to continue studying mathematics rather than statistics). You probably need to cast a wider net next cycle, given how much competition there is for international students. For example, the program that I attend is ranked in 21-30 range for Statistics, and one of the incoming first year international students this year had an MPhil AND a publication from his Masters work. So maybe it is just that the competition that has gotten much more fierce in recent years? I think maybe the Waterloo program sounds the best, based on the comment by statfan. If you perform well in graduate-level mathematics/stat courses and write a great thesis -- and your thesis advisor can attest to the quality of your thesis and how it speaks to your potential as a PhD-level researcher, then it can definitely help boost your profile.

You are absolutely not too old. You may be on the older end but it's not ancient either. And it's not unheard of either to start a PhD in your 30s. I just looked at the UC Berkeley Statistics program and saw that there is one PhD student who finished his BS in 2002 and who entered the PhD program there in 2013 (I'm only aware of this because this student was one of the winners of a Student Paper Competition at JSM ). Looking through the PhD alumni at Berkeley Stats, I also see one alumnus who got his BS in 1997 and finished his PhD at Berkeley in 2012, and is now a Professor at Harvard. And there is another one who finished his Bachelor's in 1985 and didn't complete his PhD at Berkeley until 2013 (almost 30 years after he got his Bachelor's!). So you're golden if you want to spend another 4+ years getting a PhD after you're finished with your Masters.

In general, admissions to Masters programs in Statistics should be much less competitive than for PhD programs. I am not sure about admissions to MS programs in your country, but in the U.S., it is much less competitive, besides a few selective MS programs (like Yale or Duke). Masters programs are also more forgiving about poor grades in the first few years. If you do decide to do a Masters in Statistics and do well in it, you may possible be more competitive for schools at the level of University of Florida, Missouri, Virginia Tech, and UConn, and schools ranked below that (a "reach" school for you -- ASSUMING you got a Masters and did well -- would probably be a school like Ohio State). If you want to pursue Statistics and you have the funds, perhaps obtaining a Masters degree in it is not a bad idea. That can give you a flavor of whether you enjoy it enough to continue pursuing a PhD, or if not, it will open up a lot of doors in industry.

Even for the schools you've listed, it's probably going to be tough to gain admission (not sure about SUNY or Stony Brook, but Missouri, UConn, and Penn State are going to be tough). Your work experience at Deloitte has value for future employers but won't really be given much attention for PhD admissions. To be blunt, there are going to be a lot of international students at Missouri, UConn, and PSU who have much stronger profiles (including some with impressive profiles who just weren't lucky enough to get into the top tier programs -- possibly because of one small weakness in the application, like forgettable rec letters or something). If you are DEAD set on getting a PhD in Statistics, what may help in your case is to complete a Masters in Statistics (where you also take two semesters of real analysis and one or two additional proof-based math on top of that), or a Masters in Mathematics where you complete 4 statistics courses (the year-long Masters level applied statistics sequence and the Casella-Berger mathematical statistics sequence). That is not to say that getting the Masters will *guarantee* you admission to a PhD program, but in your case, it would improve your chances from your current profile. For reference, my current PhD program has international PhD students who did their undergrad at less well-known or obscure institutions, but who got their Masters degree at a reputable university in their home country (like ISI or Sharif or one of those schools). Or they got a Masters from a reputable American university. And doing well there helped them gain admission to our PhD program.

For PhD programs in pure mathematics, I think it is indeed true that having taken graduate level courses in math will strengthen your application for math PhD programs. However, I think this is because PhD adcoms in mathematics have concerns about the level of mathematical rigor/preparation of domestic applicants (who admittedly tend to have been exposed to much less mathematics than international students). So having taken graduate math courses alleviates their concerns about domestic students being able to handle the graduate coursework. But in other PhD programs like Statistics, Economics, and Finance, I tend to think that taking more mathematics courses will boost your application moreso than taking classes in those disciplines (however, I am only hypothesizing this, I do not know if that is indeed the case).

Given your profile, I would not think of Pittsburgh as a "safety." As a non-domestic student, the bar is going to be way higher for you at top schools. For reference, one of the Associate Profs in my Department got his PhD at Stanford, and he finished top of his class from his college and scored in the top 4 in the National Mathematics Olympiad in his home country. And there are enough international applicants with stellar profiles like this (finished in the top 5% of their class, perfect scores on the Mathematics Subjects GRE, won nationwide contests and math Olympiads -- particularly from China and India) that the top schools really do have their pick of the cream of the crop. For international students like the ones I've described, Pitt might be a "safety." Also, admissions to mid-ranked/lower-ranked programs is still going to be quite competitive. I am graduating from a program that is ranked in the 20-30 range in Statistics, and the acceptance rate for the PhD program here is between 5 and 10 percent. I would not be surprised if it is like that at Pitt and other similarly ranked programs.

Fair enough. I can't say I know for sure whether a Masters in Statistics or a Masters in Mathematics looks better to a PhD admissions committee in Stats, but anecdotally, my current program has admitted a number of students into the Statistics PhD program who obtained Masters degrees in (pure) mathematics (including some alumni who are now faculty/postdocs at places like Duke University, University of South Carolina, etc.). My MS is also in Applied Math. Based on personal experience, a Masters in Mathematics would certainly not be a handicap in admissions to statistics PhD admissions. I'm not sure if a Statistics Masters is "better" for admission to PhD programs though.

Maybe it depends on the program, but there are many PhD students in Statistics who never took any stats before matriculation (maybe they took an undergrad probability class but that's it). If someone were to tell me, "I want to get a PhD in Statistics but I haven't taken much/any stats before. Can I still get into a PhD program?" I would probably tell them that this is not a huge issue as long as they have a very strong math background and great letters of recommendation. As per the OP, the UMN Statistics Masters description even recommends that those considering a PhD in Statistics take math electives to strengthen their application.

For admission to Statistics PhD programs, taking many more statistics classes and having research experience are usually not necessary (the latter is especially rare, and it is also not uncommon for students to be admitted to PhD programs who have never taken a statistics class before but who only have an extensive background in pure math). In fact, I have a suspicion that taking *too* many advanced PhD-level stat classes as a Masters student can hurt your PhD application. I think a lot of programs want to teach topics like advanced statistical inference, measure-theoretic probability, and theory of linear models/generalized linear models to you "their" way in order for you to pass their qualifying exams. But I agree with you that getting a Masters in mathematics is not going to automatically guarantee admission to a Statistics PhD program, so if statistics is the professional track the OP wants to go down, it may be more worthwhile to get a MS in Stat, supplemented with advanced math courses. If that is the case, I would try to take a holistic view at your application. If other factors were strong (math grades, GPA, GRE score), then it may have been recommendation letters that did not "stand out." You can only speculate at this point, but if you eventually do decide to apply for PhD, you should do your best to ensure that everything within your control is strong.