Stat Assistant Professor

Coolbeans. I wonder why this thread did not show up when I searched for it!

Hello everyone, I did not see a thread for this when I searched. Is anyone here planning to attend University of Florida this fall? I will be starting in the statistics PhD program, and I am very excited. Please feel free to PM me. It'd be great to chat with some folks before going.

I like MIT OpenCourseWare a lot. I have been studying for the first year qualifying exam, and the notes on that site have been very helpful.

For stats (not biostat), NCSU, Duke, Ohio State, Harvard have some great industry placement and seem to place more folks in industry rather than academia. NCSU and Carnegie Mellon seem much more into applied stats (judging from their research and the past dissertation topics). Both would be excellent places to shoot for. I think Biostats would be relatively applied most places.

What are your long-term career goals?

If you can get stronger letters of recommendation that say something memorable about you (rather than just "he was one of the top students" but that give concrete examples of your research potential, e.g. "He completed a project where he did [x y and z] and demonstrated clear ability to comprehend statistical literature"), I'd say you would be a very solid candidate for most biostatistics PhD programs. For pure statistics programs, you are a solid candidate as well, but I think your profile will be even better if you get *all* strong LORs. At the top programs, it is difficult to assess your chances, but I'd say you can definitely get into a top 20 stats program and likely a few top 10 too. Your admission chances will probably boil down to your letters though, since it seems your other stats are excellent.

How much debt would going to UW put you in? Is there no financial support from either program? If the amount of debt were similar, I would pick UW, but not if it means going into a *lot* more debt. Then I would just go to UCLA.

Perhaps one of the stat professors should chime in about how undergraduate research experience is viewed for statistics/biostatistics applications. However, for mathematics, I am confident in saying that most undergrad research is not viewed as being very mathematically sophisticated (there are of course very rare exceptions), so adcoms do not place as much emphasis on on it as they do on grades, letters of recommendation, and exposure to graduate-level math (or barring that, breadth of serious theoretical math classes taken).

Not sure about this... it probably is still unforgivable for most of the top stats programs (not sure about biostats), but programs ranked below the very top tier might overlook it. I am not really sure about how it looks for stats. However, my impression is that the admissions committees in statistics pay close attention to all the upper division and graduate courses taken in both mathematics and statistics (or if you are coming from another quantitative field, the courses taken in your original field of study). There's probably also a huge difference in perception between a single B- earned in abstract algebra but A/A-'s in all other math classes vs. a C in abstract algebra or a number of B/B-'s earned in proof-heavy classes. I think REUs may be beneficial for the student in helping them determine if they'd like to do research as a career, but most of the math professors I've talked with (as well as the math professors who chime in on mathoverflow.net and Stack Exchange) have stated that they do not place that much importance on the REU -- or undergraduate research in general (since most of the students are not mathematically mature at that stage in their training)... unless one got a poor or neutral letter of recommendation from the professor they worked with or unless one is a Terence Tao-esque genius who is already producing publishable results as an undergrad, REUs do not make or break an application in most cases for mathematics (not that they hurt). I think doing well on a Putnam exam, taking loads of graduate courses and acing them, and taking a PhD qualifying exam as an undergrad and achieving a high pass are all things that look way better to a math admissions committee than an REU.

Actually, in mathematics, not having an REU or research experience won't matter that much (as far as I can tell and based on my convos with math professors who serve on admissions committees). Many REUs do not resemble graduate-level research in mathematics, so the letters of recommendation, grades, and amount of exposure to graduate-level courses carry the most weight (though the top tier schools will also place heavy importance on the subject GRE and it's not competely unheard of for applicants to MIT, Princeton, Harvard, etc. to have math publications either). The statistics professors on this board have also confirmed that research experience isn't as important as research potential either (which they feel is best assessed by grades and letters of recommendation). Also, there is actually very little room for error for the top PhD programs in math and statistics. A poor grade in analysis or abstract algebra is unforgivable for the top programs in math, even if one has a Masters degree and gets A's in graduate analysis and algebra. As anecdotal evidence, another poster on this board and over at mathematicsgre.com had a 3.5ish undergrad GPA from a top 20 school; he obtained a Masters degree in pure math with a 4.0 GPA and achieved a very high subject test score, but he was still rejected across the board one year. According to him, his undergrad GPA and the lack of having taken a Masters level probability/statistics sequence were the main reasons for his rejections. I also had a high Masters GPA, and it was not enough to overcome a 3.5 undergraduate GPA for the top schools. I agree that this applicant and most other applicants can definitely handle the coursework -- but when there are so many strong applicants, adcoms can afford to be picky and will typically (but not always) go with the ones who have the least blips in their academic performance (of course there are exceptions). However, once you get outside the top 20 schools, there is a little bit more flexibility. I got into every school I applied to that was ranked in the 20-30 range (when you separate stats from biostats) but was rejected from nearly every school I applied to ranked above that. So I don't think I was competitive enough for the top 20, but definitely competitive for the ones ranked below that.

1. Attending: University of Florida 2. Turned down: UConn, Rice. Removed from waitlist: UCLA. 3. Reasons for attending:Fellowship + stipend for UF was more generous than the financial offer from the other two. Additionally, Florida has a great group of professors working in the areas of Monte Carlo methods and Bayesian statistics, which make it a good fit for me. Their application fee was also only $35, byfar the cheapest of all the schools I applied to.

^ That is a good point. There are a few top schools that seem to be a little bit more lenient with GPA - Carnegie Mellon and NC State come to mind immediately. So the OP could afford a few more reach schools, but in general, I think he should be targetting his search in the middle to lower range. I think Berkeley cares a lot about grades in upper division math and stat classes, actually. Their application made us submit a detailed list of all upper division math classes taken, including grade earned, instructor, textbook used, and description of topics. I think you are right though that Berkeley places a little bit less importance on overall GPA, but they scrutinize grades earned in math and stat classes pretty closely. So if that is the case, the OP's chances with Berkeley are not great.

Just being real, the top programs in math/stat seem to have little room for mistakes like poor grades in math/stat (even a Masters degree may not be sufficient to overcome this), and your chances may not be great, barring very unusual things that otherwise set your application apart (e.g. a famous LOR writer). The lower down the ranks you go, though, the more forgiving the admissions committees will be, e.g. I have a MS from a mathematics department ranked in the 60-70 range on USNWR, and they have accepted a few PhD students with low 3.0 and high 2.0 GPAs but who done very well in Masters programs in math. I'm sure you must have some very other positive attributes since you have been accepted to a top 25 PhD program in another field, so it is possible that I am underestimating your chances... but without knowing the other stuff and based SOLELY on the information you've posted in this profile, I think a realistic list of schools for you would have Ohio State (for stats) and Indiana-Bloomington (for math) as being on the upper end of places to aim for. Also, keep in mind that rankings aren't everything. There are some "lower ranked" schools that are very solid and that have great academic/industry placement -- for academia (if that is your end goal), as long as you work with a good adviser who is reputable and who has a strong publication/citation record and get some publications before going on the market, you could potentially be in better shape than someone from a top tier school who isn't as productive.

The top schools will be hard to break -- there is very little room for error unless you have something *very* outstanding in your application (i.e. you have very glowing letters from rockstar professors or you're a member of an underrepresented group). You will definitely need to address the poor grades in upper division math in your statement of purpose, but you need to put emphasis on the fact that you have matured since college. You may want your letter of recommendation writers to address your lower grades as well, but to stress that you are one of the best students they've ever had and that your grades are not reflective of your ability. You should definitely speak with your adviser and your professors to find out what would be an appropriate list of schools to apply to where you have some chance of getting in. However, assuming that you can get glowing letters of recommendations and score well on the mathematics subject GRE, I would suggest having places like Indiana-Bloomington and UC San Diego (if you apply in math - these depts have applied math/statistics reseach under their math departments), SUNY Stony Brook (if you apply in applied math), and Purdue and Ohio State (if you apply in stats) as your "reach" schools. Otherwise, you may have much more luck at mid-tier to lower-ranked or unranked programs in stats and programs that are mid-tier in math/applied math.

Congrats! Agreed with all of this.

... though I could definitely see a stats dept being willing to admit someone with a degree in physics, CS, economics, or engineering who has demonstrable quantitative skills but who lacks real analysis (and they just take it their first year in the PhD program). However, this prospective applicant is getting a mathematics degree, and the adcom may question the rigor of the math program if it doesn't require analysis to graduate. I could be totally wrong though. I agree, a stats faculty member should weigh in on the relative importance of having analysis on the transcript.

Hmmm, I did not know that this was the case. I was led to believe that real analysis was heavily encouraged from previous posts (e.g. , and I believe biostat_prof has said the same thing in other posts -- IIRC, the phrasing was that the application might be thrown out if advanced Calculus/real analysis was not on the transcript). And when I started a Masters program just under four years ago, I was also told that if I eventually wanted to do a PhD in applied math/stat, I should take analysis. So it still seems as though it is "strongly encouraged" if not necessary.

Is there any way you can get a third strong letter of recommendation? For domestic applicants, it seems as though the lower ranked or less well-known your institution, the more the letters of recommendation will be weighed. Additionally, a good score on the math subject GRE (let's say 70+ percentile) could help to mitigate the issue of adcoms being unfamiliar with the level of rigor of your undergrad. That said: this exam can be pretty challenging and requires a lot of studying, and you seem to be missing quite a few of the topics that are tested on the exam (real analysis, abstract algebra, advanced linear algebra). Additionally, I think you would have a hard time getting into a statistics PhD program given your current status unless you take a real analysis class (and perhaps a few more proof-based math classes). I've personally found that for pure stats programs, the depth of advanced mathematics taken is more important than the amount of statistics you've taken... especially since the first two years of classes in grad school will be VERY theoretical (after that, you could choose to do a dissertation in a heavily applied area, but you still have to take a bunch of classes on proofs/theory and pass qualifying exams on theory before you go do that). So you certainly must take real analysis, and apart from that, it might be helpful to take some other rigorous proof-based classes to show you can handle mathematical proofs. Based on your coursework, I think getting a Master's first would be most beneficial for you. That said, you may have much more luck with biostatistics than with stats -- it seems your profile may play a lot better to biostats. Have you thought about that? EDIT: For the SOP, just mention why you want to study statistics and mention a few areas of it that interest you. You say you've taken a time series class before -- so if MCMC, stochastic processes, etc. are things that sound cool, you could talk about those possibly being areas you want to explore.

I am going to accept Florida's offer. I will decline Rice and remove my name from consideration at UNC later today. Hope that helps someone else out!

NC State has excellent industry placement, so you would do well there. I am not sure about UCSC... you should ask the DGS for job placement information, or a list of PhD graduates (then you can search for them on Google and/or LinkedIn).

I declined one offer so far, and I did not list a reason. I said that I was very impressed by their program and that they were one of my top choices. I thanked them for their interest in me as a candidate but that I would be accepting another offer. If they ask for a reason, you can tell them, though I would definitely avoid mentioning something like reputation (fwiw, reputation did not factor that greatly for me anyway, since the places I got in are similarly ranked. For me, it is more about fit, financial offer, and possibility of finishing in 4 years -- I already have a Master's and some of those credits can transfer provided I pass the qualifying exam).

I've removed myself from their waitlist. Best of luck to you!

It differs by field. Some fields (e.g. the humanities) do not allow you to count that many Masters credits towards the PhD degree, so doctoral students in those fields end up having to repeat classes if they already have another advanced degree. However, I have found that statistics is not like that at all, and I imagine biostats is similar. Stats departments do seem to allow previous graduate course credit to transfer to the PhD. HOWEVER, not having to retake the coursework is also contingent upon passing a basic qualifying exam (so if you take the quals before entering the PhD program and fail them, you'll have to repeat the courses). fwiw, both Rice U. and UFlorida allow up to 30 credits from graduate classes to count towards the 90 credits that are needed to graduate. However, not retaking previous coursework is entirely contingent upon passing the basic qualifying exam. At UF, students who enter with a Masters degree but who either a) fail the basic exam or b ) don't attempt the basic exam at all in the August before they matriculate need to retake all the previous Masters classes (e.g. the graduate intro probability/statistics sequence).

+ Duke UNC, Duke, and NCState have the advantage of being located in the Research Triangle. I have not heard anything at all about NYU or Stony Brook placement. NYU's stats program is extremely small -- it's in Stern, and they aim to have only 1-2 new students for the Statistics PhD concentration every year (just like most management doctoral programs). This suggests to me that they aim to place people in academia rather than industry?

Yeah, although my overall undergrad GPA was a little bit on the lower end, my grades in math/stat (and engineering and CS) classes were all A or A-. My undergrad also has a reputation for grade deflation, so that is likely why the overall undergrad GPA did not -completely- kill my chances.