Stat Assistant Professor

My mistake. I thought you were referring to Indiana University of Pennsylvania ? ? Either way, this program does not seem to have any kind of national reputation as of now... so whether a degree from there will help in your job search (at a national level) may be one thing to consider.

For industry, you don't necessarily need to go to an Ivy or a top school to get a good job, but if you are aiming for a nationwide job search, the people hiring need to be familiar with the institution/program. My guess is that the graduates from IUPUI will mostly be recruited from companies in the Pittsburgh area (local companies are more willing to pluck graduates from nearby regional schools that don't have as much nationwide recognition). That's fine if you want to stay in the Pittsburgh area short-term, but your chances at getting a job in another location would most likely be better if you went to Emory. That's certainly a brand name school for big pharma. I would inquire the Emory Biostatistics department about research assistantships and other financial aid for Masters students. Some of the top Biostatistics programs in the country do offer partial tuition remission or a small stipend for Masters students who serve as RA's (you'll likely help out on projects and do coding tasks in R/SAS).

Goodness! IIRC, one of you guys went to Duke, took a bajillion graduate-level math and statistics classes and published several papers as an undergrad, yet only got waitlisted at Stanford? Another applicant that I've been messaging with personally on thegradcafe also went to Harvard and took graduate courses there, but got rejected from Stanford. They must really weigh the Mathematics Subject GRE more heavily than I had thought. Personally, I question the relevance of it for Statistics, but I guess when you have so many qualified applicants, it does serve as one method for paring down the list of applications.

Everyone's going to have a different opinion about this. See, e.g. the responses on this Reddit thread: https://www.reddit.com/r/AskAcademia/comments/73fo9w/cold_emailing_phd_students/ I don't think most PhD students would sneer at a cold e-mail from a prospective grad student, but a handful of them may not respond to a cold e-mail. However, if you go through the Graduate Coordinator, they will almost certainly find a current PhD student -- usually several -- who will respond to questions from the prospective student. That is why I would recommend going through that channel. My sentiments are similar to those made by one of the commenters on the Reddit thread I linked to above: "Best way to do it, in my experience, is to email the Department Chair or the faculty you're interested in working with (if you've already contacted them) and ask for grad student contact info to ask questions about the process. Most grad students would be more than happy to help, but it might seem a little more polite to go through an official channel."

I would not recommend sending unsolicited emails to PhD alumni or current PhD students directly. I would email the Graduate Program coordinator (who should be one of the professors in the department) and ask them to put you in touch with current PhD students. I have answered questions from prospective students at my PhD program, but it was only after the graduate program coordinator sent me an e-mail and CC'd the potential student in it, and I replied that I would be happy to answer any questions that they had about our program. As for deciding on a program, it may also be beneficial to visit the department after being admitted (if possible). Then you can talk directly with faculty and current students there. I know some people were dead-set on a certain program, but after they visited a few different ones that they had been admitted to, they changed their mind, because another program felt more "right" for them.

It matters most what journals or conferences they are publishing in and what their reputation in the statistics community is, not what their degree is in. For example, Michael Jordan from UC Berkeley is a very well-renowned researcher in both Statistics and Computer Science, even though his PhD is in Psychology/Cognitive Science -- Jordan later switched his research field from psych to mathematical statistics and machine learning... kind of random, but he was a smart guy, and I guess he taught himself statistics, because he pioneered some fields of statistics that are very active research areas (namely, in topic modeling, variational inference, and Bayesian nonparametrics). Michael Jordan's lab and his PhD students/postdocs consistently publish in top statistics journals like JASA, Annals of Statistics, etc. as well as a top CS conferences like NIPS, so working with Dr. Jordan would absolutely be very good for someone's job prospects (I know some of his former PhD students have gone straight from PhD to TT faculty at places like MIT). I would take a look at what venues these faculty members are publishing in and what the job placements of their former PhD students are.

I began my PhD at age 28 after working for some time, and I definitely think it was an asset for me personally. Years of work experience allowed me to mature and gain some real-world experience (not just work-wise but also with regard to "life skills" like budgeting, paying bills and loans, etc.), and it also gave me some more clarity about why I wanted to do the PhD. Anecdotally, most of the PhD attrition I saw was from students who enrolled in a PhD straight from a BS/BA, whereas slightly older students who had worked a few years were more likely to stick it out to the end. That said, there are other pros with going to a PhD program right out of college. You're still young so you likely don't have as many financial/family obligations and probably don't mind living on a graduate stipend for 4-6 years (in my case, it was a huge adjustment going from a well-paid industry job to earning about $25k a year as a grad student). *Most* PhD holders also do not end up working in academia (there are only a finite number of professorship jobs, and of course, there are other things like two-body problems, family/geographical constraints, etc.). There are a lot of people who finish the PhD and later decide academia is not for them, and that's quite alright too. The unemployment rate for a PhD in STEM is very low, so the chances are high that you will be able to get *some* job after getting a doctorate. So I might disagree with bayessays that you should only take the PhD offer if "you couldn't imagine yourself not getting a PhD and have an extreme passion to academic research or teach." I actually think it's perfectly fine to do a PhD for "fun" or with the explicit goal of going into industry afterwards, as long as you really want to do it *for yourself* (too many people are enamored by the perceived "prestige" of the degree, but the truth is that most people don't really care if you have a PhD, and if you work in a setting with a lot of other PhD's/researchers, the degree itself is not enough... you still have to earn their respect and demonstrate your competencies).

A couple things: 1) Even if a department leans more heavily in one direction (Bayesian vs. frequentist), your coursework will most likely expose you to "classical" statistics. Actually, at my PhD program (which leans more heavily towards Bayesian), most of the core PhD statistics classes (linear models, GLMs, inference, mathematical statistics, design of experiments) were taught almost exclusively from within the realm of "classical"/frequentist statistics, and the Bayesian courses were only offered as electives. I had to teach myself the Bayesian stuff for my research. 2) In your research, you will most likely need to do a ton of reading and surveying literature/books when you first start out, and then, it will be necessary to expose yourself to both frequentist and Bayesian strategies. Even though my research was on Bayesian statistics for high-dimensional data, I still had to learn all about LASSO, ridge regression, SCAD, MCP, etc. (all of which are frequentist methods for dealing with high-dimensional problems). If you want to learn something like Bayesian deep learning or Bayesian dimension reduction, you will almost certainly need to be somewhat familiar with neural networks, PCA, etc. from the frequentist perspective. Even if your thesis research is on Bayesian, you will no doubt have to learn the frequentist analogues and strategies for solving the same problems, and it is not that hard to pick up on it. 3) You can always change your research area and teach yourself a new area later on if you find that you prefer one approach over another. A PhD is just something to certify that you were able to conduct academic research at a bare minimum publishable level, but it does not prevent you from moving on to something totally different later. My PhD advisor's thesis research was on frequentist nonparametrics and he continued to do stuff like that for awhile, but in the late 1980s/early 1990s, he became a Bayesian. One of his former PhD students also no longer works on Bayesian statistics in her postdoc, but instead, on recommender systems and process data from a frequentist point of view.

I did my PhD research on Bayesian statistics and I am still continuing to work on that in my current postdoc. I think there are a lot of exciting developments and directions for research in this field, both theoretical and methodological/applied. In particular, a major direction of research right now is making Bayesian methods more computationally tractable and scalable to massive datasets (something that has traditionally been a challenge for implementations based on MCMC). This is pursued through various alternatives to traditional MCMC like variational inference, sequential Monte Carlo, etc. A lot of it is a matter of taste, I suppose. Some people like the Bayesian perspective, since you get a posterior distribution of the parameters and not just a point estimate. So you can get automatic uncertainty quantification from the posterior density *and* point estimates (posterior mean, median, or mode) under a unified framework. Whereas in "classical" statistics, the uncertainty quantification is based on long-term coverage rather than on actual probabilities (e.g. "if we repeated this experiment a large number of times and constructed these confidence intervals, 95% of the confidence intervals would contain the true parameter").

Thanks for the clarification. University of Chicago is well-known for grade deflation and rigor, so even something in the range of 3.5 from UChicago looks good (if it were below a 3.0, that would be a different story though). The same applies to schools like MIT and Caltech. Adcoms know these schools are very tough but also much more intellectually rigorous than average. I have no doubt that you were well prepared for your PhD program if you studied math at Uchicago and had around a 3.5. So no need to deflate your accomplishment!

It is not impossible to go straight from PhD to tenure-track faculty, but in order to get a job at an R1, you almost always need one or two publications accepted in top journals (think: JASA, JRSS-B, Biometrika, Annals of Statistics, Annals of Probability, Biometrics). The successful job candidates I have seen who went straight from their PhD to TT jobs usually had two or three in press in the aforementioned venues by the time they were on the job market. This is quite an impressive feat for a PhD student, but a handful of Stanford, Chicago, Berkeley, Columbia, UPenn Wharton, Michigan, etc. grads really do accomplish this (they really are *that* good). I have seen some PhD graduates get jobs right out of their doctorate at teaching colleges or regional state schools without top publications though. Usually one or two pubs is still preferred but they don't necessarily need to be in top journals. If you are someone who loves teaching more than research, these are good institutions to look for jobs at.

Nah. For STEM-related majors, average earnings don't vary much among the college categories. There is definite correlation between prestige of degree granting institution and salary for some fields, e.g. law or MBAs, but not usually for STEM. As bayessays pointed out, there are some hedge funds and consulting companies that may be biased towards hiring Ivy League/Stanford/MIT/UChicago/etc. PhD graduates, but I would say these places are the exception rather than the rule in industry.

Agreed with this. As long as the people hiring are somewhat familiar with your PhD institution (this would encompass most of the top public universities in the country, in addition to the top private universities), chances are good for at least getting an interview. People from my PhD department were able to get interviews for data scientist/statistician jobs at Netflix, Apple, Google, Amazon, Lyft, and all those kinds of companies. Chances might not be as good if your PhD is from a more regional school.

Based on anecdotal evidence and if that is your preference for future employment, I don't think the reputation of the program itself will matter as much then (to a point... I think any top 60ish stat/biostat program will probably be fine). So you can feel free to attend the program where you feel the most comfortable and feel is the right "fit" for your personality. My PhD alma mater is a smaller department (though it has been growing quite a bit in recent years) that is about the same rank as UIUC, and there are PhD alumni from my department who are currently working at Apple, Amazon, and Google as data scientists or research scientists. It seems like it is not a big problem to attend a school like UIUC if that is your ultimate goal.

NC State and Wisconsin are better programs and among the best in the country. In my personal opinion, UIUC is comparable to PSU and UCLA though, and certainly a quality department. I think if you are choosing between these schools, you may want to consider whether you would prefer attending a smaller PhD program or a bigger one. Each has its pros and cons.

I think it is a solid department, with some really good professors who have great PhD placement: Annie Qu, Xiaofeng Shao, and Yuguo Chen are particularly good and publish consistently in AoS, JASA, JRSS-B, Biometrika, etc. Qu and Chen seem to be doing particularly interesting current stuff on network analysis and recommender systems. It is a bit smaller department but definitely solid IMO.

Agreed with this assessment. Carvalho, Scott, and Walker are all very well-known and renowned in the field of Bayesian statistics, so having any of them as an advisor gives you a very good shot at landing a top postdoc at a world-class institution (e.g. Duke or Columbia).

Either of those would look good on a transcript, so take the one that interests you more. In either case, you would get a lot of good practice with doing mathematical proofs, which will be an essential skill in at least the coursework phase of a PhD program (if you end up doing something more "applied" for your thesis research, you might not need to bog down in so many theorems, but a lot of the core courses tend to be theoretical). I don't think you need to know much about either for statistics research... you do sometimes see imaginary numbers and Cauchy's integral formula (namely stuff related to Fourier series), and some concepts from topology, like countability and compactness (namely as they apply to parameter spaces), etc., but I wouldn't say a very deep understanding of either complex analysis or topology is needed in most cases. However, the technical skills gained from doing proofs in these classes will definitely be helpful.

UF lost a lot of bigshot professors to retirement/death in the early 2010s, e.g. Casella, Agresti, Randles, Linda Young, Daniels, etc., so it became a smaller department whose main strengths were in niche areas (specifically Markov Chain Monte Carlo theory). However, recently, it has been on the rise again. Daniels returned to be the Department Chair, the department hired Michailidis from Michigan who heads the UF Informatics Institute, and the department has been hiring a lot of new faculty in machine learning and data science. I would put UF currently in the same neighborhood as schools like UCLA, UIUC, and Rutgers.

Also wanted to add: 1) You can look at Alumni Placements for most schools that interest you to see how their PhD graduates fare. If this information is not available on the Departments' websites, you can e-mail the PhD program coordinators who should be happy to provide you with this info. 2) If you're interested in stochastic processes, it may be worthwhile to look into Applied Mathematics and Operations Research & Industrial Engineering PhD programs as well. Some graduates of Math or ORIE programs do become faculty in Statistics departments (e.g. Princeton's Operations Research PhD program and Cornell's ORIE program have both placed their alumni in Statistics postdocs/Assistant Professorships).

Most of the top-tier Statistics programs have at least a couple very strong faculty who work in probability theory: Penn Wharton, Stanford, Berkeley all have some faculty who work in probability, e.g. Diaconis at Stanford, Bhattacharya at Penn Wharton, and Pitman at Berkeley. Below that, UNC-STOR is particularly strong in stochastic processes and applied probability (e.g. Bhadimi and Budhijara). For Markov chains, University of Florida is a world leader department in the area of MCMC theory, with world-renowned faculty who work on convergence rates and complexity analysis of Markov chains (e.g. Hobert and Doss).

Those are all good programs, and NCSU is definitely a top tier program for Statistics. That is reflected in the USNWR rankings as well. These schools may have less "name recognition" overall than say, the Ivies, Stanford, Berkeley, Chicago, etc., but in the field of Statistics, they are well-regarded. When it comes to certain academic disciplines, individual programs can be more reputable than the school as a whole. For instance, University of Pittsburgh is a top 3 school for Philosophy and has academic placements on par with Ivy schools or UChicago, even if UPitt's overall rankings are not as high. Also, the PhD advisor's reputation is more important. Students of any distinguished professor from any program should be in good shape to get a good postdoc. The top programs will have, on average, more professors who are leaders in their field, but the mid-to-lower ranked programs also have some good distinguished professors who are leaders in their respective fields (off the top of my head, I can think of Dr. Marianna Pensky at UCF, Cun-Hui Zhang at Rutgers, Peter Mueller at UT-Austin, etc.)

cyberwulf makes good points. If I were hypothetically applying to Stat PhD programs again and were admitted to a Harvard, Berkeley, UChicago, Penn Wharton, CMU, etc. (or a JHU, Harvard, UWashington for Biostat), I would probably be inclined to go with one of those. Below the very top tier, I would probably be a little bit more choosy though. For instance, I quite admire some of the smaller programs which are quite strong, in my opinion, like UIUC, Rutgers and Yale -- they have some truly pioneering faculty like John Lafferty and Cun-Hui Zhang. I do not know for certain that I would pick a more "prestigious," large state school program over one of these smaller but very solid programs. Just my opinion though. Others may disagree, and of course, it is always easier to evaluate things in hindsight after you have gained more experience and knowledge of the field. ?

Prestige is something you should certainly take into account but it is not the only thing. It is probably a good idea to visit the programs you are admitted to in order to get a sense of where you would be most comfortable. In addition, even if a program is ranked "below" the top tier, it likely has niche strengths that are very strong, e.g. Rice University and UCSB Statistics are very strong for financial mathematics, while OSU and University of Missouri are both strong in spatial statistics. So if you are interested in a niche area, it makes more sense to go with the program that has particular strengths in that area. In addition, your PhD advisor matters more than the program. I attended a mid-tier program (in terms of overall USNWR rankings), but I worked with one of the most distinguished professors who has great PhD placements... former PhD students of his are now faculty at Duke, TAMU, University of Georgia, etc., and from 2012-present, this PhD advisor has had his students taks postdoc appointments at Stanford, Carnegie Mellon, Columbia, Penn, etc. So while the prestige of the institution is something to take into account, your success is ultimately up to you. Your publications and recommendation letters are what carry the most weight in academic hiring for Statistics.

Nope, I think you are in very good shape. Having degrees from ICL and Oxbridge and having already completed so much advanced math coursework will certainly work in your favor. I wouldn't be surprised if you finish your PhD in four years. Best of luck.