Texas Stats PhD Programs

[Cavalerius]

I had a question about some of the trends in the rankings of stats programs in Texas, particularly those of Texas A&M, Rice, and Austin. I know that rankings are not always the best indicator of the value or strength of a program, but acknowledging that rankings are often backward looking, I did notice that in the last few iterations of the US News rankings, Texas A&M has been falling, such that now it is outside the top 10, if that means anything in itself. But I was wondering whether anyone knows of the reason for this decline, or whether it is just attributable to a degree of randomness in the rankings. It seems that Rice does not generally move much, staying right near the 30 mark. And does anyone know what to expect from Austin, given its recent introduction to the rankings. Would you expect the program to continue to gain recognition or to stay where it is? I ask this not only in terms of the rankings themselves but also in terms of the quality of the programs and the extent to which rankings might be some reflection of that quality. Any insight into this situation is greatly appreciated, and as these programs seem in general to be the three highest ranked programs in Texas, for those who wish to go to school in Texas, I hope that any insights shared might prove helpful to them as well.

Edited February 16, 2019 by Cavalerius

[bayessays]

Texas A&M, like Iowa State, is a very old and large program - they still turn out a decent number of grads who land good positions, but if you go there you will have to find one of the more well-known advisors and publish (as you would anywhere, of course). I suspect they are decreasing in the rankings because they have trouble recruiting top students and faculty to live in College Station, which should probably be your main factor in your decision. 

UT Austin has some of the biggest names in Bayesian stats, and if you go there and work hard with Scott or Carvalho or Walker as your advisor, you'll be in great shape for a good post doc - there first PhD grad got one with Michael Jordan at Berkeley. They've lost two faculty members including their department chair in the last couple years.

I don't know a ton about Rice, but none of their faculty have stuck out to me and I don't often see their grads as professors, but that may be because of their size.  I would say they're a step below the other two schools.

[Cavalerius]

Thanks, bayessays, for the analysis. I guess that College Station is not as desirable a location as Houston or Austin, but it seemed like a nice place to live, while still being somewhat close to some major cities. I really want to live in Texas or the southwest, which limited my options in selecting programs, but I thought that applying to the combination of Texas A&M, Rice, and Austin would provide a nice cross section with respect to program size, city type, and faculty research interests.

Edited February 16, 2019 by Cavalerius

[bayessays]

Yes there is a nice variety there. I looked a little more at Rice. Schweinberger publishes a lot in top journals, but most their other professors publishing in top journals are very young - they don't have the big names UT has.

My decision process, personally, would be like this:

1) Do you want to be a Bayesian? If so, go to UT, no question. If you definitely don't, don't go to UT.

2) if you don't care about the Bayesian thing either way, take a look at your political beliefs and preference for surrounding area.  I think environment is huge when you're making a decision on where to spend 5 years.  If you're a liberal hippie, go to Austin. If you're more conservative or like small towns, you might enjoy College Station. Houston is more moderate and a normal big city. 

3) Weigh the above factors with career goals. If you want to be a data scientist, go where you'll enjoy it the most, and look at the professor research interests. If you want to be a professor at a top department, I think going to UT and working with one of their top 4 guys has the highest ceiling, but I've looked at Rice's placement and they've put multiple people in TT jobs at PhD granting places recently, including TAMU.

[Schweinchen]

Texas A&M has had a well-known and long history in Stat but UT just started their Stat program recently. On the other hand, UT is very strong in machine learning. The caveat is that those ML faculty are all in CS and ECE, even Math, but not Stat. That graduate who went work for Jordan is likely a student of someone in CS or CSE.

[bayessays]

2 hours ago, Schweinchen said:

Texas A&M has had a well-known and long history in Stat but UT just started their Stat program recently. On the other hand, UT is very strong in machine learning. The caveat is that those ML faculty are all in CS and ECE, even Math, but not Stat. That graduate who went work for Jordan is likely a student of someone in CS or CSE.

No, he was a student of James Scott, one of the statistics professors. The department is new but the statisticians are not. UT has at least 4 elite, renowned Bayesians. If you work with them, it does not matter that the department is new. Out of their 4 PhD grads, 1 has gone into industry, and the other three were post-docs at Berkeley statistics, Princeton CS and UChicago biostatistics.

Edited February 16, 2019 by bayessays

[Stat Assistant Professor]

UT Austin is a great place to be a Bayesian. TAMU also has a few very strong faculty in Bayesian statistics as well, including Johnson, Mallick, Bhattacharya, and Pati. These professors consistently publish in top journals and have placed PhD graduates in top postdocs like Harvard and Duke.

Overall, I'd say that these are a few great schools you have to choose from. I personally love the city of Austin, so I would be biased towards UT Austin. But Rice and TAMU are great schools too.

Edited February 17, 2019 by Stat PhD Now Postdoc

[Cavalerius]

Thanks, Stat PhD Now Postdoc, for the feedback. Going back to the post that I created earlier on this forum about Bayesian vs. frequentist statistics, I am still deciding whether I would want to commit to a Bayesian perspective, but I have been reading Hoff's A First Course in Bayesian Statistical Methods, which the syllabi of the first-year UT Austin PhD statistics courses indicated as a good primer on the subject. The approach is very intriguing and does offer some interesting perspectives on statistics. The PhD statistics program at UT Austin is also interesting because according to the documents available on the website, about half of the core faculty are 50% appointments between mainly the business school and the new statistics department, an arrangement that is actually one of the desirable aspects of the program to me, and of course, I have heard that Austin is a great place to live: indeed, it was selected as US News' best place to live this past year.

Edited February 17, 2019 by Cavalerius

[Bayesian1701]

UT Austin is basically 100% Bayesian and very heavy on Bayesian nonparametrics.  It's fantastic if that's what you want to do but not so much if you aren't sure if you want to go Bayesian.  Austin is expensive, but the student usually get high paying tech internships (some in Austin).  Texas A&M has a Bayesian lean but has plenty of classical statisticians.  Texas A&M is also much larger than the other Texas programs.  

[Cavalerius]

Thanks for the response, Bayesian1701. I also read your post about your visits to both A&M and Austin, which helped give me some insight into the differences between the programs as well. I suppose since your interests were in Bayesian statistics, you didn't really look at Rice, which doesn't seem involved in Bayesian research. I think that I just need to spend more time considering the programs and how they fit with my research interests and overall objectives. Still, making such a big decision does not come easy, especially since it is easy to be concerned about what one's missing by not choosing certain programs.

[bayessays]

In response to Bayesian1701, Austin is expensive relative to College Station, but the student areas of town are really not that expensive. You can live on their stipend.

If it would calm your nerves a little, UT has at least one professor who publishes a lot and is 100% frequentist.  If you're open-minded to the idea of Bayesian statistics, it probably won't be a big deal.

Two other things to consider are size and coursework. TAMU will have more students which has pros and cons. I believe TAMU also has a more traditional course sequence which will include some pretty rigorous courses in theoretical statistics and measure theoretic probability, while UT has a more Bayesian oriented curriculum.

[Cavalerius]

Thanks, everyone, for the opinions and insights. It will be difficult to make a decision, so if I could add just one final question here, I would appreciate any further feedback.

 

Right now, I have narrowed down my choices to UT Austin, NCSU, UNC-STOR, and TAMU. Assuming that the funding is relatively equal given the cost of living, would choosing any one of these programs over the others provide better opportunities for the future, especially for academic jobs? I really like the small size and research focus of UT Austin, but given its newness, it is hard to tell what kind of program it will ultimately turn out to be, whereas NCSU and TAMU, though larger--with NCSU being considerably larger--are already established as top statistics departments and UNC-STOR's focus on applied probability and theoretical statistics appeals to my mathematical sensibilities--yet perhaps these subjects are not as favored as others these days.

 

(I know that the original topic was strictly about Texas programs, but I do not want to overlook other reputable programs. Nonetheless, all things being equal, I would still prefer to live in Texas!) 

Edited March 20, 2019 by Cavalerius