Stat PhD: UC Davis vs. Rutgers vs. Rice

[Aweaston]

Hi everyone,

As of yesterday, I've heard back from most of the schools I applied to. It is exciting to have been accepted to these amazing schools. My undergraduate major is Statistics and really want to learn something in Ph.D. training. I am very confused about these three programs as they rank very close and all have profs I would like to work with. During my undergraduate, I have been exposed to research about high-dimensional data, network analysis, and statistical inference. My research field is not that clear but at this stage, I prefer high dimensional data and network analysis. I am not sure how well I could do so I am not sure I could go to academia or industry. But if I do well in academics, I prefer to being a professor.

 

UC Davis

Pros:

1. rank highest among the three programs 

2. prestigious profs such as Hans-Georg Müller and J.L. Wang

3. it's California, although the corner 

Cons:

1. I could not find the placement of Ph.D. students so I'm not sure if they have a good way out

2. the research is too theoretical

Rutgers

Pros:

1. talked to one prof from Rutgers, got to know that the size of the program is relatively small (no more than ten Ph.D.) and the rank of it is significantly underestimated

2. it's New Jersey and close to NY

Cons:

1. way out to academia is not very satisfying

2. I know it's naive but the overall ranking of Rutgers is relatively low

Rice

not very familiar with this program and can not find much information about it

Pros:

1. Houston is a big city and the life there is more comfortable

Cons:

1. research area is relatively narrow as there are only 12 faculty members

 

I have confidence in finishing the Ph.D. training in either school. However, since my research area is not settled and I am not sure about academia or industry, I really need your advice, especially the details about the three programs and way out of them! Any advice or perspective is greatly appreciated. Thank you very much!

Edited March 20, 2019 by Aweaston

[hnn12]

I am in the same situation without Rutgers. As to the placements of PhD graduates at UC Davis, I asked the program coordinator and she gave me a list of placements last year. It seems like most people (80-90%) go into industry. I prefer to work in industry after graduation, so the placements at UC Davis is very appropriate for me, but Rice seem to do equally well with regards to this aspect. UC Davis is close to Sacramento, which is a much smaller city, but it only takes less than 2 hours of drive to get to San Francisco and the tech hub in Palo Alto. However, Rice has the edge for me with regards to location because I am kinda a big-city boy and Houston is nice yet low-cost city. There are many outstanding professors that I want to work with at UC Davis (Thomas Lee for one), but Rice has Genevera Allen, who is a rising star in the field (and her advisor is Robert Tibshirani from Stanford). The thing I like most about Rice is that the department allows well-prepared PhD students to take the qualifying exam right at the beginning of the first year. The exam is based on Casella & Berger, and there is a high chance that you can finish your PhD within 4 years if you pass it. Overall, I am leaning towards Rice slightly, but UC Davis gave me a really good financial package. And I have other universities in the equation, too.

[Geococcyx]

I'm just another prospective student trying to decide where to go, but I can probably provide a few details of varying quality (the actually good details are secondhand from older posters, but if I mention it now they may expound on it later):

Rutgers has Cun-Hui Zhang, who appears to do work in model selection and high-dimensional data (or at least used to -- I don't think they have a Google Scholar profile, so I'm just seeing their biggest hits).  Dr. Zhang has several really strong papers in Annals of Statistics, has something in PNAS, and so on, so I doubt you'd have too many issues in the academic job market if you worked with them.  For what it's worth, my statistics adviser spoke pretty strongly in favor of my applying to Rutgers, and he was mostly encouraging me to apply to places like Duke, Columbia, and UCLA statistics, so Rutgers is outpunching its supposed weight, ranking-wise.  (Credit to Stat PhD Now Postdoc for mentioning Dr. Zhang in previous comments about Rutgers)

I've heard Rice is strong in financial statistics, which would probably lead to pretty strong industry placements.  I don't know to what extent it would really effect the program, but I know Rice stat and MD Anderson biostat (which is right across the street from Rice) allow for some class crossover, which might(?) extend research areas if they allow collaborations with professors from the other school.  Hadley Wickham is also an occasional lecturer there, if you care.  As a mostly unnecessary note, I'm not sure that I'd rate living in Houston as more enjoyable than quasi-Sacramento or New Brunswick due to my impression of Houston's urban sprawl, but I can't claim too much expertise or experience about what it's like to live in any of these cities.

 

Edited March 20, 2019 by Geococcyx

[Stat Assistant Professor]

Second the comment about Cun-Hui Zhang who has done a lot of great work on high-dimensional regression (both parametric and nonparametric). He's the inventor of a popular alternative to the lasso for penalized/regularized regression called the MCP (minimax concave penalty). Hans-Georg Müller at UCD is also prominent for functional data analysis, a field I have been doing some research on in recently and which has some interesting applications.

I think all of programs you listed have more graduates going to industry than academia, but it doesn't really matter if your PhD research was theoretical or applied if you want to go the industry route. If you want to go into academia, it usually helps to be at least somewhat theoretical, even if your primary interest is methodological/applied. This is the case for both Statistics and Biostatistics. For Biostatistics, unless your field is something like RNA sequencing, medical imaging data, etc. (which would lend itself to publication in mainly in journals like BMC Bioinformatics, Nature Methods, etc.), a publication in a statistics journal like Biometrika, Biometrics or JASA gives you an edge in the hiring process (and papers in those venues frequently contain at least one or two theorems).

[Aweaston]

10 hours ago, hnn12 said:

I am in the same situation without Rutgers. As to the placements of PhD graduates at UC Davis, I asked the program coordinator and she gave me a list of placements last year. It seems like most people (80-90%) go into industry. I prefer to work in industry after graduation, so the placements at UC Davis is very appropriate for me, but Rice seem to do equally well with regards to this aspect. UC Davis is close to Sacramento, which is a much smaller city, but it only takes less than 2 hours of drive to get to San Francisco and the tech hub in Palo Alto. However, Rice has the edge for me with regards to location because I am kinda a big-city boy and Houston is nice yet low-cost city. There are many outstanding professors that I want to work with at UC Davis (Thomas Lee for one), but Rice has Genevera Allen, who is a rising star in the field (and her advisor is Robert Tibshirani from Stanford). The thing I like most about Rice is that the department allows well-prepared PhD students to take the qualifying exam right at the beginning of the first year. The exam is based on Casella & Berger, and there is a high chance that you can finish your PhD within 4 years if you pass it. Overall, I am leaning towards Rice slightly, but UC Davis gave me a really good financial package. And I have other universities in the equation, too.

Thanks! Have you joined the open house which was held on Feb 25? Maybe we have met with each other already lol. Yes, both UC Davis and Rice have most of their Ph.D. going to the industry. I have looked through the placement of Rice Ph.D. (https://statistics.rice.edu/doctoral_alumni). But I'm not sure the placement is good enough or not as there is no placement data about UC Davis. My main concern is whether Houston is a good choice if you prefer to work in the industry after graduation as Houston is famous for petroleum industry rather than high tech.

Dr. Genevera Allen is a potential advisor if you want to do research on statistical machine learning, but current Ph.D. students told me that it's competitive to join her lab. Hope the information helps.

[Aweaston]

9 hours ago, Geococcyx said:

I'm just another prospective student trying to decide where to go, but I can probably provide a few details of varying quality (the actually good details are secondhand from older posters, but if I mention it now they may expound on it later):

Rutgers has Cun-Hui Zhang, who appears to do work in model selection and high-dimensional data (or at least used to -- I don't think they have a Google Scholar profile, so I'm just seeing their biggest hits).  Dr. Zhang has several really strong papers in Annals of Statistics, has something in PNAS, and so on, so I doubt you'd have too many issues in the academic job market if you worked with them.  For what it's worth, my statistics adviser spoke pretty strongly in favor of my applying to Rutgers, and he was mostly encouraging me to apply to places like Duke, Columbia, and UCLA statistics, so Rutgers is outpunching its supposed weight, ranking-wise.  (Credit to Stat PhD Now Postdoc for mentioning Dr. Zhang in previous comments about Rutgers)

I've heard Rice is strong in financial statistics, which would probably lead to pretty strong industry placements.  I don't know to what extent it would really effect the program, but I know Rice stat and MD Anderson biostat (which is right across the street from Rice) allow for some class crossover, which might(?) extend research areas if they allow collaborations with professors from the other school.  Hadley Wickham is also an occasional lecturer there, if you care.  As a mostly unnecessary note, I'm not sure that I'd rate living in Houston as more enjoyable than quasi-Sacramento or New Brunswick due to my impression of Houston's urban sprawl, but I can't claim too much expertise or experience about what it's like to live in any of these cities.

 

3

Thanks for the information! It does help. Cun-Hui Zhang is prestigious in the areas you listed. But when I look through the Statistics faculty of American universities, very few of them graduates from Rutgers (actually, I have never seen faculty who graduated from Rutgers Statistics). So I wonder whether I should expect too much concerning way out to academia if I choose Rutgers.

Rice Stat is closely associated with the MD Anderson Biostat and profs from two departments have close collaborations. Ph.D. of Rice Stat have the chance to be co-advised by Profs from Rice and MD Anderson, which I think is a strength of Rice Stat.

[Aweaston]

2 hours ago, Stat PhD Now Postdoc said:

Second the comment about Cun-Hui Zhang who has done a lot of great work on high-dimensional regression (both parametric and nonparametric). He's the inventor of a popular alternative to the lasso for penalized/regularized regression called the MCP (minimax concave penalty). Hans-Georg Müller at UCD is also prominent for functional data analysis, a field I have been doing some research on in recently and which has some interesting applications.

I think all of programs you listed have more graduates going to industry than academia, but it doesn't really matter if your PhD research was theoretical or applied if you want to go the industry route. If you want to go into academia, it usually helps to be at least somewhat theoretical, even if your primary interest is methodological/applied. This is the case for both Statistics and Biostatistics. For Biostatistics, unless your field is something like RNA sequencing, medical imaging data, etc. (which would lend itself to publication in mainly in journals like BMC Bioinformatics, Nature Methods, etc.), a publication in a statistics journal like Biometrika, Biometrics or JASA gives you an edge in the hiring process (and papers in those venues frequently contain at least one or two theorems).

4

Thanks for your advice! Yes, the three programs all have more graduates going to industry than academia. So maybe I should think about the type of potential industry job I'll seek (such as tech, IT, finance), which would help me make a decision lol.

[Stat Assistant Professor]

15 minutes ago, Aweaston said:

Thanks for your advice! Yes, the three programs all have more graduates going to industry than academia. So maybe I should think about the type of potential industry job I'll seek (such as tech, IT, finance), which would help me make a decision lol.

I wouldn't infer any causation here (i.e. attending one of the three schools causes students to go into industry). There may be self-selecting bias, where the students who enroll at these programs had a preference for industry to begin with. Even at some higher ranked programs like Harvard and NCSU, a majority of the PhD graduates go into industry. That said, everybody needs a "Plan B," especially those who go the academic route. If you can't get an academic job after two postdocs, you should definitely start planning your exit strategy to industry or government.

Edited March 20, 2019 by Stat PhD Now Postdoc