Thoughts on new rankings for Stats programs?

[gainer19]

Anyone have any initial thoughts or impressions on the new U.S. News rankings for statistics programs (no longer a math specialty) released this morning? I know rankings are just one of many things to take into consideration, but I still find them informational. I was kind of surprised to see that UNC's stats department (#22) was ranked noticeably lower than their biostats department (#10) [schools with separate departments were ranked separately]. I was also kinda surprised to see Florida's stats program ranked 27th, since they were just ranked in the top 10 in the previous rankings. Your thoughts?

[sciencegal]

Wow. These rankings are pretty bizarre to me. For people intimately familiar with these programs, departments, and the field will likely disagree with a lot of the placements. The problem with rankings, of course, is what they use to calculate the score. For so many of these programs their research innovation is practically zero and yet they are in the top 10 or 20.

[rooster34]

I really didn't see that many surprises; perhaps Florida being 27th after being in the top 10 recently. Of course, part of that is that, to my knowledge, they've never separately ranked the biostat and stat programs. Doing so inflated the number of programs being ranked and, as such, pushed some "top" programs down. As for UNC, my understanding has always been that the biostat is much stronger than the stat program. I will be interested in seeing how far UW drops over the next few years if the financial situation does not improve.

As for the methodology, I couldn't find just how things were done. The link I clicked on took me to an old explanation. And I was disappointed that we weren't provided the treasure trove of information like that in the engineering rankings.

Finally, it is dangerous to just look at the rankings. For example, Ohio State is 27th (3.5) but is just 0.1 from being tied for 22nd (3.6) and, I believe, 0.2 from being tied for 20th (3.7). Point being, there is not much difference between spots.

[statsguy]

On 4/15/2010 at 8:55 PM, rooster34 said:

I really didn't see that many surprises; perhaps Florida being 27th after being in the top 10 recently. Of course, part of that is that, to my knowledge, they've never separately ranked the biostat and stat programs. Doing so inflated the number of programs being ranked and, as such, pushed some "top" programs down. As for UNC, my understanding has always been that the biostat is much stronger than the stat program. I will be interested in seeing how far UW drops over the next few years if the financial situation does not improve.

As for the methodology, I couldn't find just how things were done. The link I clicked on took me to an old explanation. And I was disappointed that we weren't provided the treasure trove of information like that in the engineering rankings.

Finally, it is dangerous to just look at the rankings. For example, Ohio State is 27th (3.5) but is just 0.1 from being tied for 22nd (3.6) and, I believe, 0.2 from being tied for 20th (3.7). Point being, there is not much difference between spots.

That's the one thing about Stats departments: there are the 5-6, top-tier schools (Stanford, Harvard, UChicago, UWashington, Berkeley etc...) and then there is a huge cluster of very good programs (in these rankings, they would be the schools rated approx. between a 3.5 and 4.0) It's really hard to differentiate the schools in that area. What was suprised me was NC State's ranking, which, after removing the preceeding Biostats rankings, comes in around #7.

I'm really glad that USNews seperated Stats and Biostats, although it would have been nice had they created two seperate lists.

By the way, congratulations on OSU. I actually recieved the first year fellowship their when I applied, but did not end up going there. A really solid department.

[sciencegal]

I did this super quickly so hopefully no heinous mistakes but I separated them by stat and biostat through scores 3.5 just because I was curious as to the separated lists.

I echo the above sentiments about using the rankings as just one piece of the pie. Especially when a large portion of the score is based on a survey sent to academics and their opinion of other programs: http://www.usnews.com/articles/education/best-graduate-schools/2010/04/15/the-science-rankings-methodology.html

STAT

1. Stanford 4.9

2. UC Berkeley 4.7

3. Harvard 4.3

3. University of Chicago 4.3

3. UW 4.3

6. Carnegie Mellon 4.1

7. Duke 4.0

8. NC State 3.9

8. Texas A&M 3.9

8. University of Pennsylvania 3.9

8. University of Wisconsin 3.9

12. University of Michigan 3.8

12. University of Minnesota 3.8

14. Iowa State 3.7

14. Pennsylvania State University 3.7

16. Columbia 3.6

16. Cornell 3.6

16. Purdue 3.6

16. UNC-CH 3.6

20. Ohio State 3.5

20. UCLA 3.5

20. University of Florida 3.5

BIOSTAT

1. Harvard 4.6

1. UW 4.6

3. Johns Hopkins 4.4

4. UNC-CH 4.0

5. University of Michigan 3.9

6. UC Berkeley 3.8

7. University of Minnesota 3.6

8. University of Wisconsin-Madison 3.5

[rooster34]

I did this super quickly so hopefully no heinous mistakes but I separated them by stat and biostat through scores 3.5 just because I was curious as to the separated lists.

I echo the above sentiments about using the rankings as just one piece of the pie. Especially when a large portion of the score is based on a survey sent to academics and their opinion of other programs: http://www.usnews.co...ethodology.html

STAT

1. Stanford 4.9

2. UC Berkeley 4.7

3. Harvard 4.3

3. University of Chicago 4.3

3. UW 4.3

6. Carnegie Mellon 4.1

7. Duke 4.0

8. NC State 3.9

8. Texas A&M 3.9

8. University of Pennsylvania 3.9

8. University of Wisconsin 3.9

12. University of Michigan 3.8

12. University of Minnesota 3.8

14. Iowa State 3.7

14. Pennsylvania State University 3.7

16. Columbia 3.6

16. Cornell 3.6

16. Purdue 3.6

16. UNC-CH 3.6

20. Ohio State 3.5

20. UCLA 3.5

20. University of Florida 3.5

BIOSTAT

1. Harvard 4.6

1. UW 4.6

3. Johns Hopkins 4.4

4. UNC-CH 4.0

5. University of Michigan 3.9

6. UC Berkeley 3.8

7. University of Minnesota 3.6

8. University of Wisconsin-Madison 3.5

From the above linked article on methodology:

Response rates for the doctoral Ph.D. sciences were as follows: for biological sciences, 15 percent; chemistry, 25 percent; computer science, 46 percent; earth sciences, 29 percent; mathematics, 34 percent; physics, 31 percent; and statistics, 67 percent.

No surprise that the stat departments response rate is the highest!

[floffel]

It seems to me this is as good as any list is going to be. Of course there are slots that seem a bit puzzling, and depending on someone's specific interests the list could look very different. But every program on that list is a good program, and I can't off the top of my head think of a great program that does not show up there.

One thing that's for sure is that UW's funding problems for Stat for this coming year is going to hurt them in the next edition of these rankings. It's such a shame considering how great that department is.

[sciencegal]

I guess my initial surprise/gruff reaction comes from a research perspective. As a science person using new methods developed in stat/biostat some programs chugging along on the same tired out dated methodology are ranking highly (higher than previous). I also researched and visited a lot of these places as I considered these fields for graduate school when I applied a few years ago, and have friends in the various departments. I understand that previous and current innovative highly regarded research doesn't necessarily mean you won't get outranked on a list like this, and really, once I broke it out into the separate lists it wasn't as jarring. For the most part I didn't see any school I would have moved up or down more than 5 places, and you can't really expect more than that I guess (especially with all the ties).

I agree about UW, I don't know what it'll do to their rankings but I was shocked to hear about their crazy funding issues this year. Who were they able to get? Last year I had friends getting great packages from them in stat/biostat. Packages that were 10K better than other publics & privates offered them!

[blabla]

I find this list more than a little odd. Columbia's ranking is just absurd.

[origin415]

If they split stat and biostat departments, why aren't they splitting math and applied math departments? :/

[sciencegal]

I find this list more than a little odd. Columbia's ranking is just absurd.

also- Rutgers and Yale not in the top 20 for stat, interesting? Historically they have been ranked in the top 20 and generally have had reputations that warranted that.

[ssong2]

also- Rutgers and Yale not in the top 20 for stat, interesting? Historically they have been ranked in the top 20 and generally have had reputations that warranted that.

I agree. They both have relatively smaller department, but well respective programs.

[sciencegal]

I agree. They both have relatively smaller department, but well respective programs.

I don't know *as* much about Yale's program, but Rutgers, again focusing on research, is doing stuff with estimating equations and semi-parametric estimation that is superior to a lot of the programs that made the top 20. I know this is one particular research area, but just to give an example. They have many notable professors. Surprising.

I hope this list doesn't hurt good programs. I meet so many students that focus heavily on rankings. Look at placement of students as a better indicator of reputation, or as another indicator of reputation!

I'll be interested to see if the rankings differ significantly the next time they do it, or if they modify their metric slightly.

[kash]

I find this list more than a little odd. Columbia's ranking is just absurd.

Why do you find it odd, should it be ranked higher?

[ajaxbel]

This might already have been mentioned, but this page describes how the rankings were done:

Also, for the first time, U.S. News is ranking Ph.D. programs in statistics. These rankings are based solely on the rating of academics at statistics and biostatistics departments.

I myself am looking into Masters programs in Statistics but now that I realize the US News rankings are only for PhD programs, I have to continue searching for some sort of guidance on which Masters programs are good.

Edited May 10, 2010 by ajaxbel