Lack of Statistics Classes/Research

[undergrad1]

I am an undergraduate majoring in pure mathematics, and am considering applying to graduate school in statistics. The thing is that I have not done any coursework or research in statistics. I have only taken graduate statistical theory class and my research was focused primarily on probability theory and random processes. How is this situation going to be viewed by the admissions committee?

[Noegenesis]

3 minutes ago, undergrad1 said:

I am an undergraduate majoring in pure mathematics, and am considering applying to graduate school in statistics. The thing is that I have not done any coursework or research in statistics. I have only taken graduate statistical theory class and my research was focused primarily on probability theory and random processes. How is this situation going to be viewed by the admissions committee?

If you are equally qualified as another similar applicant, the applicant with more statistics courses will be considered more favorably than you. There are plenty of statistics courses online. Is there any way you can strengthen your application?

[Geococcyx]

I'm just another student, so be cautious taking my advice, but I think you're probably fine --more often the issue with stat applicants is that they haven't taken Real Analysis, so your pure math background is more of a positive if anything.  Maybe biostat programs would care a little more, but probably not a lot.  Most applicants haven't done theoretical research, either, so your research might be a positive, although its hard to say from so little information.

Edited March 23, 2019 by Geococcyx
Projecting my own real analysis-taking proclivities onto others

[undergrad1]

On 3/22/2019 at 9:58 PM, Noegenesis said:

If you are equally qualified as another similar applicant, the applicant with more statistics courses will be considered more favorably than you. There are plenty of statistics courses online. Is there any way you can strengthen your application?

Do you think the adcoms really care about the online courses? I just thought that those do not really match the graduate coursework in terms of rigor and depth, and as such would not be given much consideration. 

On 3/22/2019 at 10:00 PM, Geococcyx said:

I'm just another student, so be cautious taking my advice, but I think you're probably fine --more often the issue with stat applicants is that they haven't taken Real Analysis, so your pure math background is more of a positive if anything.  Maybe biostat programs would care a little more, but probably not a lot.  Most applicants haven't done theoretical research, either, so your research might be a positive, although its hard to say from so little information.

Do you think that applicants with strong statistics background and minimal number of math courses, say Real Analysis, would be favored over those, who have done math in undergrad, and took one or two stats courses?

[Stat Assistant Professor]

14 minutes ago, undergrad1 said:

Do you think that applicants with strong statistics background and minimal number of math courses, say Real Analysis, would be favored over those, who have done math in undergrad, and took one or two stats courses?

There is no way to really answer this without looking at the whole package "holistically." If the statistics student has very strong research experience (even if it's not in statistics specifically, but in something like computational biology or computer science) and stronger letters of recommendation, then the statistics student would certainly look better. If the Statistics major went to say Stanford, while the math major went to an unknown school, then the former would also be preferred. GPA in upper division major classes is also a factor.

However, assuming all other things being equal besides the coursework (alma mater about the same level of prestige, similar overall GPA and GRE scores, similarly strong letters of recommendation), then the applicant with a strong math background but only 1-2 stats courses would typically look better to adcoms than the applicant with a lot of stat classes but minimal math classes. A lot of Statistics PhD programs have something written on their websites' FAQ's to the effect of "Although some exposure to undergraduate probability and statistics is helpful, skills in mathematics and computing are more important."

Edited March 27, 2019 by Stat PhD Now Postdoc

[bayessays]

On 3/22/2019 at 8:53 PM, undergrad1 said:

I have not done any coursework in statistics

...

I have only taken graduate statistical theory class

Which is it? I don't understand.  Have you never taken a statistics course or have you taken graduate statistics (way beyond what most applicants have)?

Applicants should have a probability glad and an intro statistics class.  You have probability and a graduate statistics class. You're gold!

1 hour ago, undergrad1 said:

Do you think that applicants with strong statistics background and minimal number of math courses, say Real Analysis, would be favored over those, who have done math in undergrad, and took one or two stats courses?

Absolutely, 100%, no. Undergraduate statistics classes are a joke and those classes will barely help when you get to grad school. The relatively large number of statistics majors in PhD programs is self-selection bias of people who like statistics.

Your research focusing on probability is a good thing! Nobody does statistics research as an undergrad. There are like 10 applicants in the country in a given year applying who have done anything resembling real statistics research.

Do not take an online statistics course unless you want to for fun.

[galois]

I was in your situation this past cycle, but more severe. The only stats course I took was 101 freshman year. The rest of my courses were pure math, like logic, algebra, and analysis, not even probability! I still got into some great programs. At visit days I learned that it's far more common to have a pure math background, although typically students have seen some level of mathematical statistics, so I do think I'm still in the minority, but I'm not quite as odd of an applicant as I initially thought. A professor at UW explained to me "we look for people that can learn statistics", as opposed to people that know statistics.

 

So, a background in pure math + graduate statistical theory class + research in probability and random processes actually makes you an ideal candidate from what I can tell.

[YF1999]

On 3/28/2019 at 1:14 PM, galois said:

I was in your situation this past cycle, but more severe. The only stats course I took was 101 freshman year. The rest of my courses were pure math, like logic, algebra, and analysis, not even probability! I still got into some great programs. At visit days I learned that it's far more common to have a pure math background, although typically students have seen some level of mathematical statistics, so I do think I'm still in the minority, but I'm not quite as odd of an applicant as I initially thought. A professor at UW explained to me "we look for people that can learn statistics", as opposed to people that know statistics.

 

So, a background in pure math + graduate statistical theory class + research in probability and random processes actually makes you an ideal candidate from what I can tell.

i am wondering what kind of math background is needed? i only took two optimization, one discrete math, one ODE, two calcular,  one linear algebra, and plan to take phd level probability next term, do i need to take more math courses?

i only learned analysisi by myself but take no class due to limited units

[MathStat]

Hi @undergrad1, I had the exact same concerns when I applied. My most meaningful research was also in probability theory and random processes and my coursework was very math heavy. I had only taken a handful of statistics courses and I can certainly say my statistics background was lacking. Btw, it is perfect that you took grad-level statistical theory!

Anyway, my applications were pretty successful (never would have expected the outcomes), but my research interests are also in theoretical/mathematical stat. So if you're planning to apply to very theoretical programs, I think you will be fine! And I also think it is fine, even if your interests end up being more applied. One thing to make sure is that you have a good coding background. 

Edited April 26, 2019 by MathStat

[galois]

12 hours ago, YF1999 said:

i am wondering what kind of math background is needed? i only took two optimization, one discrete math, one ODE, two calcular,  one linear algebra, and plan to take phd level probability next term, do i need to take more math courses?

i only learned analysisi by myself but take no class due to limited units

Well I might not be the best person to answer this question, but I'll give it a whack. I would say things like calc/linear-algebra fall into the "prerequisite"s that of course you need to have taken, and things like optimization/ODE, while not prerequisites, are almost "nice to have prerequisites", in that it doesn't hurt to be exposed to some of those computational concepts. I think when adcoms like seeing "math" they are referring to rigorous, proof-intensive courses that you would generally take as a junior or senior, things like real-analysis/abstract-algebra come to mind. Obviously, real analysis being the more important of the two, since that is actually relevant to the theoretical courses you will take in graduate statistics. But, even though something like abstract algebra probably isn't relevant at all to what you are studying, getting an A in such a class would demonstrate your ability for very abstract problem solving and theorem proving.

[BL250604]

I echo everyone's sentiments wholeheartedly. Research doesn't matter- don't worry about that. I truly am an outlier as I had very very good research and decent grades. If you have the requisite course work and did well in it, you're set. The research will set you apart if you make it a focal point, but if you want to, you need to show that you actually know stuff about your research. If your advisor really took point, maybe don't make a big deal of it apart from the research process. Hopefully this adds something to the convo from the research aspect.