PhD in Statistics

[GradSchoolNYC]

Hello all,

I would really appreciate it if someone would tell me if I have any chance to get into a PhD Statistics Program. I would like to get into a top 20.

I am not the typical student. First of all I will be completing my BA/MA in Economics this fall of 2010. My school gives you the option to take lots of statistics based econometrics class and I did all that.

In terms of classes (note, the class with a (G) after them mean that they were grad courses) I have taken Calc 1-3, Diff eq, Linear Algebra, Advanced Prob Theory (G), Sampling Theory (G), Multivariate Analysis (G), Econometrics 1 & 2 (G), and Time Series Analysis (G).

I am a BA/MA student. My overall GPA (meaning, the GPA of ALL my classes) is 3.55. My MA gpa is a 3.8 and my undergrad GPA is a 3.6.

I am writing a thesis paper which will have a heavy dose of statistics and econometrics.

I plan to do my absolute best of the GRE but haven't taken it... but for simplicity let's assume I score in the 95th percentile in all the sections.

I can get great recommendation letters from 3-5 professors, although only 1 of them will be from the Math department, the rest will be economists.

I participate in the student body, I have been a TA for the past 2 years and I play Tennis, if that matters at all. I have also been in Economics based competitions twice, one time making it to the semi finals.

I am a minority. I don't know if this helps at all...I know it sure does if one goes for the Econ PhD.

Since the earliest I can get into any program is by fall of 2011 I am planning to take an Analysis course (maybe both 1 and 2 if time permits), a computer programming course (I know VBA and a little bit of python but I have never taken any formal classes) and about 2-3 more Stat course...most definitely Math Stat but not really sure what else.

And finally, some schools, like Cornell, offer a MS/PhD track, do I have a better chance getting into those types of programs?

That is about all the information on myself that I can think of right now. Can someone please lend a hand?

Thanks so much in advance!!

[kman3201]

You have what it takes to get into a stats PhD program no problem, its the top-20 part which might be a little issue. You seem to have alot of math background ,which is what stats programs look for, as well as a computing background. The reason I say the top-20 part is hard is becuase you have to think about the other students who will be applying for the same program. Alot of them have done pure math or statistics in undergrad, and some have masters degree's in statistics as well(most of the internationals do).

But I would say definately go for it. I would be shocked if you didnt get into atleast one top-20 program with funding. As for the MS/PhD program, I am almost sure you would get into most of those which you apply you. But given your background, you are ready to go straight into the phd.

[babaloo]

I agree that you are definitely qualified for PhD programs. Having those graduate courses are very helpful. I'd try to get perhaps another letter from a math or stat professor that taught one of your advanced classes, but if that's not possible it would be nice if the economists could speak a little about your mathematical ability. Also, if you have experience with programming or statistical packages like R or SAS (even a little bit), this is something you want programs to know about

As for courses, take as much analysis as you can. So if you can take two semesters, do it, and let the schools that you apply to know that you're taking them if it doesn't show up on your transcripts. Programs will care much more about analysis than any applied stat class. If you are going take stats, math stats (i.e. probability theory, statistical inference) would be good classes to have. For a programming course, you can't go wrong with C, C++, or Fortran.

One thing I'm curious about is how you managed to take advanced probability theory without analysis. Typically you would need to have some background in analysis or measure theory, but in any case, if you did well in that course and your other graduate stats courses, it will look good.

For the GRE, most top programs will expect a perfect or near perfect quantitative score. Below a 770 might hurt, but I'm not too sure about this. Just go for >= 770. Verbal and writing shouldn't matter much, just make sure they're not too low.

Departments will probably not care much about the student body or tennis involvement , but the TA is relevant, and the econ competitions may be as well.

Being a minority could be to your advantage, especially if you're qualified (and you are). One reason is because some schools have university fellowships for strong minority applicants, so funding wouldn't be an issue for the department. Another reason is that there aren't a lot of non-asian minorities in statistics. Either way, you're well qualified so this shouldn't be something to worry about.

As for the MS/PhD thing, most PhD stat programs operate in this manner anyway, they just don't advertise it. If you're applying without a previous master's in statistics, you'll have to complete the master's requirements before moving onto PhD candidacy anyway, and many departments will award you the master's degree en route to the PhD. Also, if you don't pass your quals after a couple of tries, the department might give you a conciliatory master's degree on your way out of the program.

Hope this was helpful, and good luck!

[floffel]

Assuming you go to a school with a reasonably good reputation, you could probably get into top 20 programs with what you have already. Taking some more math and a programming course or two will give you a decent shot at some top 10 programs IMO (again, this is assuming you're in a decent program already).

[GradSchoolNYC]

Thank you guys all so much for your replies. It really helps motivate me because now I really believe that this is a viable endeavor.

I will take as many Analysis classes as I can and will try to get some programming courses.

Thanks again.

[GradSchoolNYC]

I agree that you are definitely qualified for PhD programs. Having those graduate courses are very helpful. I'd try to get perhaps another letter from a math or stat professor that taught one of your advanced classes, but if that's not possible it would be nice if the economists could speak a little about your mathematical ability. Also, if you have experience with programming or statistical packages like R or SAS (even a little bit), this is something you want programs to know about

As for courses, take as much analysis as you can. So if you can take two semesters, do it, and let the schools that you apply to know that you're taking them if it doesn't show up on your transcripts. Programs will care much more about analysis than any applied stat class. If you are going take stats, math stats (i.e. probability theory, statistical inference) would be good classes to have. For a programming course, you can't go wrong with C, C++, or Fortran.

One thing I'm curious about is how you managed to take advanced probability theory without analysis. Typically you would need to have some background in analysis or measure theory, but in any case, if you did well in that course and your other graduate stats courses, it will look good.

For the GRE, most top programs will expect a perfect or near perfect quantitative score. Below a 770 might hurt, but I'm not too sure about this. Just go for >= 770. Verbal and writing shouldn't matter much, just make sure they're not too low.

Departments will probably not care much about the student body or tennis involvement , but the TA is relevant, and the econ competitions may be as well.

Being a minority could be to your advantage, especially if you're qualified (and you are). One reason is because some schools have university fellowships for strong minority applicants, so funding wouldn't be an issue for the department. Another reason is that there aren't a lot of non-asian minorities in statistics. Either way, you're well qualified so this shouldn't be something to worry about.

As for the MS/PhD thing, most PhD stat programs operate in this manner anyway, they just don't advertise it. If you're applying without a previous master's in statistics, you'll have to complete the master's requirements before moving onto PhD candidacy anyway, and many departments will award you the master's degree en route to the PhD. Also, if you don't pass your quals after a couple of tries, the department might give you a conciliatory master's degree on your way out of the program.

Hope this was helpful, and good luck!

Babaloo, thanks so much for your in depth reply. I have a couple of follow up questions.

1- Ok so the earliest I can get into a phd program is fall of 2011. This means that I can stay at my school until spring of 2011. I am planning on graduating in fall of this year. Now my question is would it make sense to postpone my graduation until Spring of 2011 so that I can take some computer programming courses to be more prepared for phd?

2- Sort of a follow up question to #1- My school offers some computer programming courses, but there are some other schools in my area that have better csci dpts. I don't know if this is even possible., but would it be wise to try and take one of these course at these schools as a non matriculated student? If so, how will I let the schools know that I am doing this?

3- In response to your adv prob th question, I had to get permission from the advisor to take the class. In hindsight it was probably not the best idea because I ended up with a B+...do you think that negatively affects my chances? If so is there anything I should do to remedy it?

Thanks again so much, you rock!

[BongRips69]

I would suggest deferring graduation until Spring, unless you have a job or apply to programs which offer spring admission.

The only piece missing from the description of your background is the subject GRE score. I'm not that familiar with stat programs, but it's my assumption that many (especially top 20) require the subject mathematics GRE. Taking more analysis is a good suggestion, as it will prepare you for this test. It consists mainly of calculus, analysis and algebra (linear, abstract).

I'd check out over at www.mathematicsgre.com. There's a thread where people post stats and acceptances/rejections. You should get a good feel by browsing over there.

[BongRips69]

Oh yeah, and don't expect a 95% percentile on the subject exam!

[GradSchoolNYC]

Oh yeah, and don't expect a 95% percentile on the subject exam!

YIKES! You have to take the subject test in math??

That means I have to study for two gre exams?! Oh man lol.

Anyone know the average score for this exam?

[BongRips69]

If you want "top 20", then most likely you have to take the exam. Look at the webpages of departments in which you intend to apply. They will spell out admissions requirements.

As for studying for the exam, I focused most of my time on the subject exam. I scheduled the general exam for one month afterwards. This was plenty of time to get an 800 on the quant section, but I should've memorized more vocab words. About both of these tests: great scores alone won't get you into a school, but poor schools will keep you out.

Most schools don't post minimum scores (or even averages), but a few top ones in pure mathematics suggest not even applying if you're under 80%. They get a lot of applications and need to cut somehow. Again, check out the following link to get a feel for scores required for a particular school:

http://www.mathemati...c.php?f=1&t=357

Here's a list of previously published exams courtesy of UCSB math club.

http://www.math.ucsb.edu/mathclub/GRE/

Here's the current practice book from ETS:

http://www.ets.org/g...tics/index.html

http://www.ets.org/M...actice_book.pdf

Edited March 4, 2010 by BongRips69

[GradSchoolNYC]

If you want "top 20", then most likely you have to take the exam. Look at the webpages of departments in which you intend to apply. They will spell out admissions requirements.

As for studying for the exam, I focused most of my time on the subject exam. I scheduled the general exam for one month afterwards. This was plenty of time to get an 800 on the quant section, but I should've memorized more vocab words. About both of these tests: great scores alone won't get you into a school, but poor schools will keep you out.

Most schools don't post minimum scores (or even averages), but a few top ones in pure mathematics suggest not even applying if you're under 80%. They get a lot of applications and need to cut somehow. Again, check out the following link to get a feel for scores required for a particular school:

http://www.mathemati...c.php?f=1&t=357

Here's a list of previously published exams courtesy of UCSB math club.

http://www.math.ucsb.edu/mathclub/GRE/

Here's the current practice book from ETS:

http://www.ets.org/g...tics/index.html

http://www.ets.org/M...actice_book.pdf

Hey

Thanks so much. All that stuff is really helpful....Just by glancing over the exam I must say I need to get to studying!!

Thanks again and good luck on whatever it is that you are doing.

[BongRips69]

I don't mean to harp on the subject GRE... you're background looks good. Actually, it's pretty similar to mine (substitute work experience for your MA). So I'm mainly just trying to give you advice on things I didn't take seriously enough! I've already got some acceptances for math/applied math PhD programs, so you definitely can.

[babaloo]

Babaloo, thanks so much for your in depth reply. I have a couple of follow up questions.

1- Ok so the earliest I can get into a phd program is fall of 2011. This means that I can stay at my school until spring of 2011. I am planning on graduating in fall of this year. Now my question is would it make sense to postpone my graduation until Spring of 2011 so that I can take some computer programming courses to be more prepared for phd?

2- Sort of a follow up question to #1- My school offers some computer programming courses, but there are some other schools in my area that have better csci dpts. I don't know if this is even possible., but would it be wise to try and take one of these course at these schools as a non matriculated student? If so, how will I let the schools know that I am doing this?

3- In response to your adv prob th question, I had to get permission from the advisor to take the class. In hindsight it was probably not the best idea because I ended up with a B+...do you think that negatively affects my chances? If so is there anything I should do to remedy it?

Thanks again so much, you rock!

1. That kind of depends on you. If you can land some kind of relevant job or paid research position for a few months, you might be better off doing that. If not, more math and computing, and maybe a little bit of research, won't hurt.

2. I would stick to staying within your school for courses to avoid the hassle of registering, and getting transcripts and all of that. I don't know how much better these other schools CS departments are than yours, but I doubt it would matter much as long as you did well in the course at your school

3. I don't think a B+ in a graduate probability course will hurt, and if anything, I believe it would be helpful as it'll show them that you can handle graduate work.

Also, if your school has a statistics department with a graduate program (and from your list of courses it seems that it does), I would talk to some of the faculty, particularly the grad studies director and the chair if you can, and ask them all of the questions you asked in this thread, because they'd definitely have a better idea of what stat programs look for and care about.

About the GRE subject, a handful of top programs require it (Stanford and U Chicago are the ones I know of), and some programs strongly recommend it (Harvard, and perhaps others). By no means do all the top schools require or recommend it, so as someone else suggested, check the web pages of the programs in which you're interested, and plan accordingly. The good news is that you have time to prepare for it

[Pierre]

Since I'm an international (spent one semester in the US, but that hardly counts) I'm not entirely sure about how much emphasis is put on different aspects of your application etc...However, it seems as if you have a solid background for entering a PhD program. As some have suggested I would also encourage you to pick up some more analysis, especially if you want to go in a very theoretical direction. However if you are leaning more towards an applied direction for your studies, some more advanced statistics and/or computer programming surely wouldn't hurt.

Also, you could always do yourself a favor and apply to a department that conduct research influenced by finance (I can't remember if you wrote that this was something you didn't want to do, my apologies if so) and where your background would suit well with the interests of the faculty.

As for the GRE, only a few of the statistics program do require it. E.g. out of the top three (according to US News 2008 ranking) only one require it (Stanford), one recommends it if you performed well (Berkeley) and one is satisfied with a solid math background (UNC). Here you can of course show the programs that you are indeed good at math, even though the program doesn't necessarily require the test, should you feel that your grades/courses aren't doing your abilities justice. I'd also follow the other posters recommendation and really study for the Subject test if you plan to take it.

I advise you to make a list (really big at first) of places that you can see yourself applying, and start looking more carefully at every program. What professors would you want to work with at each place (is it one or ten), what type of research does the faculty seem to focus on/lean towards and how does this correspond to what you want to do, etc.. In my case this completely changed where I applied as compared to what I thought a couple of months earlier. I don't know how set you are on certain departments or so, but this "tactic" was really useful for me.

Best of luck!

[Piwi]

I just want to reiterate that most Statistics programs do NOT require the GRE Subject test. The ones I know of that do require it are Stanford and Yale (and a few others "recommend" it, eg. Columbia).

Wisconsin, Duke, and Berkeley are all top 10 schools, and do NOT require the Subject test (Berkeley may recommend it, I don't remember... but I know they don't require it).

[BongRips69]

Guess I got swayed by the two programs to which I applied with the word 'statistics' in their name. I guess they're not really true statistics programs.

[OuterMeasure]

Hello all,

I would really appreciate it if someone would tell me if I have any chance to get into a PhD Statistics Program. I would like to get into a top 20.

I am not the typical student. First of all I will be completing my BA/MA in Economics this fall of 2010. My school gives you the option to take lots of statistics based econometrics class and I did all that.

One thing to think about now in a precise way is why a PhD. and why Statistics versus Economics or Mathematical Finance, or something related. The rational is important to justify to yourself the devotion of several more years of your life to education in statistics and then research in an extremely specific topic in that field. Some distillation of this will also be very important for your statement of purpose. My rational is something like: I was a math major with programming experience who moved into Bioinformatics and saw a lot of statistics in all the methods being used. Stat is a nice way for me to study something very mathematically oriented but which is directly used with all manner of research data. I like collaborative research, having worked in that environment several years after college. A Ph.D. in Statistics (as opposed to applied math, or even bioinformatics) keeps a lot of doors to the private sector open, which is important for me since I'm somewhat older than most students and may not want to wait as long to have a stable, well-paying job as a professor (possibly without much choice of where to live). Not all of this went directly in my SOP, but that along with a description of my research projects are definitely the basis. It sounds like your masters thesis has given you exposure to mathematical research, but in addition to coursework you could consider a summer research program where you get to directly apply statistical methods to some real data (and possibly publish the results).

In terms of classes (note, the class with a (G) after them mean that they were grad courses) I have taken Calc 1-3, Diff eq, Linear Algebra, Advanced Prob Theory (G), Sampling Theory (G), Multivariate Analysis (G), Econometrics 1 & 2 (G), and Time Series Analysis (G).

I am a BA/MA student. My overall GPA (meaning, the GPA of ALL my classes) is 3.55. My MA gpa is a 3.8 and my undergrad GPA is a 3.6.

I am writing a thesis paper which will have a heavy dose of statistics and econometrics.

I plan to do my absolute best of the GRE but haven't taken it... but for simplicity let's assume I score in the 95th percentile in all the sections.

I can get great recommendation letters from 3-5 professors, although only 1 of them will be from the Math department, the rest will be economists.

I participate in the student body, I have been a TA for the past 2 years and I play Tennis, if that matters at all. I have also been in Economics based competitions twice, one time making it to the semi finals.

I am a minority. I don't know if this helps at all...I know it sure does if one goes for the Econ PhD.

Since the earliest I can get into any program is by fall of 2011 I am planning to take an Analysis course (maybe both 1 and 2 if time permits), a computer programming course (I know VBA and a little bit of python but I have never taken any formal classes) and about 2-3 more Stat course...most definitely Math Stat but not really sure what else.

This is certainly a very strong academic record, and I think you'd probably get into a "top 10" program and certainly a "top 20," program to the extent that those lists are well-defined. Programming courses are good, but Real Analysis will be more useful in terms of the coursework you'll have to take in graduate school. I learned R on my own having a good deal of experience with Perl and Python; much of learning to program at an intermediate level involves having something to work on and reading a lot of reference material. You start to need fundamental coursework when you want to do things of an algorithmic nature that have to be optimized. Real Analysis also works well in a course because it requires a lot of thinking about a comparatively small number of difficult concepts. Rudin's Principles of Mathematical Analysis will give you excellent preparation, but real analysis, however it's taught at your university, would be a great course to take in general. Again, research experience, be it through your master's thesis, a Summer program, or even a job in academia (my route) is a great way to show admissions committees that you like research. Statistics, as you may know, is very much dependent on the existence of large data sets, of which there are many.

Another note, the vast majority of Statistics departments don't require GRE subject test. Stanford does, Washington, Chicago, and Johns Hopkins Stat recommend it, and a couple departments send mixed messages, but those that even talk about it are decidedly amongst the most competitive. If you're going to take it, you'll want to wait until after that real analysis course. It will also require you to study a good deal of abstract algebra. Abstract algebra and upper-division linear algebra are two other good math courses to choose from, if you have the time. Either way, most will tell you only take the subject test if you'll do very well (75 percentile or better). This might be very difficult with only one or two upper-division math courses under your belt, but you do have a good deal of time.

And finally, some schools, like Cornell, offer a MS/PhD track, do I have a better chance getting into those types of programs?

That is about all the information on myself that I can think of right now. Can someone please lend a hand?

Thanks so much in advance!!

Such programs give you a way to get a Master's degree should you choose to not finish your PhD., and they may give students accepted into the MS program (which usually has lower requirements, but less funding) a streamlined way to transition to PhD, should a well-performing student wish to do so. Some MS/PhD programs (like Cornell) don't directly admit Masters students and have a separate, professional Masters track. Many programs do award a terminal Masters even if they don't advertise themselves as "MS/PhD" on the departmental home page. I don't think this aspect drives systematic differences in admission rates; it's just what options you have available to you degree-wise. It's good to read the "fine print" on the departmental website, since they give a good deal of information about all aspects of the program.