Stat PhD Advice for Economist

[econ_to_stats]

Seeking advice on what schools are appropriate.

The Rundown

Undergraduate Institution: Medium-size Midwest Public University. Not well known

Majors: B.A. Mathematics / Economics

GPA: 3.54

Type of student: White Domestic Male

Graduate Institution: Large Southern Public University

Program: Econ PhD program dropout. MA in Economics.

GPA: 3.25

GRE Scores: (Unofficial scores)

Q: 167

V: 167

W: unknown

Relevant Coursework:

Undergrad:

Mathematical Statistics (3 quarters): A-, A, A-

Statistical Computing: A

Linear Algebra: B

Matrix Theory: A-

Real Analysis / Adv. Calc (2 quarters): A, A

Elem. Point. Set Topology: A-

Graduate:

Econometrics (2 semester): A-, B

Limited Dependent Variables: A (In Econ Dept. Transcript shows Incomplete changed to A)

Survival Analysis: A (In Stats Dept. Transcript shows Incomplete changed to A)

Panel Data: A- (In Econ Dept.)

Time Series: D- (In Stats Dept. Undergraduate level course. I messed up here. Didn't do the homework, which turned out to be worth 60%)

LORs: Not sure yet. I'm considering one econometrician from grad school, one econometrician from undergrad, and a statistician from undergrad. Not likely to be well known by admission committee.

Papers:

I have one empirical paper published in a mid-to-low level economics journal that uses a panel regression with one co-author.

The Point:

Essentially, I dropped the ball in grad school. I was disappointed with my poor econ placements and accepted an offer at a school that was a bad fit because it was the best funding. Didn't have the maturity to deal with a bad situation and walked away leaving a mess. Cleaned up mess later and received my degree.

I'd like to go back and get a PhD in Statistics. I like Bayesian statistics and recently have been getting into Statistical Learning. I'm very pessimistic about my chances though. Partly based on how poorly I did applying to Econ PhD programs and partly because I'm switching fields. I know Standford and Carnegie Mellon have great programs in Statistical Learning, and Duke is great for Bayesian Stats, but I have no idea what would be a good fit beyond that and know that all of those schools are out of my reach.

So my question is, "What can I do?" Is it possible to go to a less prestigious university, do well, and transfer to a better one? What schools would be willing to accept a student with a transcript as blemished as mine? Help?!

[econ_to_stats]

Official GRE scores in.

Verbal: 167, 97%

Quant: 167, 96%

Analytical Writing: 4.5, 73%

Also, why no love? Over 100 views and no comments?

[m4thguYY]

I don't know how far my opinion would really go. Let's just say I had a slightly weaker undergrad, made up most of my bad grades. Then got two master's and did great in them, great LOR, 3.95 concentration GPA, stochastic/poisson process thesis applied to economics. I have a perfect QGRE score/160ish Verbal.

If you are "considering" your LOR, i don't see how they could be very good. You did relatively well in school grade-wise, but you also got a D-, which is like unheard of. I didn't even know schools actually gave that, thought it was a myth.

Also, I don't see many grad courses, especially in stats-related department. You have 6 classes, 5 if you don't count the D one. A lot of them show incomplete, and you will have to address why you gave up during a PhD. I've seen it first hand - committees don't like giving second chances to people. You do not have A LOT of the major math concept classes: stochastic processes, applied probability, queueing theory, operations research.. You may think you are well prepared for stats but people like me have 3 degrees in math/stats, and I am not even that strong of an applicant.

Given that info - I know I am applying to PhD programs in the 15-35 range. Columbia Stat and Rutgers Stat, along with Stony Brook Applied Math (Operations Research or Financial Math track). I believe that my application has a lot that yours doesn't - strong LOR from professors in the stats world, way more math/stat breadth. And although you may say that you have economics background also, many economics (most) will really go to waste .. economic theory classes (macro, micro, etc) will be useless compared to core stats classes (stochastic, probability).

Also, I have experience teaching at a SUNY school as a mathematics professor right now while I am applying. One of my master's also has a dual concentration in mathematical teaching methods.

I can only really help you by comparing you to me. I am in the 15-35 range, and I believe you should stick to that range too. I don't think you (or me for that matter) will make a top 20 like Columbia or Stern Statistics.

I honestly don't see too much for you .. but that's just me. I don't think quitting a PhD program will look good at all, and I don't think you have too much experience with serious mathematics. Again, I don't know how much my advice should count..

[wine in coffee cups]

Also, I don't see many grad courses, especially in stats-related department. You have 6 classes, 5 if you don't count the D one. A lot of them show incomplete, and you will have to address why you gave up during a PhD. I've seen it first hand - committees don't like giving second chances to people. You do not have A LOT of the major math concept classes: stochastic processes, applied probability, queueing theory, operations research.. You may think you are well prepared for stats but people like me have 3 degrees in math/stats, and I am not even that strong of an applicant.

I don't think the OP's coursework being insufficient is a well-founded point of criticism. I had a (pure) math major background with just a handful of undergrad statistics classes and never took any of the courses you listed. Mine were all at a liberal arts college, nothing at a graduate level. I had no problems getting taken seriously by top stats departments this year. I'm sure having some relevant coursework is important but I think only up to a point.

Certainly agree that the D- thing and the dropout thing are problematic, hard to guess how much of a risk programs would be willing to take on someone with that history. No specific suggestions for places to apply except to cast a wide net and to talk to your recommenders about things they might be able to mention that would instill more confidence in your ability to succeed in a graduate program. Wishing you the best of luck, econ_to_stats.

[econ_to_stats]

Thanks for the responses. I know it's going to be difficult to overcome my grad school grades. FWIW, my LORs seem to be strangely optimistic about my chances. At least, they seem to believe I can adequately explain that I went into an economics PhD program largely due to inertia, expecting to be intellectually stimulated, but that ultimately my heart wasn't in it. I don't know if it will make a difference, but my LORs believe, as I do, that my poor performance was a matter of attitude not aptitude. I'm pessimistic about my ability to convince an admissions committee of the same.

james8787, you certainly seem to have a much greater breadth of statistical experience. I'm hoping that since I don't have research interests in operations research or queuing, my lack of exposure to those concepts won't handicap me too much. (Although, who knows. Maybe if I were to take those courses, I would get really into it. I was surprised at how much I enjoyed Survival Analysis.)

So, I guess now I have four main questions.

1) What are decent programs I might have a shot at?

2) Is there any chance of excelling in a lower ranked program and transferring to a higher one?

3) What if I want to go into the private sector, not academia?

4) Is it worth it to take the GRE Math Subject test? (It seems only programs out of my league require it.)

I had assumed that, for instance, Columbia has a top 5 program, but based on james8787's comment, they are only in the 15-35 range. I would love to go there. Andrew Gelman is the man. Is there any aggregated information about programs beyond USNWR and PhDs.org that I can use to help me with my decision? I'm not opposed to doing grunt work but if something already exists I would really appreciate being able to use it.

Edited September 4, 2012 by econ_to_stats

[m4thguYY]

Again, I don't know how far my advice goes, I am myself (to no avail) asking for PhD advice. I would take what I say with a grain of salt, but I will attempt advice given what I have read/seen.

US News is the generic place to get grad rankings. Yet, those are for overall departments. For example, just looking at "math" will set NYU as #10, whereas if you choose the specialization "applied math", they are #1. Unfortunately, statistics is one of the few categories that does not have the subcategory option. US News puts Columbia at #22 for stats and #32 for biostats (they put stats and biostats departments together when they rank). Yet, NRC ranks them anywhere from #9-#22 for scholarly rank and #9-#30 for faculty rank. Google "NRC statistics rankings" and read their S-Rank and R-Rank for description.

So, from these numbers, I don't see how Columbia could be top 5, but again, I could be wrong.

I may have a greater breadth of stats, but when you go into a PhD program, the coursework at the beginning will introduce you to a lot of things and may even change what you are interested in researching. In math, I have a specific research interest, but my #1 program is Astrophysics if I make my dream program, so it is not crazy to think that students have a wide variety of things that spark their interest and passion.

To answer your questions:

1. Honestly, again, I don't see how your LORs believe that quitting a PhD program won't hurt you. I think that will be a huge setback, as will your degrees in econ, not really math(I know you have a dual major, just sayign). Your grades are good though. I would stick with wine_in_coffe_cup's advice and cast a really wide net. If Columbia is a program you'd like, then that should be in there, but maybe a big reach school. If you don't care about location, then just apply to a lot of schools that you think you would fit well with. If I didn't care about staying in NYC, I would be applying to like 50 schools. But I do, so I am only applying to like 4 or 5.

2. Programs REALLY seem to hate PhD students that want to transfer. I would definitely not talk about wanting to transfer. Ever. The school you go to and their faculty put a lot of money and teaching responsibility into you coming out with a PhD. They don't want students leaving (affects their stats), and they don't want to waste years of funding for nothing. Also, the "better school" you would want to apply to would probably prefer new students that they can mold and start fresh with. I don't think this plan is good at all.

3. I don't know much about this, I am all academia. Yet, it seems the private sector always needs statisticians. Yet, it also seems like the school you went to is much more important there. Look, I am not trying to teach at Harvard or Yale. I am trying to graduate and join a good math/physics department, do some research, and concentrate on teaching students. Therefore, going to a rank 20 or 30 school is fine for me. Yet, in the private sector, they pay big bucks for top names. They don't really pay average bucks for average schools to my knowledge. They need their people to be perfect, and with that comes top schools. Just my opinion. Look at alumni out of rank 50 schools and rank 5 schools and you will see the difference.

4. ETS has a new policy that you can pick and choose the tests you submit to each school. Therefore, taking the subject test has no possible bad side effects. Yet, reporting it is a different story. I would only take & report it if you do REALLY well. Doing average or slightly above average doesn't do much. I will leave the school name and program out, but it is a very top program in a physical science. Their advice on the Math Subject GRE was this:

"We don't require a GRE subject test, but suggest it is useful if you think you'd get a really high score on it, which would enhance your application."

I tend to agree with that. If you do really well, its worth it. If not, I personally would not. People that I have encountered seem to put the regular GRE as more of a qualification to be considered than a judging point, AKA over X and you go on in the process, under X and you don't. The GRE Subject Test is probably similar. It doesn't test you on your research interest, it tests you on very undergraduate knowledge. I would say LOR, combination of degrees & GPA, research/papers, and overall experience weigh much higher.

As a side note: Keep in mind that although Columbia might be #9-22, they are also the first ranked NYC school. People really want to be in NYC. Therefore, I am sure they get some serious competitors (such as top 5 types). Plus the name is amazing in general. Therefore, I tend to treat it around a top 10 program. I am sure they are better at some specialties and worse at others.

Also, NYU Stern has a PhD in stats that is not even ranked. Yet, they take 1-2 of 100+ applicants and is apparently impossible to get into. I don't know why they are not ranked.

Also you said:

I'm pessimistic about my ability to convince an admissions committee of the same.

That is the point. It is not your passion or desire to do well, or that anyone is saying you can't do it. It is that admissions committees see SO many applications, and to try to justify quitting an econ PhD and transferring to a stats PhD with a light background in stats is going to be VERY difficult. Again, thats just my opinion. I am going to attempt to explain a C or two, and I think even something like that will disqualify me from the beginning for some programs lol

[econ_to_stats]

Based on this post

I'm starting to wonder if I would be better off applying to Masters programs first, then moving into PhD programs once I prove myself.

Any thoughts?

[cyberwulf]

I'm starting to wonder if I would be better off applying to Masters programs first, then moving into PhD programs once I prove myself.

Any thoughts?

I think that's a reasonable idea. Make sure you go to a top-notch stat place for your Masters, and it wouldn't hurt to find a place where a non-negligible fraction of Masters students go on to do a PhD in the same department. The easiest route into a PhD program for you may be to impress enough people locally that they are happy for you to "upgrade" into the PhD track.

[econ_to_stats]

Here's what I'm thinking. Any and all advice appreciated.

Masters Programs

CMU

Columbia

Johns Hopkins

University of Chicago

UCLA

Univ of Washington

Stanford

UNC- CH

PhD Programs

Bowling Green

Ohio State

Case Western

UVA

UC-Davis

Colorado State

Do these schools seem reasonable given my profile?