The Most Difficult Thing Entering Grad School

[StatsG0d]

As admissions decisions roll out, I was wondering what is the most difficult adjustment people have had to make entering their first year of graduate studies, particularly in statistics. Are the mathematics courses so much harder / different than in undergrad? Do you find your classmates collegial or does it seem more like a competition. Are stress levels manageable?

 

In addition, did anyone do some math review the summer before the entering semester? Should one tirelessly study Rudin prior to entry?

 

Thanks a lot.

[ar_rf]

From what I've heard, probability is generally the most difficult first-year class. I think that working on the first several chapters of a measure-theoretic probability book (Billingsley, Resnick) or the measure theory portions of an analysis book would be a more efficient use of time than studying all of Rudin. I have also heard that brushing up on basic linear algebra is very beneficial. But this is all second-hand.

[dickmwong]

By Rudin I assume you mean the book "Real and Complex Analysis". It is a very nice text given that it discusses the essential topics in real analysis (measure theory), basic functional analysis and complex analysis in 400 pages in a very clear and concise manner.

 

It is surely a rewarding experience to go through the whole book if you manage to do so. That said, whether this is necessary or not is another story.

 

(Oh, and I am also applying to PhD programs this year. So this is not an advice from somebody who is in his PhD.)

[StatsG0d]

@ar_rf: Thanks for the replies. I agree that my linear algebra could probably use a little tune-up. I'm not sure if going from random variables to random vectors is that trivial. I'll take a look at those books. Any opinions on Casella and Berger?

 

@dickmwong: That was indeed the book I was talking about. I took Real Analysis, but we did not use that book and did not really touch on measure theory. 

[ar_rf]

@ar_rf: Thanks for the replies. I agree that my linear algebra could probably use a little tune-up. I'm not sure if going from random variables to random vectors is that trivial. I'll take a look at those books. Any opinions on Casella and Berger?

I think that it's a very good book,but I'm not sure you really need to brush up on it before graduate school. That's where you'll learn it.

 

To me it seems more important to brush up on fundamentals than to repeat material you will be learning your first year. Linear algebra is extremely fundamental, often needs a brush up (most people seem to take it fairly early in undergrad), and your classes will probably assume that you know it. I think measure theory is helpful because it's something that most undergrads haven't seen before. Knowledge of it is required for graduate probability, but I'm guessing that they will go over it more quickly than might be optimal in order to get to the probability stuff.

[wine in coffee cups]

Specifics depend on what courses you start with, but I think you're better off reviewing calculus and linear algebra than anything proof-driven. Having stuff like the proof of Heine-Borel fresh in your mind will not speed up work, but having a good command of calculus will. I had made a post summarizing the specific math I wish I had reviewed before starting my program

 

Get comfortable with LaTeX before you start. Even better, learn how to integrate it with R using knitr. Some of my classes required homework to be typeset, but nobody is going to teach you how to do this if you don't already know. The people who had not used LaTeX seemed to have a hard time learning as they went along, and Microsoft Equation Editor output looks like garbage so you'll want to not use that as a crutch.

[StatsG0d]

Thanks everyone so much. This is really useful.

[Karoku_valentine]

Thanks everyone. I will be attending a MS in Statistics next August/September I am trying to be ready for everything. I was studying Casella and Berger and it seems ok.

[Biostat_Assistant_Prof]

Current second year biostats student chiming in.

 

Many (most?) biostat programs won't require a strict measure theory course. Casella & Berger would be a good bet if you wanted to jump start your statistical theory, although it'll probably be what you use through the first couple theory courses. 

 

I think my best advice for anyone getting ready to enter biostats PhD in particular is to 1)Brush up on and/or strengthen your linear algebra skills, 2) ditto with calculus 3) learn a bit of R if you don't know it. 

[StatsG0d]

Thanks a lot Biostat. This is all really useful. Does anyone have any recommendations for LaTeX and/or R material? I already know SAS and Stata if that helps at all to suggest material for the latter.

[ar_rf]

Thanks a lot Biostat. This is all really useful. Does anyone have any recommendations for LaTeX and/or R material? I already know SAS and Stata if that helps at all to suggest material for the latter.

For LaTeX, this post gives some of the more popular introductory materials. I believe that I used The Not So Short Introduction to LaTeX when I first learned. But once you get the basics down, the best way to learn is simply to use it regularly. If you're going to be studying materials either way, take the time and typeset solutions to the problems that you do. If you come across something you don't know how to do (which you probably will a lot), spend some time on Google, the wikibook, and tex.stackexchange until you know how to do it.

 

It's been a long time since I've learned R, so unfortunately I don't really remember what I used. The official introduction seems decent and comprehensive, if a bit dry. Again, I suggest learning by doing once you have the basics down, if you have any data analysis side-projects or hobbies then you should try to use R with them.

[bayessays]

"The Art of R Programming" is absolutely fantastic and would be my recommendation to learn R.  It teaches you about R as a programming language, as opposed to teaching you how to do specific things in R.  Once you get that down, learning how to do specific statistical tests will be much easier and can be learned as you go in classes.

[StatsG0d]

This is fantastic. Thank you all so much. Such a great community in this forum.

[GTM-1]

baby Rudin would be sufficient.

[Karoku_valentine]

Thanks everyone. I will carefully start reviewing all your suggestions. I am a little afraid I might lose some of my ability to understand and do proofs, so I think I will continue studying Casella.

[TakeruK]

I'm not in Math/Stats but the most difficult adjustment I had to make was nothing related to the discipline at all! But maybe this does not apply if you are in a course-based graduate program.

 

For me, the hardest thing was adjusting the way I manage my time and expectations. As an undergraduate student, my main priority was my coursework -- I worked on problem sets until I felt I understood the material completely and studied for tests aiming for top grades. I'd say I probably spent 60 hours per week on coursework.

 

However, as a graduate student, all of this changes! My main priority is now research and classes are just a necessity. I had to learn how to stop working on problem sets and stop studying when I started reaching diminishing returns. I no longer try to get every question right on homework sets or understand every bit of every lecture. Often, I know I can get 90% instead of say, 80%, if I spent another couple of hours on it, but I can also be another couple of hours closer to a paper if I worked on research instead. That is, I think the hardest adjustment to make, for me, was the ability to turn in coursework that I knew was not my very best. 

 

Fortunately, we did get a lot of support with the adjustment. The professors talked to us about their expectations and made it clear that if we didn't get a few Bs here and there, then we were trying too hard on classes. We just have to prioritize what is important to us and our future career goals, which is normally research but the occasional essential class might be worth the extra effort. 

[bayessays]

TakeruK's advice, while certainly true for most disciplines, may not apply to students beginning a graduate degree in statistics.  Your main focus will very likely be your coursework for at least the first year (and likely two), which will give you some time to adjust.  Statistics is pretty unique as a discipline in this respect because few people start with the background required to be able to do research.  In a biostatistics department, you may have slightly higher research expectations early on if you are being supported by a research assistantship.

[ar_rf]

TakeruK's advice, while certainly true for most disciplines, may not apply to students beginning a graduate degree in statistics.  Your main focus will very likely be your coursework for at least the first year (and likely two), which will give you some time to adjust.  Statistics is pretty unique as a discipline in this respect because few people start with the background required to be able to do research.  In a biostatistics department, you may have slightly higher research expectations early on if you are being supported by a research assistantship.

I'm guessing that it is, however, extremely relevant for second-third year students in the midst of the transition. Though, as you said, you will probably have more time to adjust to a research schedule.

[wine in coffee cups]

TakeruK's advice, while certainly true for most disciplines, may not apply to students beginning a graduate degree in statistics.  Your main focus will very likely be your coursework for at least the first year (and likely two), which will give you some time to adjust.  Statistics is pretty unique as a discipline in this respect because few people start with the background required to be able to do research.  In a biostatistics department, you may have slightly higher research expectations early on if you are being supported by a research assistantship.

Agreed, I was just about to post the same thing! The emphasis in the first year or two of an American statistics PhD program is on coursework and learning the material well enough to pass qualifying exams, which functionally means spending a ton of time studying. You can work on research, but it's more on the back burner until your second or third year when you're done with the hard requirements.

[stats_applicant]

Agreed, I was just about to post the same thing! The emphasis in the first year or two of an American statistics PhD program is on coursework and learning the material well enough to pass qualifying exams, which functionally means spending a ton of time studying. You can work on research, but it's more on the back burner until your second or third year when you're done with the hard requirements.

This is true. However, for me the approach to coursework has still been fundamentally different from undergrad. I'm not so concerned with finishing every little problem on every homework set, and grades below A- are rarely handed out. If I want to spend time working on a coding or data analysis side project, or attending a seminar or reading group, or reading survey papers or supplementary material for a topic I don't understand particularly well, I'll sometimes choose to do it instead of finishing my homework. Yes, I need to pass my qual, but ultimately I think the main goal is to develop the tools to become a strong researcher (which is sometimes orthogonal to some of the qualifying exam material).

[StatsG0d]

This post should really be stickied because of all the great insights.

 

In terms of social interaction (whether among other graduate students inside/outside your program or otherwise), how bad is the shock? I find myself to be something of an extrovert, so I am hoping that the social shock of being a graduate student isn't that severe. Granted, I was very dedicated to my studies in undergrad and had a good social balance, but obviously it's much more serious in graduate school. Do you find your classmates collegial? I've heard for some programs/schools your fellow grad students view you as competition, and so are unwilling to help you out in times of need.

[TakeruK]

This is true. However, for me the approach to coursework has still been fundamentally different from undergrad. I'm not so concerned with finishing every little problem on every homework set, and grades below A- are rarely handed out. If I want to spend time working on a coding or data analysis side project, or attending a seminar or reading group, or reading survey papers or supplementary material for a topic I don't understand particularly well, I'll sometimes choose to do it instead of finishing my homework. Yes, I need to pass my qual, but ultimately I think the main goal is to develop the tools to become a strong researcher (which is sometimes orthogonal to some of the qualifying exam material).

 

I think this statement expresses the intent of my original post better than the way I wrote it. Also, sorry that my original post was not very practical for your programs! 

[wine in coffee cups]

In terms of social interaction (whether among other graduate students inside/outside your program or otherwise), how bad is the shock? I find myself to be something of an extrovert, so I am hoping that the social shock of being a graduate student isn't that severe.

I'm in a program that is much more heavily American than most, so my department is pretty social (we plan frequent happy hours and parties, good intramural sports participation, etc.) The other students here are great, though it took a little while to get to know people who I didn't have classes with besides my office mates.

[stats_applicant]

This post should really be stickied because of all the great insights.

 

In terms of social interaction (whether among other graduate students inside/outside your program or otherwise), how bad is the shock? I find myself to be something of an extrovert, so I am hoping that the social shock of being a graduate student isn't that severe. Granted, I was very dedicated to my studies in undergrad and had a good social balance, but obviously it's much more serious in graduate school. Do you find your classmates collegial? I've heard for some programs/schools your fellow grad students view you as competition, and so are unwilling to help you out in times of need.

 

There's no quota or anything for the quals, so it's in everybody's best interest to help each other pass.  We're allowed to work on homework together in most of my classes (it's up to the instructor) so we frequently discuss problems.  It's not competitive at all.

 

I think the social atmosphere varies by department.  My department is rather small but we still have activities now and then.  But it does take a while to meet people outside the department and outside your classes, especially first year when you have a heavy courseload.  Really its what you make of it; the workload is such that you could spend every waking moment working and not be 100% comfortable with the material, so if you decide to do that, that's what'll happen.  Not that that's a bad thing and a lot of people are satisfied with that lifestyle, but I make it a point to prioritize social events, playing sports, etc. and I don't think it detracts too much from my academic life.

[statisticsfall2014]

This is a great thread. I just made it through my first semester, but I will try and collect my thoughts and post on it.