Things to Do Before Graduate School

[whatdoido]

So recently, I've been trying to learn a few things in my free time to make me a more useful research assistant and a more useful future graduate student. So please rank the next few things in order of importance, because although I would like to take them all on, I really don't have the time.

 

()Learning Stata at a more advanced level

()Learning Matlab

()Learning SAS

()Reading a graduate microeconomics text if I don't have the chance to take Grad. Micro in undergrad (I've chosen Varian)

()Learning Python

()Learning Mathematica

 

The way I am, I always want to take on a billion things and end up only doing a little bit of each. This time I would like to only take on a few things and really learn them well.

 

Also, add anything else that you think should be on this list. Thanks.

 

 

EDIT: I'm a junior, if that changes anything.

[thewhiterabbit]

(1) advanced Stata (makes you very useful to self and others) (2) Python (lots of applications) (3) Matlab (4) grad Varian (good choice!)

 

I would not consider SAS unless you're interested in one of the few areas of research where it has an advantage over Stata, or unless you think you'll end up in private industry.

[treddi]

I agree with thewhiterabbit... stata, matlab, and python way before SAS or any similar program

[whatdoido]

Thank you! So do you know of any good STATA or Matlab books that I can look into? I'm not taking econometrics until this spring semester, so I don't know how much STATA I'm limited to at this point. If anyone could tell me how far I could learn STATA before econometrics that would be helpful as well.

 

EDIT: Another reason I would like to learn STATA better now is that I am RAing for a professor this coming semester, so I want to be of as much help as I possibly can be.

[thewhiterabbit]

Oh man, there are so many excellent Stata resources out there on the internet. One of the great things about Stata is its user base :-) Join the Statalist mailing list, for starters, as there is a wealth of knowledge passed around on there and people are very helpful. There is also a monthly column called Speaking Stata (example: http://www.stata-journal.com/sjpdf.html?articlenum=pr0003) in the Stata Journal which tends to be a nice, useful article aimed at regular users. Most (except the most recent I think) are free online.

 

You actually are not limited much at all by not having done econometrics yet. There are essentially two branches of Stata technical skill. The first is doing statistical analysis - for this you do need to have metrics. But of as much (if not more, depending on what you are precisely interested in) use are Stata programming techniques: using Stata as a programming language in order to do things like manipulate and change datasets, etc. in an automated fashion. An example lesson in this is the column I linked to above. These skills come in handy all the time, from doing data cleaning in an automated (not rote) way, to manipulating datasets/merging/etc. They are also the core skills you need in order to eventually write your own Stata programs. These skills can make you much more efficient in a wide, wide variety of contexts, and require no knowledge of metrics at all. Even if you never clean data you will be a much better Stata user for knowing these things.

 

Stata has netcourses which you can shell out $$ for as well - the advanced one goes quite a bit into programming tactics.

[kipfilet]

Books can help, but I believe the most useful way to learn coding is really learning by doing. Impose yourself some challenges and try to fulfill them in the most efficient manner (such as coding as few lines as possible). That's, at least how I learn MATLAB and STATA. I also began by reading books, but that gets boring really fast :)

[rthunder27]

But no, you should learn how to use Matlab (it's not that hard, and oh so useful)

[thewhiterabbit]

EDIT: Another reason I would like to learn STATA better now is that I am RAing for a professor this coming semester, so I want to be of as much help as I possibly can be.

 

So the advice I give the RAs that work for me is: if you think there is should be an automated way to do something in Stata, there is, and 99 times out of a hundred it's worth your while to figure it out. Even though it might take you as much time to figure out an advanced solution to some repetitious task as just doing it by hand, you'll walk away from the encounter knowing more for the next time. Plus I find it to be much more fun and interesting to work this way. You can answer many questions by googling, and Statalist, and its archives on the net, are great.

 

I would also emphasize that if you ever work on a large data cleaning task (this is most common in development work where surveys are still mostly by hand) you are going to introduce more problems and make your code harder to read and debug if you have, for example, 1000 copy-pasted "replace" commands instead of a few loops.

[thewhiterabbit]

I will never forget the time when I was a wee undergraduate RA and I pointed out to a fellow RA that you can use "save, replace" instead of going through every time and deleting each output file by hand before you run the do file :)

[thewhiterabbit]

Books can help, but I believe the most useful way to learn coding is really learning by doing. Impose yourself some challenges and try to fulfill them in the most efficient manner (such as coding as few lines as possible). That's, at least how I learn MATLAB and STATA. I also began by reading books, but that gets boring really fast :)

 

I tend to agree, except in the long run, there is sometimes a tradeoff between computational efficiency and # of lines.

 

It matters if your code takes an hour or several to run :)

[whatdoido]

Thank you very much. So would you suggest that I just go through google and try to find all the resources I can that would help me learn data management/manipulation?

[thewhiterabbit]

Well, it depends on your current level. What can you do right now?

[kipfilet]

Well, I was not properly saying that you should simply ignore the books. What I really meant is that I found much more helpful to use them as a reference, rather than going through them all at once :)

[Diplomer]

Things to Do Before Graduate School

For some reason, the first to come to my mind was something like "Make peace with life, your enemies and your family".

However, just in case you don't know it already, LaTeX may also be useful to know prior to entering gradschool.

 

Best, Diplomer

[thewhiterabbit]

@Diplomer lolz and +1 for LaTeX.

 

@OP A lot of beginner stata resources aren't specifically divided into 'programming' versus 'statistics.' Most people don't think of stata that way even if they should. Some of the advanced resources in Speaking Stata (and the Stata netcourse) do make a distinction, however.

[whatdoido]

Well, it depends on your current level. What can you do right now?

 

At this point, not really that much. I did some basic STATA training in my statistics course. I can run a basic regression. I can make some graphs. I can view data in a few different ways and I can mess around with the data a little by deleting data or creating new variables. There may be some other small things I'm forgetting, but overall not really much at all. What suggestions do you have?

[thewhiterabbit]

Well, if you were my RA I'd give you the Stata netcourse, but that's cause where I work provides the course to RAs if they want to do it. It's good but probably not so amazing that it's worth shelling out for.

 

Really, googling around till you find a tutorial that looks good to you is probably your best bet. There are a lot of good ones out there that use public datasets. You can also ask your future prof if he has suggestions, since he will know best the kind of work he'll have you do.

[whatdoido]

Well, if you were my RA I'd give you the Stata netcourse, but that's cause where I work provides the course to RAs if they want to do it. It's good but probably not so amazing that it's worth shelling out for.

 

Really, googling around till you find a tutorial that looks good to you is probably your best bet. There are a lot of good ones out there that use public datasets. You can also ask your future prof if he has suggestions, since he will know best the kind of work he'll have you do.

 

I asked my professor and she said that my best bet would be to just mess around in the STATA manual. I'll try to look around on google and find a good one.

[mathemagician]

lol... when you guys hit grad school you will wish you:

 

1) found a wife

2) had some fun before coming in

3) didn't have senioritis

[Charis]

don't forget stockpiling the gigabyte upon gigabytes of solution manuals, past HWs and prelims if they're available from your school, from where your prof graduated from, and from anywhere else that you want (but probably won't get to doing).

 

Also, my opinion was that going from some books may not have been more productive but it made trying to learn stata less arduous and stressful.

[thewhiterabbit]

lol... when you guys hit grad school you will wish you:

 

1) found a wife

2) had some fun before coming in

3) didn't have senioritis

 

Not everyone here is male or lesbian/bi and living in a state where gay marriage is legal....

 

:)

[Diplomer]

Not everyone here is male or lesbian/bi and living in a state where gay marriage is legal....

 

Is gay marriage still illegal in the US? :eek::eek::eek:

[thewhiterabbit]

It depends on the state.... some states have gay marriage, some have civil unions (same rights, different name), some have (voter-passed) amendments to their state constitution banning it.

 

It's hardly uniform in Europe that same sex marriage is legal... are you in Germany? Sounds like gay couples don't have full rights there either...

Recognition of same-sex unions in Germany - Wikipedia, the free encyclopedia

[OneArmedEcon]

A random assortment of observations...

 

The Stata commands that you will need to complete your first year assignments (at least in my program) is pretty basic and straight-forward (second year metrics courses are much more demanding). I think that having familiarity with Matlab (for Macro) is for more important. I don't think that you'll use much Python in the first year either. Obviously at some point you'll acquire many of the skills that you have listed, but you don't need to have them before you start the program.

 

If you're going to spend some time reviewing micro, then I'd actually suggest using the Rubinstein micro lecture notes rather than Varian. In my opinion, some of Varian is very helpful, but in other parts he "overexplains" the simple stuff at the expense of not thoroughly covering the more complex concepts. Rubinstein just does a rather quick overview of everything, so no doubt you will need to consult another source (for some topics Varian will be a better choice than MWG) but you won't waste time on less complicated things.

 

But most importantly... try to relax and enjoy the time before you start. Once math camp starts, you really won't get a chance to enjoy life until Christmas. And then it's another mad dash until prelims. Given how competitive admissions are, it's unlikely that you'll be admitted anywhere for which you are underprepared. That's not to say that you know everything that you will need--but that's why you take the first year courses.

[Harry Lime]

23 replies and not a single "real analysis"?

 

Me to the forum: "It's.. like I don't know you anymore."