Software to learn this summer? STATA, SPSS, SAS, R?

[MollyB]

Hi everyone. Post-MPP I'd like to go into a research associate position at a place like the American Institutes for Research or Mathematica. Has anyone here worked, perhaps as a research assistant, at either of these places (or similar)? If so, can you speak at all about the software commonly used?

 

 

I've decided to learn at least one stats software this summer, but I don't know which. Assuming I can get student pricing for a several-month license, I think STATA, SPSS and SAS can be relatively inexpensive (or at least not out of my reach). And of course R would be completely free. Anyone have an opinion on which is most commonly used in the real world? Should I bother with Python? I know nothing about databases -- would that be useful? 

 

From looking at a few syllabi, it seems like most schools teach at least a little bit of STATA. But if I want an RA position in my first year, even if not the first semester, I need to prove to profs that I have some useful skills. Thoughts or opinions? Anything would be appreciated! 

[Tupacalypse]

I know that in my quantitative poli sci class our professor had us use Stata and said that and R are most commonly used in real-world research while in his experience SPSS is used more for academic research.  But that is just one opinion.  I personally like stata because it is simple to understand, although for large datasets you have to get the larger version and the price is ridiculous.  See if you can manage to get a student discount still.

[jm08]

For what it's worth I'm a big fan of the newer versions of SPSS. Having STATA in your back-pocket is good, too, because it's used a bit more broadly than SPSS. SAS is a huge pain to learn because it "inherited" many odd quirks from its initial build that they never got around to changing; it also used primarily in the private sector (at least that was my personal experience). 

 

I've also seen decent "how-to" guide floating around the internet if you have the time to do some due diligence to find a good tutorial -- it's definitely possible to teach yourself. No idea how Python knowledge would come across to a potential professor (or employer). You'll also have a leg-up on fellow students the more comfortable you can get with these data packages if you end up taking an econometrics course in your program or intend to do something very quant-heavy post-graduation. 

 

Hope that helps!

Edited March 21, 2013 by jm08

[MPPgal]

Everyone is switching to R right now, if you can lean one, do that!

[Abc_adams]

R is the future.

SPSS is the past.

SAS is the present.

You decide based on who you plan to work for :

If you work alone - R. If you work with professors - SPSS. If you work with the industry - SAS.

[MollyB]

Thanks everyone for the anecdotes / impressions! It's all very helpful. 

 

@ abc_adams Thank you for the breakdown! It's good to know how I should prioritize. I've dabbled in R (but I have undergrad background in STATA and comp sci, so I don't suggest it to total beginners), but I know zero about SAS. 

 

In re Python: Admittedly, it would impress (or much less mean anything) to very, very few employers / professors. For anyone curious, it's used a lot by some social scientists to write scripts for data scraping and some data cleaning. I've personally used it occasionally to write very basic scripts to clean data (could also be done in R, etc). I imagine you'd have to be working in a very research heavy place for this to ever come up (obviously I'm not there yet).

 

Btw, UCLA has an awesome guide to learning STATA that many of my friends have sworn by http://www.ats.ucla.edu/stat/stata/

Edited March 21, 2013 by MollyB

[biscuits]

For those who are interested, Codeacademy has a free series of courses for Python: http://www.codecademy.com/tracks/python

Their courses are especially good for folks (like me) who learn via practice. 

[Dinosaur7]

Really interesting feedback on this thread, as it is different than what I've seen!

 

I am fairly familiar with AIR/MPR and that sort of industry. What I've seen is that most kids coming out grad school know STATA, but that isn't really preferred by employers--but they are learning STATA in grad school, so presumably that's what their professors use?

 

People really like SAS for its abilities to handle massive datasets. It definitely has its quirks and is not intuitive, but I think it is the most powerful of all the options. Also if you go into federal contracting, you will find that there are contracts that you need to use SAS for. So if you already have that skill, that's a bonus.

 

I'm really intrigued to hear people say R because I have not seen that much in the workplace yet (but used it back in undergrad). It would make my life easier if that was where things were headed!

 

Depending on your field of interest, ArcGIS is a good skill to pick up in addition to statistical programming. That type of analysis is becoming more popular and not many people have it.

[MPPgal]

R is slowly gaining support among both SAS, STATA and SPSS users! The fact that it is free and has all the functions of math lab and stata combined (plus mathematica, etc) makes it most people prefered program, but as it involves a bit more programing some are scared of it!

[JFactor]

So it seems there is no consensus? I have a feeling based on my experiences that STATA and SPSS are more popular with professors/in academia, but it seems R is becoming more popular especially in the private sector?

[MollyB]

What I've seen is that most kids coming out grad school know STATA, but that isn't really preferred by employers--but they are learning STATA in grad school, so presumably that's what their professors use?

 

I get the sense that in academe economists use Stata, the rest of the social sciences use SPSS. And b/c most PP professors are economists, they end up teaching Stata. I'm not savvy enough to know why economists prefer it and everyone else likes SPSS though...

[biscuits]

According to my economist fiancé, STATA is better for programming and big panel data sets. SPSS is much more of a push-button interface and thus harder to tweak - it's less flexible. Although, apparently IBM recently bought SPSS, which means there might be some changes coming down the pike. SAS is very expensive compared to the other programs, which means it may be hard to experiment with it until you're in a situation in which it is needed for work.

[MPPgal]

As an economist I will say this:

STATA is the preferred program by economists.

SAS by people in the business sector

SPSS is the "user friendly" for lack of a meaner word of STATA (used by non-economists, it is by far the easiest to learn but also the most restrictive)

R is the smart and future version of STATA and SAS with the added value of being capable of data mining, of being free and of having all the assets of math lab (and its equation and graphing solving capabilities), Iris (and its graph capabilities) and wolfram alpha´s mathematica (and all it´s data).

I have used all 4, I am currently learning R and it is a bit harder (as it is a programming language and as a non-engineer it is hard) but after a month of self taught (with an expert coworker who I can go whenever I have questions) I can safely say it is BY FAR the best one as you can do all of them and so much more plus it is free! I will also say the free stats course at carnegie mellon may help a lot in teaching yourself as it teaches the very basic.

[state_school'12]

I'd say, figure out what you'll learn in your MPP program, and teach yourself a little of a different language, if you want to start before the fall. That way, you'll have some experience with both.

[brooklyn11]

I use SAS and R all the time and have used STATA extensively in the past.

 

In my experience, basically no one uses SPSS anymore other than in some into data/stats courses. Its true that R is the future of this stuff and I use it for analysis and graphics all the time because its so flexible.  But its also true that its not as good as SAS with extremely heavy data processing steps. Ive gotten around this in R-only environments connecting R to an SQL server.

 

SAS has been the standard for some time (except economists who prefer stata, statisticians who seem to prefer are, and some others).  Not just in industry, but in the public and nonprofit sectors and at thinktanks that do research.  Just go to any job search engine and type "SAS" compared to "STATA" and you will see what I mean.

 

For summer practice, teach yourself R because its free. Try http://www.statmethods.net/.  In reality, the skills and basic concepts transfer between the various packages.  Plus you will not be able to learn any of these fluenty without on-the-job practice anyway.