STATA or R for statistics software?

[iowaguy]

Hi, I'm enrolling for fall PhD classes, and can choose between two nearly-identical intermediate-level statistics classes.  One uses/teaches STATA, and the other uses/teaches R.

 

Which would you recommend (STATA vs R)?  I have a basic knowledge (and license) of SPSS already...  Thanks in advance!

[bedmas]

i learnt R first and prefer it 100x more than STATA. To me, R is a lot more intuitive. The help-files are great, there's so many discussions online if you need more help, and there's a package for everything. 

[ak48]

Never learned STATA but took a stats course in undergrad (I'm a physics major) that used a lot of R to do Markov modeling. I liked it a lot. Kinda reminds me (in a good way) of Matlab in terms of ease of use, straightforwardness of coding, and package/documentation support.I

[IRToni]

I would say it depends on what you want to do, and how much you think you'll end up using Stats, aside from programming experience etc.

 

STATA has a GUI, and is therefore usually easier to deal with as a beginner (and more similar to SPSS, I believe). The learning curve for R is, IMO, a lot more steep, and if you've never used programs without GUI, it might be harder to use.

At the same time, STATA can be annoying exactly because of this issue, the GUI, I find, can get in the way, and using do-files doesn't necessarily help. If you're planning on using a lot of stats in your research, biting the bullet and learning R migt be worth it in the long run. Otherwise, STATA is, IMO, easier to learn.

[ANDS!]

R.  So much more robust, and you'll seem (to others) smarter for it.  If you want a GUI, get RStudio.

[SeriousSillyPutty]

My stats class didn't really try to teach us how to use any program, but did examples in SPSS.  Personally, I would have liked to learn how to use R, because it's free but not intuitive enough for me to learn on my own, having never learned something without a GUI.  So I say do R, then you can do your work on your own computer, and learn a program that you can use again when you're not a student, without having to pay an arm and a leg.

[wine in coffee cups]

R.  So much more robust, and you'll seem (to others) smarter for it.  If you want a GUI, get RStudio.

This is a very good suggestion. Download and install R and then download RStudio. Both are free. RStudio offers a clean interface to organize your code files, console output, workspace objects (you can double-click on your dataframes/matrices to look at them!), and plots. I'm really surprised it isn't more widely used.

RStudio is also well-integrated with Sweave/knitr so that I can do my data analysis in the same file as my write-up. Then all my tables, regression coefficients, plots, etc. end up right in my document or slides without any copying/pasting or manual typing of Word tables if I use R packages like xtable. Just click a button and it spits out a PDF. Definitely a learning curve there but saves a ton of time in doing homework or making presentations once you have the hang of it.

[MAME5150]

Ya, I also recommend R. I think it's really at the vanguard of statistical software. 

 

I generally think that GUIs are a horrible way to go. The logic usually follows that these are good to learn on. However, I feel like I rarely encounter people that try to move away from GUIs once they learn how to use them. Scripting forces you to know a lot of your analysis. Learn how to script and you'll know your analysis like the back of your hand. This leads to my next point: Your data should be reproducible, so regardless of what field you are in you should be using scripts with annotations and some kind of version control to have a clear space that charts the evolution of your findings. 

 

With that said I will say that something I have encountered is that depending on how niche your analysis is, you might run something in one program, write up the results and send it out for publication only to have a reviewer ask you to redo the SAME analysis in another program. 

 

Ideally, you learn them all. This is why i recommend R. Once your learn a scripting language taking those principles across programs is very easy. I script in STATA, but I can bet that the GUI is how that program will be taught. 

[Zrome]

Between the 2 I'd recommend R; however, my preference has always been MatLab. 

[fuzzylogician]

All things being equal, I'm with the others -- R is so much more powerful. However, before choosing you should find out what others in your lab or department use. Ask your advisor if s/he has a preference, and ask more advanced students what they use. It'll make life easier for you when you need to communicate with others, when you need help with your code and when (if) you inherit code from others. 

[Monochrome Spring]

I like R, since that is what I learned in my owns statistics course, and because that's what everyone in my lab group uses.

[juilletmercredi]

STATA is easier, but R is free and everyone is using R now.  I would take the R class, because STATA is easier to teach yourself IMO if you want to learn it later.  And there is indeed code for everything in R already online.

 

GUIs in stats programs are just annoying.  Learn the code.  It gives you so much more control over your analysis and it's much easier to make individual tweaks and SAVE your work.

 

Whether or not you need to use the same package as others in your lab will really depend, honestly.  A large enough department will have students who use everything.  I think pretty much every major stats package is well-represented in my department at this point.

[SiraRaven]

I've never used STATA, but I really like R.  It's not always straightforward, and there are weird typing issues that I've run into, but there are a whole lot of resources online and a lot of nice packages.  It's able to make some really beautiful graphs if you know what you're doing, and I've ended up making good use of RCaller, a Java library for calling R.  However, I would agree with what others have said - if everyone in your lab uses STATA, that's probabaly the way to go, since you can ask your labmates for advice.

[Guest |||]

Hands down R

[SensLu]

Agreed, I recently had to make the same decision, chose R because I found it to be referenced more by professors in environmental studies and it's a free program, so after my class I can still use it.

[NoontimeDreamer]

I'm definitely a fan of R because, like others have mentioned,

 

(1) It's rather intuitive once you get the basics

(2) Lots of helpful forums and the "? query" help function

(3) Really strong plotting (I'd suggest the package ggplot2 for your graphing needs)

(4) It's free!!

Edited July 18, 2013 by NoontimeDreamer

[kmitch]

From what I hear (from several computational neuroscience and other clinical psychology professors), R is becoming the industry standard. And its free!   

[jeffster]

Don't make the mistake of thinking you'll learn one statistical package and forget the others though; it's more important to learn what you're actually DOING to the data when you analyze it.  You can ALWAYS learn new sets of commands, if you understand what's happening in the background.  In other words, don't think of statistical software like a little black box that spits out magic answers when you click/type the right combo.

 

This is very important because you don't always get a choice.  Stata is very popular in academia, but different fields and even different employeers within the same field use different statistical packages.  For example, if you're going to work for the government in DC you simply must know SAS.  I personally use Python for everything, but I'm currently working on research for a professor who uses Stata, but we're dealing with a dataset that comes only in SAS format.  So I have to know a little of all three for my work.

 

Tinker with as many of them as you can (if data analysis is what you'll be doing a lot of, that is).  Or even better yet, if you're clear on what you want to do with your degree, find out what package is used in that area and stick with that as much as possible.  Otherwise, there is no right answer.  They will all get the job done in the end, and the one you'll like best is the one you know how to use well!