statistics software

[iowaguy]

Do you have a favorite statistics software package that you have on your own personal computer?  Or do you use the same software as the department offers in their computer lab?

 

I'm hoping to use these upcoming months before I start grad school to brush up on my statistics.  Ideally I could learn a software package or two...

 

Pros/cons of the common/popular statistics software packages?  Thanks!

[lilgreycells]

I think this can be very field specific. I trained in my undergrad and Master's programs using SPSS (psychology and higher ed) but in my PhD program we use STATA because of it's superior abilities to handle regression (which is more popular in my field since we often rely on econometrics). So the best program for you is probably dictated by the programs the natural resource field uses. I have personally used SPSS, STATA and R and can say that they all have their worthwhile pieces it just really depends on what you're doing with them.

Oh and depending on the package your field uses you may want to wait on purchasing the program for your personal use. STATA, for instance, is extremely expensive so you would want to wait to buy it through your school.

[Eigen]

A lot of this depends on what type of statistics/data analysis you want to do. 

 

I use Origin and Prism for most of my data analysis, and they do all the statistics I could ever see myself needing, but there are a lot of specialized packages (as mentioned above). 

[ohgoodness]

Let me just say that it certainly is not excel.. I just spent 12 hours manually entering data into a new matrix and now I feel like I really need to rethink my life choices..

Seriously tho - I love STATA. I used to like SPSS but it becomes old fast and people tend to misuse it, i.e. using it for mixed methods projects or so. If you can get a good license from your university for STATA then you should be solid. SAS for data handling, Stata for regression and descriptives, excel for charts, ArcGIS formaps and then we are all good.

I want to develop into R + latex but another day, another project.

[fuzzylogician]

I use R. It's much more flexible than other software I've used in the past (mostly SPSS) and it's what others in my program use. It has a steeper learning curve initially than other software, but once you're past that it's also much more flexible. I still create graphs for papers in Excel, but there's not much to learn there.

[judowrestler1]

At my undergrad we always used SAS. It's pretty expensive though.

[sisyphus1]

R is the way to go. Free, elegant, flexible.

SAS is still popular (able to handle larger data sets more easily).

[TheFez]

I am not even sure that software is "program specific" - I think it often just depends on the preferences of the instructor.

We used SAS's Enterprise Guide On Demand system which is nice since it's online and very feature rich.

I like SAS's newer program JMP. It has some really powerful modules like a DOE module, structural equations, conjoint, etc. For SAS users it also has a SAS interface. It has an Excel add-in, and is very user friendly and very fast with large data sets.

Many also use SPSS.

Those (including me) who do econometrics prefer STATA

There are the R folks, Matlab folks and even Mathematica folks.

I don't think it matters all that much - find a program that's free, and used in your discipline.

[lilgreycells]

I am not even sure that software is "program specific" - I think it often just depends on the preferences of the instructor.

We used SAS's Enterprise Guide On Demand system which is nice since it's online and very feature rich.

I like SAS's newer program JMP. It has some really powerful modules like a DOE module, structural equations, conjoint, etc. For SAS users it also has a SAS interface. It has an Excel add-in, and is very user friendly and very fast with large data sets.

Many also use SPSS.

Those (including me) who do econometrics prefer STATA

There are the R folks, Matlab folks and even Mathematica folks.

I don't think it matters all that much - find a program that's free, and used in your discipline.

Great point though it depends on your program as to whether the software is program specific I would say. You have to give extreme justification in my program as to why you would use anything that wasn't STATA. For us, STATA is the only software allowed in 5 out of our 6 baseline statistics courses. But if your program is more flexible I definitely agree just find something that works for you.

[Eigen]

Also keep in mind that if you use what other people are using, you have a lot more resources when it comes to troubleshooting. 

[booksnlooks]

I'm a huge fan of STATA, which is particularly great when it comes to the Help Manual. It explains everything well and it also has some useful examples. I also found the coding fairly intuitive but that could be because I used to use SHAZAM.

[Armadilla]

Let me just say that it certainly is not excel.. I just spent 12 hours manually entering data into a new matrix and now I feel like I really need to rethink my life choices..

Seriously tho - I love STATA. I used to like SPSS but it becomes old fast and people tend to misuse it, i.e. using it for mixed methods projects or so. If you can get a good license from your university for STATA then you should be solid. SAS for data handling, Stata for regression and descriptives, excel for charts, ArcGIS formaps and then we are all good.

I want to develop into R + latex but another day, another project.

ohgoodness, why is STATA goog for regression, in your opinion?I've just been recently trying to reduce everything to one software package (or at least two, like matlab and Arcgis), but then I noticed a lot of people give accolades to stata, so no I am tempted...  Other than regression and descriptive stats, would you recommend it for anything else?

[lilgreycells]

armadilla,

STATA was built specifically to handle regression and I think that's why so many people like it. Plus the options available to create complex econometric models are top notch.

[ohgoodness]

I would be very interested in hearing if anyone has any good ideas about useful programs for qualitative work? Specifically for mixed-methods or interviews. 

[Quant_Liz_Lemon]

R!

[juilletmercredi]

My lab is mixed-methods and we use NVIVO for our qualitative work.  I don't use it  because I do the quantitative side of the mixed-methods, but I've seen it in action and it looks pretty cool.  The rest of our med center (lots of medical sociologists and anthropologists) use Atlas.TI, which is another qualitative software.  I think it's mostly a matter of preference between the two, but I've never used either.

 

As far as quant packages go, R is the most robust as far as handling a lot of different kinds of models well and it's FREEEEEE which is always nice, since stats packages cost in the hundreds.  After that, SAS and Stata are both good for different things.  I think Stata is more intuitive, but I taught myself how to use SAS because it was on all the computers and that's what we used in our Analysis of Change class.  It's been useful because a lot of the health departments want people with SAS skills instead of Stata, for some reason, but I think Stata is better for certain types of data.  SAS is great at manipulating data though.  I would wait to get on campus before purchasing, though.  Your university probably has a discounted licensing scheme (here students can do a 12-month license of SAS for $99) and your advisor may pay for your license out of his grant (that's how I got SPSS, although I've been begging for SAS.  I always come at the wrong time - our license starts in August and it is not prorated.)

 

I think SPSS is the worst of the crew, BUT it's the easiest to get on a student computer and I think it's the easiest for learning stats because of all the point and click and neat little tables.  They have GradPacks where you can buy up to a 3-year license for student use.  You can find some here: http://studentdiscounts.com/spss.aspx?gclid=COXI2Mqyj7UCFQ2i4AodIhkAqg

 

There are specialized packages for special functions, too. My advisor is fond of MPlus because he does a lot of latent variable and longitudinal modeling.  But MPlus is expensive!  Good news is that there's a student price that is like 1/3 the original cost.

[Armadilla]

I would be very interested in hearing if anyone has any good ideas about useful programs for qualitative work? Specifically for mixed-methods or interviews. 

Atlas.ti! It seems to be pretty good for interviews and any language-related data (such as audio and transcribed interviews). I also heard they have a pretty good video and photo integration option but I am not familiar with it. Back in the days, they had a free trial package for students, they still might offer it now.

Edited January 31, 2013 by Armadilla