« New Approaches in Single-Firm Event Studies | Main | "Empirical Holy War" -- Eisenberg v. US Chamber of Commerce »

23 September 2009

Comments

Seth Chandler

I make extensive of Mathematica and am a huge fan of the product. The new statistics capabilities in Version 7 such as GLM, Probit and Logit regressions mean that empiricists ought to take a fresh look.

Four places to see what can be done include http://reference.wolfram.com/mathematica/guide/Statistics.html, http://demonstrations.wolfram.com/search.html?query=statistics http://www.wolfram.com/solutions/Econometrics/ and http://www.wolfram.com/solutions/Statistics/

Tracy Lightcap

I should preface this by saying that my experience with computers pre-dates the point-and-click era. I did actual programming in BASIC, COBAL, FORTRAN, SPSS, and (my favorite still) BMDP when I started work in computer stats. The idea of programming, iow, is not particularly off-putting.

I've pretty much switched over to R for most of my projects. That doesn't mean I use it the same way as I used STATA, however.

I don't use it for data entry or major data manipulation; I use spreadsheets for that then read the data in. And I don't use R base for too much of anything: I use Gary King's ZELIG for almost all the linear model stuff and other "libraries" for most other purposes. ZELIG simplifies the entire programming effort for linear models to next to nothing and other libraries allow you to re-cap SPSS, SAS, and STATA functions for most purposes without undue work.

As an R novice, I have found Robert Muenchen's R for SAS and SPSS Users very useful in pointing out R's differences from other stat apps and a good set of warnings against any general weirdness. Gelman and Hill's Data Analysis Using Regression and Multilevel/Hierarchical Models is pretty good on R and STATA too; they usually have programs for both languages set out for each technique. (I have only begun to work through the first part of G/H, so don't hold me to this.)

Why take on the additional burden, you might ask? Well, R isn't easy and it is so flexible that it can be a royal pain in the tush at times, but it has several real advantages. It's free and it isn't platform dependent. It has the best set of graphic routines extant, albeit that they are hard to work with in some incarnations. R has a useful and pretty sophisticated menu driven GUI, John Fox's R Commander, so I can just breath a sigh of relief and point and click for some jobs. People are adding to it all the time and it always has cutting edge techniques available. And, I believe I mentioned, it's free! I got tired of paying the stat app companies a fortune every time I turned around.

I still have STATA 8 and I use it fairly often. I also still use DATA DESK for many purposes. But I'm trying to go all R all the time. I might add that the students don't find it that intimidating (they only use the Commander) and they love it that they can load it up for free and use it on their own. My colleagues are a bit more reticent, but they are interested.

Theodore Eisenberg

I find the main benefit of Stata 11 to be in the improved programming environment. "Do" files now automatically have line numbers and color code comments and other features. Loops using features such as "foreach" are flagged.
Substantively, the new ability to do competing risk models could be very helpful. If one is studying time to an event, such as trial or settlement, there are always competing risks, such as pretrial dismissal.

Jason

Stata 11 crashes with a slightly higher frequency than Stata 10, however I am still happy. As for switching to R, you should only do that if you don't co-author papers (the vast majority of potential co-authors use Stata).

David Stras

I have Stata 9 and it sounds as if Stata 11 is a huge upgrade from that earlier version. But I have also been throwing around the idea of using R. Does anyone have any opinions about the relative utility of Stata 11 versus R?

Bob Lawless

I went ahead and got Stata 11 and mainly for the graphics as Frank suggests. The new version can use the fonts on your system, making it easier to integrate graphs from Stata with the formatting in a paper. Also, I needed to get the "SE" flavor and figured as long as I was upgrading I might as well get the new version. If I did not have these two reasons, I probably would not have gotten Stata 11.

Now for my question . . . is anyone making good use of Mathematica? I've had a few occasions to use it, but nothing fancy. I was wondering if anyone was making creative uses of it in ELS circles.

MH

Excellent point, Frank. I'm coming to the upgrade decision from v.10.

frankcross

Don't be scared. I've noticed a few annoying differences in the dropdown menus, where you have to look up the answers, but nothing different when you just type in commands. What was your last version of stata? I think the big upgrade/change for most purposes was in the graphics for stata 10. I'm not sure it's worth an upgrade from 10, but I never got 10, and it's worth an upgrade from 9

The comments to this entry are closed.

Conferences

July 2014

Sun Mon Tue Wed Thu Fri Sat
    1 2 3 4 5
6 7 8 9 10 11 12
13 14 15 16 17 18 19
20 21 22 23 24 25 26
27 28 29 30 31    

Site Meter


Creative Commons License


  • Creative Commons License
Blog powered by Typepad