Monday, August 24, 2009

Data Analysis With Stata

The Unknown Family went to the Unknown Sister-in-Law's family's house in an adjacent state (their youngest daughter is going off to college, and Unknown Wife wanted to see her before she leaves for the Big Adventure). So, I got a couple of days to myself. Nothing very exciting - I've been grinding data during the day, and went on a couple of longish bike rides (I'm up to 25-30 miles at a time at what for me is a pretty good clip).

On the data analysis front, I finally took the plunge and started using Stata. It's a pretty amazing package of tools. I work with a lot of large and complicated data sets, and there's always a lot of data manipulation before I get to the point where I'm running statistical analyses. When it comes to moving data around (merging data, sub-setting, mean adjusting, etc...) SAS wins hands down. And I've put a lot of time getting my SAS chops, so I'd put off learning Stata for a long time.

But I now understand what so many of my friends have been telling me for so long - once you get to the point that your data is all nice and neat, Stata rocks. I was able to do many permutations of regression models (fixed effects, random effects, robust and/or clustered errors, etc...) in about a quarter of the time it would take in SAS. And while it's possible to work in batch mode by writing "do" files, you can do quick and dirty analyses with drop down menus.

I have seen the statistical light, and it reveals that I'll be doing a lot more with Stata in the future.