The SAS software corporation thinks so, to the tune of $5 million a year; as mentioned on page 3 of this
NYT article. The article provides some useful insight into how major corporations motivate highly-skilled employees; SAS seems to be a more sedate version of Google. There is also some discussion about current developments in the world of statistical software: SAS has modified its software so any program written with R works seamlessly with SAS technology.
A more sedate version of Google? Stata perhaps. I spent much of the late 1980's using SAS on a mainframe (Look it up in a history book) before there was a PC version. It was better than SPSS. I think they have moved away from the emphasis on statistics into more general information management.
ReplyDeleteAccording to the NYT article - of the 100 largest companies worldwide - 92 use SAS software. But maybe this is for information management more than statistics.
ReplyDeleteDuring the stats-course I did at Michigan this summer, SAS was most often used (followed by R, then Stata), but it is quite obvious that Stata is the economists' weapon of choice; and R is the statisticians' poison.
I'm aware that SAS is usd by the CSO (and in a lot of American universities). But the NYT article also says that SAS has modified its software so programs written with R work seamlessly with SAS technology. This motivates me to look no further than R when considering the use of a package besides Stata.
R is not that easy to use from what I can recall. But because stats people like it there are lots of applications (packages) written it. Roger Koenker has written the quantile package I think so you know you are getting the real deal. There are also nice packages for non- & semi-parametric modelling which some of our colleagues here use.
ReplyDelete