blinn@furilo.dec.com.UUCP (04/02/87)
In a message to sci.math.stat and sci.misc posted 30 Mar 87 16:04:31 GMT, Douglas Clark of Bath University in the UK wrote.. > When we get our new UNIX machine next year I am interested in the possibility > of bringing in either SAS or PSTAT as an alternative to SPSS-X. I understand > that SAS can, as at Cambridge University, cause system problems if not > installed on specific IBM machines and would welcome comments thereon. > I have the impression that Roald Buhler's PSTAT acts well in interactive > situations but have the impression that he is a one-man-band. But my real > problem is that I don't really know much about what users can actually do > with these packages. Any information is welcome. Or should we stick to SPSS-X. > > Thanks in advance. > -- > Douglas Clark Voice: +44 225 826826*5214 > Computer Unit, Bath University, JANET: Clark@UK.AC.AUCC > Claverton Down, Bath, Avon, UUCP : seismo!mcvax!ukc!bath63!cc_dgdc > England BA2 7AY ARPA : cc_dgdc%ux63.bath.ac.uk@ucl-cs.arpa I've worked with Roald Buhler in the past, and while it's true that he has a rather small company, he is very dedicated to what he's doing, and he is very good at porting his software to a wide variety of environments, so there's a good chance that he'll be able to get it running on the Unix environment you're going to have (you didn't say in your message WHICH Unix system you'll be getting -- they are not all the same, you know..). Almost all of the P-STAT code is written in ANSI Fortran, so if you've got a good ANSI Fortran compiler on the target system, you should get reasonably good performance. P-STAT is a fairly complete package, with plenty of useful stuff for users. It works well both interactively and in batch. I believe that most of the SPSS-X system is written in ANSI Fortran, as well, with some routines (typically I/O) written in some other language (often Assembler) for specific target machines. (I used to maintain the SPSS conversion for Burroughs Medium Systems under MCPV, back in the days before SPSS-X; much of the SPSS code was used in building SPSS-X, albeit with a new user interface.) So, it's possible that there will be an SPSS-X for the environment you're going to have. You should ask SPSS, Inc. about this BEFORE you get the Unix system and discover that SPSS doesn't run on it. They are much less interested in outside conversions of their code than they were in the past, and if they don't have access to the target, they may not have done the conversion. Once again, not all Unixes are the same. I gather from your posting that you're already familiar with SPSS-X. As for SAS, I'll be amazed if they have it running under Unix. Much of the code in SAS is (or at least was) written in PL/I, and I don't think there are any Unix implementations with a production quality PL/I compiler. So, if you're in the mood for SAS, you should ask SAS Institute whether their software will run on the system you're purchasing. Some problems have been observed by VAX/VMS sites running SAS, but these appear to be due to some constraints in the way the software is installed, not due to any inherent deficiency in either the SAS software or the VAX/VMS environment. Since the SAS conversion for the VAX/VMS environment is relatively new, it's not surprising to see some "teething pains". I can't speak for any other SAS environments. SAS is a very complete and powerful package. In many cases, people choose the computer system based on whether or not it will run the software they want, rather than choosing the software based on the computer system they're stuck with. The former is the better approach. Obviously, you should get the user documentation for all three of these packages and compare it. I think you'll find that they all provide all of the basics, but differ in some of the more esoteric routines, as well as in the user interface to their facilities for data management / manipulation. Choice of statistical package is a lot like choice of religion -- each has its vocal adherents as well as vocal critics. In the end, you pays your money and you takes your pick. Standard disclaimer about the above being my personal understanding, and not in any way the official position of my employer. Regards and good luck! Tom Dr. Thomas P. Blinn <-- Ed. D. in statistics U. S. Volume Operations Digital Equipment Corporation Two Iron Way -- MRO3-3/R18 Marlboro, Mass. 01752-9003 Internet: blinn%furilo@decwrl.dec.com Easynet: FURILO::BLINN Usenet: {decvax,ucbvax,allegra}!decwrl!furilo.dec.com!blinn Voice: (617) 467-5562 Note: Opinions expressed herein are my own, and do not necessarily represent those of my employer or anyone else, living or dead, real or imagined.