As I approach the end of my graduate math studies I have been experimenting with lots of different statistical software packages, trying to build skills to help me find work for after graduation. The software I like best is Systat. It seems powerful and easy to use, and with good documentation. Statistica, with its powerful SVB language, is also very good (if you forgive its tendency to crash sometimes). Those are the two tools that I would prefer to work with.
But when I search on job websites I find that nobody seems to be hiring for Systat or Statistica users. On the job market SAS seems tops by far, SPSS second, with SPlus/R running about third.
My question for this forum is: is there some good reason that Systat and Statistica are so unpopular, or is it simply that SAS and SPSS do much better marketing to the industry?
it's a good question, I would like an answer too, but I don't know if anyone can provide any. There are lot's of good statistical packages like Stata, JMP, Minitab, Systat, Statistica, XLSTAT,....but I noticed too, that the biggest majority of companies work with SAS and SPSS. Don't know why, I mean SPSS is very easy to use and easy to get addicted to, SAS is VERY powerful, but there must be another reason.
Maybe it's the way you're searching. I find jobs when I go to monster.com and search by keyword. For example, type in "STATISTICA" and you’ll find 12 job postings.
You could also try looking at a StatSoft's customer listing and then going to those company web pages directly to find STATISTICA job opportunities.
I don't consider 12 Statistica jobs to be a significant number, for the whole of the U.S. For comparision, a quick check today shows SAS mentioned in 1607 listings on Monster.com.
My question was really an attempt to get opinions about why a few statistical software makers dominate the industry to the extent that the many other excellent software choices are practically invisible on the job market.
I was also trying to discover whether those other software choices are more-widely used than my searches of the job market would suggest.
I think WeeG hit on this with SPSS is very eay to use, as is SYSTAT. Both really only require an understanding of statistics and the ability to read a user manual. With SAS and S-PLUS/R and other that require knowledge of a GUI type interface (and as often said there is a steep learning curve involved).
So, as I see it companies using SPSS , SYSTAT etc... might be able to get away with on the job learning to operate these packages, others require a bit more assumbed and specilized knowledge often not afforded in the work place in terms of retraining staff.
R and the like (at least in my area) are becoming industry standard because of the power and graphics etc...
So I think in industry SYSTAT and SPSS seem to be phasing out (again, at least in my area) and it is often assumbed that graduates have experience using dropdown software experience, so is not not required as a specilaity selecttion criteria.
popularity = marketing + utility
SAS and SPSS are very useful and have very strong marketing programs. If you've ever been to a statistical meeting, chances are that it was sponsored either by SAS or SPSS. R has essentially zero marketing, but is extremely useful. R has other nice qualities that are absent with SAS and SPSS. Namely, R is free to use, modify, redistribute, and has a sophisticated system for extension packages. Every function that R executes can be verified/checked immediately by inspecting the relevant source code. Also, if you have an R question, you can simply ask the actual R developers via the r-help mailing list.
I don't use Systat or Statistica, but perhaps they are lacking in either marketing or utility or both. However, I did speak with a Systat rep. at the Joint Statistical Meetings Expo this past August, and I am very impressed with their graphics.
SAS is so popular, because it is best for datawarehousing. It has connectors to any big datababase and offers database integration solutions which are without real competition in the market. E.g. you can integrate a DB2 database in china, an oracle database in houston and an MS Access DB in London and then use it, as if it was one. This is, why a lot companies run their datawarehouse with SAS! Further SAS is the most perfomant statistical package to join big datasets. And this was still 5 years ago a big issue.
R would be much more popular, if it had a name, which would be understood by search engines. One character is simply to short. E.g. "R-Stat" would drive a lot more traffic on R websites.
Consuli
I don't know all of the capabilities of SAS, but I just want to clarify that all of the statistical software choices mentioned so far in this thread have ODBC import capabilities, and so they, also, can inquire from "a DB2 database in china, an oracle database in houston and an MS Access DB in London and then use it." SAS may be better at combining data sources (I don't know), but I didn't want to leave the impression that other ODBC-compliant software is restricted to reading only local data sources.
I think that all the packages mentioned here can do most of the common things statisticians need. I think that the differences are:
1. How easy is it for a non programmer.
2. How much can the software be extended (procedures that don't exist in the menus)
3. Is there support for modern statistical methods like Robust Statistics, etc,...
4. How common the software is, "if everyone use it, I will use it", maybe this is what managers say.
5. Price
Posting Permissions
|
|
Reply With Quote