Statistical software is in a phase of rapid change. The graph on the left shows the citation rates for the major statistical packages on Google Scholar. It's taken from an interesting post here.
It is clear that the traditional packages – SAS and SPSS – are in sharp decline (though the decline in SPSS seems to have slowed.
But what is also clear is that both Stata and R are increasing solidly. Both of these packages have a key advantage over the competition: users can write commands that add to the package's functionality. This means that you are likely to be able to access cutting edge procedures in these packages before they are implemented (if they are implemented at all) in other packages.
From the teaching point of view, both are attractive too. While SPSS and SAS have annual licenses, Stata student licenses are permanent, and R is, of course, free. With SAS or SPSS, a graduating student rapidly ends up unable to afford to renew the license for the software and so finds they have learned a package they cannot now use. Not a problem with R – it's open source – and Stata users will be able to keep whatever version they have indefinitely.
However, a major consideration is research output. While there is considerable interest in Stata in among PIs in College, a package that emerged with a solid and enthusiastic fan base was GraphPad Prism. BCSS gets a small but steady trickle of support requests for support with Prism, which is a package ideally suited to the workflow (and thought flow!) of lab-based researchers. And both we and the users agree that the manuals are remarkably well written and informative. So we plan on supporting Prism as long as this enthusiasm continues.
College will have to consider seriously how we invest money in software, taking into account the teaching potential, the research productivity and the accumulated expertise. We will be cavasing opinions, both by survey and by contacting PIs, but in the meantime we would be interested to hear from people with with views or interests.