Stats packages

Russ Poldrack rpoldrac at s.psych.uiuc.edu
Wed Mar 8 11:51:56 EST 1995


jon.laake at basalmed.uio.no (Jon Laake) writes:

>Try STATISTICA for MacIntosh (Statsoft, Tulsa, Oklahoma). I've only tried the Windows 
>version, but think The Mac version is the same. It has excellent graphics and very
>user friendly statistics options. Most kinds of ANOVA will work well, but STATISTICA 
>does not offer the maximum likelihood algorithm, which reduces its usefullness with respect 
>to unbalanced designs. In my experience only BMDP does that well, which is sad because 
>its such a terribly unelegant DOS thing. 

A while back there was a news item in _Science_ about how Leland Wilkinson
of Systat accused the folks at StatSoft of stealing some of his
algorithms from Systat.  I think there was also some news about
Statistica producing incorrect results for some analyses, though that
may have been fixed in subsequent releases.  It was always my
impression that Wilkinson was a real standup guy and that he wouldn't
accuse someone of something so serious without strong evidence, so my
personal decision would be to stay away from Statistica.  Your mileage
may vary, however.

I have been very happy with StatView and SuperANOVA on the Macintosh.
Both have very nice interfaces, and SuperANOVA is by far the easiest
way to do ANOVA that I have seen.  The new version of StatView is
native for PowerMac and is very fast, while SuperANOVA has not yet
been released in a native version.  I heard rumors a while back that
SuperANOVA might become a plug-in module for StatView, but no
corroboration has arisen for that.  Both of these packages have
flexible transformation options and can deal with large data sets.

rpoldrac at s.psych.uiuc.edu



More information about the Neur-sci mailing list