Bugs in statistical software also always occurred. Best known are statistical programs from the 70ies and early 80ies that ended up with negative variances (which is statistically impossible) for particularly small numbers that resulted in internal computer register underflows when these values were squared. Another problem that occurred very frequently was related to random number generators that generated anything else but random numbers - in many cases the numbers repeated themselves in exactly the same order after only a few hundred or thousand values inbetween.
So - whatever type of software you have - never expect it to
be bug-free. Under this point of view, evaluate Webstat at
http://www.stat.sc.edu/~west/webstat/version1.0/ and answer
the following questions [15 Points in total]:
You may want to work on this particular
exercise in groups of 2 or 3 to be most effective.
Also make notes what kind of a computer you used for your software test, which Web browser you used (Netscape Version x.x, Internet Explorer Version y.y), and what your exact inputs were.
The intention is to summarize all your comments and send an
e-mail that contains your bug reports to those people
that develop WebStat. Typically, software
developers, in particular at universities, are very thankful for
any comment they get on their software!