there should be a new version of Regression.pm (0.5) that should soon make its way to cpan. It includes computing standard errors, which was quite some effort to debug. I am waiting for Ivan to package it---I don't know how.

pfeldman: actually, despite my misleading error message, the error was not internal but your own. You entered an observation that had no Y realization. I have changed the error message for the next version.