R help - Mar 2006 - Scaled or unscaled variance covariance matrix

Hi, does anyone know which of the scaled or unscaled version of the
variance-covariance matrix of regression coefficients (as produced by
ls.diag) is the correct one, or when one is better than the other? I am
interested in univariate linear regression (lm). Thank you very much.

           Dipl.-Psych. Johannes Ullrich

            Philipps-Universit?t Marburg
                 Fachbereich Psychologie
                    AG Sozialpsychologie
                           35037 Marburg
                                 Germany
                  T. ++49 6421 / 2823654
                  F. ++49 6421 / 2823789