Dear Ondra Whether the covariance matrix is singular is only a problem if it concerns the independent variables, or if you want to use the standard multivariate tests (Wilks lambda etc.) that include inverses or determinants of the covariance matrix.
In your case, where the shape variables are the dependent variables, you can use the standard regression methods. For testing the statistical significance, use an approach such as permutation tests with a statistic such as the sum of squared covariances between centroid size and the shape variables. Only the testing requires special methods. I hope this helps. Best wishes, Chris At 09:11 PM 2/7/2007, you wrote: >Dear morphometricians, > >I want to perform multivariate regression of landmark-specific signed >asymmetries on the centroid size. The problem is I am dealing with the >structure with object symmetry (namely, ventral side of a skull). Thus, >dimensionality of an asymmetric shape variation is reduced and its >covariance matrix is singular. I would greatly appreciate your advice >how to modify standard multivariate regression procedure in this case. > >Thank you in advance > >Ondra Mikula >([EMAIL PROTECTED]) > > >-- >Replies will be sent to the list. >For more information visit http://www.morphometrics.org *************************************************************** Christian Peter Klingenberg Faculty of Life Sciences The University of Manchester Michael Smith Building Oxford Road Manchester M13 9PT United Kingdom Telephone: +44 161 275 3899 Fax: +44 161 275 5082 E-mail: [EMAIL PROTECTED] Web: http://www.flywings.org.uk *************************************************************** -- Replies will be sent to the list. For more information visit http://www.morphometrics.org
