Virtual orbital energies

 I would appreciate your views on the following.
 We have been doing some semi-empirical (PM3) and ab initio calculations
 (6-31G* basis set) on a series of ligands (pyridines and imidazoles) to
 try to find correlations between the properties of the ligands themselves
 and the stability constants for their binding to an iron(III) porphyrin.
 Our question is the following.  In addition to being sigma donors, the
 ligands are potential pi-acceptors from, and pi-donors to Fe(III).  Hence
 we are interested in the energies of the frontier orbitals.  Of the
 occupied orbitals, two have pi symmetry and one sigma symmetry.  The two
 virtual orbitals (LUMO and LUMO+1) both have pi symmetry.  Now the
 energies of the filled orbitals presumably have some physical meaning.
 But do the computed energies of the virtual orbitals?  Are we justified in
 using these energies in multivariate correlations with the
 experimental stability constants?
 Thanks for your time
 Helder Marques
         | Prof. Helder M. Marques, Department of Chemistry|
         | University of the Witwatersrand, P.O. Wits      |
         | 2050 Johannesburg,                              |
         | South Africa                                    |
         | Fax: +27+11+339-7967; Tel: +27+11+716-2303      |