Abstract
It is unrealistic to assign a single Coefficient of Variation (COV) to a geotechnical parameter. The COV depends on the quality of the site investigation and the property evaluation methodology. It is also important to recognize that the COV of a design parameter may be reduced because of: (1) availability of multiple test data in a typical site investigation and (2) spatial averaging along the critical slip curve. It is necessary to characterize the correlation structure in both cases to be able to realize COV reduction, which can be used to justify design savings using the reliability-based design framework.