The linear viscoelastic dynamic modulus, |E∗|, has become the primary material property of interest for asphalt concrete mixtures. The shift towards linear viscoelastic characterization of asphalt concrete is due in large part to national efforts to develop new fundamentally based pavement design tools and techniques. Within the pavement community, there is a substantial interest in using predictive models to estimate | E∗| because key structural design decisions are based on its value. Moreover, for many projects, these critical decisions must be made even before the materials are selected. In response to the need for predictive capability, numerous equations have been developed, but most require knowing the asphalt binder shear modulus, |G∗|. This property is currently measured as part of the purchase specification process, but it is not measured at enough temperatures and frequencies to be directly useful with these predictive models. Instead, agencies that want to use the predictive equations must complete additional testing that may require several days to complete and requires the purchase of more sophisticated equipment. It is the purpose of this paper to show ways that the specification data can be processed to provide nearly the same information as a more complete testing suite. This effort is possible because although the specification data are sparse, they still cover the range of conditions that need to be assessed. Additional surrogate models are needed to fully apply the proposed methodology, and the development and verification of these models are presented as well. The ability of the limited data calibration process to match characterization from more complete testing is demonstrated. Finally, it is shown that observed differences between the complete and limited calibration processes are reduced when applying two |E∗| predictive models.