How can the parameter estimation procedure for the averaging model of information integration be improved using contemporary numerical techniques?

Estimating parameters of Anderson's (1981) averaging model of information integration theory requires solving an optimization problem by means of numerical estimation techniques. Up to the present day functional measurement researchers utilize AVERAGE (Zalinski & Anderson, 1987), a FORTRAN program using Chandler's (1959) STEPIT iterative least squares to estimate parameters of factorial designs. However, despite the attractiveness of the averaging model to allow fully separation of weight and scale values and to provide empirically sound estimates, the use of the estimation procedure remains limited. The complexity of determining starting and bounding values along with the lack of adequate model information criteria are often reported as drawbacks of the AVERAGE estimation procedure. Contemporary advances in numerical estimation techniques and computational power may improve the effectiveness of the procedure and contribute to its spread among scientists involved in decision-making research.