Probabilistic scaling (Proscal) is a powerful and flexible analytical tool for representing and understanding how people perceive and make choices involving complex objects.
The power of Proscal comes from its sophisticated modeling of peoples' perceptions and of their choice processes.
The underlying preference model used throughout Proscal is based on the unfolding model developed by C. H. Coombs.
Estimates of all the parameters are obtained using maximum likelihood methods.
Proscal models both objects, such as consumer products, and people or market segments as probabilistic distributions in a multidimensional space.
These probabilistic distributions measure the precision with which people view different objects and different attributes.
Probabilistic modeling allows Proscal to capture the complexity of people's perceptions and to test hypotheses concerning those perceptions.
By realistically modeling the richness of peoples' perceptions, it is possible to obtain quality estimates of their choices.
In a business context, Proscal can estimate the perceptual shares for existing and experimental products. It can also carry out what-if modeling which makes it possible to evaluate different product development strategies.
Proscal can utilize a variety of input data, types such as: product profiles, liking ratings, choices, similarities and preference ratios.
These different types can be analyzed individually or in combination with other data types.
Models can be constructed for up to sixty objects in spaces having up to six dimensions.
Euclidean or city-block metrics can be specified.
Proscal comes in two versions - an interactive windows version built around S‑PLUS and a command line version that can run in a MS-DOS environment.