CHEN, YONGCHANG; PHD

                         UNIVERSITY OF PENNSYLVANIA, 1987
 
                         BUSINESS ADMINISTRATION, MARKETING (0338)
 

                         Most new product diffusion models proposed in past two decades are deterministic models. This
                         dissertation proposes adaptive and varying-imitation stochastic parameters models. The adaptive model
                         can accommodate sigmoid and nonsymmetric shapes with double concavities. The maximum
                         discrepancy between deterministic and adaptive stochastic model increases as q/p, where q and p are
                         imitation and innovation coefficients, increases and it is almost constant as q/p is constant. The
                         varying-imitation model can accommodate sigmoid, exponential and nonsymmetric shapes with double
                         concavities. In the analysis of variance of Monte Carlo simulation we find prediction performances of
                         adaptive and varying-imitation models are not sensitive to data points, but the Bass model is sensitive to
                         data points. We use black and white TV, color TV, clothes dryers, freezers and dehumidifiers in the
                         empirical application. We also study the implications of optimal advertising policy and risk behavior under
                         stochastic diffusion. The last chapter is the summary, conclusions and directions for future research.
 

Social Systems Simulation Group P.O. Box 6904 San Diego, CA  92166-0904 Roland Werner, Principal Phone/FAX  (619) 660-1603