Resumen de Estimating a Markov Model That Incorporates First Visit Decisions and Varying Repeat Frequency
Jay Beaman, Tzung-Cheng Huan, Metin Kozak
Recent articles have demonstrated the value of approaching the analysis of repeat travel using Markov models. The tourism literature identifies behavior that can be introduced into Markov models by expanding the state space. The authors have formulated such models. This article addresses the estimation of parameters of a model for first-time and repeat visitors; if the visitors are repeat then being a frequent or less frequent repeater is addressed. Estimation is for 1991 Travel Industries “In-flight Inbound” data for 1991 Japanese pleasure visitors to the US. Because the estimation is not possible using a standard program, a special estimation program was written. The logic and results of the estimation program, including criteria for determining a reasonable fit, are discussed. It is found that parameters of the model can be estimated and that the fit to the data is possibly good enough that residual variance is random. Unfortunately, other models involving different behavior can be expected to fit the data equally well. So, while a practical implication for researchers is that expanded state space models can be estimated, the challenge raised is that research must establish the actual structure of behavior. Only with that structure defined can valid information be produced for managers.
