Abstract
This paper presents a method to estimate and predict the numbers of participants in sports and recreational activities incorporating the results of cohort analysis of paticipation rates with predicted population sizes. The effects of age, period, and cohort on participation are estimated from data classified by age group and survey period using the Bayesian logit cohort model which successfully overcomes the identification problem in cohort analysis. Based on a scenario concerning future period and cohort effects and keeping the estimated age and cohort effects constant, the participation rates and then the numbers of participants are predicted incorporating with future population sizes. The method is illustrated by predicting the numbers of participants in golf in Japan. Although the predicted numbers mainly depend on the future period effects set up, the proposed method can separately consider the influences of the factors of change in participation and the trends in population size.
