This paper explores the transferability of the Multiple Discrete-Continuous Extreme Value (MDCEV) model for activity type and duration using various transfer methods and sample sizes. This study employs the data of travel demand studies in two major cities, Shiraz and Mashhad in Iran. The model is first developed for Shiraz and then transferred to Mashhad. The adopted transfer methods are transfer scaling, Bayesian updating, combined transfer estimation, and joint context estimation. Aggregate and disaggregate transfer measures are adopted to examine the transferred models' general prediction and policy predictability. The results indicate the joint context estimation method's superiority in terms of estimation and policy prediction powers. The available massive data to the authors enabled measuring the value of sample size in this study. The sample size sensitivity analysis revealed a decrease in the marginal gain of the transferred model's performance as the sample size increases. Remarkably, the transferred model outperforms even the locally estimated model when 1) advanced transfer techniques are applied (i. e., the combined transfer estimation and the joint context estimation), and 2) the application context sample size is large enough (i. e., more than 30 percent).