Estimating joint preference: A sub-sampling approach

摘要

Individual and joint preference inform about different aspects of a product's marketing strategy. While individual preference is easily measured, joint preference is expensive to obtain. The author proposes a sub-sampling approach that uses MCMC and data imputation techniques to estimate individual and joint preference. It requires individual data from the entire sample and joint data from a fraction of the sample. Empirical evidence suggests that the sub-sampling approach works well when joint data are collected from 25% of the sample. Predictive and correlation tests demonstrate the superiority of the proposed approach. Greater than 50% reduction in data collection cost is shown.