Efficient Methodology for Generating Synthetic Populations with Multiple Control Levels

This paper details a new methodology for controlling attributes on multiple analysis levels in a population synthesis program. The methodology determines how household- and person-level characteristics can jointly be used as controls when populations are synthesized as well as how other multiple-level synthetic populations, such as firm and employee or household and vehicle, can be estimated. The use of multilevel controls is implemented through a new technique involving the estimation of household selection probabilities on the basis of the probability of observing each household, given the required person-level characteristics in each analysis zone. The new procedure is a quick and efficient method for generating synthetic populations that can accurately replicate desired person-level characteristics.