Sample design for the Housing Assistance Supply Experiment

This Note proposes a sample design for field surveys to be undertaken as part of the Housing Assistance Supply Experiment. The design provides a plan for determining the numbers and characteristics of residential structures at each experimental site to be selected for annual surveys of the structures, their owners, and their tenants; the plan is illustrated with data for the Saginaw, Michigan, Standard Metropolitan Statistical Area, one of the candidate experimental sites. The study concludes that a sample of 5,000 housing units allocated according to the design will allow for the reliable estimation of supply responsiveness for all high-impact sectors of the housing market; exceptions are eight strata of large structures, whose sampling quotas cannot be filled because of the low incidence of such structures at the experimental sites. For these, 100-percent sampling is recommended.