The goal of this research was to identify certain classes of records on the Master Address File to exclude from the American Community Survey housing unit frame in the future. I focused primarily on a certain type of record on the Master Address File termed "Excluded from Delivery Statistic" by the United States Postal Service. The suspicion being the reason the Postal Service excluded these records from their delivery statistic would also be reason for the American Community Survey to exclude the same records from the housing unit frame. Prior research had shown that removing "Excluded from Delivery Statistic" records also removed over-coverage from the housing unit frame. We use 2009 Address Canvassing results as a way to gauge whether any record stays on the housing unit frame.
We find that while certain classes of "Excluded from Delivery Statistic" records should be excluded from the housing unit frame, the risk of under-coverage for identifying those records in the next decade is too high from using just information available on the Master Address File.