The 2012 American Community Survey Main Phase Housing Unit Frame included ungeocoded records from the United States Postal Service Delivery Sequence File for 730 counties where ungeocoded records were previously excluded from the frame. These 730 counties contained at least 80 percent city-style addresses2. To assess this change we looked at the total interview rate and type C3 non-interview rate and how they changed by using these records.

In the 2,135 counties where we used ungeocoded Delivery Sequence File records in both 2011 and 2012 we found a difference of 0.15 percentage points for the total interview rate and a difference of 0.17 percentage points for the type C non-interview rate when including the ungeocoded Delivery Sequence File records. In the 730 counties where we started using ungeocoded Delivery Sequence File records in 2012, the difference in total interview rate was 0.18 percentage points and the difference in type C non-interview rate was 0.19 percentage points. All of these differences were within predefined acceptable limits.

Since all differences were within acceptable limits, we recommend that we continue to use ungeocoded Delivery Sequence File records in counties with at least 80 percent city-style addresses.