Traditional census design allows for late additions to the address list. Although the primary address list building activities occur prior to the enumeration stage, addresses may be added later from the United States Postal Service files, local government updates for new construction, the Local Update of Census Addresses Appeals program, and other operations designed to improve coverage. During the 2010 Census, these types of addresses were added to the list while census enumerations were already occurring in the spring of 2010. Under the original plan for the 2010 Census, which was similar to the Census 2000 design, these late additions would not have received a census questionnaire. Instead, their first chance for a census-initiated enumeration would have been during the Vacant Delete Check operation in July 2010.

In January 2010, the United States Census Bureau Director expressed concerns regarding the workload for the Vacant Delete Check operation. At the time, the Local Update of Census Addresses Appeals universe was approaching 1.8 million cases, a larger than expected workload. The Director was concerned about our plan to enumerate these cases so late in the process and potential respondent recall bias given the enumeration would occur three months after Census Day (April 1, 2010). During that time, the Census Bureau was also heavily promoting mail response through the March to the Mailbox campaign. This caused additional concerns regarding public reactions to the campaign for those respondents who had not yet received a questionnaire to return.

To address these concerns, the Director tasked the Decennial Management Division with developing a plan that would allow for an earlier enumeration of these cases slated for the Vacant Delete Check operation. The Decennial Management Division organized an interdivisional team to explore the options and make a recommendation. On January 29, 2010, with the Director’s approval, the team proceeded to implement the 2010 Census Late Adds Mailout operation. The operation would allow a subset of the “late adds” addresses from the Spring 2010 Delivery Sequence File, the Ungeocoded Resolution operation and the Local Update of Census Addresses Appeals operation the opportunity to receive a questionnaire return in the mail.

Final census results indicate that 1,351,091 housing units from the Late Adds Mailout universe were included in the final 2010 Census counts. This represents 66.1 percent of the Late Adds Mailout universe. Inclusion could have resulted from the mailback of a questionnaire or through enumeration in the Vacant Delete Check operation. Results for the “Spring 2010 Delivery Sequence File” and “Ungeocoded Resolution” housing units were higher than the overall, with 83.7 percent of housing units and 90.5 percent of housing units in the final census respectively.

The results for “Local Update of Census Addresses Appeals” housing units were much lower than the other two sources at 54.5 percent. Through all stages, the appealed housing units appeared to have less success than the addresses from the other two sources. Presumably high undeliverable as addressed rates affected the overall response rates and led to the inclusion of most of these addresses in the Vacant Delete Check operation which ultimately resulted in low in-census success rates.

Overall, the Late Adds Mailout operation met with some successes. Namely:

• Census Bureau staff were able to successfully work together to plan and implement a mailout operation for over two million addresses during the height of 2010 Census activities.
• The operation reduced the Vacant Delete Check workload by more than half for two of the “late adds” sources, the Spring 2010 Delivery Sequence file and the Ungeocoded Resolution. Had these sources represented more of the Late Adds universe, results could have been significant.
• Approximately $2.5 million was spent for the Late Adds Mailout operation and it generated approximately $20.6 million in savings.