Tailoring Outreach to Boost Mail Self-Response in Geographic Areas with Similar Low Response Scores

• Activity Number: 593
• Type: Topic Contributed
• Date/Time: Wednesday, August 12, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Government Statistics Section
• Author(s): Darryl Creel*
• Keywords: response rates ; low response scores ; tailored outreach ; statistical learning ; reproducible research

Abstract

Each percentage point increase in the mail response rate to the Decennial Census saves the federal government about $85 million. The Census Bureau's 2014 Planning Database contains low response scores (LRS) by geographic area. These are summary scores that reflect an area's predicted mail response rate. The populations in geographic areas with similar LRS can have different characteristics such as the proportions of non-English speakers, proportion of transient population, or the number of 18-24 year olds. These characteristic differences may require very different outreach techniques in order to boost response from mail. Using the 2014 Planning Database, this research will investigate effective methods for incorporating the LRS into a screening process to identify geographic areas for further review. We will examine how the low response score can be used in a supervised learning context to explicitly identify the most salient characteristics for describing the geographic areas, and suggest tailored outreach programs based on a particular area's characteristics. This research will be conducted with an emphasis on answering the question: Why? In so doing, it will use the reproducibl