The purpose of this dissertation is to identify ways to improve the effectiveness of the communication procedures used by the U.S. Bureau of the Census to elicit household self- responses to the American Community Survey (ACS). The ACS, which seeks responses from 3.5 million randomly sampled households each year, serves as the main source of information for comparing the characteristics of people across cities, states and regions of the United States. Although overall response rates to the ACS have not declined dramatically in recent years, as has been the case for many surveys, it now takes more effort at greater cost to get a sampled household to self-respond. In this dissertation, I note the shortcomings in the current process used by the Census Bureau to innovate the ACS methodology to gain self-response. I then review existing theories of human response behavior and utilize them to conduct a content analysis of five ACS mail communications. Based upon this analysis, and recommendations derived from relevant literatures, I proposed a revised sequence of communications for future testing against current communications efforts. It is hypothesized that use of this comprehensive redesign will improve the speed by which responses are obtained and lower data collection costs by reducing the amount of in-person follow-up interviews as well as increase participation from reluctant respondents, reducing non-response bias. These recommendations will apply directly to future ACS testing, but will also be applicable to government and non-government surveys that utilize a similar multi-mail-contact web-push methodology.