In 2011, the U.S. Census Bureau designed two experimental tests offering an Internet reporting option in addition to the traditional paper form for the American Community Survey (ACS). An Internet reporting option was offered as a way to maintain or improve self-response, while reducing costs. As part of the first test, a qualitative follow-up interview of both nonrespondents and respondents was conducted to determine what motivated a response and why some did not respond at all.

In this paper, we examine the groups of individuals who have been typically hard-to-interview in the self-response phase of the ACS. Joshipura (2008) found that renters, those with less than a high school education, non-White, and Hispanic respondents are more likely to need an interviewer-assisted interview to complete the ACS. We examine the data from the two tests to determine if those types of respondents were more likely to report using the Internet compared to mail-return response rates. We also examine the qualitative data from the follow-up study for these people to determine why they did not respond.