The goal of this research was to identify a set of possible stopping rules to use in the American Community Survey (ACS) Computer Assisted Personal Interviewing (CAPI) operation to reduce the burden that sample households may feel due to the level of effort that we currently make to obtain high response rates. We analyzed data from the 2012 ACS to understand:

• how frequently we contact sample households,
• the proportion of households that express reluctance and how we address that reluctance,
• the cumulative burden that sample households might feel due to the sequential mixed mode design of the ACS,
• the quality of CAPI data collection in terms of response rates, item - level completeness of responses, and total number of completed sample interviews, and
• the number of hours spent during CAPI to convert noninterviews.