A total of 131,704,730 housing units were enumerated in the 2010 Census. Of those in mailout/mailback areas approximately one-third (33.5 percent) did not respond to the initial or replacement mailings (Letourneau 2011). This left approximately 47 million housing units in the Nonresponse Followup workload (Jackson et. al. 2011). These housing units required an enumerator to collect the Census information (using personal visits or phone calls) as part of the nonresponse followup operation.

The objective of this experiment was to understand the effects of reducing the maximum number of Nonresponse Followup contacts in a census environment from the usual six, to either five or four, which could provide large cost savings through a reduction in field work. This will be answered with the following research question: Can a reduction in the number of contact attempts still maintain the same level of data quality in order to save costs in Nonresponse Followup? By conducting the experiment in the 2010 Census, the results are more predictive of what to expect in the 2020 Census with regard to the feasibility of reducing the maximum number of Nonresponse Followup contact attempts.

The production enumerator questionnaire consisted of questions designed to procure the same data as the 2010 Census mailout/mailback questionnaires, as well as a “record of contact” section to document information about each contact attempt. In particular, the front page of the questionnaire had space to provide data on up to six contact attempts including mode (personal visit or telephone), date, time, and outcome. Two experimental questionnaires were modified to provide data for up to either four or five contacts, respectively. All content questions were the same as those found on the standard enumerator questionnaires.