Nonsubstantive Change Request (Quality Control Interviews)

Nonsubstantive Addition to OMB Submission for 2013 Census Test

The original submission did not contain a reference to quality control measures that might be employed in the test. We propose to do some checking of completed cases for falsification. The checking involves analysis of paradata and, only if necessary, direct contacts with sample units to confirm test participation. The analysis of paradata identifies completed cases in which interviews are notably short. An average time and standard deviation for all interviews is calculated in order to identify outliers. Audit trail files for these cases are examined to see if the data support a conclusion that an interview occurred. If prima facie evidence suggests that an interview might have been falsified, the sample unit will be contacted to verify the interview.

The interview takes about 6 minutes on average. In an extreme circumstance where we need to contact a large number of cases (e.g. 100) to verify participation, the total burden hours added to the project would be ten. This estimate is the upper bound for possible burden hours increase. The much more likely circumstance, because paradata analysis will clear up questions, is that very few or no direct contacts with sample units will be necessary. In this case, there would be no increase in burden hours or the increase would be less than one hour.