Coverage Improvement from Experimental Residence Questions
Anne Kearney, Roger Tourangeau, Gary Shapiro, Lawrence Ernst
RR 93/09, 02/28/94
ABSTRACT
One important source of coverage error in censuses and surveys is the lack of clarity of the roster
questions. These questions depend heavily on household informants deciding who "usually" lives
in the household. This paper reports on the results of an experiment in which one-third of the
sample was asked the usual residence questions, while two-thirds were asked an extensive set of
questions on who stayed any time in the last month, who else usually stays at the household, and
who else ate there during the last week. This paper discusses how many extra people were
picked up through these questions who would otherwise have been missed. The paper also
reports on the results of relaxing the requirement of giving full names.
