Abstract:
|
The National Immunization Survey (NIS) uses an RDD telephone sample from all 100-banks of telephone numbers containing at least one directory-listed residential telephone number. A 100-bank is 100 consecutive telephone numbers (i.e., 301-458-4500 to 301-458-4599). Increasing the number of directory-listed residential numbers from one or more listings per 100-bank to two or more may produce cost savings in the form of fewer nonworking and nonresidential numbers encountered by interviewers (only 100-banks with more residential numbers would be sampled). The increase in efficiency may introduce bias in the vaccination coverage estimates, since low-density100-banks would be excluded from the sample frame.
Simulated NIS data are created for samples from 100-banks, with 2+, 3+, etc. directory-listed numbers using information on the number of listings per 100-banks.. We examine the extent of bias in vaccination coverage estimates and projected cost savings by comparing the simulated data with the current NIS data. Because the number of directory listings vary by geographic area, we compare vaccination coverage estimates at the state and selected urban-area levels.
|