The Accuracy of Small Area Sampling of Wireless Telephone Numbers

The value of telephone surveys for assessing effects at geographic areas is impacted by increasing wireless telephone use. Highly accurate landline samples may be drawn for national, state, county, or even smaller areas; however, wireless samples have less geographic precision. This requires additional data collection effort and screening costs in order to ensure the appropriate geographic area is surveyed. In this paper, we examine the accuracy of wireless sample frame from the 2010-2011 National Flu Surveys. We illustrate differences and variations in wireless sampling accuracy for different geographic areas, focusing on variability by area in placement of wire centers related to residences. Our results suggest that the accuracy of wireless sampling may be dependent on differences in geographic areas with the accuracy of wireless sampling decreasing as the level of geographic aggregation gets more specific; landline accuracy remains relatively stable regardless of geographic specificity. To explain this phenomenon, we examine patterns of geographic dispersion of wireless telephone numbers related to telephone switch centers and geographic area. Based on the evidence from these surveys we present several options to estimate the geographic specificity of an area prior to sampling.