How Can We Produce Estimates When We Can't Call You? Revisiting Methods to Adjust for the Phoneless Population

• Activity Number: 35
• Type: Contributed
• Date/Time: Sunday, August 9, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Survey Research Methods Section
• Author(s): Meena Khare* and Nadarajasundaram Ganesh and Kennon Copeland and Wei Zeng and Xian Tao and James A. Singleton
• Companies: National Center for Health Statistics and NORC at the University of Chicago and NORC at the University of Chicago and NORC at the University of Chicago and NORC at the University of Chicago and NCIRD/CDC
• Keywords: noncoverage ; RDD telephone survey ; weighting ; propensity score

Abstract

Prior to the explosion in the cell-phone only (CPO) population, Keeter's method was used to adjust for noncoverage in landline RDD telephone surveys. Approximately 98% of adults and children live in a household with access to either cell phone or landline telephone service (NHIS, 2014) and more than 40% of households are CPO. Only a small proportion of the U.S. population (~4%) lives in a landline only telephone household and phoneless population remained smaller being less than 3%. Use of Keeter's method has been complicated by the reduction in the size of the landline sample in dual frame surveys and the fact that interruption in telephone service does not apply to cell phone sample cases. Furthermore, the sample of landline cases with interruption in landline services has decreased to such minimal size that it cannot represent the phoneless population. We use data from 2012 NIS, and 2012-2013 NHIS to explore the potential bias resulting from noncoverage of the phoneless population, discuss alternate methods to compensate for noncoverage in dual-frame telephone surveys, and present results relative to bias reduction and impact on the variance of survey estimates.