Abstract, For legal and operational reasons, wireless phone numbers are,usually ,excluded ,from ,RDD surveys. Wireless-only households,are rapidly growing,and about 6-10% of the U.S. households,are wireless-only. Estimates from,RDD surveys are subject to potential bias due to noncoverage ,of households,without ,landline telephones. We use ,combined data from the 2003-2004 NHIS to compare
... [Show full abstract] characteristics of adults living in phoneless ,households ,with those living in wireless-only households,or households with interruptions in telephone,service. We selected ,a random ,sample ,of telephone,households ,and ,evaluated ,ratio- and ,propensity- based weighting,methods,to compensate,for noncoverage,of phoneless,and ,wireless-only households ,using information on interruptionsin landline telephone,service and presence,of wireless telephones. To assess bias, resulting estimates are compared,with the annual NHIS estimates. Keywords: Weighting methods, propensity score methods, ratio adjustments, RDD telephone survey 1.,Introduction: Random-digit-dialing (RDD) telephone surveys are the quickest way to collect data and investigate emerging public health issues. Estimates from RDD surveys are