Compensating for Noncoverage of Nontelephone Households Using the National Health Interview Survey

• Activity Number: 405
• Type: Contributed
• Date/Time: Wednesday, August 10, 2005 : 10:30 AM to 12:20 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Karen Davis*+ and Meena Khare
• Companies: National Center for Health Statistics and National Center for Health Statistics
• Keywords: noncoverage ; propensity score model

Abstract

Telephone survey estimates are subject to bias from noncoverage of nontelephone households. To compensate for such bias, sampling weights typically are adjusted using ratio adjustments to match the demographics of the population, weight adjustments of households with interruptions in telephone service to account for nontelephone households, or weight adjustments of telephone households based on logistic regression estimates of the propensity for each household to have been without service. The interruption method assumes that characteristics of telephone households reporting interruptions in telephone service during the previous year are more similar to nontelephone households than to households with continuous service. The propensity score methods indirectly make a similar assumption, because the probability of being a nontelephone household is often predicated by whether the household had service interruptions. Using the 2001--2003 National Health Interview Surveys, we will use the interruption method and propensity score method to develop sampling weights for telephone households and show how adjustments based on interruption in phone service can reduce bias.