An Empirical Evaluation of Two Methods for Non-Telephone Bias Adjustment in RDD Surveys

• Activity Number: 96
• Type: Topic Contributed
• Date/Time: Monday, August 12, 2002 : 10:30 AM to 12:20 PM
• Sponsor: Section on Survey Research Methods*
• Author(s): Robert Wright*+ and Martin Frankel+ and Meena Khare and Michael Battaglia and K. Srinath and Wenxing Zha
• Affiliation(s): National Center for Health Statistics and Abt Associates, Inc. and National Center for Health Statistics and Abt Associates, Inc. and Abt Associates, Inc. and National Center for Health Statistics
• Keywords: Random Digit Dialing ; Telephone Interruption ; Non-Telephone Bias ; Propensity Weighting ; National Immunization Survey

Abstract

This paper evaluates the performance of two recently proposed methods for the reduction of non-telephone bias in RDD surveys. The first method was suggested by Keeter (1995) and is based on information collected from respondents in the RDD survey about "interruptions" in household telephone service during the past year. The second method makes use of a logistic regression model developed from an ancillary survey, which examines characteristics that are related to the presence of a telephone in the household.

The National Health Interview Survey (NHIS), conducted by NCHS, CDC, was used for the evaluation of these two methods. The NHIS is based on an area probability sample and includes both households with and without telephone service. The survey also includes questions about possible interruptions in telephone service during the past year. The 1997 and 1998 NHIS surveys were used to build the logistic regression model. The 1999 survey was then used to evaluate the two adjustments in terms of bias reduction and overall mean squared error.