Assessment of Nonresponse Bias Through Interviewing Effort Analysis in a Dual-Frame RDD Telephone Survey

• Activity Number: 305
• Type: Contributed
• Date/Time: Tuesday, August 2, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Section on Survey Research Methods
• Author(s): Wei Zeng*+ and Ben Skalland and James Singleton
• Companies: NORC at the The University of Chicago and NORC at the The University of Chicago and Centers for Disease Control and Prevention
• Keywords: Cell Phone survey ; non-response bias ; weighting

Abstract

In RDD telephone surveys, direct approaches to assess non-response bias are typically not possible because little information is known about non-respondents. Information linked to landline telephone exchanges do not exist for cell phone samples. An indirect way to evaluate non-response bias in landline and cell phone surveys is a level of interviewing effort analysis, assuming that high-effort respondents may resemble non-respondents. For example, converted refusal cases can be compared to non-refusal cases and respondents can be compared by number of calls made. The National Immunization Survey (NIS)-a nationwide, list-assisted RDD survey, monitors the vaccination rates of children 19 through 35 months. A national cell phone sample for NIS was conducted, targeting households with age-eligible children that only or mainly used their cell phones. This paper uses level of effort analysis to assess the potential non-response bias in the estimates of vaccination coverage in the NIS cell sample, and compares the findings with similar analysis of the NIS landline sample. We will also evaluate whether appropriately weighting the sample can be effective in reducing the potential bias.