Collapsing Weighting Adjustments in the National Immunization Survey (NIS) - Is There an Advantage?

• Activity Number: 586 - Weighting and Variance Estimation
• Type: Contributed
• Date/Time: Wednesday, August 2, 2017 : 2:00 PM to 3:50 PM
• Sponsor: Survey Research Methods Section
• Author(s): Elizabeth Ormson* and Philip J. Smith and Zhen Zhao and Qiao Ma
• Companies: NORC at the University of Chicago and Centers for Disease Control and Prevention and CDC and NORC at the University of Chicago
• Keywords: weighting ; estimation ; NIS ; telephone ; survey ; variance

Abstract

The National Immunization Survey (NIS) tracks state and national vaccination coverage through a dual-frame landline and cell telephone Random Digit Dial (RDD) sample design. Demographic data are collected during telephone interviews, with a mail survey sent to providers to collect vaccination histories when consent is obtained. We investigate a method that collapses three of the initial NIS weighting adjustments associated with the CASRO response rate (nonresolution of phone numbers, screener nonresponse, household interview nonresponse) into a single step using adjustment cells, while keeping the remainder of the weighting process intact. By collapsing three adjustments into one, the sample weights were expected to be somewhat less variable and yield smaller variances for NIS vaccination coverage estimates. This method was evaluated using the 2015 NIS data and found little impact on variances or their estimates, with larger effects on intermediate weighting steps. These results raise the question whether there is an advantage in making separate adjustments for all CASRO factors involved in constructing NIS weights.