Weighting Methods for the 2010 Data Collection Cycle of the Medical Monitoring Project

• Activity Number: 87
• Type: Contributed
• Date/Time: Sunday, August 4, 2013 : 4:00 PM to 5:50 PM
• Sponsor: Survey Research Methods Section
• Author(s): Lee Harding*+ and Ronaldo Iachan and Christopher Johnson and Tonja Kyle and Jacek Skarbinski
• Companies: ICF and ICF International and CDC/NCHHSTP and ICF and CDC
• Keywords: population-based surveys ; health surveys ; multistage sampling ; non-response adjustment ; weighting

Abstract

This paper describes the weighting procedures developed for the Medical Monitoring Project (MMP), a nationally representative surveillance system for HIV-infected persons receiving medical care in the U.S. Unlike traditional surveys, MMP collects data on sampled persons from three different data sources: face-to-face interviews, medical record abstractions (MRA), and an extract of data reported to the National HIV Surveillance System. In the 2009 data collection cycle all three datasets were weighted independently for differential probabilities of selection and non-response and adjusted for multiple opportunities to enter the sampling frame(s). In 2010 we improved our weighting process by using a unified classification process for eligibility that uses information from all three datasets and parallel weight adjustments for the three sets of weights. We also used a single estimated population total to harmonize weight sums across the three sources. In addition to describing weighting methods, we describe design variables (strata and clusters) defined for variance estimation applicable to all datasets as well as to an analytic dataset of matched interview and MRA records.