Comparative Effectiveness of Administrative Data and County-level Aggregates in Nonresponse Adjustments for Surveys of Low-income Populations

• Activity Number: 162
• Type: Contributed
• Date/Time: Monday, August 9, 2004 : 2:00 PM to 3:50 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Frank Potter*+ and Mourad Touzani and Ronghua Lu
• Companies: Mathematica Policy Research, Inc. and Mathematica Policy Research, Inc. and Mathematica Policy Research, Inc.
• Keywords: nonresponse adjustments ; response propensity ; low-income populations

Abstract

In the evaluation of social programs, low-income populations are surveyed to study the program effectiveness. In these surveys, limited data are available from administrative data to use in the nonresponse adjustments (generally, age, gender, and sometimes race of the program participant). The data can be supplemented by using aggregated data at the county level or at the zip code level (e.g., Area Resource File or Census Bureau data). These data provide contextual information on where the program participant resides, but are not directly related to the individual's propensity to respond. In addition, using these aggregated data can add to the costs of the nonresponse adjustment activities. Our paper will investigate the benefits of using the additional aggregated data in terms of response variation explained by the response propensity models when the aggregated data are used relative to models using only the available administrative data. We will also evaluate measures of model fit for applicability to similar nonresponse logistic modeling efforts. Our study is based on more than 50 nonresponse models developed for a 10-state evaluation of a social program targeting children.