Investigating the Value of Appending New Types of Big Data to Address-Based Survey Frames and Samples

• Activity Number: 354 - SPEED: Big Data, Small Area Estimation, and Methodological Innovations Under Development, Part 2
• Type: Contributed
• Date/Time: Tuesday, July 30, 2019 : 10:30 AM to 11:15 AM
• Sponsor: Survey Research Methods Section
• Author(s): Paul John Lavrakas*
• Companies: Independent Consultant
• Keywords: Auxillary data; nonresponse bias

Abstract

Advances have occurred in survey science by using data that can be appended to frames. These include census block group data and non-public big data. The latter come from commercial sources and are not without error. Despite the errors in these household variables some are found to be reliable predictors of response. Although household data are not available for all addresses, the missingness of such variables can be a reliable predictor of response. We used a national probability sample of nearly 100K cases for our study (response rate of approx. 22%). To each sampled address we appended block level census data and some household characteristics. We report on the nature of the nonresponse bias in this survey using these characteristics. Then we appended a much larger set of nontraditional household and person level characteristics. These demographic and psychographic characteristics come from commercial “big data” sources. We then conducted a second nonresponse bias investigation and compared the results to our first investigation. We will show that as more big data can be appended to addresses, researchers are able to conduct more valuable nonresponse bias investigations.