Data Quality in the Retrospective Reporting of Addresses
Bachtell, Kate and Tangel, Virginia and Latterner, Michael and English, Ned and Haggerty, Catherine (2010) Data Quality in the Retrospective Reporting of Addresses. In: 65th Annual AAPOR Conference, May 2010, Chicago, IL.
MetadataShow full item record
While tracking the movement of respondents has always been crucial for panel studies, the increasing popularity of geographic analyses has furthered the demand for both accurate and systematic address collection. This paper advances the existing literature on retrospective reporting in surveys by focusing on the collection of respondents‘ past addresses. It features data from the third wave of Making Connections, a ten-year, neighborhood-based survey funded by the Annie E. Casey Foundation. The wave three questionnaire featured a new series that solicited a detailed history of respondents‘ movement during the previous three years. Recovering previous addresses presented challenges beyond those typically associated with the systematic recording of physical addresses because recalling information from the past is inevitably more difficult than describing the present (Kennickell and Starr-McCluer 1997). We designed an experiment to test different methods of maximizing data quality in the retrospective address series collected in 2009 in White Center (Seattle), Washington. Households were randomly assigned to two treatment groups. Addresses collected from the first group underwent administrative data cleaning (using Google Maps, etc.) while those in the second group received intensive follow-up calls by a team of field interviewers and managers. We compare the results of these treatments to the original data collected in White Center and investigate the efficacy of each method for producing addresses that can be successfully translated into geographic coordinates for spatial analyses. We find that the retrieval effort – while more costly to execute – was far more successful in returning ‗geocodable‘ addresses. This supports the argument that successful collection of retrospective addresses depends on an interactive process between the interviewer and respondent involving a variety of probing techniques. Our findings may inform improved methodologies for collecting retrospective data in both panel and cross-sectional surveys.