|
Abstract:
|
The U.S. Census Bureau is researching the use of alternative data sources to assist in the editing and imputation procedures used in the American Community Survey (ACS). The expectation is that using already available data to fill-in missing values in lieu of statistical procedures, may improve data quality. The first phase of this research was presented at the 2019 Joint Statistical Meetings, and showed that alternative data are available to replace missing values for the demographic characteristics of age, sex, race, Hispanic Origin, and place of birth. The research found that using alternative data sources improved item allocation rates. However, when matching the values from alternative sources to the values allocated by the ACS editing process, there was misalignment for some of the demographic items, particularly the age item. This research builds off the research presented last year, by exploring deeper into this misalignment. Additionally, it incorporates the next phase of the research, which analyzes the impact on the data after the simulated dataset with both 2016 ACS response data and alternative data (in lieu of missing values) are run through the ACS edits.
|
