Abstract:
|
The U.S. Department of Agriculture’s (USDA) National Agricultural Statistics Service’s (NASS’s) June Area Survey (JAS) is an annual survey based on an area frame, which has complete coverage of the contiguous U.S. Data for this survey are collected via in-person interviews. NASS employs manual imputation for JAS nonresponse, which is becoming increasingly costly as response rates are declining. Moreover, it can be difficult to measure the data quality resulting from these efforts. We are proposing a new automatic imputation approach that uses a unique combination of data sources, including historic satellite imagery, digital geospatial archive of the sampled areas of interest, and administrative data. This paper evaluates the quality of the proposed automatic imputation approach.
|