Abstract:
|
The literature has no standard method for estimating the coverage of address-based frames, especially for subnational areas such as segments in area probability designs. Versatility is desirable for different study needs, but standardization improves comparability and informs frame supplementation decisions. Many estimates are simple ratios of counts of addresses in the frame to control totals; others are model-based predictions. Limitations to these methods include geocoding error, outdated control totals, errors in the address frame, model lack of fit, and systematic exclusion of types of addresses. We tested various estimates on segments selected for the 2015 Residential Energy Consumption Survey. "Truth" was based on field enumeration or a field-based frame supplementation procedure. We matched field and frame addresses to determine which housing units were covered. We investigated relative costs and presumed accuracy of control totals. Here we share the results of our investigation and provide a public resource with net coverage estimates for counties.
|