Calibration on partly known counts in frequency tables with application to real data

M. Sverchkov and Richard B. Tiller

Abstract

Deville and Särndall (1992, Section 4) considered calibration on the known counts (cell counts or marginal counts) of a frequency table in any number of dimensions (generalized raking procedure). In this paper, we show that a similar procedure can be applied to the case of partly known overlapping counts. As an example we consider calibration of areamonth- year unemployment estimates to month-year totals from a time series model of State estimates from the Current Population Survey and area-year totals from the American Community Survey.