|
Abstract:
|
Business surveys are usually stratified by industry, geography and size. Estimation is done for domains of interest that may or may not correspond to the stratification. Auxiliary data can improve the reliability of estimated totals via calibration as follows: calibrating across a number of strata or calibrating to domains of interest. These methods have a number of advantages and disadvantages. This paper will compare these methods via an extensive numerical study illustrated using Statistics Canada's Survey of Employment, Payrolls and Hours, which provides a monthly portrait of earnings, the number of jobs and hours worked at different industrial and geographical levels.
|
