Abstract

This study attempted to generate SE estimates for some indicators for the Annual Survey of Establishments using STATA, CENVAR and EXCEL. SE estimates for employment variables is lowest among other variables. This supports the appropriateness of the sampling design used in the current annual survey of which employment size is the stratification variable. Generally, computed measures of design effect (deff) further confirm that the ASE sampling design using stratified simple random sampling is appropriate in terms of (1) using employment as stratification variable compared to other variables and (2) sampling selection within stratum is done using simple random sampling. For hotels and restaurants, the deff for revenue is lower compared to other variables, i.e. employment. The figures imply that for this sector, stratification by revenue, instead of employment, may be a better alternative.