Summary

Almost every day, the Australian Bureau of Statistics (ABS) puts out information describing different aspects of social and economic conditions in Australia. Much of this information hits the news headlines, drives the money-markets, and acquires a particular interest in the middle of election campaigns!

The body of statistical information built by the ABS provides

• a basis upon which the government may develop and implement sensible policies;
• useful data for decision-making in the business sector;
• for the general community, it opens a window on the performance of governments, which is instrumental in maintaining our open system of government.

Mathematical statistics is a vital ingredient to the production of statistical information in the ABS, providing tools for efficient collection design and for analysis aimed at extracting patterns in the data, from which inferences can be drawn.

This paper describes some of the mathematical statistics underlying the design and analysis of the Bureau's largest household survey, the monthly Labour Force Survey. It shows how mathematics have been used to minimise costs of the survey, to test alternative operational strategies, to extract signal from noise in the analysis of the survey data, and in recent research to investigate the use of smoothed labour-force estimates at the regional level.