Online Program

A Comparison of Hierarchical and Hidiroglou-Berthelot Macro-edits
*Keith Farwell, Australian Bureau of Statistics 


Keywords: macro-edits, Hidiroglou-Berthelot edit, selective macro-edits, hierarchical macro-editing, significance editing, expectations

Several statistical organisations have examined or implemented macro-edits based on the popular Hidiroglou-Berthelot edit. It is based on the analysis of ratios of estimates and has the advantage that it was designed to control the effects of swamping and masking when selecting anomalous ratios.

ABS has recently developed a suite of selective macro-edits based on a significance editing framework, called hierarchical macro-editing. The approach is based on comparisons of observed estimates with expectations of them. The significance of the deviations from expectations is measured in terms of the impact of the deviations on higher level estimates. However, the suite contains an option for selective macro-editing when no expected estimates are available. This option provides an alternative to the Hidiroglou-Berthelot macro-edit.

This paper will examine both kinds of macro-edits and will provide some results using data from the Australian Agriculture Survey. It will outline three variants of the Hidiroglou-Berthelot macro-edit which were developed as part of the study. It will look at the similarities and differences of both types of macro-edits.