Keith Farwell, Robert Poole, Stephen Carlton
Australian Bureau of Statistics
GPO Box 66A
Hobart 7001
Australia
Email: keith.farwell@abs.gov.au
The term "significance editing" refers to a general editing approach which incorporates survey weights and estimation methodology into edits and maintains a link between individual responses and output estimates. The Australian Bureau of Statistics (ABS) has been using significance editing in varying degrees over the last decade. When significance editing is applied at the input stage of a collection it is called 'input significance editing' within the ABS. Rather than using edits that result in messages such as "we should check that value", input significance editing uses a score to prioritise editing effort. The larger the score, the more significant the respondent reporting error is to possible bias in estimates. Respondents can be ordered or ranked in order of decreasing score giving a prioritised list of units to edit. This method was first applied within the ABS by super-imposing it on the edit failures of an existing input editing system. Scores were generated for the edit failures and used to target editing effort towards a subset consisting of those most likely to provide the largest reduction of reporting bias on estimates. This approach has resulted in a noticeable improvement in editing efficiency.
This paper will outline further progress on input significance editing. It will outline progress on a technical framework for input significance editing which uses a model to describe the effect of editing unit record data. It looks at the joint problems of minimising reporting bias for a fixed cost and minimising cost to achieve a fixed level of reporting bias and shows that, in the case of uniform cost per unit, both problems have analytical solutions.
This paper will also outline the results of an empirical study set up to verify the applicability of the above framework to the editing of actual data from the Australian Survey of Employment and Earnings. The work investigates some of the practical aspects of estimating the model.