1504.0 - Methodological News, Jun 2012  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 12/07/2012   
   Page tools: Print Print Page Print all pages in this productPrint All

Investigating a List-based Sampling Approach for ABS Household Surveys

A major initiative underway at the ABS is the construction of an Address Register (AR) to increase our efficiency in producing Census and other population statistics, and to broaden our range of analytical and statistical products.

One particular area of gain is the potential for smarter and more flexible household survey sample designs. Currently, ABS household survey sample designs are constrained by the need to select geographic areas as the first stage of selection. Recently, Household Survey Methodology (HSM) conducted an investigation to explore the types of sample design efficiencies possible for household surveys under a list-based sampling framework. Three main areas of potential gain were investigated:
1. De-clustering our household samples
2. Using auxiliary information to stratify a sample by specific population characteristics, along with an optimal allocation to these strata
3. Using auxiliary information for targeting specific subpopulations of interest

The results from the HSM investigation show that a list-based sampling framework for household surveys is worth further investment, and would have major benefits for sample efficiencies if a list-frame was augmented by auxiliary information that could be used in design and estimation. While previously the use of auxiliary information has been constrained to the areal level, an AR enables us to also utilise address or person-level auxiliary data for design and estimation. Specific findings include:
· Reductions in sample size may be realised through de-clustering of our samples, however these need to be coupled with a modal change of enumeration (such as increased use of Telephone Interviewing or E-forms) to realise significant monetary gains.
· While reductions in sample size may be realised through de-clustering our sample, these reductions need to be balanced against increased cost of enumeration (or impacts to follow-up processes and contact rates) due to a more spread-out sample.
· There are potential benefits afforded by a list frame which has been augmented by auxiliary data that can be used in design (through stratification and allocation) and estimation.
· A major advantage to having auxiliary information available is that it facilitates the targeting of subpopulations of interest which typically have needed to be found through expensive techniques such as oversampling or screening. This leads to the potential of surveys reassessing their output requirement needs, rather than having outputs constrained by what could be collected under a cost-effective area based sample.
· The quality and type of auxiliary information attached to a list frame influences the magnitude of efficiency gains possible. For example, using auxiliary information that is highly correlated with variables of interest will provide greater gains. Similarly, inaccuracies in the auxiliary data (for example, by having out-of-date data) will reduce the gains possible.

It is believed that an AR can help the ABS Household Survey Program achieve ABS goals of reduced costs (and time) and growing the business through new statistical products and services. However, the implementation of the AR needs to be considered holistically along with other initiatives in the wider household program. For example, moves are afoot to introduce an internet mode of collection for the Monthly Population Survey, as well as the consideration of moving surveys to a more integrated approach (such as an omnibus survey vehicle). Other current areas of investigation across the wider household survey program could include an evaluation of a greater use of administrative data (across both the design and analysis phases).


Further Information
For more information, please contact Rebecca Farrow (07 3222 6467, rebecca.farrow@abs.gov.au) or Lyndon Ang (02 6252 5279, lyndon.ang@abs.gov.au)