METHOD OF ESTIMATION
LFS estimates of the number of people employed, unemployed and not in the labour force are calculated in such a way these totals add to independently estimated counts (benchmarks) of the usually-resident civilian population aged 15 years and over. These benchmarks are based on Census of Population and Housing data and are adjusted for scope differences and under-enumeration, and updated for births, deaths, interstate migration and net overseas migration.
Two sets of benchmarks are used in the LFS. The first set of benchmarks classify the population by state or territory of usual residence, part of state of usual residence (capital city, rest of state), age group and sex. The second set classify by region of usual residence and sex (known as 'regional benchmarks'). The regions published in the labour force estimates until January 2014 are based on aggregations of Australian Standard Geographical Classification (ASGC) (cat. no. 1216.0) 2006 regions but will be replaced with Australian Statistical Geography Standard (ASGS) (cat. nos. 1270.0.55.001 to 1270.0.55.006) Statistical Area Level 4s (SA4s) from January 2014. This is discussed in more detail under the Regional Estimates section of this information paper.
To derive labour force estimates for the entire population in the scope of the survey, expansion factors (weights) are applied to the sample responses. The weighting method ensures that LFS estimates conform to the benchmark distribution of the population by age, sex and geographic area. This reduces sampling variability and compensates for any under-enumeration or non-response in the survey. It does not overcome any bias arising from non-response.
The LFS estimation method is a form of composite estimation and exploits the overlapping design of the LFS sample. It does this by combining the previous six months' sample responses with the current month's responses to produce the current month's estimates.
Initially, composite estimates for 11 sets of key LFS estimates are derived using the sample responses for the current month and previous six months.
Weighting factors (or multipliers) are applied to the seven months' sample responses. They determine the extent to which the responses over the seven month 'window' contribute to the current month's LFS estimates. They are based on the correlation structure observed in historical LFS data. While taking account of the multipliers, the weights of the sample responses are adjusted to align with current month population benchmarks. As a result, the weight assigned to a sample response is dependent on the geographic area, age and sex of the respondent, the month in which the response was collected and the number of months the rotation group has been in the sample. The set of composite estimates are then produced from the seven-months weighted dataset.
Finally, the current month's sample responses are weighted to both the population benchmarks and to the set of composite estimates produced from the seven-months' weighted dataset. The current month's estimates are produced from this weighted dataset, where the estimates for each characteristic of interest are obtained by summing the weights of the persons in the sample with that characteristic.
Further information about the LFS estimation method can be found in the information paper Forthcoming Changes to Labour Force Statistics, 2007 (cat. no. 6292.0) which is available on the ABS web site <http://www.abs.gov.au>.
RELIABILITY OF ESTIMATES
The accuracy of a sampling estimate refers to how close that estimate is to the true population value. The variation between the two is referred to as 'the error of the sampling estimate'. The total error of the sampling estimate results from two types of error:
- sampling error, which occurs because data were obtained from a sample rather than the entire population; and
- non-sampling error, which arises from imperfections in reporting, recording or processing of the data that can occur in any survey or census.
One measure of sampling error is given by the standard error of the estimate, which indicates the extent to which that estimate might have varied by chance because only a sample of dwellings was surveyed. There are about two chances in three that the estimate that would have been obtained if all dwellings had been included will differ by less than one standard error from a sample estimate, and about 19 chances in 20 that the difference will be less than two standard errors.
Expressing the standard error of an estimate as a percentage of the estimate to which it relates offers another useful measure of sampling variability. This is known as the relative standard error (RSE).
Standard error estimates published in association with any LFS results are mathematically modelled after each sample redesign using estimates on a range of sub-populations (e.g. broken down by sex, age group, geography, etc.) from 12 months of survey responses.
For further information, refer to the Impact on Standard Errors
section of this information paper.
Changes to the LFS, including the introduction of self-enumerated electronic online collection methods at the same time as the 2011 sample may result in some non-sampling errors, however the ABS is using its experience developed in introducing previous changes to the LFS and other statistical collections to minimise any potential impact. In addition, the ABS will monitor any impact of the move to online collection through a measurement strategy. Refer to the article, Upcoming Changes to the Labour Force Survey
, in Labour Force, Australia, July 2012
(cat. no. 6202.0), for further information on the upcoming changes. For details of the LFS electronic collection refer to the Transition to Online Collection of the Labour Force Survey
article in Labour Force, Australia, Apr 2013
(cat. no. 6202.0). Any identified non-sampling errors in a given month will be included in the Explanatory Notes section of the monthly Labour Force, Australia
publications (cat. no. 6202.0).
More information is available in the Article Archive
sections of Labour Force, Australia
(cat. no. 6202.0).