WEIGHTING, BENCHMARKS AND ESTIMATION PROCEDURES
Weighting is the process of adjusting results from a sample survey to infer results for the total in-scope population. To do this, a 'weight' is allocated to each sample unit; that is, a person or a household. The weight is a value which indicates how many population units are represented by the sample unit. Separate person and household weights were calculated, as only one adult and one child per household were enumerated. The steps used to derive person and household weights are described below.
Initial household weight
The first step of the weighting procedure was to assign an initial household weight to fully responding dwellings. The initial household weight was calculated as the inverse of the probability of the household's selection in the sample. For example, if the probability of a household being selected in the survey was 1 in 600, then the household would have an initial weight of 600 (that is, it represents 600 households).
The initial household weight was then adjusted as described below.
Initial person weights
After obtaining adjusted initial household weights, initial weights were assigned to fully-responding persons based on the sub-sampling scheme deployed within households. Initial person weights were calculated by multiplying the person's household weight by the probability of the person being selected. For persons 18 years and over, the household weight was multiplied by the number of adults aged 18 years and over in the household, and for persons 0-17 years old, the household weight was multiplied by the number of children, of selected age applicable to the survey, in the household.
In developing the survey weights, information available for responding and non-responding households was used to conduct quantitative investigations into explicit non-response adjustments. No explicit non-response adjustment was made to the weighting however, as the effect of the investigated non-response adjustments to the estimates was negligible.
Investigations were also made into non-response for particular voluntary components where lower levels of response were achieved.
- In 2011-12, 37.1% of respondents aged 5 years and over participated in the biomedical component. A biomedical weight has been produced in order to adjust for the significant non-response in this voluntary component and to derive the required population estimates. Further analysis of fasting status however indicated that an adjustment or an additional weight was not required in order to produce representative fasting data.
- In 2011-12, 52.8% of respondents aged 5 years and over participated in the pedometer component, and 49.0% met the adequate response threshold for use in reporting (see Pedometer Steps chapter). Therefore, pedometer steps data presented as part of the NNPAS relates to the participating population only. Generally sex did not appear as a factor in non-participation, and there was a small increase in participation by age. Significance analysis indicated there was some difference between participants within subgroups (state, age groups, sex and Australia) used in the analysis in association with risk factors (including smoking, physical activity self-assessed health). However, calibration results showed that the pedometer status did not introduce bias into estimates, and the subsample with pedometer data is representative of the population.
- In 2011-12, 82.8% of respondents aged 2 years and over had their height and weight measured. Therefore, BMI data presented as part of the AHS relates to the measured population only. Analysis of the characteristics of people who agreed to be measured compared to those who declined indicated that age and sex were factors in non-response. Females were more likely to decline, and non-response increased with age.
- In 2011-12, 82.3% of respondents aged 5 years and over agreed to have their blood pressure measured. 81.4% were able to get at least one reading and 81.0% had a valid blood pressure reading obtained according to the systolic and diastolic calculation criteria used (see Blood Pressure chapter). Blood pressure data presented as part of the AHS relates to the valid measurement population only. Analysis of the characteristics of people who agreed to be measured compared to those who declined indicated that age and sex were factors in non-response. Females were more likely to decline, and non-response was most prevalent in children and then tended to increase with age.
Person and household weights are calibrated to independent estimates of the population of interest, referred to as 'benchmarks'. Weights calibrated against population benchmarks ensure that the survey estimates conform to independently estimated distributions of the population rather than to the distribution within the sample itself. Calibration to benchmarks helps to compensate for over- or under-enumeration of particular categories of persons and households, which may occur due to the random nature of sampling or non-response.
All person weights (including biomedical) were benchmarked to the estimated resident population living in private dwellings in non-Very Remote areas of Australia at 31 October 2011, based on the 2006 Census of Population and Housing. Excluded from these benchmarks were persons living in discrete Aboriginal and Torres Strait Islander communities, as well as a small number of persons living within Collection Districts that include Aboriginal and Torres Strait Islander communities. The benchmarks and hence the estimates from the survey do not (and are not intended to) match estimates of the total Australian estimated resident population (which include persons living in Very Remote areas of Australia and persons in non-private dwellings) obtained from other sources.
Calibration to household level benchmarks
The household benchmarks used in the weighting of the 2011-12 NHS, NNPAS and combined NHS/NNPAS survey files were preliminary household estimates for 31 October 2011, based on the 2006 Census of Population and Housing, scoped to the surveys.
The calibration levels used for benchmarking were State by Part of State by household composition (numbers of persons 0-14 years old, numbers of persons 15 years and over).
Calibration to person level benchmarks
Two types of person weights were produced for the 2011-12 NHS, NNPAS and combined NHS/NNPAS survey files:
- a person weight was applied to all responding persons in the surveys
- a biomedical person weight was applied to all selected persons who provided biomedical samples.
The person benchmarks used in the person and biomedical weighting of the 2011-12 NHS, NNPAS and combined NHS/NNPAS survey files were preliminary population estimates for 31 October 2011 based on the 2006 Census of Population and Housing, scoped to the surveys as appropriate.
The calibration levels used for benchmarking were State by Part of State by sex by age group. For NNPAS person weights, a seasonal adjustment was also incorporated.
The age groups (years) available for use in calibration were: 0-1 years, 2 years, 3 years, 4 years, 5-9 years, 10-11 years, 12-14 years, 15-17 years, 18-24 years, 25-29 years, 30-34 years, ... 80-84 years, 85 years and over.