|Page tools: Print Page Print All|
This document was added or updated on 06/11/2015.
Systolic Blood Pressure is located both in the Person level folder...
Any special codes for continuous data items are listed in the Data Item List.
Perturbation of data
To minimise the risk of identifying individuals in aggregate statistics, a technique is used to randomly adjust cell values. This technique is called perturbation. Perturbation involves small random adjustments of the statistics and is considered the most satisfactory technique for avoiding the release of identifiable statistics while maximising the range of information that can be released. These adjustments have a negligible impact on the underlying pattern of the statistics.
The introduction of these random adjustments result in tables not adding up. While some datasets apply a technique called additivity to give internally consistent results, additivity has not been implemented on this TableBuilder. As a result, randomly adjusted individual cells will be consistent across tables, but the totals in any table will not be the sum of the individual cell values. The size of the difference between summed cells and the relevant total will generally be very small.
Please be aware that the effects of perturbing the data may result in components being larger than their totals. This includes determining proportions.
Some tables generated within TableBuilder may contain a substantial proportion of very low counts within cells (excluding cells that have counts of zero). When this occurs, all values within the table are suppressed in order to preserve confidentiality. The following error message below is displayed at the bottom of the table when table suppression has occurred.
ERROR: The table has been suppressed as it is too sparse
ERROR: table cell values have been suppressed
Field exclusion rules
Certain groups of similar variables are restricted from being used together in a table. These restrictions are referred to as field exclusion rules, and are in place in order to protect confidentiality. The collection of similar variables restricted in this way are called field exclusion groups.
For the Australian Health Survey, there is one field exclusion group. This consists of the 2006 and 2011 geographical and Socio-Economic Indexes for Areas (SEIFA) data items (see below for items).
2011 Geographic Items
2006 SEIFA Items
2011 SEIFA Items
There are three weight variables visible on the TableBuilder file under Summation Options categories:
The NNPAS is a sample survey. To produce estimates for the in-scope population you must use weight fields in your tables. If you do not select a weight field, TableBuilder will use 'Persons (Benchmarked weight)' by default. This will give you estimates of the number of persons. To produce estimates of the number of households, you would have to change the weight field to 'Households (Benchmarked weight)' by adding it to your table from the Household level under Summation Options.
The Household Weight was benchmarked to the Household Level while the Person Weight was benchmarked to the Person level. To produce estimates for NNPAS persons who participated in the National Health Measures Survey (NHMS), the 'Biomedical persons (Benchmarked weight)' located on the Biomedical level must be used. When using a Weight/Summation from a level that is different to that of the variables in the table, please be careful in interpreting the results.
Note that the Biomedical level contains non-biomedical participant records, however their biomedical weight is set to 0 so they will not contribute to estimates when the Biomedical persons (Benchmarked weight) is used. However, if the Persons (Benchmarked weight) is used with biomedical data items, then these non-participants will contribute to estimates. When using biomedical variables in conjunction with other variables on the Biomedical level or with variables from other levels, the Biomedical persons (Benchmarked weight) should be used.
For example, a table of reported 'Month of biomedical collection' using the 'Persons (Benchmarked weight)' will show the 'Month of biomedical collection' for the entire National Nutrition and Physical Activity Survey. Note that the 'Not applicable' persons include those people who did not participate in the NHMS. The population for this table presents the weighted estimates for the population aged 2 years and over.
The same table using the 'Biomedical persons (Benchmarked weight)' will show the 'Month of biomedical collection' for only persons who participated in the NHMS. Note that in this case, no-one is in the 'Not applicable' category. People who did not participate in the biomedical component do not have a biomedical person weight and therefore do not contribute to the table when this weight is used. The biomedical population now presents weighted estimates for persons aged 5 years and over.
You can use a weight field with classificatory fields from other levels, but should take care when interpreting the results. Below are some examples which you can use as a guide.
MEANS AND MEDIANS
Means, medians and sums of continuous data items are automatically calculated at the level of the continuous data item. Due to current functionality of the software, a weight from another level cannot be brought into such calculations. The "subject" of means, medians and sums calculated in TableBuilder is therefore the statistical unit associated with the level of the database on which the continuous data item is stored. The weights used for these calculations are not visible, other than on the Person level, but are referenced in the 'Weighted by' statement with continuous variables, as per:
Means, medians and sums across levels
Means, medians and sums of continuous items are automatically weighted before the mean, median or sum is calculated. As TableBuilder only allows one weight to be included in a table, all other items in the table will inherit the weight applied to the mean, median or sum. This has implications when using means, medians and sums from one level with items from another level. For example, if you cross tabulate "Weighted mean of Age" (a Person level data item) with "Total cholesterol status (mmol/L)" (a Biomedical level data item), the default weight applied to the table will be "Persons (Benchmarked Weight)" because this weight is automatically included in the mean "Age of person" calculation. As a result, the biomedical item, "Total cholesterol status (mmol/L)" will also be weighted to "Persons (Benchmarked Weight)" not "Biomedical Persons (Benchmarked Weight)".
ITEMS LOCATED ON MULTIPLE LEVELS
Where items are available on more than one level, an additional number is added to the label to indicate the level version. For example, a (1) indicates it is a Household level version, a (2) indicates a Persons in household level version, a (3) indicates a Person level version, and so on. These are identified in the Data item list labelling as well as the item in TableBuilder. The numbering is based on the ordering of levels found in the File Structure page of this product.
Care should be used to ensure the correct version of the item is used, particularly with regards to demographic items located on both the Persons in household and Person levels. See below for more details.
PERSONS IN HOUSEHOLD LEVEL VS PERSON LEVEL ITEMS
The Persons in Household level contains data for every person in the household while the Persons level only contains data for the selected persons. Both levels are children of the Household level - that is, they are siblings and are not linked by person but by household (see the File structure page of this product for further information on structure). This means that there is a many-to-many link between records at these levels (persons on the Person level are linked to all the people in their household on the Persons in household level). When summing the Person weight (which is stored at the Person level) the meaning of the estimates produced when disaggregating by another data item at the Person level will not be the same as the meaning of the estimates produced when disaggregating by a data item at the Persons in Household level.
For example, disaggregating by Sex and Marital status at the Person level will produce estimates of the type "Number of persons who are Male and Married". These estimates will be additive (aside from the effects of perturbation) as shown below.
On the other hand, disaggregating by Sex and Marital status at the Persons in Household level, and using the Persons (Benchmarked weight) from the Person level, will produce estimates of the type "Number of persons in households containing one or more persons who are Male and Married". These estimates will usually not be additive, as shown below.
These documents will be presented in a new window.
4324.0.55.002 - Microdata: Australian Health Survey, Nutrition and Physical Activity, 2011-12 Quality Declaration
Latest ISSUE Released at 11:30 AM (CANBERRA TIME) 18/12/2013 First Issue