A concordance, in statistical terms, is a product that allows a user to convert data from one geographical region (under which data have been collected) to a new geographical region.
The primary benefit of concording data to a new geographical region is that it allows a user to make direct comparisons between two datasets that may have been collected under different geographical classifications.
Types of concordances
In order to convert data from one geographical boundary to another, each region in the new boundary is assigned percentages of data from the old regions. These percentages in the concordance can be constructed using any number of variables. The most commonly used variables include population, dwelling counts and area.
A population-based concordance derives its percentage splits using population. This type of concordance is useful when concording demographic-based datasets such as labour force and family composition.
A dwelling-based concordance derives its percentage splits using dwelling counts. This type of concordance is useful when concording dwelling-based datasets such as tenure type of dwelling structure.
An area-based concordance derives its percentage splits using area. This type of concordance is useful when concording area-based datasets such as agricultural production.
Whilst these three concordance types are the more commonly used concordances, a concordance can be created from any variable to create a more specific concordance. Examples of these variables can include Indigenous persons or a particular age-by-gender cohort.
Creating and applying concordances
There are several approaches to creating a concordance file. Determining the best method will depend upon how it is intended to be used.
Hierarchical geographical datasets
Concordances for hierarchical geographical datasets, such as certain parts of the Australian Statistical Geography Standard (ASGS), are simple to create, because each child geographical unit will sit perfectly within a parent geographical unit. There are no percentage splits for a concordance of this type because they are purposely created to align exactly.
Non-hierarchical geographical datasets
Concording non-hierarchical datasets is a more complex task, because there are several methods for doing so.
The most accurate method for creating a concordance for non-hierarchical datasets is to use the smallest geographical dataset available as a building block for calculating the percentage splits. Examples of some of the smaller geographical datasets that can be used include the Geocoded National Address File (G-NAF), Mesh Blocks (MB) or the Digital Cadastral Database (DCDB).
Another less common and less accurate method is to apply a physical split of one region using another region, and calculating the percentage of the original area within each split region.
Assumptions
When using concordances that have been derived from smaller polygons, it is assumed the polygon’s associated population is distributed evenly over the area.
When using concordances that have been derived from point data, it is assumed the associated population is proportionally distributed over a specific area at the same rate as the point distribution.
Caution should therefore be exercised when using concorded data as this assumption may have implications if it is apparent that there are distinct clusters of a specific sub-population cohort within the smaller areas.
Derived concordances
For a list of QGSO-derived concordances, see the Geographies and maps page.
For a list of ABS–derived ASGS concordances, see their correspondences page.