Comparing the average number of persons per bedroom over time
There has been a change in methodology used to calculate the average number of persons per bedroom between the 2011 and 2016 Censuses. This page outlines the differences in the methodology to help users understand how to interpret and analyse this information.
To enable time series comparisons to be made on a consistent basis, in August 2017 the 2016 Time Series Profile (table T02) was updated with calculations made using the new methodology.
The method used to calculate the average number of persons per bedroom for the 2016 Census is summarised in the following equation. This method produces a representative dwelling based measure:
To calculate the average persons per bedroom, the ABS creates the ratio of number of persons usually resident in a dwelling to the number of bedrooms in a dwelling. This ratio is rounded to increments of 0.1 for processing purposes. The average persons per bedroom is then calculated in the standard manner as the sum of these ratios divided by the total number of dwellings.
This calculation uses the Census classifications:
(i) BEDD number of bedrooms in private dwelling, and
(ii) NPRD number of persons usually resident in dwelling. <link to Census Dictionary entries>
The method incorporates the following adjustments:
- it excludes dwellings where not stated and not applicable responses were found in either the BEDD or NPRD classifications.
- it excludes visitor only, other non-classifiable and not applicable dwellings using the HHCD household composition classification.
- it weights the number of people usually resident in dwellings by whole numbers based on the NPRD classification. The category of 8 or more people in this classification is weighted to 8.
- it weights the number of bedrooms in private dwellings by whole numbers (based on the BEDD classification). The category of 0 bedrooms (includes bedsitters) receives a weight of 1 to avoid scenarios involving division by 0, and the category of 30 bedrooms or more receives a weight of 30.
- this weighting approach is necessary because processing the ratio calculation can only occur on a discrete variable with a finite range.
For the 2011 Census, the process of creating this discrete variable was based on rounding the ratios to whole numbers. For the 2016 Census, the ratios have been rounded to increments of 0.1. This improves the accuracy of the average persons per bedroom calculation by minimising the downward bias present in this approach. To illustrate, for 2011 Census output, the ratio for 3 persons living in a private dwelling with 2 bedrooms was rounded to 1; for 2016 Census output, this ratio is rounded to 1.6.