Australian Bureau of Statistics prepares for 2016 Census

Australian Bureau of Statistics prepares for 2016 Census

SHARE

With the successful implementation of GIS in its recently completed 2011 Census of Population and Housing, the Australian Bureau of Statistics (ABS) is exploring the uses of additional GIS capabilities for the upcoming 2016 Census.

The Australian Bureau of Statistics (ABS) has recently completed the 2011 Census of Population and Housing. Geographic Information Systems (GIS) and geospatial data were utilised across all phases of the project including the design of collection and output areas, management reporting of progress with eCensus take up, data processing and geocoding, and ultimately in the information dissemination systems.

Additional functionality is being developed for use in the 2016 Census, particularly associated with the move to developing an address register to mail out instructions to access the eCensus form to around 60% of the population. Network analysis has been tested and implemented in some parts of ABS’ business areas to improve allocation of the field workforce and the use of GPS and mobile mapping applications are also being investigated to assist with enumeration.

Towards 2016 Census

Preparation is well underway for the 2016 Census. Although final decisions are still to be made, a new Census process based on a mixed mode of delivery is envisaged. The process will utilise an address register and mail out of Census materials for the more urbanised areas of the country, while traditional manual delivery will occur in the more difficult to enumerate areas. This change in methodology will also drive other changes in the use of GIS for the 2016 Census which are outlined below.

Development of an address register

An address register is being developed to support the mail out of materials to around 60% of all households. The address register will be based on a Geocoded National Address File (G-NAF), which contains the most comprehensive list of official addresses in Australia. One significant issue, however, is that G-NAF contains the addresses of almost all properties, and not just the residential dwellings required to be contacted in a Census. One of the challenges in developing an address register suitable for supporting the Census, and in the longer term other household surveys, will be to identify a way of maintaining a land use code against the address. This code will need to differentiate between dwelling addresses and other types of non-dwelling addresses, such as vacant sites, industrial sites, etc. Applying this land use classification, and ensuring the overall currency and accuracy of the register, will require a large body of quality assurance work. This is likely to entail field canvassing of many of the addresses, prior to using the register to construct a “Census Frame” for the 2016 Census.

Address management will therefore be a critical component of the Census in terms of form delivery and follow up, and also in the coding of addresses to a geographical area. There are actually five addresses captured and geocoded as part of the Census, as outlined below.

Type of Address

Census Year (2011)

Dwelling Address

9,300,000

Usual Address – (where different from address on census night)

1,000,000

1 year ago address

2,000,000

5 year ago address

9,000,000

Workplace address

10,000,000

The approximate number of addresses requiring geocoding for 2011

Use of network analysis to improve allocation of field workforce

Esri’s Network Analyst together with the NAVTEQ networked road dataset for Australia has been successfully utilised to calculate road distances between a matrix of addresses. The output from this is currently used in an optimisation process for the allocation of survey interviewers to interviewees. Based on the success of this trial, other applications are being developed and tested, including:

  • Developing a process for ordering addresses within an area into a logical list for field staff to “Check List”. The list is sorted into order that minimises walking / driving times.
  • Optimising the creation of adaptive workloads as a basis for field checking non-responding dwellings during the Census enumeration process.
  • Optimising workload area design to minimise travel outside a workload area to complete enumeration.

Use of GPS and mobile mapping applications to improve enumeration

Use of mobile electronic devices (tablets and or smartphones) has the potential to create significant advantages, such as:

  • Providing maps to assist with locating the correct dwellings.
  • Undertaking data validation and coding processes in the field.
  • Linking a statistical instrument (Census or survey form) to a location in the field at the point of entry using a GPS enabled mobile device.
  • Reducing costs associated with data entry, including scanning and editing forms.
  • Facilitating the more efficient communication with and management of field staff.

Some testing of devices has already taken place which has identified many issues associated with their use. The issues being evaluated include data security, reliability of devices, costs and occupational health and safety, to name a few.

It is also anticipated that e-Census will become the preferred method of response to the questionnaire in 2016, with take-up rates expected to increase from those obtained in 2011

Map of 2011 e-Census take up rates

Conclusion

Effective use of GIS within an organisation such as the Australian Bureau of Statistics relies heavily on the access to good base data, strong internal management of that data, flexible software tools and most importantly staff with geospatial capability to manage the Statistical Spatial Framework and increasing application of GIS in the organisation. Successful application of GIS in the 2011 Census is a step along the way to the additional capabilities being developed for 2016.

(This paper is an abbreviated version of the paper submitted to the World Statistics Conference in 2013)