The primary goal for the processing of the 2016 Census is to ensure the timely release of data, whilst maintaining or improving data quality. Release of data to the public is expected to be in a similar time frame to 2011.
The processing system in 2016 will be enhanced to ensure optimal results from the digital approach to the Census with its new methods of delivery and collection. For example, address information used in planning the delivery and collection activities will be integrated into the processing system to assist data reconciliation and address coding.
Processing the Census includes all steps from receipt of Census responses in either online or in paper form through to the production of a clean Census data file.
A Data Capture Centre (DCC) will be established to capture and process the data from the paper forms. Data will be captured and processed using the same technologies as in 2011, namely imaging and Intelligent Character Recognition for paper forms, and an online instrument for electronic collection. Once captured and loaded onto secure ABS systems, responses are then coded using automatic and computer assisted coding techniques.
Processing of data will proceed as follows:
Receipt and registration of forms: Upon arrival at the DCC, Census forms will have their Form ID electronically captured. This information will be immediately communicated to Field Staff to eliminate contact with people once their forms are submitted and received. Follow-up of missing forms will be rigorous to ensure coverage is complete.
Data capture: Forms will be scanned and a reconciliation process conducted to ensure that all forms which have been received at the DCC have been captured.
Coding: All responses on the Census form are coded from indexes. Coding techniques include automatic, computer assisted and manual coding.
Imputation: Imputation is the process whereby information for a small number of non-responding dwellings is included in the Census dataset based on credible source information. Improvements have been made to the accuracy of the imputation methodology for missing values for sex, age and marital status. These improvements will be made in concert with the need to streamline existing systems and harness different information collected under the new enumeration model.
Editing: Editing is a process that looks to correct errors in the data and will be undertaken as part of the validation strategy to produce a consistent, valid dataset. The kinds of error which editing procedures can detect are limited to responses and/or codes which are invalid, or which are in conflict with Census definitions.
Quality Assurance: A variety of quality assurance measures will be implemented across the various systems and processes throughout the processing cycle. This will provide an accurate, consistent and coherent final Census output file.