Completed Census forms are delivered to the Data Processing Centre (DPC) as soon as possible after Census Night. They are then put through a number of processes.
The first processing stage in the 2006 Census is precapture. During this process forms are checked to ensure key fields have been completed and extraneous material removed. The forms are then prepared for data capture.
The data capture stage is the second stage of input processing. This stage encompasses a number of processes, including:
The second stage includes Automatic Coding and Computer Assisted Coding (CAC). All hand-written textual responses are examined automatically to see if a classification code can be allocated based on the response provided. Where a classification code cannot be allocated automatically, CAC is used to allocate the classification code.
- Scanning, which captures an image of each page of each form;
- Intelligent Character Recognition (ICR), which converts any mark box or hand-written responses found on an image into machine processable information;
- Repair, which is a mixture of automatic and clerical processes aimed at correcting any data not confidently captured by ICR;
- Balancing, which ensures that the numbers of dwelling and associated person(s) provided to the data processing centre have been captured by scanning and ICR. Once this stage is completed, the forms and record books are stored and are not included in further processing stages.
Quality control checks are constantly made for coding accuracy. The editing process is largely automatic, with some fields being reset based on other responses on the individual Census forms. All Census data are extensively validated before data are released.
See also Data release, Derivations and imputations, Intelligent Character Recognition (ICR).