In considering this question in the context of the DPC, the most obvious point to be made is: coders can code only what is there in the forms. No matter how advanced and sophisticated the technology, no matter how expert, well-trained and dedicated the staff, no matter how effective the systems for checking internal consistency, the fact remains that one gets out what one puts in. And what is put in is a very partial representation of the Indigenous ‘facts’, couched in the categories of the mainstream. It was somewhat disconcerting to see such care, thought, expertise and expense being devoted to the processing of this data—in particular, the household data.

The data are partial in at least two senses. Firstly, because of the intercultural difficulties of the enumeration exercise, many data are missed, or presented incoherently. The forms are filled in incompletely or the answers are hard to interpret. Secondly, the data are partial in the sense that it captures only a partial representation of Indigenous lives and circumstances. Some would argue that this is necessary and desirable—the state needs to know only so much about its citizens in order to plan and deliver policy and services. What is more, the census is a broad-brush instrument. It is not designed to capture the subtleties of social life. There are, however, unhelpful cultural biases in this partiality that could be addressed, so that a more accurate partial picture emerges—one that would make Indigenous people more recognisable to themselves in the output of the census. To that end, I have gone into some detail on the subject of the representation of the family structures of Indigenous households.