User:BenM/DVWG notes
Statistics
Caution should be taken with regards to ABS statistics, especially in comparision to police statistics. Having worked at the ABS and seen the way their figures are coded, as well as the way data is abstracted, it can be presented in a manner which is not really indicative of the scope of the subject matter it is related to.
For example, with violent crime such as sexual assault, if a woman is assaulted in three separate incidents in a given month, the ABS will count her once as there is a single victim. Even through there were multiple events, possibly with multiple perpetrators (though there need not be multiple perpetrators). Whereas law enforcement statistics will count the three separate events because those statistics are focussed on specific incidents or crimes, regardless of who the victim is and regardless of who the perpetrator is.
Which isn't to say the ABS statistics aren't useful, they most certainly are exceedingly useful, you just need to keep a sharp eye out for precisely what they're actually tabulating and how that might be interpreted by others. The classic example being, "oh look, the figures show there's less victims than you said, it's all a feminist plot, blah, blah, blah ..." Well, no, the figures show that the disparity between incidents and victims prove that in most cases there is evidence of a series or pattern of behaviour aimed at dehumanising and controlling the victim. And so on, and so on, and so on ...
Another example would be the, now dated, figures on suicide. They entirely ignore unsuccessful suicide attempts and other deaths which were probably suicide, but not ruled that way due to a lack of compelling and incontrovertible evidence. See the accompanying documentation to the deaths in Australia publications from the ABS, in particular the methadology used to compile the figures.
The methadology in that case is very similar to the type of thing used when collating figures from the census. Bear in mind that the census did not classify REDACTED as either stockbroking or trading because the ABS middle management hadn't heard the term used in relation to REDACTED exchange and trading. This was less than four years ago! The only major difference between census coding and coding suicide and violent deaths is it's even more strict.
Note: due to the terms of employment (and the Census and Statistics Act 1905) I can't name the term as it appeared on at least one person's form. If I do reveal what the redacted terms are, I face the possibility of 2 years in prison; this restriction will last until the year 2110. So I can tell you after my 135th birthday. This does, however, highlight another problem we face with this issue. There are many cases where there is information which would aid us and anyone else trying to find a solution, but which, for one reason or another is unable to be utilised. I'd like to say more on the statistics thing, to demonstrate certain dangers in taking it as gospel or interpreting it one way or another, but I can't without risking prison. So I have to bite my tongue while people make understandable mistakes because they assume the ABS methadology is consistently logical (actually it is, but they tend to take a much longer view than any other department - to the extent that they don't care about solving current problems, they only care about figures which can be reliably referenced hundreds of years after we're all dead).