Concoction, inaccuracy and irrelevance

[Chris Lloyd]

We all know the adage that 57% of statistics quoted in an argument are made up and the other 44% are inaccurate. The third member of the holy trinity of misuse of statistics to my mind is irrelevance. And wrong headed conditioning is usually the best way to end up with an irrelevant statistic. To whit

“Domestic violence is the single most likely cause of preventable death for women under the age of 45”.

This week in class, I had an interesting discussion about this statement, which I have heard or seen many times. A few years ago it appeared on billboards around Melbourne and it was in a Green Left Weekly editorial last December.



I could not Google a source for this statement but I am prepared to assume it is true for the sake of argument.

Whenever*a statistic or probability is quoted, it should answer a*question, or at least shed light on the answer. The title of the GLW piece is

There is no room for any doubt that Australia is suffering from an epidemic of domestic violence

So presumably we want to quote statistics in order to address the question “how high is the rate of domestic murder of females?” The statistic is being quoted to suggest that the rate is high – higher perhaps than you might have expected - but it is suggesting it very indirectly.

The quoted statistic says that, for those women under 45 who die, domestic violence is more probably than any other single cause. In terms of symbols, let D be the event of death, A be the age and let us list the various possible causes of death such as

CDV=domestic violence

CMV=motor vehicle accident

CMS=murder by stranger

CTI=terminal illness

CO=other cause


So the statistic is really saying the Pr(CDV|D,A