FUZZY FIGURES: The construct of a “typical” person is meaningless.Then Bureau of Statistics has claimed to reveal the “typical” n.One nationally, eight for each of the states and mainland territories.
The typical n, we were told, is a 38-year-old woman, born here but “of English ancestry”, with two kids, three bedrooms and two cars. She wears red lipstick but only every other Saturday. (OK, I made that up, but only the last sentence.)
The entire exercise was a stunt. Not fake news, but certainly linguistic and statistical trolling.The clue, if one was needed, was in the ABS using scare-quotes around the everyday word “typical”.The construct of a “typical”person, concocted from over a dozen different measurements such as age, gender, ethnicity and household size, is meaningless.
The ABS might as well announce that the typical n meal is a Vita Brits sandwich of chocolate, chicken and salad. Yes, Vita Brits are popular for breakfast and chicken for dinner. But putting them all in one plate doesn’t make a meal any more than the ABS’s potpourri creates something resembling the “common man”.Your chances of meeting the ABS’s “typical”person yesterday was probably about one in several hundred.
If we said the “typical”dog in was a Staffordshire bull terrier, we’d have a factoid.According to aNational Geographic survey in 2012 that was the most popular breed. You may even meet a few in a week of park walks. The Staffy, however, might represent 5 per centof all dogs, but it is hardly typicalof anything.At best this would be a fudge for saying which breed was the “mode”. That is, the category that has the most numbers. Indeed, measuring only for recognisable breedsmisses the reality that the mongrel/mixedcategory may be the most common of all.
It gets worse. Dog classifications are one, straightforward metric. Yet the ABS’s “typical”person is constructed from 14 more subtle measurements.Almost all lie on a continuum.Finding a “typical”person on even one measure alone requires a concept of the average or the most common.Gender for the vast majority is a binary issue. But saying the “typical”n is a woman is odd when the split is about 50.5 to 49.5.
Why would the ABS engage in such floss? Cynics will say they are looking to rebuild their popular appeal after“CensusFail” in 2016. But almost 98 per centof the population was recorded in last year’s census. So for all the finger-pointing about planning and outsourced IT, the census turnout was excellent.
Ultimately, the ABS was just looking for some headlines to promote the release of the first tranche of census data. In the end, it’s just a bit of fun, surely?Yes. And no. That the nation’s leading statisticians would fudge statistical concepts isn’t reassuring, even accepting it’s a world where even the public service competes to brand itself.
I’m just a law academic, albeit with a maths degree. Why should I be commenting?First, we should care how officials use words, whether it is legal lingo or concepts such as “typical”.Second, our society has a dearth of basic mathematical knowledge. I see it regularly, amongbright students who dodged or rote learnt maths at school. They are set for professional degrees. But like many of us, they are ripe for such fudges as the use of a fuzzy idea such as “typical”to obscure the more nuanced world of means, medians and modes.
Graeme Orr is a professor of law at the University of Queensland.