August 9 is Census night, and the Census is Australia’s unique opportunity to get the facts on Australian life.

No one says it better than the Australian Bureau of Statistics: “The aim of the Census is to accurately collect data on the key characteristics of people in Australia on Census night and the dwellings in which they live. In 2016, the Census will count close to 10 million dwellings and approximately 24 million people, the largest number counted to date.”

Market analysts, researchers, policy makers, government agencies, developers, planners and just about every type of business in the country will soon be poring over the results of the 2016 Census as the results begin to dribble out later this year. They will be examining in detail how our incomes – both personal and household – are faring, how we use our housing, how we get to work, and subtle changes in society in everything from religion to ethnicity to family types.

As a source of accurate, unbiased and comprehensive data, the Census is priceless. So why is it so regularly abused?

By “abused”, I mean the manner in which data is frequently pulled out of context or subject to contorted interpretation, often in support of a pre-determined agenda. Confirmation bias is the process of searching out information that confirms a pre-existing view, and discounting information that doesn’t support that view. The Census is a happy hunting ground for commentators looking to confirm their bias.

Here’s an example: we read often that ‘the nuclear family’ is in decline, and that single person or alternative households are the fastest growing household type. It is a comment so widely repeated that it’s almost become a truism, and gullible policy makers have been known to fall for it without calling for the context. This truism has been called upon in support of wholesale housing change – the argument being that the declining ‘traditional family’ model (couples with kids) means we need fewer typical suburban homes and more high density dwellings for the burgeoning numbers of singles or group households.

You can find examples here, and here. The trick seems to be to take a trend, extrapolate it indefinitely until a minor group becomes a major one, and then attach some quirky acronym to the demographic for maximum media interest.

So what do past Censuses actually say?


The table above summarises household types in Australia according to the 2006 and 2011 Censuses. If you want confirmation bias that the nuclear family is in decline, you would seize on the figure that shows ‘couples with children’ grew by ‘only’ seven per cent in the five year period – the slowest percentage growth of any of the household types (excluding the not classified). You might then seize on the nine per cent growth rate of lone person households, and maybe also the 14 per cent growth rate for group households, to argue that alternative household types are the fastest growing household types and that policies should change to reflect this. This is especially so if you think extrapolating these trends forward decades into the future is a valid approach to forecasting. It isn’t, but that doesn’t stop a lot of people.

A more impartial observation (but one unlikely to achieve as many headlines) would be that couples with children are the biggest single household type, followed closely by couples without children. Couples without children can include young couples in the pre-children phase, or older couples without children any longer in the home – but both are still what you would define as traditional family types. Combined, they represent 56 per cent of all household types in the country – a proportion which has barely changed in the inter census period.

Plus, they have each grown numerically more than single person households, so can hardly be described as in decline. A further valid observation is that lone person households would include widows and given our ageing population; there’s a valid reason for expecting the number and proportion of lone person (including widowed) households to rise further. Suggesting some connection between the rise of single person households and rising inner urban hipster appetites for single bedroom apartments is not something directly supported by the data.

Much longer term analysis of Census data will show larger changes in household types. True, the nuclear family type is less prevalent now than in 1971, for instance. True, there are more single households (mainly driven by ageing). True, more families are having children later in life, and the number of children per household has fallen. But these are changes in the very long term, with more than one explanation possible. The impact of housing affordability and economic uncertainty on decisions about when, and how many children to have, is not something the Census will reveal. Neither does it suggest that raising a family within a conventional two-parent family has fallen afoul of the majority of people (as some suggest), or that decisions about housing type and location are being made because of preference (as opposed to limited choices or affordability).

This is just one example of how Census data will be interpreted differently – sometimes because of pre-determined agendas or ideologies, sometimes by headline chasing analysts and commentators keen to exaggerate minor changes at the margin, and less frequently by impartial observers keen to understand the truth and ready to ask further questions.

No doubt as Census 2016 results are released, the rush will be on to confirm social theories and ideological agendas, and we will be bombarded with commentators ‘expert’ conclusions, referencing Census data, in support of predetermined views. We are a society increasingly awash with information – and some of it very high quality as with the Census – but this doesn’t mean we necessarily benefit from quality analysis or logical conclusions. There are too many agendas out there in need of dubious statistical support.