Key takeaways

  • The 2016 Australian census has a large focus on digital and online submission.
  • Digitising surveys lowers costs and can boost accessibility and language support.
  • A digital census could also unlock many new potential insights and capabilities, such as how questions are answered.

This week, Australians will take part in one of the country’s biggest online events: the 2016 census.

Described by the Australian Bureau of Statistics (ABS) as the “nation’s largest peacetime logistical operation”, the census will seek to capture an up-to-date snapshot of the nation, gathering demographic and socio-economic information that will be crucial for future evidence-based policy making.

In a bid to keep pace with changing technology, this year’s census also has a large digital focus. According to the ABC, up to two-thirds of Australians are expected to complete the census online using a computer, smartphone or tablet. If this projected turnout is accurate, Australia could become a ‘world leader’ in digital censuses.

Compared to its paper-based predecessor, a digital census – like any online survey – can unlock many new capabilities. These could include reduced costs, faster processing times, and the presentation of questions in a much wider range of languages and accessibility standards.

As PwC Australia’s Chief Data Scientist Matt Kuperholz points out, this is only scratching the surface of what a digital census can do. I sat down with him to discuss the several opportunities of a fully digital census and what lessons this could provide for businesses looking to conduct surveys of their own.

The data
behind the data

While the ABS estimates the digitising of the census could save taxpayers up to $100 million a year, a lesser-known feature of a digital survey is the chance to gather information about how the information itself is accessed and completed.

“If you do a survey mostly on paper, you really have no idea about how it’s being filled in,” says Kuperholz. “But if it’s digitised and designed to do so, all of a sudden you can get additional information beyond simple answers to the questions.”

This additional information could include how long users take to complete the census, what time of day they sit down to do them, or the frequency and duration of pauses between answers.

Other data that could be gathered includes basic information sent from a user’s internet browser (screen resolution, plugins, language, etc) through a process known as ‘browser fingerprinting’, and public details about their IP address, which can help broadly geo-locate demographics by their internet service provider or if they are visiting from inside a corporate network.

No further
questions

For Kuperholz, however, it’s the information about the survey process itself that can be extremely valuable.

“When I analyse digital surveys, the one thing I consistently encounter is how people write surveys to be far too long,” he continued.

“Because the survey itself is assumed to be the most important part of the process, survey writers usually don’t put themselves in the shoes of the respondent. This can result in too many similar questions with highly correlating answers, blowing out the length of the survey and creating fatigue.”

A census
about the census?

Analysing this sort of survey data can produce many new insights, especially when these are combined with the answers. As an example, Kuperholz cites the time taken between answering questions.

“Instead of just trying to analyse the overall pattern of the data, other phenomena such as respondents taking ten minutes to answer one question, then answering ten more in ten seconds – faster than they could have read the actual questions – can be observed.”

If this information is then combined with some of the survey data itself, even more nuanced questions could be investigated.

“One example could be: how does the level of attention and care correlate demographically or socio-economically? In other words, do highly educated professionals rush through the census? Or is it the opposite? For other demographics, is a language barrier an issue?”

While respondents will always be more likely to pay less attention to the final questions in a survey, this data could nevertheless lead to the creation of more effective survey structures. By going digital, the census can simultaneously function, in effect, as a census on census-taking.

Deltas
of change

If these innovations are taken on board and accepted by the public, we could be looking at a future of shortened census cycles.

“One of the challenges data scientists and statisticians face is that the census occurs so infrequently. Data such as the growth of different demographics within a certain area must often be ‘filled in’ between known data points – a process known as interpolation. This makes it less accurate,” notes Kuperholz.

“If we could move towards a secure and encrypted census that focuses on ‘what has changed’ – incentivising people to submit their deltas of change rather than filling it all in afresh – those are some of the things that could be investigated in a fully digital census environment.”

By the
numbers

The census – and the technology that enables it – has come a long way. From the five-yearly censuses of the Roman Republic that calculated taxation and those fit for military service, to Australia’s introduction of its electronic census in 2006.

While we don’t know what digital innovations have been incorporated into this year’s online census, there are many possibilities for an end-to-end improvement. Multiple languages could be displayed on command, accessibility could be optimised for the visually impaired and additional explanatory information could be made available through hyperlinking.

What’s more, with the census being taken by one of the largest and most diverse audiences available, a full spectrum of usability issues could be reported in a short period of time, providing invaluable feedback about how Australians interact with online interfaces.

Indeed, a digital census of this size provides the rare opportunity to improve the art of online surveying itself.

Of course, these innovations, while available now, can only be made possible if they are utilised.

“If all that happens is the paper version is put online,” adds Kuperholz, “then some potentially valuable opportunities are going to be missed.”