Sunday, 10 July 2016 19:27

'Horse-race polling' not ABC mandate Featured


The ABC had a much wider sample of digital polling data than commercial polling companies during the recent Australian federal election, but did not use the material to say which party was ahead at any particular point in time because "that is not our mandate", a spokesman for Vox Pop says.

Comments from Clifton Van Der Linden, founder and chief executive of Vox Pop Labs, the people behind the ABC's Vote Compass survey that ran during the campaign, provided to iTWire by Sally Jackson, ABC media manager for news and current affairs, said: "As to why we don't use the data to engage in horse-race polling during the election campaign, that is not our mandate and we do not collect the variables necessary to run horse-race polls (e.g. TPP). Our data are unique and we use them to offer unique insights into public opinion on a host of salient public policy issues in a given campaign."

iTWire had written about, and, later raised with ABC, some questions about Vote Compass, the ABC survey that ran during the campaign, asking how an opt-in poll could claim to make categorical statements about the entire Australian population, given that the submission of personal identifying characteristics was not mandatory.

In response to queries, Jackson replied: "Vote Compass only draws from records where the required socio-demographic weights are reported, which is why the sample size reported in the methodological note that accompanies each article is lower than the overall number of users registered on Vote Compass at a given time."

She added: "We then weight the remaining sample by population-level estimates as per the Census, which is both an industry standard practice and a technique that has been validated in the world’s leading scientific journals.

"We actually weight on far more variables than do pollsters, given that the size of our samples allows us to do so without introducing large distortions."

Asked how the ABC could claim that Vote Compass could not be gamed, when it was indeed possible for two people to vote from the same IP address, Jackson said: "We treat the data after it has been collected and before it is reported and remove multiple entries from the dataset prior to analysing the data. And we do so using a much more complex detection strategy than IP addresses alone."

She did not specify what this "complex detection strategy" was.

iTWire wrote back asking: "If, as you say, the ABC had much better data than the professional polling companies, why didn't the ABC issue its findings on the support enjoyed by the major parties, same as Newspoll or Fairfax-Ipsos? After all, if the data was so much better, why waste it merely on what, in the context of the election, are actually quite trivial generalisations? The only beneficiary of knowing what support was enjoyed by the parties would have been the owners of the ABC - the Australian people - isn't it?"

The ABC had made some generalisations about the Australian population during the campaign, such as "80% of Australians back a federal corruption watchdog"; the word "trivial" was used to compare this conclusion to the weightier question of which party was ahead in the polls at any given time in the campaign.

To that Van Den Linden wrote: "We hardly think this 'trivial'; quite the opposite, in fact. We try to broaden and deepen the electoral discourse by using our data to increase the focus on public policy as opposed to the horse-race - to give voice to Australians on the issues that concern them in this campaign."

He added: "It is misrepresenting our argument to say the ABC or Vote Compass is claiming to have 'better data than the professional polling companies'. Our argument is that Vote Compass sample contains orders of magnitude more observations than conventional samples collected by commercial polling firms.

"We trust that these firms have reliable methods for making representative inferences from the sample they collect. The size and structure of our dataset are different, but the techniques we apply to the data to weight it to the general population are theoretically consistent with those of commercial polling firms."

Van Den Linden did not say whether the techniques applied to the Vote Compass data were also practically consistent with those of commercial polling firms.


26-27 February 2020 | Hilton Brisbane

Connecting the region’s leading data analytics professionals to drive and inspire your future strategy

Leading the data analytics division has never been easy, but now the challenge is on to remain ahead of the competition and reap the massive rewards as a strategic executive.

Do you want to leverage data governance as an enabler?Are you working at driving AI/ML implementation?

Want to stay abreast of data privacy and AI ethics requirements? Are you working hard to push predictive analytics to the limits?

With so much to keep on top of in such a rapidly changing technology space, collaboration is key to success. You don't need to struggle alone, network and share your struggles as well as your tips for success at CDAO Brisbane.

Discover how your peers have tackled the very same issues you face daily. Network with over 140 of your peers and hear from the leading professionals in your industry. Leverage this community of data and analytics enthusiasts to advance your strategy to the next level.

Download the Agenda to find out more


Sam Varghese

website statistics

Sam Varghese has been writing for iTWire since 2006, a year after the site came into existence. For nearly a decade thereafter, he wrote mostly about free and open source software, based on his own use of this genre of software. Since May 2016, he has been writing across many areas of technology. He has been a journalist for nearly 40 years in India (Indian Express and Deccan Herald), the UAE (Khaleej Times) and Australia (Daily Commercial News (now defunct) and The Age). His personal blog is titled Irregular Expression.



Recent Comments