Code of practice offers greater clarity on polling
A new code of practice will offer greater transparency on how polling companies obtain samples and analyse data.
After the 2019 federal election, Newspoll made significant changes to its methodology to address the discrepancy between published poll numbers and the final outcome of that election. These changes have been tested at two state elections since – in Queensland 2020 and Western Australia 2021. Despite these elections being very unusual due to the Covid-19 pandemic, Newspoll performed very well at these elections which indicates that the changes we made are working.
YouGov, the data and analytics group which conducts Newspoll for The Australian, has advocated for greater transparency in Australian public opinion polling. For this reason we strongly support the Australian Polling Council’s new code of practice, which is designed to provide Australians with more information about how polls are conducted.
This is similar to the British Polling Council, which helped the British polling industry turn things around with a very credible performance in the UK 2019 election and the Scottish 2021 election.
The Australian Polling Council’s code of practice will require polling companies to provide more information about how they obtain their samples, how they analyse their data and how they structure their questionnaires. This will allow journalists, academics and the general public to make informed decisions about whether or not polling is being conducted properly.
Newspoll has always had a degree of transparency not seen in other polls, because The Australian prints the questions in full in the tables that accompany the news stories. But the new rules mean that additional information can be provided.
Once upon a time, polling was conducted face-to-face, by knocking on people’s doors. When Newspoll began in 1985, it was the first major Australian poll to be conducted entirely by telephone. This was controversial at the time, because the so-called experts suggested that the new method was inferior. However, the growth of the internet, mobile phones and especially call screening have meant that telephone polling is less feasible and the samples obtained less representative. For this reason, many global polls – including Newspoll – have shifted to be conducted online.
This gives us significant advantages over the methods used previously. We control for more variations in the sample by ensuring that groups who are less willing to do polling than others are properly represented. People can’t opt in to do a survey – we send invitations to a large panel of Australians in a targeted way to ensure the sample represents the Australian population as closely as possible.
For instance, YouGov now targets, then weights our sample by levels of education, household income and what sort of electorate they are in. Groups like non-tertiary educated younger people, very high income people and people in rural areas are harder for pollsters to get to do surveys – therefore our weighting ensures these groups – and other groups – are properly represented in the sample.
Another major difference is that we now have a much more sophisticated questionnaire which means the presentation of the questionnaire varies according to the survey participant’s electorate. This means we can show participants the choices they are likely to have on the ballot paper, such as whether they have a Nationals or a Liberal candidate (or in some electorates, both), whether they are likely to have a One Nation or an Animal Justice candidate to choose from, etc. This attention to detail wasn’t common in Australian polling, and Newspoll is now one of the most robust polls globally.
Polling is based on science, not magic, and it always has a degree of uncertainty attached to it. That science means that we are making as estimate, not gazing into a crystal ball, and good science is transparent and scrutable. Sir Winston Churchill once said that democracy is the worst form of government – except for all those other forms that have been tried from time to time. Similarly, polling isn’t perfect, it’s just far better than any other way of measuring democratic opinion.
Campbell White is head of public affairs and polling – Asia Pacific – for YouGov
To join the conversation, please log in. Don't have an account? Register
Join the conversation, you are commenting as Logout