Public policy is often determined by polling and for interested followers of politics, the fortnightly Newspoll in Australia is an institution. These predictions have become increasingly important to politicians, the media, and the public, with some of Australia’s most important decisions in recent history being made as a result of the public sentiment captured by pollsters.

As rumours of a 2021 federal election grow louder, the team at Nexus is here to help answer your most pressing questions about Newspoll and opinion polling in general.

How does opinion polling actually work? How accurate is the data really? And most importantly, how much weight should we be putting in these political polls and why (or why not)?


A public opinion poll is a type of interview or survey designed to measure the population’s views regarding a particular topic.

In this case, Newspoll is just one of the political polling companies designed to estimate the public’s opinion of current politics and, most importantly, determine the next election’s outcome.

It should be noted that there are two types of polling: 1) private polling commissioned by political parties or organisations for private research and also 2) public polling – usually published in media outlets.  In this article, we will focus on public opinion polling.


The mission of political polls is to be as accurate as possible. To do this, trained interviewers ask specific questions to a group of people chosen at random to represent the larger population.

It is essential that the people surveyed for the poll are chosen at random. The most popular methodology for achieving a random sample of the population is random digital dialling (RDD).

Companies like Newspoll use a continuously updated database of all the listed telephone numbers in the country for both cell phones and landlines. Polling companies then use computers to analyse the database to identify all ‘active blocks’ of numbers. These active blocks are the area codes and exchanges (the second three-digit) actively in use. The computers are then programmed to randomly dial every possible number in each active block.

To get a truly random sample when dialling landlines, pollsters need to choose random numbers and choose random respondents within the household. American research indicates that women and older Australians are far more likely to answer the phone than other people.  Interviewers will often ask to speak to a specific member of the house and sometimes ones with the most recent birthday.

Once a political polling company has collected responses from a sufficiently random sample, it must adjust or weight that sample to match the most recent census data about the sex, age, race, and geographical breakdown of the public.

For example, a pollster might find that only 5% of people included in the poll were “First Nations peoples, compared to the 10% of the Australian population.  As such, pollsters might use weighting to effectively account for First Nations peoples’ responses twice towards the poll’s overall results.


The idea for a public survey to predict election results and measure opinion is an old one, but the method has evolved significantly over time.

The modern Australian news poll was based on the ‘Gallup Method’, invented by American George Gallup. Gallup was the first to try sampling a randomly selected, statistically average group of people as representative of the larger population. Gallup’s first poll in 1932 successfully predicted the outcome of the local election in Iowa. More importantly, in 1936, Gallup went up against a more respected ‘straw poll’ conducted by the Literary Digest. The Literary Digest used its data to predict that Franklin D. Roosevelt’s opponent Alf Landon would win. Gallup’s method worked, and Roosevelt won in a landslide.

It was not until 1985, with the creation of Newspoll, that a seemingly more accurate type of political polling appeared in Australia. Initially started by News Limited and Yann Campbell Hoare Wheeler, Newspoll is now published by The Australian and administered by international market research and data analytics group, YouGov.

Newspoll is well-known for the regularity with which it conducts research and releases data, despite perceived inaccuracies in recent years. The company currently uses a mixture of telephone and online polling to collect sample data.


Newspoll’s election predictions and political polling have the reputation of being the most accurate poll in Australia.

Since 1993, Newspoll has accurately predicted the winner of Australia’s state and federal elections about 80% of the time. In 2010 and 2016, the polling almost precisely matched the election result.

This level of accuracy led some in politics to take Newspoll results very seriously, with the most high profile consequence being the change of leader to Malcolm Turnbull from Tony Abbott after the latter had lost the two-party vote in 30 consecutive Newspolls in 2015.

These actions had a karmic return for Turnbull, as in 2018, he also had trailed behind Labor in more than 30 Newspolls before his government replaced him with current Prime Minister Scott Morrison, as the 2019 election loomed as a possible defeat for the Coalition.  The 2019 Federal Election was naturally the big surprise result that was not accurately picked by public polling.


In recent times, Newspoll has lost some of its accuracy, with traditional methods no longer yielding results that are representative of the broader population. This has seen three consecutive major elections in Australia (VIC 2018, NSW 2019 and the 2019 Federal Election) being either incorrectly predicted or the magnitude of victory underestimated.

These results led polling companies, including Newspoll, to reconsider how opinion polling is administered, with new quantitative methods and new sources of information needing to be utilised to bring opinion polling into the 21st century.


Online polling has also become more prominent in political polling and is now an essential aspect of Newspoll’s sample collection. Online polls are a cheap, fast alternative to live-caller interviews, but they can still contribute to inaccurate data. In 2020, the Australian Bureau of Statistics stated that 2.5 million Australians do not have access to the internet. This means pollsters must add another layer of weighting to try and get close to a representative sample of the public. Recent studies show that even after weighting, online polls tend to overrepresent men and the unemployed, which works to skew the poll results.

Ultimately, companies like Newspoll can use various methods to make their poll results as accurate as possible. Still, numerous factors decrease the validity of their predictions that cannot be accounted for.

So, when Newspoll releases its next 2021 election predictions, be aware of their data’s duality; it is both the most accurate and inaccurate polling on Australia’s political opinion.