There are two polls out today, one showing good news for the government, the other showing bad news for the government. How can this be the case? It’s all in the different methodologies used by the respective pollsters and a statistical item called the ‘margin of error’.
Newspoll, published by The Australian, shows increase in support for the Liberal–National Party to 48 per cent on a two-party preferred basis (up 2 per cent from March 7), which means a decrease for the Labor Party to 52 per cent.
On the other hand, the monthly Ipsos poll published in the Sydney Morning Herald, shows a widening of the polls, 47 per cent of the two-party preferred vote to the LNP, and 53 per cent to the Labor Party.
The Mediapoll aggregation from all pollsters is 47.7 per cent for the LNP, and 52.3 per cent for the Labor Party. There has been a very slow tightening of the polls since August 2018, when Malcolm Turnbull was deposed as Prime Minister, but even if that trajectory continues, with all other issues being equal, it will only reach 47.1 per cent at the time of the election, which is due on either May 11 or May 18 this year.
However, whichever poll or metric is used, the Government is still facing a landslide defeat, if these poll numbers continue.
The lowest two-party preferred vote to win a federal election is the 49.0 per cent achieved by the Liberal–National Coalition in 1998, followed by the 49.3 per cent result in the 1940 election, also achieved by the Coalition.
An election victory with a two-party preferred vote of less than 50 per cent has only occurred on five occasions in federation in 1901. Four of these have been by the Coalition, with the last occasion being in 1998, and once by the Labor Party in 1990. And on all of those occasions, a government has been returned from a position of strength (albeit with a fall in votes), while the current LNP Coalition is coming from a position of weakness, holding minority government since November 2018.
|Election Date||ALP||Coalition||ALP seats||Coalition seats|
|21 September 1940||50.3%||49.7%||32||36|
|29 May 1954||50.7%||49.3%||57||64|
|25 October 1969||50.2%||49.8%||59||66|
|24 March 1990||49.9%||50.1%||78||69|
|3 October 1998||51.0%||49.0%||67||80|
Graph: Winning elections with less than 50 per cent of two-party preferred vote.
The upshot is: this Government will need to break many electoral records and overturn historical precedents if it is to be returned for a third term. Based on the 168 consecutive losing polls for the LNP since August 2016, it’s difficult to see where an advantage could be for the Government to turn these figures around.
While The Australian has argued the recent boost in the Newspoll numbers for the Coalition is all due to the electorate’s favourable response to the 2019 Budget, budget announcements don’t have the same impact they’ve had in previous decades, and any influence on polling figures, if any, takes weeks to advance.
Optimists for the LNP can point to the ‘Kinnock Principle’, where UK Labour opposition leader lead the polls for over 18 months, up to the polling date in 1992. The Conservative Party had ousted Margaret Thatcher as Prime Minister in the previous year, and Labour was expected to win. But new Prime Minister John Major managed to hold on to government, losing only 40 of its 359 seats and keeping a majority of 21 seats.
However, British elections are based on a ‘first-past-the-post’ system, and polling has become a more exact science since 1992. Of course, the cliche of polling is there’s only one poll that counts, and that’s the election itself. But the almost three years of negative polls for the Coalition is unlikely to change within a 33-day campaign, especially if the electorate has already made a decision on the performance of the incumbent government.
Governments seeking re-election always have the two-fold difficulty where they are not only being assessed on what they can offer the electorate in the future, but their
Based on these polling figures from Newspoll and Ipsos, and assuming a uniform swing, Labor would win an election, picking up between 10–18 seats, and winning between 81–90 of the 151 seats available.