Menu Close

Making sense of the polls

Victorian Election Preview

The Victorian election will be held in just one week. While I have written about the Victorian election previously, I have yet to do a full preview post. This post includes some material from previous articles.

The Victorian lower house elects 88 members representing single member electorates, using the same compulsory preferential voting system that is used for the Federal House of Representatives. At the 2010 State election, the Coalition won the statewide Two Party Preferred (2PP) vote by 51.6-48.4, and won a bare majority of 45 seats to Labor’s 43. During the last term, Liberal Geoff Shaw became an Independent, and held the balance of power in Parliament.

Since the 2010 election, there has been a redistribution of electoral boundaries. As a result of this redistribution, three seats Labor won in 2010 are notionally Coalition held, so the Coalition starts with a 48-40 notional majority. While Labor needs to win more seats after the redistribution, the swing required to win is only 0.9%, implying that Labor could win with a 2PP of only 49.3%.

However, it would be unwise to take this uniform swing calculation literally. This is because MPs will generally have a personal vote. When a party makes many gains from the other party’s sitting members, as the Coalition did at the last election, the on-paper margins of new members will be bolstered by a double personal vote effect: their own personal vote and the lost vote of the former member. Peter Brent and Kevin Bonham have both modelled the effects of personal votes, and both think that Labor will actually need about 50.6% 2PP to win the election.

Polling this term has had Labor ahead since the initial honeymoon for the Baillieu government wore off, other than a brief Coalition lead when Denis Napthine became Premier. If anything, polling during the campaign has shown a slight move to Labor, and Labor now leads by 53.0-47.0 in both the Poll Bludger’s Victorian BludgerTrack and Kevin Bonham’s Victorian aggregate.

Yesterday, two polls were released. A Galaxy poll has Labor leading 52-48, unchanged from late October. Primary votes were 40% for the Coalition, 39% for Labor and 13% for the Greens, the only change being a 1% gain for Labor. If a 2PP is calculated from these primary votes using last election preferences, we would have a Labor 2PP of 52.7%, suggesting that this poll was rounded towards the Coalition. This poll was conducted on Tuesday and Wednesday nights from a sample of 920.

However, a Morgan SMS poll conducted on Wednesday and Thursday nights from a sample of 1040 gives Labor a 55-45 lead, a 1.5% gain for Labor since a Morgan taken on the 7-10 November. Primary votes are 35.5% for Labor (down 0.5), 35% for the Coalition (down 3) and a very difficult to believe 19.5% for the Greens (up 1). I think that Morgan’s methodology is greatly overestimating the Greens support, and it is more reasonable to think they are on the Galaxy figure of about 13%.

Using the 2010 State election preference flows is probably underestimating Labor’s actual 2PP from given primary votes. At the 2010 election, Federal and state Labor governments were in office, and after 11 years the state Labor government had become unpopular with environmental and social left wing voters, so Greens preferences did not flow as strongly to Labor as they did in last year’s Federal election. At the upcoming state election, with state and Federal Coalition governments, I think Labor will benefit more from Greens preferences than they did at the 2010 election, especially given the level of Greens’ disdain for Tony Abbott.

The major battlegrounds at this election will be the so-called “sandbelt” seats in Melbourne’s southeast. Frankston, Carrum, Bentleigh and Mordialloc were all taken by the Coalition in 2010, and Labor probably needs to win two of them back to be confident of victory. Another seat to watch is Ripon, where a notional Liberal margin of 1.5% will be further bolstered by the retirement of the 15-year Labor sitting member. Given the 4.6% 2PP swing to Labor since the 2010 election in BludgerTrack, Labor should not have any difficulty retaining other seats that are notionally Coalition held.

Another question is whether the Greens can win lower house seats. In 2010, the Coalition’s decision to put the Greens behind Labor on their How to Vote cards was critical in allowing Labor to hold onto the inner city seats of Melbourne, Richmond and Brunswick. The Coalition are following this policy again in 2014, so the Greens will need to increase their primary vote in these seats at the expense of Labor to win. Most statewide polling has Labor and the Greens both increasing their vote, but the picture could be different in the inner city. Labor would absolutely not want to rely on Greens support in the lower house.

Victoria elects its upper house through Senate style proportional representation, using eight regions with each region electing five members. Group voting tickets for the upper house were lodged last Sunday 16 November, and the biggest surprise is probably the Greens’ decision to put PUP ahead of Labor in four regions. For their part, Labor has non-Left parties ahead of the Greens in four regions.

The cynicism of these deals shows that the group voting ticket system currently used for the Senate and Victorian, SA and WA upper houses should be abolished. Fortunately, Victoria only requires voters to number 5 boxes below the line to cast a formal vote. I recommend that Victorians control their own preferences by voting below the line, rather than let parties control their preferences.

In Victoria, Federal Labor leads by 57-43 according to BludgerTrack. With only a week until the election, it is unlikely that the State Coalition can recover much ground given the Federal Coalition’s unpopularity in Victoria. Labor should win a comfortable lower house majority. In the upper house, Labor and the Greens should have a combined majority, but preference deals could mess up that scenario.

Want to write?

Write an article and join a growing community of more than 165,200 academics and researchers from 4,637 institutions.

Register now