Today’s Advertiser carries two interesting electorate level polls with impressive samples of around 550 voters, which provide support for two items of conventional wisdom about the coming South Australian election that the Rann government will win a handsome majority, and that quirky independent MP Peter Lewis is gone for all money in Hammond. The first proposition is backed by a poll of voters in Stuart (top), a vast electorate covering the eastern part of the state in which Labor are competitive thanks to their strength in Port Augusta. It has Labor’s Justin Jarvis on 47 per cent of the decided primary vote against 42 per cent for Liberal veteran Graham Gunn (veteran seems almost an understatement for a man who entered parliament in 1970) for a lead of 52-48 on two-party preferred.
The Hammond poll (bottom) backs up Malcolm Mackerras’s assertion in Saturday’s Australian (where he predicted that Labor would pick up six seats overall) that Peter Lewis would "get done like a dinner". It shows Lewis in distant third place on 13 per cent of the decided vote compared with 48 per cent for Liberal candidate Adrian Pederick and 25 per cent for Labor (who unveiled James Peikert as their candidate earlier this week). If these figures are accurate, the best preference flow in the world would not be enough to boost Lewis to second place ahead of Labor, and even if it did the Liberal vote is close enough to 50 per cent to assure Pederick of victory. It also suggests that a majority of the voters abandoning Lewis are heading for Labor rather than Liberal together with the Stuart results and Saturday’s Advertiser poll, this undercuts the notion that the Labor swing will be confined to Adelaide.
Bass | Braddon | Denison | Franklin | Lyons | Total | |
Labor | 39 | 41 | 35 | 37 | 48 | 42 |
Liberal | 38 | 42 | 28 | 29 | 30 | 32 |
Greens | 20 | 10 | 36 | 24 | 18 | 22 |
Also just to hand is a detailed breakdown of the EMRS poll for the Tasmanian election (above) which was published in the Mercury on Saturday. There are a number of reasons why these results should be treated with caution. While the total sample of 1002 is substantial, the margin of error blows out significantly when it comes to the seat-by-seat breakdowns. Even more troubling is the extremely high undecided rating (23 per cent in Saturday’s poll) that is a consistent feature of EMRS polling, which suggests they are making no effort to twist the arms of voters reluctant to declare a preference (larger polling agencies ask undecided voters who they are "leaning towards", and they usually get an answer). The following table, showing the course of aggregate EMRS polling over the past year, is probably more useful.