Aggregated results from the last two weeks show that Labor’s recent weakness in Newspoll has been driven by dire results from New South Wales.
Today’s Australian brings state and demographic breakdowns from the combined results of the Newspoll surveys of August 9-11 and August 16-18, which respectively came in at 52-48 and 54-46 in favour of the Coalition. The overall sample is 2826 respondents, with sample sizes for each state ranging from 458 to 659. The narrowness of the range suggests the super-sized sample in this week’s poll was used to boost the numbers from the smaller states, by way of reducing the margins of error on today’s state breakdowns, the largest of which is 4.6%. The salient points:
• New South Wales looks to have done the damage in Labor’s weak ratings of late, the published two-party preferred coming in at 57-43. As you can see from the sidebar, this is a fair bit worse for Labor than the published and unpublished state-level numbers from other pollsters which have been used to determine the current BludgerTrack results.
• Victoria on the hand swings heavily the other way, a 54-46 lead for Labor suggesting only a swing to the Coalition of a little over 1%. This includes a 17% result for the Greens which most would consider a bit hard to credit, given the 12.7% result from 2010 and the general trend of the party’s fortunes.
• The numbers show Labor looking alive in all-important Queensland, a 53-47 lead to the Liberal National Party implying a swing to Labor of around 2%.
• The Western Australian results on the other hand paint a very different picture from one that has long seemed overly favourable to Labor in BludgerTrack. The two-party result is 59-41, implying a swing to the Coalition of around 2.5% off an already very high base. It should be noted though that it’s around here that the margins of error start to push north of 4%.
• A 54-46 lead to the Coalition in South Australia is in line with talk that Labor should be concerned about Hindmarsh and perhaps one or two other seats in the state, suggesting as it does a swing of about 7%.
• Personal ratings don’t show a huge amount of interstate variation for Kevin Rudd, with Victoria being effectively even with his home state for his best net approval rating. His approval rating is higher among men (39%) than women (35%).
• Tony Abbott on the other hand rates considerably lower in Victoria (a net approval of minus 20%) than in New South Wales and Queensland (minus 5%).
I’ll be running all that through the BludgerTrack updatermator later today. You can view the full tables on voting intention here. You can also view aggregated state breakdowns for Essential Research here if you’re a Crikey subscriber, as you should be.
UPDATE: The Guardian has a Lonergan poll of Kevin Rudd’s seat of Griffith which is raising a few eyebrows by showing his Liberal National Party opponent Bill Glasson leading 52-48, from primary votes of 38% for Rudd (down six on 2010), 47% for Glasson (up 11% on the LNP vote in 2013) and 11% for the Greens (down four). However, it’s well worth pointing out that Lonergan’s own blog reprints an article from Adrian Beaumont at The Conversation which suggests we “trust the national polls much more than the marginal seat polls because the national polls have a good track record at predicting elections, while the robopolls are fairly new”.