The Adelaide Advertiser did an interesting telephone poll  of the South Australian Labor held marginal seat of Kingston on Wednesday night – and they deserve a round of applause for  both spending the resources on it and undertaking the poll in fairly robust manner. One of the problems we often have with some of these individual seat polls undertaken by non-pollsters is that they often have a ridiculously small sample size on the one hand, and aren’t demographically weighted properly on the other– but the Tiser folks did the hard yards on this, giving us a sample of 605 and weighting the respondents by age and gender – so, a little golf clap please!

A sample of 605 gives us a margin of error that maxes out around the 4% mark – which is pretty good for an individual seat poll. The top-line results are almost unbelievable, coming in like this:

If a Federal election for the House of Representatives (or Lower House) were held tomorrow, which party or candidate would receive your first preference?


kingstonvotes


It suggests that Amanda Rishworth, the incumbent Labor member, is getting a nearly 13% swing towards her on the two party preferred, and an 11% swing towards her on the primaries (which have been rounded up to the nearest whole number).

We see the same large gender gaps coming out that we’ve been seeing in all the polls of late, with the female/male split on the primary vote for Rishworth coming in at 61/55, and 24/27 for the local Liberal Party candidate Chris Zanker. The Advertsier also asked a few other questions, the first of which was on the Better PM:

Regardless of the Party they represent, who do you think is better suited to be Prime Minister?

The results came in 68/22 for Gillard over Abbott, with the women running 73/17 for Gillard and men coming in at 62/28 for Gillard. Also interesting was that 26% of Liberal voters preferred Gillard over Abbott.

On asylum seekers, we had:

Which Party, Labor or Liberal, do you think is best equipped to handle the asylum seeker issue?”

The results here came in with the ALP leading  44/34 among all voters, 43/39 among men and 46/29 among women.

Finally, the Tiser also asked how people voted at the last election and broke the results down by voting intention in the poll itself. So we have things results like 84% of respondents that stated they would vote for Labor if an election were held tomorrow also state they voted for Labor in 2007, and 38% of people that stated they voted Green last election said they would vote for Labor if an election were held tomorrow. There’s a full spread of that data, so using a bit of back engineering, the results of the 2007 election came out as a two party preferred of 58/42 to Labor (give or take a point), which is about 4 points overcooked for the ALP compared to the 2007 result of 54.4%.

What makes this result interesting is that it’s about the average level of overestimation that occurs when you ask people how they voted at the last election, after the outcome of the event is known. Generally, when a question like this is asked, the person who actually won the election ends up getting an overestimation of around  5% – people like backing winners with hindsight I suppose. It’s something we see pop up regularly in things like the Australian Election Study.

This makes me wonder just how inflated the poll result might be for Labor. On the one hand, the poll has a big chunk of the WTF! factor going on – 67/33 is an enormous lead for Labor by any definition of the word. On the other hand, if it were heavily overcooked for Labor, we’d also expect to see a much higher vote coming out in the poll results on the question of how respondents voted for Labor at the 2007 election.

But we don’t, respondents said they voted at the 2007 election at levels that are pretty much in the ballpark of where we would expect to find the responses – with 5% overestimation thereabouts.

The other thing here worth noting is that South Australia was the State where Rudd was generally getting the largest swing towards him for most of the period of his large poll leads before October 2009. If we use the Newspoll quarterly demographic breakdowns to look at the two party preferred swing Labor was receiving at the state level up to the end of 2009, we can see South Australia at the top of the pack (click to expand):

earlyswing

If a state was going to go berko for Labor at the election, SA would be the place where the last few years of polling evidence provides the most support – but this lead in Kingston is huge, probably abnormally huge – yet the data consistency in the poll is as we would expect it to be if this lead was actually true, which makes it rather extraordinary.

Yet, even if it is overcooked a few points the Labor – a 57/43 result is still bad for the Coalition, let alone a 60/40 or, as the headline results here suggest, 67/33. Andrew Southcott in neighbouring Boothby and Christopher Pyne up the road in Sturt must be getting pretty nervous about now.

UPDATE:

Peter Brent (of Mumble fame) via twitter tells us:

A ‘Tiser Kingston poll at same point in 2007 campaign was pretty on the mark http://bit.ly/9RrTLt

(Visited 21 times, 1 visits today)