The West Australian has published another small-sample Patterson Market Research-Westpoll survey (401 respondents) to follow on the poll of June 12, which had the federal Coalition with a gaping two-party lead in WA of 62-38. The newer poll paints a much rosier picture for Labor, who are up 8 per cent on the primary vote to 36 per cent and have narrowed the two-party deficit to 54.5-45.5. This would mean a 1.2 per cent swing to the Coalition, which would only threaten Labor in Hasluck and leave them well clear in their other three seats. In contrast to every other poll since the leadership change, this one shows Labor’s gains coming at the expense of the Coalition, who are down seven points on the primary vote to 49 per cent. The Greens are steady on 9 per cent, but the result in the earlier poll did not square with last week’s Newspoll quarterly geographic breakdown which had it at 16 per cent. The Nielsen survey of late last week included a sub-sample of 100 Western Australian voters, which had the Coalition on 50 per cent, Labor on 42 per cent and the Greens on 5 per cent.

UPDATE: Roy Morgan throws a curve ball: a phone poll of 600 respondents conducted between Friday and Monday which has the Coalition leading 51.5-48.5 on two-party, and 45.5 per cent to 38.5 per cent on the primary vote (with the Greens on 9 per cent). It should be stressed that this is a phone poll as distinct from the weekend face-to-face surveys Morgan usually publishes on Fridays, which are the most Labor-leaning in the business. The results of this poll and the one from Friday should thus not be compared, though the Morgan press release does just that. The last Morgan phone poll was conducted May 26-67, and had Labor at 37.5 per cent on primary, the Coalition on 43 per cent and the Greens on 11.5 per cent, with two-party on 50-50. The margin of error on the poll is about 4 per cent. For those confused by this apparently aberrant result, Possum offers the clarification that “exogenous shocks have a large random component to the resultant impulse response function”.

UPDATE 2: Julia Gillard’s atheism having emerged as an issue, I thought I’d crunch some Australian Election Study survey data on church attendance and voting behaviour, as there have been suggestions Labor will suffer the loss of Christian voters attracted by Kevin Rudd. Defining church attenders as those who go at least once a year and everyone else as non-attenders, 2007 was unusual out of elections going back to 1993 for the narrow gap between the Coalition church attender vote and the total Coalition vote – 2.6 per cent, whereas in other years it had ranged from 5.5 per cent to 7.5 per cent. However, the Labor vote was unexceptional: 1.0 per cent lower for church-attenders than the Labor vote overall, in keeping with an overall range from 3.9 per cent lower to 0.3 per cent higher.