Morgan and the Senate

Praise be to the Roy Morgan organisation, which has apparently been asking respondents a bonus question on Senate voting intention since mid-August and has only now chosen to unload the results. The cumulative outcome favours Labor even more heavily than their House of Representatives surveys, with the Coalition down 5 per cent from 2001 to 37 per cent and Labor up 2.4 per cent to 36.5 per cent, while the Greens are up 4.9 to 12 per cent. The big surprise is that the Australian Democrats are on 6 per cent, three times what they have managed in Morgan’s normal polling this year. What follows are Morgan’s takes on these figures, followed by the Poll Bludger’s explanations of why they’re wrong.

New South Wales: The Senate seats would most likely remain unchanged, with the L-NP winning three of the six seats up for election and the ALP winning two seats. The Greens would retain their seat.

Firstly, there is no seat for the Greens to retain. It is Aden Ridgeway of the Democrats who is up for re-election. Secondly, Morgan’s figures have the Coalition on 37 per cent, well short of what they would need to secure the third seat which Morgan predicts here (compare and contrast with their finding for Western Australia). The Coalition’s surplus over their second quota would be 9 per cent, with which they would have to fend off the combination of Family First, One Nation, the Democrats and Liberals for Forests, who have all put each other ahead of the Coalition. That would add up to at least 12 per cent on these figures. Fred Nile also favours Family First ahead of the Coalition. The real outcome would be two Coalition, two Labor and one Greens, plus one from the others – probably Family First.

Queensland: The ALP and L-NP would each have won two seats with Pauline Hanson also winning a seat, resulting in the loss of an L-NP Senator. The remaining seat going to either the Greens or the Democrats – preferences following a vote for Pauline Hanson will be important.

Again Morgan appears not to realise that it’s the Senators elected from 1998 rather than 2001 who are up for re-election. The Coalition in fact only won two seats in 1998 and thus would not "lose" a Senator on the basis of this result. If Hanson indeed wins a seat, her preferences will not be "important" as she will be elected by a narrow margin and will have only a small surplus to pass on. If she fails narrowly, making it through to the the final round as she did in 2001, her preferences will not be distributed at all. But Morgan’s figures in fact give good reason to think she won’t be elected, their assessment to the contrary being based on an apparent spike in the "independent/other" vote from 7 to 13.5 per cent in the most recent weekend’s opinion polling. Firstly, the 13.5 per cent figure would be based on a sample of about 100; secondly, it would still include support for Hetty Johnston and Family First, among others. With just about everyone putting Hanson and One Nation last on preferences, there’s not enough there for her to match the combined vote of the Democrats (7 per cent), Johnston and Family First, who are all preferencing each other. This is a hard one to pick, but the Poll Bludger’s reading of Morgan’s figures is three Labor, two Liberal and one Democrat (namely John Cherry).

Western Australia: The ALP and Liberal Party would each win two seats with the Greens also winning a seat, resulting in the loss of a Liberal Senator. The remaining seat would fall to either the ALP or the Australian Democrats.

Morgan’s Coalition vote from the past two weekends is 43.5 per cent, which is three quotas with 0.5 per cent to spare. How on earth were they able to conclude that the Liberals would not win a third seat, and that Labor, with 30 per cent of the primary vote, might? The outcome based on Morgan’s figures (and indeed on most other reasonable assessments) is as plain as plain can be – Liberal three, Labor two, Greens one.

South Australia: The Liberal Party would win two of its three seats up for re-election, the ALP would win two seats. The remaining two seats would be a contest between the ALP, Australian Democrats, Greens and an Independent – not the Liberal Party.

What independent? Let’s give them the benefit of the doubt and say they mean Family First. We then move on to the problem of their figures, the most recent of which show Labor with an unlikely lead over the Liberals of 45 to 31 per cent. In failing to predict Labor will win three seats, Morgan are either confessing that they have no confidence in their figures (and here the Poll Bludger sympathises with Morgan, as he does not begrudge them providing such figures despite the small samples), or demonstrating a failure to understand the Senate election system. If the results are accurate, Labor would win three seats and the Liberals two, with the other going to the Democrats or Greens – the "independent/other" vote of 6 per cent being too low to sustain Family First, despite this being their home state.

Tasmania: The Liberal Party would retain two seats and lose one seat, with the ALP winning two seats (unchanged) and the Greens retaining their one seat. The remaining seat would be contested between the ALP and the Greens.

One more time: this election will replace Senators elected in 1998, not 2001. The result in 1998 was Labor three and Liberal two, plus Brian Harradine. Two seats for Labor would not be "unchanged"; there is no seat for the Greens to retain; and if the Liberals "retain two seats" they will not also "lose one". The Greens’ "one seat" – that of Bob Brown – will remain no matter what happens on October 9. As for the figures, once again the Poll Bludger sympathises with the small samples and will not mock the enormous discrepancy between the results from August and those from during the campaign. But assuming the current results are correct, Morgan appears to be over-rating Labor as well as the Greens, who have been known to fantasise about winning two Tasmanian Senate seats at one election, and missing the combined effect of the mutually preferencing Democrats (4 per cent) and Family First (surely a large part of the 7.5 per cent "independent/others" and the natural inheritors of a part of Brian Harradine’s base). Bearing in mind that a preference deal between Family First and Labor has confused the situation horrendously, these figures if accurate would produce a result of two each for Labor and Liberal plus one for the Greens, with Labor, Liberal, Family First and the Democrats all in contention for the final place.

Victoria: The ALP and L-NP would have each won two seats with the Greens also winning a seat, resulting in the loss of an L-NP Senator. The remaining seat would either go to the ALP or the Australian Democrats.

Correct. Well done.

UPDATE: It appears I did not make it sufficiently clear that the above assessments are an interpretation of Roy Morgan’s figures, rather than what I believe to be the actual state of play. An insight into where the poll might have gone wrong is provided by the "analysis by House of Representatives voting intention" table at the bottom of the page, which indicates implausibly large numbers of people voting for different parties in the House and Senate. Respondents would have been asked about Senate voting intention after they had already given an answer for the House, and many would have felt compelled to nominate a different party because they did not wish to appear unsophisticated. In many cases the Democrats would have been the first name that entered their heads. Having been conducted early in the campaign, these results would also under-estimate Family First, who have been building what Lisa Simpson might reluctantly describe as "the momentum of a runaway train".

Author: William Bowe

William Bowe is a Perth-based election analyst and occasional teacher of political science. His blog, The Poll Bludger, has existed in one form or another since 2004, and is one of the most heavily trafficked websites on Australian politics.