THIS time, the extrapolations published in the papers are correct — if state election results followed the polling figures shown in the EMRS poll reported in the Examiner and Mercury on November 25, a hung parliament would very probably result.

Furthermore the most likely distribution of seats would indeed be 10-10-5. Indeed if the results exactly followed the seat-by-seat breakdowns given by EMRS that result would be a no-brainer as all the electorates would be straightforward 2-2-1 splits and none would be especially close.

As noted in one of my previous articles (A premature hanging) seat-by-seat breakdowns are very unreliable for several reasons, and it is encouraging to see such reasons now being regularly acknowledged in the mainstream media. If an election really returned figures of Labor 40% Liberal 36% Green 20%, however, it is extremely unlikely the Government would win a majority whatever the breakdown.

In general, Labor had been travelling reasonably (sometimes well) in the polls until this one. So does this rather sick-looking poll for the Government suddenly mean a hung parliament is overwhelmingly likely? Not necessarily, for the following reasons:

Firstly, the November EMRS survey showing a 40-36-20 distribution follows one in August showing 50-29-17 (which would almost certainly result in a landslide Labor win). Peter Tucker (Tasmanian Liberals must climb a mountain) notes that the results of EMRS polls are more prone to move about from poll to poll than those of Morgan polls. The most likely reason for this is that EMRS polls typically have a much narrower survey window, in this case two weeks. They are thus more likely to reflect short-term fluctuations in support for parties, often based on particular issues or good or bad weeks for particular parties at times when voters’ minds are not concentrated on the full range of issues. It is also possible that, within the margin of error of a few percentage points, the August poll overestimates Labor support and the November one underestimates it.

Secondly, according to the Mercury,

“About one-fifth of all voters, according to the poll, remain undecided about which way to vote. It was only when these undecideds were asked to predict which way they were leaning that the votes for the Liberals and Greens climbed close to the levels of support shown for Labor.”

An issue that remains important despite the above is the exact size of the Green vote.

This means that the figures for the Liberals and Greens include a greater proportion of soft support and therefore that Labor would actually do better and the Liberals and Greens worse than those figures suggested. Allocating the leanings of an undecided voter as a full vote to the party mentioned becomes misleading when the preferences of undecided voters are unevenly distributed, because that vote might really be (say) a 60% chance of a vote for one party, a 30% chance for a second and a 10% chance for a third. If the undecided voters are strongly favouring the Greens and Liberals then the real shape of that supposed 40-36-20 distribution might be more like 42-35-19. That is not at all likely to be a majority-winning position, but it is only a few percentage points of uniform swing short of one, assuming that the usual north-south bias in voter intention is maintained.

An issue that remains important despite the above is the exact size of the Green vote. The 1998 election showed that if the Greens miss out in most electorates then Labor does not need much lead over the Liberals to win a majority. However if the Greens win in every electorate, Labor needs to beat the Liberals by around a quota (16.7%) in each of three different electorates to have winning chances. In 2002 Labor achieved this difficult feat in every electorate, missing the extra seat in Bass (where it beat the Liberals by 17.77%) because of leakage. In some polls Labor has been well placed to win by enough in three electorates, and in some it has not. The Labor lead over the Liberals moves around greatly, but the Green vote remains consistently rather high. The article by Tucker, linked above, shows that with a distribution of 47-34-16 (a typical result of three Morgan polls in the year to August 2005) Labor would probably retain its majority. The average of the last three EMRS polls (May, August and November 2005) is 44-34-18, which would make an outright win for Labor possible, but unlikely.

Speculation about the Greens winning two seats in Denison

Not surprisingly, given that the government’s ability to win a majority will itself be an election issue (many voters will vote Labor for stability if they think it can win), a Labor source has released (Saturday Mercury November 26) selected details of Labor internal polling, including a claim that Labor would win 62% on a two-party preferred basis vs the Liberals. If the Government is really that heavily favoured it will win, but it is a rather strange statistic to release given the overwhelming likelihood that most Green votes either won’t be distributed or will be distributed only as surpluses at greatly reduced value, and furthermore a third of those that are distributed will exhaust. It would be useful to know more about this poll’s results and in particular its treatment of undecided voters. I suspect that this is yet another relatively short-window poll like the EMRS ones, and if someone does one in the first two weeks of December they might get yet another result altogether.

In all, there is still not much to get excited about either way, until we get a few polls in a row showing consistent trends rather than polls oscillating deadlock, landslide, deadlock. However this is the first poll that has shown the Government not even close to winning. The next Morgan poll, due out in early January, will be interesting.

Lastly, there has been some speculation about the Greens winning two seats in Denison following the announcement that Judy Jackson will not recontest and the preselection of Cassy O’Connor as #2 candidate for the Greens. This is premised on the fact that the Greens beat the Liberals in Denison last time and the belief that former Jackson supporters will defect to the Greens. While the Greens did indeed outpoll the Liberals in Denison in 2002, this was following a singularly awful campaign by the Liberals and there is no reason to believe the Liberals will poll so badly again. Secondly Judy Jackson’s primary vote in 2002 was a very ordinary 1551 votes, so the question that must be asked is: what supporters? (And why would they not just vote for new Labor candidates with similar views?) Thirdly beating the Liberals again is not enough; the Greens would need to increase their Denison vote to around 28% to also be in a position to beat Labor. I don’t see this happening and I think it is extremely unlikely the Greens will win two seats in any electorate.

Kevin Bonham is primarily an invertebrate research and conservation consultant, when not being inordinately lazy. A lifetime studying creatures with no spine and limited intelligence equips him well to discuss the behaviour of both voters and politicians.