Hanging loose

Earth to journalists: a hung parliament is not the most likely outcome

RECENTLY I wrote an item for this website with the title Lies, Damn Lies and Opinion Polls in which I maintained that political opinion polling had to be interpreted in the context of its mathematical and procedural limitations.

I wrote that as a response to media reaction to a November 2005 EMRS opinion poll on Tasmanian voting intentions which showed that support for Labor had dipped to 40 per cent, while the Liberals and the Greens had climbed to 36 and 20 per cent respectively (all after excluding undecided voters).

There were breathless predictions in the press (and this website) that the Government was on the skids, the Liberals would gain at least 10 seats and that a hung parliament at the 2006 election was close to a certainty.

My point then was that one poll a summer does not make. The variables and margin for error inherent in surveying methodology means that only the courageous (foolish?) would make predictions based on just one poll.

Especially when the poll in question contradicted the trend that that polling company’s three previous polls had been indicating: for 2005 a strengthening for Labor and the Greens and a softening of the Liberal vote, as the following table shows.

EMRS polls for 2005 – First preferences (undecides and others excluded)
	Feb	May	August	Nov
Labor	45	45	52	41
Liberal	40	37	30	36
Green	15	18	18	21

So had voters, in a fit of pre-Christmas pique, suddenly abandoned Labor and thrown their lot behind the opposition parties, or was the November poll a statistical glitch?

We need to turn to the other firm that regularly polls Tasmanian voters, Morgan Polls, for some guidance. Morgan released its Latest Survey on Christmas Eve. It showed that for December Labor was on 45.5 per cent, the Liberals on 27.5 and the Greens 18. To give a direct comparison to the EMRS results where undecides are excluded, that translates to 49 Labor, 29 Liberal and 19 Green.

Compare that outcome with, firstly, the EMRS result for November and then for August. Can voter sentiment really chop and change that much from month to month?

My view is that the Tasmanian electorate is not swinging crazily. I am not prepared to say categorically the November EMRS poll is wrong; rather that it is not supported by the evidence of other polling.

If we look at the Morgan results since the 2002 election (below) do we some support for my stance.

The graph shows that during 2005 support for Labor and the Greens rose slightly, while support for the Liberals softened. This is much the pattern that EMRS was giving us before the “rogue” November poll.

My conclusion is that support for Labor has dropped since the 2002 election, and appears to now sit in the mid to high forties. Notwithstanding, that represents a slightly strengthening position by the end of 2005. The Liberals, after an improvement late 2004-early 2005, may have dropped back by the end of 2005 to be nearer to their 2002 level of support, while the Greens continue to poll strongly at above a quota.

Finally, what does this mean in terms of seats? In an Earlier Paper I used the Morgan poll results for August to “construct” a hypothetical election and concluded if an election was held in that month Labor would have returned 12 seats, the Liberals eight and the Greens two, with three seats doubtful. That translated into 87 per cent probability of a Labor majority and a 13 per cent probability of a hung parliament.

Using the same methodology with Morgan’s December poll I now find the following:

December 2005 hypothetical election: based on Morgan Poll results
		Labor	Liberal	Green	Other	Doubt-ful
Bass	1st Pref %	45.5	33.5	16	3
	Quotas	2.72	2.00	.96	.18
	Seats	2	2	1	0	0
Braddon	1st Pref %	49.5	33.5	11	3
	Quotas	2.96	2.12	.65	.18
	Seats	3	2	0	0	0
Lyons	1st Pref %	51.5	28.5	16	3
	Quotas	3.08	1.71	.96	.18
	Seats	3	1	1	0	0
Denison	1st Pref %	47.5	25.5	23	3
	Quotas	2.84	1.53	1.38	.18
	Seats	3	1	1	0	0
Franklin	1st Pref %	48.5	26.5	19	3
	Quotas	2.90	1.59	1.14	.18
	Seats	3	1	1	0	0
Total seats		14	7	4	0	0

My analysis shows that if an election was held in December, and swings were assigned to each electorate uniformly, then the result would be exactly the same parliament as we have now: a Labor majority government.

I stress that this analysis is educated speculation. I am not saying the above or any result will occur; rather that this is my best guess casting a jaundiced eye to the opinion polls.

The Hare Clark electoral system contains too many ifs, buts and maybes to make predicting elections anything other than a fool’s game — so call me a fool. But I do think those who say anything like “the polls are indicating a hung parliament” are bigger fools than I.

Certainly a hung parliament is a possibility — it usually is under Hare Clark — but there is no way it is the most likely outcome on any considered reading of the available opinion polls.

Peter Tucker is a PhD candidate at the University of Tasmania’s School of government, and a past advisor to the Tasmanian Liberals in both government and opposition.