Into the Swing of Things Part II
Posted by Possum Comitatus on June 16, 2007
Following on from Part One, which looked at the primary vote swing history of all governments going back to December 1985, we now move on to the Opposition primary vote swing so that we can eventually combine the two to pull out some interesting voter behaviour which has some pretty chunky implications for the forthcoming election.
The above graph measures the swing in the primary vote of the Opposition by taking the difference between the Oppositions primary vote as estimated by Newspoll each month and the Oppositions primary vote at the election previous to the monthly observation.
Some clear patterns emerge over the last 22 odd years.
There was a consistent swing away from the Opposition up to the 1990 election and there was a consistent swing to the Opposition from the 1990 to the 1996 election.
For the ALP since the 1996 election however, the pattern has been mixed. To explain this better, we need to combine the two swing graphs. To refresh the memory, here is the Government primary vote swing from Part 1.
Now lets combine those two graphs below. This might seem a bit counter intuitive to start with and be a bit hard on the brain so try and bear with me. If we take the Opposition Primary Vote swing (the red line) and put its values on the left hand side of the graph we get a the usual swings to the opposition being in the top half of the graph and swings away from the Opposition being in the bottom half of the graph just as they were in the graph at the top of the post. If we then overlay the Governments primary vote swing (the blue line) but put its values on the right hand side of the graph and invert the values, we end up with swings to the government being in the bottom half of the graph, and swings away from the government being in the top half of the graph – the opposite of what happens with the Oppositions swings. This is what we end up with.
The reason we do it this way is simple: If the two lines stay very close to each other it means that one side of politics is losing votes and the other side is picking those votes up. If however a gap appears between the two lines it means that one of two things:
1. A major party is losing votes but the minor parties are picking those votes up
2. A major party is gaining votes, but gaining them from the minor parties rather than the other major party.
Before we go any further though, one thing needs to be pointed out. The One Nation Effect that is marked in the graph had persistence beyond the 98 election because of the way the swing is calculated. Due to the Coalition getting a relatively low primary vote in 1998, when One Nation collapsed shortly thereafter its primary vote flowed back to the major parties (but mostly the Coalition) so that we see a large gap between the blue and the red lines over the 1999-2001, and where both major parties had swings TO their primary vote. That’s the One Nation effect washing out of the system.
If we go back to the lead up to the 1990 election, where the ALP got returned off the back of Green preferences, we can see how that Graham Richardson strategy worked in practice, and why he took it. In that period, both the Government (blue line, right hand axis) and the Coalition (red line, left hand axis inverted) were getting swings away from them, opening up that gap, and with the Greens and the Dems taking most of that vote. The preference deals of the ALP pulled most of that Gap back into the ALP two party preferred vote to get them over the line at the 1990 election.
If we now head along the timeline up to the last 7 years we can see some interesting dynamics at play. For those that haven’t read Part 1, I suggest you do now and come back because the accelerating swing away from government and arguments over why and how the Coalitions vote is soft become important here.
The slow but accelerating swing away from the government that started in 1999, and which has continued to snowball hasn’t been completely transferred across to the ALP vote. And likewise, big swings to the ALP primary vote haven’t completely come at the expense of the Coalition, but have come at the expense of the minor parties vote.
If we take the Latham period for instance, we clearly see in the lead up to the 04 election a large swing to the ALP, but it wasn’t matched by a similar sized swing against the Coalition (the gap opened up between the lines).What this tells us is that Latham was getting a large chunk of his headline primary vote swing at the expense of the minor parties like the Greens. It wasn’t coming from the Coalitions vote base.
But more interestingly, in terms of importance for the coming election, in early 2005, there was a large swing away from the Coalitions vote that wasn’t picked up by the ALP. This swing against the government was picked up by the minor parties. It wasn’t until Rudd became leader that these AWOL former Coalition supporters again changed their vote rapidly from the minor parties to the ALP. Rudd has also picked up a different group of voters than ex-Coalition supporters from the minor parties which can be seen at the very end of the graph where the swing to the ALP is larger than the swing away from government. These are minor party voters that are slowly switching to the ALP from their minor party homes.
This has a few implications.
Firstly, a large chunk of those voters have already deserted Howard. They deserted him under Beazley Mk II and changed their vote to the minor parties rather than to the ALP. Howard has already lost them. The idea of him getting them back is, short of the ALP and minor parties imploding, a ludicrous proposition.
These voters have already decided to vote against the government – they decided it over 2 years ago. The only question for them now is the decision where other than the Coalition to put their votes.
Secondly, apart from the 1990 election, the minor party vote grows during the election campaign, and is nearly always underestimated by the polling organisations at all times. Likewise the methods the polling organisations use to distribute preferences in the opinion polls are borderline rubbish when you compare their preference distribution estimates in the final poll of an election campaign to the actual preference flows at the election itself.
A better idea of preference flows can be ascertained by satisfaction rating dynamics, and for anyone interested I suggest they check out the ALP Victory Index to see how they work in terms of election results.
What this means is that as we get closer to the election, the usual behaviour of primary votes of the two parties converging will start to play out. But it also means that the two party preferred estimates from the polls will start to converge. That’s all fair enough and to be expected, but the underestimation of the minor party vote by the polls combined with the behaviour of that group that changed their vote from Howard to the minor parties and then again to Rudd will be underestimated in the TPP figures because of the poor preference distribution results from the polling organisations.
What this actually means is that as we approach the election I strongly suspect that the two party preferred estimates of the polling organisations will actually underestimate the ALP and overestimate the Coalition, by as much as 1.5% if the satisfaction dynamics play out like they have at previous elections in terms of real preference flows. The larger the size of the negative net satisfaction ratings, which is roughly the electorate seeing the Opposition leader being better than the Prime Minister which is calculated as:
(PM satisfaction – PM dissatisfaction) – (Opposition leader satisfaction – Opposition leader dissatisfaction)
… the larger will be the size of the underestimation of the ALP two party preferred vote by the pollsters.