On Saturday 15 March South Australia goes to the polls with the Labor Party struggling to win re-election after 12 years in office.
Next week I hope to publish my election site for the South Australian election. In the mean time I am publishing the electoral pendulum the ABC will be using for the 2014 election, all margins re-calculated to take account of the 2012 redistribution.
I am publishing the pendulum now to provide an explanation of why the ABC will not be using the pendulum previously published by the Electoral Districts Boundaries Commission (EDBC).
I explain below the pendulum why my calculations differ from those published by the EDBC and also why the ABC will be using my estimates.
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
The South Australian Electoral Districts Boundaries Commission is the only redistribution authority in the country that is required to take into account voting patterns and votes by party in drawing boundaries. Under provisions of the South Australian Constitution Act, the Commission must draw boundaries that are 'fair', attempting to guarantee that if one side of politics wins a majority of the vote at the next election, it will also win a majority of the seats.
All this complex boundary drawing came to nothing at the 2010 election when Labor recorded only 48.4% of the statewide two-party preferred vote, but won a clear majority with 26 of the 47 seats.
In its 2012 deliberations, the Commission chose not to radically re-draw boundaries for the 2014 election. It ruled that the 2010 result came about because of campaign, candidate and other local issues and did not reflect some underlying unfairness in the electoral boundaries.
In 2010 the Labor Party effectively 'sandbagged' its marginal seats, suffering big swings against it in safe Labor seats, but only small swings in marginal seats, even increasing its majority in several key seats completely against the statewide swing.
However, rather than fairness criteria, the point of this post is to explain why the EDBC's methodology for calculating new margins does not provide a useful starting point for election analysts like me trying to measure swing at the next election.
The EDBC's methodology may help it meet the agreed fairness criteria used at hearings for two decades, but it is not useful for someone like me setting up a computer system to analyse swing at the subsequent election. It is a lesson I've learnt over two decades of using the EDBC's margins.
For example, the EDBC has some strange factors related to demographic change that it uses to alter the margins in electorates, even when boundaries are unchanged.
For example, the state seat of Adelaide will be contested on unchanged boundaries at the 2014 election. Yet the EDBC has adjusted the margin for Adelaide. Where it had a Liberal margin of 4.2% at the 2010 election, the EDBC has adjusted the margin of 3.9%. I won't be using this adjusted margin, as when calculating swing, you should be comparing with the 2010 election, not a mythical estimate of what the 2010 result might have been in 2014.
For more than two decades I have been using the ABC election night system to analyse results booth by booth to get an accurate swing, To do that, the result in an electorate has to match the total of the polling places and declaration vote. You don't start analysing the result with a margin that does not reflect the underlying votes cast by polling place at the last election.
The second problem occurs with the method used by the EDBC to allocate the declaration vote (postal, absent, pre-poll etc) from an old electorate to new electorates. The EDBC apportions the declaration vote from the old electorate to each new electorate, taking no account of the voting patterns in the polling places transferred.
For example, two ultra-safe Labor polling places in the electorate of Light have been transferred to Napier. The Labor 2PP in both polling places was more than 70% in 2010, and you would expect the Labor 2PP in all declaration votes from that area would be roughly the same. The declaration vote in Light was 56% at the 2010 election, which means the EDBC transfers a portion of the declaration vote to Napier at 56%, not the 70% at which polling place results indicate the declaration vote should be transferred.
The EDBC has estimated the margin for Light falls from 5.4% at the 2010 election to 4.2% on the new boundaries. My estimate weighting the declaration vote produces an estimated margin of 2.8%, the value the ABC will be using.
The same problem occurs with the electorate of Bright. Labor won the seat with 50.4% of the 2PP in 2010. On the new boundaries the EDBC estimates that the Labor 2PP in the new Bright is 49.9%, making it a notional Liberal seat. My estimate of the Labor 2PP% is 50.5%.
Again, this comes about because the EDBC simply apportions the declaration vote. The area transferred to Morphett in the redistribution was covered by the West Brighton booth and voted 38.7% Labor 2PP. The area covered by the Somerton Park East booth transferred from Morphett to Bright voted 41.7% Labor 2PP. On those booth figures it makes sense that the new margin for Bright should be about the same for Labor or slightly higher than at the 2010 election, not lower as estimated by the EDBC.
I'm not the only person to have noted the discrepancy. Similar calculations carried out independently for the South Australian Parliamentary Library are very close to mine and differ from the EDBC's estimates in a similar manner.
So the pendulum the ABC will be using for the 2014 SA election still has Bright as a Labor seat. It also has Ashford, Hartley and Light as more marginal seats than calculated by the EDBC.
On the EDBC pendulum, Labor would be reduced to a minority by the loss of two seats on a swing of 1.7. On my pendulum, it is three seats on a swing of 0.6%. On the EDBC pendulum Labor has five seats on margins under 3%, on my pendulum seven.
So until my election site is published next week, I hope this pendulum helps provide a guide to the 2014 South Australian election.
Some of the more important difference in Labor marginal seat are:
Ashford, EDBC 1.5%, my estimate 0.6%
Bright, EDBC LIB 0.1%, my estimate ALP 0.5%
Elder, EDBC 1.7%, my estimate 2.0%
Hartley, EDBC 0.5%, my estimate 0.1%
Light, EDBC 4.2%, my estimate 2.8%
Mawson, EDBC 4.9%, my estimate 4.5%
Wright, EDBC 4.6%, my estimate 4.9%
Antony, slightly related in that it applies to the SA election, I received advice yesterday from the electoral commission that the revised nomination fee of 3000 dollars will apply not only to legislative counsel candidate but also to all house of assembly candidates. The other thing they confirmed was that the new requirements for 250 signatures for LC does not apply to registered political parties.
COMMENT: I think the application of the higher deposit to the Assembly as well as the Council might have been a drafting error. The fact parties do not need to use nominators is exactly the same as in every other electoral act in the country. I think that nominators should be re-introduced for party Senate nominations under the Commonwealth Electoral Act.
Posted by: Jonathan pietzsch | January 22, 2014 at 01:11 AM
If the state desired that the party with the most votes win the most seats, was there ever any thought put towards a system other than single member electorates?
COMMENT: It has been discussed but not with much serious intent.
Posted by: Mark Roth | January 22, 2014 at 08:46 PM
So, in the SACA, they explicitly use the word "fair"? Considering that is a highly subjective term with absolutely no definitive rules, outlines or method to coming to a conclusion that is fair but not biased, who has final authority on what is considered fair at any one time?
This person would then go and vote at the election in the boundaries they have set up for a political party they prefer. Of course it's going to be biased.
The results will be biased if there is an ambiguous term used to define laws. Period.
Anyway, the question is: as a resident of SA, I find this rather distressing, are there any plans in place by either party to combat this?
COMMENT: I would point you to Section 83 of the constitution act which attempts to set out a measure of what 'fair' means. However, it struggles as a definition not because 'fairness' is not defined, but because there is dispute over what is meant by 'groups of candidates' and also because the application of fairness has to be about something happening in the future, and as we saw in 2010, you can't always predict future voting patterns.
There appear to be no plans to remove this provision from the constitution.
Posted by: Nick | January 23, 2014 at 03:43 PM
Antony,
I find it extraordinary that the SA Electoral commission could so flagrantly violate the state constitution.
This issue as to whether the constitutional amendments passed in 1990 were wise is a separate matter. The fact is that they were passed and the intention was clearly that a group of candidates with 50% of the vote should win a projected 50% of the seats based on a uniform swing from the previous election. Isn't his the reason redistributions are now conducted in SA every term?
The commission argued that the boundaries were fair and that the result in 2010 was due to superior Labor campaigning in marginal seats. That is an opinion only and I can't understand how an opinion by appointed commissioners can outweigh the state constitution.
A side note: Even as an opinion it seems about as sound as the seat margin calculations that you mention above. My opinion is that there is no gerrymander in SA but a natural bias to Labor based on the state's geography, namely the Liberal vote being locked up in safe country seats, where as Labor's vote is more evenly spread (but dominant) in Adelaide, where the bulk of the seats are located. On your table above there are 15 ALP seats with margins of under 10% and only 4 Liberal. On the other hand, the Liberals have 14 seats +10% vs only 11 for Labor. The Liberals have won the 2PP in 5 of the last 6 elections, but have held government only twice. This looks structural.
COMMENT: 'Flagrant violation' is hyperbole. If the Commission was flagrantly violating the state constitution as you suggest then it would be a relatively easy legal case to force them to do otherwise. Your paraphrasing of the both the constitution and the Commission's argument over-simplify arguments that fill many volumes of Commission reports.
You're arguments about the Liberal Party's problem with over-concentration of support is correct, but the only way to fix it using single member electorates is to abandon any idea that electorates have communities of interest. The Commission went down this path with the first redistribution under the fairness criteria in 1991 by putting Kangaroo Island in Flinders and there was outrage.
On uniform swings, the 2010 election had the lest uniform swing I have ever seen. My election website to be launched next week has graphs of the two-party vote in each electorate going back three decades. The seats Labor clung on to to win office in 2010 were completely at odds with previous voting patterns. The results in Light and Mawson in particular were very much at odds with previous voting patterns.
If you draw boundaries to deal with a bias issue, should you pay attention to long established patterns or do you ignore those and only draw boundaries based on one election? The Constitution doesn't specify and I suspect the answer of participants on which approach to take may change from redistribution to redistribution depending on what they think is in their own self-interest.
Posted by: Peterjk23 | January 25, 2014 at 01:10 AM
Hi,
A simple question (hopefully) in terms of definition of swing and how it is effected by the redistribution of the seats done by the commission. What is the starting point for the swing - is it the last election result or the nominal 50:50 position created by the Commission.
For example, recent 2PP polling has the Liberal Party leading 53:47. If that were replicated in the election would that be considered a 3% swing from the nominal 50:50 starting position or a 1.4% swing based on the actual 51.6:48.4 result of the last election?
COMMENT: The swing would be 1.4%.
The issue the Commission has traditionally tried to deal with is matching the swing in marginal seats required to deliver a majority of seats.
When you hear a swing quoted for a party to lose or form government, it is calculated by assuming a uniform swing and counting the seats gained and reading off the swing required. So Labor holds 26 seats, if it loses three seats it loses its majority, and those three seats correspond to a swing of 0.6% on my figures. So from a starting point of 51.6% Liberal 2PP in 2010, and assuming a uniform swing, then the Liberal Party need a statewide vote of 52.2% just deprive Labor of its majority.
The Liberal Party argued the boundaries should have been drawn so that the Liberal Party had a notional majority of seats going into the 2014 election and Labor needed to increase its vote by 1.6% to get back into government.
The current situation has arisen because in 2006 Labor polled 56.8% of the 2-party vote. After the subsequent redistribution, the pendulum indicated that a 6.9% swing against Labor would have delivered the Liberal Party a majority of the 2-party vote, and deprived Labor of seven seats and its majority. Instead, the swing was a larger 8.4%, but Labor lost only two of the seven seats that were supposed to fall, plus Adelaide that fell on a larger swing. The 2010 result was an astonishingly non-uniform swing, Labor winning because despite the loss of state wide vote, it didn't lose the same proportion of vote in the marginal seats it was fighting to retain.
Posted by: Dominic | January 30, 2014 at 05:16 PM