There has been much speculation over the winner of the fifth and final seat in the North Metropolitan Region of the Victorian 2010 Legislative Council election.
Most of the speculation was due to the ABC, which provides an election calculator that showed the fifth spot was likely to go to Crikey founder Stephen Mayne, who received 1% of the vote and then looked to be riding favourable prefences from other parties to the required quota of 16.7%.
The closed-sourced ABC calculator however, has several limitations, such as poor handling of ungrouped candidates and also ignoring leakage of preferences by below-the-line voters. It does format the results very nicely though (I'm stilling working on that)
So on Wednesday night, I took the current results from the VEC results website and put it through pycassandra, using below-the-line predictions provided by Stephen Mayne himself. The results are a bit different. Due to preference leakage, especially from Joanne Stuart in Group A, Mayne can not quite overhaul the final Liberal Candidate, and so pycassandra predicts the final makeup to be 2 ALP, 2 Liberal, 1 Green.
This morning, the Victorian Electorial Commission (VEC) began a recount of the votes, which means the tallies may change.
The Count:
Elected MIKAKOS, Jenny of Australian Labor Party from quota
Elected GUY, Matthew of Liberal Party from quota
Elected ELASMAR, Nazih of Australian Labor Party from quota
Elected BARBER, Greg of Greens from quota
At this point, no candidates had quota, so started eliminating
Eliminated 19 very low ranking candidates
Eliminated WHITEHEAD, Adrian of Whitehead (quota: 0.0054)
Preferences: 61.4583% flows to BHATHAL, Alexandra Kaur (Greens), 19.7917% flows to MURPHY, Nathan (Australian Labor Party), 9.375% flows to ONDARCHIE, Craig (Liberal Party), 9.375% flows to MAYNE, Stephen (Group C),
This Round: (candidate | group | quota)
ONDARCHIE, Craig | Liberal Party | 0.6543
MURPHY, Nathan | Australian Labor Party | 0.5937
PATTEN, Fiona | Australian Sex Party | 0.2029
KAVANAGH, John | Democratic Labor Party | 0.1665
CONLON, Andrew | Family First | 0.136
BHATHAL, Alexandra Kaur | Greens | 0.0754
MAYNE, Stephen | Group C | 0.0623
JANSON, Vickie | Christian Democratic Party | 0.0495
STUART, Joanne | Group A | 0.0304
ARCHIBALD, Kevin | Country Alliance | 0.0287
The rest of the count is after the jump...
Stephen Mayne's chances for the final spot in the North Metropolitan Region (Victorian 2010 Legislative Council election) Thursday, December 2, 2010
Election 2010 Tuesday, September 7, 2010
Ironically I was doing other stuff during the election period rather than focusing on the Senate. Basically, the place split as expected.
The most interesting results were in Victoria, Tasmania and the ACT.
In Victoria the Democratic Labor Party (DLP) won a seat from Family First, meaning the Coalition only won 2 seats in the state. In Tasmania the ALP/Greens won 4 seats, leaving the Coalition with only 2. The other states split 50/50 left/right. And in the ACT, the Liberal's candidate was almost pushed to preferences, setting up a fight with the Greens in the next election.
The Sex Party, Liberal Democratic Party and the Shooters Party all had decent swings to them. The Sex Party, the political arm of the Eros Foundation, in their first Senate election, ran a highly visible and well funded (well, it was funded!) campaign. The Liberal Democratic Party was probably the beneficiary of being a stable entry with a solid niche (the libertarian vote) as well as a name change from the Liberty and Democracy Party to the much more popular Liberal Democrats name ... containing not only the word Liberal, but the word Democrat and also the popular UK party name in the title.
The Senate now swings from left to right in July next year. This means the Coalition only have 7 months to annoy the government until the Greens get their turn.
Quotas are too high in the Senate.
Kernot likely to take seat from Labor or Coalition - not the Greens Friday, July 30, 2010
Can Cheryl Kernot, or indeed any micro-party or independent candidate, win a Senate seat in New South Wales (NSW) this election? Based on my simulations, it does not look like it. However if Kernot can get about 5% of the vote she will have a chance to take a seat from Labor or the Coalition, meaning that we might see both an Australians Greens senator and an independent returned from NSW - an amazing result for a state which split 3 Labor 3 Coalition at the last election.
Predicting a senate result is hard. However, we now know several things about the contest on 21 August.
- We have a list of candidates.
- We know the ALP will get roughly 39% of the vote.
- We know the Greens will get roughly 12% of the vote.
- We know "roughly" means probably a margin of error of about 3.5%.
- We know that the ALP is preferencing the Greens above all other parties.
- We know (from my previous work) that randomly allocating preferences returns mostly the same result as a real election with preferences allocated by the parties.
- And finally, we have results from 2007 for a host of micro parties such as the Christian Democratic Party (CDP).
Here's a graph of the results:
So what is happening in this graph?
[more after the jump]
Greens to win two Senate seats in Tasmania and other quarterly Newspoll results Wednesday, July 21, 2010
A collapse in the Labor primary vote would crush Labor in the Senate and give the Greens a serious chance of winning two seats in Tasmania if the results of the most recent quarterly Newspoll figures were repeated at the forthcoming federal election on 21 August 2010. The Liberals would stage a dramatic turnaround, beating their 2007 result after spending most of the last two years in the doldrums.
Despite these wild swings in the number of seats won in the Senate, the effective result has not changed. The Coalition will lose its blocking power in the Senate and the Greens will gain control of the cross benches, giving them balance of power along with Labor and the Coalition.
Read more after the jump...Virtual Bridge Walk for Reconciliation Thursday, May 27, 2010
To commemorate the tenth anniversary of the Reconciliation Walk across Sydney Harbour Bridge, Australians for Native Title and Reconciliation (ANTaR) commissioned me to build a virtual bridge walk as part of their "Are We There Yet?" campaign.
The website is written in python and uses the django web framework. It has some features of which I am particularly proud. The avatar-building component provides over 30,000 different combinations. The puzzle page provides each user a unique image composed of over 350 different avatars. I use inkscape + PIL to composite avatars from about 30 base images. To avoid crushing browsers on low end computers, I used (abused?) image maps to create the puzzle. Overall the result is a very solid, easy-to-use project with solid features that had a fast turn around and low cost.
My own personal idea of the Reconciliation is that white and black Australia needs to agree on what happened in the past before we can move forward. Only by "reconciling" the two different views of history we seem to have in this country (settlement vs invasion) can we chart a way forward.
Coalition out of danger zone in Western Australia Thursday, April 22, 2010
Newspoll has released its quarterly detailed breakdowns of people's voting intentions, including state-by-state breakdowns of primary votes. This allows us to extrapolate an interesting state-by-state swing to or from the major political parties, the Australian Labor Party, the Australian Liberal Party and the Australian Greens. Applying this swing to the 2007 Senate election results and running a batch of simulations using pyapollo we can see what trends are likely to be apparent in the Federal election later this year.
If an election was held today, the Senate results would possibly be Labor 19, Liberals 15, Greens 5, Independent 1. If Independent Nick Xenophon does not stand a friendly candidate and sits out the 2010 election (which is likely), the Greens are favourites to win the final seat in South Australia.
SA 2010: Coulda, Shoulda, Woulda Edition Tuesday, March 30, 2010
The final result for the South Australian (SA) 2010 Legislative Council election appears to be beyond doubt.
The split looks to be Liberals 4, Labor 4, Greens 1, Family First (FF) 1 and Dignity for Disability (D4D) 1.
My prediction was Liberals 5, Labor 4, Greens 1, Family First 1. The Liberals did slightly worse than I expected, falling 2% short of the final seat on about 6.5% of the vote at the last count (the quota is 8.33%). I think this near miss is not bad for my very first attempt at predicting a result!
The winners look to be (from the ABC):
David Ridgway (Libs), Paul Holloway (ALP), Stephen Wade (Libs), Gail Gago (ALP), Terry Stephens (Libs), Bernard Finnigan (ALP), Jing Lee (Libs), John Gazzola (ALP), Tammy Jennings (Greens), Robert Brokenshite (FF) and, excitingly, Kelly Vincent (D4D)
Rita Bouras (the fifth candidate for the Liberals) was not exactly unlucky to miss out, but I am going to say she came "12th" in the election (there are only 11 seats available).
As a follow up I have plugged the results into apollo and, now we have actual results and preference tickets, tried a few different scenarios to see what might have been. I hope you enjoy these "What If" scenarios.
SA 2010: Overview and Predictions Thursday, March 18, 2010
Tomorrow is the 2010 South Australian election. My coverage here has been exclusively on the South Australian upper house, the Legislative Council.
Overview
Blog posts I written about #sa2010 are:
- initial overview
- what vote a minor party needs to win a seat
- analysis of the group voting tickets - which parties did best (and worst)
- simlulation using the GVT and newspoll and galaxy
- batch simulations teasing out the micro-party independent field 1 2
I predict that the result from the Legislative Council election tomorrow will be: Liberals 5, Labor 4, Greens 1 and Family First 1.
However, I would not be surprised if the Liberals only get 4, with the extra seat going to the Australian Greens, Dignity for Disability or David Winderlich.
Good luck to all the candidates and their workers - your hard-work, often unrewarded, keeps this country great!
Now let's never mention SA again.
SA 2010: If all micros are created equal ...
The other day I ran a batch of simulations based on my own impressions of the South Australian Legislative Council Election. Basically I gave the micro parties and independents a slightly higher vote if I had heard of them (I live outside of SA) and a lower vote if I had not. The big winner was David Winderlich, who on 1% of the vote rode preferences from Dignity for Disability and the others to win a seat in nearly 40% of outcomes.
But what if all the smaller parties started with the same basic vote? Would Winderlich still outlast Dignity for Disability? Just who would rise to the top?
So I reset all the micros and independents to 0.5% of the vote and ran 20,000 simulations with the votes varied by normal distribution each simulation.
The big winners in this scenario are Dignity for Disability and the Greens. The losers are David Winderlich and Family First.
Read more after the jump...
Legislative Council around the traps Tuesday, March 16, 2010
Tool for helping you vote below-the-line
Official Google Australia Blog: Be a Cluey Voter
SA 2010: Batch simulation Monday, March 15, 2010
Methodology
A batch of 20,000 simulations with a standard normal distribution applied to each party's vote (with a underlying primary vote based on a sythnesis of galaxy and newspoll). Basically, we run 20,000 simulations and each time change the vote of each party by a small random amount. This should allow us to see which parties are most likely to rise to the top. As a rule of thumb, if I, a non-South Australian, had heard of the micro party candidate I gave them 1% of the vote, else 0.1% of the vote. So David Winderlich, Save RAH, FREE Australia, Gamers4Croydon, Shooters, DLP all got 1%.
Results
The most common outcome, occurring 48.3% of the time, of the 20,000 simulations is Family First 1, Australian Greens 1, Labor 4, Liberals 5. The second most common outcome, occurring 39.5% of the time, is Family First 1, Greens 1, David Winderlich 1, Labor 4, Liberals 5. A distant outside chance, occurring in 10.2% of simulations is the Greens winning 2 seats.
Read more after the jump...
SA 2010: Greens vs Family First for final Legislative Council seat
With the group voting tickets now published for the South Australian Legislative Council election, and newspoll and galaxy providing reasonably up-to-date polling of the electorate, it is possible to run a simulation of this weekend's Legislative Council election. In fact, we can run two, one for newspoll and one for galaxy.
Results
If Newspoll is accurate, next weekend will result in the Liberals winning 5 seats, Labor 4 and the Greens 2. If Galaxy is accurate, result will be Liberals 5, Labor 4, Greens 1, Family First 1.
Methodology
Taking the declared candidates, the group voting tickets lodged with the South Australian electoral commission, and the predicted vote for each candidate provided by the latest polls, I simulate the election day by counting the votes in a manner similar to the one used on the day. Candidates with a quota 8.33% are elected to a seat. If, after electing all candidates with a quota, there are seats unfilled, we eliminate the candidate with the lowest number of votes. These votes are then distributed to their next preference. We keep eliminating candidates until somebody has quota, and eventually all the seats are filled. I don't do split tickets very well, but then this is just all idle speculation.
Newspoll
SA 2010 Legislative Council analysis - group voting tickets Sunday, March 14, 2010
Groups that have fared well in the 2010 South Australian Legislative Council race for preferences include Dignity for Disability, United Party and a slew of independents (including David Winderlich). Dignity for Disability in particular have done extremely well.
Groups which have fared very poorly include the majors such as the Liberals, Labor and Greens parties and some esoteric minors such as the Anti-Abortionists, DLP, Ultra Progressive Frayne Coombe, Shooters Party and both Euthanasia parties. Labor, Save the Unborn and One Nation are in particular on the nose with other candidates.
Middle of the pack include the FREE Australia Party, Democrats, Games4Croydon, Family First and Save RAH Party.
The full table is available below:
Whose preferences do the Australian Greens need the most? Friday, March 5, 2010
The other day I looked at which party would benefit most from Green preferences in the senate at the next election. The result was that of the three battleground states simulated, Green preferences only really matter in Western Australia, where they look to play a decisive role in whether the Australian Labor Party (ALP) or the Liberals/Coalition pick up a third seat.
Today I am going to reverse the situation and ask the question: Whose preferences do the Australian Greens need the most?
Q. Do the Labor and Liberal parties need Green preferences? A. No Saturday, February 27, 2010
From my previous simulations, we can see the Senate election this year is going to be a lot less interesting than in 2007. Solid Green voting and lower Coalition voting means most states will return 3 ALP senators, 2 Coalition and 1 Green. However, of the slim pickings available, the most interesting Senate battles this year are probably in Queensland, Western Australia and New South Wales.
Since we use above-the-line preferential voting in counting the ballots cast for the Australian Senate, political parties are able to do deals with each other to rank their candidates higher or lower. This "saves" the vast majority of voters from having to do it themselves (Luke's tip: ALWAYS VOTE BELOW THE LINE).
Above-the-line voting leads to a lot of back-room dealing before an election whereby parties try to gain favourable preference deals, hoping to ride preferences into the Senate (ala Steven Fielding in Victoria in 2004). This wheeling and dealing is generally an unpleasant affair since strategists have to weigh up the desire for a good deal versus doing a deal with ideological enemies.
Despite the angst caused by the desire to get good preference deals, in the majority of cases the stress counts for naught. Put simply, if you get enough votes you get elected. Preference deals are far behind in effectiveness. It is only in borderline elections that the arcane black magic of party preference deals actually matter, and even then, with so many independent actors in the system, for an individual party, preference deals may as well be random. This is backed up by repeatedly by the results of my advanced election simulator, apollo, which is going to be applied to today's question:
Do the major parties need Green preferences in the Senate this year?
Can David Winderlich win? Sunday, February 21, 2010
Yesterday, one reader commented on my simulation of the forthcoming South Australian state election that it was a "big assumption" to assume minor-party/independent (MPI) preferences could flow equally to the Greens as well as each other. The implication being that a candidate like David Winderlich would harvest more MPI preferences than the Greens and is therefore a higher chance of winning than the 20-30% I gave the micro-parties/independents.
Let's see, shall we? (Yes, I do requests!)
Since I had already set up the scenario for SA 2010 for the previous run, it was simple enough to make adjustments and run two new scenarios. The first new scenario is a set of simulations where Winderlich's vote climbs from 0% to 6% and everyone's preferences are distributed randomly. The second new simulation is similar. Winderlich's vote climbs from 0% to 6% but the random preference distribution is filtered so that minor party/independent candidates always preference him above the Greens.
Tight Race for South Australia Legislative Council State Election 2010 Wednesday, January 20, 2010
In 1985 the South Australian Legislative Council adopted an electoral system "almost identical" to the method used in the Australian Senate. The main difference being that instead of six seats available at a "normal" senate election, in a South Australian state election, eleven seats are up for grabs. The similarity to the Senate makes it perfect for my election scenario designer cassandra and my advanced simulation software apollo to try and predict the outcome of the 2010 South Australian state election.
If the election were held today, of the 11 seats available, Labor would win 5 seats, the Liberals would win four seats and the Greens would most likely win 2 seats but with a good chance that the micro party with the highest non-major vote would win the last seat.
Disaster looming for Coalition in Queensland and Western Australia Friday, January 1, 2010
If the latest Newspoll figures are correct, then the Coalition is in danger of losing two traditionally held Senate seats in Queensland and Western Australia, with the collapse in the conservative vote benefiting Labor and the Greens. This would be a remarkable result, particularly in Queensland, where in that state the Coalition face the prospect of going from 4 seats at the 2004 election to only 2 seats in 2010.
Using apollo, my advanced simulation software, I take the Newspoll figures and simulate the 2010 Senate election in each state. The results are as follows:
NSW: ALP 3, Coalition 2, Greens 1
VIC: ALP 3, Coalition 2, Greens 1
QLD: ALP 3, Coalition 2, Greens 1
WA: ALP 3, Liberals 2, Greens 1SA: ALP 3, Liberals 2, Independent 1
TAS: ALP 3, Liberals 2, Greens 1
ACT/NT assume split 1/1 Lib/ALP (no simulations run)
For a total election result:
ALP 20 (+4 over current senate and a total of 32), Coalition: 14 (-7, total 32), Greens: 5 (+3, total 6), Ind: 1 (+1, total 2)
This is a dramatic (in Senate terms) change since my last simulation was run in November. At that time, apollo predicted the ALP would win 20 seats, the Coalition to win 16 seats, the Greens to win 3 seats, and Friends of Xenophon 1 seat.
The big changes since the last simulation (the two lost Coalition seats) are in Queensland and Western Australia.
Queensland
In November, the Liberals were neck and neck with the Greens for the sixth and final senate seat in Queensland (50% for the Greens, 33% for the Coalition). However, a rise in the Greens vote has solidified the minor party's lock on the sixth seat (they win in 66.4% of simulations) and seemingly pushed the Coalition out of contention (below 14% - output was garbled d'oh).
Western Australia
In Western Australia, the strongest state for the Liberals at the 2007 election, the result is nothing short of catastrophic. Going from an absolute lock on a third seat at the next election, the Liberals are facing a shock loss of their safest "third seat".
The following two images need a bit of explanation. With Liberal and Labor expected to gain a full quota for two seats each, the fifth seat becomes a contest between the parties with the highest remaining quotas. By this stage, the Greens have a higher quota than either of the majors and so win the fifth seat. This leaves the Liberals and Labor fighting over the sixth and final seat. The increase in the Labor vote has seen them edge past the Liberals, winning a third seat in 49.5% of simulations compared to the Liberal's 39%.
This is still an extremely close battle, but the fact the seat is no longer a lock for the Liberals must be extremely worrying for them, and strengthens the case for Rudd to risk a double dissolution election.
The Liberal senator most at risk appears to be Senator Mathias Cormann, who was third on the Liberal's Senate ticket in 2004.
Conclusions
The falling Coalition vote has turned their previously winnable third seat in Queensland to a likely (but not totally solid) Green's seat. This is a remarkable turn around since these senators were last up for election in 2004, where the Howard led Coalition won an impressive four seats in Queensland (and overall control of the senate). They have also lost their lock on a third seat in Western Australia, with the seat now winnable by Labor.
About the methodology:
I took the swings in the ALP, Coalition and Greens votes recorded in the most recent newspoll and applied it to the 2007 senate vote each party received. For each state, I run 20,000 simulations with random preference tickets (a similar run on the 2007 results shows a very high accuracy of simulation) and tabulated the results.
Posts
