Archive for April, 2011

After analysing the overall preference flow of the Legislative Council count last week, I have moved on to taking a look at the results in each electorate. When analysing electorate results, most energy is expended looking at the Legislative Assembly vote, but the Legislative Council vote is fascinating as it isolates many factors in individual seats, such as candidate quality and ballot paper position. Every voter in the state gets the same ballot paper and vote for the same parties.

Most patterns in the Legislative Council vote mirrors those seen in the Legislative Assembly, but there are some fascinating elements.

The Coalition topped the poll in the upper house in 80 of 93 districts. In comparison, the Coalition won the most primary votes in 70 lower house races, winning 69 of those seats, and only being overtaken on preferences in Balmain.

The ALP overtook the Coalition in twelve of their strongest seats: Auburn, Bankstown, Blacktown, Cabramatta, Cessnock, Fairfield, Keira, Lakemba, Liverpool, Mount Druitt, Shellharbour and Wollongong. The Coalition managed to overtake Labor in the upper house vote in the seats of Canterbury, Heffron, Kogarah, Macquarie Fields, Maroubra, Toongabbie and Wallsend. The Coalition also won the upper house vote in the independent-held seats of Sydney, Lake Macquarie and Northern Tablelands, and in the Greens-held seat of Balmain.

In the seat of Marrickville, the Coalition came third, with the Greens coming first. While the ALP’s Carmel Tebbutt outpolled Fiona Byrne by 2.23% in the lower house, the Greens beat Labor by 8.28% in the upper house, with the Coalition trailing behind.

In Balmain, the Greens outpolled Labor by only 0.56% in the lower house, allowing them to overtake the Liberal Party on preferences. In the upper house, the gap was more than 10%.

This result certainly indicates that, in the most interesting and complicated race in the state, the personal vote for local Labor MP Verity Firth played a large role in blunting the swing and bringing her close to winning. In the neighbouring seat of Marrickville, the Labor vote was 8.16% lower in the upper house, a similar figure to that in Balmain, indicating that Labor held on in Marrickville largely due to the sitting member.

I broadened this analysis to see if this trend appeared in other seats. The ALP polled more votes in the lower house in 63 of 93 seats. The difference was more than 5% in 25 seats, and in twelve seats the lower house vote was at least 7% higher than in the upper house. All twelve of these seats had a sitting Labor MP running for re-election. In seven of them the ALP retained the seat, while in the other five the Labor Party lost the seat. It does indicate that in some contests a strong local MP managed to hold back the tide. This analysis has previously been demonstrated elsewhere when looking at differential swings. This suggests that the anti-Labor tide was just as strong in these seats, but were held back by local Labor MPs who campaigned virtually as independents.

It’s also worth noting that the upper house vote in Balmain is much lower than the other seats on this list. With a vote of 22%, it was the eighth-worst vote in seats previously held by Labor. The seven seats with with lower upper house votes than Balmain (as low as 19.1% in Menai) all tended to have roughly similar lower house votes, indicating they were some of the seats most heavily hit by the anti-Labor swing. These include Menai, Coogee, Miranda, Heathcote, Blue Mountains, Drummoyne and Riverstone. This seems to indicate that Firth turned what would have been a massive defeat in Balmain into a narrow loss due to her personal vote and effective campaign.

Apart from shining light on the effect of a personal vote on the Labor vote in each seat, the upper house figures include some other interesting statistics.

The Greens overtook Labor in sixteen seats, many of which were the same seats that the Greens overtook Labor in the lower house. This includes Marrickville, and fifteen seats where Labor came third and the Coalition came first.

The Greens were overtaken by other minor parties in 15 seats. In twelve seats they were overtaken by the Shooters and Fishers. In one of these twelve, Barwon, the Greens came fifth behind the Shooters and Fishers and Pauline Hanson. In the seats of Mount Druitt and Blacktown the Christian Democratic Party came third, and in John Hatton’s old seat of South Coast the former independent MP came third with 10.89%.

I was interested in investigating where the increased vote for the Shooters and Fishers came from, geographically. In 2007, the Shooters Party (without the Fishers) polled 2.8% statewide, while the Christian Democratic Party polled 4.4%. This time around the Shooters and Fishers increased their vote to 3.7% while the CDP vote fell to 3.1%. The CDP decline is easily explained by the 1.5% vote for Family First, headed up by former CDP MP Gordon Moyes. Why did the Shooters and Fishers vote increase?

My original hypothesis was that the vote was due to the party taking on the issue of fishing. The fishing issue had become a major political issue on the north and south coasts at the 2010 federal election and the recent state election. When you break down the vote for the Shooters and Fishers, however, you find that most of their vote is concentrated in the inland country areas.

I divided the state between the metropolitan areas including Sydney and areas as far north as Newcastle and as far south as Shellharbour. I then divided the remainder between 12 coastal seats and 17 inland seats. Averaging out the votes in each of these seats in 2007 and 2011 produced the following figures:

While the Shooters did gain a swing in coastal areas affected by the fishing debate, the swing was much bigger in inland areas. The biggest swings were in Murray-Darling (5.93%), Murrumbidgee (5.06%) and Albury (4.12%). As a comparison, the Fishing Party (which did not welcome the Shooters Party changing its name) polled more strongly on the coastal strip, but not by a substantial margin.

I blogged earlier tonight on what the actual result was in the Legislative Council. Now I want to turn to analysing that result and how that reflects on the campaign and coverage of the result.

In the end the result of my projection was accurate, even if the exact margin wasn’t so accurate. The problem was that my calculator needed to make an assumption about the number of formal below-the-line votes, which it did by looking at the proportion of above-the-line votes that were counted. Through most of the count it assumed that the final number of BTL votes would be between 60,000 and 80,000 votes, but the final figure ended up being over 91,000. It appears that a lot BTL votes were left until the end of the process to be data-entered, which makes sense as they take much longer to process than ATL votes. So while I had Hanson’s percentage of BTL votes about spot-on, the much larger total number of BTL votes meant that I had underestimated her final primary vote lead by about 9,000 votes.

If I had known that Hanson would end up leading by 15,000 votes I would have been far less confident in my belief that she would lose, which would have been reasonable considering how close the final figures were. All the same, it still remained far too early to call the result yesterday, as some media outlets did.

In addition, for reasons that I have yet to understand, there are more ATL votes in the final count than there were in the original count posted on the website. The Coalition had 5,000 more votes, the CDP almost 5,000, Labor almost 4,000, the Greens 2,000 and Hanson 4,000. Thus the overall percentages were different, whereas my calculator assumed that these two figures would eventually coalesce.

In the end, the key preferences deciding the result certainly weren’t the Greens (despite the shrieking of David Penberthy and Labor figures), nor were they Labor preferences. It is true that the Greens and Coalition needed every set of preferences they got, but the decisive preferences came from John Hatton and Gordon Moyes. After all of Labor’s talk about preferencing the Greens and the need to keep out Hanson, only 9.4% of Labor votes at the key point went to the Greens. It’s hardly particularly impressive, and may have been just as high if the party hadn’t made a big song-and-dance about the issue. Maybe it would have been higher if the party had actually preferenced the Greens in all seats, as they said they did. Luckily it was just enough to win.

This points to how difficult it is to know how these things will play out. It remains the case, as it was before the election, that preferences won’t make a difference in the Legislative Council unless the primary vote count is extremely close. Some of us may have been flippant in dismissing the possibility of this close race happening, but you have to admit that this contest was extremely close, and that very few preferences did actually flow. Of the eleven groups that had their preferences distributed, the exhaustion rate was on average 78%, ranging from 67% for Socialist Alliance to 85% for the ALP. It was a perfect storm. I still believe that Greens preferencing Labor would have been counterproductive and resulted in neither Labor or Greens winning one of those last seats, but it is true that the scenario Labor painted could have come true, and one bearing some relationship to it did. Yet despite all of Labor’s talk about needing to swap preferences, 85% of Labor voters did not give a preference.

Most of the media commentary about which candidates were going to be in the final contest was inaccurate. Following election day the media talked about a race between Buckingham, Ferguson and Hanson, even as Ferguson’s chances slipped away. In the end Ferguson came 27th, being excluded before the No Parking Meters Party. Meanwhile the media (myself included) largely ignored the danger to the Coalition’s eleventh seat, and missed the key role Family First would play in pushing the Greens and Nationals ahead of Hanson.

It is quite disappointing how often mainstream journalists seem to show a serious lack of understanding of Australia’s electoral system. Regularly during the campaign journalists ran with stories that would have been non-stories if the journalist understood the electoral system. The Sydney Morning Herald ran a story about Pauline Hanson claiming a huge amount of public funding from running, despite the fact that she had little chance of polling 4%, and if she did she would have certainly been elected, providing far more access to public funds through a parliamentary office and expenses for eight years.

The Herald also ran a story about the Greens denying preferences to the ALP in Marrickville with no understanding of the fact that, as a race between Labor and the Greens, preferences from the Greens would have no impact on the result in Marrickville.

The Australian Associated Press ran a story on April 3, when it was clear that Labor was not going to win, swallowing Anthony Albanese’s highly dubious claim that the Greens not preferencing Labor would result in Hanson getting elected. David Penberthy went further on The Punch, but I think it’s more likely that his claim was malicious, rather than due to a lack of understanding, considering his outright refusal to reconsider his statement when the facts were presented to him clearly.

Just yesterday the ABC and AAP both reported that Hanson basically had the Legislative Council seat in the bag. They quoted Antony Green’s factual statement that Hanson was leading by 6,000 primary votes, and then spun that into a story saying Hanson would win, despite the evidence from the 2003 election, confirmed today, suggesting that the Greens could easily overtake a 6,000 vote lead. The ABC is usually better in covering electoral matters, and it’s a little bit scary that they can drop the ball this badly when Antony Green leaves the country for a few days.

Most of the time these stories come from journalists who usually show a strong understanding of politics in New South Wales, but sometimes it doesn’t extend to understanding the electoral system. I’m not talking about young journalists who don’t understand policy, as was an issue during the federal election. Often you need to conduct your own research to know the consequences of preference flows, or to know whether a particular outcome is possible or probable. I worry that the modern political journalist has little time to test these outcomes and do this research, and it often can fall by the wayside, buying the line of a politician who has an agenda to push (like Luke Foley or Anthony Albanese). Maybe major media outlets need to invest in someone like Antony Green who understands the process and can follow the count and give proper advice to political journalists. No, I’m not volunteering, although journalists can always call me to talk about these things.

It is important to Australian politics that journalists actually understand how elections work, and what way a result is likely to go. Often we can’t say for certain which way a result will go, today’s result was a perfect example of that. But understanding preference flows and where candidates stand in the count can allow you to understand how likely particular outcomes are. It can have a significant impact on the narrative of Australian politics when the media gets a story wrong. Just look at the reaction to yesterday’s stories saying Hanson was on track to win in the Legislative Council. Of course she had a chance, and came very close, but in the end it was far early to call it, and it turned out to be wrong.

Pauline Hanson, in her press conference this morning, criticised the electoral system and claimed she would have won under a system that made it easier for people to vote for her. It may surprise you, but I agree with her. I believe that Hanson received the largest below-the-line vote of any candidate because many voters would have been confused and not realised that they could vote for her above-the-line. Her above-the-line box was only marked with the words “Group J”, which have no meaning for any voter. In contrast, every other group except John Hatton got their party name above the line. Below-the-line votes for Hanson needed to have at least 15 squares numbered formally. Any “Just Vote 1″ votes for Hanson below the line would have been counted as informal. A survey of votes when I was scrutineering at the counting centre indicated that many informal votes were attempts to vote for Hanson. She would have likely been elected if those attempts to vote for her had been counted.

Hanson is not the best ambassador for electoral reform, but there are serious problems with how we conduct Legislative Council ballots that make it hard for voters to understand. As a first step, it should be at least possible for independent groups that have met the high bar to appear on the ballot to have “John Hatton independents” or “Pauline Hanson independents” appear above the line in the place of a party name. Independents running for the Legislative Assembly are as easy to find on the ballot as party candidates, the same should be true in the Legislative Council.

I would go further, and abolish the need to number all 15 squares below-the-line. This would also allow parties to run far less than 15 candidates, which would make the ballot paper much simpler and avoid having such huge numbers of candidates, the vast majority of whom never came close to being elected, and probably wouldn’t have wanted to be. If you wanted to go further, you could abolish below-the-line voting entirely. Below-the-line voting made no difference at this election. You might say “but Hanson wouldn’t have won on above-the-line votes”, but those votes would have probably been cast above-the-line if it was clear who they were voting for. It remains the case that a candidate further down a party ticket has never won a seat on a surge of below-the-line votes.

Instead, we could have a system where each party has one box, and you can only number preferences for parties as a whole, and preferences would then flow the way above-the-line preferences flow now. In the process you would make the process much simpler and much easier for people to vote. It would also make the counting process much quicker, without having a real effect on people’s ability to exercise choice. Despite the theoretical possibility of voting for an individual candidate against the party’s wishes, it currently doesn’t have any real effect on the count, and significantly confuses and complicates the voting process for many voters.

This is the first of two blog posts I will write tonight on the Legislative Council result. The second will focus on the spin and analysis around the count.

Today’s result in the legislative Council count couldn’t have been any closer. The final two seats went to the Greens’ Jeremy Buckingham and the Nationals’ Sarah Johnston by the smallest of margins, edging out independent Pauline Hanson, despite Hanson’s substantial lead on primary votes.

The count began by electing seventeen candidates from the parties that had polled over a quota: 10 Coalition candidates, 5 Labor candidates and 2 Greens candidates. Following this, 278 candidates who were either ungrouped or a candidate in an unwinnable position were excluded without a significant shift in the count.

After count 296, only one candidate remained from each of the 16 groups with candidates above the line. The key preference distributions that decided the result happened after that point.

The following chart shows the vote for the three candidates in the race for the final two seat. It ignores the CDP’s Paul Green and the Shooters and Fishers’ Robert Brown, who had not polled a quota but were polling well above Hanson, Buckingham and Johnston.

Votes at counts 296-308 for key candidates Jeremy Buckingham (green), Sarah Johnston (blue), Pauline Hanson (purple), showing the count as the last candidate of each party is excluded.

At count 296, Hanson was 9720 votes ahead of Johnston, and 16592 votes ahead of Buckingham. At every point of the count Buckingham and Johnston gained more preferences than Hanson, with a few candidates playing a key role.

The Greens gained boosts from the exclusion of Socialist Alliance candidate Peter Boyle (1609 vote net gain on Hanson), Democrats’ Arthur Chesterfield-Evans (3074 votes) the ALP’s Andrew Ferguson (3580) and independent John Hatton (3983). Johnston particularly gained votes from the Democrats, No Parking Meters and the Fishing Party, but were gaining votes slower than Hanson.

When John Hatton was excluded, Buckingham overtook Johnston. When Gordon Moyes of Family First was the only candidate remaining, the vote was:

• Hanson – 102,466 votes
• Buckingham – 102,276
• Johnston – 101,183
• Moyes – 64,738

While a vast majority of Moyes’ votes exhausted (52,101 votes) and over 4000 went to the Christian Democratic Party, Moyes’ preferences allowed both Buckingham and Johnston to jump over Hanson, leaving the final figures:

• Buckingham – 105,472
• Johnston – 104,341
• Hanson – 103,035

At this point Hanson was excluded, leaving four candidates for the four remaining seats.

Attached is the table of the preference distribution, beginning with the exclusion of “Restore the Workers’ Rights Party”.

The primary reason for this practice has little to do with political parties wanting their voters to follow their preference instructions, beyond voting for their party.

The primary reasons for HTVs is historical, and related to party exposure. Election campaigns in the United States and the United Kingdom do not focus on handing out pieces of paper to voters on election day, instead focusing on get-out-the-vote efforts. There are efforts made to put up signs and have a presence at polling booths, but the primary aim of the campaign is not to hand a particular new piece of election material to voters as they come to vote. Even in Ireland, the country closest to Australia using a system of preference voting, I don’t believe they hand out how-to-votes.

It was previously the case that it could be quite difficult to fill out a ballot paper without the help of your political party. For much of the twentieth century, all mainland state and federal elections were held using compulsory preferential voting, and with no identification of a candidate’s political party on the ballot. Voters for a major party would need to number every box, and may not be able to identify little-known candidates for such parties as the Communist Party. A how-to-vote would allow voters to number every box, even if they had no idea about a number of the candidates.

This is less of an imperative now. In state elections in New South Wales and Queensland, optional preferential voting means that voters can cast a formal vote without numbering many boxes. In most electorates the major parties advise voters to “just vote 1″. In other states, and in federal elections, party labels on the ballot paper make it easier for voters to number every box. Ballot papers also make it clear when voters need to number every box. Of course, many voters may lack the confidence or intelligence to work this out for themselves, but it remains the case that it is much easier to vote.

The main reason why political parties focus on handing out how-to-vote cards is exposure. It is the last opportunity to have contact with voters, and everyone else does it. It becomes a kind of arms race, with parties in marginal seats focusing on trying to have more workers than the other party, even if they are not all necessary to reach every voter. While it wouldn’t have a serious impact on elections if no party were to hand out cards, it would likely be detrimental for a political party if they stopped doing it unilaterally.

I’m sure it is true that a secondary reason for handing out HTVs is to ensure all of your voters do cast a formal vote. This would particularly be an issue in areas with low literacy rates or language diversity, where many voters may not be fluent in English. A how-to-vote can allow them to copy out their vote without understanding the instructions and can also help with confusion in states where the rules vary between federal and state elections.

If a party is one of the top two polling parties in a district, their preferences will not be distributed under any circumstances. Except for rare three-cornered contests such as Balmain, it is usually clear who will come in the top two well in advance, so the major parties know their preferences won’t be counted.

In some cases, preference decisions are part of a deal with another party. A party might swap preferences between a district where they won’t win and a district where they could win, or for preferences in the upper house. Even if preference decisions are made based on principle, they are usually not the primary motivation for handing out HTVs.

Discussions on preferences often dominate airtime for smaller parties, and in some places can be the only media a small party can receive. Rather than being an opportunity to influence politics, it can often prove a distraction, preventing the party from discussing policy issues that may have motivated them to run in the first place.