unspinning group voting tickets

The strange allocation of preferences in the Senate, where Howard has associated himself with a fundamentalist confessional party (and perhaps legislated to gain their support) could be cured easily. The NSW 1999 election was also skewed by an outbreak of microparties with unlikely names and platforms who exchanged preferences in a fairly promiscuous manner.

Antony Green's New South Wales Legislative Council Elections 2003 (PDF) tells us:

While the new voting system retained the 'above the line' or group voting option, groups had to nominate full lists of 15 or more candidates to have access to a group voting square. Votes cast using the group voting method only counted as preferences for the selected party, and could not be directed to other parties. Like-minded parties running against each other would therefore split their base vote. Previously, like-minded parties had been able to compete for votes against each other, sure of their ability to swap preferences. At recent elections, several parties have used this tactic to elect MLCs despite receiving quite small totals on the primary count.

A new form of 'above the line' voting was also introduced, allowing voters to order parties above the line, in an analogy with the way candidates can be ordered 'below the line'. Data on ballot papers is not yet available, but from the details provided in the distribution of preferences, it appears that less than ten percent of voters took advantage of this new option.

A consequence of the changes was that only 15 groups nominated in 2003 compared to 80 at the last election. However, every group nominated 15 or more candidate, compared to just three groups in 1999. So while the number of groups fell, producing a much more manageable ballot paper, the number of candidates rose from 264 to 284.

and:

Two findings are clear from the detailed distribution of preferences in Section 1.

(1) Preferences played no part in the final outcome. Under the previous operation of group ticket voting, preferences flowed strongly between groups on the ballot paper, as more than 90% of votes had been cast using the group voting option. Under the new system, 80-90% of preferences exhausted between groups. The number of members elected from each group at the 2003 election was determined entirely by the level of primary vote support for each group and was unaffected by the distribution of between group preferences.

(2) Parties that divided their core support were disadvantaged by the new system. The Shooters Party, Independent Pauline Hanson, One Nation, the Fishing/Horse Riders/4WD ticket, and Australians Against Further Immigration, probably share a similar support base. Together they polled 1.63 quotas. Under the old electoral system, this support could have been accumulated using ticket voting, giving an outside chance of electing two MLCs between the groups. Under the new system, these parties split their vote, no preferences flowed and John Tingle from the Shooters Party was elected with less than half a quota, edging out Pauline Hanson for the final vacancy.

The way to resolve this is to abolish group voting tickets and applyRobson rotation:

Of special interest is a feature of the Tasmanian electoral system whereby through a process of rotation each candidate gets a share of the position at the top of a particular column. This system has been in use since 1979. This is an attempt to even out the donkey vote (simply voting up or down the ballot) which is said to favour surnames early in the alphabet, or candidates early in the list. This system of rotation was championed by Hon. Neil Robson MHA, and is often known as 'Robson rotation'

Under the current electoral process a draw is made for the position of Party or independent groups across the ballot paper. Other candidates are classed as 'ungrouped' on the far right of the ballot paper. Next the rotation process is applied. Since 1996 this has been achieved by batch printing which first places candidates in a random sequence in each vertical column, then 'rotates' the names evenly in the positions available.

On polling day only first preference counting occurs; after postal votes arrive the cut-up of preferences commences. Candidates who achieve or exceed a quota of first preferences are declared elected.

Family First have proved the flaw in group voting tickets - people's votes are going to parties and candidates they have no intention of voting for. We know they have no intention of voting for them because no-one, but no-one, ever knows the registered orders of preference in any detail. The NSW solution is not bad, but its discriminatory between GVT voters and individual voters. Pure Robson rotation, on other hand, skews things to strongly against GVT voters.

What we really want is a system that's neutral between Robsonian individualism and GVT colllectivism. If I had my druthers, and there is absolutely no prospect of that happening, I'd:

• get rid of the line altogether;
  
• list all candidates and tickets in the same area of the ballot paper and apply Robson rotation across the board;
  
• allow voters to give as many or as few preferences as they wish.
  

Boring technical points
There would need to be rules that :

• because each candidate would appear individually and on each ticket, a candidate could only receive the highest preference shown for them on each ballot.
  
• a ticket preference would count as the next available preference for each ticket candidate and then revert to the voter's individual order
  
• ideally the same rules would apply to House and Senate elections with the quota calculated as V/S+1 (rounded up) where V= the number of voters and S=the number of seats to be filled.
  
• ideally we will one day elect our president this way.