The reason the vote in the ACT Legislative Assembly appeared to wander all over the place over the past week and a half was the freakish closeness of the vote.
Indeed, it was so close that Labor came within 350 votes of obtaining a majority of nine seats.
The ACT Electoral Commission did a computer crunch of the preferences after each day’s keying in of ballot papers. It treated all the votes counted as if they were the whole electorate and counted all the preferences of all of them.
Each day a different set of “successful” candidates would appear, sometimes flopping backwards a forwards between candidates. Given that by after Day 3, more than 20 per cent of ballot papers had been keyed in, you would expect that high sample to be fairly conclusive of the final vote. It was not, because so few votes separated candidates. In several cases the difference was just 0.1 per cent of the vote.
Ginninderra is a good example.
In a Hare-Clark count, the over quota (in this case that of Jon Stanhope) is distributed first. Then the preferences of the candidates with the least vote are distributed successively.
In the final count, at one stage after Stanhope, Wayne Berry and Bill Stefaniak were over the line we were left with the Liberals’ Harold Hird and Vicki Dunne, Dave Rugendyke (Ind), Roslyn Dundas (Democrat), Shane Rattenbury (Green) and Labor’s Susan McCarthy fighting for the remaining two seats.
The first to go at that stage was Hird. When Hird was eliminated he was just 55 votes behind Dunne. It is apparent that in earlier counts he would have been just ahead at that stage and stayed in with Dunne going out. It would have meant Dunne’s preferences (not Hird’s) would have spilled down affecting the later count, perhaps causing it to alternate between Rattenbury and Dundas.
Fifty-five votes is such a small margin that it would always fall within the sampling error in a total of 55,000 votes, until about 95 per cent of the vote had been counted.
Later in the count, just 48 votes separated Dundas and Rattenbury when Rattenbury was excluded. Even at the end, when Rattenbury’s votes were distributed there was fewer than 350 votes separating Dunne and Labor’s McCarthy. Who knows, if Hird had stayed in the trickle down would have been different and Rattenbury or McCarthy could have won.
He last two seats in Ginninderra were decided on margins as fine as the two most marginal seats in the House of Representatives.
In Brindabella it was different. Once the Greens were out and the Democrat Jeanette Jolley did not pick up enough preferences, Labor had the last seat won. The margin was not huge, but enough to remain constant once about a third of the vote was counted. Brindabella stayed stable throughout the count.
Molonglo was a bit like Ginninderra. So few votes separated candidates at critical stages that it would not matter what sample of the total vote you took, the likelihood would be that you would get a different result. And given that order of exclusion in the middle part of the final count affects the spilling down of preferences, the differences are often magnified to affect the election of the last two or three candidates, not just the last one.
The moral of the story is that if you want to have a full say in who represents you, vote preferences right down to the last candidate, do not stop at just five or seven preferences.
In this election 55 people (perhaps people who had voted Labor or Green as a first preference) determined whether Dunne or Hird got up and just 48 decided the fate of the Green.
And by the way, yesterday’s ACT result came quicker than the declaration of the last three House of Representatives seats in 1998 by three days.