Well, there you have it a day of joy for the Tasmanian Greens as their most endangered sitting member, Kim Booth, just scrapes over the line in Bass by 136 votes. I might modestly note that it was my own gut feeling that this would be the case, as expressed in this post from early last week, despite news reports to the contrary as recently as this morning (in the Mercury and apparently also on the ABC). However, the real prize must go to Kevin Bonham, whose correctness was based not on lucky guesswork, but on psephological modelling that outperformed some formidable competition. Note his running commentary in the Tasmanian Times and contributions to the comments thread from this site’s previous post, in which he boldly predicts an imminent victory for Booth in the face of ongoing scepticism.
The key to Booth’s win was not just the rate of Labor leakage, as was stressed in my earlier commentary, but also the high rate of Labor exhaustion (thanks to Geoff Lambert for making this clear to me). Following the 2002 election, it was widely argued that Labor’s surprise late-count defeat was caused by the fact that they had fielded six candidates, when voters are only required to number five boxes. However, Antony Green argued that the real cause was leakage of Labor votes to Liberal candidates, particularly Sue Napier, and Labor was not discouraged from fielding six candidates again at this election. But this time, it does appear that Labor voters who numbered five of the six Labor boxes and left the remaining candidate hanging really did cost their party an extra seat. This compounded a repeat of Labor’s problem from 2002 when the number of Liberal votes leaking to their own candidates was subdued because only two of them had a sufficiently high profile.
Elsewhere, Paula Wriedt’s victory has been sealed in Franklin, which the Mercury did get right this morning, while all bets seem to be off regarding the final seat in Lyons. The final result is a remarkable status quo outcome in each of the five electorates 3-1-1 in Denison, Franklin and Lyons, 3-2 in Braddon and 2-2-1 in Bass.
Great stuff from the Poll Bludger as always but let’s call a spade a spade.
The Greens got their analysis right on this one. From election night on they said it would be close. Bob Brown in particular said again and again that it was stupid to call the result early. He and other Greens said that it would go down to the wire with Kim Booth probably winning Bass in a very close race.
Many others (not Poll Bludger or Tasmanian Times), with axes to grind against the Greens, wrote Kim off in hope more than study.
Lots of humble pie to be consumed in the morning. In particular it is Labor and Liberal strategists & spinners that must do the eating.
Yes Ben, but…..
Many of the analysts who rated Kim Booth with a low chance (me too) did so on pretty firm grounds- except for one thing. They didn’t know and apparently nobody except Bruce Taylor the Electoral Commissioner noticed, what the rate of ALP wastage would be until it happened at the Cripps cut-up. I think even the scrutineers also did not see this coming.
For the record, at the Cripps cut-up, the ALP lost 770 of the 4122 votes lodged with Cripps- 178 went to the Greens, 350 (!) went to the Libs and 242 exhausted. This changed the ALP lead over the Greens from 523 to a deficit of 425. I don’t think this was on anybody’s radar.
Who were these wasters?
* Personality voters who voted across the paper?
* ALP stalwarts who filled out only 5 of the 6 ALP squares?
* TFP & Rochester voters who filled out their 5 squares by voting erratically among ALP-LIB-GRN?
Probably?, probably? and probably?
Geoff Lambert
It’s not unknown for the last Labor candidate excluded with one Labor candidate left in the race, to leak in the high teens, although Cripps is one of the more severe cases. In 2002, the last Labor candidate out in Bass was Lyons who leaked 746 of 4301 (17.3%) with 431 to the Libs and 304 to exhaust; there were no Greens left in the count at that stage.
I looked at the issue of leakage to exhaust at the final Labor exclusion in Bass after a poster on Tasmanian Times, Tom Nilsson, raised Labor running six candidates as an issue and suggested it could cost Reissig “about one fifteenth of quota” which is almost 700 votes (comment 4 at http://tasmaniantimes.com/index.php/weblog/comments/lennon-wins-as-green-vote-softens/ ). I pointed out numerous errors in his assumptions (comment 15) and showed that in a vaguely comparable case (Denison 1998) leaks to exhaust from Labor had totalled 338 votes. I expected it to be significantly less this time with six Labor candidates rather than seven in the case I had looked at I therefore didn’t think it would cost Reissig his potential seat – but only because at that stage my projections had Booth 600 votes behind so the level of leak to exhaust didn’t look a big enough factor to save him whatever it was.
As it turned out my election-night projections for both Bass and Franklin were garbage because they relied on statewide estimates rather than actual past histories in the electorates in question. There are some differences in things like how many Green preferences go to Labor, and how many Labor votes leak, between different electorates. Think I only estimated 10% leakage from Labor to all sources for Bass on the night, but every single Labor Bass candidate in 2002 leaked more than that.
That the leak to exhaust this time was not small suggests that it is mostly across-the-paper voters and mostly not Labor ticket voters stopping at 5. Further very strong evidence that this is the case comes from Denison 2002. Labor ran only five candidates and when Crotty (5461 votes) was excluded with three(!) Labor and two Liberal candidates still in the count, a whopping 202 votes exhausted. That’s 3.7% to exhaust with three ticketmates still in the race, compared to 5.9% with one left in the Cripps case. So the six-candidate thing looks like a non-issue and the number of voters voting 1-5 from 6 Labor candidates then stopping should be just a few dozen per electorate. The real issue is that Labor needs to push its third candidate really hard rather than have three candidates polling a few thousand each and then getting trashed by leakage.
Whoops! Forgot to add that the real credit should be William’s for getting this one right at a time when everyone else including me (bar the Greens who generally say they’ll probably win even when it is virtually impossible) was getting it wrong. In particular William was right in picking Labor’s high vote in Bass as nothing but a personal vote for Michelle O’Byrne. I assumed it was a backlash against the Greens’ stance against the pulp mill, but it seems more likely that the pulp mill had no influence at all.
If you take the AVERAGE leaks ALP–>others, and LIB–>Others over the counts from 36-43 and assume NO exhaust, and then go back and use them to project forward from the time Courtney was excluded, you come up with an ALP victory over the Greens for #5 position by 118 votes, instead of a Greens victory over the ALP by 136. The difference, 254 votes, is almost exactly equal to the Cripps exhaust (242).
So, from this perspective, you could say the high exhaust rate at Count 36 made the difference.
Geoff Lambert
It’s something like that, but if the 242 Cripps voters whose votes exhausted were mostly across-the-board sprayers (as is very highly likely), then had they continued to the point where their votes were available to one of the five remaining candidates, it is not clear that their votes would have been for Reissig in sufficient proportions to make the difference. I suspect what’s going on is that when a candidate has been in the count for a very long time they have a higher proportion of non-party-loyal votes than a candidate excluded earlier whose vote at exclusion is dominated by their primaries and possibly surpluses from the same party. This should mean that late-excluded candidates leak more to other parties, but because there are so few candidates left in the count, they instead leak more to exhaust.
The exhaust rate in Tassie Hare-Clark cutups is typically very high where the last candidate from a party fielding five or more candidates is excluded, and otherwise stays very low for a long time and then rapidly escalates in the final few exclusions. Someday when I have way too much time on my hands I will do a model where a certain proportion of voters spray across party lines in proportion to primary vote (as an indicator of profile) but otherwise randomly, stopping at 5. I suspect such a model would display very much the same pattern.
I was really right for the wrong reasons with this one because as it turned out the proportion of leak from the ALP to all sources needed to get Booth over the line was indeed higher than the 12% I expected. I expected the Napier surplus from Fry via Gutwein to behave similarly to the Fry #1 votes and the Gutwein surplus from Fry based on past similar cases (presuming that increased leakage would more or less cancel out the tendency of those voters to be more strongly Liberal and hence more anti-Green). The flow from Napier’s surplus to Reissig was far stronger than I anticipated, meaning that Reissig in fact caught up almost 300 off Fry and the surpluses and not the 50 or 100 I expected.
I also may have overlooked the impact of ATV in my earlier calculations re Fry. Prior to ATV, Fry’s primaries would have been thrown as a single parcel, and would have been the votes soaked up most in the elections of Gutwein and Napier, and his votes received from elsewhere, which would have been less anti-Green, would have played a greater role in the outcome.
I don’t want to nag you, since I am sure you have another life beyond the site, but I wondered if we would see a preview of the Wellington and Rowellan Tasmanian upper house elections. Rowellan is pretty straight forward, but I have not been able to find out anything about two of the candidates in Wellington, and one of the things I most admire about this site is its ability to collect local information and info on candidates.