Tuesday, October 28, 2014

Hobart City Council Count (Includes Some Coverage Of Other Councils)

Wrapup (Saturday Nov 1):  

With the pressing of the final button in Launceston a few hours ago it's pretty much time to bring a close to my coverage of the Hobart and other Tasmanian local government elections and turn my attention to the deluge of Victorian and other state polls that I've neglected while this has been on.   If there are any post-count resignations then I will put up separate threads for the recounts to fill them.

Interest levels have been tremendous with both this page and the Hobart guide logging nearly 8,000 pageviews so far, each from nearly 3,000 unique visitors. Unique pageviews on Tuesday and Wednesday were at levels comparable to the busiest days of the federal election campaign.  I think this all says something against the idea that people are not interested in local government.  I'd like to thank readers for their interest, especially those who threw in a few hundred dollars in donations between them, which I definitely felt like I'd earned after 13 hours working more or less flat out on Tuesday.  It would be great to provide this level of coverage for more councils in 2018, but to do that there need to be more of me!

So what have we found out, especially from Hobart but also from the other results?



Hobart

Most of the Hobart results were about what was expected.  The mayoralty was pretty close and Hickey was the challenger she seemed to be and then some.  The surprise was that, from what seemed to be a winnable primary vote position, Damon Thomas lost on preferences to Hickey.  I've attributed this below partly to gender voting, but it is probably more complex than just that, with the Myer bailout (generally opposed by Greens) and the challenge of following a very popular predecessor being some other possible factors.  I had thought Hickey's Liberal connections might hurt her more in the race for left preferences but perhaps the word that from a Green perspective there is not much difference in voting records between the two had got out more widely than I thought!

The deputy mayoral race was as expected very drawn out.  Incumbent Christie, who I flagged as one of the top two chances, retained fairly comfortably in the end despite a feeble aldermanic primary vote, the performance of his main challenger Jeff Briscoe being weaker than expected on both ballots.
On the Hobart aldermanic ballot a major theme was the strong performance of off-council and especially younger candidates, while the sitting aldermen beside the three predicted frontrunners all polled sluggishly.  Two sitting aldermen lost (my prediction was "one or maybe two") while a third had a narrow escape for the second election in a row.  It will be interesting to see the age and gender stats on voting vs non-voting when they come out.

The defeat of Harvey was the biggest surprise; I've probably said enough about that already in this article.

Statewide

Overall we saw a major increase in candidate numbers, brought on by the all-in all-out system, the move to four year terms, and the opportunity to run directly for leadership positions.  I thought the increase may have been even greater but it was still large.

Despite the increase in names to vote for, total turnout was virtually identical to 2011.  Turnout increased in seventeen councils (including the votes received after Tuesday for Flinders Island) and declined in twelve.  The twelve showing declines included one with no councillor contest (Central Highlands) and one with only a single councillor vacancy (Latrobe).  The remaining ten were clustered by location - six in the south-east (Derwent Valley. Clarence, Sorell, Southern Midlands, Brighton, Glenorchy) and four in the north-west and west (Burnie, Devonport, Circular Head, West Coast).  That list to me (eg that those particular south-eastern councils went down while Hobart and Kingborough went up) suggests to me that lower socio-economic areas in many of those councils may have found the longer ballot papers and formality requirements all too hard.

A general message from the results is that running for mayor works.  82 of 91 candidates who ran for mayoralties statewide were elected as aldermen/councillors.  This is skewed by the tendency of very well established aldermen to run for mayor, and the only mayoral candidate without previous council experience to actually become mayor* was the obvious one (high profile businessman Michael Kent deposing recently controversial incumbent Bertrand Cadart).  But there are plenty of examples of strong aldermanic performances by off-council candidates who also ran for mayor to suggest that it is a huge profile benefit for an electable candidate in the larger cities especially.  Mass media coverage tends to focus heavily on the mayoral candidates but provides much less opportunity for new councillor candidates to get their message across.

(* There was also at least one winner from off council but with prior council experience and hence eligible under the old rules - Peter Coad (Huon Valley), who had previously been on the council from 2000-5).

The system

My biggest concern about this year's elections was the spike in informal voting.  The informal vote rate increased in all 28 councils that had councillor contests.  It was up by more than 50% in 22 of these councils including ten in which it more than doubled.  The informal rates in Hobart (7.47%, up from 3.24%), Launceston (6.19% from 3.48%) and Clarence (5.68%, up from 3.22%) were especially severe.  Some voters simply did find it all too hard - yes, they have time to vote carefully at home, but maybe many can only find time to vote quickly.

In the case of Hobart I have seen that the increase was largely because of unintentional informality.  While it might be rather cynically argued that a system that puts elderly voters at increased risk of voting informal helps cancel out the tendency of young voters not to vote at all, it's not good to be losing this much genuine voter intention.  Either the number of squares required for a valid vote should be reduced or there should be savings provisions for unintended errors.  State parliament must fix this; the good news is there are four years to do it in.

Another concern is the risk of uncompetitive candidates slowing the leadership counts by running for leader positions in large numbers.  We saw this with Hobart's tedious DLM count in which the Ron Christie team cluttered the ballot paper with uncompetitive candidates, to no evident benefit to anyone.  Fortunately the only person whose victory was delayed by this tactic was Ron Christie.  However, it may be that there will be more uncompetitive candidates running for the top jobs next time (there is even the potential for people to vandalise the system by running packs of pointless candidates a la the so-called Christian Democrats in the Bradfield by-election). I'd like to see this sort of thing deterred by increased requirements for unique nominators for leadership candidates, introduction of deposits, or both.

There will be, and already are, the usual calls for the illiberal and unnecessary extension of "compulsory voting" and (on various grounds) for the abolition of postal voting.  I'd like to point out again that voluntary on-the-day booth elections had truly pathetic return rates before postal voting replaced them.  However, it would be great if by 2018 we can have a fair portion of the voter base voting securely online ("securely" being the challenge here) so as to reduce resources required for either data entry or manual counting.

At this stage I am not aware of any problems with the results released under the Computer Count system and given that it appears to be more accurate and more objective while taking about the same time as the same counts (for that many candidates and surpluses) it seems, for now, that its trial has been successful.  But if I become aware of any issues with the results I will mention them.

Thankyou!

As well as the thankyous at the top I'd like to thank other scrutineers on the Hobart and other counts for their help both with scrutineering info and at times transport, especially Ross Copping, Peter Tucker, Peter Bull, Chris Merridew, Christian Street and Lili Sun, but also others too numerous to mention.  Thanks for their understanding to those who allowed me to be prioritised to a degree on Tuesday night so I could do sampling which was then shared with other scrutineers.   I'd also like to thank the TEC and AEC staff working on the counts and especially Kate Roach for their enormous help with information about the progress and processes of the various counts.  Thanks to the P Bonham taxi, thanks to Eva Ruzicka for authorising me as scrutineer so I could watch all this up close, and I'd like to say well done to Bob Vincent, for calling it for Hickey early despite my small and, as it turned out, rogue preference sample on Tuesday night.

Back for more in 2018.

===============================================
Note: Comments on other council preference cutups are now being added in the other councils section (scroll way down)

Welcome to my live coverage thread for the Hobart City Council count, which will also have some comments on other councils when I find time to look at them.  My candidate guide and preview was here and was viewed by about 10% of Hobart voters.  Updates will be added below the dotted line; check back regularly through the week for comments.  The introductory comments will stay at the top.

If you have a question about another Council after there are figures available please ask it in comments.

During scrutineering I will often have my phone off or on silent.  Media and candidates wishing to talk to me should send a text or email if they have a simple question, or leave a message;  I will get back to them when I can.  From past experience covering the election will cost me $150-$200 in phone costs and transport, and any donations to help with these costs (see sidebar), plus the obvious downtime from other work, are very much appreciated.

Current summary for Hobart

LORD MAYOR: Hickey defeated Thomas by 908 votes.

DEPUTY LORD MAYOR: Christie defeated Harvey by 1521 votes.

ALDERMEN (12): Hickey, Thomas, Burnet, Zucco, Christie, Briscoe, Sexton, Cooper, Ruzicka all called before button press.  Reynolds, Denison, Cocker elected on button press.

Count process for Hobart and all councils

The counts for Lord Mayor and Deputy Lord Mayor (and for mayors and deputy mayors of other councils) will be conducted as normal with counting expected to start from about 3 pm Tuesday.  Because of the large number of candidates for DLM it may be a slow count and preference distribution, especially if close.  It will be hard to scrutineer (even for me!) Hopefully by the end of the night we will know the LM winner and at least have a fair idea of who is Deputy.  We will probably also know several aldermanic seats.

The count for Hobart aldermen will be conducted using automated data entry.  This is scheduled to start from midday Tuesday (though that's likely to be a bit optimistic) but don't expect meaningful official results until the first posting on Tuesday night.  I may have some early indications from scrutineering before then, but they will not be all that reliable, and my attentions will be divided between this count and the mayoral and Deputy ballots.

Councils using automated data entry for councillor positions are Hobart, Glenorchy, Clarence, Kingborough, Derwent Valley in the south and Launceston, Dorset, Devonport, George Town and West Tamar in the north.  The remaining councils will be hand-counted with a final primary figure being obtained and then a preference distribution being conducted manually as in past elections.  The Flinders election count is being delayed for two days because of postal delays.

What will happen through the count for Hobart (and the other nine councils using automated entry) is that at a few stages of the count the Electoral Commission will post current primaries, together with a preference distribution based on the votes available so far.  This is expected to happen with around 20% counted at around 8 pm on Tuesday, with about 40-50% counted on Wednesday night, and then with a final release at the end of counting sometime over the weekend.  Between the second release and the final release (when they push The Button and the winners are determined, after a certain artificially programmed dramatic pause for effect - I'm not kidding!) there will apparently be no more releases of progressive primary figures

The 20% and 45%-ish counts will be indicative only.  For starters, if a candidate is projected to win by fewer than 50 votes based on 20% of the count, it's quite possible that by normal random sample error, that candidate will actually not win.  It might be possible based on these results to call, say, nine of the twelve seats, but not the last few.  They will be waiting a few days.

Random sample error isn't the only reason why results will change.  There's another one, and here it is:

Beware, Big Word: Stratification!

It's very important to be aware that interim primary vote figures for the various Councils using automated data entry won't necessarily be unbiased samples of the whole count.  Votes do not reach the Electoral Commission in random order, but rather arrive on particular days, and within each day's sample they may be further stratified by post office of origin (for instance).  There are some political differences between the pool of voters who tend to vote early in the count and those who vote late in the count.  Also, campaign events that happen during the postal voting period can have more impact during some parts of the count than others.

The TEC may well make vigorous efforts to reduce this stratification, but nothing will get rid of it completely.

Hence when the TEC release primary figures for various percentages be aware that these are not true representative samples and that past experience shows that substantial changes may occur.  Any result that is even remotely close will therefore not be called on 20% of primaries.

Please note that I only "call" a seat when I consider it virtually certain that the candidate in question has won.  Calls for Hobart seats will appear in the current summary.  I ask that people not describe me as having called any seat where I have not explicitly used the word "called" for it.

======================================================================
Latest Comments for Hobart Council

midnightish:  OK, here's the rundown on the closer exclusion points for major candidates for Hobart.

Cong: In both previous cases the quite important exclusion Cong vs Mao Ding has been very close but Ding did much better in the final count than the 20% and 50% counts and Cong was excluded by 64 votes - not vs Ding but vs Foley.  22% of her preferences went to Ding moving him well up in the count.

Foley: In the final count Foley was excluded 23 votes shy of Dutta (who really did poll very well, outpolling five sitting aldermen including the DLM on primaries).  Not a very relevant exclusion as neither was going to get elected anyway.  Dutta is then out 14 votes shy of Stansfield,

Harvey: The margin in the Cocker/Harvey contest blew out to 74 votes.

Ding: This is the really close one: after the Cooper surplus, Mao Ding is out by 3.6 votes in all of Hobart against Tanya Denison.  Ding's preferences then put Denison above two of the remaining competing Greens and create a cascade of surpluses for Sexton, Ruzicka, Reynolds and Denison.  As Denison's preferences would surely have put Mao over the line, this is the closest outcome for a seat since Barry Fisher defeated Dee Alty by 3.22 votes in 1996.

Andrew: After Harvey's defeat (first case I know of of a Green being displaced by an intra-party contest in any council) Cocker just gets over the line as well, defeating Andrew by 51 votes.

9 pm Friday - The Button!!! I forgot to mention TEC considered it possible they might get Hobart done tonight and that has happened.  Elected in order are Hickey, Thomas, Burnet, Christie, Zucco Briscoe, Cooper, Sexton, Ruzicka, Reynolds, Denison, Cocker.  From the 50% list there are no changes to the winners lists at all!  Cooper has moved up ahead of Sexton and Christie ahead of Zucco and the other big change is that Mao Ding must have beaten Cong at their exclusion point; he has finished 14th.  Sitting aldermen Foley and Harvey are defeated and there are three new aldermen Cooper, Reynolds, and Denison. 

The loss of Harvey rather than Cocker seems the big surprise here but it becomes less surprising when one hears stories of regular Greens sign houses refusing him on account of his votes on something or other.  I also suspect he lost independent-left votes to Cooper who has polled a remarkable fourth place on primaries (sixth after preferences) with 704 votes, more than half a quota in her own right.  I'll bet a lot of the outside candidates who didn't run for Lord Mayor are kicking themselves even harder now!

With the retirement of Freeman that means three females have replaced three males and that underlines something that seems to have been the hidden theme of this election: dissatisfaction with Council as a boys' club, as seen in regular preference flows between female candidates from all points of the political spectrum.  The new Council, assuming there are no challenges, will have a historic 6-6 gender ratio, a substantially lower average age, and probably a very different dynamic.  Its political balance hasn't changed all that much, though I expect Cooper will be well to the left of Foley.

I will have more comments when I have finished perusing the 205 page (aaargh!) scrutiny sheet.  That's likely to be tomorrow sometime. 

2:00 pm Friday: The button press for Hobart is expected around mid-Saturday. The swings from the 50% sample seen in Glenorchy (in which Stevenson jumped from 10th to 6th including turning a projected 111-vote deficit compared to Lucas into a 126-vote lead) do suggest that Rachel Andrew's chance of improving from fourth Green to third and being one of the three current Green winners is more realistic than I first thought, though it still probably won't happen.  Bearing in mind my previous comments about stratification, if for some reason votes at the tail end of the count were greener than those at the front end, then it's just possible the Greens could win four.  However there was no change to speak of in the Green vote between the 20% and 50% count.

7:40 pm: Sorry I've been a bit slack about putting up the final DLM margin which has been up for hours but Christie won by 1521 votes with 54.9% 2PP.  A weaker preference flow from Briscoe than Zucco last time but a comfortable margin in the end and a slight swing on his 2011 final result.  This oddly marks the fifth time in the past six that the Deputy's 2PP has been around 54-55 (the sixth one was the ultra-close Burnet-Sexton result in 2009).  The last person to win DLM by more than that was, er, my mother, way back in 2000.

3:00 pm: Back to the aldermanic count now and some news about informal vote rates.  First data entry is complete but figures for it aren't released and there is no further release of figures until the very end.  The very end will likely be Saturday, but we will see how that goes tomorrow.  Currently the TEC are checking votes initially flagged as informal (as of the 50% count they were 7.1% of total.)  Some were just votes requiring more scrutiny that have been readmitted but at current rates readmissions will make up about 0.5% of the total vote, so the final informal rate might be about 6.6%.  This is depressingly high (compared with 3.2% in 2011) and most of the informals are unintended.

Doubling numbers inside the first 12 is the lion's share of them, with skipped numbers and failure to fill sufficient numbers also substantial culprits.  There are of course the usual deliberate informals (complete blanks with or without scribbles) and the usual hopeless cases (eg voters who just tick 12 boxes) but most are resulting from mistakes made down the list in positions 7-12.  It is especially remarkable to see how many voters stopped at 11.  Maybe these voters consciously thought that near enough was good enough and brought their fate upon themselves, but the requirement to number 12 boxes with no room for error is trashing hundreds of votes that display plenty of voter intention and would be formal if only, say, 1 to 6 were required.

Like Antony Green with the Senate GTV disaster, it's time for me to say "told you so".  Because in my submission on proposed Local Government Act changes (written in late 2012) I warned that "The doubling of the number of candidates to be elected [..] will indeed lead to a significant increase in the rate of unintentional informal voting". I suggested the formality rules would need to be relaxed if the number of boxes required was increased.  However, nothing was done.  As both a Labor-Green parliament and a Liberal majority parliament have passed laws on local government without addressing this problem, and they got through the LegCo too, practically the entire Tasmanian parliament is to blame for this situation.  I haven't looked at informality rates round the whole state yet but they are high in Clarence too.

12:45 pm: And as usual when one sample is too low the other one is too high and the errors cancel out - Briscoe's gain rate on the Sexton primaries is only 0.184 which means he will definitely not catch Harvey, and the final throw will be Christie vs Harvey on Briscoe's preferences.  We already watched this movie in 2011 when it was the same two candidates on Zucco's preferences, and in that case Christie came from 979 behind and won by 1370.  Maybe Christie won't gain that steeply off Briscoe, maybe he will; it doesn't matter a bit because this time he will start something like 600 ahead.  That makes it fork o'clock for this contest.  Ron Christie has managed to sufficiently distance himself from a troubled mayoralty (or helped make it troubled, take your pick) and also his past voting pattern, and he will continue as Deputy Lord Mayor.  Check back for the final margins on the TEC site and here later this afternoon, I won't be waiting here hours for the Briscoe throw.

12:20 pm: Still some life here for Briscoe as his gain rate on the Sexton non-primaries looks higher based on the piles than my sample indicated, and might be close to .1, which would be enough if he also did slightly better than my sample off the primaries.

11:30 am: I don't have official figures for the last exclusion but it was something like Christie 4727 Harvey 4671 Briscoe 3854 Sexton 3730.  This means Briscoe needs to catch Harvey at a rate of .219 votes per Sexton vote.  Sampling I did of Sexton's preferences had the catch rate at .291 for the primaries (sample size 207) but only .02 for non-primaries (sample size 150, though these could be stratified a bit by candidate of origin).  With only 57% of Sexton's votes being primaries I project the catch rate overall at .174 putting Briscoe's chance of staying in at this count at about 5%. Christie is upgraded to "extremely likely", the confidence level lurking one layer of token doubt below the final "called".

10:30 am: Sexton will be out on this count as he is making only trivial gains on Briscoe. Harvey is actually doing well on this count (the more left-leading side of the Ruzicka support base having him as an obvious choice while the more right-leaning side has three options) making it doubtful that Briscoe can even catch Harvey off Sexton's preferences with substantial numbers exhausting.  This is looking very strong for Christie either way but I will hold off calling it just in case Harvey is third and his preferences flow really strongly to Briscoe (perhaps off Briscoe's stance on the cable car, for instance.) 

Thursday 9:30 am: Welcome to this morning's coverage for the DLM count final stages but I'm sorry folks, it's looking like a fizzer.  The throw of Ruzicka's preferences is pretty much scattering and no-one is making an impression on Christie who may even be gaining on this count; also I am doubtful that Sexton is making the gain on Briscoe (126 votes) that he needs to stay in the race.  My suspicion is that if it comes to a throw of Harvey's preferences between Christie and either Briscoe or Sexton, that Greens supporters may find that choice much of a muchness anyway, and if Harvey actually makes the final two then it's a walkover.  So I think it's very likely Christie has this in the bag.

Wednesday evening (long update!): Fortunately today's aldermanic scrutiny sheet was only 24 pages long so I was able to print it out and stick it together quite nicely.  As noted before the big change is that the Cong preference surf from 19th place no longer happens and instead Tanya Denison gets that seat.  If we look at the count closely there are some very close exclusions.  Mao Ding is excluded 4.7 votes behind Cong and 6.55 behind Dutta.  Stansfield is excluded 1.31 votes behind Cong and 8.12 behind Foley.  Cong is excluded 25 behind Foley.

I'm really not at all convinced that the order of any of these exclusions matters.  For instance Ding is relatively low profile in the broader community compared to Cong so if she is excluded before him it's hard to see why he would get elected (if she wouldn't from the same sort of position).  Probably even if any of these exclusions shift Denison is still elected, barring changes in the proportions of primary votes.

For the Cocker vs Harvey contest not much has changed - Harvey now gets excluded by 15 votes. That is still very capable of being turned around on the remaining sample.

The basis for the change for Denison is that her primary vote in this sample was 36 votes higher than expected based on the 20% sample.  This is the biggest difference in the pack except for Thomas (who had a high base, implying higher differences even by random chance.)

Thought I'd look at how close any candidate is to exclusion at any point, bearing in mind that these gaps would all else being equal be expected to double:

Cooper avoids exclusion by 191 votes at the exclusion of Foley.
Ruzicka by 152 at the exclusion of Harvey
Reynolds by 59 at the exclusion of Harvey
Denison by 38 votes at the exclusion of Cong
Cocker by 15 votes at the exclusion of Harvey

As I see it even if their primary votes go down by, say, 100 compared to projection on the final total, Cooper and Ruzicka still don't lose.  They could probably get no more primaries from here and still win if they continued to perform the same way on preferences, and preference flows are not as affected by stratification (which, incidentally there was quite a bit of with the earliest votes intensely strong for Hickey, a trend that has since dropped off slightly).  I am virtually certain both will win so they are called.

Reynolds is very slightly less clearcut though in all probability Reynolds will win.  The risk of her losing to one of the other Greens is statistically extremely low and within-ticket contests also shouldn't move around as much as anything else.  However if Harvey beat Cocker, maybe Cocker's preferences would strongly favour Andrew over Reynolds (for instance).  So I'll hold off on calling this one just for now.  (Likewise I can't be absolutely sure that Andrew will miss out, though her chance of bridging 66 votes to Cocker and Reynolds is very slim.)

I'm not going to call the last two at this stage.

-------

5:00 The final margin of Hickey to Thomas is about 900 votes and should be up on the TEC website before too much longer, they are just chasing the odd stray vote.  It will be something like 52.5% two-candidate preferred to Hickey excluding exhaust.  Close but not mega-close.  I'm going home now and will digest the councillor results and post more comments tonight.  Need a bit of a rest first!

4:10: Prior to Zucco exclusion it was Harvey 3759 Christie 3521 Briscoe 2870 Ruzicka 2803 Sexton 2651 Zucco 2283.  Zucco's votes have gone heavily to Christie putting him to the lead, but have also put Sexton over  Ruzicka whose preferences will be thrown tomorrow morning, with the DLM count finished for the day after the long day yesterday. We're looking at about Christie 4150 Harvey 3917 Briscoe 3220 Sexton 3095 Ruzicka 3060.  The cutup should be done by early tomorrow afternoon. The Sexton scenario isn't out of the question - he needs to pass Briscoe on Ruzicka's preferences, pass Harvey on Briscoe's preferences and beat Christie on Harvey's.  I think he now has a better chance than Briscoe. The Lord Mayor distribution of Burnet is still going.

3:10: Prior to Denison's exclusion official figures were Harvey 3662 Christie 3242 Briscoe 2665 Ruzicka 2491 Sexton 2325 Zucco 2103 Denison 1723.  Awaiting confirmation and possibly missing some small parcels I'm expecting something like H 3759 C 3520 B 2880 R 2803 S 2651 Z 2283, with Sexton needing to beat Ruzicka by about 150 from Zucco's to stay in the race.  But if he does that (which is quite plausible) it's likely he'll beat Briscoe on Ruzicka's preferences too and make the last three and still be in contention.  Anyway Christie still looks pretty tough to beat.

2:45: A new Hobart aldermanic count with 50% counted is up.  The difference in this one is that Denison has replaced Cong in the list of winners, beating her at the exclusion point where Cong previously survived by nine votes.  It will take me some time to examine the new scrutiny sheet carefully. The other eleven provisional winners are for now the same.

2:40: The first real test of the idea that Briscoe can catch Christie and on Denison's preferences it doesn't seem to be happening; if anything, optically based on pile size Christie might be gaining more.

1:55: Ding is out of the Deputy count and Christie's margin over Briscoe is almost 600 with over 8600 to throw before the expected crunch point for those two.  Christie has no ticketmates remaining.

1:25: Preferences of Cocker and Cooper have been thrown and continued to favour Hickey over Thomas.  The final margin could be several hundred votes.  The centre and right of the Hobart voter base could not decide and the left has been left to make the choice and is choosing the challenger.

12:40: Campbell's preferences favoured Denison and Ruzicka (gender voting strikes again!) Next to be excluded is Ding and after that we will go to Denison, then sitting aldermen.  Once the Christie ticket are all gone we'll get a clearer idea of whether Briscoe can do this or not.  Note that Sexton and Ruzicka are still not totally out of it either - I don't think Ruzicka can get the preference flow but perhaps if Sexton catches Ruzicka off Zucco's preferences he could keep going.

12 pm: Barlow also being thrown for DLM.  Despite Barlow being from the same ticket as Christie, Briscoe is gaining on these and the margin will close to about 532.  Harvey still leads but this is probably meaningless (even if he keeps his aldermanic seat!)

11:40 am: One candidate excluded from the DLM count (Siena).  No-one got more than 13% of them.  Christie's margin to Briscoe increased by 12 to 560. There was little change in checking of the DLM figures except that a 50 vote bundle for Zucco was found in the informal pile.  Cocker about to be thrown for LM.

11:15 am: Scrutineering samples of Cocker and Cooper also show flow to Hickey, probably slightly stronger than from Burnet.  On our samples Hickey will just about certainly win.

10:50 am: Merged samples of Burnet votes by three expert scrutineers including me are now all showing a slight preference flow to Hickey - around 55:45 of those votes not exhausting with a sample size of 460.  On this basis Hickey becomes somewhat the favourite while we wait to sample Cooper and Cocker.

10 am: While my own small sample yesterday had Thomas doing mildly better than Hickey on Burnet preferences, I've now spoken to another scrutineer who was sampling properly and had the reverse result also on a small sample.  So we need to regard this as a tossup until we get bigger preference samples.

9:40 am: I have had confirmation that the councillor counts for automated data entry councils are based on first entry only.  They have not been rechecked by second entry and it is to be expected when they do that the currently very high "informal" rate will go down.  Rechecking is expected to occur after the first data entry for the 50% sample.

9:20 am Wednesday: I'm at Elwick for the Hobart LM/DLM counts and will stay with them until they're clear and then see what more I can do with the Councillor count, which may not be much until next release.  At the moment LM/DLM primary votes are being rechecked and this will probably cause only very minor changes.  I'm hoping to sample some more Burnet, Cooper and Cocker preferences during this process as yesterday's samples were not big enough to call the outcome of the close mayoral race.  Don't expect too much movement at the station for a little while.

2:00 am Wednesday (sigh) After just spending two hours lovingly hand-transcribing a 172-page scrutiny sheet I have some insights into the interim preference distribution.  Aside from the three in with quota, all of Zucco, Christie, Briscoe, Sexton, Ruzicka, Cooper and Reynolds win sufficiently comfortably in the provisional distribution that there would have to be very major changes in vote patterns over the remaining votes to dislodge any of them.

There is a question mark over the defeat of Harvey and election of Cocker.  At the point of his apparent exclusion he is 11 votes behind Foley and 12 behind Cocker.  If he gets over Cocker at this point, either directly or on Foley's preferences, then Cocker loses.  Quite likely Cocker's preferences would then go to him (as his go to Cocker but not significantly to Andrew) and he would go on to be elected.  So I think these two are fighting for one spot.  Andrew has polled very well but looks most likely to just miss out.

There are many question marks over the apparent election of Cong.  Cong amazingly comes from 19th on primaries to win the 12th seat thanks to preference-sponging and a lot of luck.  I identify three tipping points that contribute to this:

* After the exclusion of Griggs and the surplus of Zucco, Cong leads Dutta by two votes and Mao Ding by six.  If Cong is excluded instead at this point, I don't believe Dutta can make it, but flow of preferences between Cong and Ding based on both being Chinese might then boost Ding.  If Cong loses at this point this frees up a seat that could go to any of Ding, Stansfield, Foley or Denison depending on her preferences and later preference flows.

* Foley is excluded 10 votes behind Denison.  Perhaps if Foley actually beats Denison he can get somewhere on her preferences, though I think it is unlikely.

* Denison is excluded 9 votes behind Cong (and probably irrelevantly a few votes behind Andrew).  If Cong is instead excluded then Denison or Stansfield may win that seat.

Thus there are two seats very much up in the air.  It's interesting to see there's a serious chance Hobart could have a council with a 6-6 gender split.

By the way the informal rate for Hobart is running at a shockingly high 8%.  In scrutineering I have seen a lot of unintentional informals (as mentioned before) but I am not sure if the votes have actually been double-entered or fully checked as yet, and hence if the formality level might change.  I will have more information on this tomorrow.

11:00 There is a provisional 20% result up for Hobart with the following order of election: Hickey, Thomas, Burnet, Zucco, Christie, Briscoe, Sexton, Ruzicka, Cooper, Cocker, Reynolds, Cong. (Stansfield is 13th). There is a distribution in a very user-unfriendly format which I will try to make sense of later tonight but those well down this list should not be counting their chickens yet! Likely there are some very close exclusions that are sensitive to slight changes. The shock so far is the provisional non-election of Harvey, but we need to see how this goes with more counting.  It is a shock even that his position isn't assured.  It is not so surprising that Foley is in trouble.  It is interesting that in the provisional sample Cong comes up from way down on primaries and I will have to check how she does this.  More later tonight,

10:20 Getting close to final figures for DLM.  It could look something like this, subject to checking:

Harvey 3449 Christie 2968 Briscoe 2423 Ruzicka 2193 Sexton 2126 Zucco 1907 Denison 1411 Ding 984 Campbell 545 Barlow 293 Siena 203

Still very unclear.  I'd expect Christie to probably catch Harvey, the question then being whether Briscoe or someone else catches Christie first.  Zucco can't do it.  Very difficult on these numbers for Ruzicka and Sexton too.

The final LM sample is looking like it will leave Thomas and Hickey virtually tied on primaries and it will all be down to preferences.  My sampling continues to show these favouring Thomas slightly but that is off small samples and there's a lot more work to do here.  Burnet will be about 1500 behind them and won't catch them despite getting c.50-60% flows off Cooper and Cocker

9:30 Just scrutineering the Christie ticket (HHT).  Of 100 Barlow, Ding and Campbell votes 16 exhausted without reaching Christie or anyone outside HHT, 11 went to Christie and 73 went to one of the six non-HHT candidates.  It's not looking like Glenn Druery needs to look for a new job yet!  Significantly Denison is getting a lot of HHT preferences, which could lift her in the aldermanic race.  If Christie doesn't have any real help from his ticketmates he is vulnerable to being run down by Briscoe, who could win from third.

8:10 Here are expected DLM primaries (unchecked totals) in current counting

Harvey 1848 Christie 1494 Briscoe 1255 Ruzicka 1184 Sexton 1100 Zucco 946 Denison 741 Ding 502 Campbell 246 Barlow 159 Siena 111

If it stays like this I think it is very likely Christie will retain his position.  He's the best of the rest behind Harvey, not far behind him and with four ticketmates care of his preference harvester.  Even if the flow from them only weakly assists him it will be all he needs. (Reverse ferret: see above!)

Lord Mayor - expecting the new primary figures to show Hickey and Thomas more or less locked with Hickey ahead 3166-3122 and Burnet further back on 2355.  The scenario of Hickey being knocked out seems off the table now and while Thomas is getting slightly friendly preferences from Burnet (and presumably also from the other two though I've not checked yet) it is still too close to call. 

7 pm: I believe stuff is being posted for other councils on the TEC site, I'm too busy to even look at it! We have some 20% count figures for Hobart:

LM: Thomas 1216 Hickey 1144 Burnet 998 Cooper 357 Cocker 233
On these figures Burnet would be likely to overtake Hickey who would be eliminated in third place, her preferences would then elect Thomas.  Even if Hickey outlasted Burnet Thomas would win on Burnet's preferences.  But still early days given the issues with sampling I mentioned before

DLM: Harvey 810, Christie 537, Briscoe 527, Ruzicka 480, Sexton 421, Zucco 339, Denison 203, Ding 201, Campbell 134, Barlow 114 Siena about 50.   As expected a long and unclear cutup to see who might be able to haul in Harvey on preferences.  His primary lead is only 7% so I would think someone should catch him, but again wait for more.

Councillor: Patterns below were maintained as my sample reached 831 votes with Hickey still leading Thomas and Burnet.  The order of the rest after surpluses on my projection started Zucco, Christie, Briscoe, Ruzicka and Sexton tied, Stansfield, Harvey, Cooper with Reynolds,Andrew, Foley not far behind.  Cocker seems to be struggling a little compared with Reynolds and Andrew and Foley may also be struggling but the numbers of votes sampled for all the borderline candidates is very small - even the TEC 20% sample won't tell us much for sure.  I'm increasingly confident Briscoe will be returned as an alderman - not quite calling yet - but those below him remain unclear.

Taking a break to try to bludge food now! Will do some LM/DLM pref sampling later.

5:35 pm: Dutta who was surprisingly in the mix in my initial sample is dropping back in further sampling.  Not much else changing although perhaps the dominance of Hickey over Thomas is easing up a bit.  A possible LM scenario is Hickey leading Thomas on primaries with Thomas relying on Green preferences to try to catch her, but it will be a long time before we know if that's how it is.  For DLM the issue is none of them are doing all that well on aldermanic votes so who knows what will happen!  That one will be a long slow cutup. 

4:45 pm: Lord Mayor/DLM counting will not start til c. 6 pm.  Just about to start sampling Alderman again, will update around 6 maybe.

3:45 pm: I have a sample of 590 aldermanic votes and am heading out to LM/DLM in the Pru Bonham taxi.  Hickey has come back to 3.1 quotas in my sample with Thomas on 1.9 and Burnet 1.3.  Based on preferences, in my sample Zucco and Christie come up to close to a quota after surpluses while Briscoe and Sexton are not doing so strongly.  Harvey is not clearly the second Green in with Harvey, Reynolds, Andrew, Cocker all in the mix (Cocker struggling).  Assuming the Greens do get three, which I haven't fully checked out, we have five remaining seats for which the following are most prominently in the mix in my sample: Briscoe, Stansfield, Sexton, Ruzicka, Cooper, Dutta (!), Foley.  Others are competitive but struggling at this stage. Oh and the rate of accidental informal voting is disgusting - seeing a vote for your candidate binned because the voter only numbered to 11 is a sad experience and a damning indictment on the politicians who retained a requirement to number as many spaces as spots to be filled.

Tuesday 2:30: Just about to resume counting after lunch. Not much more sampling news.  I've decided to sample Councillor samples for another hour before starting on LM/DLM.  Some early reads: despite his feeble primary in my samples Christie should be OK - surplus from Hickey if it doesn't put him in will put him high enough to get preferences off his ticket.  Generally the outside candidates (Stansfield another one going OK so far) are doing better on primaries collectively than I thought but it will be ages before we know which have the best chances of grabbing the vacant seat.

For LM, the size of primary for Hickey in aldermanic sampling so far is promising but too soon to say much. For DLM, none of them are doing much in aldermanic primaries.  I predict Harvey will lead on DLM primaries and then it will be a question of how the "blues" go chasing him down - keep an eye on Sexton here based on what I've seen of the performance of the others so far.  (Update later - Sexton probably not going to get enough DLM primaries.)

Tuesday 1:30: Counting has started.  My first scrutineering sample of a mere 234 votes has Hickey on about 3.5 aldermanic quotas, Thomas on about 1.5 and Burnet a shade over 1.  All others nowhere near quota in this sample and too early to say much else except that Christie is polling badly on primaries in this small sample so it will be interesting to see if that holds up, Cocker is not polling better than the various minor Greens so far; Cooper doing quite well but sample too small and perhaps not representative to mean much yet.  I am trying to get up to a sample of about 700 primaries before heading out to LM/DLM.

Comments On Other Councils


Satuday 9:33 pm: Kingborough Button Pressed:  A turnaround in the Kingborough count and Flora Fox is saved! Samantha Scarlett misses out.  Investigating ...

Saturday 4 pm: Launceston Button Pressed: The last council count is over with Launceston returning no changes from the list of candidates elected at the 50% count, and all in the same order too.  There were some changes in exclusion order down the list.

Saturday 1 pm: Kingborough Button Pressed: The count in Kingborough is over with no changes to the list elected from the 50% count.  Richard Atkinson moved up from sixth to fourth.  We now await only Launceston of the ten Computer Count councils.  So far, with the possible exception of Derwent Valley for which I missed the 50% figures, we have not seen any council with more than one change from the 50% list of winners, and we've now seen four with no changes.

By the way I've been alerted to another close finish: Flinders Island Deputy Mayor, margin of one vote.

Friday 9:20: Devonport Button Pressed: The count in Devonport is over with one change in the list elected from the 50% count - Grant Goodwin displaces Peter Hollister.

Friday 8:40: West Tamar Button Pressed: West Tamar is in and not only have all the candidates elected in the 50% count got up, but they've done so in the same order, and the order of all bar the second and fourth excluded candidates is the same too!

Friday 7:00: Close One In Dorset: My attention has been drawn to the Dorset cutup in which at one exclusion point votes were Christine Booth 245.05 Leon Quilliam 243.16 Mary Schramm 241.13.  Having survived by 2.03 votes Quilliam then defeated Booth by 13.25 on Schramm's preferences.

Friday 3:20: Clarence Button Pressed: Well just as well I didn't call that one as - and apologies for anyone who got their hopes up here - my "confident prediction" was wrong!  Cusick has got up instead of Peter Cooper; the other winners are unchanged.  Having a look at why this has happened, there was a 44-vote gap at an exclusion point in the 50% count, and gaps of this size can get overturned as we have seen (though they usually won't).  Having too quick a look at it, I was so focused on where Thurley might or might not lose that I didn't notice a candidate above her was in danger.  In the end Cusick beat Cooper at this exclusion point by 10.5 votes.

Friday 3:00: Dorset Button Pressed: Only the last seat changed from the 50% result in Dorset with Quilliam winning in place of Schramm. The order of two other winners changed.

Friday 2:20: George Town Button Pressed: Again, where else?  In the George Town councillor election, candidate Nicholls has gone from losing in the 50% count to being elected third (more than trebling his primary), with the candidate originally elected sixth (Gibbons) now out.  These sorts of turnarounds are more likely in small councils, both for reasons of sample size and greater risk of geographic stratification.

Friday 1:30: Glenorchy Button Pressed:  Glenorchy is all done and all those elected at the 50% count did indeed win. The big surprise in the final count is that Matt Stevenson, winning the last seat by 51 votes in the 50% count, jumped from tenth to sixth, ahead of a bunch of candidates who he trailed by between 39 and 82 votes in the 50% count.  The most extreme swing (Stevenson up on projection by 126, Lucas down on projection by 111 at the same point) does indicate that 100-ish vote gaps for final positions in the 50% counts, like Thurley over Willink in Clarence, can potentially be turned around, though it's not all that likely in practice.

Friday 1:15 am: Also expecting all those currently shown as elected to get up in Glenorchy, with a little less confidence as there's a final 51-vote margin at the moment.

Friday 12:30 am: Thought I'd have a look at the Clarence aldermanic 50% count as its scrutiny sheet comes in at just 12 pages and it's probably the one next after Hobart that people I know are asking about.  In this count Chipman is elected on primaries with over three quotas, and Campbell and James are just over quota.  Then like Hobart it takes a while for the other incumbents to start getting in.  There are some very close exclusions early (most notably Richardson excluded before Holland by 0.03 votes!) but these are presumably irrelevant as candidates with such low totals aren't going to get up from there anyway.

The final margin is Thurley over Willink by 110 votes - which given that would be expected to double in final counting is quite a bit.  That's actually as close as Thurley or anyone else who is elected gets to being excluded at any stage.  Before that Willink edges out Cusick by 10 votes, which could conceivably go the other way (and did on the first distribution) but with Thurley 90 up on Cusick it's extremely hard to believe Cusick could catch Thurley on Willink's preferences and with four candidates to throw to.  And indeed in the first sample this did happen and it didn't do Cusick any good.  Before that there is a 15-vote exclusion between McManus and Sheridan but again with these candidates 100 below anyone else it's extremely difficult to credit that the order of exclusion here can matter.

I'm not going to call anything here as it's not the council I've been following and there may somehow be things I do not know that matter (seems wildly unlikely, but ...)  However I am going to confidently predict that all 12 currently projecting as elected will get up. We'll see how that prediction goes.  (edit: bzzzt! wrong! What happened was that Cusick edged Cooper at an earlier stage and Cooper's preferences were highly friendly to Cusick.)

Thursday 9 pm: The George Town tie is broken with a margin of four votes.

I have archived the 50% counts and scrutiny sheets for the following Computer Count councils: Dorset, Devonport, George Town, West Tamar, Launceston, Hobart, Glenorchy, Clarence, Kingborough (I was too slow with Derwent Valley).  I will be using these to see how many changes there are in results from 50% counts to final results; it may be a useful guide for the future.

Thursday 6 pm: The first Computer Count to finish, Derwent Valley, has seen former Greens MHA and former mayor of this council, Tim Morris, defeated in his bid to return to the Council.  Morris was left in ninth position for the eight-candidate council, missing out by 28 votes out of 3762.

Thursday: There is some kind of further checking (unsure if it is a formal recount) still going on in the Huon Valley mayor election.  I believe the basis for it is that while Peter Coad won by 193 votes, he did so after Mike Wilson beat Bruce Heron by just four votes for second place.

Wednesday: A freak scenario in the Deputy Mayor count for one of Tassie's most controversial councils, George Town.  There have potentially been two ties, one at an exclusion and one provisionally for the final result.  One candidate, Dean Gibbons, has already been excluded by lot.  Now Heather Barwick and Tim Harris are tied on 1326 votes apiece.  If rechecks fail to split them one will win the election by lot.

(Note: this differs from exclusion ties in Hare-Clark multi-member councillor positions - in those ties can be broken by countback of who was ahead at the most recent prior count. For mayors and deputies this isn't an option.)

4:45 pm Have heard a fellow expert scrutineer has called Glenorchy Mayor for the Andrew Wilkie-endorsed Kristie Johnson, apparently winning massively in a sample of 3000.

-------------------------------------------------------------------------------------------------------------------------
Disclosures:

1. I have been appointed as a scrutineer by Alderman Eva Ruzicka. I am acting in a voluntary capacity.

2. I was hired by the Tasmanian Electoral Commission as a freelance consultant to conduct pre-testing and make recommendations on the Computer Count system.  My full report on testing of the system can be viewed here:

2014 Tasmanian Local Government Elections: Review Of Computer Count System

11 comments:

  1. Hi Kevin, please share Glenorchy results when/if you can.

    ReplyDelete
  2. And by next election, well prior, we'd better make sure you have a team of well trained scrutineers and well funded (all disclosed) set up for you, so you can be reporting from across the State. Kevin, this is really good reporting and your stratification comments are spot on.

    ReplyDelete
  3. What is "automatic" voting. On ABC radio this morning Sue Hickey said she was doing well in it.

    ReplyDelete
    Replies
    1. She may have been referring to the automated data entry for councillor papers.

      Delete
  4. Hi Kevin, can you give any insight into why Harvey polled so well for DLM but not as alderman?

    ReplyDelete
    Replies
    1. Yes. He was the sole Green candidate for DLM so polled most of the DLM votes of those who voted for various Greens for aldermen. His DLM vote isn't anything special, it just looks good because the others are so spread. But it seems Greens voters this time decided he was too moderate compared to their other aldermanic candidates. In the DLM field there are really not even any other left-wingers apart from Harvey (I count Ruzicka as centre) whereas the aldermanic field was packed with left independents as well as other Greens.

      Delete
  5. I agree with your comments about the informals - hope this election is at least a wake up call!

    ReplyDelete
  6. Hi Kevin, it was nice to chat to you the other day :) just want to let you know you i think you have done a amazing job, i have been checking your blog every hour to see if there are any updates. thank you for doing this !

    ReplyDelete
    Replies
    1. Thankyou and commiserations on your extremely close loss.

      Delete

The comment system is unreliable. If you cannot submit comments you can email me a comment (via email link in profile) - email must be entitled: Comment for publication, followed by the name of the article you wish to comment on. Comments are accepted in full or not at all. If you submit a comment which is not accepted within a few days you can also email me and I will check if it has been received.