Sunday, July 3, 2016

House of Reps Postcount 2016: Melbourne Ports

Melbourne Ports (ALP, Vic, 3.6%)
Michael Danby (ALP) vs Owen Guest (Lib) and Steph Hodgins-May (Green)
Outlook: Danby Retain (Awaiting official confirmation)

Key questions (updated Saturday 16 July, 2:30 pm):

1. Can the enormous declaration vote rate in Melbourne Ports cause Guest to beat Danby on the two primary preferred?  No.

2. Can Hodgins-May overtake Danby on the preferences of left-wing micro-parties?  Awaiting official confirmation - provisionally Danby has survived by about 800 votes.

3. If Hodgins-May overtakes Danby, who wins out of Hodgins-May and Guest?  Too close to call on scrutineering information available to date but irrelevant



======================================================================

Welcome to my post-count thread for the most intruiguing three-cornered seat of the 2016 federal election, and the one likely to cause the most confusion, Melbourne Ports.  A very large number of seats are already on the various lists of seats in doubt but this is one that generally isn't, and in my opinion should be, at least until we know some more.  Melbourne Ports, noted for its high (13%) Jewish population and also having a large LGBTIQ population and high Green vote. It has been held by Labor's Michael Danby since 1998 but at recent elections Danby's primary vote fell from 42.9% to 38.2% to 31.7% and at this election he is running at only 27.1%.

(I was curious whether any Labor candidate had ever won from a primary so low - Malcolm Baalman very kindly checked this and Labor won Riverina from 21.68% and Calare from 26.57% in 1940, and there was also the very comparable case of Richmond in 1990 from 26.68%, when Helen Caldicott as a green independent almost took the seat.)

On the night, Melbourne Ports was an easy Labor retain to the world of mass-consumption election coverage since Danby leads Guest on the two-party preferred vote by 52.7% to 47.3%, which seems an unassailable lead (though we don't even know that for certain, see below).  However during the campaign there has been a lot of attention on what would happen if the Greens overtook Danby (with two commissioned polls suggesting they could do so).  In almost any other seat, Hodgins-May would win easily under this scenario, but Melbourne Ports is different, as Israel/Palestine issues make Danby a stern critic of the Greens.

During the campaign Danby stated he would preference the Liberals and at times made good on that by distributing how-to-vote cards that did so, despite head office disapproval.  At other times it was clear the approved HTV was being handed out, but I don't know what the proportion of "normal" HTV cards was.  Hodgins-May was also criticised furiously in the local Jewish media (or some of it at least) after declining to attend an event hosted by Zionism Victoria.  The Greens attacked Danby repeatedly, alleging that his preference suggestions could cause a Liberal Government, and as things stand ... you can see why they were saying it.

Someone may have information that shows Danby will win this easily and if so I would welcome it, but I think this seat is well worth its own thread as another sign for why we should be careful of two-party calls by the computer on election night, and for comparison with Prahran.

The primary vote

As of July 3, primary votes in Melbourne Ports are:

Guest (Lib) 40.92%
Danby (ALP) 27.11%
Hodgins-May (Grn) 25.14%
Smyth (Animal Justice) 1.90%
Holland (Ind) 1.70%
McKenzie-Kirkbright (Drug Law Reform) 1.42%
Von Doussa (Marriage Equality) 1.37%
Myers (Ind) 0.45%

The on-the-night primary vote in Melbourne Ports saw only 60.44% counted, with just a few mobile booths outstanding.  In 2013 the ordinary count was 65% (turnout reaching 90%) so there has been a somewhat increased prepoll.

Melbourne Ports is already noted for high rates of pre-poll voting because of the election being held on the Sabbath.  It might be thought this would help Danby, and it probably helps him a bit, but not that much.  In 2013 the net changes from the booth result to the final primary result were a gain of 1.38 points primary vote for the Liberals, Labor down 0.62 points, Greens down 0.95.  On this basis the prepolls (25% of the total vote) caused a 1.74% swing in the two-party preferred vote.  If this swing is extrapolated to a c. 30% prepoll count, Danby's 2PP lead over Guest would be pushed down to about 50.6, so if postal voting strengthened still further for the Liberals they might even catch Danby on the 2PP.  However I am sceptical that this will happen as usually an increased non-ordinary count is accompanied by more representative voter behaviour, and as the patterns in Melbourne Ports are extreme enough already.  Also I don't yet know which category the increase in the uncounted vote mainly falls in - if it is mostly absents or out-of-electorate prepolls then it won't make so much difference.

Danby's primary vote lead over Hodgins-May shouldn't, based on the above, increase much.  It might be stretched towards three points should it increase at all.  The question then is whether the flow of preferences from the left-wing micro-parties Animal Justice, Drug Law Reform and Marriage Equality, as well as the independents, can get Hodgins-May in front of Danby.  Peter Holland, the higher-polling Independent, ran mainly on the single issue of the Palais Theatre, and has preferenced Danby on his website.  Animal Justice and Marriage Equality both preferenced the Greens (though their level of HTV activity and preferences in Melbourne Ports are not known to me at this stage) and Drug Law Reform preferenced nobody, but if you vote for a party called Drug Law Reform there's a fair chance that we know who you might preference.

Ambiguously Gendered Child Of Prahran

We've actually seen a film a lot like this before.  In one of the most exciting postcounts I've ever seen, the Greens pulled off a win in the overlapping state seat of Prahran from a rather similar position.  Trailing Labor by 1.1% on primaries, the Greens outlasted Labor in the cutup by 0.11%, with the turnaround especially down to a 57.6% to 19.0% split of Animal Justice preferences (2.3%).  If Hodgins-May could get a similar gain off the three micro parties mentioned and hold steady on the independents (which might not happen) she would reduce Danby's current lead to 0.2 points.  But in this case the voters for these left micros are not choosing between the Greens and a left-wing gay Labor candidate; they are choosing between the Greens and Michael Danby, who has red cards from the activist left on Israel and preferencing Liberals, and at least a yellow card on same-sex marriage.  It might therefore come down to what level of handout effort these micro-parties were making (if any) and what happens with the balance between Danby and Hodgins-May in the remaining votes.  It might be super-close, or a larger than expected boost to Danby in the postcount may just kill it off.

At the start I'm ever so slightly sceptical about whether the gap will be closed, but what would I know?  I initially didn't think the Greens would get the flow they got in Prahran.

If Danby Is Eliminated, What Then?

If Danby is eliminated, on current primaries the Greens need 73% of all preferences to beat Guest.  In Prahran they needed 83% and got 84%, but because of Danby's preference antics and the nature of his support base it is not clear they will even get 73% here.  The serious problem for the Greens is the likely further drifting in the postcount.  If I apply the same primary vote corrections as in 2013 then the share required goes up to 77%, and it might be it even goes higher based on the higher postcount.  That sounds hard because of the Danby factor, and might have been hard even without it.  (I have heard that the preference flow they are actually getting from Labor is 77% based on a large sample of booths, and it would probably be slightly stronger from the micros, so if this actually happens it could be very tight, but the increased postal rate could be the problem.)

What next?

 Many declaration votes will be thrown and will start being added from probably Tuesday.  These may put the seat out of question fairly quickly, especially if they are postals and help Danby.  On the other hand, it's possible the 2CP makeup won't be known until the throw of preferences (as was the case in Prahran).  In the latter case the seat could take a long time to resolve.

Real scrutineering sample info from verifiable sources (confidentiality guaranteed) on this contest is very welcome and can be emailed to k_bonham@tassie.net.au, or sent to me on Twitter (@kevinbonham).  Please no fake stuff this time! Ta.

Update: Wednesday

Today the first postals (1520 of which 38 were informal) were added and they split 53.37% Liberal, 28.07% Labor and 12.28% Green.  The Liberals performed 12.5 points better than their ordinary vote (compared to 9.1 points in 2013), Labor 0.9 points better (compared to 1.3 points worse) and the Greens 12.7 points worse (compared to 7.3 points worse).  On a 2PP basis Guest (Lib) performed 10.4 points better compared to 9.7 points in 2013, an insignificant difference, and I often notice in these counts that the very first postal batch breaks very strongly to the Coalition.  Such strong postals are consistent with Melbourne Ports finishing up as a very marginal seat but not with Guest actually winning the 2PP against Danby.  The gap from Danby to Hodgins-May is up to 2.5 points already (which could be difficult to bridge on the micro votes) and the winning target for the Greens to 74% of preferences.  So the chances of the Greens winning this seat seem a lot weaker after adding these postals (especially with so many more to go).

Update: Thursday

More postals were not much better for the Greens who now trail Danby by three points, and there are still a massive number of postals to go.  I am now assuming (but not calling) that Danby will win the seat and will only return to this thread if there is something interesting worth adding or I detect signs of life later on.

Update: Monday 11th

There has been some belated media interest in the seat.  The report says the Greens are "becoming more confident" they can get into second, but no evidence is given on the key question of the gap between Danby and Hodgins-May, which has already expanded to 3.4 points. The current preference flow required for the Greens if they are second is 75.3% but I project this to increase.

I have done three projections, which are very rough because of the usual uncertainties about whether all votes in particular categories will be received:

* On the assumption that postal vote flows end up the same compared to the ordinary vote as last time (so far they are much better for Labor and weaker for the Greens), I would expect the primary gap between Danby and Hodgins-May to finish around 2.2 points - this would give the Greens good chances of making the final two.

* On the assumption that postal vote flows end up the same as they have so far, I would expect the primary gap between Danby and Hodgins-May to finish around 3.8 points - this would make it almost impossible for the Greens to make the final two. It might even be higher.

* On the assumption that postal vote flows end up more or less in between the expected flow based on the ordinary vote swings and the flow so far, I would expect the primary gap to finish around 3.0 points.  This would make it very difficult for the Greens to make the final two.

In all these cases I project the final percentage of all preferences required by Hodgins-May at about 77.5%.

Of course stronger or weaker relative flows on absent or other votes could interfere with this.  However even in the worst case above it is still in-theory possible the Greens might get into second, depending on the strength of flow from micro parties.  In the best case it is reasonably likely they would do so, but this requires remaining postals to be better for the Greens than last time.

Monday 8:30 pm: Some postals and absents were added today and the postals were extremely favourable for Guest, taking his share of all postals over 50%.  The absents also helped him (even splitting in his favour!) but that does not mean anything without knowing where they're from.  Strong postals for the Liberals are bad news for the Greens as they increase the share of preferences the Greens require if second.

Wednesday 3 pm:  More postals and absents have been added.  The patterns previously seen on postals have moderated slightly with the Labor-vs-Green gap down from 15.3 points down to 14.4 points. On absents, the gap is four points to the Greens, exactly in line with the projection with half of them now counted.  The Greens need the gap on postals to continue coming down, because on current projections the Labor-Green gap will blow out from its current 3.17 points to something upwards of 3.7 and perhaps even well over 4 (depending on how many remaining postals arrive).  While the gap has closed, the Greens' failure to outperform the projection on absents is taking away a lot of the remaining doubt here.  As for the required rate if the Greens are second, it is now 76%.

I have seen a claim of very high card-follow rates among the micro-parties, but it was not supported with figures despite me requesting them.

Wednesday 10 pm: The Labor-Green gap on postals has continued to moderate (now 14.01 points) but so has the Green-Labor gap on absents (now only 2.81).  The overall Labor-Greens gap has risen slightly again to 3.25 points (with 7.71 points of micro-party preferences now available) and the required rate if the Greens are second is 76.2 points.

Thursday 11 pm: Most of the out-of-electorate prepolls were counted today.  The Greens' performance relative to Labor was very slightly worse than the 2013 projection suggested, and although the overall Labor-Greens gap is now down to 3.05 points (out of 7.24), the likelihood is still that this will blow out substantially when all postals are included.

I am getting quite a range of views on this seat from interested observers, from those who can't believe I haven't just called the thing already to a minority who think a Danby loss is still seriously possible or even likely (in one case, certain!).  While this divergence of views persists I think it's best to keep an eye on it.

Preferences will have to be distributed to know the outcome in the seat, but it shouldn't take that long to throw 4500 or so of them to get to the point where we will know for sure whether Danby has won or lost.

Friday 1:15 pm:  My current projection is only finding the Labor-Greens gap stretching to about 3.25 points assuming current postal trends continue (which they might not).  As noted by CJ McDonald in comments, an interesting feature of the Absent count so far has been the large increase in votes for the three left-wing parties that may well act as feeders for the Greens.  The collective Others vote is now on 7.24% and I project it to finish up around 7.35%.  Moreover the proportion of it provided by the left-wing micros has risen from 69% to 71% and is still rising.  The percentage of all the Others votes that the Greens need to beat Labor by is currently 42%, and I project this could rise, but not much (maybe finishing at 44%).

Based on past "similar" cases that still seems like too much - the independent Holland's votes presumably won't break strongly especially with the donkey vote on board (and might even assist Danby), so we could be looking for 70-20-10ish splits or more from the left-wing micros.  But because of the Danby factor and given that there are no previous data at all for the Marriage Equality party, I can't completely rule out that even that might happen.

It should be noted that Michael Danby has said his scrutineers say there is "no chance" the Greens will overtake him.  What it interesting there is that he refers to something that may stand him in good stead when the AJP preferences are thrown - he has indeed made lots of noises about animal welfare.

Friday 5:25 pm:  Another batch of postals have been thrown and these continued the pattern that the later postals are less bad for the Greens than the early ones, with the gap on postals overall now down to 13.7 points.  The gap from Labor to the Greens is 3.22 points out of 7.27 (44.3%).  The share of preferences the Greens require if second is now 76.6%. There are still at least 1619 postals left.

I have seen an unconfirmed report that there will be a preference throw to attempt to determine second and third tomorrow.

Saturday: I've been sent a scan of an indicative 3CP throw which shows Danby outlasting Hodgins-May by 802 votes, suggesting Hodgins-May overall gained by about 0.29 votes/vote.  Subject to official confirmation, on this basis, Danby has retained.

(Incidentally based on the preference flow recorded, the on-the-night figures would have produced a tie (to the nearest .01%) between Danby and the Greens.  The postcount, and especially the woeful postals, has done for them.  Also note that the above is not the final margin as some votes are to be added, but there is no prospect of the remaining votes changing the result.  A second source has also advised that Danby won this 3CP count by this margin.)

10 comments:

  1. If I might comment on Melbourne Ports, where I lived for 30 years. There’s a common misconception that Michael Danby gets all the Jewish vote and that’s why he hangs on to the seat. In fact most Jews vote Liberal, although Danby certainly gets more Jewish votes than any other Labor candidate would. This is relevant when we consider the 17,000 outstanding postal votes. These are cast mainly by religious Jews who can’t vote on Saturdays. The religious Jews are more conservative than the secular ones, so these votes will favour the Liberals (by 54 to 46% in 2013). But they will favour the Greens least of all, since very few Jews will have voted for a party they see as anti-Israel and therefore anti-Jewish. So if the critical question is whether the Greens can overtake Labor and come second, the postal vote will work heavily against that happening. That’s why the Danby camp tell me that they are confident of retaining the seat. AC

    ReplyDelete
  2. Libs are up 4.3% on TPP on the first batch of 1500 postals, not sure if it will be enough but if that holds it will be close. Postals are up from 13,000 to over 18,000 this election, absents are up from 5,900 to 7,900.

    Greens look gone on this batch, they're down 1% on on their 2013 effort.

    ReplyDelete
  3. Clearly we need to stop old conservative people voting by post!

    ReplyDelete
  4. I doubt there are many more postals to be counted - they haven't yet been received. Isn't the deadline for receiving postals 13 days after election? (ie today?)
    I'm still calculating +44% 3PP required. While the latest votes added saw GRN doing worse than projection, "Others" did much better than projected hence the 3PP% hardly changed.

    ReplyDelete
    Replies
    1. There are 3482 postals awaiting processing, of which some small proportion may be rejected. There are in theory a further 3739 outstanding but the great majority of those presumably won't arrive. In 2013 the average non-return rate was about 1000/electorate but doubtless in electorates with a huge request rate the non-return rate will be proportionally higher.

      Delete
    2. I am not sure the how the 'blow out' for the Greens from postals holds if there are also nearly 5,000 pre-poll decs. absentees and provisonals that also need to be counted. Surely the preference switch is still on - especially with the increase in "Others' on pre poll decs.

      Delete
    3. If 2013 is any guide the majority of provisionals will be rejected. 458 provisionals were accepted in Ports in 2013. There were changes in the last couple of election cycles that slashed the number of provisionals accepted. I've been following the slightly more advanced Herbert count where over time the "awaiting processing" figure has been going down and "rejected at provisional scrutiny" has been rising.

      The most important issue in terms of the Labor-Greens gap is that while absents, DPPs and provisionals help the Greens only slightly (thus far .028, .011 and I project from 2013 .007 votes per formal vote respectively), postals have been breaking at .14 votes per formal vote to Labor, and there are almost as many of them as we are likely to actually see of the other three groups combined.

      That said the estimates of the size of the blowout I'm now getting are smaller than before - I now get the final Labor-Greens margin at about 3.25 if nothing changes in the remaining postals.

      Delete
  5. My figures of 44% based on the following votes arriving with the same percentage primaries as we've already seen in 2016
    - postals (total up to 14621 of them, or an additional 3533)
    - pre-poll 1485 more, total of 5416

    Provisionals - 492 being accepted and formal with same post-ordinary primary swings as in 2013

    AFTER these are added I get the 3PP required split at +44%. For Prahran, the total 3PP split was:
    GRN: 49% (for a net split of +27%)
    ALP: 22%
    LIB: 29%

    The Prahran split for Animal Justice preferences was:
    GRN: 58% (for a net split of +39%)
    ALP: 19%
    LIB: 23%

    However, other 3PP splits in Melbourne, Batman and Ports in 2013 showed a +25% for GRN only from Left candidates.

    TruthSeeker assessment: ALP leading... Just!

    ReplyDelete
    Replies
    1. Yes I just also projected 44%.

      In terms of splits off third parties a bit of an eye-opener is Wills (federal 2013). Socialist Alliance 58-11, independent 60-12, Sex Party 57-22. In the first two cases there were more than three parties in the mix and the flow from the indie included donkey votes. Sex Party is not a straightforward Greens feeder either, because of stoushes with the Greens over their preselection of candidates who support the "Nordic model" of banning the purchase of sex. Overall the 3PP flow to the Greens was about +28 but that includes Family First and PUP, so take those out and throw in the Danby factor ...

      Delete
  6. I am not sure where this belongs so I'll just put it here, since we're talking about 3-cornered contests.

    The preference distribution has seen the Liberals take second place in the seat of Melbourne. So instead of being a marginal Grn v Labor, it will be a safe Grn v Liberal seat going in to 2019.

    ReplyDelete