2025 House of Reps Postcount: Melbourne

MELBOURNE (Green 6.9% vs ALP)

Adam Bandt (Green) vs Sarah Witty (ALP)

Labor win after 2CP realignment

WARNING: Explaining what is going on with the Melbourne count is complex.  This page is rated Wonk Factor 4/5

Click here for link to Reps postcount hub and tallyboard page.  

-----------------------------------------------------------------------------------------

I am doing postcount threads in a capricious order as something provokes me to jump to a different seat next and en route to the fine mess that is Monash, the ambulance has been rerouted to Melbourne.  This follows the ABC projecting the seat as 52-48 to Labor after projecting it as 51.5 to the Greens last night.

Why are we even here?  For some reason the AEC preset the Melbourne count 2CP as Greens vs Liberal, a baffling decision when Labor beat the Liberals at the 3CP point by 8.6% in 2022 and when a Greens vs Liberal 2CP is about as informative concerning who will actually win the seat (even post-redistribution) as the 2CP between One Nation and Tim Smith.   This is the single worst 2CP preset decision by the AEC that I have ever seen - generally I am a great fan of all the work done by the AEC but this one's an exception; this has let election watchers down and needs to not happen again.  

As a result we had no useful election night preference estimate between the Greens and Labor in the Greens Leader's seat. We now do have a preference count for 3926 postals and 150 hospital votes, which shows a massive Labor lead, but the postals are unrepresentative.

The current primary figures with just 62.1% counted are:

Bandt (Green) 41.39
Witty (ALP) 31.35
Hunt (Lib) 18.35
Koutoufides (IND) 3.14
Casey (ON) 2.22
Huang (Fusion) 1.93
Smith (Ind) 1.63

Melbourne was redistributed, which made it less favourable for the Greens and made adjacent Wills more so.  There is a small primary vote swing away from Bandt (currently running at just over 3%) and a large swing (running at nearly 6%) to Labor.  There is also a less supportive mix of minor candidates for Bandt, and all that adds up to trouble.

In the postal sample, Bandt has 26.08%, Liberals 29.87%, Labor 33.21%.  The 2CP is 64.34 to Labor, meaning Labor is getting 76.5% of preferences in a sample that is slightly more Liberal (73.3% of non-2CP) than the count as a whole (67.3%).  If Labor gets that flow rate across the count as a whole, that's a 52.2% 2CP to Labor for the current count as a whole.

The ABC notes that these postals are a highly Liberal sample, but what's more notable to me is that they're a bad sample for the Greens.  In general a party does better on preferences in booths (or postcount samples) where its primary vote is higher.  This pattern can be seen in the 2022 results for Melbourne using the standard graph I use for checking for booth preferencing errors:

In the case of the counted postal samples, Bandt is polling 15.3 points below his average for the seat of Melbourne as a whole.  In 2022, in a seat where Bandt was polling 15.3 points below his average primary vote, he would according to the relationship above received a preference flow that was 8.6 points below the preference flow he actually received.  If that applied to the 2025 case, Labor would receive 67.9% of preferences across the votes counted so far, which would mean that on the current primaries Bandt would be ahead, albeit by a minuscule 50.15-49.85.  So before we take any notice of the ABC's projection, we need to see more votes from across the whole seat to see what the flow is like outside of an unrepresentative subsample of unrepresentative postals.  (Early postals are usually more conservative than later ones - in 2022 Bandt's primary in all postals was 5.2 points below his vote across the electorate, and his preference share was 2.0 points below).  I will post some projections as we get more booths realigned.  Perhaps the relationship will be less steep than last time for some reason (possibly the lack of AJP and Vic Soc candidates as preference sources) and Bandt will still come out in trouble, perhaps not.  

The other thing to note here is that the Melbourne primary count is very incomplete.  It is currently at 62.1%.  There are probably still 12,000 or so postals to come and in 2022 there were over 6000 absents and over 7500 out of division prepolls.  That said the absents and out of division prepolls had little difference from the ordinaries in 2022 and a likely shift to more of the latter could be a problem for Bandt.  The early postals being very very bad for Bandt may also not be a good sign concerning what's to come.  

Updates will follow and the outlook will be updated over coming days.  

Monday 9 pm: The postal count has been updated and it has come down slightly to 64.18% 2CP (now 3777-2101) but it is not coming down as fast as Bandt would like given that nearly another 2000 went in.

Tuesday 1:30 pm: Booths are being realigned making a live count estimate for this seat more useful.  I now project Bandt to be at 50.57% in the live count.  

Tuesday 3:21 pm: Things looking much worse for Bandt now; he has had some bad preference flow booths including an 83% flow to Witty in the Burnley booth.  I now project him at a mere 49.1 which would get worse after postals.  Advantage Labor.

Tuesday 6:05 pm: More incredibly bad booths for Bandt, in Fawkner Park he got only 11% of preferences! I now project him at 47.6 and I think Labor has won.  It does occur to me that absents could be better for the Greens than before on account of previously strong areas being moved out of the electorate, but that won't be enough to change things. 

Wednesday 12:30 am:  In a blog post Ben Raue noted that the flow to the Greens in the one prepoll booth in this seat is very strong at 47%.  Well I am always keen to check my reasoning (this is what good scientists do, we attempt to destroy our own hypotheses) so I checked this out.  Could there be a strong difference between prepoll and on-the-day preference flows for whatever reason that would save Bandt?

Answer: no.  Checking the 2022 Melbourne flows by booth there were two small prepoll booths that stood out as outliers on the Green vote vs preference share graph while the larger prepolls were relatively normal.  In 2022 these were Footscray MELBOURNE and South Yarra MELBOURNE, in 2025 the culprit is Brunswick MELBOURNE.  What do these have in common?  Answer - while counted as ordinary votes they're not actually in the electorate!  So Liberal voters voting in them would most likely not have seen the Liberal how to vote card for Melbourne, resulting in a higher flow to Bandt.