Friday, February 11, 2022

New South Wales Feb 12 By-Elections Live And Postcount

Bega (Lib 6.9%) - CALLED 8:48 pm Holland (ALP) gain.
Monaro (Nat 11.6%) - CALLED 8:02 pm Overall (Nat) retain.
Strathfield (ALP 5.0%) - CALLED 10:50 pm Li (ALP) retain.
Willoughby (Lib 21.0%). CALLED following Saturday as James (Lib) pulled ahead on postal votes. 

March 1st: James ended up winning 53.3-46.7.  Preferences split 38.5% to Penn, 8% to James, 53.5% to exhaust.  Had preferencing been compulsory, Penn would have needed 69.8% of the votes that exhausted to flow to her to win, so it would have been much closer (hard to say if it would have happened or not.)

Friday 25th: James now ahead 53.5-46.5.

Saturday night The margin is now out to 53-47 as James continues to get around 55-45 on the postals so far.  This may drop off with remaining postals and I expect not much more change from here.

Saturday 1:19 Votes showing on the ABC site (but not yet on NSWEC) confirm Liberals will retain as the margin is out to 52.8-47.2 after the first 10000 postals.

Thursday 7:15 The projected gap has closed to 51.8 to James (ABC has 51.7) mainly off a very strong preference performance by Penn at the large North Willoughby EVC prepoll, a .482 votes/vote gain rate.  This wasn't echoed in the other prepoll added so far which had only a .230 gain rate, but should make Liberals nervous about the preference split on postals if they are not otherwise favourable.

Tuesday 6:55 The projection has stayed around 52-48, based on which Willoughby can't be called yet and we will need to wait for postals.  I want to highlight this comment on Twitter as an example of the things we don't know about this count.  We have seen elections with high postal counts under COVID before, but not with widespread community spread, so we shouldn't rule out that the postals might even skew left until we see some.  They probably won't, but we'll see.  We also don't know whether preferences in postals might be stronger than the booth votes, given the lackadaisical approach to preferencing on how-to-vote cards - though that alone will not be enough.
Tuesday 5:40 The Willoughby re-alignment has progressed.  Ignore the raw totals from the booths counted (which currently have Penn very narrowly behind) as these are from unrepresentative booths.  What to watch for is the rate at which Penn gains on James.  In the six booths counted so far it has been around .320.  She needs .464 and has (unsurprisingly) not been near that in any booth to date.  My projections so far match the ABC's, ie 52-48 on current primaries, but it is possible some way of projecting this better based on good and bad booths will emerge (possibly in James' favour).  

Tuesday 12:30 With all prepolls added in Bega, the swing to Labor has come back to 12.3%.

Tuesday: Slow progress here but in Strathfield, a very favourable Burwood prepoll has pushed Li out to 55.5% 2PP, now a slight swing in Labor's favour.  

Sunday: I've remembered that thanks to the wonders of NSWEC data entry I can look up a Liberal vs Penn 2CP from 2019 (it was 73.75%) and find the preference flows from that election for those candidates from everyone else (Lib 11.6% Penn 35.5% rest exhaust).  Applying that to the current primary figures I get the same projection as the ABC (53.1% to James.)  There are some reasons why this might be wrong in either direction - the absence of an ALP candidate and James' lower profile on one hand, but also the lower rate of how-to-vote preferencing.  Incidentally the Liberal vs Green 2CP swing is currently running at 7.5%.  The 2019 preference flow for these parties from all sources combined was 37.4% to Greens, 11.6% Liberals, rest exhaust.  The 2022 flow on votes so far is 32.7% to Greens, 13.2% to Liberals, rest exhaust, meaning that it is actually slightly weaker (but this may well be as a result of Liberals switching to Penn).  It would be possible to extract 2022 flows by party between the Liberals and Penn but I haven't so far done this.  If the ABC's projection is correct the 2CP swing from the Liberals to Penn is 20.6%.  However this is somewhat likely to decline on postals.  

Also a note on postals in Strathfield: in the pandemic so far there has been a strong tendency for the difference between on the day votes and postals to reduce when postals are a very large share of the vote.  For example in the Tasmanian seat of Rosevears in 2020, the extent to which the Liberals' postal vote outperformed their booth vote fell by 5% to 1.5%.  In Strathfield even if the difference holds at the 2019 level that probably wouldn't be quite enough to overturn Labor's lead.  

Late Night Wrap: All Over The Place

By-elections tend to result in media spin-fests but they are weak predictors of state election results let alone federal, and it is easy to rave about swings that would be notable at a general election but happen quite a lot at by-elections.

My overall impression of this one is that there's not a lot to see here - the results are very variable between the different seats and both major parties have recorded a mixed bag.  No overall message should be drawn from these results in the context of recent NSW by-election history.  The 2PP swing in the three 2PP seats is running at an average of 6.7%, actually below the average of recent decades but still hardly a concern for Labor overall.  

In Bega, Labor has had a storming win with another popular local GP (shades of Freelander in Macarthur and in Tasmania Bastian Seidel) and weren't shy about exploiting his medical charm.  Large as it is, the 14% swing is one we have seen pretty often against NSW governments in by-elections.  The seat has fallen to Labor for the first time since its 1988 re-creation, but it was on a fairly slender margin and seats often fall for the first time in by-elections.  As noted in my intro, the seat only became vaguely competitive for Labor in the 2010s (in which time Labor hasn't won an election), so the long period of Liberal tenure means nothing.  

In Monaro, the Nationals have done very well indeed (or Labor badly) to get away with only a 6.6% swing against them after the loss of John Barilaro and in a historical bellwether seat.  It's likely candidate choice was a big factor here.

In Strathfield, there has been very little swing.  Swings tend to be smaller in opposition vacancies (because the opposition is losing the personal vote of its departing MP) but all the same this seems lacklustre for Labor, or good for the Liberals.  The competition from Elizabeth Farrelly (with many of her votes exhausting) may be a factor.  

Finally in Willoughby much has been made of the 13.5% swing against the Liberals but primary vote swings of this size are common when NSW Premiers depart: 12.5% for Willis, 22.2% for Wran, 14.6% for Greiner, 5.8% for Carr, 15.7% for Iemma and 24.2% for Baird.  The exception was Kristina Keneally whose departure saw a 17.7% primary vote swing to her party (but off the baseline of the 2011 calamity and with no Liberal candidate contesting.)  No-one should be surprised that the Liberals have been taken to preferences here.  The surprise, however, is that the vote for other candidates was concentrated enough in one opponent (Larissa Penn) to make a reasonably close result possible.  That by itself is no reflection on the government.  

Of other competitors, that one independent performed well (as an inconclusive taster for the climate indie attempts to come), but nobody else made much impression.  The Greens' result, albeit in unsympathetic seats, was lacklustre and anti-lockdown/mandate candidates polled insignificantly.  

We could have seen something different here, as a backlash against Liberal governments either state or federal, but the backlash here is just within the normal by-election range.  The loss of Bega is a nuisance for the government, knocking it down to 45 seats out of 93 (with two more held by ex-Liberals who are reliable votes with it), and meaning it needs to find a vote somewhere else to pass legislation.  However, the government is not close to falling.  They will now face the challenge of winning the seat back, the size of the swing suggesting this won't be straightforward.

Updates will appear here with the latest at the top

10:42 To win, Penn currently needs a gain rate of .464 votes per vote.  For comparison, in Lismore 2019 Janelle Saffin (ALP) managed a gain rate of .474 and won.  But in that case 70% of the preferences were from the Greens (50% here), and the Greens were preferencing Labor with a high rate of on the day voting and hence how-to-vote exposure.  In this case very few voters will have seen a how-to-vote card and the Greens, for whatever reason, have not preferenced anybody.  It's not realistic that Penn could win if the primary vote numbers stayed as they are.   

10:37 But now they bounce back winning a second Strathfield prepoll by over 30% on primaries.  That should put them back where they were, provided that it is not an error.  

10:20 Labor now down to 53.4 after that Strathfield prepoll.  Inclined to wait a little longer though in all probability they're still fine.  

9:52 First prepoll in Strathfield.  The Liberals have done pretty well in this one, with a 43.1-40 lead, compared to 39.0-49.4 on prepolls overall last election.  But because their preference performance in this seat is poor, all that will do is water down Labor's lead on the day.  

9:46 The third prepoll in Willoughby is in.  The swing in the Liberal vote in prepoll overall is the same as in the on-day voting, but the prepolls are a few points less concentrated in Penn as the main opponent.  The ABC is projecting a 53-47 win to James, which seems much more likely than the Liberal estimates, although it may not be that close.  

9:39 The swing in Monaro is dropping with more prepolls added.

9:05 Still waiting for first prepoll in Strathfield.  I have had a report of a high exhaust rate in Willoughby.  Labor and the Liberals respectively have claimed these seats; still waiting for official figures based off which we can confirm this.  

8:48 After a long wait the first prepoll in Bega has matched the booth swings so I'm giving that to Labor now.

8:20 The second prepoll in Willoughby has almost identical results to the first (50.4-21.6) but there isn't a matching booth for a swing.  James' lead is growing and this would now be not easy to overturn under CPV.    I should note that the swing against James does not surprise me at all; the surprise is that the vote for other candidates is concentrated enough in one to make that candidate vaguely competitive.

8:10 First prepoll in Willoughby and a 12.1% swing against Liberals with 13.7% to Penn - at the lower end of the swings Penn is getting.  

8:03 First prepoll in in Monaro and there's basically no swing there so I've called that seat.  

7:55 We're in that teatime of the psephological soul that is the wait for prepolls.  Made longer by a faster on-the-day count.  

7:40 Willoughby: 42.8% to 32.6%.  Trent Zimmerman has said they have a 61% 2CP in scrutineering but that's extremely hard to see off those primaries.  

7:35 Extremely hard to see the Nationals losing Monaro.  I want to see one prepoll of decent size before calling it.  

7:25 ABC website has called Bega with a projected 56-44 result.  I'd like to see a few prepoll booths.
7:15 The swing in Bega has bounced back up and is now at 14% on the ABC projection.  The on-day votes need to be very unrepresentative for the Liberals to retain the seat.

7:11 The swing in Bega is dropping, now down to 9.8% (still enough but how will it go when other vote types go in).  Trent Zimmerman says that the Liberals expect the Willoughby swing dropping to 13% from the current levels.  The Nationals are looking fine in Monaro. 

7:01 No projected 2PP swing in Strathfield at the moment but let's keep an eye on how that goes on prepolls.  Labor has a net loss to Farrelly. 

Note that the NSWEC are counting Willoughby as Liberal vs Green, so it will have to be re-aligned.  

6:55 In Willoughby better booths have come in for Tim James with the swing against him currently down to 13.6% primary but with a 22.5% swing to Penn.  If that holds up it projects to about a 43.4-32.4 lead, which would be difficult under compulsory preferences, but James would probably just be OK under OPV.  Notably there is a very low Liberal Democrat vote and most of the preferences are Green or Reason so this is still interesting at this stage.

6:52 In Monaro with 4.1% counted the swing against the Nationals is substantial (around 9% on primaries) but it is spraying and is not currently as high as Labor needs.

6:50 Votes pouring in in Bega and at this stage there is a massive swing to Labor which is so far up 10.3% with the Liberals down 13.7%.  But how representative are the on the day votes?  

6:38 We have votes.  In Bega there's about a 10% swing in early primary votes.  There is also a strong vote for Penn in Willoughby.  I can't find the booth St Mark's Anglican which Penn has won in the 2019 results but the swing against James in the Chatswood Guide Hall booth is only 12.2% which would be harmless.  A third booth Naremburn has an 18.8% swing against the Liberal primary.  This is going to take a while to settle down and especially we will need to see prepolls (where the Liberals claim their primary is holding up).

5:35 Antony Green has posted that the NSWEC will stop counting pre-polls at 10:30 and that counts are unlikely to pass 40% tonight.  That could still give us a good idea, but with reports of hardly anyone voting at booths, beware the possibility that booth votes could be very unrepresentative.  Prepolls may give a better guide.  Prepolls will be finished on Monday.  



Welcome to my live (-ish) thread for the four Feb 12 NSW by-elections.  This thread will follow the four by-elections at least until the results for all are clear, and with any notes of interest beyond that.  Coverage will start from around 6:30 pm Saturday night.  Once numbers come in refresh maybe every 10-15 mins for updates.  

I am not sure yet how detailed this coverage will be as I am in the process of moving house at the moment (a rare and protracted business in my case) but I will try to provide useful information as the counts unfold.  These will be especially slow counts because the NSWEC has mailed out postal votes to every elector following the iVote debacle in recent local government elections.  The iVote system failed to cope with demand, resulting in enough voters being potentially disenfranchised for the NSWEC to petition against three results, and several others being capable (with an apparently low probability) of being affected.  The postal votes cannot be counted until Feb 19.  With the postals already at 28.1% in Willoughby, 26.4% Strathfield, 21.3% Bega and 15.3% Monaro, and many more postals still to arrive, if seats are even remotely close on the night we will be waiting at least a week to get a much clearer idea of them.    This said, experience with other pandemic counts with huge postal tallies has been that the postals are more like booth votes than normal and don't do very much.  The broader use of postal voting dilutes its normally conservative-leaning nature.

Expectations for the by-elections have been that the government will win Willoughby (Lib 21.0%, ALP not contesting) easily and also hold Monaro (Nat 11.6%), that Labor should retain Strathfield (ALP 5%) but might not, and that Bega (Lib 6.9%) is in serious doubt.   (The betting odds for the incumbents are 1.02, 1.08, 1.40 and 1.65 respectively, and there have not been any useful polls reported.) 

Some of the arguments about these seats have seemed pretty woolly.  For instance, Bega is supposed to have been gradually slipping from the government for some time for demographic reasons, but in fact the previous drift of Bega towards the state 2PP stabilised at the 2015 and 2019 elections.  In the last three elections it has run 4.4, 3.9 and 4.9 points above the state 2PP.

The real reason the government should be concerned about Bega and should probably consider a win in Monaro by any margin acceptable is the feral history of by-elections in NSW since 1999.  In Bob Carr's first term (1995-9) his government did very well in by-elections, taking Clarence on a double-digit swing and almost taking Orange in a similar result.  But since 1999 almost every NSW by-election that has had a 2PP contest has seen a double-digit 2PP swing against the government, even when the government was only headed for a small swing against it at the next election.  (I make it thirteen out of fifteen, the exceptions being the artificial Northern Tablelands case with an independent vacating and last year's Upper Hunter contest.  Both these saw 2PP swings to the incumbent government.)

Why have recent NSW by-elections mostly seen huge 2PP swings against sitting governments?  One reason is optional preferencing.  Under OPV, a swing between the major parties on primary votes is slightly amplified as a 2PP swing, because of exhausting minor party votes.  But a bigger effect that is seen sometimes is when one side loses primary votes to minor parties.  Most of these votes don't return as preferences under OPV.  

The history of NSW by-elections does suggest that a little nervousness about Monaro should be in order no matter how good the feeling on the ground - after all this is a previously marginal seat where John Barilaro's larger-than-life leadership of the party saw a massive 9.1% swing to his party in 2019.  The seat is, as Antony Green notes, a traditional bellwether, and bellwethers are easily lost in by-elections.  The Nats will have to hope that the qualities of their candidate Nichole Overall, and/or the amount of pork thrown at the district under Barilaro, see them safely over the line.

As concerns Strathfield, while Jason Yat-sen Li does hail from out of the electorate, a win would be such a remarkable result for the government all things considered that it's hard to see the seat as in play, especially against the current federal backdrop.    There is some speculation Elizabeth Farrelly's independent run could take votes from Labor but I am not sure that Farrelly will poll such a big vote after a rather strange campaign, or that she will draw that much from one side rather than the other.  In Willoughby it's hard to see any particular opponent getting enough primary vote to be a serious threat to the Liberals (the indie Larissa Penn is the most interesting to watch as a test of concept for climate indies in the federal election), though it won't be surprising if their primary vote takes a large whacking in the absence of Gladys Berejiklian.  That said, the seat was nearly won by an independent mayor when Berejiklian first won it.  

Something else that is interesting about these by-elections is the absence of any really significant attempt by anti-mandate, anti-lockdown or anti-vax candidates.  One would have thought if these movements had any serious support a by-election would have been a great chance to get a result to boast about, but nothing has come of it.  A touted campaign by Craig Kelly-backed far-right Russian nationalist Simeon Boikov for Strathfield did not result in Boikov's name making the ballot, but I have not seen any explanation as to why he's not a candidate.  

I have not looked at the impact of federal government drag on state by-election swings, but given that, for instance, state governments more often gain seats in by-elections when in opposition federally, it is likely to be an unhelpful factor for the state government.  Given everything I have said above, if the government retains its three seats by any margin that should be considered an excellent result.  One loss would be unremarkable, even two losses acceptable if the swings aren't too enormous, but three would be a serious disaster.  For Labor, given that Jodi McKay would not have had much personal vote in Strathfield as she was not leader at the last election, the target should be at least a few points extra on their margin.  By-elections are not that predictive of the election to follow, and any losses will only make the Perrottet government's life harder rather than bring it down.  

Finally, incumbent governments have performed very well in pandemic-era by-elections so far, but the pandemic has been around for a while and there is a question of whether it is still helping (especially with the chaos of the Omicron outbreak).  

See also Tally Room and Pollbludger  guides.


  1. Kevin, you've talked quite a bit about "federal drag" for state elections. Is there such a thing as "local drag"?

    Queanbeyan-Palerang essentially just elected a Labor council, with a working majority of Labor, Greens, and the ticket led by Kenrick Winchester, who's an ALP member that ran on an independent ticket. The ALP candidate is sitting member of that group

    Meanwhile the successful Nationals candidate is the wife of the mayor that retired at the last election.

    1. If there was then what I would expect is that parties would have swings against them at council level when they were in state government. It's not something I've looked at in detail, though I have looked at Brisbane City Council where it seems not to be much of a thing.

  2. Now that we know that the 2CP preference flow is roughly:
    Exhaust 47%, preference to Penn (IND) 43%, and James (LIB) 10%. I would love to see polling that asked people if they gave a preference between Penn and James, and if not, if in hindsight they did actually have a preferred candidate out of the two of them.

    My dislike of OPV is in part because I think there are plenty of people who are just too lazy in the voting booth the express a preference on independants because they think it's not going to matter. To a degree I think OPV also discourages people from researching all of the candidates. So, I wish there was someone who was doing the type of polling that would help test my hypothesis.

    I understand the argument that if there are lots of candidates on the ballot, it's unreasonable for people to know enough about all of them. But I think they can manage if it's, say 8 or fewer. So, perhaps there is a middle ground where the minimum number of boxes that has to be numbered is the same as it is the Senate (for consistency).