Monday, May 17, 2021

Tasmania 2021: What Was The Point Of This Election?

LIBERAL 13 LABOR 9 GREEN 2 IND 1 (Unchanged from before election)

This is another state election where I think a post-election washup post is in order, some of which will cover similar ground to the 2018 washup.  I started writing this a few days after the election but found it was hard to assess the result without knowing for sure who had won and lost.  I ended up rewriting this piece quite a few times and have stripped out a lot of commentary because the article was getting way too long.  

This election's been a crazy and fascinating ride.  The interest levels in it, especially Clark, have been extremely high on this site.  The post-election Adam Brooks news just adds more weirdness to the mix, and now we have a Labor leadership election as well (the first since Michael Field defeated Neil Batt in a party room ballot in 1988, and the first under new rules with voting by members and delegates).  

Why Even Bother?

Before the events that led to this election, Tasmania had a 13-9-2-1 parliament.  After Sue Hickey quit the Liberals and Madeleine Ogilvie joined them, it was still 13-9-2-1.  And after the election it was yet again 13-9-2-1.  All the election seems to have achieved is replacing Sue Hickey with Kristie Johnston and letting Labor get Dean Winter and Janie Finlay into parliament, without any renewal (unless one counts Ogilvie) on the Liberal side.  The Liberals sacrificed ten months of their term and risked a lot in the process, winning the final seat in Clark by just 2.2%.  Was it all worth it?  The artificial falling into minority and trying so not hard to get out of it?  The nonsense theories about the early election? The bizarre Adam Brooks storyline and the claims that it tainted the result?  The criticism of calling an election while the Opposition Leader was pregnant?  The criticism over undelivered promises and over outlining plans for the first 100 days only to be told they could have been doing those things instead of having an election?  All this ... just to get rid of Sue Hickey?

Well, in the case of Brooks it wasn't worth it. The Liberals would have won three in Braddon anyway (which is why the "tainted election" claim is incorrect, though Braddon voters have good reason to feel stuffed about by the scandal).  They didn't get that close to four, and the unseemly dissembling and dead cat throwing in Brooks' defence could all have been avoided by just not picking him in the first place.  It just turned out to be a very bad preselection.  But overall, the point of the election from the government's perspective is this: the next election is now due in 2025.  Winning majority government in Hare-Clark isn't easy.  If you can reset the clock and not have to worry about it again for a while, then you do.  As the government again has a one-seat majority, we will have to watch out for any further rogue MP issues, but the biggest problem (Brooks) has been avoided.  

A close election?
The Liberals only just won a majority in Clark, with a victory margin of 1429 votes (2.2%) over Sue Hickey.  But they were nowhere near losing any seat outside Clark, and with 12 seats they would most likely have governed anyway.  A two-party swing around 7% would have been needed to cost them a seat to the Greens or maybe Labor in Lyons.  

Winning majorities in Tasmania at all is challenging because it usually requires a primary vote in the high 40s.  The Liberal Party outpolled Labor by 20.7 points. On primary votes this is a third consecutive thrashing, which I estimate at 57-43 two party preferred.  This compares with my estimates of 60-40 for 2014 and 56-44 for 2018.  

However, the dynamic around minority government in Tasmania often drives lopsided voting share results.  Voters who do not want hung parliaments (especially those involving the Greens) herd to the party that appears to be winning.  So we can't assume that in a single seat system these Liberal governments would have won as heavily.   Since the Greens first captured the balance of power in 1989, primary vote margins exceeding fifteen points between the big parties have occurred at six of the next nine elections, compared to just two of the twelve up to and including 1989.  (There had been a batch of lopsided elections around World War II, as in some other parts of the country.)

Facts, stats and the odd interpretation ...


This is the first time the modern Liberal Party has won three straight Tasmanian elections.  An earlier but different party also called the Liberal Party won three straight elections, and in all the non-Labor side won elections continuously until Labor won in 1925.  However the non-Labor side did not always rule through this time (eg the Liberal government elected in 1913 collapsed with Labor governing from 1914-6).  

The Liberals' 13-seat result ends a streak of 17 consecutive cases where a state (not territory) government of the same party as the federal government at the time has lost seat share in the parliament at an election.  11 of these 17 governments lost the election.  The last case of a same-party state government not going backwards was WA 1996, but that was in both the state and federal governments' first terms.  The last case where a same party state government that was beyond its first term didn't lose seat share was Victoria 1976 and the last such case where neither government was first term (as in Tasmania this year) was in Victoria 1967. (Queensland Nationals 1972 gained seats but the parliament had been expanded, so that doesn't count.)  I think that this underlines the difficulty of winning a majority at all at this election, suggesting that without at least one and perhaps all of COVID, Labor being a mess and the emphasis on majority governments in this state, it wouldn't have occurred.  

Peter Gutwein's primary vote of 48.2% is the second highest in percentage terms of all time, behind only Doug Lowe's 51.2% (which was before the adoption of Robson Rotation).  His 32,482 votes was over 5,000 ahead of Will Hodgman's 27,184 in 2018, which in turn passed Doug Lowe's record for the highest number of votes by a candidate - a record that had withstood 39 years of enrolment growth.  


Labor's vote in Clark of 22.1% is the worst Labor primary vote in an electorate ever (the previous worst was 23.1% in Braddon 2014).  

This election was the third case since World War 2 of an opposition winning fewer seats than at the previous election.  This previously happened in 1979 (also an early election) and 2002 (first term government).  However, in this case Labor's seat loss came via a mid-term defection (Ogilvie) and they held all the seats they went to the election with.

Vince Taskunas has noted that Labor won only one booth (St Marys) north of Kempton.  Even in 2014 when Labor's vote was lower overall, Labor topped some northern booths it did not win this time, such as Waverley, Ravenswood and Acton.  In fact in the north in 2021 Labor won the same number of booths as Craig Garland did (Garland won Sisters Beach).  For non-Tasmanian audiences who want to know what "north of Kempton" looks like, it looks something like this:

As spotted by Mark Bacon on Twitter, Rebecca White polled exactly the same primary vote (16,338) in 2021 as 2018.


This was the first election since 1989 (and before that 1959) with primary vote swings against both major parties.  It also saw the largest vote for candidates who were not Labor, Liberal or Green (including "Green Independents") since Norm Sanders and Doug Lowe won seats in 1982.  

The Greens recorded swings to them in every electorate except the one where they turned out to most need a swing, Bass.  As it happened the final result in Bass wasn't even remotely close with Janie Finlay winning by 5.2%.  Had the Greens matched their swing in other seats and done so entirely at Labor's expense they may have just won Bass.  Although much was made (especially by them) of the overall 2.1% swing to the Greens, their statewide result was still 2.3 points below their historic average.

The lead Green candidate received less than half the Greens' vote in Bass (47.8%), Braddon (47.7%) and Lyons (45.8%), however none of these were as low as the 42.2% in Lyons in 2018.  Low vote shares for the top Green candidate make the Greens more susceptible to leakage, and indicate the problems that the party currently has with attracting high-profile candidates outside the south.  

Independent Kristie Johnston has been elected in Clark.  This is the first time an independent has won at the ballot box since the House was reduced to 25 seats in 1998.  Proto-Greens excluded, independents had been rare even in the 35 seat system, with Doug Lowe (ex-Premier, 1982) and Bruce Goodluck (ex-federal MP, 1996) the only exceptions since the 1950s.  (Kevin Lyons in 1969 was elected as "Centre Party", a Country Party branch rebadge.)

Justin Stringer (Australian Federation Party) recorded the lowest vote tally at this election, 150 votes in Clark.  However that's the highest statewide low score by a candidate since 1955.

On the exclusion of the Animal Justice Party candidate in Lyons, 32% of preferences flowed to the more or less diametrically opposed Shooters, Fishers and Farmers (who had drawn an adjacent ballot paper column).  In Bass the figure was 25.6% and in Franklin 16.8%.  In Clark, the Shooters were excluded first and 13.8% went to the AJP, who were in an adjacent column.  It appears the AJP attracts some votes from voters who either don't know what the party stands for (like the low-information voters who used to be attracted by Family First) or else really can't stand Labor, Liberal and Greens.  


Madeleine Ogilvie is just the second MP elected for both the Labor Party and a major non-Labor Party, after Carrol Bramich whose defection triggered the 1956 election.  (I frequently cited Bramich as a prior for Ogilvie winning but Bramich's win was far more emphatic.)  Several MPs have been elected for multiple conservative parties or for some party and as an independent.   Ogilvie incidentally has led a mostly charmed electoral life with close within-party wins over Amos (2014) by 331 votes, Cox (2019 recount) by 201 and Behrakis (2021) by 342, but also the one serious loss against Haddad (2018) by 731.  

All five incumbents were returned in Lyons.  This is the first such case since 2006 in which all incumbents were returned in all of Braddon, Franklin and Lyons.  All five winners at the 2018 election were returned in Braddon, but one of them (Adam Brooks) had resigned partway through the term, with his seat being filled by Joan Rylah then Felix Ellis before he very narrowly won it back.  Braddon has hence had a pass-the-parcel Liberal seat that has gone Brooks to Rylah to Ellis to Brooks and now presumably to Ellis again.  

35 Seat Conversion 

I've been asked a lot about the 35-seat result for this election.  My estimates (Liberal - Labor - Greens - IND) are as follows:

Bass 4-2-1 (close to 5-2 but Liberal vote leaks)
Braddon 4-2-0-1 (Garland) or 5-2 depending on candidate effects
Lyons 4-3-0 (because Labor would have three incumbents)
Franklin 3-3-1
Clark 2-2-1-2

Total 17-12-3-3 or 18-12-3-2

However in the 35 seat system the government would have realised Bass and Braddon were crucial to majority and strategised to increase their chance of at least one 5-seat result.  Some of the contests in this election, especially Braddon and Clark, really deserved a seven-seat contest and several of the candidates who missed out would have been deserving winners.  

Polling and pundit accuracy

This election saw a serious polling drought.  The lack of quality polling during the campaign makes it impossible to say (unless you trust reports of party polling) whether the government was ever on track to win by more than they did, though there have been no reports of internal polling otherwise.

The only actual poll released during the campaign was a uComms robopoll commissioned by The Australia Institute.  As the table below shows, the February EMRS was a better predictor of the election result than it was:

(MPG = major party gap)

Especially the uComms poll was inaccurate on the major party gap because it severely underestimated the Liberals and overestimated Labor.  Both errors were outside the poll's margin of error (in the Liberal case more than doubly so).  The EMRS poll was taken before the announcement of significant independents so had EMRS polled later they may have been even more accurate.  In uComms' case, the error may have been magnified by the sample of "undecided" voters (in their lingo voters who are leaning to a party but need prompting) being unusually pro-Labor, suggesting that the Labor vote was soft.  However even with "undecided" redistributed proportionally their results would still have been less accurate than EMRS (2.8% average error, 6.8 point miss on major party gap.)

I refused to make a formal projection of this election because of the unsatisfactory level of polling data.  However the aggregate attempts I issued anyway performed very well with the version incorporating house effects predicting the seat total correctly and the version excluding house effects getting every party's votes within 0.5%.  What I found impressive was that Tasmanian political commentators who attempted seat predictions, including Sean Ford in the Advocate/Examiner, and all of Brad Nowland (FontPR), Becher Townshend (FontPR), Alex Johnston (Win) and David Killick (Mercury) on the Fontcast, in the end converged on the correct outcome of 13-9-2-1 - although some of these had the Liberals higher for a while.  So at this election Tasmania was rather well served by its more informed pundits (as distinct from those who wrote hopeful and one-eyed op eds barracking for minority government) in the absence of polls.  Betting markets also performed very well (for once) with betting successfully predicting that there would be no change to any party's total in any seat, and successfully predicting a Liberal majority at all times.  And the reader Not-A-Polls on the sidebar of this site also nailed 13-9-2-1 as the average result.  

Legislative Council

I should also make some comments about the Upper House elections.  These were contentiously held on the same day as the state poll for the first time ever.  Mike Gaffney was re-elected unopposed, the first MLC to be so since Ruth Forrest (Murchison) in 2011.  There were four unopposed seats between 2005-2011 after a 12-year gap from 1993.  The Liberals gained Windermere on the retirement of Ivan Dean and now hold four seats compared to five for Labor and six independents.  As my voting patterns assessment shows, four of the independents are currently to the left of Labor while two are somewhat (but not very) right of centre.  This means the Government normally needs support from two of the four left indies on votes where Labor is opposed.

There was some thought that the blanket party advertising would result in voters voting the same way in both houses.  The voters of Windermere, however, weren't having a bar of it and the Liberal primary vote in the Legislative Council was 17 points lower in matching booths than it was for the Assembly.  Labor's result in Derwent was also strong given lacklustre Assembly results in the Lyons booths in Derwent and the trashing inflicted by indies in the Clark booths.  It seems that holding the elections on the same day did not result in voters in them voting much differently to normal in primary vote terms.  Whether or not it made a difference to the outcome in Windermere (where Will Smith missed the final two by 1.5%) we will probably never know.  

Added May 20: Did The Brooks "Terry" Saga Make A Difference?

With the release of booth voting figures by candidate it is possible to look for signs of the Adam Brooks "Terry" saga involving alleged fake dating profiles affecting his personal vote.  Basically, there is no evidence it did.  Brooks polled 16.1% of postal and prepoll Liberal votes and 15.2% of on-the-day Liberal votes in Braddon.  That might look like something happened given that in 2018 he polled slightly better on-the-day than before (28.3 vs 27.7).  However, it isn't necessarily evidence of anything to do with the "Terry" saga, as Jeremy Rockliff who had no such scandals also showed a significantly greater difference in favour of prepoll and postal voting than in 2018.  (51% vs 46.6% in 2021, 46.1 vs 44.9 in 2018).  It turns out that the factor most likely driving these shifts for both Rockliff and Brooks is the performance of Felix Ellis, who polled only 12.7% of Liberal prepolls and postals but 16.9% of Liberal votes on the day.  In 2018 Ellis had a much lower vote and a much lower difference in this regard.  Two possibilities here (and it could be a bit of both) is that Ellis's appeal to voters peaked late in the campaign, and that Ellis, as a younger candidate, was supported strongly by young voters and other voter types who are more likely to vote on the day. 

There's also not much evidence the saga hurt the Liberal vote in Braddon on the day, and if it did it wasn't by much.  Their on the day vote was 2.6 points lower than prepoll/postal compared to 1.5 points lower in 2018 but this could be explained by many things including Liberal support increasing towards the end of the 2018 campaign.  


  1. You often say that Tasmanian politics is more about the individual candidates than the parties they're running for. I think Janie Finlay is probably the reason for the decline in the Green vote in Bass.

    1. I think she is a big part of the reason. I think loss of incumbency in the seat and not having a candidate with a high existing profile are also relevant.

  2. Thanks for the summary and stellar coverage as always Kevin.

    I know the AJP pseph team has done a bit of work trying to map preference sources/destinations in previous elections. The expected patterns are usually evident (high flow to/from greens and left micros) but the flows aren’t super tight by any means. Plenty of strange preference sources/destinations making it clear voters don’t think in strict left/right terms (I.e. pious vegans voting Christians then AJP or visa versa).

    Likely just the adjacent column factor here more than anything, but good reminder that voters don’t always view/preference politics/parties in the way we logically think they would.

  3. This was a trick... Chance of reelection higher now than normally scheduled.... So Confect a reason for an early election.... Did maddie really decide to join the libs after the election was called.... Not to mention Mr Brooks...? Maybe this is their last term..
    Should be larger parliament and fixed terms

  4. The crowing from the Greens about their wonderful results is weird. This is their second worst result this century. And interestingly, their second worst result in Clark/Denison this century.

    To carp and cry that "the Greens are back" seems a trifle premature.

    It seens, dare I say it, cultish.

    But I'm sure it's a great result for the local Green luminaries. There's always votes in (white supremacist) NIMBYism so Cassy will assured of a seat for life.

    1. The Greens do often lay on talk of their successes with a trowel. Their worst result this century was last time, so second worst is an improvement. It does mean that the extent of Greens` poor result last time was not an indication of collapse but of an ALP policy (on pokies) that attracted Green voters. The Greens faced more competition than previously in Clark/Denison, with competition for dissatisfied ALP voters from Johnson and for teal voters from Hickey but their vote still went up.

  5. Non-first term same party governments not loosing seats:

    The Victorian Parliament expanded in both 1967 and 1976 , although the LCP/Liberal government (Country Party on the crossbench) probably would have gained in a stable Parliament (although malapportionment was an issue, particularly in 1967).

    Legislative Council:

    I don`t think a Derwent result held on a different day would have been different. Windermere, however, is probably a different story. Given that Smith only missed out by 1.54%, a non-simultaneous election could well have seen him overtake the ALP candidate and get a higher rate of preferences than he ended up giving the ALP, with the Liberals getting a lower preference rate and having a slightly lower vote to start with.

    1. Yes even a very small primary vote difference in Windermere might have shifted the results (the view seems to be Smith would have won had he got into second, though I've seen no figures in support of it). So I've added a note to that effect.

      Re parliamentary expansions, the metric I use to cater for that is the proportion of seats held by the government, which in both cases increased slightly.

    2. If I was a voter in Windermere who supported independents being a major force in the Legislative Council, I would be considering challenging the result on election spending/party campaigning grounds.