Saturday, December 25, 2021

Psephology And The Palace Letters

It's a nearly annual tradition on this site to release something every Christmas Day.  (Click the Xmas tag for previous examples.) It's also an annual tradition when I write the annual review, to list pieces that languished in the draft section, some barely started, others more or less complete but never quite seeing the light of day.  When I did this last year there was an unusual level of interest (a whole three enthusiastic requests!) in one of the unreleased items, about "psephological themes in the Palace Letters" and I decided to make it this year's Christmas present to readers.  This article mostly focuses on the 1974-5 cycle although the letters cover the 1975-7 cycle as well.   By the way, I am not the only psephologist dropping down chimneys with offerings this year.

The Palace Letters are a series of correspondence (not all of it related to federal politics) between the Queen's Private Secretary Sir Martin Charteris and Governor-General Sir John Kerr in the leadup to and the aftermath of the November 11 1975 sacking of the Whitlam Government.  They also include a large number of media articles that are public documents but would not be well known to a modern audience and that are fun to read over.  The letters were released after a very prolonged legal stoush, especially thanks to the persistence of Whitlam biographer and political historian Jenny Hocking.  

The letters were not the smoking gun for direct involvement by the Queen that some were hoping for, but they do show that the Queen's Private Secretary Sir Martin Charteris provided extensive advice to Governor-General Sir John Kerr, and that that advice in general was that Kerr could, if he eventually saw the need, do what he did.  It isn't news that Kerr genuinely feared Whitlam might sack him before he could sack Whitlam, and really this race-to-sack aspect has little to do with the (now largely dormant) republic versus monarchy debate and is more a question of finding an adequate structure that avoids it under either.  I think much of the debate last year overweighted the effect of the Charteris letters themselves and ignored the extent to which legal opinions of very much the same nature were swirling in the public debate in 1975 generally.  

All the same on many occasions reading the letters back and forth I thought, regarding a wide range of specific political subjects: why are Kerr and Charteris, especially the latter, even discussing this? Do such general political discussions ever continue between the Palace and the Governors-General today, only by telephone or email rather than letter?  Or was the voluminous and at times rather needy correspondent Kerr a special case?  (Note added: see Anne Twomey here re surrounding GGs).

The course Whitlam was embarking upon when dismissed was hopeless.  There was little real chance that a half-Senate election in late 1975 - even if supported by Supply and the co-operation of all the State Governors -  would have put Whitlam in a position to pass Supply before June 1976.  Kerr says - and I am surprised more is not made of this - that if Whitlam was still unable to get Supply, then Whitlam would have agreed to a double dissolution in early 1976, giving Australia two gigantic manual Senate counts within a few months of each other.  That would have been ... interesting ... 

The electoral backdrop

The Whitlam Government's issues with the Senate were a distant legacy of the 1961 election.  In 1961 the Menzies government squeaked through with a two-seat majority.  Menzies called an early election in 1963 in search of a larger majority and succeeded, but this put the two houses out of alignment with half-Senate elections being held in 1964, 1967 and 1970 and Reps elections in 1963, 1966, 1969 and 1972.  As a result when Whitlam won the Reps election quite convincingly in 1972, there was no accompanying half-Senate election for Labor to do similarly well in, and the Senate was still the one elected in 1967 and 1970 (apart from one vacancy in Queensland which the Liberal Party retained).  The numbers in this Senate were 27 Coalition, 27 Labor, 5 DLP and three independents - Labor having lost two contests in 1967 by less than 1%.  

The obstructive Senate resulted in the 1974 double dissolution, which Labor won - but more narrowly than 1972.  The Senate result was 29-29-2 with the two crossbenchers being SA Liberal Movement Senator Steele Hall and Tasmanian independent Michael Townley.  Labor was unable to turn its small overall voting intention advantage into any 6-4 splits, missing out on one in NSW by 1.4%, but the Coalition managed a 6-4 split in Queensland, just 4911 votes in that state (0.47%) separating Labor from a 30th seat and the ability to pass anything supported by Hall.  Another unlucky breakdown for Labor and so close electorally to a completely different history.

Labor's problems from that point on are well known.  Townley joined the Liberal Party.  One Labor Senator resigned and another died.  The NSW government filled its casual vacancy with an independent, Cleaver Bunton, arguing that the then uncodified same-party convention for casual vacancies did not apply to resignations.  This ended up not affecting Labor's numbers on Supply. The Queensland government, however, rejected Labor's nominee and instead installed Albert Field, an unknown Labor Party member who was promptly expelled from the party.   While Bunton supported passing Supply, Field supported deferring it.  Field's eligibility was challenged and he went on leave, but the numbers still stood at 30-29 for at least deferring Supply unless the by this stage scandal-plagued government was willing to call a House of Representatives election.

The letters note that Malcolm Fraser's predecessor as Opposition Leader, Billy Snedden, had appeared set on blocking Supply and forcing an election early in the year and that Whitlam had been resigned to (or perhaps given it was Snedden looking forward to) a rinse-and-repeat of 1974 until Snedden was rolled.  Even in February, and before the Queensland Government's appointment of Field materially altered the numbers, Kerr wrote: "The Ministers to whom I talk appear to accept that this is likely to happen and to be not entirely confident of the outcome."

The proposed half-Senate election

Gough Whitlam's proposed attempt to resolve the deadlock was to not call a Reps election but instead call a half-Senate election to try to alter the numbers in the Senate.  However, irrespective of the outcome, only six seats would have been filled immediately, with the rest of the new Senators taking their places in mid-1976.  These seats were two each (newly created) in the ACT and NT and the ex-Labor casual vacancy seats held by Bunton and Field.  

Had pro-Supply candidates won five out of these six seats, Whitlam would have had the numbers to pass Supply.  Had they won four, the Senate would have been able to reject supply but not to continue deferring it, and the Liberals may have gone to water in this situation.  A 1-1 split in the NT was inevitable, but Labor had hopes that former Liberal Prime Minister John Gorton, running as an independent in the ACT, might deprive the Liberals of a seat there.  The actual results, with Gorton getting only 11%, suggest that this was a pious hope even if there was a major backlash against the Liberals' supply tactics.  It kickstarted a long tradition of perennial speculation that the Liberals might miss out on an ACT seat (something that has only come close to happening twice so far, and that was under Group Ticket Voting).  

The biggest curiosity would have been the race to fill the casual vacancy seats.  Queensland and New South Wales would have had six-seat elections with the first five winners elected for the 1976-1982 term and the sixth winner serving until 1979 (assuming the Senate did not determine some other principle to the at that stage generally accepted order of election method).  The sixth seat would normally be a consolation prize but in this case it would be desirable because of the Supply crisis.  Normally the least successful of the two major party forces would win the sixth seat.  

To show what a rabbit-hole this could have become, in NSW in 1974 Labor polled 3.5 quotas for a six-seat contest and the Liberals polled 2.92.  In Queensland, the combined Country/Liberal ticket polled 3.52 quotas and Labor polled 3.06.  All else being equal, Labor would win seats 1, 3 and 5 in NSW and the Coalition would win 2, 4 and 6, while in Queensland the result would be the reverse.  

In order to try to win the sixth seat, it would be in the interest of the leading side in either state to deliberately campaign badly (without overdoing it) and try to throw about 4% of the vote to their opponent.  However, the other side could do that too, so it probably wouldn't work.  The Coalition in Queensland at least could have tried the cute trick of running separate Country and Liberal tickets, with the aim of getting them below 2 quotas each.  If Labor did well enough the order of election would then be Labor, Liberal, Country, Labor, Labor with the Coalition taking the 6th seat despite its greater vote share.   However Labor could avoid this by getting its own vote far enough below three quotas to just avoid crossing on minor party leakage, meaning that one of the Coalition partners would be excluded, electing the other in 5th with Labor elected 6th on Coalition leakage.  Parties could also respond with Hare-Clark style vote-spreading strategies by handing out different how to vote cards in different parts of each state and thereby attempt to spread the vote such that their candidates took as long as possible to cross quota.  For instance, if the side with 3.5 quotas manages to spread that load evenly between its top four candidates, and the side with 3.0 quotas does the same, at some point the fourth candidate for the less popular side gets excluded. However, given that minor party preferences will have put the latter just over 3.0 quotas total, this exclusion will elect all the less popular side's remaining candidates in positions 1, 2 and 3, and the more popular side will win positions 4, 5 and 6.  However, if the more popular side does a little better than expected, it might find itself hitting 4 quotas during the cutup, winning it four seats out of six but causing it to not win the sixth seat.  Other possible strategies could have included propping up a friendly flank party or independent. 

The Palace Letters include some discussion of the absurdity of the scenario, for instance a 1976 article by David Butler, but the discussion only scratches the surface of what a strange contest it would have been.  Of course, there was also the possibility that Labor would have done well enough to not get the numbers but still cause the Liberals to pull their heads in; as Butler notes:

"But of course the half-Senate election would have served as a plebiscite. If the votes on December 13 had gone decisively Labor's way, it is overwhelmingly probable that the continuing Senate would not have kept its stomach for the fight."

I don't think it's likely though.  The history of half-Senate elections was that governments tended to poll disappointingly in them.  Labor had surged in polling as the Supply crisis developed, but only to a competitive position rather than a strong one, and even that was based on limited evidence (see below).

A further well-known curve ball is that the Governors of NSW and Queensland might have been advised by their Premiers to muck about with the issuing of the writs for their half-Senate elections in order to make it more difficult for Whitlam to improve his position when he wanted to.

Kerr The Poll Junkie

One of the revelations of the Palace Letters is that Governor-General Kerr was an avid poll watcher, especially of polls that might affect perceptions of his own position.  After the Dismissal he became something of a cheerleader for the Fraser Government, viewing it as personal vindication when it was winning, whereas previously he had been an interested but relatively detached observer.  Kerr seemed to have a high degree of faith in most of the polling, in spite of it having been rather inaccurate in 1969.

The  most regular voting intention polling in 1974-5 was Gallup (later Morgan Gallup and nowadays Roy Morgan), but there were also five Saulwick voting intention polls for The Age, which sometimes painted a slightly better picture for Labor and sometimes didn't.  At the head of the article I have posted a graph based on all the individual polls, except that when polls by each company covered about the same time period I have averaged them.  The graph is bouncy because it is based on single polls, a few of which look a bit rogue, but the overall picture is that the Whitlam Government trailed out of the blocks and rarely reached equality.   During March 1975 it was competitive as the Opposition replaced Billy Snedden with Malcolm Fraser, but from June onwards the government crashed to the worst polling position seen for any government in Australian history.  As the initially unpopular Supply crisis arose the government surged back to competitiveness, gaining something like 12% 2PP in less than two months (!) but once the Governor-General had called an election things quickly blew out again, and the polls proved reasonably close to the mark.  This - perhaps the biggest blowout in a campaign proper (there was a similar but less dramatic one in 2013) - is often attributed to Fraser's campaign focusing on the economy while Labor continued to bang on about the Dismissal, which had lost salience as it did not affect voters personally. I suspect it is also the case that the Dismissal itself removed the sense of gridlock and gave voters previously concerned about the Coalition's tactics license to vote for them freely.

I should note here that there are significant issues with trying to obtain a single set of Gallup figures.  There are often multiple results that are somewhat different, in particular some with the Coalition parties split and some not, but also with other issues.  For this article for voting intention I have used the figures that at one stage were archived on the Morgan website, having also found some evidence in the letters that at least some of these were published at the time (as well as some others that contradicted them).

The narrowing of voter views also played out in Gallup net approval ratings through 1975 (note that the time axis gaps before Sep 27 are not uniform in size):

As the Supply crisis developed through September into early November, Whitlam's net rating rose steadily while Fraser's crashed from net +25 to net -24 in three weeks.  However once Whitlam was sacked his ratings again fell while Fraser's, despite the controversy, climbed back to a respectable final net -2.

As a sign of what a tumultuous campaign it was, as of the Nov 22-3 Gallup poll, Labor had a 47-43 lead on the question of who voters thought would win the election.  A week later they trailed 33-52 on the same question. 

The letters also include a Gallup Senate poll from Nov 15.  It underestimated the Coalition's eventual result by 5.7%, accurately estimated the vote for Labor, and overestimated all minor parties canvassed.  (The Australia Party, polled at 3%, received only 0.48%).  On this basis 32 seats were projected to the Coalition, 30 to Labor, 1 to Steele Hall and a question mark over the Liberals vs Gorton in the ACT.  As the government's vote grew through the final weeks it ended up winning six-seat slates in both Victoria and WA as well as the predicted Queensland (thereby winning 35 seats), while Labor also lost an predicted seat to Brian Harradine in Tasmania and only got 27.  Still by Senate polling standards this wasn't bad and the poll was probably very accurate at the time it was taken.

Many polls showed voters thinking that the Coalition should allow Supply.  Newspapers, such as the AFR editorial of 30 October, argued that the view of 70% of voters to this effect and the rapidly improving position of the Government in polling meant the Opposition should back down.  History shows that its decision to ignore public opinion and continue with the mutual brinksmanship was the right one, and that polls were capturing only how the public reacted to the crisis as it stood, and not how they would react to what happened later.  A Gallup poll at the same time showed voters said they did not want an early election (55-38) but what voters might say about that and how they might vote in one are very different matters.  (Voters pretty much never want an early election.)

Poll-Shaped Objects, 1975 Vintage

On December 3, Whitlam claimed that the party's polling showed that four Melbourne marginals gained in 1972 and two gained in 1974 were "safe" and that there were fine prospects of gaining four more.  The Victorian Labor President said of the public polling "Don't be pessimistic about these polls.  Don't believe them.  Our own polls show that we are running at a level about the same as that in 1974."  He referred to a claimed 17-20 % level of undecided voters (I have no idea if any polls actually showed this.)  As it happened, the public polls were accurate and these Labor internals, if they existed at all, were bunker stuff.  Labor lost all six of the Melbourne marginals they had gained in 1972-4, with margins between 1.6% and 9.1%.

ANOP conducted a "swinging voters" poll series with a miserable sample size of 120 voters per poll, based on voters who had been picked out as having weak party commitment and based on demographic indicators.  The results of this series proved to be spectacularly false as a guide to voter behaviour overall, as I would say has also since been the case with many if not most "undecided voter" panels picked by pollsters for numerous leader debates.  56% of the swinging voters said that Labor deserved to win the election compared to 29% for the Coalition.  The most important campaign issue (chosen by 41% of the sample) was "rules and conventions", ie the Dismissal and surrounding events. (Economic management scored 21% and leadership 15%).  The sample almost unanimously disapproved of how Fraser became Prime Minister (72-20) but was very evenly divided about whether Kerr had been right to sack Whitlam.   Aside from using such a tiny sample size, another thing this poll might have been doing wrong was limiting its sample to capital cities. 

There is much more I could cover in this article, especially the material in the leadup to the 1977 election (in which the polls performed much worse than in 1975).  However I think that is enough for one day and especially for this day.  Secular season's greetings to all readers and best wishes for election year 2022!


  1. Hi Keven,

    Thanks for the excellent article today - I have enjoyed reading it, as I do for all your articles.


  2. The Whitlam Government`s half-Senate election strategy would have been much more credible if there were 3 seats per territory, meaning the territory seats that were needed for the strategy to work were actually within the realm of realistic possibility of being up for grabs, something that would likely have favoured or been neutral to non-right parties in subsequent years. It may even have worked without NSW and Queensland holding half-Senate elections, unless Albert Field turned up to vote on supply.

    Another interesting counter-factual is would the Whitlam Government have got half or more of the Senators if the 1967 New England Statehood Referendum had passed and resulted in New England Statehood?

  3. Thanks for the mention, Kevin! Not sure if I could fit down a chimney without breaking myself or something around me...

    In all honesty, I planned that piece to drop after the next Resolve poll, but I realised the aural similarity between "Resolve" and "Rudolph" and couldn't resist.

    Enjoy the holidays, and the very best of wishes for next year as well! It looks to be a pretty big year for Aussie election-watchers, especially given how close the state SA and federal elections will likely be.

  4. For this post reader one of the absolute delights of the palace letters, finally being released (story there too), particularly having listened to some bang on endlessly about 'seeking to engage the Palace', is the solid documented history that the only one of Kerr, Whitlam, and Fraser, who directly lobbied the Palace to intervene was.... wait for it.... it really is priceless... not Kerr, .... not even Fraser, .... but WHITLAM. Think I laughed on an off for a month when I read that.

  5. This comment has been removed by the author.

  6. To add to all the mess there was also the possibility of the ballot paper being flooded with loads of extra candidates. In 1974 voters in New South Wales had the joy of having to number 73 different candidates entirely accurately - no ticket voting, no optional prefencing, no savings provision, no allowance for accidental duplication or any other later provisions even though over 90% of voters voted for either Labor or the Coalition. A lot of conservative activists (including a youngish Fred Nile) had nominated in the belief that Labor voters were more likely to make mistakes in such circumstances and a high informal rate could cost the ALP a seat. The narrow failure to win a sixth seat suggests their tactic had worked. But what a group on one side of politics can do in one election can be duplicated by others in the next.

    A clear conclusion from your piece is that STV is too easy to bastardise. The initial election should always be for seats of exactly equal weight with nothing hinging on the order of election (if a split is needed have some follow-on mechanism) and parties should benefit from getting more votes instead of how well they could do at game theory.

  7. Indeed, the 1974 Senate election had a particularly high informal rate because of ballot flooding and the lack of savings provisions you mention.

    The Senate system at the time was especially easy to bastardise because of the way casual vacancies were then dealt with - this has since been changed so the incentive to win the last seat that figured in the Whitlam government's 1975 strategy has gone. However there remain serious problems with allocating long and short terms after a DD (or for that matter a Senate expansion a la 1984) in a way that reflects the will of the voters rather than the will of the big parties.