Saturday, March 2, 2013

Not-A-Poll: Best Tasmanian Premier of the Last 30 Years: The Verdict

Federal note: I've updated my piece on Lilley following a new poll; see Silly Lilleys: Is Wayne Swan Losing His Seat?

 Note on WA election: I don't expect to be covering the WA state election on the night because of other commitments on the same weekend, and I don't have much to say about it except that I concur with the general judgement that the incumbent government is extremely likely to be re-elected, probably easily.  I recommend those interested head directly to The Poll Bludger.  I may have follow-up analysis of Upper House seats however, and sporadic comments on Twitter (@kevinbonham).

 The opt-in "Not-A-Poll" of reader opinions on the best Tasmanian Premier of the last 30 years has just finished after running through February with 150 votes received.  Huge thanks to all who voted; it has been a fun and I hope worthwhile exercise. 

Origin of the concept

I was inspired provoked to run this Not-A-Poll and write up the outcomes by the mainstream "reporting" of the Galaxy "best prime minister in Australia in the last 25 years" poll released in late January.  That was a credible formal opinion poll based on the views of 1000 respondents, but many of the writeups were abysmal.  The figures for that poll, as originally published in The Mercury, can be seen here.  John Howard topped the poll with 35% of votes, or 44% once uncommitted voters are excluded, but that was no call for nonsense such as:

"Australians overwhelmingly rate John Howard as the country's best prime minister of the past quarter century - and Julia Gillard the worst - in a new poll." (Gemma Jones, The Mercury, Jan 25).

In fact the poll suggests that most Australians do not rate John Howard as the country's best PM of this time (there is not even a non-overwhelming majority of decided voters doing so) and all the poll found was that Howard is the favourite of more voters than each of the other four.  Furthermore the poll did not even ask for opinions on who was the worst PM of this time, so to conclude that Gillard was the worst just because the least people thought she was the best is quite unwarranted.  Indeed, had the same sample been asked which of the five PMs had been the worst, Gillard probably would have topped it, but Howard would have scored highly on that question too.

The reason for Howard's large lead in the Galaxy poll is extremely obvious - he was the only Liberal.  While some voters picked a PM from the other major party, most stuck to party lines, resulting in the four Labor PMs racking up 45% of the vote between them compared to Howard's 35%.  Furthermore, the high undecided rate among Labor supporters (14% compared to 6% for Coalition supporters) most likely resulted from some of the Labor supporters being undecided which Labor PM they preferred the most, not from them being undecided between a Labor PM and Mr Howard.

It is not even clear from the Galaxy results whether Howard is Australia's favorite PM of the last quarter-century at all.  Those preferring Rudd, Keating and Gillard, plus the undecided Labor supporters, might very well prefer Hawke over Howard heavily enough to more than cancel out the 20-point primary gap between Howard and Hawke.

I decided it would be fun to run a similar exercise just among readers of this site for Tasmania, and then give an example of how to interpret such results. 

The weaknesses of opt-ins

Opt-ins, including this Not-A-Poll, and most so-called newspaper website polls, have four common weaknesses that make them a useless indicator of the general views of the public.

Firstly, the readership of any given site is typically not politically representative.  If an opt-in survey is conducted of readers of a right-wing newspaper, then you're likely to get a right-wing response.

Secondly, opt-ins are prone to "motivated response".  This applies more to opt-ins that require some effort to vote in (such as those dreadful "elector surveys" circulated by politicians, sometimes as a form of push-polling), but in general those with strong opinions about the subject of the poll are more likely to bother to vote upon seeing it.

Thirdly, online opt-ins are very prone to organised stacking.  An activist with an interest in a Tasmanian opt-in poll can easily use social networks to get people from outside Tasmania or even Australia to vote on an opt-in question on a website/

Fourthly, and most significantly, it is often possible for a voter to vote multiple times.  Some websites take precautions to discourage this, but there is probably no completely failsafe and practical method.  Many opt-in polls are being deliberately stacked not only by activists of both sides, but also by people with an interest in exposing their unreliable nature.

Media Watch recently featured the adventures of Ubermotive, aka Melbourne software engineer Russell Phillips, who has systematically gamed many dozen online polls to create deadlocked results, favour an unlikely response, or reverse the existing result.  The results of these gamed opt-in polls have then been unsuspiciously reported by their host sites as news.

Luke McIlveen of has responded: "I have not published a story about the hacker’s activities because I believe this individual should not be afforded any publicity."  But Phillips is not "hacking" as such, and deserves considerable and favourable publicity for exposing ugly truths about opt-ins and encouraging mainstream media websites to lift their game. It is really the original results of "polls" of such kinds that did not deserve publicity of the sort that they were given (and yet got it). Obviously, if Phillips can do what he did, then political parties can do it too.

The relatively small vote number in my own little exercise suggests that if there was any stacking or multiple voting it was on a very low scale, and watching the votes come in I didn't detect anything all that suspicious.  That said, it would have been very easy to do if anyone had wanted to, and I was curious to see if anyone would.  The main reason my results are not representative of Tasmanian voters as a whole is the first reason above - an unrepresentative audience.

The Labor lean in the results

The five Labor Premiers scored a combined 100 votes, exactly twice as many as the three Liberal Premiers.  To a degree I believe this represents a political leaning to Labor over Liberal of people voting on this site.  This arises from hits on this site in the last month coming from three main sources:

* Pollbludger, which for whatever reason tends to attract an audience of mainly Labor supporters with relatively few Greens and Liberals.

* Twitter, which is generally considered to be somewhat left-wing on the whole (although #politas, the main source of hits other than my follower base, sometimes feels like a bit of an exception to this.)

* Tasmanian Times, posters on which have Green, left-independent or greener-than-Green tendencies. I think the reader base on TT and the poster base are somewhat different but would still expect the reader base to lean left.

However, a lean in the preferences of readers is not the only possible explanation for the two-to-one Labor-to-Liberal margin.  Tasmania last had a Liberal Premier in 1998.  Readers in their early 30s and younger have never voted in an election at which a Liberal was Premier, and voters would have to be in at least their early 40s to have voted in the Robin Gray days.  So, some  readers would be making a choice between the more recent Premiers, all of them Labor, without being that familiar with Gray, Field, Groom or Rundle.

Also, the three Liberal Premiers in the survey all have tainted legacies - Gray as the subject of an adverse finding in a corruption inquiry, Groom as the one who lost his majority after just a single term, and Rundle as the one who co-operated with Labor to reduce the size of parliament, thus sending his party into Opposition for what turned out to be a very long time.

Some other comments on the results

One result that may surprise some is the strong performance of incumbent Premier Lara Giddings, given that she leads one of Tasmania's least popular governments.  Possible explanations for this result include:

* This site would be visited by several readers who have a personal interest in Giddings' political fortunes and who would therefore be inclined to vote for her whether they actually considered her the best of the list or not. 

* It's possible that Green voters would consider Giddings to be the best Premier of those listed on account of her being very socially "progressive" and also having given the most ground to their agenda.  Non-Green voters concerned with gay rights may also rate her the best on policy grounds.

* While the Government is extremely unpopular, there is no evidence that Giddings herself is unpopular.  Approval ratings have not yet been polled in Giddings' term as Premier, but her preferred-premier scores have been about what would be expected given her government's dire polling position.  

I'm inclined to reject the idea that being the only female gave Giddings a big advantage.  This certainly wasn't the case for Gillard in the Galaxy poll.  I noticed Giddings polled especially strongly at two times - when the poll had just started, and when I released analysis of the EMRS poll.

Paul Lennon's distant last place with just four votes (2.7%) - one of them a probable sympathy vote -  may well produce an "ouch!" response, and be seen to give further credit to the view that he was spectacularly unpopular while in office.  In fact, the final-poll 17% result for Lennon often claimed to have been his "approval rating" was actually his preferred-premier score, and it was very likely deflated by Labor voters who preferred David Bartlett to be Premier picking "none of the above".  A true approval rating for Lennon at the time would have been poor, but not that poor.  I reckon it would have been about 30.

I also suspect that the kinds of voters who would be least unlikely to think that Lennon was the best Premier of the last 30 years - most likely elderly, somewhat socially conservative, traditional ALP voters - would not be likely to read websites such as this one.  It's also possible that a lot of the Bacon voters would rank Lennon second. 

(The least popular Premier in Tasmanian polling history, by the way, was Harry Holgate, who in one old Morgan poll polled an approval of 11 and a disapproval of 66, and then followed that with an approval of 15 and a disapproval of 74.  I am not aware of worse figures than this being recorded by any incumbent Premier or PM, although Keating at one stage came close.)

And the winner is ...

And finally, here are the results, together with a quick electoral form guide for each of the eight Premiers in question:

1. Jim Bacon (ALP, 1998-2004) 41 votes
Elections as party leader: 1998 (Won outright from opposition), 2002 (Won, retaining majority)
Elections as candidate: 1996, 1998, 2002 (won all)
Highest personal vote: 35.5% in 2002, as Premier.

2. Lara Giddings (ALP, 2011-) 23 votes
Elections as party leader: None yet
Elections as candidate: 1996 (won), 1998 (lost seat), 1999 (Pembroke LegCo - lost), 2002 (won), 2006 (won), 2010 (won)
Highest personal vote: 15.1% in 2010, as Deputy Premier.

3. Ray Groom (Lib, 1992-6) 21 votes
Elections as party leader: 1992 (Won outright from opposition), 1996 (Lost majority and resigned, but party remained in office)
Elections as candidate: 1986, 1989, 1992, 1996, 1998 (won all)
Highest personal vote: 26.8% in 1992, as Opposition Leader.

4. Robin Gray (Lib, 1983-9) 20 votes
Elections as party leader: 1982 (Won outright from opposition), 1986 (Retained majority), 1989 (Lost majority and government)
Elections as candidate: 1976, 1979, 1982, 1986, 1989, 1992 (won all)
Highest personal vote: 42.2% in 1986, as Premier.

5. Michael Field (ALP, 1989-92) 19 votes
Elections as party leader: 1989 (Won minority government from opposition), 1992 (Lost outright), 1996 (Lost, but Liberal government lost majority)
Elections as candidate: 1976, 1979, 1982, 1986, 1989, 1992, 1996 (won all)

Highest personal vote: 15.6% in 1989, as Opposition Leader.

6. David Bartlett (ALP, 2008-11) 13 votes
Elections as party leader: 2010 (Lost majority but remained Premier)
Elections as candidate: 2002 (lost, but later elected on countback), 2006 (won), 2010 (won)
Highest personal vote: 15.9% in 2010, as Premier.

7. Tony Rundle (Lib, 1996-8) 9 votes
Elections as party leader: 1998 (Lost outright)
Elections as candidate: 1982 (lost), 1986 (won), 1989 (won), 1992 (won), 1996 (won), 1998 (won)
Highest personal vote: 23.0% in 1998, as Premier.

8. Paul Lennon (ALP, 2004-8) 4 votes
Elections as party leader: 2006 (Won, retaining majority)
Elections as candidate: 1992 (lost, but later elected on countback), 1996 (won), 1998 (won), 2002 (won), 2006 (won)
Highest personal vote: 26.1% in 2006, as Premier.

UPDATE (10 March): Vote theft scandal!  Oddly, the poll as displayed on the site now shows Bacon on 40 votes.  


  1. One more reason I'd think the votes would be weighted by is the recency of the various premierships. I don't go back further than Jim Bacon, so although I may have heard good / bad things about earlier premiers, I have no experience of them and are unlikely to nominate them. Obviously Lara picks this up to the greatest degree, and more so because the two immediately before her don't rate well.

    I suspect that even if you filtered for Liberal voters, Jim Bacon would still feature prominently in the results.

    Paul Lennon was Jim Bacon's hard man, before he had to step up due to circumstance. Not too dissimilar to Tony Abbott in that sense; obviously different circumstances forcing the move. But the hard man role is remembered, and personal popularity is never the same.

  2. I didn't see this or vote in it. I was surprised to see Gray get so high given his interesting fiscal legacy - but then I saw who was under him.


The comment system is unreliable. If you cannot submit comments you can email me a comment (via email link in profile) - email must be entitled: Comment for publication, followed by the name of the article you wish to comment on. Comments are accepted in full or not at all. Comments will be published under the name the email is sent from unless an alias is clearly requested and stated. If you submit a comment which is not accepted within a few days you can also email me and I will check if it has been received.