Saturday, March 24, 2018

Legislative Council Voting Patterns 2014-8

Advance summary:

1. This article presents a revised analysis of voting patterns in the Legislative Council (the upper house of Tasmanian Parliament) based on contested divisions involving the current MLCs in the last four years.

2. Although there is a degree of independence in all Legislative Council voting, the Council continues to have a fairly clearly defined "left wing" consisting of the four Labor Party MLCs, and independents Mike Gaffney, Ruth Forrest, Kerry Finch and Rob Valentine.

3. Excepting Rosemary Armitage and Tania Rattray (and Jim Wilkinson, who does not vote) the remaining MLCs (independents Ivan Dean, Robert Armstrong, Greg Hall, and Liberal Leonie Hiscutt) can all be clearly placed in a strongly-defined right-wing cluster.

4. A possible left-to-right sort of the Council is Forrest, Valentine, the four Labor MLCs (Farrell, Lovell, Siejka and Willie in no particular order), Gaffney, Finch, Armitage, Rattray, Hall, Armstrong, Hiscutt (Liberal), Dean.  However some of the exact positions in this list are debatable.

5. Going into the 2018 elections, the left holds an absolute majority in the current Legislative Council, although the fact that four of the left MLCs are independents means it will not necessarily be realised on every specific issue.


After the slight distraction of an election for the house of parliamentary theatre, it's almost time for another episode of Tasmania's house of serious power (at least when it comes to stopping bills), the Legislative Council.  Last year's adventures saw Labor win the eastern shore Hobart seats of Pembroke and Rumney from the right, and as a result the left of the Council holds an outright majority.  This doesn't seem that likely to improve for the Hodgman government at this year's elections.  Conservative independent Greg Hall is retiring, and the washup of a major redistribution creates an unpredictable vacancy in the new mostly rural seat of Prosser.  In Hobart, former Lord Mayor Rob Valentine is up for his first defence of a typically left-wing inner-city seat.  So far Valentine has at least three challengers and there are three declared candidates for Prosser, which both major parties are contesting.

As my traditionally super-wonky curtain-raiser for the LegCo season, this piece is my analysis of the voting patterns of the current Legislative Council.  Click the link to 2013-17 for last year's effort, which in turn links back to earlier years.

For the most part I've followed the same methods as in earlier years, but the red tide sweeping over the red couches has created a few little problems.  In the past I've measured alignment by tendency to agree with the members on one side or the other, but the four Labor MPs in the Council have so far voted together every time they could.  (That's not to say they will always do so, but no conscience issue has come up in the last few years on which the Labor team in the upper house has been split.) Using the methods I've used in the past, that would create the impression that they were the most left-wing members of the chamber, when it's really just a result of Labor's tendency to vote as a block.

Also, the methods I've used in the past would create the impression there are substantial political differences between the various Labor MLCs, which might be true, but it's not clear from the data that that's actually the case.  Sarah Lovell has voted on the same side as Leonie Hiscutt (now the sole Liberal) only 15% of the time compared to Craig Farrell's 28%, but this doesn't prove that Lovell is way to the left of Farrell.  After all, if we look at just the divisions Lovell voted on, Farrell only voted with Hiscutt on 15% of those.  The real reason for the difference is changes in the issues mix and possibly in Labor's tactics.

The solution I've come up with is to treat all the Labor candidates as having the same position in the absence of evidence otherwise.  To compare that position with other candidates I could just use data for Farrell, who has been there longer than the others.  An argument (though not terribly strong) against doing so is that it's possible that election of new Labor MLCs could have pushed Labor's decisions in some direction or other because those new MLCs acquire a vote in the Labor caucus.  I've decided for this year to weight the Labor figures by the number of votes for each MLC recorded in my dataset (105 for Farrell, 48 for Josh Willie and 27 for Lovell. Jo Siejka has also voted on six contested divisions but I decided that wasn't enough of a sample size for this year.)

The other details are generally the same.  I use only the last four years of data (which means no Jim Wilkinson as his votes in that time have been entirely determined by speakership precedent) and I only use recorded divisions with at least two votes (including pairs for absence) on either side.  There was only one significant lone dissent division in the last twelve months, which was Rob Valentine on the bill to grant access to parts of Mt Wellington to the proponents of a cable car.  (Divisions are not always recorded for lone dissents.)  Multiple divisions on the same issue on the same day with identical voting patterns are treated as a single division.

Since last year's review there have been 32 contested divisions that I have added to the sample.  Law and order issues, mostly in the field of sentencing, accounted for eleven of these including several amendments to motions.  Other issues to get a run included: forestry, local government rates, the Aboriginal Relics Act, anti-discrimination laws, workers' compensation, finfish regulation, and some electoral recommendations to the Lower House.  On the surface it might seem that the government didn't have too bad a run, being on the winning side of 19 of the 32 divisions, but those they "won" were often unimportant, and included some tied motions at preliminary stages where the President voted to break the tie so as to keep the debate going.

For this year I've only used one ratio figure (average agreement rate with the left-wing MLCs, counting Labor as one, vs average agreement rate with the right-wing ones).  The second, stepwise-comparison-based, method I've also used is a pain in the neck to code, and in this year's case I have good reason to believe it would give exactly the same result.

Agreement matrix and left-right sort

This chart shows how often the Legislative Councillors agree with each other on contested votes.  For instance, the chart shows that Forrest and Rattray agree 52% of the time.  As usual I've highlighted agreement scores over three-quarters in red and dark blue, and scores close to that mark are highlighted in orange and pale blue.  Given that I've removed the 100% agreement scores for the four Labor MLCs with each other, the highest remaining agreement scores are Armstrong and Hall 87%, Forrest and Gaffney 85%, Forrest and Labor 84% and Armstrong and Dean 83%.  The lowest are Hiscutt (Liberal) and Labor 24%, Valentine and Dean 25%, Valentine and Hiscutt 26% and Forrest and both Hiscutt and Dean 28%.

Once again there are two obvious clusters of high agreement scores.  The strongest one in this year's sample is the right-wing cluster of Hall, Armstrong, Hiscutt (Liberal) and Dean.  The left cluster of Forrest, Valentine, Labor, Gaffney and Finch is a little more weakly defined, but bear in mind that this includes four Labor members who have so far always voted together.  Two MLCs, Rosemary Armitage and Tania Rattray, don't form part of any cluster, though both agree slightly more with the right than the left.

The Left and Right figures give the average agreement rate with the MLCs in the left and right clusters (again, counting Labor as one). The Ratio figure (comparable to Score 1 last year) gives the ratio of the higher of these figures to the lower, coloured in red if the left is higher or in blue if the right is higher.

Compared to last year the main changes are: the Labor position appears to be more strongly left-wing and Mike Gaffney's less so, Rattray has moved back a little towards the conservative side, and Greg Hall comes out as much more consistently right-wing.  In Hall's case, this is partly because Tony Mulder, who he had a relatively low agreement percentage with, is no longer part of the right-wing sample after being defeated in 2017.

As usual, some of the closer pairs are arguable.  I've previously commented that Valentine is both less likely to agree with the right MLCs than Forrest and also less likely to agree with the left ones.  In this year's sample the first difference has reduced and the second difference has increased so I have swapped the two around this time.  A similar situation arises on the right, where Armstrong is more likely to side with the right-wing MLCs than Hiscutt but also more likely to side with the left-wing MLCs; this probably reflects cases where conservative independents take a different view of something to the Liberal Party.

Overall, the left now holds eight seats.  This means where the four left independents and the Labor Party agree on something, they will always carry the day.  Even if the Government gets the votes of Rattray, Armitage and one of the left independents, that is still not enough to pass motions that alter the status quo.  As a result the Hodgman government is struggling to get anything remotely controversial through.  A difficult term awaits unless it can improve relations with some of the left-wing independents or somehow have the numbers turn in its favour.

In total in 2017 there were 11 contested divisions where all the right-wing MLCs voted together and all the left-wing MLCs voted together, nine more where all the right voted together, seven more where all the left voted together, and five where neither side completely followed the pattern.

In two dimensions

Once again I've run a principal components analysis to give a 2D impression of the voting patterns though the value of this exercise is somewhat debatable. For those unfamiliar with graphs of these sorts, a principal components analysis aims to represent patterns in 2D with as little distortion as possible.  Both the angle of different lines to each other and the distance of different data points from the centre are relevant here.  The angles between different candidates indicate whether or not they display different kinds of voting patterns and the distance indicates how strongly each pattern is realised.  Even if two Legislative Councillors appear opposite each other, if one is close to the centre they will still agree fairly often.  If two Legislative Councillors are at a similar angle and a similar distance from the centre then it is likely their political views are rather similar. The two axes chosen by the analysis do not necessarily mean anything in particular and are not predetermined by me, but it's obvious in this case that the x-axis corresponds pretty closely to "left-right".

(ALP - Labor (Craig Farrell, Sarah Lovell, Jo Siejka, Josh Willie), Armi - Rosemary Armitage, Arms - Robert Armstrong, Dean - Ivan Dean, Fin - Kerry Finch, For - Ruth Forrest, Gaf - Mike Gaffney, Hall - Greg Hall, Hisc - Leonie Hiscutt (Liberal), Ratt - Tania Rattray, Val - Rob Valentine)

The left-right axis explains 79% of variation in this year's sample, and that's a lot.  While last year's graph picked out a Rattray factor as the second axis and put Armitage close to the centre, this year's has gone the other way round.  If we imagined it as a 3D image, the Rattray line would be off at close to right angles to the rest.  My impression is that the graph picks up MLCs who are not part of either cluster as its second and third axes because there is really nothing else beyond the left-right axis going on.  While the MLCs who are part of each cluster all vote independently (except the four Labor members) there doesn't seem to be any consistent pattern in how they break from left-right lines when they do.  It may as well be random.

That's that bit done for another year, and shortly I will be rolling out electorate guides for this year's contests for Hobart and Prosser.


  1. could the results of the 2 elections make any difference to the legistrative council left.... right balance

    1. Hobart - only if the Liberals win it (unlikely in my view)

      Prosser - whoever wins is effectively replacing Greg Hall (conservative independent) so if Labor or a Labor-ish independent wins then that will make a big difference; anyone else, probably not all that much.

  2. Obviously there is no voting data, but do you know where Jim Wilkinson would roughly sit on this spectrum? Are there many instances where he has had to use a casting vote?

    1. We do have data on how Wilkinson voted back before he was President ( and at that time I assessed his voting pattern as centre-right. In his time as President I can quickly find at least four cases where he has used his casting vote plus another six where votes on amendments were tied in the committee phase (the President currently does not participate in the voting on amendments meaning all ties are automatically lost).