JWS Research has conducted an automated telephone poll of 1391 Victorians of state voting intentions. The poll was commissioned by the Herald Sun. The full report for the JWS poll can be found at the JWS Research website here. The Herald Sun article on the poll can be found here (behind pay wall).
The poll makes much better reading the Liberal and National supporters in Victoria than the Newspoll released a few weeks ago which has the Coalition TPP at 45%. The primary vote shares reported in the poll are as follows with our TPP estimate. The swings in the below table are compared against the 2010 election result.
As per usual we will undertake a seat projection based on this poll as state polls are few and far between. However we have applied our own interpretation to the poll due to a few unique characteristics.
The headline TPP figures of the JWS poll are subject to interpretation as JWS have asked respondents to allocate their own preferences. The JWS report also provides a headline TPP calculated on preference flows at the last election, but the latter calculation is based on a state election result adjusted for two by-elections in Melbourne and Niddrie which the Liberals did not contest skewing the figures slightly.
In the experience of this blogger the allocation of preferences as per the flow of preferences at the last election provides the most accurate reflected of the TPP vote on election day and thus we use a TPP figures calculated on this basis for use in the seat projection.
Our TPP estimate based on the JWS poll primary votes is based on unchanged figures from the 2010 election and separate allocation of preferences for both Greens and Others. The result is a Coalition TPP of 52.8% which reflects a state-wide swing of 1.2% to the Coalition from the last election.
Swing by Margin Categories
The JWS poll also provides TPP swings for categories of seats based on the marginality of the seats. In the JWS poll report TPP swings are provided on a respondent allocated basis for four categories of seats, Safe Labor, Safe Coalition, Marginal Coalition, Marginal Labor and Safe Labor. Marginal seats are defined as seats with a TPP margin of 6% or less.
When reported by these four categories the TPP swing to the Coalition is concentrated in its safe seats with a swing of 4% to the Coalition on a respondent allocated basis compared to swings of 1% or less to Labor in all other categories.
However these categories have a small sample sizes. Of the four categories sorted by major party and marginality have MoEs between 4.8% & 6.6% which is too large for our tastes.
When seats are grouped by marginality alone we get larger sample sizes and lower MoEs of just under 4%. On a respondent allocated basis the swings reported by JWS were 0.7% to Labor in marginal seats and 0.3% to the Coalition in safe seats. The MoEs for these categories are a bit better at just 1 percentage point above the statewide MoE, so we will use these swings adjusted for difference between respondent allocated preferences and our TPP estimation. Our adjustment results in no swing in marginal seats and a swing to the Coalition of 1.6% in safe seats. It is these swings which are used in our seat projection instead of a single state-wide swing
With our input figures settled the seat projection is below. The methodology for the seat projection can be found at this post.
As the swing in marginal seats is zero the only change in the seats from the last election arises due to the fact that there are more Labor marginal seats with smaller margins than the Coalition, therefore in a situation where there is little to no swing in the marginals the Coalition has a better chance to pick up seats than Labor. This is accounted for in the sum of probabilities which is the basis of the seat projection.
Once again we should not the above projection does not take into account the sophomore effect as we do not know which first term MP’s will recontest.
It should also be noted that a redistribution of Victorian state electorates will occur shortly, changing the margins of Victorian electorates before the next state election.