Tasmania is a federal polling black spot. It is not because the major pollsters don’t poll in Tasmania on federal voting intention, they do. It is because with 2.5% of enrolments on the national electoral roll any national sample of between 1,000-2,000 should only poll between 25-50 Tasmanians, a sample far too small for any meaningful analysis.
Yesterday, Roy Morgan released an aggregation of its face-to-face polling from the past 3 months. The aggregate national sample was 10,800 respondents, which is pretty hefty. In comparison Newspoll’s last quarterly aggregation was for 6,884 respondents. When we have such a large sample, a Tasmanian sample starts to become slightly more relevant, albeit with caution. If the Tasmanian sample within yesterday’s Roy Morgan aggregation was proportion to the Tasmanian population vis a vis the rest of the country, the sample size should have been 270. The margin of error for such a sample is 6.1%.
The Morgan Poll is generally but not always conducted on a face-to-face basis. Readers of this blog will know when calculating the Poliquant Australia Poll Average face-to-face polls are not included the poll average. This is because in the experience of this blogger face-to-face polling has provided unreliable results and is too susceptible to social desirability bias. Other methods of polling such a phone polling or internet panel polling are less personal and thus less prone to social desirability bias. This is an interesting topic in its own right and deserves its own blog post and we will come back to it in future.
Nevertheless what we are looking for in the Roy Morgan polling results is the relative position of Tasmanian voters compared to the rest of the country. For this task the method of conducting the poll is not so important because any bias in the face to face results is constant throughout the country as it is an issue of methodology not geography.
Therefore what we are going to look at is the deviation of each state from the national swing contained in the same poll. This is calculated using an average, weighted by sample size, of each states’s difference with the national TPP swing in the aggregation for each monthly result (June-August). The deviation is calculated in terms of the TPP swing to the Coalition. A table is set out below.
A positive number indicates a greater swing to the Coalition than the national swing, a negative number indicates a lesser swing to the Coalition than the national swing.
We have also calculated implied margins of error for each state’s sample. The Roy Morgan finding did not specify the sample sizes for each states. However we can assume state sample sizes are broadly proportional for the relative populations of each state. We therefore assume sample size which equate to the states’ share of the national population multiplied by the national sample of the aggregation. The margin of error is calculated in the usual way. We use the term implied as the margin of error for the state sample sizes is not provided by Roy Morgan, it is our calculation based on the above assumptions.
As you can see the TPP swing in Tasmania is 7% greater for the Coalition than the national swing according to the Roy Morgan aggregation for June-August. However this is not the whole story, we have to take into account the margin of error for the small sample. This creates range of the deviation of between 0.9%-13.1%. Therefore we know the TPP swing to the Coalition is greater in Tasmania but we don’t know with much precision by how much. It is tantalising information with a definite implication but a lack of a precise figure.
The correct position of the federal TPP swing is of great importance. We have pegged the national swing to the Coalition at 5.1%. The 4 Labor held seats in Tasmanian seats are all on margin exceeding this figure, these margins are set out in the table below. The size of the greater Tasmanian swing to the Coalition will determine whether any or all of these seats fall to the Coalition.
For now, we can only say with any degree of certainty there is a larger TPP swing to the federal Coalition in Tasmania than Australia as a whole. The implication for our Australia poll average and seat projection is that Labor’s Tasmanian seats may not stay with Labor. However once again we do not have hard data with sufficient precision to adjust the seat projection model accordingly.
In the mean time, you can substitute your own views about the Coalition swing in Tasmania. Currently, the seat projection model is giving 3 seats to Labor 1 to the Coalition and 1 to Wilkie. If you are convinced that Labor is facing a wipe-out in Tasmania, just take three seats off Labor and add the same number to the Coalition. If you think that the swing to the Coalition in Tasmania is only slightly larger than the national swing then you do not need to adjust the seat projection numbers.
If there is any public federal polling in Tasmania with a decent sample size prior to the election poll-watchers will be quite fortunate. If not, then federal election night in Tasmania will be very interesting to watch.