Oz Politics Blog

Morgan’s poll of 1795 voters on 30 April, 1 May and 7-8 May found two-party preferred support for the Labor party was 52 per cent to the Coalition’s 48 per cent.

Primary support for Labor rose 2.5 percentage points to 42.5 per cent. Support for the Coalition remained stable at 42 per cent.

Among the minor parties, support for the Greens was 8 per cent (down 0.5), Australian Democrats 1.5 per cent (down 0.5), Family First 1.5 per cent (unchanged), One Nation 1 per cent (unchanged) and other parties and independent candidates 3.5 per cent (down 1.5).

The old phenomenon appears to be back. There is a gap between Newspoll and Morgan, with Morgan consistently to the left of Newspoll in terms of the two-party preferred population statistic (see chart here).

I have discussed this deviation between Morgan and Newspoll many times before. I am still of the view that there is some form of systemic bias at work here. Interestingly the systemic bias appears to be manifest on four fronts (in the following order of significance).

1. Morgan under estimated the primary vote for the minor parties (other than the Greens), compared with Newspoll. The error bars give it away. At least one of the Morgan or Newspoll population estimates series is wrong. Anyway, it is implausible that Newspoll’s population estimate is 1.67 times Morgan’s most recent.

   

2. Morgan over estimated the Green primary vote, compared with Newspoll. Again the error bars give it away. There is something wrong with at least one of the pollsters when it comes to the Green primary vote.

   

3. Morgan under estimated the Coalition primary vote, compared with Newspoll. Although some of the data points from the two pollsters are at least consistent with each other.

   

4. Morgan over estimated the primary vote for the Labor party, compared with Newspoll. However, this difference appears less pronounced than the three above. There are more occasions when the data points are consistent.