Wednesday, December 26, 2018

Our aggregation remains at 46.2 per cent for the Coalition

As the year draws to a close, let's have a final look at our model of the two-party preferred (TPP) vote share for 2018.

If we look over the last eight weeks, the polls have been in the usual three to four point range (in this case, with the Coalition ranging from 45 to 48 per cent). The aggregation sits within that range at 46.2 per cent. There appears to be a slight downward trend, over last eight weeks in the median aggregation estimate.

For the Coalition to give itself a better than even chance of winning a May 2019 election it will need to improve its polling position by 0.9 percentage points each month. By the end of January 2019 the Coalition would want to be on 47.1 per cent. At the and of February they would want to be at 48 per cent. At the end of March, 48.9 per cent. And at the end of April 49.8 per cent. While theoretically possible, this is an unlikely degree of sustained change in public sentiment. The more likely scenario is that the Coalition will be polling somewhere between 46 and 48 per cent of the TPP vote share come the end of April. Not enough to win.



Turning to the primate vote models ...





And our estimates of TPP based on these aggregations ...


Tuesday, December 4, 2018

Steady as she goes - Coalition on 46.2 per cent

Today's Essential poll (Coalition 46 v 54 Labor) has not substantially changed the Bayesian poll aggregation. I have the Coalition on 46.2 per cent of the two-party preferred vote, roughly where it has been for about two months now. If an election were held now, it is highly likely that Labor would win with a sizable majority.



Turning to my quick and dirty aggregation of primary votes, we can see that Labor is maintaining a higher primary vote share under Morrison than it did under Turnbull, and the Coalition is maintaining a lower vote share under Morrison.









The implied two-party preferred vote share from these models is very similar to the straight aggregation above.