Newspoll 49-51

Today's Newspoll of 1409 respondents estimated the national two-party preferred (TPP) vote shares as 49 per cent for the Coalition and 51 for Labor. This had little effect on my standard aggregation model, which finds the new Newspoll series has a pro-Labor house effect of around one percentage point.

The aggregated estimation of national TPP voting intention, assuming the house effects across polling houses cancel each other out, has the Coalition on 49.9 per cent and Labor on 50.1.

If we anchor the aggregated TPP series to the poll outcome at the last election, the national TPP estimate is 50.6 per cent for the Coalition and 49.4 for Labor. In round terms, the house effects of the combined polls are worth around half a percentage point to the Coalition.

The comparison chart of the medians from these two aggregations follows.

There is of course a problem with the anchored model. The published TPP estimates up until the 2013 election were based on preference flows at the 2010 election. Since the 2013 election they have used the preference flows at the 2013 election. The 2013 election saw a sizable change in preference flows compared with previous elections. The interesting question is whether the change in preference flows at 2013 was a response to Mr Abbott and therefore temporary or something structural and permanent in the nation's preference intentions.

In the next two charts, I estimate the TPP series from an aggregation of primary votes. Both charts assume the house effects across polling houses cancel each other out (they sum to zero). In the first chart I take the 2013 preference flows and in the second chart I use the 2010 preference flows. If we revert to the 2010 preference flow pattern at the 2016 election, it would be worth around a percentage point to the Coalition.