Updated Poll of Polls – two myths exposed

The race to become the next government is effectively over, but there is intrigue in some of the secondary numbers

National retains a very large lead in our latest Poll of Polls, and that lead is growing wider still as we move closer to the election.

The headlines tend to concentrate on its head-to-head lead over Labour, which has climbed up to almost 24% in our poll of polls. But even more important is the lead that National and ACT together hold over the grouping of Labour and the Greens.

That gap has grown steadily over the last year, from a low of around 10% in November 2010 up to its current level of over 17%. This translates into a seat advantage of around 21 seats, making a right-leaning government massively more likely than a left-leaning government (about a 94% probability according to ipredict). As remarkable as that kind of a lead is, it is nothing especially unusual. This has been the broad political situation in New Zealand since early 2009.

Despite the fact that little has changed in the main race to form the next government, there are two items in the poll of polls of particular interest.

First, some commentators have been suggesting that ACT’s gambit when it took on Don Brash as its leader has already proven a failure, and there has been no improvement for ACT in the polls. That claim is false.

In April, we estimated ACT’s support at around 0.8%, whereas for the last two months we have estimated it hovering between 1.9% and 2.2%. The ACT vote has more than doubled since Don Brash became leader, and the rising trend has been fairly consistent. Of course, this is a far cry from what Dr Brash promised – I recall the number 15% crossing his lips in yet another ill-advised outburst – but it is not nothing.

We will have to wait and see whether ACT’s very public internal spat over cannabis policy negatively influences their poll ratings in current weeks.


Second, there has been a suggestion that Labour’s loss is the Greens’ gain, and that the drops we are seeing in Labour’s support are due to left-leaning voters deserting it for the fresher-looking Greens. Our data from 2011 shows some support for this idea, illustrated in the chart above. Using very simple regression, I find that a one percent change in Labour support translates, on average, into about a 0.3% change in the vote for the Greens.

As important as this effect is, it is worth acknowledging that the other 70% of Labour’s lost vote during 2011 is estimated to be due to decreases in the total support for the left writ-large, which has dropped around three points since February.

Our Poll of Polls currently does not include the Fairfax polls, because they are new kids on the block. Including them, however, would make little difference to our predictions – less than 1% for each of the three largest parties, and substantially less than that for all the smaller parties.

We may try and add them for future iterations as we get deeper into the campaign.