Final Election Forecast update

I’ve updated my Election Forecast with the results of the final Colmar Brunton and Reid Research polls. Either National or Labour could form government, but they would both need the help of NZ First, Labour would also need to add the Māori Party to the mix.

My forecast has the Greens creeping closer to, but still below, the 5% threshold. I feel that my weighting against them has been too aggressive, and I would be surprised if they don’t make it back to Parliament.

While National have gained ground in the last two weeks, it will still be a very, very close result. Labour will make massive gains, and with a good ground game on Saturday could still take the prize.

It’s certainly been the most interesting election I remember in years.

My genuine thanks go out to all the candidates, staff and especially volunteers. No matter what party you stand for, the work you’ve done over many months ensures that New Zealand has a healthy democracy. Get some sleep on Sunday!

My electorate predictions

Electorate level polls in New Zealand are problematic.

They almost always use tiny samples, and pollsters have significant problems knowing the people they’re calling are actually enrolled in the seat they’re asking about.

And taking all that into account, it’s never done regularly so you can’t even look for a simple trend.

So, for those reasons, predicting electorate level results is difficult at best, and it’s why I haven’t built any change in electorates into my Election Forecast.

However, I promised a few people I’d write a list of which seats I think will change hands next Saturday. Here they are:

Auckland Central – Labour gain from National

Ōhāriu – Labour gain from United Future

Maungakiekie – Labour gain from National

Christchurch Central – Labour gain from National

Waimakariri – Labour gain from National

Whanganui – Labour gain from National

Waiariki – Labour gain from Māori Party


So there you have it, seven seats I think Labour will pick up on the day. And I no longer think any of Labour’s seats (eg Hutt South, Te Tai Hauauru, Napier) are at risk.

If I’d written this list two months ago it would have been very different!

Election Forecast update – things are tightening

I’ve updated my Election Forecast with the results of tonight’s TVNZ/Colmar Brunton poll. The update results in Labour gaining two seats, one from National and one from the Māori Party. The Greens are still forecast to be under the 5% threshold.

This weekend I’m going to go through my data and make my predictions for which electorates are going to change hands. I’ll be very interested to see what others guess as well.

Nine days to go.

Election Forecast updated – more bad news for the Greens

I’ve updated my NZ election forecast with the results of tonight’s Newshub/Reid Research Poll. On current numbers either National or Labour could form a government, but both would need the support of NZ First, and Labour would also need the Māori Party.

More bad news for the Greens with their predicted party vote result dropping to 2.5%. Ouch.

Election Forecast updated – and what Parliament would look like

I’ve updated my Election Forecast with the results of tonight’s Colmar Brunton poll. While the gap has significantly closed between National and Labour, the Nats are still slightly ahead in my forecast. Both major parties could form a government with the support of NZ First.

I’ve also done a breakdown of who would actually make it into Parliament on these numbers. With Labour polling almost 20 points higher than they did in 2014, it’s a fairly safe bet that they’d pick up additional electorate seats. For the purposes of this I’ve made a guess that they’d pick up Auckland Central, Ōhāriu, Maungakiekie, Christchurch Central and Waimakariri – which were all less than 3,000 votes out of reach in 2014.

If you disagree on which seats would change hands, you can use my National and Labour calculators to see who would make it in at any level of support.

Here’s what Parliament would look like:

Continue reading “Election Forecast updated – and what Parliament would look like”

Election Forecast – the end of the Greens?

I’ve updated my Election Forecast with the latest round of polling. Jacindamania is well and truly happening – Labour have made huge gains at the expense of every party.

There is one massive change in this update – the Greens are forecast to only get 4.7% of the party vote, which would mean they would fail to get back into Parliament.

While only one poll has shown the Greens below 5%, my model adjusts for known house biases – this is essentially replicating the phenomnea whereby the Greens perform worse on election day than in polls.

There would only be five political parties in Parliament and the landscape would be significantly different. Both Labour and National could potentially form governments, with the two options for forming a government would be National + NZ First (plus potentially Act and/or the Maori Party), or Labour + NZ First + Maori Party.

Can the Greens lift their game and get back over the line? Quite possibly – my forecast is only showing them 0.3% off the mark.

The next three weeks are going to be fascinating.

Huge movement towards Jacinda and Labour

I’ve updated my Election Forecast with the results of two recently released polls – one from UMR which ended three days ago, and one from Roy Morgan which ended on the seven days ago.

Neither of these polls taken individually offer much interesting new information – they both show a closing of the gap between Labour and National as swing voters start moving towards Jacinda Ardern’s Labour Party.

However, what is interesting is that adding these two new data points has tipped my election model substantially.

It’s designed so that individual polls can’t have considerable impact, but the sheer weight of support now behind Labour has meant that with these two new polls my forecast for Labour seats has sky rocketed from 36 to 44 – with them coming from National, NZ First and the Greens.

On these numbers, Jacinda Ardern could form a government with the backing of NZ First and the Greens, and Bill English could form a five-headed hydra of a government with ACT, the Maori Party, United Future AND NZ First.

I would be very worried if I was a National back bencher.



Election Forecast updated

I’ve updated my Election Forecast with the results of today’s Colmar Brunton poll. It’s good news for Labour, and bad news for National and the Greens.

You may have noticed that Colmar Brunton have also done an electorate-level poll in Ohariu which puts United Future’s Peter Dunne 14 points behind Labour’s Greg O’Connor. It’s only one poll with a small sample, but it does not surprise me. If there is any further polling in that seat showing similar results, I’m going to update my forecast to give the seat to Labour, meaning United Future and their over-hang seat will be gone from Parliament.

Two important changes to my Election Forecast

I’ve made two changes to my Election Forecast.

Firstly, I’ve managed to import all the UMR polling data that has been released into the public domain, going back to January this year (though polls older than 90 days will have no impact on the forecast). Given they are polling at least fortnightly I’m really hoping their results continue to be released.

Note that UMR polls don’t release numbers for ACT, Maori Party, United Future or any other minor parties.

I’ve also fixed a bug that was impacting the weightings. The impact of this will be that the forecast will be more heavily impacted by newer, bigger polls. Older and smaller polls won’t have as much of a factor in the result. While it wasn’t my intention, this means the forecast will start to show the impacts of the Jacinda Ardern affect sooner.

Bring on more polls!

Election Forecast updated

I’ve updated my Election Forecast with the results of today’s astonishing  Newshub / Reid Research poll.

Because of the nature of my forecast, polls as old as 90 days will factor into the result. Which means a significant sudden changes won’t be as noticable as they are in raw numbers.

So while Labour is up 9% and Jacinda’s preferred PM ratings will have Bill English very worried, my forecast is very conservative and only shows one seat changing hands from National to Labour (which is still a very good movement).

As the campaign progresses and polling becomes more regular I may look at doing a version of my forecast with a much shorter half-life, so giving significantly more importance to more recent polls.

Let’s do this.