Election predictions

Polls are generally snapshots of opinion in the past, but they are often used to try to predict future election results. This can be interesting but accuracy is obviously not guaranteed.

An interesting approach here to New Zealand general election forecasts

This page provides experimental probabilistic predictions for the 2017 New Zealand General Election. It draws on multiple opinion polls, but goes a step beyond a straightforward poll aggregator in that the estimated voting intention from successive polls is used to forecast the chances of each party to actually win seats on election day, taking into account uncertainty. Polling results are also adjusted to take into account different polling firms’ past performance in predicting different parties’ results.

Every election can have it’s own unique pattern of lead-up polling so past trends won’t always predict future trends correctly, but this is as good as predictions can get.

It has some flaws but I expect some of these to be dealt with.

Oddly that doesn’t include a National coalition with NZ First.

What I like about this site is that it shows ranges of probabilities, showing that there is always margin for ‘error’ or inaccuracy.

This shows that a range of outcomes are possible for all parties. I presume it will be refined as we get closer to the election – it will be interesting to compare these predictions with later ones.

The poll results and trends show how volatile the polls have been for Labour, Greens and NZ First.

Perhaps ominously for National their results have been tightening in a downward trend since John Key stepped down, but it’s too soon for poll responses to the budget.

For the term to 2014 National had been trending down but that turned around, but well in advance of the election. They will be hoping for a late upswing this year but there is no sign of that yet.

NZ First have had late and significant upswings for previous elections after dipping mid term, but this term their poll support has held up. It’s difficult to predict whether this pattern will repeat this year or not.

Leave a comment


  1. Kevin

     /  30th May 2017

    NZF coalition with Labour? I call BS on that.

    From the link:
    “…the assumptions of allocation of nine key electorate seats”

    So basically nine assumptions?! With regards to predictive analytics I believe if you’re making even a few assumptions you’re on a bit of shaky ground, but nine?

    Personally PG I reckon the guy is a closet Labour supporter. 😉

    • PDB

       /  30th May 2017

      Obviously a left-winger……..

      By (deliberately) leaving out the most likely outcome (National-NZL First) it makes it look like the left will romp in as they are only comparing that possibility against the current govt arrangement which will have very little chance of remaining in govt as they are. This alone makes all the analysis a waste of time.

      Winston has already ruled out working with the Mana party (and the Maori party) as far as I’m aware so that appears flawed as well.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s