In advance: Preferred PM means nothing

Posted: December 12, 2016 in democracy, elections, New Zealand
Tags: , , ,

I’m sure we’re going to see some very excited pollsters looking to be the first to compare Bill English and Andrew Little. (It’s probably going to be Colmar Brunton in first, as they’re the only ones who have been regularly running polls, but I expect the leadership change will prompt the bit players to show up suddenly)

In advance of that, (because saying it afterwards doesn’t have the same impact) and as someone who is frequently on record as saying that polling is relatively accurate1, I would like to remind people that Preferred PM almost literally means nothing. Almost the only important part of NZ polls is the party vote question, which actually has some reasonable predictive power for the general election.

Preferred PM also doesn’t do a particularly good job of predicting who will be in Government, (most PMs lose the Treasury benches while still being more popular than the other person, because a PM is usually more popular than their party as a whole) rather it’s more an indicator of whether a leadership coup is likely for either the Government or the opposition. The only time it generally means anything is if the Leader of the Opposition actually overtakes the PM during the election campaign and stays consistently ahead, in which case, we would expect a change of government because that’s a Big Deal.

It’s also a metric that behaves a little differently for left-wing and right-wing leaders. The Left tends to not be very keen on uniting. (which, fortunately, is actually an advantage under MMP, so long as all your parties are over-threshold or winning an electorate) The Right loves it, so long as they view their leaders as successful, so even if you restricted answers to actual leaders of political parties, (which pollsters do not, it’s an open question, hence why Helen Clark kept showing up even after she was long gone) you would still likely get a bunch of people saying they want Winston Peters or Metiria Turei or even Hone Harawera to be PM, wheras even Bill English’s previous preferred PM polling leading up to his disastrous loss was still higher than Labour’s has been in opposition.

What will be interesting in the upcoming preferred PM results is how many people simply answer John Key because they don’t follow politics. I expect John Key will fall off the preferred PM polls faster than Helen Clark did, but he will probably show up at a significant level for at least the next two.

So, even if Little outperforms English in the next Preferred PM poll, remember:

  • That’s only important if English is actually enough of a drag on the Party Vote to bring Labour into fighting distance of forming a Government.
  • A lot of people won’t be able to name Bill English unprompted just yet, even if they think he’s not doing badly.
  • There won’t really have been time for more than an initial impression, and it’s actually more important how the trend shows up over time. (ie. whether Little and the Opposition in general start gaining ground over time)

And likewise, if English outperforms Little, that doesn’t even necessarily mean he’s on his way for a win, just that he hasn’t immediately lost all of National’s former support. There’s still a possibility that his support will trend downwards if people want to give him a chance but don’t like what they see as time goes on.

This is also why Bill English rather smartly has ruled out commenting on the timing of the next election until next year, as he will want to see not only how National is polling immediately after the transition, but he’ll want to see whether the government’s poll numbers, both in internals and public polling, are moving up or down. If it’s down, expect an early election, if it’s up in early polling, expect to wait.

1 By which I mean, it’s very rare for polling to be outside of its margin of error. Even the Clinton-Trump result was in the margin of error in predicting the popular vote for most polls, it was just how the states played out that was surprising.

This does require some understanding of political poll margins of error, but basically, the largest a margin of error is ever likely to be is 3.1%, so long as you don’t look into demographics within the polling. (New Zealand public polling is already underfunded enough that it’s not well served outside of election year in general, so demographic analysis is a bit too fancy to ask for public polling) This is essentially because polling 1,000 people is the most economical tradeoff between accuracy and cost, and the margin of accuracy forms a curve based on standard deviation where a 50/50% polling result has a margin of error of 3.1%.

 

As an aside, I’m working on some maths which does useful polling averages and reporting with margins of error which I will probably start posting to my blog after the next poll.

Advertisements

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 )

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s