Public polls are all wrong, and this is why

I was sitting in the car waiting for someone to come out of a meeting and had a talk back station on.   As I tuned to the station, the very first statement I heard was (from memory)

All the public polls are biased in favour of National.

I immediately reared up.  What a load of bollocks.  But as the caller continued I actually came to agree with him.

We all say things like

  • National will drop 5% off their poling on election day
  • Labour will drop 3-5% off their polling on election day
  • The Greens always poll better and generally do 1-3% worse on election day
  • NZ First almost always polls poorly and do 2-4% better on election day

These are truisms and we make a quick mental adjustment to any poll result to ensure we don’t get excited, because we, as pundits, know better than what the poll is telling us.

The caller went on (I paraphrase)

Before the 2014 election National was polling around and above 50%.  There was no way they were going to get that.  And of course, they never did.

It was roughly at that point that I stopped debating against the caller in my head, and asked the question:  Why is it that we accept poll data that we know to be wrong?

To some degree, Labour is suffering a persistent string of poor polls while they say their own polling is better.

But the real losers are the smaller parties.

All we hear is that TOP, ACT, Conservatives and United Future are languishing in the kinds of numbers that makes them immediately irrelevant to the voter.

Ask polling companies, and they will explain that “this is what people say they will vote if an election was being held today“.   But that’s a cop-out.    If you consistently measure voting intention and it is consistently different from what really happens, then why are you accepting your methods as being accurate in any way?

In the end, when the polls consistently report National’s result higher than what they will actually achieve on election day, and the results for, say, ACT are consistently lower than what they will get on election day, why are we accepting these polls at all?

Parties such as TOP may very well have 3-4% support out there (perish the thought) but is reported at 1%.  Potential TOP voters will now choose not to vote TOP because they can see at 1% they are wasting their time.  Tell them that TOP is at 4%, and a potential TOP voter may be motivated to try and get it over the line.

There is a precedent for this with the Conservative party in 2014.   They were polling in the 1.5 to 2.5% range, but were well into four percent on election day.  Had the publicly damaging MacGregor resignation not happened, who knows what could have resulted.

It may suit “us” in general to have these broken polls, but the question is a fair one:   Why are the MSM commissioning polls that they themselves, as they report them, know to be a poor representation of what will eventually happen on election day?

A large event that will affect politics between now and election day aside, most of the pundits can just about predict the election result now, within 1-2%, by making the above-mentioned adjustments to public polls.

So yes.  Public “professional” polls are consistently biased in favour of National.   It’s not even the case that it under-reports, even once.

And that’s not good polling.

The question now becomes:  why?    Why is this self-delusion going on?  Why are Commie Dan and Paddy standing in front of their monthly results and not demanding that the companies fix their broken polling so that they aren’t forced to constantly prop up National with results we all know won’t be a reality on election day?

 

– Pete


Do you want:

  • Ad-free access?
  • Access to our very popular daily crossword?
  • Access to daily sudoku?
  • Access to Incite Politics magazine articles?
  • Access to podcasts?
  • Access to political polls?

Our subscribers’ financial support is the reason why we have been able to offer our latest service; Audio blogs. 

Click Here  to support us and watch the number of services grow.

38%