Is there a problem with political polling?

There is growing concern about political polling.

The Australian election result a week ago defied the polls. The Brexit referendum, Donald Trump elected president of the USA and the New Zealand election in 2017 all delivered different results to what polls were predicting.

I think the biggest problem here is the word ‘predicting’. Media has become obsessed with trying to predict election results. Polls are only designed to be approximate snapshots of voter leaning or intent, with typical statistical margins of error of 3-4% (as at the time the poll was taken, not on the election date).

ODT:  Is there a political polling problem?

In each of these examples, the Left has appeared stronger in polling data than the Right and, more importantly, the Left has polled higher than what the electorate has ultimately delivered.

It is worth noting that, in the above examples, polling data was very close to the final results. A little swing this way or that, added to margins of error, could be all the explanation required.

Don’t forget movement on voting intentions as voters close into making an actual decision on election day.

There is a school of thought, however, suggesting there is a trend at play here. The theory posits much of the major media organisations around the country and the world are staffed – at least on the “shop floor” – by a majority who swing left politically.

Is it possible an element of that presumed political thinking comes through in reporting?

There are two separate issues – polling, and reporting of polls.

Is it possible consumers of that news then feel it is more acceptable, when asked, to align themselves to the tone of the news stories and causes of the day, rather than more conservative views which may out them as morally outdated?

Is it possible the highly visual social campaigning undertaken by some on the political left – current strikes and marches are a fair example – compel those polled to err towards the left? And that, months later, in a private voting booth with just themselves, their personal views and a list of options in front of them, they opt for their own views – even if those views are more conservative than they’re willing to admit out loud?

There could be many reasons why this apparent trend in polling is resulting in a mistaken skew leftwards. It could well be the sample size listed in this piece is far too small to be worth analysing. Perhaps there is no issue at all.

I think there is very little issue. Polls don’t decide elections, media don’t decide elections (despite them appearing at times to do their best to influence them rather than report on them) – voters decide elections. It isn’t a contest between pollsters and voters.

Political arguments, intellectual disagreements and challenges to our world view are generally tiring and difficult. Is that what’s at play here?

Is the Left winning the publicity and polling battle, but losing the war?

That’s a different issue again.

And it isn’t entirely accurate – in the last New Zealand election the slightly more right leaning National Party won the election battle, but lost the coalition war to Labour, NZ First and the Greens.

There’s too many variations and variations to make any sort of statement about problems with political polling.

The best solution is to polls as approximate indicators of support prior to elections, and to ignore most media overstatements about their importance.

The media need to learn that they don’t decide election results beforehand. Voters have the only say.

Leave a comment


  1. Zedd

     /  27th May 2019

    The ‘fault’ with most political polls.. they are still basically FPP results: Lab V Natl, with the ‘minor parties’ as an add-on. BUT as is clear, we do have MMP & the obvious issue is, the total lack of ‘friends’ that Natl currently seem to have.. as they still have their heads in FPP.. US v the rest. :/ 😦

    • Duker

       /  27th May 2019

      Well thats partly correct, because MMP relies on nationwide voting numbers its a good match for nationwide polling. Actual results for each polling company vary enormously however.
      FPP is a much harder to pick from polls because each seat needs its own poll. That rarely happens. The old days of two main parties going head to head which suited polling are gone. Smaller parties can split off the vote, which shows up in NZ electorate seats where Greens-labour vote exceeds that of national but they win the seat, Auckland Central a prime example. Australia has preferential voting which means that labour can often win on Greens and others preferences, which works for Liberals too with its allies, the nationals.

      UK polling is a nightmare with partys like Lib Dems who are strong in some areas coming into mix , same as Scotland where the SNP faces both Conservatives , Labour and Lib dems ( in some areas)

  2. duperez

     /  27th May 2019

    Polls are simply a function and symptom of the type of world we live in.

    The hope that they are or can be an intelligent part of an intelligent world and intelligent political discourse is a vain one.

    • Duker

       /  27th May 2019

      I see turnout at EU elections this week in UK was about 25%. How can a poll by randomly calling people get anything but a majority for ‘Dont know/dont care’

  3. Duker

     /  27th May 2019

    Australia polls had a different sort of problem, the voters use preferences so the pollsters have to use ‘recipes’ to come up with a ‘two party preferred’ number.
    And thus they did for labour in lead -who still had less 1st preferences than coalition for years- and no PG they did run polls in the last week before the election which don’t really differ from election day.
    It would be interesting to look at the polls ‘first preference’ numbers and see how they compared with actual result from ballot box.

    Trump result wasnt really a failure of the nationwide polling, as they predicted that correctly, but president isnt chosen from a popular vote. The media used the perception that the person with the lead in nationwide vote would aslo win the electoral college ( mostly that happens too)
    Another fail from the media was the use of ‘statistical chance’ to see a result. It doesnt work for climate change and it doesnt work to come up with absurd predictions like XXX candidate /party will have 95% chance of winning. ( Like climate change they run 100s of simulations and the use stats theory to pick the number that is 95% ‘correct’. However its a prefect example of how comparing predictions to results should invalidate the whole process)

  4. Maggy Wassilieff

     /  27th May 2019

    Keeps a bunch of psephologists in employment, though.

    • Kitty Catkin

       /  27th May 2019

      (pretends to know what that means)

      Did you know that acnestic means the part of the back that can’t be reached when it itches? I love the OED.

      • Maggy Wassilieff

         /  27th May 2019

        I did not…

        • Kitty Catkin

           /  27th May 2019

          Opsigamy;marrying late in life.

          Gobemouche; one who believes anything, no matter how stupid.


Leave a Reply

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

You are commenting using your 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