David Farrar and his false polls



Being the Government mouthpiece and official pollster, Farrar must at all times strengthen the methodology of his land line telephone polls.

This rehash of an older post Farrar has written from a pollster blog is effectively him yelling that his version of reading entrails is the most valid, no other oracle shall be considered.

Let’s see how the master of lite right spin justifies his flawed methodology.

It is worth recalling the stat – you’d have to call almost 17 cell phones to find 1 person who doesn’t have a landline.

No. As Andrew at Grumpollie notes, the supposed 6% who only have landlines is hypothetical. Using that hypothetical to justify a statement is nonsense especially when we have research showing the number is much higher, as Brian Rudman points out

The last household Census in 2006 highlights that the worst affected areas coincide with significant clusters of state housing.

The Pt England Census block topped the poll in Auckland with 105 households having no access to telecommunications systems of any sort. Close behind came Pukekohe North (99), Clendon South (93), Otahuhu West (87), Harania East (87), Papakura East (81), Otara East (78), and on goes the list through the lower socio-economic areas.

TDB Recommends NewzEngine.com

Farrar’s next point is a clever twist. He asserts that the difference in vote for cell phone only voters would have to be extreme to make much difference. Seeing as it is mostly young people and the poor who have no landline at all, that demographic in terms of age and class and socio-economic status IS going to be wildly different from the middle class landline pool that telemarketers are fishing in.

Farrar claiming it would have to be large (when it is) is just an attempt to downplay it and his claim that pollsters take these factors into account when telephoning misses the point that it’s increasingly difficult to get that group’s opinion. The weighting placed upon the one conservative Maori who is at home with the wealth to have a landline is going to generate a perception that may have little to do with reality.

The real power of these polls is what they do to the wider psyche of the electorate. Constantly repeating that the Government are in majority creates a spiral of silence, where the voter ends up being desensitized and intimidated into not voicing opposition. The mainstream media spent 3 years telling NZers that Key had over 50% and that resulted in the lowest voter turn out in 120 years. The criticism is that these polls are not reflecting public opinion, they are manipulating it.

As the Producer and host of the only TV Election show that called the 2011 election closest by not relying on landline polls, I think I’m allowed to put the boot a bit into Farrar’s fantasy.

The Newspaper industry has fallen to its knees hemorrhaging cash and its future because of their flawed methodology destroying the credibility of their revenue valuations. Once upon a time Newspapers could con and dupe advertisers into paying vast sums of cash for a full page advert because they could falsely claim a million eyeballs would see that advert. The cruel efficiency of the internet can now tell advertisers exactly how many people have seen that advert and the numbers are far less than the pretend numbers Newspapers once used.

As the landline becomes a dinosaur technology and the pollsters get further away from reality, perhaps Market Research will face the same demise Newspapers have.

The recent flurry around TV3s recent Reid Research Poll utterly failed to notice that the very same company were out all three weeks leading into the 2011 election, yet it was enough for Patrick Gower to invent a narrative that National are unassailable.

How convenient for the status quo.


  1. Also, who answers the landline? I have one (because it got bundled with internet) but I generally won’t answer, as the line’s unlisted and I don’t give out the number, so any call is intrinsically spam.

    If a family with adult kids has a landline, do all the kids count as having access to it even if they don’t answer the phone.

    And also, what the pollsters do is to apply correction factors based on polls before the last election(s) vs their raw poll results. That’s static, it assumes that the underpolled (like non land-line owners) keep thinking the same as they did at the last election. It doesn’t measure change.

  2. This is the scary thing about the polling that we see coming from the MSM:

    A large proportion of non-voters cited the polls predicting the National Party’s victory, and decided the election was a foregone conclusion. The percentage of non-voters who said this was a factor was far higher in 2011 than in 2008.

    They are, quite literally, destroying our democracy.

  3. How do they deal with the landline/cellphone divide in the US where some pollsters were exceedingly accurate?

    • If you are referring to Nate Silver, I think his methodology was to take a sample from all polls, but when taking said samples take into account the standard ideological biases that each different poll often has. Rather than relying on a series of individual polls, as most political media in New Zealand does (i.e the song and dance that occurs when a single new poll comes out, showing changes within the margin of error from the last one) Silver’s work was more accurate through being more comprehensive. From memory he also managed to pick the electoral college winner in all 50 states, showing the value of his non-partisan methodology.

    • All of the polls in the US were wrong. The only one who was accurate was Nate Silver and he was doing far more complex stuff.

  4. I am predicting a Labour/Greens landslide victory at the next election. The current polls don’t reflect the huge non voters (800,000) from the last election turning up to vote National out.

     More than anything the Government ‘not’ the opposition parties & their policies, will get people out to vote.

  5. The census just undertaken will reveal how many households rely on cellphones only, without a landline.

    If it’s anywhere around 10%, Farrar and other pollsters can kiss goodbye to their credibility.

    The only accurate polling is going door to door. That may be more expensive and time-consuming, but no other method will capture a section of all potential voters.

    • I thought the cell phone question in the census was poorly worded and gave me pause before answering it. It asked if there was a cell phone in the household most of the time.

      Cell phones are attached to individuals and not households. Mine goes where I go.

Comments are closed.