Devious Polls

Just before the 2005 election, a poll was released by TV1/Colmar Brunton that showed Rodney Hide (30%) well behind Richard Worth (44%) in Epsom.

This news was covered extensively by media and blogs across the country, mostly gleeful that ACT was finally dead and National's chances of forming a government were dying with us. Even Kiwiblog covered the poll, with David Farrar publishing the results with no further comment.

The catch? Rodney wasn't behind Richard Worth, the poll was miles off because the question it asked was awful. The poll didn't ask people which candidate they were going to vote for, it asked them which PARTY that the candidate they were going to vote for was from. Rodney Hide pointed this out immediately, as did ACT on Campus on this very blog, but it was no use.

The media coverage of the poll hit ACT hard in the party vote polls and probably cost the party a couple of extra MPs at the election a few days later, as many ACT supporters switched to National, worried that their vote would be wasted if they voted for ACT and Rodney lost to Richard Worth.

Seeing this risk, ACT paid for Roy Morgan to run a new poll in the last days of the election campaign and the result showed Rodney Hide in a clear lead to win Epsom - which of course eventuated.

But did the media cover this poll as widely as the first? No. Did anyone write an opinion piece comparing the two questions and questioning the polling companies methods? No, not until months after the election, at least. In fact, the only media coverage the poll did get, accused ACT of faking a poll in an attempt to help Rodney Hide win the seat - assuming that because ACT paid for the (independent - not internal) poll, that ACT must have set the question, run the poll, processed the results and spun it to show Rodney winning. Not PC made similar claims on his blog, only to apologise and retract them later.

Why point this out now? So many years on? Well, we don't know, and we never will know, the motivation behind the original Colmar Brunton poll and the author of the question that was used. But we can know what to look out for when reading polls in the future, and media should do the same.

Who RAN the poll is much more important than who PAID for the poll.
The question asked is VERY important. If the question that was used isn't publicised, this should be an immediate red flag.

This brings us to a more recent poll that was 'leaked' to media, supposedly a poll of Epsom votes, asking them who they will vote for as their local MP at the next election. The poll is (probably) of people in Epsom, but that's as far as the accuracy of this poll goes.

At the time, the poll was widely reported in the media (sound familiar?) as a poll of centre-right to right leaning Epsom voters, showing that, if they ran, John Banks, Colin Craig, Don Brash and Winston Peters would all beat Rodney Hide in Epsom. Columnists and left-leaning blogs picked up on the story and crowed about the upcoming death of ACT and the potential defeat of National because of ACT's demise (again, sound familiar?).

The only problem? No-one knew who paid for the poll, who ran it, or what the question was. A few days later, Fran O'Sullivan wrote about the poll in the New Zealand Herald and revealed more information.

The poll was commissioned by Colin Craig and, as mentioned earlier that shouldn't in and of itself disqualify the poll, however it is a bit suspicious when the person who commissioned the poll is also the one who 'leaked' it to the media, as admitted by Fran O'Sullivan. The poll was by Research First who have a reasonable reputation, however, the question that was asked was still not released.

At this point, if you've just clicked on that NZ Herald link, you might be a bit confused, as the question has been added to the article since it was first published. The question was NOT included in the initial article and i remember commenting on various friends Facebook pages that, without the wording of the question being made public, the poll shouldn't be trusted. Now the wording has been released and added to Fran's article, we can assess it.

Remember, this is a poll that was 'leaked' to the media and purported by the leaker and the media to ask Epsom voters who they would vote for in Epsom at the upcoming election.

However, now we learn that the question was actually:

[The question,] asking all poll respondents to rank their preferred leader for a new party in place of Act, put Banks at the top with 37 per cent. Craig had 27 per cent, Brash 21 per cent, Peters 16 per cent and Hide was on 14 per cent.

Does anyone else here see the problems with this question?

1) The question has nothing to do with Epsom.
2) The question has nothing to do with voting, or the 2011 election.
3) The question asks respondents to ASSUME that ACT (and therefore Rodney Hide) are being replaced.

I'm surprised that ANYONE picked Rodney, given that question. Heck, if you asked me who I wanted to lead a (new) party to replace ACT out of that list, I'd probably go for Brash - I would assume Rodney wasn't an option, given you've told me he's being replaced.

So why did NO media pick up on this issue? Why did Fran O'Sullivan not figure out what was going on, given she had all the information and was writing an article on it? In fact, Fran simply pointed out in her article that February polling by Research First showed similar results. Of course it will, if they're still asking that same silly question.

Do the media (in general, not just Fran O'Sullivan) not understand how polls work? Do they not understand how they can be twisted to fit any purpose (especially when you don't know what the question was)? Does it fit the media's narrative or personal preference better to ignore the facts? Or is it just easier to write a story that someone hands to you on a platter, rather than doing the research and actually thinking for yourself? I'll let you decide.