Something’s not right

Yesterday, TVNZ’s One News released their latest opinion poll conducted by Colmar Brunton (video above). Newsmedia all over New Zealand reported it as being disastrous for the Green Party who, based on that poll, were in danger of being turfed out of Parliament. But then we found out they were rounding the result for the Greens downward. And then I also noticed something far more troubling.

The poll results as reported really were disastrous for the Greens, who fell from 15% to 4%. National fell three points to 44%, and Labour surged from 24% to 37%, a clear indication that changing leaders really helped Labour, especially because in the “Preferred Prime Minister” beauty contest poll, Labour’s Jacinda Ardern is now tied with National’s Bill English at 30%. In the last Colmar Brunton poll, former Labour Leader Andrew Leader was tied with his then Deputy Ardern on 6%, meaning they were tied for third place. Things definitely are looking up for Labour.

But based on the reporting on this latest poll, Labour may have trouble forming a coalition, especially if the Greens are out of Parliament. The trouble is, the reporting was deeply flawed.

The Greens are actually on 4.3%, but journalists rounded it down to 4%. In ANY other field, rounding to the nearest whole number would make sense, but in political polling it creates a false impression of relative strength/weakness. After all, the Greens are closer to 4.5% (a figure some journalists used) than they are to 4%.

This matters first because it creates the false impression that the Greens will be out of Parliament. In New Zealand, a party must win 5% of the Party Vote or win one Electorate Seat in order to be in Parliament. While the Greens are pushing hard for the Nelson seat, it’s been held by a National MP who’s been in Parliament for two decades and been a minister many times. If the Greens lose that seat, they'll need to get 5% of the vote.

The reporting made it sound like they had to gain an entire percentage point when, in fact, they need 0.7%. That’s not splitting hairs because it’s a much lower number of votes they’ll need, and because one percent sounds much more difficult than seven tenths of one percent, and in election campaigns impressions influence voter behaviour (there is no President Hillary Clinton, for example, in part because journalists were reporting that polls showed she was certain to win). Of course, when you factor in the margin of error alone, the Greens could possibly be doing much better (or much worse) than was reported, but that complicating factor was never mentioned, either, as far as I can tell.

This fudging of the numbers made me look a little harder at the reported numbers, and I noticed something even more disturbing.

Look at all thse percentages reported: National 44%, Labour 37%, NZ First 10%, Greens 4%, Māori Party 2%, and TOP (The Opportunities Party) 2%. That adds up to 99%. So, we can guess that the other 1% must be for all other minor parties, including the Act Party, which is in Parliament because of deal with the National Party that allowed them to win the Electorate Seat of Epsom here in Auckland. On Election Day in 2014, Act got 0.69%, so if true this time that would mean that 0.31% must be divided among all the other very minor and fringe parties, right?

And this is where there’s a HUGE problem with the reporting of the poll: Most reports, incuding One News’ own televised report, only included the headline figures, which includes ONLY those who have decided who they’ll vote for. The reporting completely omitted any mention of the number of undecided voters. In fact, the poll found that 13% of voters were undecided (down from 20% in the July poll; this was reported on their website, not on television). That 13% could decide everything—the fate of the Greens, whether Labour or National forms government, how many seats any minor party that wins an Electorate Seat might get. Or, they might not vote at all, but no evidence supporting that possibility was reported, either.

Undecided voters matter for the Greens because their voters who abandoned the party did not all go to Labour. The Greens’ 15% figure from the last poll included disaffected Labour voters who “came home” in this poll, thanks to Jacinda Ardern, and that accounts for some of the Greens’ loss in this poll. But the Green’s natural level of support is around 8% of the electorate, give or take, and—trust me on this, because I’ve seen it first hand—hardcore Greens supporters don’t particularly like Labour (I’m being nice), so it’s highly improbable that all the sudden they jumped to Labour for the first time ever.

My suspicion is that a chunk of the Greens’ support called itself “undecided” in this poll, because they truly don’t know who else they’d vote for (and some of the Labour supporters who were previously “undecided” are now again supporting the party; given poll movements, this is a reasonable assumption). The most common party I heard Greens talking about switching to was Mana, which would be included in that 1% of “also ran” parties. So, clearly Greens voters didn’t actually go there.

This is not Colmar Brunton’s fault, of course, but the fault of One News and subsequent re-reporting. The undecided figure should have been widely reported because it matters so much—or were they more interested in spinning a narrative that the poll is “disastrous” for the Greens and they “could be out of Parliament” when the same poll, especially when undecideds are taken into account, also suggests that the Greens could do just fine?

The reality here is that, historically, undecided voters in New Zealand don’t make up their minds until late, sometimes in the last two weeks. This means that the election is still anyone’s game, and it was irresponsible for New Zealand news media to report this poll as drama when the data doesn’t support that as the only conclusion, or even necessarily the most likely.

We all deserve better.

Addendum – August 19: Since I posted this, I realised there’s one more point I should add: To win Government, Labour does not need to take votes from the Greens—or New Zealand First, for that matter. The undecideds are up for grabs, and if Labour gets a good share of that, and takes some votes from National, then it can form government with the Greens whose vote will likely recover to their regular levels of support if we get balanced and responsible reporting. Those who are promoting the narrative that Labour is trying to “steal” votes from the Greens are often mischief-makers on the Right, as well as some on the Left who didn’t realise, thanks to poor news media reporting, that there were so many undecided voters to draw votes from.