One - by Reid research - was pleasing for National. Another - by Colmar Brunton - pleasured Labour.
Over on The Standard, the former was met with howls of despair and derision. Reid Research, apparently, were fully paid up members of the Illuminati-Space Lizard conspiracy. Stephen Joyce personally wrote in the numbers. Or something like that.
The second poll was give a much warmer reception. Indeed, I think some commentators on The Standard would have married the poll, so entranced were they by its appeal. Someone, writing under the anonymous NOTICES AND FEATURES nom-de-plume, rushed out a
Whoa, whoa, whoa! It does no such thing!
The claim that a Colmar Brunto shows a Reid Research poll to have been completely inaccurate is not correct and whoever is writing under the NOTICES AND FEATURES byline should be ashamed of themselves. I suspect they know they are doing a Bad Thing but decided to put it out there anyway for purposes of propaganda.
Comparing the findings of a Colmar Brunton poll and a Reid Research poll does not show one is accurate and one inaccurate. Even comparing several CB polls to the RR poll does not invalidate it. It merely shows that the two companies have different sampling methods and / or process the data differently and obtain different results.
The only things that would show RR’s poll to be a rogue would be further RR polls that show significantly different results; or the election result itself.
You’ll recall that in Britain, Survation was mocked when its polls showed Jeremy Corbyn’s Labour Party closing in on May’s Conservatives. Other companies showed the Conservatives well ahead. It turned out that Survation was correct and they called the result more accurately than the other companies, by considerable margins. You do not know which polling company has the right formula (or the least wrong, more like) until the real votes are counted. Until then, you can only regard the forecasts with bemused curiosity or phlegmatic disdain.
Do not dismiss the latest RR poll until it is invalidated by further RR polls. The alternative is psephological madness.