Polls could be all wrong, says expert

People who are undecided or unlikely to vote in this year's election could yet make a monkey of the political opinion polls, a Massey University expert says.

Marketing lecturer Janet Hoek says these are the two critical things that the polls do not tell us but are the very things that could decide who wins on September 17.

The undecideds, she said, could be genuinely undecided, just plain not interested or of the mind-your-own-business variety.

If there are many of them, they could influence the polls in ways that are not being seen. As for the don't knows, Dr Hoek said their numbers were reflected in the rising number of people who had not voted since the first MMP election in 1996.

Since then the no-vote has climbed by 11 per cent, so that by last election just over three-quarters of those eligible to vote did so.

Whether that will change this election because of the tax cuts and tax relief on offer will not be known until after the vote.

However, Dr Hoek said with groups such as students traditionally not voting in large numbers, Labour might have a problem.

"All of this makes the results of the polls hard to accurately interpret, because unless we know the answers to these two questions, we can't say for sure that any single poll is giving us the right answers."

Dr Hoek said even the latest One News-Colmar Brunton poll, which put National eight points ahead of Labour on 46 per cent of the vote was not as clear-cut as it seemed for these reasons.

Its "true value" could be in the 43-49 per cent range, not making it hugely different from the last Fairfax/ACNielsen poll, which put National on 44 per cent, three points ahead of Labour.

So a trend or a rogue poll?

Dr Hoek said Colmar-Brunton had been tracking a National rise and Labour decline for some time.

Whether that is a trend or something peculiar to how the poll operates will not be known until after the election results are in.

Dr Hoek said it was a "fair comment" to say the latest Colmar-Brunton poll is a rogue based on statistical probabilities.

However, that did not mean it was wrong.

As for the influence of the polls, they could make people who are interviewed for them more interested in the outcome.

They could also have a "bandwagon" effect -- prompting people to jump aboard what looks like the winning party, or do the reverse, persuading voters to vote for the underdog.

- nzpa