Erratic election polls blamed on sample size

Many journalists are not fit to report on election polls, says marketing expert Janet Hoek, of Massey University.

The media have already begun asking predictable questions about whether the opinion polls "got it wrong", Professor Hoek said.

"Instead of asking whether the polls were wrong, we need to start asking whether the media are competent to assess this question."

Most poll coverage failed to include important information such as the response rate to the poll - that is, how many telephone numbers the pollsters had to call to get a sample of 1000 voters, she said.

"If these people [who refuse] differ in some material way from those who did participate, the survey estimates could be affected."

Pollsters interviewed by the Weekend Herald reported response rates ranging from 16 to 65 per cent, meaning up to 84 per cent of the households either did not answer the phone or refused to participate.

Professor Hoek said it was "little wonder that the public have become cynical about poll results and more reluctant to participate in surveys when these have been superficially reported and analysed".

DigiPoll director Gabriel Dekel said his own final poll, published in the Herald the day before the election, gave a larger margin to Labour than actually occurred, but the DigiPoll published a week earlier was much closer to the final result.

He believes the final week of a campaign may be simply too "volatile" for polls to be published at all.

TV3's polling firm, TNS, came closest to the actual result, giving Labour 40.5 per cent in their final-week poll and National 38.7 per cent.

Election night gave the Labour Party 40.74 per cent and National 39.63 per cent.