Rely more on the real election result

Throughout election campaigns, pollsters seek to communicate public opinion. But polling can raise more questions.
... How should voters, and those who communicate to them, interpret poll results?
. ..What role do polls play in voting behaviour? Does superficial analysis influence behaviour?
... What is the methodology used?
... Are the polls designed to give genuine information - or to create news entertainment?
... Did the poll sample less people without landlines?
... Did the poll sample less people with poor grasp of English? - e.g. are some migrants and some marginalised people not having their voices heard?
... Did the increasing number of people who refused to answer affect the poll's validity - e.g. would the opinions of those who would not answer be similar to those who did answer?
... If only 30% of those asked to respond, do so, does this bias the result?
....Or is volatility in poll reports really reflect volatility in the electorate?
... When should the quantitative measure offered by opinion polls (e.g. a phone call to your home at dinner time) be balanced by qualitiative measures (e.g. a pre-planned focus group session)?

Factors such as the questions, timing and sample of polls can significantly alter results. In assisting voters to understand what poll reports may be doing to their judgement, it is important they be reminded to compare only like polls – and to look for trends.

A case study: the NZ 2005 general election
In the New Zealand 2005 general election the polls contradicted one another frequently.

In the final days of the campaign, a One News Colmar Brunton poll put National six points ahead.

A New Zealand Herald DigiPoll gave Labour a lead of almost seven points.

A Fairfax New Zealand ACNielsen poll earlier gave National a six-point lead. When Fairfax poll results were adjusted to include only those who said they intended to vote, the poll results predicted the election night result within its “margin for error”.

A TV3 TNS poll gave a possible one-seat lead to Labour – that poll, at least, reflected the election night result.

Analysis
Whilst individual polls gave different results to other polls taken at about the same time, political analyst Colin James said the trend in a "poll of polls" pointed to a similar result to that given by the voters in the "poll that counts" - the election.

Reporter Simon Collins said he found that his interviews with many New Zealanders pointed to North Islanders being more interest in Treaty of Waitangi, migrant and with roading issues than were people in the South Island. He found that older Labour Party voters had started to shift to vote National, and that young people, attracted by Labour's student loan policy, had tended towards Labour. "

Analysts and politicians sometimes use the quantitative and qualitative information they get publicly and from private polls to shape their judgements - an important factor in understanding how opinion polls work.
Updated 25 September 2005