Media release

24 November 2011
Horizon Research

More intend to vote, Labour gains

The number of people saying they will definitely vote in this weekend’s general election has risen 7% in the five days to midnight November 23.

Labour is the main beneficiary of the higher intended turnout, gaining an additional 1.8% of people who have a decided party vote, are undecided with a preference all of whom are enrolled and definitely intend to vote.

An Horizon Research poll of 2,807 of these Definite Voters, conducted between 7.30am Tuesday November 22 and midnight Wednesday November 23, finds New Zealand First has held onto substantial gains made among this group since its leader, Winston Peters, announced the party would not enter a formal coalition agreement with National or Labour.

The nationwide Definite Voter snapshot indicates New Zealand First will hold the balance of power with up to about 13 seats, but subject to a 1.8% margin of error.

Some 82.1% of the 18+ population now say they will “definitely vote”. This is up from 75.1% in the HorizonPoll conducted between November 19 and 21.

At the 2008 election 75.1% of those aged 18+ turned out, or 79.5% of those aged 18+ who were enrolled.

As the number intending to vote expands, it dilutes the vote share of parties even if they are retaining much or all of their 2008 vote share.

The survey is weighted by age, ethnicity, education, employment status, personal income and party vote 2008 to ensure a representative sample of the New Zealand population aged 18+. The maximum margin of error at a 95% confidence level is ± 1.8%.

The November 22-23 survey finds parties with the following Definite Voter party vote shares:

ACT New Zealand
2.7%
Conservative Party of New Zealand 5.2%
Green Party 12.4%
Labour Party 28.6%
Mana Party 2.8%
Maori Party 1.1%
National Party 33.4%
New Zealand First Party 10.8%
United Future 0.9%
Other party 0.7%
Choose not to vote 0.0%
Don't Know 1.3%

Horizon says the result shows National is holding at about or just above its share of votes among the adult population 18+ at the last election (32.9% in 2008), while Labour has increased its total 18+ population vote share from 25% in 2008 to 28.6% in the past two days.

Leaders’ debates – possible influence

The current survey started the morning after Monday night’s Labour-National leaders’ debate on TV3.

An HorizonPoll of 1000 electors taken the day after the first main party leaders’ debate on October 31 delivered Labour a 4.7% support lift to 30.4% of net potential voters at November 1 and this eroded again later. National suffered a 3.6% fall in support the day after the first leaders’ debate and recovered some of this later. On November 1, Horizon found Labour leader Phil Goff had won the debate by 7.4% over National leader John Key.

A similar post-debate support spike may have followed Monday’s debate this week. The survey captured about 240 respondents who saw the leaders’ debate last night on Television One.

Horizon polling between November 8 and 15 found 59.1% saying their party vote was being influenced by discussions they have heard between party leaders. 56.8% said leaders’ personalities were having an influence.

But the main influencing factor, however, at 90.8%, was confidence in a party’s ability to manage the economy. 88% also said the “state of the economy” was an issue influencing their party vote choice.

National has focused strongly on economic management.

Labour’s campaign has focused significantly on asset sales. Some 64.4% said the partial sale of assets would influence their vote. Unemployment (74.4% influence) is another area pin-pointed by Labour.

Labour support:

Between November 22-23 Labour was retaining 64.7% of its 2008 voters. 17.6% had gone to Green, 6.6% to National and 6% to New Zealand First. It was picking up 7.3% from National and 12% from Green.

Among those who chose not to vote in 2008 – an important group of more than 180,000 electors if they vote on Saturday - Labour picks up 38.4% of those who are now Definite Voters, National 22.9%, the Greens 18.7%, New Zealand First 8.3%. This indicates that a higher turnout, if it includes 2008 non-voters, will favour Labour more than any other party.

New Zealand First:

New Zealand First has highest voter loyalty at 84.6%, compared with National at 71.3%, the next highest.

Of its support, 16.7% comes from those who voted for “other minor parties” in 2008, 6.2% from National, 11.3% from Labour, 10.5% of Jim Anderton’s Progressive Party, 5.8% Green and 2.2% from the Maori Party.

Conservative party:

The new Conservative party is showing at just over 5%, but the 1.8% margin of error indicates it could be either above or below the threshold to have seats in Parliament if its leader, Colin Craig, does not win the Rodney seat.

As the party is new Horizon says its ability to weight results for it against 2008 party vote is reduced and the number may be less reliable. However, analysed against party voting in 2008, the Conservatives are drawing 25.1% of their support from people who voted for minor parties in 2008. This “other minor party” category excludes NZ First, United Future andJim Anderton’s Progressives but includes parties like the Family and Kiwi parties of 2008.

The result among Horizon’s Definite Voter group indicates no main party will be able to govern alone.

The results, as other Horizon ones have found throughout and before the campaign, will depend very much on candidate-vote support for minor parties in seats like Epsom, Rodney and Ohariu, and final party vote support levels, specially for the Greens, New Zealand First, Act, Conservative, Maori and Mana parties.

HORIZON’S METHODS

Horizon’s Definite Voter poll result cannot be directly compared with other published polls which are typically for samples of 750 to 1000 respondents, who have answered a telephone landline. All publish decided voters as a percentage of 100. Some do not publish the remaining undecided vote, or measure which way the undecided group is leaning. Some do not publish results for the undecided group at all. All do not publish results for those who will not say or will choose not to vote. None of those published weight on 2008 party vote. Their margins of error are typically 3.6% on their total sample and there is some risk of under measuring minor party support.

Horizon weights on 2008 party vote to ensure any over-response from parties’ 2008 supporters is weighted for, in order to provide a representative sample of the 18+ eligible voting population. For example, an over representation of voters for three parties in the current sample has been corrected by weighting.

Other published polls do not ask undecided voters which party they prefer and put those with a preference into a “decided/ undecided with a preference” group. Horizon then further filters this group by voter enrolment (essential to determine of those with a view can actually express it at the ballot box) and whether or not they will vote, including options ranging from definitely to may and probably will vote to definitely will not.

Horizon does not criticise others’ polling or methodologies. Horizon also conducts telephone, mail and other forms of research.

Some allege the HorizonPoll national panel is self-selected.

Horizon’s second largest on-going investment is in recruiting people to the HorizonPoll panel by personal e-mail invitation. The invitations are sent to people selected to match the 18+ population at the 2006 census. 95% of Horizon’s HorizonPoll adult population panel is recruited in this way. 5% have self-selected by joining via the www.horizonpoll.co.nz web site. Horizon has procedures, some of them industry standard, to review panel membership.

The HorizonPoll can select those who will be invited to respond to a survey in different ways: An invitation can be sent to all members to complete a survey, or lesser sample sizes can be selected, depending on response rates required. The samples are then pre-weighted to represent the 18+ population. After responses are received, all are then weighted by up to six criteria at one time to again ensure the final sample represents the community being sampled.

Results for much of Horizon’s research, for large corporate and other clients, compares accurately with those for research conducted by other professional researchers, or the statistical reality within the population, where the same or very similar questions are asked.

For example, post-Budget polling accurately predicted savings intentions. Results for recent polling on production of eggs using battery caged hens exactly matches research done by another company earlier.

Specialist polling of wool producers exactly predicted the likely subscription to a share float.

Horizon’s research of the labour market research on issues like intended permanent migration overseas produces results similar to official migration statistics. Readership research results come within 1% of other monitors.

Figures for mobile phone companies and banks' market shares, for example, exactly match industry numbers and there are many other examples. Scientifically conducted online research has performed well in UK and USA elections.

NO LANDLINE TELEPHONE

Horizon Research earlier this year found 6.4% of New Zealanders 18+ do not have a landline telephone at home. This rises to 19.6% among 18 to 24 year-olds; 18.8% for those earning $100,000 to $200,000 a year and 12.5% for those earning less than $20,000 a year.

12.9% of business managers and executives no longer have a landline at home, along with 17.2% of students and 14.6% of labourers, agricultural or domestic workers. 13% of those flatting and boarding and 11.6% of one parent families have no landline at home.

HorizonPoll research finds 95.5% of its panellists have access to the internet at home, 39.3% at work, 19.1% by mobile, 7.3% at an internet café and 8.4% at a wireless hot spot.

© Scoop Media