Reid Research.
Source reports and survey context. Accuracy ratings are not yet available.
2026-10-06 · RNZ
Fieldwork 2026-09-24 to 2026-10-01 · 1000 respondents · online
Eligible New Zealand voters. RNZ reporter’s original report syndicated by The Spinoff. Quota sampling; weighted age, gender and geography. Excludes 3% undecided/don’t know and 2.6% would not vote. All seven named parties reported; their rounded total is 98.6%. The calculator/average derives a labelled 1.4% residual for unreported parties; it is not an observed Other result.
2020-10-16 · Newshub
Fieldwork 2020-10-08 to 2020-10-15 · 1000 respondents · mixed
National party-vote survey; detailed voter eligibility not specified in retained commissioner article. Original Newshub commissioner article preserved by Stuff. 700 telephone and 300 internet-panel respondents. The article reports TOP 1.3% and Maori Party 0.6%; preserve those primary-source figures rather than swapping them to match secondary tables. Seven main party shares total 96.0%. New Conservative 1.7% and Advance NZ 0.3% are preserved separately in the source audit; the article's nine named shares total 98.0%, leaving unreported categories. Undecided treatment and the decided response base are not specified. No combined Other share or missing-party zero is manufactured. Denominator is marked unknown and the record is excluded from model training.
2017-09-21 · Newshub
Fieldwork 2017-09-13 to 2017-09-20 · 1000 respondents · mixed
National party-vote survey; detailed voter eligibility not specified in retained commissioner article. Original Newshub commissioner article preserved by Stuff. 750 telephone and 250 internet-panel respondents. Seven reported party shares total 99.2%; minor-party detail and respondent-base/undecided treatment are not supplied. No zero or residual is assigned to missing categories. Denominator is marked unknown, so the record is excluded from current estimates and model training.
2014-09-17 · 3 News
Fieldwork 2014-09-09 to 2014-09-15 · 1000 respondents · unknown
Voters; further population definition not given in accessible broadcaster report. Factual transcription of the original commissioning broadcaster report archived by the Internet Archive. Original commissioning broadcaster report gives dates, sample and party shares. Conservative share is stated in the article body, though omitted in its Party standings list. Mode, undecided and the party-share denominator are not defined; left unknown. No Other residual inferred. Capture is from 2015; the article explicitly retains its September 17, 2014 publication date. No revised party figures are claimed. Retained for inspection; excluded from polling aggregates and Bayesian inputs pending denominator verification.
2014-09-10 · 3 News
Fieldwork 2014-09-02 to 2014-09-08 · respondents · unknown
Voters; further population definition not given in accessible broadcaster report. Factual transcription of the original commissioning broadcaster report archived by the Internet Archive. Fieldwork dates are explicitly given for the preceding poll in the original September 17 broadcaster report. Sample size, mode, undecided and denominator are not stated in the accessible September 10 report; left unknown. No Other residual inferred. Retained for inspection; excluded from polling aggregates and Bayesian inputs pending denominator verification.
2011-05-29 · 3News
Fieldwork 2011-05-20 to 2011-05-26 · 1000 respondents · phone
Voter-eligible New Zealanders aged 18+, nationally including rural, secondary urban and main urban areas. Original pollster report: cover embargo 29 May 2011, methodology page 3, complete May column on page 4. Telephone random-number generation with age, sex and geography quotas. Published decided-party base is 871 of 1,000; 11.1% undecided and 1.8% would not vote are preserved separately. Rounded party results total 99.8%; modelling normalises only the decided vector. Publication on 29 May is corroborated by the contemporaneous Curia summary. For backtests, use the conservative 9 June availability boundary: a dated academic document links this exact original PDF. Earlier table columns are not backdated as new poll releases. Progressive is displayed as a dash and is not assigned an invented separate share.