Public support for cancer prevention and care policies in Aotearoa New Zealand: methods of a national survey

This paper is intended to provide the methodological foundation for a co-ordinated series of analyses examining public support for cancer prevention and care policies in Aotearoa New Zealand. By outlining the study design, sampling approach and analytical framework, it supports transparency and consistency across the accompanying reports. The use of a standalone methods paper also enables a clear and consistent description of the study design across the series, reducing duplication and strengthening the interpretability of findings.

Cancer remains one of the leading causes of mortality and morbidity in Aotearoa New Zealand, and effective prevention, early detection, treatment and supportive care policies are essential for reducing the burden of disease.¹ Public support is increasingly recognised as one contributor to the legitimacy and political feasibility of cancer prevention and care policies in democratic settings.² Understanding public views across key domains of cancer prevention and care is therefore relevant to both policy development and implementation in Aotearoa New Zealand. This study was designed to assess public views across multiple topic areas relevant to cancer prevention and care, using a national weighted survey.

Overall, this framework provides a robust and practical basis for examining public views on cancer-related policies in Aotearoa New Zealand and contributes to the evidence base informing population-level prevention and care strategies.

Study design and methods

A nationwide, self-administered, cross-sectional survey was conducted to assess public support for cancer-related policies in Aotearoa New Zealand. The survey used structured, predominantly closed-ended questions (Appendix A) and was delivered online.

Study population and sampling

Participants were recruited via an established online research panel provider (Dynata). Dynata’s panel comprises individuals who have opted in to participate in survey research and includes 297,000 panel members within Aotearoa New Zealand.³ Panel members are recruited through a combination of direct recruitment (e.g., online advertising and invitations to join the panel managed by Dynata) and partnerships with third-party platforms (such as loyalty or rewards programmes and other online networks), resulting in a multi-source panel. Members voluntarily enrol and provide profiling information, which is used to identify and invite eligible participants to specific surveys. Participation is incentivised, with respondents typically offered points, rewards or similar compensation for completing surveys. Panel members are typically verified and subject to quality control procedures designed to support valid and reliable responses.

Sampling for this study was conducted by inviting panel members who met predefined demographic criteria and to achieve sufficient representation of Māori and Pacific participants. Quota sampling was used to align recruitment targets for the initial sample with population benchmarks for age, gender and ethnicity based on the 2023 New Zealand Census.⁴ To ensure adequate representation of Māori and Pacific peoples within the sample, minimum target sample sizes were specified for these groups. Where these targets were not achieved during initial recruitment, additional targeted invitations were issued to panel members from these groups (augmentation), resulting in intentional oversampling.

Additional samples were combined with the initial sample to achieve an analytic sample of at least 1,500 participants, with targeted recruitment of Māori and Pacific peoples to enable ethnicity-specific analyses. Minimum sample sizes of 356 participants per ethnicity group were identified, providing sufficient precision for ethnicity-specific estimates (prior to data collection and without accounting for design effects, this would allow 95% confidence intervals ±2.5% for the total population and ±5.2% for Māori and Pacific estimates in the worst-case scenario where prevalence was 50%).

Following the calculation of weights (design and poststratification), approximate design effects were calculated, giving an effective sample size of 1,161 in terms of the equivalent number of responses from a simple random sample. This allowed for a worst-case margin of error of 2.9% overall. The equivalent margin of error within prioritised ethnicities⁵ was 5.7% for Māori and 6.2% for Pacific peoples. As recruitment was conducted via an existing online panel, a conventional response rate (overall or by sub-group) could not be calculated, limiting assessment of potential non-response bias. This is consistent with other Aotearoa New Zealand panel-based surveys. Online panels are widely used for population-level assessment of policy attitudes where probability sampling is impractical.⁶

Eligibility criteria

Participants were eligible if they were aged 18 years or older, resided in Aotearoa New Zealand, had internet access and were able to read and write in English.

Recruitment and consent

The research panel provider (Dynata) managed recruitment using its standard procedures. Participants provided consent electronically prior to beginning the survey.

Data collection

Data were collected using a Qualtrics-hosted online survey. Questionnaire completion time was generally between 15 and 20 minutes. Data collection occurred over a 6-week period in November to December 2025.

The survey instrument was tested with five individuals before being submitted to the research panel provider, and modifications were made to improve question clarity and understanding. A soft launch with a target of 100 panel members was then conducted to further assess question clarity and survey flow. Feedback from this phase led to the refinement of only one question (about the current Government’s repeal of the 2022 Smokefree legislation, where we were concerned about the understanding of “repeal” in this context) prior to the full launch. The earlier collected data for this question were not used, reducing the sample size by 107 only when it is reported on.

Survey instrument

Survey variables were finalised in consultation with the Cancer Society of New Zealand and subject matter and methodological experts, and they were informed by a review of relevant literature. Topic areas included:

• Cancer prevention (tobacco, unhealthy food and sugary drinks, alcohol, ultraviolet radiation, human papillomavirus [HPV] vaccination)
• Early detection (screening)
• Treatment and supportive care
• Lobbying and political influence

Validated instruments were used where available, and all measures were self-reported. The full survey instrument is provided in Appendix A of the current paper. For each brief report, full question wording and the abbreviated versions used in tables are provided in the corresponding Appendix A of the respective papers. Appendix B for each paper contains supplementary tables.

Socio-demographic variables included age, gender, ethnicity, region, household income, parental status (including children’s age groups), personal or close-other experience of cancer, voting intentions and political leaning. Participants were able to skip any question by selecting do not want to answer and withdraw at any time. Derived variables included both total response and prioritised ethnicity⁵ (in order of Māori; Pacific; Asian; Middle Eastern, Latin American and African [MELAA]; Other; and European), age groups (18–20, 21–29, 30–39, 40–49, 50–64, 65+), income (using quintiles of household equivalised income)⁷ and children (any children aged under 18). We collected information on voting intention to enable analyses and reporting of findings across political affiliations.

Māori consultation

Māori consultation was undertaken in accordance with institutional procedures and principles consistent with Te Tiriti o Waitangi.

Data analysis

Statistical analysis was performed in Stata version 19.5,⁸ accommodating the survey design through weights as explained below. All descriptive statistics for socio-demographic variables are presented as actual frequencies, with both unweighted-sample and census-weighted percentages. Design and poststratification weighting was applied to align the achieved sample with the Aotearoa New Zealand resident population by age, gender and Māori ethnicity using 2023 Census data from Stats NZ Tatauranga Aotearoa.⁴ Weights were calculated based on age-gender-ethnicity strata as described below, with strata definitions adapted where sample sizes were small. In those cases, broader age strata were used to ensure stable weight estimates and avoid undue influence of extreme weights, consistent with recommended practice in survey weighting.

For non-Māori men and women, 5-year age groups were used from 18–19 years to 80+ years; for Māori men and women, 5-year age groups were used from 18–19 years to 65+ years; and for participants identifying as non-binary, broader age categories (18–24, 25–49 and ≥50 years) were used with Māori and non-Māori combined due to smaller counts.

Results were examined overall and by key demographic sub-groups (gender, age group, prioritised ethnicity and voting intention).

Reporting

Results presented in this paper describe the characteristics of the study sample prior to the weighting. Seven topic-specific reports then present findings across policy-specific areas, including tobacco, alcohol, unhealthy food and sugary drinks, ultraviolet radiation, lobbying and industry interference, vaccination and screening, and treatment and care, interpreted in relation to the Aotearoa New Zealand policy and practice context. Tables reporting weighted responses by age group, gender, prioritised ethnicity⁵ and political affiliation are presented in supplementary tables in Appendix B of each paper.

Ethical considerations

Ethical approval was received from the University of Otago Human Ethics Committee (25/1184). All participants provided electronic informed consent before starting the survey. Responses were anonymous, and no identifying information was asked in the survey. The research panel provider did not share any personally identifiable information with researchers. Demographic data used for analyses were collected directly within the survey. Participants were also provided with the research team’s email address should they wish to request a copy of the results.

Sample characteristics

The target population for this study was adults aged ≥18 years living in Aotearoa New Zealand; however, as the sample was drawn from an opt-in online panel, the achieved sample represents individuals willing to participate in survey research. Weighted estimates are presented to approximate the demographic composition of the Aotearoa New Zealand population in terms of selected variables.

The achieved sample included participants from all regions of Aotearoa New Zealand, across the adult age range and spanning the political spectrum, supporting the relevance of the findings for national policy discussions. Limited descriptive results are presented here to characterise the achieved sample prior to weighting; substantive findings are reported in the accompanying brief reports.

In total 1,635 New Zealanders participated in the survey, although one indicated that they were not eligible to vote in an election and without further explanation. A description of unweighted participant characteristics is provided in Table 1. Participants were broadly distributed across adult age groups, with the largest proportions aged 25–34 (n=288; unweighted 17.6%) and 35–44 (287; 17.6%), and fewer respondents in the oldest age groups, including those aged 75 years and over (134; 8.2%). Slightly more participants identified as female (856; 52.4%) than male (775; 47.4%), with a small number identifying as another gender (4; 0.2%). Most respondents identified as European (1,036; 63.4%), with Māori (385; 23.5%) and Pacific peoples (368; 22.5%) also well represented using total response ethnicity. When prioritised ethnicity was used, European (752; 46.0%), Māori (385; 23.5%) and Pacific peoples (316; 19.3%) remained the largest groups. Over half of respondents lived in cities (927; 57.0%), with a further 31.3% residing in towns (509) and 11.6% in rural areas (189). The most common region of residence was Auckland (625; 38.2%).

Participants (n=1,634 due to the person indicating that they were not eligible to vote) reported voting intentions across a range of political parties, including Labour (685; unweighted 41.9%), National (317; 19.4%), the Green Party (143; 8.8%), NZ First (143; 8.8%), ACT (64; 3.9%), Te Pāti Māori (74; 4.5%), Opportunity (13; 0.8%) and other parties (4; 0.2%), with some respondents reporting that they were unsure, did not know who they would vote for, or were not intending to vote (191; 11.7%). Political views across the left to right spectrum were also reported, with responses spanning all categories.

Just over half of participants were in paid employment, either full-time (709; unweighted 44.1%) or part-time/casual (263; 16.4%), while 35.4% (570) reported not currently working or studying, and 4.1% (66) were engaged in study or training. One-third of respondents reported having at least one child under 18 years of age (551; 33.7%).

A majority of respondents reported that they or someone close to them had been diagnosed with cancer (1,192; unweighted 72.9%).

View Table 1.

Strengths and limitations of the methodology

This study has several methodological strengths. It uses a large national sample and applies quota sampling and weighting to align the achieved sample with key population benchmarks for age, gender and Māori ethnicity based on the 2023 New Zealand Census. The survey instrument was developed in consultation with subject matter experts and the Cancer Society of New Zealand, informed by a review of relevant literature, and incorporated validated measures where available.

Oversampling of Māori and Pacific peoples was undertaken to enable ethnicity-specific analyses with sufficient precision, consistent with equity-focussed research priorities in Aotearoa New Zealand. While greater precision and equal explanatory power for Māori and Pacific population estimates would have been desirable, the achieved sample sizes were sufficient to support meaningful ethnicity-specific descriptive analyses while maintaining adequate precision for overall population estimates.

When calculating weights to address the risk of non-representativeness in both the panel’s sampling frame and those responding to invitations to participate in our survey, we took a reproducible and transparent approach. We based our weights on Māori and non-Māori ethnicity alongside gender and age group, rather than using all six of the Level 1 ethnicity groups in Stats NZ Tatauranga Aotearoa’s classification along with other respondent characteristics. The weighted characteristics of our sample are shown in Table 1 and show consistency with 2023 Census data⁴ for those aged 18+ in terms of age and gender, but suggest some imbalances, including in terms of some ethnicities, geography and voting intentions (with the last of these differing from concurrent political polls).^9,10 Our results were planned to be stratified by ethnicity and voting intentions, as well as by age and gender, and, as shown in the accompanying brief reports, levels of support for policies remained broadly consistent when this was done, increasing our confidence that the choice of weighting approach is unlikely to affect our substantive conclusions.

Data collection was conducted using a structured, self-administered online survey with testing and refinement prior to full launch to support clarity and survey flow.

Self-reported voting intentions and political views showed considerable diversity across the sample, including uncertainty about voting intentions. As with other sensitive self-reported measures collected via non-probability online panels, these data are presented descriptively and interpreted cautiously.¹¹

The study has some general limitations across all domains. Recruitment via an existing online panel may introduce both selection and participation bias, as the sampling frame is limited to individuals with internet access and who are willing to engage in panel-based research, and our sample is limited to those who agreed to participate with this particular survey. The survey was conducted in English only, which may have limited participation and response quality among some communities, including Pacific people. Although weighting was used to align the sample with population benchmarks in terms of age, gender and Māori ethnicity, residual differences may remain within age categories, and these weights cannot achieve representation in terms of other characteristics. In particular, we based weights on Māori and non-Māori ethnicity (alongside genders and age groups) rather than using all six of the Level 1 ethnicity groups in Stats NZ Tatauranga Aotearoa’s classification. Another limitation is that we had too few in some groups, including but not limited to non-binary/other gender participants and those of MELAA ethnicity, to be able to discuss their support for policies with confidence. Future research should investigate these views, which may require purposeful sampling. Furthermore, although we do not provide margins of error for sub-group estimates, due to the smaller numbers we would expect these to be larger than 2.9%.

The survey did not include measures of individual health-related behaviours and characteristics such as smoking or alcohol use; while these may act as mediators, effect modifiers or confounders in some contexts, the primary aim was to estimate overall population-level views rather than examine differences between sub-groups.

Finally, the use of an online panel precluded calculation of response rates overall and by key sub-groups, limiting the ability to assess non-response bias. All data were self-reported and therefore subject to reporting bias, particularly for more sensitive questions such as voting intention. This is a descriptive analysis, and we have avoided making formal comparisons between groups.

Conclusion

The study draws on a large national sample recruited through an online panel, with quota sampling and targeted recruitment to ensure representation of Māori and Pacific people, followed by weighting to align with population benchmarks. These approaches support the generation of population-level insights across a wide range of policy-relevant domains.

View Appendix.