Methods

Item generation

The Commission has included an open-ended question on the reason for giving the trust in the Public Service score in the Kiwis Count survey since late 2020. This has generated a large quantity of responses relating to trust reasons, particularly on the responsiveness and reliability of services, as well as the integrity and motivations of public servants. Transparency and openness are sometimes mentioned as factors that could improve trust. These statements were a useful starting point to generate items for measuring factors that influence trust.

In addition, an element of the OECD framework, fairness was important to include. The concept of fairness overlaps with the Public Service value of impartiality. Other values in the Public Service Act are accountability.

Overlap between the OECD framework and Public Service Act shown below, along with questions tested.  The stem for all new items was ‘Thinking about the Public Service overall, how much do you think they…’. The definition of the Public Service was listed above the questions.[1] Responses were measured on a five-point Likert scale (strongly disagree to strongly agree).

The dependent variable, trust in the Public Service, is measured with the question “Overall, to what extent do you trust the Public Service?” Response scale has five options ranging from ‘1- do not trust them at all’ to ‘5 – trust them completely’. Only the end points of the response options are labelled. All questions also have a don’t know response option that is excluded from analysis.

The full Kiwis Count questionnaire is published here.

Procedure

The Commission ran a pilot test of the survey in May 2023, through the research provider Gravitas, using the Dynata panel. Methodology and results from the survey are presented in detail in a published technical report.

The sampling procedure includes targets for key demographic variables: age, gender, ethnicity, and region. This ensures the achieved sample is representative of the population of New Zealand.

Participants

The final sample had 2006 participants who were asked the drivers of trust question, 45% of whom were males and 55% were females. People of another gender were also included in the survey, but they made up less than 1%. The ages of the participants were distributed as follows: 18% in the 16-24 years group, 20% in the 25-34 years group, 19% in the 35-44 years group, 13% in the 45-54 years group, 10% in the 55-64 years group, 12% in the 65-74 years group, and 8% in the 75+ years group.

Respondents were asked to identify which ethnic groups they belonged to from a longer list (choosing as many as applied) and these were then aggregated into the following grouped ethnicities: New Zealand European (68%), Māori (16%), Pacific (11%), Asian/Indian (13%), MELAA (Middle Eastern, Latin American, and African 3%), and Other (8%).

In terms of education level, 8% of respondents had no qualification, 29% had a school level qualification (e.g. National Certificate of Educational Achievement (NCEA) levels 1-3, school certificate, bursary), 29% had a post-school certificate or diploma that does not require a degree, including trade qualifications, 34% had a degree or postgraduate qualification.

The household income levels for respondents were aggregated for this analysis, to 28% with income of $0 to $50,000, 32% with income from $50,001 to $100,000, 32% with income over $100,000, and 8% who selected Don’t Know for their response to the question.[2]

Records with missing values in the analysis variables were dropped from the dataset to allow the comparison between subsets of the same size, leaving a final working dataset of 1806 responses.