Census 2018: Chief Statistician Liz MacPherson has resigned
Tuesday, 13 August 2019
Chief Statistician Liz MacPherson has resigned after a review criticised 'ultimately ineffective' leadership of Census 2018.
An independent review into the long-delayed Census 2018 identified numerous problems which led to a failure to count one-in-10 New Zealanders.
The 80-page review, released on Tuesday, said the leadership of Census 2018 was 'overly complex and ultimately ineffective', a strong plan for the national survey was deviated from, and too much focus was put on the online census.
'I want to publicly apologise for what went wrong with the 2018 Census. I am truly sorry,' MacPherson said on Tuesday.
**READ MORE:
* Census 2018: Stats NZ says full results to be released mid-2020
* One in seven failed to complete Census 2018
* Chief Statistician ordered by MPs to produce information on Census 2018
* 365 days and still counting: Census 2018 results nowhere to be seen
* Jonathan Milne: Statistics NZ boss should be nervous**
'We shouldn't have been here, we shouldn't have had to do this remedial work.'
MacPherson, chief executive Statistics New Zealand, continued to offer assurances the now exceptionally delayed results from the first digital census, held on March 6, 2018, would produce a viable data set.
There has been widespread concern the data will be irreparably compromised.
The review found Statistics New Zealand faced problems in the lead up to Census 2018, and there was an 'optimism bias' within the organisation and key decisions were not raised with executive leadership.
Cuts to the staff on the ground were too aggressive considering the uncertainty surrounding the digital census.
'Communication subsequent to census day have not been transparent about response rates and plans for treating gaps, and have left many stakeholders frustrated,' the review said.
MacPherson said she felt the failure to reach Māori particularly keenly.
'The buck stops with me,' she said. 'Put simply, these decisions meant that the census, as carried out on the ground, did not stick to the original plan … When things went wrong, we didn't fix them correctly.'
State Services Commissioner Peter Hughes said he supported MacPherson's decision, and asked her to remain in the role until Christmas.
'I want her to stay on because she is the best person to finish the remediation work.
'I expect public service chief executives to own it, fix it, and learn from it … This is not an easy thing for anyone to do, and I want to thank MacPherson for her integrity.'
Census expert Connie Graziadel, an author of the review, said: 'It would be easy to conclude the census was a failure. We do not believe this would be a fair assessment. There were many things that went well.'
Māori Council director Matthew Tukaki welcomed the chief statistician's resignation and called for a 're-run' of the census.
Calls to hold another census before the planned 2023 survey have been dismissed. Such an effort would be too risky, according to the review released Tuesday.
MacPherson's resignation was deemed a move of integrity by the Taxpayers' Union and the Public Service Association.
The first census results are expected in September, almost a year after its expected released, with full results expected in mid-2020.
STRONG DESIGN, POORLY EXECUTED
The review said Census 2018 was adequately funded, but could have required additional funding due to risks and issues that arose that presented. There were problems with IT systems and staff were displaced by the Kaikōura earthquake.
Census 2018 had a solid design but lacked testing, the review said, meaning Statistics NZ did not fully understand how the system needed to work.
The reviewers recommended reconstructing the senior leadership of census operations, reconsidering the size of the 'field force' and that increased use of Government administrative data continue.
The method for reaching Māori should be co-designed with iwi, reviewers said, and Stats NZ should consider fully installing enumerators – boots on the ground – in places like South Auckland, Northland and the East Coast.
Some elements of Census 2018 had been a success, including the use of NZ Post and an online system which captured 80 per cent of the response to the census.
HOW THE DEBACLE UNFOLDED
MacPherson in April announced Stats NZ had counted 4.7 million New Zealanders, and would release its first data set in September – 18 months after the census was held.
Stats NZ at first celebrated a successful census in March 2018, but three months later conceded there were issues with the data.
The results were to be patched with the Government's Integrated Data Infrastructure, which includes health and income records. In April 2019, Stats NZ said the records had been used to completely 'imputate' the data set.
But many of the data sets usually available after the census may be compromised. Iwi affiliation, used by iwi to understand population and relevant for Treaty settlements, was the first to be ruled out.
MacPherson continues to remain confident the population data will be robust, and adequate for use by the Electoral Commission for redrawing electoral boundaries and for District Health Board funding decisions.
Stats NZ in July published the Census 2018 collection response rates, showing 4.2 million responses were received – of 4.7 million people – with 5 per cent of the responses partial.
Of the traditional method for census collection, 83.3 per cent of the country were counted. The result was lesser for Māori, at 68.2 per cent, and Pacific peoples, at 65.1 per cent.
Te Mana Raraunga, the Māori Data Sovereignty Network, said it was an 'appalling' result.
'The very low Māori response rates mean that a significant share of Māori data in the final census data set has been pulled from alternative sources.
'While that leads to improved coverage rates, it raises important issues about equity, trust and confidence.'
The Government has budgeted an additional $6 million to fix the broken census, with another $10m allocated to begin the development of Census 2023.
* Comments are now closed on this article.