Despite the threat of being in contempt of Parliament, Government Statistician Liz MacPherson says now is not the time to release some information about the 2018 Census.
MacPherson has twice refused to provide Parliament's governance and administration committee with the number of partially and fully completed responses to the 2018 Census.
The National Party's statistics spokesman Jian Yang called it an "extraordinary stand-off" but MacPherson said that without context, the numbers would be open to misinterpretation.
Yang told the Weekend Collective that MacPherson has refused twice to answer the committee's question twice.
"It is important for us to know some basic information about Census 2018, as one of the key responsibilities of the committee is to ensure our government agencies are working properly."
He says that the Census was botched and "is a shambles", and they need to know how bad it was. Yang says the participation rate was the lowest ever, with around 90 per cent of people taking part. "Over 400,000 people were not counted in the Census."
MacPherson has instead intended announcing at the end of April when Statistics New Zealand would release the first set of results for the Census.
"It is my hope that the committee will appreciate that I have made this determination after careful thought and application of statistical best practice," MacPherson said in a statement.
According to Stats NZ's interim calculations, around 10 per cent of New Zealanders didn't fill out their Census forms, resulting in the lowest participation of the past five surveys.
The calculations, released last year, showed that full or partial information for only 90 per cent of individuals was received, compared with 94.5 per cent for the 2013 Census.
The low participation has caused a delay in the release of information collected, which was due to be released in October, then March, now on the date yet to be announced.
Stats NZ will also have to use other sources of information and different imputation approaches to fill in the missing data.
The 2018 Census was the first time Stats NZ used a digital-first approach, which encouraged respondents to complete the census online.
MacPherson had told the committee on February 13 that the number was available and would be provided but then declined to provide it, saying it needed more context.
She again declined to provide the number when she appeared before the committee last week.
"The committee has taken the extraordinary step on invoking Standing Orders and requiring an answer," Yang said.
"Parliament is entitled to know the size of the statistical hole from those who did not participate, as well as those who did not complete Census 2018. The refusal to provide this information is inconsistent with the Government's pledge to be the most open and transparent ever."
MacPherson said her announcement at the end of April would outline what would be included in the first release and provide context.
"In light of the select committee's request, I will be providing information on the breakdown of census forms that were full and partial responses.
"Stats NZ is an open and transparent organisation. When we produce information for the public we outline the methodology and limitations of any data produced," she said.
The select committee's annual review of Statistics NZ, released last week, said it had sought from MacPherson information about the proportion of Census forms that were fully completed.
"We believe this is important data around the performance of the Census, and that if a large portion were unanswered it would raise concerns about the quality of the Census output, and also about the value for money in the overall expenditure."
The committee had formally required her to respond with an answer under Standing Order 227 (2). That Standing Order requires a witness to answer a question.
National MP Nick Smith said if the information was not provided by April 10 he would seek to refer MacPherson to the privileges committee for contempt of Parliament.
The Government allocated $113 million to the 2018 Census, compared with $72m for the 2013 Census.