The Ministry of Justice has released the data confirming the government's claim it met its target to reduce the number of violent crime victims by 20,000 by 2029.
The government held a press conference on Wednesday afternoon without the data to back it up - until the Ministry released it later that night.
But there are concerns around the reporting of the data, and Labour says the government has decided it has a "narrative on law and order", and it's "looking for data to fit that narrative".
The Crime and Victims survey measures the number of victims and incidents of crime for the 12 months to October.
Every quarter, the government is given estimates, also for a 12 month period.
The latest data showed in the year to February, there were 34,000 fewer victims than in the year to October 2024.
That's 28,000 fewer than the October 2023 baseline the government has used.
The reductions are primarily in Auckland and Canterbury - though this was offset by increases in Wellington and the Bay of Plenty since 2023.
It also showed that while the number of victims of non-family violence is now below the October 2023 baseline, the number of victims of family violence remains higher.
In February 2025, there was a 24 percent decrease in the number of victims of family violence compared to October 2024. But when compared to October 2023, there is a 6 percent increase.
Concerns
Victoria University senior criminology lecturer Dr Trevor Bradley said the NZCVS had limitations depending on the offence type, due to under-reporting in areas like family or sexual violence.
"They've never had an accurate picture, they arrive at best guess-timates," he said.
He said the government's announcement should be taken with a "grain of salt", and the caveat the data was volatile was a "get out of jail card" if the numbers were to go up later.
"They want to get in now, while the numbers suggest the targets are being met. Let's wait for the full cycle, to see what the much more comprehensive data tells us," he said.
Labour's justice spokesperson Ginny Andersen said the data release "still raises more questions than it provides answers" in terms of where the decrease of violent crime is coming from.
Labour's justice spokesperson Ginny Andersen says the data release "raises more questions than it provides answers". Photo: RNZ / Samuel Rillstone
She also said it seems "quite confusing to people trying to understand what's going on here".
"The Crime and Victims survey was always released annually. Now the government is asking for this to be released quarterly, even though the data is not designed to be released quarterly."
She's concerned the government has "decided it has a narrative on law and order, and it's looking for data to fit that narrative".
Andersen's main worry is whether the reduction is as a result of a reduction of police involvement in - and less reporting of - family violence callouts.
She claimed, through information from written parliamentary questions, that police have stepped back from attending family violence call outs "unless they're of a very serious nature".
When Andersen was in government as Minister of Police, she explained, the main driver of violent crime she saw was an increase in reporting of family violence.
"I would be really concerned if this government has seen a reduction of family violence reporting and is now trumpeting that as a victory.
"If we want to get on top of the problem of family violence in New Zealand, we need to be encouraging reporting, not driving it underground."
Considering the decrease in victims of family violence in the year to February compared with October 2024, Andersen said if that meant there was a reduction in reporting, "it means people who are reaching out for help aren't getting that".
The data explained
The NZCVS full report covers the October to October period each year, released the following February, going back to 2018. This covers a two-year period, because it surveys people over the course of a year, about their experiences with crime over the previous year.
The government - with its quarterly plans and its targets - has been asking for data from it to be released on a quarterly basis. These figures also cover a year, but are published each quarter by the government alongside other data for its other targets.
The latest data - for the year to February 2025 - is what the government is referring to when it says it has passed its target. Because of the timeframes, it applies to people's experiences of crime between February 2023 and February 2024, and for much of that time Labour was leading the government.
However, this data has not had the full (and more expensive and thorough) statistical analysis treatment, including weighting for population changes, and follow-up interviews. These quarterly figures are also not usually released in a full publicly available report, only fed into the government's reporting on its targets.
The Ministry of Justice has released the data in question. Photo: RNZ / Anneke Smith
The Ministry of Justice website has a note about the methodology used for the full report versus the data reported on each quarter.
"Outside the 12-month-to-October datapoints, the quarterly NZCVS release contains data collected across multiple survey periods. This means that number of respondents per region may not be proportional to the population of the region and the resulting weights may not be correctly calculated," it states.
There's another problem too.
"The 2024 quarterly estimates are currently calibrated using the 2023 annual benchmarks. If there has been substantial population change in 2024, this may result in some difference in the weights calculated. As result of these misalignments with the survey design, additional caution is required when comparing the August 2024 estimates to the October 2024 estimates."
As with many surveys, differences are considered statistically significant at the 95 percent confidence interval (meaning the data can be relied on with 95 percent confidence in its results).
A third problem also pops up because the quarterly reports use overlapping periods of data - e.g. August 2023-24 and February 2024-25.
"As such, while the movements between the October-2023 and October 2024 estimate can be compared for statistical significance, we do not recommend referring to differences as statistically significant or not when comparing the October-2024 estimate to other quarterly datapoints."
The Ministry of Justice, in response to RNZ's questions yesterday, confirmed those concerns applied to the data spoken about by the ministers.
"You are correct that the caveats on methodology apply to today's release as this is quarterly data," a spokesperson said. "On the release of the [year to February] data, we are looking to publish this soon, hopefully [Thursday]."
All this being said, the data is still useful. It does give an indication of trends, it's just not as reliable as the full reports released in February and it's best if it is not relied on - including in comparison to the full report data.
"A recent evaluation of the NZCVS found that its survey methodology and design aligned with international standards and produced robust and reliable estimates. However, with any data source, there are a number of limitations to be aware of when understanding and interpreting the data. This is particularly the case when looking at the rolling year-to-month quarterly updates because this a new application of NZCVS data."
Another question, however, is whether the data presented is an appropriate way of assessing whether crime has fallen in response to the government's actions.
The timeframes involved already put this in question, but surveys like this are typically intended to lay out broader trends over time, rather than strictly comparing one date to another - as the government is doing in this instance.
This article was first published by RNZ
ISSUE
