21 Jun 2015

Profiling for Child Abuse

From Insight, 8:12 am on 21 June 2015

New Zealand is leading the world with ground breaking research that uses government-held data to try and stop child abuse before it happens.

But an Insight investigation has found this form of profiling is also raising questions about some of the things people hold most dear: their children and their privacy and of unintended consequences. 

We consider the key questions below.

Father crouches smiling by push chair with young son

Richard rejects the idea that his personal information would provide an insight into how he treats his children. Photo: RNZ / Teresa Cowie

Listen to Insight: Child Abuse or Big Brother?

In 2012, as part of its crusade against child abuse and following recommendations from the White Paper for Vulnerable Children, the National government commissioned Auckland University to undertake groundbreaking research to try to find out if information it holds about citizens could be used to predict whether a children might be abused or neglected.

The initial research, done by Professor Rhema Vaithianathan, is being closely followed by international researchers. The idea of profiling, or Predictive Risk Modelling, is not new. But basing the model on information held by the government in order to predict child maltreatment is.

The latest research suggests it is fairly accurate. Of the top 1 percent of children it predicted to be at high risk, 42.2 percent were found to have have been maltreated by the age of 5.

But opponents say that's not accurate enough and families will be harmed if they are identified incorrectly.

Patrick Kelly is a paediatrician and the clinical director of Auckland Hospital's child abuse unit. He is involved with police on a daily basis trying to make collaborative decisions about what risk a child might be in.

I  am well aware of the deficiencies in risk assessment and I think that if you are going to introduce a tool like this based on population data into front line practice there's a real, serious risk of unintended consequences."

The likelihood of beneficiaries being singled out is a worry for advocate, Kay Brereton, who believes such profiling  would stigmatise beneficiaries as child abusers.

Advocate stands outside WINZ office

Kay Brereton says beneficiaries will be unfairly singled-out as child abusers. Photo: RNZ / Teresa Cowie

She said there is more data held on beneficiaries because they use more social services and that means they are more likely to be picked up by the system.

The Head of Philosophy at Auckland University, Tim Dare, did an ethics review of the research and he said it could stigmatise Maori too, but he thinks the stigma could be minimised if it's handled properly.


How would the profiling work?

Stop sign stuck to a door saying restricted access only

Secure data room sign - Access to the room where AUT researchers hold people's personal data is strictly controlled Photo: RNZ / Teresa Cowie

There isn't a set system yet because it is still being decided if or how it would be used. But in the latest research the Ministry of Social Development used 13 indicators or pieces of information about a child's family to work out what it calls a 'risk score' of 1 to10.  Those with a score of 1 were at less risk of abuse, those with a score of 10 were more likely to be harmed.

What government information would it use?

So far the researchers have used benefit data, from all the forms and information people hand over when they get support or services from Work and Income. They also look at Department of Corrections data to see if you've been in prison, medical records and Child, Youth and Family files to see if any of the parents of other children have been in care, or if the parents themselves have been in care, as well as Births, Deaths and Marriages information.

What information about me that the government holds could 'red-flag' me as a potential child abuser?

Here's the full list of indicators from the Ministry of Social Development's most recently published research:

Predictor Variables (Source: Ministry of Social Development)

Variable Variable values
Gender of child 1=male
Low birth weight or pre-term 1=yes
Parenting demands 1=high parenting demands1,
2=no other children
3=other children but not high demands
Other children with care and protection history 1= yes
2= no
Police family violence notifications/contacts 1=events in one of the last 12 months
2=events in more than one of the last 12 months
3=no events
Caregiver's age 1=under 20 years
2=aged 20-24 years
3=aged 25-29 years
4=aged 30-34 years
5=aged 35-39 years
6=aged 40 or above
Benefit caregiver is not a birth registration parent 1=yes
2=no birth registration
Single parent 1=single parent
2=single parent and no father listed on birth registration
3=not a single parent or partnership status unknown
Time on benefit in the last 5 years 1=more than 80%
2=between 20% and 80%
3=up to 20%
4=no time
Caregiver with care and protection history 1=yes
Benefit address changes in the last year 1=no address changes
2=1 or 2 address changes
3=3 or more address changes
Mental health in the last 5 years2 1=substance abuse issues
2=persistent substance abuse issues
3=mental health issues other than substance abuse
4=persistent mental health issues other than substance abuse
5=no known mental health or substance abuse issues
Behavioural or relationship difficulties as a child3 1=yes
Corrections history in the last 5 years 1=non-custodial sentence
2=custodial sentence for non-violent crimes
3=custodial sentence for violent crimes
4=no history
Child, Youth and Family site 43 Child, Youth and Family sites


1 - The child is a multiple birth child, there are three or more other children in the family, or one or more other child aged under 3 years in the family.
2 - Based on incapacity codes recorded for caregivers who had received incapacity-related benefits.
3 - Based on substantiated findings recorded for the parent or caregiver in Child Youth and Family data.

Could the government use it to take my children off me, before I've even done anything?

No, the Ministry of Social Development says it would only be used to help families with a high risk score by giving extra help and social services. It says it would be completely up to the family to decide if they want to get involved and take that extra help.

What sort of help would a family be given?

The Ministry is still trying to figure out how social workers might use it to help high risk families. It is running some trials at the moment and will be reporting back the results at the end of the year.

Would the score replace a social workers judgment of how a family is doing?

The Ministry says no, it wouldn't. It says social workers look at the whole picture and the profiling model is only limited to the data that is held. Also, people's family situations are changing all the time and social workers can get the best and most up to date understanding of any potential risks.

Research on profiling for child abuse from The Ministry of Social Development's website

Follow Insight on Twitter