A new ‘predictive modelling’ system being used by welfare officials to identify young people on the road to dependency fails to meet basic principles for dealing with youth, Simon Denny writes.
Newsroom published “For all the Jaydens out there” written by a deputy chief executive at the Ministry of Social Development. The article described how predictive analytics can be used to target youth at high risk of remaining on the benefit. In the article Nic Blakeley describes a hypothetical 16 year old, “Jayden”, who dropped out of school after being suspended multiple times. In the scenario, predictive modelling identified Jayden as high-risk and provided Jayden with a ‘youth coach’ who then helped Jayden get in to further training in construction.
While the hypothetical scenario paints a rosy picture on the use of predictive modelling, Mr Blakeley shows a naivety and lack of understanding of some the basic principles in working effectively with so-called ‘high-risk’ youth. To illustrate some of the problems around this approach, imagine the likely conversation between the youth coach and Jayden’s mum:
Jayden’s mum: Hello
Youth Coach: Hi, my name is Nic and I work for MSD. Your son Jayden has been identified as being high risk for ending up on the benefit for some time and not getting into work.
Jayden’s mum: What? What do you mean? You mean a ‘dole bludger’?
Youth Coach: Well, yes. MSD uses an algorithm to identity young people who are unlikely to get off the dole and in to work or training
Jayden’s mum: But that’s not fair, there are heaps of young people around here who are on the benefit. Why target Jayden?
Youth Coach: Well he has been identified as high-risk.
Jayden’s mum: How?
Youth Coach: Well we don’t really know as the computer does it. But it used information about you and your family to identify Jayden.
Jayden’s mum: What sort of information?
Youth Coach: Oh, things like who else in the family is on the benefit, and if Oranga Tamariki have been contacted because of concerns about you not looking after your children.
Jayden’s mum: Are you accusing me of abusing my children?
Youth Coach: No, no. I am sorry. I don’t mean to say you are a child abuser but this is just one of the variables the computer uses, with other information. I don’t know why the computer selected Jayden, but it used information about your family to identity Jayden as at high risk on not getting off the benefit.
Jayden’s mum: So it is because we are on the benefit?
Youth Coach: I guess so.
Jayden’s mum: Well I guess it is nice that someone wants to help Jayden. But why not help all the other young people on this street who want to get a job?
Youth Coach: Well this is cost-cutting. We can’t afford to help all school leavers who don’t go into further training or work, so we just target those at high risk.
Jayden’s mum: That’s not fair. I know a number of Jayden’s friends who would appreciate help getting a job.
Youth Coach: I wish we could help all of them.
Jayden’s mum: What sort of jobs are available for Jayden
Youth Coach: Not much, to be honest. And the jobs out there are hard work with low pay and few prospects for advancement.
Jayden’s mum: So a computer used information about me and my family to identify Jayden as a future dole bludger? Did you ask my permission to use my information like this?
Youth Coach: Ah, no we didn’t.
Jayden’s mum: Did you know that Jayden dropped out of school to help me with his grandfather, because he is unwell following a stroke and I needed help at home?
Youth Coach: No I didn’t
Jayden’s mum: Did you know that the reason Jayden was getting suspended at school, was due to bullying from other students and Jayden sticking up for himself?
Youth Coach: No I didn’t
Jayden’s mum: Did you know that Jayden has a learning difficulty that led to him acting out at school when he felt bored or like he was getting nowhere?
Youth Coach: No I didn’t
Jayden’s mum: Well why don’t you go tell your fancy computer that life is not all about benefits and dole bludgers. Good bye.
What is unfortunate about the MSD use of predictive modelling in the context of young people is that is contradicts government ministries’ guidelines for working effectively with young people. The Ministry of Youth Development published ‘The Youth Development Strategy Aotearoa’, which outlines the principles of working effectively with young people. One is a consistent, strengths-based approach. One of the problems with predictive modelling is that it is based on the information held by MSD which is deficit based and only a very small aspect of most families’ lives. I would encourage MSD to consult more widely with the youth sector and their own officials before embarking down a path of prediction modelling to identify high-risk youth.
*An earlier version of this article had the youth coach saying Jayden was at risk of “remaining” on the benefit and that “ethnicity” of the family was taken into account. MSD advises that the predictive modelling applies to those under 18, before they could go on a benefit – and ethnicity is not considered within the model.
Dr Simon Denny is an Associate Professor and Paediatrician working at the Centre for Youth Health in South Auckland and University of Auckland. He has published extensively on youth health issues in New Zealand and was past chairs of the Adolescent Health Research Group and Society of Youth Health Professionals Aotearoa New Zealand.