The data science activist
- 22 October, 2015 08:30
Data is now hailed as the new oil. Its utilisation can help spur the development of startups whose values have ballooned to billions of dollars, or uplift the bottom lines of established corporates even during tough times.
For James Mansell, data represents something else – using it responsibly to deliver both social and economic gains.
“I am a philosopher by training with a full-time job as data science activist,” says Mansell, smiling.
The latter is an advocacy role Mansell took on, first when he was working in government , and continues as an independent adviser through the consultancy Noos which he founded last year.
“I try and focus the government on being evidence-based and outcomes focused and I will talk to anybody who wants to push that agenda along,” he says. “It does not matter whether they are left, right or green.”
Mansell is part of a growing volunteer group taking on this advocacy, doing it from the corridors of government, through to not-for-profit organisations, and forums that reach both public agencies and enterprises.
In 2010, Mansell, as director of innovation and strategy at the Ministry of Social Development (MSD), pushed for increased use of shared data as part of the program to reform the welfare system. This included using advanced analytics for segmentation and risk profiling and using real time experiments or campaigns to measure return on investment on government funding for these programs.
Mansell completed an honours degree in philosophy from the Victoria University of Wellington. In 2011, Mansell was awarded the public sector’s Leadership Development Centre fellowship for his work with the MSD. He used this scholarship to complete an adaptive leadership course at Harvard University. He also attended an executive negotiation workshop ‘Bargaining for Advantage’ at The Wharton School of University of Pennsylvania and completed a course on strategy and leadership at the Centre for Creative Leadership in Colorado.
Why are we not systematically aggregating data to force (and support) the government to see the full picture to make better social investment decisions?
In 2013 he was appointed to the New Zealand Data Futures Forum, an independent think tank commissioned to consider how to build a safe and high trust data sharing ecosystem within New Zealand to drive value for citizens, the economy and improve government services. The forum is chaired by former Treasury Secretary and World Bank executive director John Whitehead.
Through his own consultancy, Noos, Mansell has worked with the likes of the Department of Human Services in Melbourne where he reviewed the agency’s analytics capability, work-shopped options to move towards outcomes focused in child protection.
Leaving a full-time government role over a year ago has allowed him to work on this goal with more organisations. He works with central government agencies, including the Inland Revenue Department, Ministry of Education, Treasury, MSD and New Zealand Productivity Commission. He is currently on the board of Te Pūnaha Matatini, the Centre of Research Excellence (CoRE) dedicated to the study of how to transform complex data into knowledge.
Mansell traces the root of his data science advocacy to 2004, when he joined the Child, Youth and Family Services (now part of MSD).
He was told to read some cases to familiarise himself with child protection. He ended up reading three cases.
The last case was about a child living in the West Coast who was sexually abused at age two by his stepfather. Child Youth and Family were notified and the abuser was removed. The mother was grateful. But then this young boy was abused again, this time by a cousin. Again, CYF intervened and sent the cousin packing. This happened twice more with different sexual predators.
In social work jargon, this is a severe case of ‘non-protective parenting’, he states. Then the boy himself exhibited harmful sexual behaviour when he started school. The case history stopped about then, at aged 12. There were also notes from a social worker who queried on what they could do for this boy without sufficient funding to help the family with care or to rehabilitate this kid.
“I read his case and that was enough,” says Mansell. “It was too traumatising.”
He thought how the boy had been let down by “a broken system”.
“It is not that the CYF was broken, the whole system is broken,” he states.
“There are really good people on the frontline who are trying to help,” he explains. But the way they were allocating budget, and what they were accountable for were getting in the way.
“The whole system is working against them doing a good job to some extent.”
“It is heartbreaking to see money being wasted, to see things getting in the way of good intentions. At that point, I wanted to help.”
This begun a journey to see what could be done, he says. Mansell ultimately ended up advocating the use of shared data and the use of analytics to help the social sector focus on outcomes and in particular, longer-term outcomes.
There is nothing in principle stopping the social sector itself ‘Uber-ing’ big government.
Meet Marc Smith
Advocating for increased use of shared data and analytics to drive value is not easy.
“A lot of people try and fail to get it used. It’s a big shift to practice and the operating model for government – not merely tacking on some data science. To drive change I learnt (after four years of failure to make progress) that one thing is try to ‘add heat’ to remove complacency within government and the state sector.”
When Mansell was with the MSD, he spoke at various forums locally and globally, and one of the cases he always brought up was that of Marc Smith.
Marc Smith (not his real name) is an actual person. Joining up the data from services he received from various government agencies reveals a life of physical and emotional trauma: He was found by CYF to have behavioural difficulties at three years of age and by age five was found to have been sexually abused. The cycle of physical and sexual abuse went on, and moved into youth offending and increasingly long periods away from parents in the care of the state. In spite of this, he progressed at school to finish with subjects at NCEA level 2. This is no mean feat given this background, says Mansell.
Marc started receiving the unemployment benefit at age 18.
“You can see the pathway of being abused several times. Of course, he had a bigger chance of offending.”
When he added Corrections data, he found that Smith had started offending as a young adult.
“I use this kind of example to raise challenging questions for the state sector,” says Mansell.
“People in various services like child protection services, Work and Income and Corrections were all accountable for little pieces of the puzzle, but no one was accountable for Marc Smith.”
“Marc is not alone. There are thousands of people in New Zealand who have this kind of profile,” he says.
“The top few thousand New Zealanders with a similar profile to Marc Smith have on average over $300,000 invested and still [result in] very poor social outcomes – long-term unemployment and offending. Most of this investment is late and by that time improving outcomes is very costly and of limited success.”
This was the idea of MSD CEO Brendan Boyle, he says. “It is putting a burning bridge in front of ministers and agencies. ‘Hey, look, here are 10,000 people. They cost us $5 billion dollars. They cross all of our agencies. What are we going to do about them because they are going through a pathway like Marc Smith?’”
He asks whether Marc Smith’s life would have turned out differently if only a few thousand dollars was spent earlier when he was aged eight or nine, and not when he was aged 25 and a long-term recidivist offender.
“Do we want to keep paying for prison or spend money early on to have a better outcome?
“But although traditional research promotes early intervention, governments do not do long-term investment very well. Without shared data and the use of analytics to make people accountable and provide the KPIs and tools for longer term pathways, it is hard to focus the social sector systemically on value.
“Why are we not systematically aggregating data to force (and support) the government to see the full picture to make better social investment decisions?”
He says this is not achievable unless government can track people across services. “We can also use shared data to question which services work.
“Marc Smith is my way of telling the story of those first cases I read – of that 12-year old. But by tracking educational and justice sector outcomes as an adult it also finished the story by tracking the social and fiscal cost of poor early investment.”
He says the current system actually works for some people, but people like Marc Smith can fall into the gaps. It should work for everybody, not just the ‘average’ person.
Mansell cites the Ministry of Social Development is using operational analytics and joined up data to understand who their clients are and how to achieve better outcomes earlier.
“We are starting to make progress,” he states.
“We invested heavily in a integrated client view which is where the data is joined up,” he states. Initial attempts to get funding and formal support through the policy group failed.
“I have consistently found that policy folk are slowest to get it or accept the change because most of them are at risk of being disintermediated by big data.”
Mansell credits MSD CIO David Habershon, manager data warehouse Mike O’Neil and chief data Scientist Kip Marks for taking a punt and helping with the integration of data systems that allowed him to join information from across MSD and across agencies in the social sector.
“They gave me a lifeline,” he states. “We needed to build that stuff to get ready for welfare reform and they pitched in to get it done.
“There was a lot of opposition within MSD because it is such a radically different way to look at your business. They took some risks on behalf of a good idea when they did not have to,” says Mansell. “They got behind me and we had skunk works going on, they built the data and integrated that with the analytics and bought some of the new tools which was what we need.”
He says MSD CEO Brendan Boyle also supported the initiative.
“It got even better because Brendan was a breath of fresh air in MSD and understands this kind of change deeply, so he is a great champion. It’s good to work for a CEO deeply committed to public value and not merely managing risk for ministers.
“This kind of work requires some champions to succeed,” says Mansell. He cites that Social Services Minister Paula Bennett and Finance Minister Bill English also supported the initiative, with both speaking in forums highlighting the importance of analytics at the government top table.
“I learn a lot from them. I think it is because they talk to people out in New Zealand every day. We sometimes forget they probably talk to more real New Zealanders in a year than your average data scientist or policy wonk does in a career. They stop us getting too abstract!”
Mansell says work on the project continues as analytics is continuing to be integrated into the ministry’s operating model.
“Analytics disrupts the status quo,” he explains. “It is not just a technical problem.
“We have been building predictive models for 20 years but no one used them until people were building this capability in support of a new kind of business model. That’s really when it gets used.”
The role of technology is to create a smart closed loop learning system, as part of a customer needs and outcomes centred business model, he says.
“That is very different and requires a very different channel strategy, a different kind of funding, and development of data structures.
“Yes, we were trying to build a new way of looking at data, a longitudinal view of data. But it is all about the about the government being more targeted, a bit more evidence based about where to target investment and accountable for the short and longer term outcomes. It’s about adding value through social spending.
“At the moment what people are accountable for is doing something fast or cheaply, and you are just cost shifting into the future.”
What makes sense for MSD makes sense for the social sector as a whole.
The real value of this approach is in applying it above the servicing level to make better investment decisions across health, justice, education and social development, he says.
This idea came off the back of identifying limitations in investment approach in MSD, he states. “MSD is really only optimising investment across a narrow service offering.
“Many times welfare beneficiaries need better health or education or child protection services, not better CV writing courses. So some of the case management investment in Work and Income should be moved elsewhere – re-invested. This wasn’t a popular view inside Work and Income.”
So, in 2012, he pitched the idea of a Whole of Government Analytics at the centre of government to drive structural shift in the way government investment is managed. This means building accountability and budget allocation around needs and outcomes on the basis of an independent (system level) understanding of pathways across services. This has led to the creation of an advanced analytics function at the centre of government (Treasury).
In addition, significant new funding was also received to expand the role of the ‘Integrated Data Infrastructure’ managed by Statistics New Zealand and enable remote access for a wider range of users. Organisations like Motu now have much better access to integrated case level government data to do research with, he says.
Mansell and New Zealand Data Alliance members are currently working with ministers and Treasury to apply a whole-of-sector view of the social system. This information will be used to shift to population based funding and accountability for outcomes. Elements of these are now being trialled for the 2016 Budget.
“It is about being smarter about where you target,” he says. “Everybody who uses the service gets a better service there is less wastage, everybody wins …except service providers whose services are low value.
“The goal is to make everyone accountable for outcomes so that they start to be innovative about having to solve the problem, how to avoid bad outcomes.”
Advocacy and risks
What keeps him up at night now are the risks attached to all of these. Along with the big benefits of using shared data to improve outcomes, come big risks. This is deeply personal information and needs to be handled sensitively, he states.
He is working with ministers and senior officials on both sides of the Tasman on how state agencies can safely use data to focus on outcomes and innovation. This, he says, including advocating for the right kind of national data ecosystems, is needed for responsible use of data science and data sharing.
One kind of solution is promoted through a document called Handing Back the Commons (2015). This report, commissioned by the New Zealand Productivity Commission, recommends the formation of a Social Data Commons. The data commons is collectively owned by the social sector composed of open source community members, researchers, NGOs and social service providers.
The central government will be involved, but as one voice among the many members of the sector, he explains. The paper argues that ownership and control of sector data by citizens and frontline providers, kept at arms-length from the state, will increase confidence, trust and the level of innovation in the social sector.
“I shudder to think what the centre of government will do if it has unfettered use of personal citizen data. The government cannot be both gate keeper and user of social sector data. To do so invites a monopoly and so stagnation and I think, a drift to more coercive uses of citizen data.”
Last year, he founded the New Zealand Data Alliance, a volunteer group of data scientists providing independent, non-commercial and non-partisan advice to community leaders on the responsible use of data to improve outcomes for New Zealanders. So far their work has included advising senior ministers and not-for-profits.
The real benefits of big data for Maori, for marginalised groups, for researchers, for innovators on the front line are huge.
He says a lot of the New Zealand Data Alliance’s work is unpaid, like work with NGOs who often ring Data Alliance members for advice.
“I think the focus now is to move outwards. We have probably pushed the centre of government as far as is possible.”
“We are doing that for free because we want to give them non-commercial advice and empower them with their own data. The real benefits of big data for Maori, for marginalised groups, for researchers, for innovators on the front line are huge.”
“We want to give them non- commercial advice because you often get advice from big software vendors, and they are always trying to make some money on sales. There are cheap options to get started.”
If GPs, schools and other social sector players can take control of their own data then they will have a powerful platform to challenge the hegemony at the centre of government, he says.
“There is nothing in principle stopping the social sector ‘Uber-ing’ (disintermediating) big government. The result would be a much safer, less fragmented, more innovation and a more powerful provider network.”
He admits being seen as a “lightning rod” when he was advocating all this from inside government.
“I was not popular. I did as much I could do, so I felt it was time to move on.
“The more you get involved in this stuff you start to realise several things. That ‘I don’t necessarily have all the answers’, sometimes you can get too caught up in your own mission and importance," he states.
“So much of this is about asking the right questions rather than thinking you have the right answers. This is very new and letting go means enabling other people to ask questions and innovate. I think I am learning (slowly!) that you get more back by letting go.”
He says this work is being continued by a lot of people in both government and private sectors. “The Data Alliance members are in the analytics and big data space, but a lot of them are entrepreneurs, pushing things along in their respective organisations.”
“Most of us have kids and we just want to make New Zealand a better place” he says, on what drives him in the role. “This is one way to do it.”
Profile of a man I respect ... doing great things for New Zealand. https://t.co/sFOJhNHiJM— Geoff Beynon (@BeynonNZ) October 22, 2015
Super work being done by great people. https://t.co/V9xomdDnem— Hermione (@MioneSparkle) October 22, 2015
Send news tips and comments to firstname.lastname@example.org
Follow Divina Paredes on Twitter: @divinap
Follow CIO New Zealand on Twitter:@cio_nz
Click here to read the Spring 2015 issue of CIO New Zealand