The New South Wales government was an early adopter of behavioural science, and its Behavioural Insights Unit has now started to consider applying the popular approach to bigger, more complex public policy challenges.
The BIU has more than doubled in size since 2012, when a tiny team of four started looking at just a few areas such as health. Now it has about a dozen staff who are setting their sights higher and that’s “partly because of the successes of the first round of projects,” director Alex King tells The Mandarin.
“We’re moving into new topic areas, which is really exciting for us,” he says. “Originally the team was mostly focused on the kind of one-off transactional behaviours, things like paying a tax [bill] or a fine or attending a hospital appointment.”
“We’ve actually started to move into the more complex and ongoing behaviours like obesity, returning people to work, domestic violence — that kind of area which is something that involves multiple behaviours over a long period of time, which is a lot harder to shift. But if you can shift it … it can be a really powerful output.”“Whatever we do, we’re trying to encourage people to do one thing or the other but at the end of the day, you can always ignore it.”
The NSW nudge unit’s recent two-year update report, which came out just before Christmas, shows it is consistently getting results from basic approaches like tweaking letters or text messages in a range of areas.
Building on work done in the United Kingdom, for example, the NSW BIU showed new text message reminders reduced no-shows for St Vincent’s Hospital appointments by just under 20% compared to the hospital’s standard reminder message, by telling the patient something about the financial impact.
The two best texts framed the information as an “avoided loss” scenario: “By attending the hospital will not lose the $125 that we lose when a patient does not turn up.” But it pays to be concise. Even just nine extra words — “this money will be used to treat other patients” — reduced the effect slightly.
King says the effect was about the size the team had hoped for, based on the UK example.
New letters reminding women to book pap tests could encourage 7500 more to attend screenings every year, an increase of about 2.5%, according to another randomised controlled trial. That result is “pretty good for the relatively low cost of changing the letter” in King’s view, and he says agencies can often achieve “a bump of a couple of per cent” at least, in more or less the same way.
“We hadn’t done anything directly in that space [before] but the kinds of messages, again, is consistent with other kinds of letters where we’re looking at a one-off behaviour, especially a preventative measure like getting a cancer screening appointment,” he says.
The report explains the value of a nudge such as this is linked to the scale at which it can be applied:
“One of BIU’s primary criteria for selecting projects is the ability for our partner agency to scale the evidence generated from the trial, which can vary depending on the nature of the intervention.
“In this case, the Cancer Institute NSW’s Cervical Screening Program already had a state-wide information dissemination channel set up, which enabled them to rapidly scale up the findings generated from the trial to all reminder letters.”
Advisory role and collaboration
King’s team also provides “upstream policy advice” across government, including in situations where agencies can’t easily run a randomised controlled trial — although the BIU prefers to do so wherever possible, since behaviour depends on a lot more than just one factor such as the wording of a message.
“There’s a lot of factors that can pull you one way or the other and it’s quite hard to know upfront which ones are most effective, so we … prefer to try a couple out, design a couple of things and then put them through a proper trial process to try and understand whether it’s having that benefit,” says King.
“It’s not always possible; sometimes there are various practical barriers [that make it] technically challenging to do that, so we just provide some advisory work where we can.”
Other NSW agencies are welcome to contact the BIU for simple advice, perhaps on a one-off initiative or new policy, and some regularly do, says King.
“There’s a lot of standard learnings and principles and tools that will work in many contexts,” he says.
“Not always in every context, but there’s often, in almost any challenge where there’s a clearly defined [goal like] ‘we’re trying to change this behaviour’, we can normally give some thoughts or some suggestions about things that worked in similar contexts, by analogy and that kind of thing.”
The DPC BIU isn’t the only one nudging, as other agencies have been building the skill set into their own capabilities all along, and nor is it the only nudge unit in NSW. The Department of Family and Community Services also set one up in late 2014 and it is looking at some complex topics, the report explains:
“The focus of the FACS Unit is on major BI projects, including in the areas of child protection and social housing, as well as building the capability of FACS staff to apply BI principles in their work.
“DPC and FACS BIU meet regularly and exchange ideas. In 2016, DPC BIU supported FACS BIU to implement their first randomised control trial, which launched in July. Results of the trial are due by the end of 2016 and are expected to inform operational reforms within FACS, demonstrating evidence-based service design within NSW Government.”
A FACS spokesperson told The Mandarin the trial is “looking at how behavioural insights can be used in feedback letters to mandatory reporters to better respond to children and young people at risk of harm, as early as possible. Specifically, for reports to the Child Protection Helpline that do not meet the significant risk of harm threshold.”
The department now expects to get the results midway through this year, and they will feed into its ChildStory project, although it isn’t clear if they will be made public.
The BIU also runs “master classes” and has built a community of practice where behavioural insights can be shared within the NSW public service. King says some agencies like the Office of State Revenue are working with the BIU less, but continuing to use behavioural science as much as ever as they build their own capability.
Transparency, choice and sustainability
As well as offering a behavioural science perspective alongside other public servants advising government on bigger, more complex policy challenges, King says he would like to do more studies checking how well the first generation of nudges are holding their effects over time.
“It’s not atypical to lose about half of the impact,” he says, based on some work done by other teams, “but it’s rare that you lose all the value of that.”
There’s growing interest among public sector behavioural science squads in the issue of how much nudges lose their punch over time. There’s also an inclination within the burgeoning international field to keep it highly transparent, evidence-based, and true to its roots in academia.
“Whatever we do, we’re trying to encourage people to do one thing or the other but at the end of the day, you can always ignore it,” King explains. “It’s not something that you have no choice about.”
Choice is what it’s all about. And for the moment, the field of nudging is relatively unified around the world.
“There’s teams springing up all over the place and the great thing about that is by sharing our results [with] other teams that generally follow the same model, we’re able to learn about a lot of these great theories,” King says, explaining that each society has different factors that influence the effectiveness of nudges.
“The great thing, I think, about there being a lot more people in this space is that ability to try many more different things that we can learn from. Certainly a number of our trials started off as having heard about an interesting result overseas, and wanting to build on it and extend it.
“And I’m sure people have done the same from our work — taken it and tried to take it further — and there’s a great body of knowledge being developed about what does work and when does it work and why.”