BX 2017

19 July 2017BIU Team



For behavioural insights (BI) fans, like us, the Behavioural Exchange conference is an annual highlight. It is two full days of talking about BI, seeing what is happening around the world in BI and sharing our experiences from NSW.


We recently attended, and presented at, BX2017 in Singapore. Below are some of the highlights from the conference.

Highlight 1 - Dean Karlan

Dean gave an engaging presentation about how commitment contracts can be used to bridge the intention-action gap. Dean is the co-founder of stickK – a website where users can set a goal and place a value on sticking to it. It works by having you sign a binding contract that requires you to commit to a penalty for not achieving that goal – for example, donating $500 to an “anti-charity” such as a political party you despise. As part of the process, you also nominate a referee who has the power to hold you accountable. Dean explained that the concept is based on making failure expensive.

Dean talked about a trial he ran in the Philippines where commitment contracts were used to encourage people to stop smoking. Smokers were offered a savings account in which they deposited funds for six months, after which they take a urine test for nicotine. If they succeeded their money would be returned; otherwise, their money went to a local orphanage. The study found that 30 per cent more people quit smoking if they signed the contract, than those that did not have a contract. The effect persisted in surprise tests 12 months later, indicating that the effect was enduring. 

Highlight 2 – Rayid Ghani

Rayid, a computer scientist at the University of Chicago, uses data and analytics to solve large-scale and high-impact social problems. Rayid’s work applies machine learning to problems to improve outcomes in situations where traditional policy work has been less successful. Some of Rayid’s achievements include:

  • working with police departments to build early intervention systems that can help identify officers (and dispatches) at risk of adverse interactions with the public
  • predicting which children are likely to get lead poisoning and then targeting proactive inspections for lead hazards to their houses
  • working with school districts to identify students in need of extra support to achieve different educational outcomes.

Rayid emphasised the importance of data analytics teams and behavioural science teams working closely together. He believes that working together would achieve better outcomes: the behavioural science teams can develop the interventions, while the data analytics teams can prioritize who should be targeted by a particular intervention.

Highlight 3 – How health is using BI

In a panel dedicated to BI interventions in health, we heard how small BI interventions are leading to large improvements in people’s health and wellbeing. Associate Professor Jason Doctor from the University of Southern California presented his research on nudging doctors to prescribe fewer antibiotics. His research showed that small, timely reminders and posters used as commitment devices could change how doctors prescribe antibiotics.

Zee Yoong Kang, CEO of Singapore’s Health Promotion Board, presented the National Step Challenge. The challenge encouraged Singaporeans to walk more by providing very small financial incentives. Over 200,000 Singaporeans signed up for the challenge and moved more. Lastly, BIU’s own Alex King presented several of our health trials (PDF, 507 KB). We have used BI techniques to change simple one-off behaviours like increasing attendance at hospital appointments to more complex behavioural issues like childhood obesity.

Highlight 4 – Colin Camerer

Colin Camerer, Robert Kirby Professor of Behavioural Finance and Economics at the California Institute of Technology, shared his latest research on the biological underpinnings of human behaviour. Colin is researching neurological signposts that seem to appear when animals learn habits. He presented results from research that looked at how rats’ brains lights up when moving habitually through a maze (they use MRI – magnetic resonance images – to see how the rats’ brains light up). Colin found that when the rats undertake automatic behaviours (activities or behaviours that animals including people do without conscious self-control such as driving to work or yawning when others yawn), only a very small part of their brains lights up. Given the biological similarities between rat and human brains, this has very interesting extrapolations for how much we actually use our brains every day.

Highlight 5 – Games

While the conference was packed with incredible speakers, we also had the opportunity to participate in a game to better understand bandwidth tax. The game, O$P$ (owe money, pay money), was developed by the Civil Service College Singapore to teach government employees what it is like to live on a low income. The game simulates real life experiences where players have to choose how they to manage their meagre finances, while dealing with life events, such as ill health and family issues that have a financial cost. In a really short time you gain an insight into the complexities, vulnerability and stress of living on a low income. As people leave the Civil Service College and go back to their roles in government, they take these insights with them and incorporate these experiences into their work.

Highlight 6 – Lessons learnt

Many of the lessons learnt highlighted by the speakers mirrored what we have learnt here in NSW: the importance of initial strong senior support; concerns about the effectiveness of BI, and unwillingness to try new things, can be overcome by focusing on quick wins such as letters trials; and, the importance of involving operational staff to help you design for workability and scale from the beginning.

Another lesson was that it is often better to break large trials into multiple stages, for example by having a ‘winner stays on’ model rather than a single multi-arm trial, or by testing interventions at small scale or in the lab before running a full trial. Something we have been doing here in NSW over the past year or two.

What’s next for BI?

What will the future frontiers of BI be? The speakers talked about new frontiers for BI globally, with four main themes:

  1. The increasing scope to apply BI to upstream policy advice – for example in BIT’s involvement in the sugar tax debate in the UK.
  2. Applying BI to organisational behaviours and structures rather than to individual citizens – for example, our own trials around returning injured workers to work attempted to bring in a culture of growth mindsets.
  3. BI moving into increasingly complex areas – our work in domestic violence or obesity for instance.
  4. Integrating BI with other tools such as machine learning, data science, design thinking and other approaches. We’ve not done much in this domain but we’re on the lookout for good opportunities to try them out!

With BX2017 over, we are already looking forward to BX2018 next year - which will be here in Australia, hosted by the Australian Government’s BETA team!