#BELHP2014 Plenary 3, Michael Hallsworth, UK Behavioral Insights Team

[Ed. Note: On Friday, May 2 and Saturday, May 3, 2014, the Petrie-Flom Center hosted its 2014 annual conference: “Behavioral Economics, Law, and Health Policy.”  This is an installment in our series of live blog posts from the event; video will be available later in the summer on our website.]

Today’s lunchtime plenary, Applying Behavioural Insights in Theory and in Practice, was presented by Michael Hallsworth, Principal Advisor, The Behavioural Insights Team (BIT)

Michael described the BIT as a “social purpose” company, originally founded within the UK government, but separated from it (in part) in February 2014.  The company is now partly owned by government, partly owned by its employees, and partly owned by the innovation charity, Nesta.  Michael indicated that when the Team was started, there was genuine doubt about whether behavioral interventions could make a difference or whether this was just a trendy new fad.  The Team responded by implementing rigorous methods of testing, measuring, and evaluating its proposals.

What does the BIT do?  Michael explained that its goal is to incorporate empirical findings about behavior into policy making. Although it has been colloquially referred to as the “Nudge Unit” and Richard Thaler does indeed advise them, the BIT is not actually a nudge unit.  Its first question is not how to nudge but rather how to solve policy problems.  It is a fact that policy tends (and is intended to) influence behavior.  Behavioral insights can allow governments and other policymakers to enhance and assess policy options, and offer new ones.  Put another way, Michael explained that there is not a distinction between policy-making and influencing behavior, they are one and the same.

Michael also argued that we should use behavioral insights as a lens through which to see all government action.  Moreover, whenever the government has decided to act, it should do so in way that it is actually most effective; there is a moral duty to maximize effectiveness and to spend limited government resources wisely.

Michael then went on to describe seven different ways of applying behavioral analysis to show that the best option is to:

1) Stop and revise an existing action that may be unintentionally influencing behavior

As an important note, Michael pointed out that it is not the case that policymakers are only influencing people when they are actively trying to influence them.  As an example, he discussed incidence of violence and aggression in hospitals that was determined to be caused largely by frustration experienced in the hospital environment.  One effective approach was simply to improve signage.

2) Decide not to change behavior, but shape policy around it

An example in this context is trying to deal with the problem of people going to ERs for nonemergency issues.  Rather than trying to get people to change that behavior, the solution was to co-locate emergency and nonemergency care.  People could keep doing what they were doing, without the bad outcome of unnecessary and inappropriate healthcare expenses.

3) Prevent counterproductive policy from being introduced

Here, Michael provided the example of carefully evaluating how to regulate e-cigarettes.  They may have their own risks, but they should not be regulated without thought as to whether/how much they will reduce deaths from regular cigarette smoking.

4) Address flaws in policymaking process itself

Michael emphasized the importance of nudging the government itself, which is also subject to biases and other decision-making errors.

5) Influence healthcare providers

Michael discussed the problem of medical errors caused by use of paper charts. The solution was to redesign charts – instead of writing out milligrams or micrograms, providers were asked to just circle one, resulting in massive improvement with little cost and no additional burden on providers.

6) Help people find new strategies for taking personal responsibility

Here, Michael discussed the example of “mindless eating,” the phenomenon that people will eat as much as is put in front of them.  There are various solutions to this problem related to limiting plate/serving size, avoiding exposure to tempting food, etc.  It is possible to give people information on these nudges that they can use themselves as rules of thumb to adopt in their own lives to help influence their own behavior.  This is not reducing autonomy, just making it more effective.

7) Prevent harm to others

Michael discussed two examples under this heading.  The first had to do with figuring out ways to get people to stop missing medical appointments, causing waste and preventing others from taking those slots.  One solution is to send people reminder texts, but what should the content of those texts be? BIT tried four different messages: (1) if you can’t make it, please call the number on your appointment letter – 11.7% missed appts; (2) if you can’t make it, please call *this* number – 11.5% missed appts; (3) 9 out of 10 people keep their appts or cancel them – 11% missed appts; (4) your missed appointment will cost the system X – 8.3% missed appts.  This final, most effective text message option would avoid 41,000 missed appointments/year in the hospital where the texts were tested.

Another example had to do with the impact of paracetamol packaging on suicide.  Smaller packages led to 765 fewer deaths between 1998-2009.  Michael pointed out that this intervention had similarities to the Big Gulp ban in the sense that the idea there was to stop people from buying large sodas to protect their health (but still allowing them to buy as much soda as they wanted), and here was to stop people from buying large packages of medicine to protect their health (while allowing them to buy as much as they wanted). In abstract, they operate on the same principle, but the different public response demonstrates how important case by case analysis is in this context.

Michael explained that behavioral insights can be used to influence (1) no one, (2) professional groups, (3) policymakers, and (4) health care users.  BIT tests its ideas through RCTs, since you never know which influences on behavior will work, and which will have unintended effects. He concluded by emphasizing that some nudges allow for learning and autonomy, and others can be manipulative.  Again, case by case analysis is essential.

Be Sociable, Share!