«EASY ATTRACTIVE TIMELY SOCIAL EAST Four simple ways to apply behavioural insights Owain Service, Michael Hallsworth, David Halpern, Felicity Algate, Rory ...»
Box 1.1: Pension defaults Studies in the US, Chile, Mexico, Denmark, and Sweden show that automatically enrolling individuals onto retirement plans and allowing them to opt out (rather than expecting them to opt in to existing systems) is a highly effective way of increasing pension savings — as well as being popular amongst employees.4 In October 2012 UK employers started automatically enrolling their workers into a pension. The scheme started with the largest UK employers (250 or more workers) and by 2018 will cover all employers. Initial results show that the overall participation rate rose from 61% to 83% and 400,000 more people now have a pension.5 In all of these areas and many more, there is much potential to consider how the options presented are currently configured, and whether the de facto default is in the public’s or individual’s interests. Appropriately, there has been much debate about who should set defaults, and how. The use of behavioural insights does not eliminate the need for vigorous democratic debate. Indeed, establishing public preferences and permissions is key to selecting defaults in the first place (as it was for adopting an opt out pensions default in the UK).
1.2 Reducing the ‘hassle factor’ of taking up a service When it is not possible to change the default so no action is required, policy makers can still reduce the effort required to perform an action. The main way of doing this is to reduce the costs or ‘friction’ associated with acting. Here are
a few examples:
· Sending taxpayers directly to a form, rather than a webpage that contains the form, increases response rates by four percentage points (see Box 1.2).
· University attendance amongst under-represented groups rose by eight percentage points when forms were filled in and submitted on behalf of the applicant (see Box 1.3).
Of course, if reducing hassle costs makes an action more likely, then increasing these costs has the opposite effect. This is well-known in the field of crime prevention, which often focuses on ways of increasing hassle to deter opportunistic crime (for example, locking windows and building higher fences).
But policy makers should recognise how even small increases in required effort can have major effects. A study found that deaths from paracetamol poisoning fell by 43% after new legislation required larger portions to be presented in blister packs (resulting in 765 fewer deaths between 1998 and 2009).6 It appears that the extra effort required to release each pill individually was enough to discourage the self-harm attempt altogether.
Box 1.2: Increasing response rates by changing the default web-link The Behavioural Insights Team has run a series of trials with HMRC that sought to improve tax collection rates by making it easier for individuals to pay. One of the simplest interventions involved testing the impact of directing letter recipients straight to the specific form they were required to complete, as opposed to the web page that included the form.
This change only slightly reduces the difficulty of responding. Yet we found that this simple act of making the action easier had an unexpectedly large impact. Sending individuals directly to the form increased response rates from 19% to 23%.
% Response rate to letters directing people to webpage vs form Both public and private systems can be re-designed to reduce the burden on users – and the people operating them. Redesign does not need to mean additional expense; it may be about realising how a particular aspect of the service, which can be easily removed, is actually limiting overall effectiveness.
In this sense, thinking about hassle factors can be highly cost-effective.
Box 1.3: Simplifying the university application process Process simplifications have recently been shown to help underrepresented groups go to university, in particular by making it easier for potential applicants from low-income groups to apply for financial assistance.7 This change involved providing personal assistance to individuals (parents of potential university applicants or the applicants themselves), which included automation and streamlining of parts of the process and testing this against no help and the provision of information alone (with no streamlining or automation). Applicants whose parents received the support were eight percentage points more likely to have attended university in the year following the intervention. In contrast, simply providing information did not increase attendance.
Impact of application assistance on university attendance
Many other opportunities to benefit from reducing hassle factors are yet to be fully explored, such as making it easier for consumers to switch suppliers (see Box 1.4 below).
Box 1.4: Making it easy to switch suppliers Many markets are getting more complex: there are more than 6,000,000 mobile phone contracts available, and the number of energy tariffs in the UK has risen to over 300. In April 2011, the Behavioural Insights Team and the Department for Business, Innovation and Skills (BIS) launched ‘midata’, a programme to give consumers access to their personal consumption data in portable electronic format. This programme was backed by enabling legislation in the 2012 Enterprise Act.
Midata will enable consumers to use these data to make better choices for themselves. It will also make it easier to help consumers to get better deals, such as switching energy suppliers. The example below shows an early midata prototype that enables users to access their personal consumption data on their mobile phone, which they can then use to switch energy suppliers, without having to type in lots of information about typical monthly usage and type of contract.
1.3 Simplify messages The Behavioural Insights Team has conducted dozens of trials with Government departments that examine ways of increasing response rates to forms or letters.
Making a letter from HMRC or the DVLA really easy to understand often results in a 5% or 10% increase in response rates – usually because the main request has been made clearer. For complicated services that require thousands or millions of communications every year, this kind of increase in response rates can mean large cash and time savings, including saving confused citizens the hassle of follow-up phone calls and letters.
Health and healthcare also offers good examples of the benefits of dealing with
complex information efficiently:
· Reducing the potential for hospital prescribing errors by simplifying the forms used (see Box 1.5).
· Developing and discussing a clear self-care plan with patients during hospital discharge reduced readmission rates by 30% over the following month, compared to usual discharge procedures. This approach has been shown to work in both acute care and nursing homes.8 · Using a ‘fast and frugal tree’ to reduce the number of factors that clinicians have to consider significantly increases the accuracy of predicting heart attacks, compared to more complex systems that try to account for many risk factors.9
Reducing prescription errors Prescribing errors affect an estimated 50% of admissions in hospitals using paper-based prescription charts (Lewis et al. 2009). There are concerns that such forms lead to medication errors by hindering clear communication between professionals. For example, it may be impossible to distinguish between milligrams and micrograms when written out by hand in a hurry.
A study by Imperial College London, funded by the Behavioural Insights Team, sought to reduce these errors by redesigning forms to make them clearer and simpler (King et al. 2013). As the chart below shows, the microgram/milligram problem was addressed by creating distinct options that simply had to be circled. In simulation testing, the new charts were found to significantly improve correct dose entries, supporting information, and provision of contact information. If adopted more widely, improvements like these are likely to lead to reduced medical errors and better patient outcomes for little cost.
The Team has identified five main lessons from its simplification work:
· Make sure that the key message is presented early, ideally in the first sentence or subject line;
· Keep language simple;
· Be specific about recommended actions;
· Provide a single point of contact for responses;
· Remove all information that is not absolutely necessary for performing the action.
Our work has shown that it’s especially useful to identify how a complex goal (e.g. ‘stopping smoking’) can be broken down into simpler, easier actions (e.g. ordering a quit kit). This is based on the insight that it is easier to affect change through simple steps (‘stages of change’) and that we learn by using simple ‘chunks’ of information.10 For example, the goal of eating more healthily was broken down into the simpler task of ‘eating your five a day’. Not only are such messages easier to understand, but they also appear achievable. Simple, discrete actions can then be fused together to form more complex ones, which in turn become easier to perform – just as, when learning to drive a car, the discrete actions of ‘ignition’, ‘clutch’, ‘handbrake’ become ‘start the car’.
Policy makers need to identify how the desired action can be boiled down to specific, simple steps.
2. Make it Attractive The private sector is particularly adept at making things more attractive.
Whether it is an offer in a supermarket, the way in which an online retailer tailors its advice to you based on recent purchases, or the free gift you receive with a new kitchen, we have all bought things because they were made more attractive to us.
We think that the public sector can also find ways to make its services more attractive. Some of these are relatively intuitive (such as personalising messages);
others are more sophisticated (such as reframing the way in which incentives are presented). Fundamentally, though, making an action attractive is about two main things: drawing attention to it, and making the action more appealing.
In other words:
· Attract attention · Design rewards and sanctions for maximum effect While we treat these goals separately, they often support one other: the prize in a lottery is both eye-catching and appeals to our tendency to over-weight small probabilities, for example.
2.1 Attract attention Behavioural scientists use the term ‘salience’ to describe the way in which people are more likely to respond to stimuli that are novel, simple and accessible.11 In other words, we are more likely to do something that our attention is drawn towards.
There are many ways to attract attention. One option is a more cognitive approach that finds new ways of highlighting the consequences of the behaviour, making the costs and benefits salient. But we are also attracted by many other (less direct) factors, such as the feelings and associations triggered by how an
object or idea is presented. Examples include:
· Putting a handwritten post-it note request on envelopes, which increased response rates to a survey by the Irish Revenue from 19.2% to 36.0% after 15 working days.12 · Showing the nutritional value of food through a simple colour-coded system, to draw on our instinctive responses to certain colours (e.g. red).13 · Including the image of the recipient’s car in letters to people who have not paid their vehicle tax to increase payment rates (see box below).
· Media campaigns that have an emotional impact, such as HIV awareness campaigns of the 1980s; smoke detector installation campaigns of the 1990s;
or anti-smoking campaigns of the 2000s.
Box 2.1: Attracting the attention of owners of untaxed vehicles The Behavioural Insights Team worked with the DVLA to test the efficacy of different messages upon individuals who fail to tax their vehicles. It is estimated that there are around 250,000 unlicensed vehicles in Great Britain, representing around £40m in lost revenue, so this is a significant problem.14 We tested the original DVLA letter against a new letter with simpler, harder-hitting messages (such as ‘Pay Your Tax or Lose Your [Make of Vehicle]’).
Simplifying the letter does not seem to have had an impact for this group, perhaps indicating the clarity of the original letter. What does seem to work well, though, is including an image of the vehicle within the letter, which attracts the attention of recipients and makes the idea of losing their vehicle more salient. Note that this image not only attracts attention, but also involves a strong element of personalisation (see section below), as drivers are shown an image of their own car.
DVLA relicensing rates
We think there is particular scope for attracting attention through personalisation. Governments have, for a long time, emphasised the desire to create more ‘personalised’ services.15 But only recently has the behavioural impact of tailoring the messages sent by government and others started to be quantified.
We all develop strategies for noting information that may be relevant, and our reaction to names offers a good example. As certain names (like our own) take on significance for us, our attention is drawn to them quickly and effortlessly when they occur.16 At the same time, personalised messages make it easier for the recipient to imagine the costs or benefits of a particular action — in other words, ‘what this means for me’. When a message deals with possible negative consequences, personalisation may make those consequences seem more likely by conveying that government has accurate and detailed records (and will act on them).
Although personalised messages will usually require slightly more work on the part of the sender compared to a more generic message, the increasing power of data analytics means that sophisticated segmentation is becoming cheaper and easier.
Examples of the power of personalisation encountered by the Behavioural
Insights Team include:
· Simply adding someone’s name to an otherwise generic text message increased the amount of money paid in a trial conducted with HM Courts Service (see Box 4.1).
· Using personalised emails, rather than generic messages, significantly increased the proportion of people who gave a day of their salary to a good cause, as part of a trial with a major investment bank (see Box 2.2).
· Tax letters that ensured that doctors knew that they were being focused on as a distinct group, rather than as generic taxpayers, considerably increased response rates (see Box 2.3).
Box 2.2: Giving a day’s salary The Behavioural Insights Team ran a trial with a large investment bank to test different ways of encouraging people to donate a day’s salary to charity.17 The control group received generic emails and leaflets encouraging people to participate. This approach was tested against a range of new interventions that included: offering people sweets branded with a charitable giving message; and making the email messages more personalised.
Personalisation and the use of sweets represent the two dimensions of Make it Attractive: attract attention and design rewards effectively.
They were found to be highly effective and cheap ways of increasing uptake – and showed an even greater impact when they were combined.
(The giving of sweets also incorporates a social element of ‘reciprocity’, as explained in the next section.) % Giving one day’s salary