Government must tackle bias in decision making

The Behavioural Insights Team made its name using behavioural economics and social psychology to understand how citizens make decisions to help government make smarter policies. In their new report, written in collaboration with the Institute for Government, they turn their lens onto policy makers themselves.

Like everyone else, officials and politicians have a range of biases. They look for information that confirms their existing views. A study of Danish politicians found they were good at interpreting data on ‘School A’ and ‘School B’ but bad at interpreting the same data when the schools were labelled ‘public’ and ‘private’. Their own views shaped what they saw.

The report points to ‘groupthink’ as another form of bias. People who speak first in meetings can sway group opinion, which enables bad ideas to mushroom. Views often become entrenched when decision-makers meet with people outside their group, for example, from another department.

Politicians and officials are also over-confident about implementation – referred to as ‘optimism bias’. They over-estimate ‘their own abilities, the quality of their plans, and the likelihood of future success’. They think they know more than they do, and troublingly a study found this gets worse the more senior they become. Universal Credit is a good example of this. Ministers and officials were hugely over-confident about implementing major IT changes on time and within budget. It is a frequent problem across major projects.  

What should government do to prevent bias creeping in?

The report puts forward several approaches. Departments should use ‘collaborative red-teaming’, a tactic adapted from the military, which involves tasking groups of officials to find weaknesses in a proposal. Government has started to do this – the Infrastructure and Projects Authority (IPA) often reviews major projects in this way; and the Commercial Function has started using this method to stress test bids – but it should be used more. The report also suggests conducting ‘pre-mortems’ on projects to explore doubts and counter optimism bias. Such an exercise was clearly missing from the development of Universal Credit, whereas it was a major feature of the successful delivery of the Olympics.

But at the launch event for the report, panellists said the bigger question is what these insights imply for ‘institutions, structures and processes’. How can departments be set up to prevent bias in the first place? Based on our work on evidence and policy making, here are three areas government should think about:

1. Transparency - It is impossible to assess bias when policy is made behind closed doors. Departments must publish the evidence base behind policy decisions – as we have argued as part of our work on an Evidence Transparency Framework. Too many still do this poorly. But Government should go further and open up the policy making process, exposing its thinking as early as possible. This has been a key aim since 2012 but there has been little progress. As the Institute has argued, if ‘open policy making’ is to work ministers and top officials need to put much more faith in it.
2. Accountability – It is often unclear which senior official in the department is responsible for the quality of evidence used to inform decisions. This makes it easier for bias to slip through. Heads of Policy and Chief Scientific Advisers should oversee how evidence is used and report on it annually to the Permanent Secretary.
3. Challenge in the system – Decisions made based on poor (or misinterpreted) evidence often go unchallenged. This is partly about bias, but also to do with incentives. Challenging those above you – political or official – is not always the fastest way to get promoted. Every department must ensure they have independent voices in place who can challenge their decision-making to ensure it is rigorous. In some areas this can be done by another government body, as the Infrastructure and Projects Authority does for major projects. But it needs to happen across all policy making. Externally appointed Chief Scientific Advisers should be among those playing this role, but they often lack influence. Independent expert bodies can also help test officials, but these are not used consistently across Whitehall.

Tackling bias is not easy. Government makes complicated, political decisions and biases are hard to spot in real time. They are even harder to weed out. But recognising the problem is the first step. Departments must now find ways to stop bias creeping in to begin with.