Professionals beware – your decisions are noisy!

Professional judgements are subject to 'noise' and it is louder than you think!

To Err is human. It has been understood for some time that systematic errors in human thinking, called ‘cognitive biases’ (by Cass Sunstein and Richard Thaler in their highly influential book ‘Nudge’), play an important part in mistakes made in professional judgements. To some extent introducing diversity into the decision making process can counter these cognitive bias effects, although this brings with it the challenge of reaching consensus. In itself a process that is far from perfect.

What is less well known is that disagreement between people who are called on to make professional judgements create organisational ‘noise’. This is what Danial Kahneman and his colleagues who studied this area, defines as the variability of judgements that should be identical. The result of this noise can be highly damaging and have far reaching affects on many people and, it turns out, there is a lot more of it around than we first think!

Studies of underwriters in the insurance industry have shown that the expected variation of their judgements for exactly the same claim was expected to be of the order of 10%. The actual variation was 55%. Similarly a study of US judiciary system found a variation of c50% between the sentencing decisions of federal judges for the same case. These are staggering differences and the impact on the people affected can be huge!

What is interesting is that when you think of a single court case, or insurance claim, you can identify cognitive bias but you will never identify ‘noise’. This requires an organisational study at scale.

The Illusion of Consensus

The reality is that we all view the world through our own eyes and the assumptions, perceptions and emotions we experience are unique to us. It is clearly unreasonable to expect everyone to think and feel the same, yet this is often the unspoken assumption that underpins much of the decision making processes that exist in all organisations.

Why is this ‘noise’ issue invisible in most organisations? Part of the reason is that organisations are designed to create the illusion of consensus. People talk things through using a process that encourages agreement and side-steps any heart-felt disagreement. In addition, many people are ‘conflict averse’, they want to be seen as a positive team player and as a result will go to extraordinary lengths to avoid disagreement. Especially in situations where there is a risk of embarrassment in front of peers or worse, a risk of embarrassing their boss.

Those characters who are blessed with personalities that do not conform to this norm, are often branded as ‘difficult’. They find they are not invited to many of the meetings where paradoxically their opposing views can add the most value!

Another reason that ‘noise’ persists across many organisations is the total lack of feedback. The professionals who make judgement calls rarely get to see, or hear about the impact of their decision. As a result a lot of professional judgements have no real connection with reality. For example, scientists working in academia expect their work to be peer assessed before publishing. In contrast very few scientists working in research and development departments in the corporate world expect the same treatment. Interesting?

We need to realise that wherever there is a subjective judgement to be made, there tends to be a lot of ‘noise’. The question for all leaders is how much noise should we tolerate?

So what and now what?

Like all embedded change challenges understanding the issue exists at all is a good starting point. Heightening leaders sensitivity to the reality of noise may not be welcome, but is a necessary first step. Only when the problem is diagnosed can the remedy begin.

AI has a role to play when the judgements are simple and factual, and sufficient data exists to program and train algorithms to make decisions. These decisions will in future become automated tasks. For example scanning an X-ray and diagnosing a cancer is best left to AI, it is proven to be more accurate than a human in spotting cancer cells.

This leaves a large amount of judgements which cannot be automated, or we do not want to automate for ethical reasons. In such cases options are the key.

I have long held the view that there is nothing as dangerous as a single idea, firmly held by a senior person. The chances are this idea will progress unchallenged so we need to have a process that counters this risk.

Leadership Disciplines

Leaders need to explore ‘noise prevention’ tactics and create new habits and disciplines in some of their decision making processes. For example:

  • Start to systematically take feedback from previous decisions and tease out insights that can inform current and future decisions.
  • Adapt a learning mind-set, rather than a performance mind-set, when faced with a decision making challenge.
  • Ask the ‘difficult’ characters who are comfortable to disagree in groups to step into the decision making process and make their views clear.
  • Develop at least 3 potential options for every key decision and agree objective criteria that clarifies what success looks like. Then rationally apply these criteria to each option surfacing the cognitive bias in the room as you go.

These tactics however come at a cost, most take time and require suppressed ego’s to work well. Professionals may not like them, as many are not used to having their judgements questioned. The adaptive leadership skills needed to reduce organisational noise will keep those of us working in the leadership development field entertained for a while yet I believe.

If you are interested in exploring this topic some more in relation to your organisation then contact me and let’s talk.