Skip to main content

Noise: A flaw in human judgement (Social Work book review)

Posted: March 24 2023

Social Work student & staff projects, Social Work
Social Work student & staff projects, Social Work

Interested in Social Work and want to learn more about the subject? The book reviews written by our Social Work students and staff help you identify the best literature to advance your learning.

This week:

  • Title: Noise: A flaw in human judgement
  • Publisher: Little, Brown and Company
  • ISBN10: 0316451398
  • Author/s: Daniel Kahneman, Olivier Sibony and Cass R Sunstein
  • Originally published in 2021
  • Reviewer/s: Dalbir S. Chana, MSc Social Work student
  • Originally published in the Journal of Social Work
book cover of a flaw in human judgement

Noise: A Flaw in Human Judgment was written by Daniel Kahneman, Nobel Memorial Prize in Economic Sciences (2002) and writer of the bestselling book Thinking Fast & Slow, one of the most cited legal scholars in recent history, Cass R. Sunstein (co-author of Nudge), and Olivier Sibony, Professor of Strategy & Business Policy at HEC Paris. The book highlights the contributing factors to ‘noise’ in decision-making and proposes tools and strategies to ameliorate noise and therefore improve judgment at an individual and organisational level.

Should all decisions be the same? Clearly not. Diversity in thought and human judgement is essential to many human endeavours ranging from philosophy to literature, to science. However, there are areas where we would like consistency between decisionmakers. For example, in the diagnostic prognoses of several physicians examining the same patient, judges deciding sentencing for similar criminal cases, or social workers assessing the same service user. To illustrate this problem, let’s imagine a patient who visits the hospital complaining of chest pains; the doctor orders a few diagnostic tests and upon analysing the results, recommends immediate heart surgery. The hospital’s procedures require two other opinions before major surgery is allowed; the second physician disagrees with the severity of the risk and suggests a different procedure; the third recommends more tests before any treatment can be given. These unpredictable professional disagreements when examining the same information are what Kahneman and co, refer to as ‘noise’; or otherwise put – undesirable variability in human judgement. Readers will perhaps be familiar with the concept of bias which, from a statistical perspective, is defined as a systemic deviation from the true value in a particular direction, much like darts thrown consistently towards the right of the bullseye. Noise, using the same analogy is simply random scatter, and thus has no discernible pattern. In this case, noise should be thought of as a random deviation in outcomes of the decision-making processes, and not an underlying factor, which causes this deviation.

Noise can be thought of as the less charismatic sibling of bias, however, according to the authors, noise may account for more of the overall error in professional judgement than bias. This means that if we want to move towards better decision-making, we may be better served to address the noise than attempt to correct the biases that we may be facing – this forms the book’s primary thesis ‘wherever there is judgment, there is noise – and more of it than you think’ (p. 12).

On the one hand, it is quite natural and normal that people in their personal judgments are different. Professional judgments are informed by expert/professional knowledge, whose range and quality may be different among individuals, their overall professional experience, and often idiosyncratic factors such as personality and unique ways of filtering, combining, and processing information. However, the authors of Noise cogently assert that ‘You may believe that you are subtler, more insightful, and more nuanced than the linear caricature of your thinking. But in fact, you are mostly noisier’ (p. 120). In order to reduce noise, they decompose noise into its constituent parts and propose the concept of ‘decision hygiene’ as well as provide extensive guidance on how to structure and arrange information. In addition, the authors suggest a number of ways for improving judgment in general. These include relatively straightforward methods such as averaging multiple independent judgments, developing simple guidelines, and designing contextually appropriate decision making scales. Somewhat controversially, Kahneman and co, propose the development of formal decision algorithms for organisations to mechanically combine multiple inputs – a method, which eliminates noise. However, such models are sensitive to biases, which have been inadvertently designed into them by their human constructors, leading to unwanted and potentially unethical outcomes if the bias is not noticed and swiftly corrected.

Although this book is aimed at a popular audience, it does illuminate important issues for professional decision makers within organisations, and perhaps more so for those whose decisions have life-altering consequences for others, such as judges, doctors, and social workers. There is a strong ethical dimension to this work, that is, an implied moral imperative to reduce noise in professional judgement as a means of ensuring fairness for those people subject to the decisions. This idea is perhaps best expressed by the maxim ‘treat like cases alike’ which is foundational across professional disciplines as a normative and positive duty to avoiding arbitrariness when deciding cases. Within social work, especially in the life-changing area of safeguarding and particularly in child protection decision-making, this Aristotelian conception of justice/fairness is certainly compatible with the professional ethics espoused by professional bodies such as the British Association of Social Workers (BASW) and the International Federation of Social Work (IFSW). If there is substantial variability in social work judgments, does this not provide a moral impetus for managers and organisations to minimise this to within tolerable limits? The key point that Kahneman and colleagues make is that without measuring noise, one cannot know its extent and pervasiveness, and the variability is likely to be far greater than one might expect.