Abstract
Some judgments are biased; they are systematically off-target. Other judgments are noisy, as people who are expected to agree, end up at very different points around the target.
—Noise (p. 2)
Whenever there is a judgment, there is a noise and more of it than you think.
—Daniel Kahneman
Noise is authored by Daniel Kahneman, a famous psychologist and 2002 recipient of the Nobel Prize in Economic Science. Kahneman’s work in cognitive biases and the art of decision-making in the face of ambiguity led him to develop the prospect theory, which deals with decisions under uncertainty. Kahneman and co-authors Oliver Sibony and Cass R. Sunstein examine noise as error that might influence our decisions. The book Noise is about errors in human judgment, which the authors have separated into two categories: bias and noise. Bias refers to systematic deviations from ideal human behaviours that are predictable, whereas noise refers to errors that are random/chaotic deviations from desirable human behaviour that are unpredictable. To distinguish between bias and noise, the authors employed a bull’s-eye metaphor. Bias occurs when all shots are systematically off-target in the same direction; noise occurs when shots are all over the place. To comprehend the error in judgment, one must comprehend both bias and noise. This will aid readers in recognizing the many fallacies that might cloud their judgement and provide strategies for overcoming them. On the one hand, noise can be a more frequent and serious issue at times, but on the other hand, it is rarely acknowledged in public discussions on human error and in organizations around the world. The reason is, because of its unsystematic nature, noise is difficult to identify and deal with. That is why practically all organizations fail to recognize and deal with it. The star of the show is bias. Noise is a minor character that usually appears offstage. Thousands of scholarly studies and hundreds of popular books have been written on bias, but only handfuls have addressed the noise issue. The authors in addition to identifying different causes of noise and providing strategies for its prevention, also try to restore the balance between the literature work on bias and noise.
The authors have provided numerous examples of realworld decisions in which the level of noise is frequently scandalously high. Here are a couple of such instances.
Medicine is noisy. When confronted with the same patient, several clinicians arrive at different conclusions regarding whether the patient has skin cancer, blood cancer, heart illness, depression, and various other ailments. For example, in an early investigation including 426 state hospital patients independently diagnosed by two psychiatrists, there was only 50% agreement about the kind of mental disorder present. Another early research with 153 outpatients revealed that 54% agreed. The source of the noise was not stated in these experiments. Intriguingly, however, it was shown that certain psychiatrists tend to allocate patients to particular diagnostic groups. Some psychiatrists were more likely than others to diagnose patients with depression and anxiety, respectively.
Personnel decisions are noisy. Job interviewers make wildly disparate assessments of the same individuals. Employee performance appraisals are likewise very varied, relying more on the individual conducting the assessment than on the performance being evaluated.
Bail decisions make a lot of noise. Whether an accused individual is granted bail or imprisoned pending trial is partially determined by the identity of the judge who hears the case. Some judges are much kinder than others. For example, According to a 1974 survey of 50 judges who handed down judgments in similar (hypothetical) instances, ‘lack of unanimity was the rule’. Moreover, there were significant differences in the penalties from judge to judge. The same heroin dealer may get a sentence ranging from 5 to 18 years, an extortionist could receive a term between 1 and ten years, and a bank robber could receive a sentence ranging from 3 years with no fee to 20 years with a $65,000 fine. More studies conducted in 1977 and 1981 were very much like the earlier ones, and all of them came to the same alarming conclusions.
The book focuses on how to identify and quantify noise, as well as how to prevent such problems when making decisions. The book is broken down into six sections. In the first half of the book, the author defines and distinguishes between bias and noise and provides examples of noise in both public and private enterprises while making decisions. In the second half, the writers look into the nature of human judgment and how to assess accuracy and inaccuracy. The notion of a noise audit is introduced in this section of the book, which can be used to identify noise in a judgment. In the third section of the book, the authors delve deeper into predictive judgment. When creating predictions, this book examines the key advantages of rules, formulae, and algorithms over individuals. How noise can affect prediction quality and how to measure the noise that impedes prediction quality? The fourth and fifth parts of the book examine the causes of noise in judgment and ways of strengthening our judgment and avoiding errors. The final section of the book addresses the subject of what is the appropriate degree of noise and whether we can manage to eradicate noise from our judgment.
The core of this book focuses on two aspects, one is the cause of noise in judgment, and the other one is its prevention. The three main causes of noise, according to the book, are substitution biases, which lead to evidence misweighting; conclusion biases, which lead us to either ignore or consider evidence in a distorted manner; and excessive coherence, which magnifies the effect of initial impressions while reducing the impact of contradictory information. All of these biases can cause both noise and statistical bias. This book also discussed judgmentenhancing techniques, or how to avoid noise. The first step, according to the authors, is for organizations to realize that noise in professional judgment is a problem that has to be addressed. They recommend a noise audit to get to that position. Multiple people judge the same issues in a noise audit. The variability of these assessments is referred to as noise. In the latter section of their book, the authors discuss noise-reduction measures. They offer the concept of decision hygiene: the strategy they advocate for reducing noise in human judgments. They give case studies in five distinct subject areas. In each area, the authors investigate the ubiquity of noise and the horror tales it inspires. In addition, they assess the effectiveness or failure of noise-reduction initiatives. The decision hygiene tactics they proposed for coping with noise are listed below.
Sequencing information, has been explored in relation to forensic labs. The book highlighted several instances in which examiners’ judgments in forensic labs generate a great deal of noise since, in every judgment some information is significant and some is not. More knowledge is not always preferable, particularly if it can influence decisions by causing the judge to create a premature intuition. In this spirit, the new processes that should be used in forensic labs strive to maintain the examiners’ independence by providing them with just the information they need at the time they require it. In other words, the laboratory should conceal as much information as possible regarding the case and disclose it gradually.
Aggregating multiple independent judgments, in the instance of forecasting, demonstrates the importance of one of the essential noise-reduction techniques: the aggregation of several independent judgments. The wisdom of crowds idea is based on the arithmetic mean of several independent assessments, which is guaranteed to decrease noise. There are more strategies for aggregating judgments beyond simple averaging. The simplest technique to aggregate several projections is to take their average. Mathematically speaking, averaging reduces noise by dividing it by the square root of the number of averaged assessments. This indicates that if you average 100 assessments, the noise will be reduced by 90%. If you average 400 judgments, the noise will be reduced by 95%, effectively eradicating it. This statistical rule powers the wisdom of crowds method.
Judgment guidelines, in the case of medicine, may be a significant noise-reduction device since they directly minimize between-judge variability in final judgments. Prominent suggestions ask for more uniform diagnostic criteria. These include (a) making diagnostic criteria clearer and moving away from vague standards; (b) making reference definitions of symptoms and their level of severity, based on the idea that when clinicians agree on the presence or absence of symptoms, they are more likely to agree on the diagnosis; and (c) using structured interviews with patients in addition to open conversation.
Shared scale grounded in an outside view, is also a decision hygiene strategy in case of performance assessment in business organisations. Efforts to minimize noise there illustrate the fundamental relevance of utilizing a common scale based on an outside perspective. This is a critical decision hygiene method for one simple reason: judgment includes translating an experience into a scale, and if various judges use different scales, there will be noise. Various raters have radically different conceptions of what good or great entails. They will only agree if we provide them with real examples to use as anchors on the rating scale and other instruments for dealing with noise, such as rules and algorithms, to avoid noise in judgment.
However, the authors have agreed that no matter how much unwanted noise may be, the cost of removing it may occasionally outweigh the advantages. They have identified several barriers to reducing or eliminating noise, including the fact that reducing noise can be costly, that some noise-reduction strategies may introduce errors of their own, that some noise-reduction strategies may encourage opportunistic behaviour, and that people do not want to be treated as mere objects or cogs in a machine. Some noise-reduction measures may stifle innovation and demoralize people.
Noise: A Flaw in Human Judgment is a well-written book that tackles a fresh issue. After reading it, I believe it is a must-read for managers and decision-makers who want to learn about many defects in human judgment, particularly noise, as well as how to identify noise in judgment and several strategies to eliminate it. It will assist them in making better decisions and being more accurate in their judgment and prediction. Further research on this innovative topic can aid in identifying noise in various public and private organizations and understanding the nature of noise, which can lead to the development of mechanisms for noise avoidance, resulting in more informed and principled decisions.