Noise: A Flaw in Human Judgment – Understanding the Hidden Errors in Decision-Making

noise in judgement

Let us say you’re holding a stopwatch. Try to stop it when you think exactly 10 seconds have passed. Do it a few times. Did you hit exactly 10 seconds every time? Most likely, no. Sometimes you’re a little early, and sometimes a bit late. These little inconsistencies are random errors. This randomness is called noise, and while it seems trivial in this stopwatch example, noise can lead to serious consequences when it seeps into our daily decisions.

In Noise: A Flaw in Human Judgment, Daniel Kahneman, Olivier Sibony, and Cass R. Sunstein expose how noise—random variability in judgments—undermines decision-making in ways we barely notice. Let’s explore the concept of noise, how it differs from bias, why it’s so costly, and what we can do to reduce it.

What is Noise in Human Judgment?

Noise refers to random inconsistencies in judgments made by individuals or systems, even when the situation remains constant. To make it relatable, let’s use an example:

Imagine two students, Anjali and Ramesh, applying to a prestigious university. They both have similar grades and extracurriculars. Anjali’s interview is on a sunny day, and she’s warmly welcomed by the admissions officer. Ramesh’s interview happens on a gloomy, rainy day when the officer is dealing with a headache. Weeks later, Anjali gets in, but Ramesh doesn’t.

What happened? It could be occasional noise—unpredictable external factors like the interviewer’s mood, weather, or other unrelated events influencing the outcome. According to studies, judgments are often swayed by irrelevant factors such as whether it’s before or after lunch or even the day of the week.

Noise doesn’t just impact college admissions. In Miami immigration courts, one study found that judges granted asylum at rates ranging from 5% to 88%, depending on the individual judge. Similar patterns are seen in criminal sentencing, medical diagnoses, and employee evaluations.

Noise vs. Bias: What’s the Difference?

Many people confuse noise with bias. While both are sources of error, they’re fundamentally different.

  • Bias refers to systematic errors in judgment that consistently skew results in a particular direction. For instance, if a hiring manager has a tendency to favor candidates from a specific college, that exemplifies bias.
  • Noise, on the other hand, represents random variability in decisions that do not follow any consistent pattern. Picture 10 different hiring managers assessing the same candidate and reaching vastly different conclusions.

Bias is like consistently missing the bullseye in the same direction. Noise, on the other hand, is like hitting different spots all over the target, with no predictable pattern. While bias is often easier to identify and correct, noise is subtle, making it harder to detect and eliminate.


Types of Noise

The book breaks noise into three main categories:

  1. System Noise: Variability in judgments across individuals in the same system. For example, two doctors might give different diagnoses for the same symptoms.
  2. Occasion Noise: Variability caused by external factors, like time of day or mood.
  3. Level Noise: Differences in judgment due to individual tendencies. Some people are naturally stricter, while others are more lenient.

In any system, these types of noise combine to create inconsistencies that can be costly and unfair.


The Real-World Costs of Noise

Noise might sound abstract, but its impact is anything but. It can lead to life-altering mistakes in fields like law, medicine, hiring, and more.

1. Justice System

Consider bail hearings. Judges often decide whether a defendant should await trial in jail or be released. Research shows that these decisions can depend on irrelevant factors like the time of day or the judge’s mood. Algorithms, which base decisions on objective data, outperform human judges in predicting outcomes like flight risk. This isn’t because algorithms are smarter—they simply lack the variability caused by human emotions and biases.

2. Healthcare

Inconsistent medical diagnoses are another glaring example. Studies reveal that different doctors can provide different diagnoses for the same symptoms, leading to unnecessary treatments or missed illnesses. For example, one doctor might diagnose chest pain as a heart attack, while another might attribute it to acid reflux.

3. Corporate Decisions

Noise affects hiring, performance evaluations, and promotions. If three managers rate an employee’s performance, their assessments can vary widely. This inconsistency undermines fairness and leads to poor organizational decisions.

4. Insurance and Finance

Underwriting insurance policies or approving loans should be consistent processes. However, studies show significant variability in these decisions. This noise translates into financial losses for companies and unequal treatment for customers.


Why Do We Ignore Noise?

Humans tend to ignore noise because it doesn’t fit into tidy stories. Bias, on the other hand, is more apparent since it follows a pattern. For example, when women consistently earn less than men, the bias is obvious. However, noise—those random variations—can be much more difficult to identify.

If a student’s college application gets turned down, we might attribute it to their grades or think the interviewer simply didn’t like them. It’s rare for us to consider, “Perhaps the interviewer was just having a bad day.” Noise often goes unnoticed, hiding in plain sight.


How to Reduce Noise

The good news? Noise isn’t inevitable. Here are some practical strategies to minimize it:

1. The Wisdom of Crowds

Remember the classic experiment where people guessed the weight of an ox? While individual guesses were wildly inaccurate, the average of all guesses was almost perfect. This principle works in decision-making too. By averaging multiple independent judgments, you can cancel out noise.

For instance, when hiring candidates, involve multiple evaluators and average their scores rather than relying on a single person’s opinion.

2. Use Algorithms

Algorithms often outperform humans in consistency. In fields like bail decisions, hiring, and even diagnosing illnesses, algorithms remove the variability caused by human emotions and external factors.

3. Implement Decision Hygiene

The authors of Noise advocate for structured decision-making processes. This includes:

  • Standardized Checklists: Ensuring everyone follows the same criteria when making decisions.
  • Guidelines for Evaluation: Reducing subjectivity by using predefined scoring systems.
  • Blind Evaluations: Eliminating irrelevant information that could sway judgments, such as names or appearances.

4. Train for Consistency

Organizations can reduce noise by training employees to align their judgments. For example, if multiple doctors assess a patient’s condition, they should use the same diagnostic criteria to minimize variability.


Why Reducing Noise Matters

Eliminating noise isn’t just about fairness—it’s about making better decisions. Inconsistent judgments lead to errors, inefficiencies, and lost opportunities. By addressing noise, we can:

  • Ensure fair treatment in the justice system.
  • Improve healthcare outcomes.
  • Enhance workplace decisions.
  • Save money in industries like insurance and finance.

Moreover, reducing noise builds trust. When people see that decisions are consistent and fair, they’re more likely to accept the outcomes, even if they don’t agree with them.


Summary Infographics

noise book summary

Read more about judgements in our blog page Here

Leave a Comment

Your email address will not be published. Required fields are marked *

Shopping Cart