Notes on (Tversky and Kahneman 1974) – Judgment under Uncertainty: Heuristics and Biases

Main Topic or Phenomenon

This paper examines how people make probability judgments and predictions under uncertainty, focusing on the systematic errors that occur when individuals rely on intuitive heuristic principles rather than normative statistical reasoning.

Theoretical Construct

The paper introduces three key heuristics:

Representativeness: People evaluate probabilities by the degree to which an event or object resembles or is representative of a particular mental model or category. For example, judging the probability that Steve is a librarian based on how well his personality description matches the stereotype of librarians.

Availability: People assess the frequency or probability of events by the ease with which instances or occurrences can be brought to mind or mentally constructed. More memorable or imaginable events are judged as more frequent or likely.

Anchoring and Adjustment: People make estimates by starting from an initial value (anchor) and adjusting from this starting point. However, adjustments are typically insufficient, leading to final estimates that remain biased toward the initial anchor.

Key Findings

1. People systematically ignore base rates (prior probabilities) when making representativeness judgments, even when this information is explicitly provided
2. Sample size effects are largely ignored - people expect small samples to be as representative of populations as large samples
3. People show misconceptions about randomness, expecting local representativeness in random sequences
4. Availability biases lead to overestimation of easily recalled events (famous names, recent occurrences, vivid scenarios)
5. Anchoring effects are robust and occur even with random or irrelevant starting points
6. People show overconfidence in predictions, creating overly narrow confidence intervals
7. These biases persist even among statistically sophisticated individuals when making intuitive judgments

Boundary Conditions or Moderators

• Statistical training: Reduces some elementary errors but doesn’t eliminate biases in complex problems
• Motivation and incentives: Monetary rewards for accuracy don’t significantly reduce anchoring effects
• Type of information: Worthless evidence is treated differently than no evidence (base rates ignored vs. properly utilized)
• Method of elicitation: Different procedures for assessing subjective probabilities yield systematically different results
• Expertise domain: Even experts show biases when making intuitive judgments outside their formal analytical procedures

Building on Previous Work

This paper synthesizes and extends scattered findings about judgmental biases into a coherent theoretical framework. It challenges the normative decision theory assumption that people naturally follow statistical principles, instead proposing that systematic deviations from rationality follow predictable patterns based on underlying cognitive processes.

Major Theoretical Contribution

The paper establishes heuristics as fundamental cognitive processes that, while generally adaptive and efficient, produce systematic and predictable biases. This shifts focus from viewing judgmental errors as random noise to understanding them as arising from the structure of human information processing. It provides a descriptive account of how people actually make judgments rather than how they should make them normatively.

Major Managerial Implication

Managers should be aware that their intuitive judgments about uncertain events are systematically biased in predictable ways. Key implications include:

• Don’t rely solely on how “representative” an outcome seems
• Consider base rates explicitly in decision making
• Be aware that recent or memorable events may be overweighted
• Question initial estimates and actively seek disconfirming information
• Use formal analytical procedures for important decisions rather than relying on intuition

Unexplored Theoretical Factors

Several factors that could influence heuristic use were not explored:

Individual Differences: Cognitive style, need for closure, analytical thinking disposition, cultural background, or personality traits that might make some people more or less susceptible to these biases.

Emotional States: How mood, anxiety, time pressure, or other affective states might influence reliance on heuristics versus more systematic processing.

Social Context: How group settings, social proof, or interpersonal dynamics might amplify or reduce heuristic biases.

Domain Expertise: More nuanced examination of when and how domain knowledge interacts with heuristic processing.

Metacognitive Awareness: Whether people’s awareness of their own judgment processes influences bias susceptibility.

Decision Importance: How the stakes or consequences of decisions might moderate heuristic use.

Information Format: How the presentation, framing, or structure of information influences which heuristics are activated.

Reference

Tversky, Amos and Daniel Kahneman (1974), “Judgment under Uncertainty: Heuristics and Biases,” Science, 185 (4157), 1124–31.