Subadditivity effect

The subadditivity effect is assigning a lower probability to a whole event than the probabilities assigned to its mutually exclusive parts add up to.

WikipediaKnowledge quiz Self assessment Microcourse

Mechanism

How it works

Unpacking an event makes several concrete pathways easier to imagine, so their estimated probabilities can each receive too much weight. A broad category feels less vivid and gets a smaller estimate, even though it logically includes the named subcategories.

Examples

Where it shows up

A person assigns 10% each to several mutually exclusive causes of a delay, then says the chance of any delay is only 20% even though the parts sum higher.
An insurer estimates a broad cause of loss as unlikely, but gives larger combined estimates when that cause is split into memorable named scenarios.
A team says an incident is unlikely in aggregate but assigns high chances to several distinct routes by which it could occur.

Consequences

What it can distort

Risk estimates become internally inconsistent, and the framing of a question can change a decision without changing the underlying evidence.
Breaking down only one option can also bias comparisons by making that option feel more probable or more risky.

Countermeasures

How to work around it

Define whether categories are mutually exclusive and exhaustive, then check that part estimates sum to the estimate for the whole.
Use the same level of detail for competing options; do not unpack only the outcome you want people to notice.

Caveats

Critiques and limits

Some apparent violations reflect ambiguous categories or legitimate uncertainty about overlap. The effect is clearest when the parts are explicitly mutually exclusive and exhaustive.

Taxonomy

Fields of impact

Political Science & Public Policy

Psychology & Social Science

Economics & Finance

Decision Science & Risk Management

Evidence

How solid is the research?

Robust — replicates reliably

Subadditive probability judgments are a well-established result in support theory, though framing and category clarity strongly affect the magnitude.

Research

Relevant papers

Support theory: A nonextensional representation of subjective probability

Tversky, A., & Koehler, D. J. (1994)

Psychological Review, 101(4), 547–567

A Belief-Based Account of Decision under Uncertainty. In J. B. Shanteau, B. A. Mellers, & D. Schum (Eds.), Decision Research from Bayesian Approaches to Normative Systems: Reflections on the Contributions of Ward Edwards (pp. 155–183)

Fox, C. R., & Tversky, A. (1998)

Springer

Case studies

Real-world patterns.

Real-world examples showing how Subadditivity effect manifests in practice

Case study

Underestimating Launch Risk: When 'Everything' Feels Safer Than Its Parts

A real-world example of Subadditivity effect in action

Context

A mid-stage SaaS company prepared to release a major new feature to its enterprise customers. Executive leadership wanted a single-number assessment of the risk of a problematic launch to decide whether to delay and allocate extra engineering and customer-success resources.

Situation

The product manager was asked in the weekly steering meeting: 'What's the probability this launch will cause a major customer-impacting incident?' She replied, 'About 12%.' Separately, engineering and operations teams were asked to list and estimate probabilities for specific failure modes during a planning exercise.

The bias in action

When asked for a single, global likelihood the product manager gave a relatively low number (12%). However, when engineers estimated individual failure modes—database migration rollback (10%), API rate-limit misconfiguration (8%), third-party auth outage (7%), edge-caching mismatch (6%), and a critical client-specific integration bug (12%)—the sum of those parts amounted to 43%. Decision-makers treated the 12% holistic estimate as the authoritative figure and did not act on the higher cumulative number generated by the decomposed estimates. This mismatch is a textbook example of the subadditivity effect: people judge the probability of the whole (a major incident) to be less than the sum of its part-events when those parts are enumerated.

Outcome

Leadership approved the launch with only light contingency staffing and no extra rollback rehearsals. Within two weeks a combination of the listed failure modes occurred (an auth provider outage triggered a cascade that exposed a client-specific integration bug), producing a major outage that required a hotfix and rollback. Recovery took 27 hours, some customers experienced data access disruption, and the company reputation suffered among key accounts.

What's inside the full case study

Unlock the deeper breakdown with real-world impact, measurable effects, lessons learned, better-approach recommendations, and relevant fields.

Real-world impact

Affected groups, timeframe, and measurable outcomes.

Lessons learned

Practical takeaways and a better path forward.

Full case breakdownEmail access

Want the full analysis?

Request access to the complete case study, including measurable impact, lessons learned, and the recommended better approach.

Recommended books

Thinking, Fast and Slow

Daniel Kahneman, 2011

Related biases

Nearby patterns.

Conjunction fallacy

The conjunction fallacy is judging a detailed story as more likely than a broader category that includes it.

Base rate fallacy

The base-rate fallacy is judging a specific case from a vivid clue while neglecting how common the outcome was before the clue appeared.

Study on Microcourse

Learn the wider pattern.

Dive deeper into Subadditivity effect and related biases in Reasoning and Logical Fallacieswith structured lessons, examples, and practice exercises.

Study

Practice

Test your knowledge.

Apply what you have learned and reinforce your understanding of Subadditivity effect with a short quiz or self-assessment.

Understanding quiz Self-assessment

Entry last reviewed 2026-07-19 · sources verified against the published literature — methodology