Conjunction Fallacy in Business Decisions

The classical Linda problem (Kahneman + Tversky, 1983):

'Linda is 31 years old, single, outspoken, very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination + social justice + participated in anti-nuclear demonstrations.'

Which is more probable?

• A. Linda is a bank teller.
• B. Linda is a bank teller AND active in feminist movement.

~85% of people choose B, the classic result from Tversky and Kahneman's Linda problem.

The math:

• B is necessarily LESS probable than A.
• If 100 women fit Linda's description, P(A) = X bank tellers (whatever X is).
• P(B) = X bank tellers AND feminist subset = some subset of A.
• By definition: P(B) ≤ P(A).

Why we get it wrong:

• Adding specific details makes a description feel more representative.
• Representativeness heuristic - we judge probability by similarity, not by counting.
• B 'sounds right' because it fits Linda's character profile.

1. Startup roadmaps:

• 'We'll grow ARR 20% MoM AND raise Series A in 12 months AND launch in 3 new markets AND hire 30 engineers.'
• Each step has its own probability: P(20% MoM) = 30%; P(Series A) = 40%; P(3 markets) = 60%; P(30 engineers) = 50%.
• Conjunction: 30% × 40% × 60% × 50% = 3.6%.
• But the plan SOUNDS achievable because each individual step is plausible.

2. Investment theses:

• 'This stock will outperform IF (A) management executes well + (B) market conditions remain favourable + (C) competitors don't catch up + (D) regulators don't intervene.'
• Each 'if' might have 60% probability.
• Conjunction: 60% × 60% × 60% × 60% = 13%.
• Far lower than the gut feel of 'this is likely to work'.

3. Project plans:

• 'Project will be delivered on time IF (A) technical implementation goes smoothly + (B) third-party integrations work + (C) stakeholder review is fast + (D) no scope changes.'
• Each 'if' might be 70%.
• Conjunction: 70% × 70% × 70% × 70% = 24%.
• Conjunction fallacy explains why projects routinely miss deadlines despite 'reasonable' planning.

4. M&A acquisition plans:

• 'Acquisition will succeed IF integration goes smoothly + cost synergies materialise + talent retention works + cultural fit holds + revenue projections meet target.'
• 5 conjuncts × ~70% each = ~17% combined probability.
• Most M&A deals fail to deliver projected value - conjunction fallacy is a key reason.

The fallacy is most pronounced when:

• Number of conjuncts is large (3+).
• Each conjunct is individually plausible.
• The conjunction tells a coherent narrative.
• Decision-maker is invested in the outcome.
• Group setting where dissent is suppressed.
• Time pressure - no opportunity for systematic analysis.

Why narratives amplify it:

• A coherent story feels more believable than separate facts.
• 'Founders will execute well + market will grow + technology will work + customers will pay + competitors won't catch up' tells a believable story.
• The same five facts as bullet points feel less compelling.

Why group settings amplify it:

• 'Groupthink' suppresses dissent on plausibility of individual conjuncts.
• Anchoring on first-presented optimistic estimate.
• Social pressure to share enthusiasm.

1. Plan involves multiple 'AND' connections: 'We'll do X AND Y AND Z AND...'.
2. Single rosy summary of plan probability: 'This is achievable' rather than 'each step has X% probability'.
3. Plausibility intuition feels high: 'This makes sense, we can do this' - check the math.
4. Narrative is coherent: each step follows from previous; smooth story.
5. Multiple uncertain dependencies: success requires X to work + Y to work + Z to work.
6. Stakeholders are invested: motivation to overestimate compounded with overoptimism bias.
7. Plan summary uses phrases like 'we just need to': signal of multiple sequential requirements.

1. Decompose + multiply:

• List each independent conjunct.
• Estimate probability of each conjunct independently.
• Multiply (or use Bayesian model for dependent conjuncts).
• Compare to gut-feel of overall plan probability.

2. Identify weakest link:

• Which conjunct has lowest probability?
• Can you focus extra effort on improving it?
• Is there a backup plan if it fails?

3. Premortem analysis:

• Imagine the plan failed.
• Ask 'what went wrong?'.
• Reveals weak conjuncts + previously-ignored failure modes.
• See our premortem guide.

4. Reference class forecasting:

• How often have similar plans succeeded in the past?
• Base rate of comparable startup roadmaps achieving stated goals: ~10-30%.
• Use this as ceiling on your own plan's probability.

5. Decision journal:

• Write down predictions + actual outcomes.
• Reveals systematic over-confidence over time.
• Calibrates future estimates.

Important caveat: not all multi-step plans are conjunction-fallacy-prone.

Necessary sequential dependencies:

• 'Plant seeds AND water AND sun AND grow vegetables.'
• Each step is required for the next.
• Multiplying probabilities is appropriate.

Strategic optionality:

• 'Try strategy A OR strategy B OR strategy C.'
• Probability of at least one succeeding: 1 - (probability all fail) = often quite high.
• DON'T multiply here - sum the alternatives.

Mixed dependencies:

• Some real plans have BOTH conjunctive (AND) + disjunctive (OR) structure.
• Decompose properly + apply appropriate probability operations.
• The conjunction-fallacy concern applies to the AND parts.