Decision Tree Analysis

tool · management · organizing-schema

Branching diagram of choices, chance events, and outcomes with associated values.

Decision Tree Analysis represents a sequential decision problem as a branching diagram with three node types: decision nodes (squares, where the decision-maker chooses an action), chance nodes (circles, where nature determines the outcome with given probabilities), and terminal nodes (triangles, with associated payoffs or utilities). The tree is solved by 'rolling back' from terminal nodes — computing expected values at chance nodes (probability-weighted averages of subsequent values) and selecting the maximum-value branch at decision nodes — yielding both the optimal first-decision choice and the expected value of the entire problem. The technique was substantially developed in Howard Raiffa's Decision Analysis (1968) within the broader Bayesian decision-theoretic framework that emerged from the Harvard Business School and Stanford operations research traditions in the 1950s and 1960s. Decision trees are foundational in operations research, management consulting (particularly oil and gas, pharmaceuticals, and capital projects), and the related field of real options analysis.

Originators

Howard Raiffa (decision-analysis formalization); Ronald Howard (Stanford decision analysis school) high

Year / Decade

1960s formalization; 1968 (Raiffa Decision Analysis) high

Primary sources

Raiffa, H. (1968). Decision Analysis: Introductory Lectures on Choices Under Uncertainty, Howard, R.A. (1966). 'Decision Analysis: Applied Decision Theory' high

Core components

Decision nodes (squares, choice points)
Chance nodes (circles, with probability distributions)
Terminal nodes (with payoffs or utilities)
Roll-back solution method (backward induction)
Expected value computation
Sensitivity analysis on probabilities and payoffs
Connection to Bayesian decision theory
Real options extension

Primary use case

Capital budgeting and investment decisions under uncertainty; pharmaceutical R&D portfolio decisions; oil and gas exploration; clinical decision analysis; structured analysis of sequential decisions; foundation for real options analysis.

Common criticisms

Requires probability assignments that are often subjective or unavailable
expected-value maximization may not match risk-averse decision-maker preferences (driving utility-function refinements)
tree complexity grows exponentially with decision and chance variables
doesn't easily handle continuous variables, requiring discretization
assumes decisions are made at clearly defined points rather than continuously
backward-induction logic can produce optimal decisions that violate intuition (St Petersburg paradox and similar)
behavioral departures from expected-utility maximization (prospect theory) are not captured.

Lineage

Siblings: Pugh Matrix