Escalation of Commitment

frame 02 / 09

Provenance

Where it was first filed

Barry M. Staw, 1976 — Knee-deep in the Big Muddy: A study of escalating commitment to a chosen course of action. Staw had 240 business-school students role-play a corporate financial officer allocating R&D funds. He manipulated (a) whether the participant had made the initial losing allocation themselves (high vs. low personal responsibility) and (b) whether the subsequent outcome was positive or negative. Participants who were personally responsible AND faced negative consequences committed the most additional money to the original division. The title quotes a Pete Seeger anti-Vietnam-War song; Staw framed the work partly as a way to understand persistent commitment to failing policies such as the Vietnam War.

frame 03 / 09

Tested variation 1

Staw 1976 - the founding responsibility x outcome experiment

Greatest reinvestment occurred in the high-responsibility / negative-consequence cell, demonstrating that personal responsibility for a failing decision drives further commitment to it.

Barry M. Staw, 1976 · yield significant responsibility x consequences interaction (exact effect size not reported in modern standardized terms in original) · 240 business administration students

frame 04 / 09

Tested variation 2

Arkes & Blumer 1985 - sunk cost effect (theater field study + radar-plane scenario)

Full-price patrons attended more plays in the first half-season (~4.11 vs ~3.3), and 85% chose to finish the doomed plane project when sunk costs were mentioned vs. ~17% when they were not - direct evidence that past, unrecoverable spending drives continued investment.

Hal R. Arkes & Catherine Blumer, 1985 · yield theater: 4.11 vs ~3.3 plays in first half-season; scenario: 85% vs ~17% choosing to continue · ~60 theater patrons (field study); college-student samples for scenarios

frame 05 / 09

Tested variation 3

Sleesman, Conlon, McNamara & Miles 2012 - meta-analytic determinants

Confirmed escalation as a real, multiply-determined effect. Estimated true-score correlations: ego threat rho ~= .47; time investment rho ~= .43; proximity to completion rho ~= .39; sunk costs rho ~= .24; personal responsibility for the initial decision rho ~= .26. Ego threat and prior time investment were the strongest drivers - stronger than sunk cost or responsibility.

Dustin J. Sleesman, Donald E. Conlon, Gerry McNamara & Jonathan E. Miles, 2012 · yield corrected correlations rho: ego threat .47, time investment .43, completion .39, sunk costs .24, responsibility .26 · 166 independent samples aggregated (from an initial pool of 917 articles)

frame 06 / 09

Method

How the judgment forms

The dominant account is self-justification (Staw): once a person owns a decision, persisting lets them avoid admitting the original choice was wrong, protecting self-image and public reputation. The 2012 meta-analysis supports this - ego threat (face-saving) was the single strongest driver, and effects are larger under high personal responsibility. Complementary accounts include prospect theory / loss aversion (a failing project is framed as a loss domain, where people become risk-seeking and 'gamble' to break even), self-perception (people infer a positive attitude from their own persistent time investment, consistent with time investment being a top driver), and the sunk cost 'don't waste' rule (Arkes & Blumer). At the organizational level, agency problems, project momentum, and social/group identity add further pressure.

Self-justification vs. prospect-theory/loss-aversion vs. sunk-cost 'waste-avoidance' vs. self-perception accounts are not mutually exclusive; the meta-analysis suggests no single theory explains all variance and that situational drivers (ego threat, time invested) outweigh any one psychological mechanism. Some researchers (e.g., a 'preference-based' account) argue the classic responsibility effect partly reflects pre-existing preferences rather than pure self-justification.

frame 08 / 09

On record

Documented in the wild

• Shoreham Nuclear Power Plant · 1965-1989

  Long Island Lighting Company (LILCO) announced the plant in 1965; by completion in 1985 costs had reached roughly $5.5 billion, vastly above the original estimate. In 1989 LILCO agreed never to operate it - the first completed U.S. nuclear plant decommissioned without ever going online - with most of the ~$6 billion cost passed to ratepayers. A textbook case of continued investment despite mounting negative signals.

• U.S. escalation in the Vietnam War · 1965

  Repeated increases in troop commitment and spending despite negative battlefield feedback are the paradigmatic real-world escalation example; Staw explicitly invoked the war in naming the 1976 paper, and Staw & Ross (Science, 1989) analyze it as an escalation situation. George Ball's 1965 memo to President Johnson warned against deepening commitment.

• Expo 86, Vancouver · 1986

  Cited in the escalation literature (Staw & Ross) as a case where commitment to staging the world's fair continued despite escalating projected deficits and negative financial feedback.

• Berlin Brandenburg Airport · 2006-2020

  Construction continued through repeated failed openings; the project ran roughly EUR 6.5 billion over budget and about a decade behind schedule before finally opening, a modern documented escalation case.

• Denver International Airport automated baggage system · 1990s

  The automated baggage handling system ran roughly two years late and about $2 billion over budget before being largely abandoned, cited as an escalation-of-commitment / IT-project case.