Where does escalation of commitment collide?

The tendency of a decision maker who is responsible for a failing course of action to commit still more resources to it, in order to justify the original decision rather than to cut the loss. Staw's 1976 study Knee-Deep in the Big Muddy gave participants a prior investment decision that had turned out badly, and found that those who felt personally responsible for the setback poured the largest sums into the same option on a second round. His 1981 review named the pattern the escalation of commitment and located its engine in self-justification: people escalate to prove to themselves and to others that their earlier choice was not a mistake. The referent here is a motivated process in a person who owns a prior decision, not a cold arithmetic error and not the emotional bond an employee feels toward an organization.[1][2]

Unit

A responsible decision maker's added investment in a failing course of action

Recovery means

Escalation eases when responsibility for the prior decision is separated from the choice to continue, so a fresh decision maker can stop the project

Measured by

Amount of additional resources allocated to a previously chosen, now-failing option, compared across levels of personal responsibility

The tendency to continue an endeavour once money, effort, or time has been invested in it, even when the prior cost cannot be recovered and should be irrelevant to the choice going forward. Arkes and Blumer's 1985 paper The Psychology of Sunk Cost showed the effect across many small experiments: people who had paid for a theatre ticket or a ski trip were more likely to use it in bad conditions, and people told a project was nearly complete kept funding it. Their account frames the behaviour as a cognitive bias, a violation of the economic rule that only future costs and benefits matter, rather than as a drive to defend one's own past decision. The referent is a reasoning error about past outlays that can appear even when the person did not make the original decision and has no reputation to protect.[3]

Unit

A person's willingness to continue an endeavour because of unrecoverable prior investment

Recovery means

The bias weakens when the prior outlay is reframed as unrecoverable and the choice is posed only in terms of future costs and benefits

Measured by

Choice to continue or abandon an option, compared between conditions that differ only in a prior, non-recoverable cost

The continued funding of a troubled software project long after the evidence says it should be stopped, studied as the practical face of escalation inside real organizations. Keil, Tan, Wei, Saarinen, Tuunainen, and Wassenaar's 2000 cross-cultural study asked information systems managers to judge scenarios in which a large development project was going badly, and measured their willingness to keep it alive. The work carries escalation out of the laboratory and into project governance, where a runaway project consumes budget and staff while completion stays out of reach, and where the interest is as much in de-escalation, the act of pulling the plug, as in the drift that got there. The referent is an organizational decision about a specific project, shaped by responsibility, sunk cost, and the difficulty of admitting failure, and not the loyalty a developer feels toward the firm.[4]

Unit

An organization's continued funding of a specific failing project

Recovery means

De-escalation, the deliberate decision to redirect or cancel the project, often after new leadership or an outside review breaks the prior commitment

Measured by

Managers' willingness to continue funding a failing project scenario, and case tracking of budget and time before a project is stopped

A formal game in which the structure of the payoffs, rather than any private motive, drives players to keep investing in a contest they cannot win cleanly. Shubik's 1971 The Dollar Auction Game auctions a dollar to the highest bidder under one added rule: the second-highest bidder also pays their bid and gets nothing. Once two players are committed, each raises to avoid the larger loss of stopping, and the bidding can climb well past a dollar, a paradox of noncooperative behaviour and escalation with no self-justification and no cognitive slip required. The referent is a property of an incentive structure that traps rational players, so it explains the same escalating pattern from the outside, through the rules of the game, rather than from inside a single decision maker's psychology.[5]

Unit

A player's continued bidding to avoid the larger loss of quitting a contest

Recovery means

The trap is avoided by refusing the first bid or by players coordinating, since once two are committed the payoff structure rewards continuing

Measured by

Bids relative to the prize in the dollar-auction game, and the escalation point at which total bids exceed the value at stake