旧实验：

说一个人，为了证明一种叫...invest bias(好像是啊)的理论，设计了一种实验。把投资人分为两组，分别投资到两个项目中去。然后告诉一组人，他们的投资的其中一个项目回报率很低，结果很多人把更多的钱投入到这个项目中去。告诉另外一组人，他们的他一个项目投资回报率很高，结果他们不愿意花更多的钱投入原来的项目了。

新实验：另外一个人不同意他的结果，认为这种结果不是general 的情况，而且他的实验设计本身存在问题。所以又设计了另外一种实验，也是两组人两种投资，但是投资的项目不同了，好像不只两种，得出的结论就不一样了。

※背景知识：

Escalation Bias: Does It Extend toMarketing?

Escalation bias implies that managers favorreinvestments in projects that are doing poorly over those doing well. Wetested this implication in a marketing context by conducting experiments onadvertising and product-design decisions.

Each situation was varied to reflect eithera long-term or a short-term decision. Besides these four conditions, weconducted three replications. We found little evidence of escalation bias by365 subjects in the seven experimental comparisons.

Research on sunk costs shows that decisionmakers often continue to invest unwisely because of commitments to priorinvestments (e.g., Arkes and Blumer 1985). Escalation research, startingwithStaw (1976), yields evidence consistent with research on the sunk-costfallacy. In addition, escalation research has concluded that committed managersare likely to reinvest more in project when they receive information that theproject is doing poorly than if told it is doing well. This has been called"escalation bias." This bias of "throwing good money afterbad" has intuitive appeal. Should marketing managers be concerned about escalationbias

Common sense would say "yes." We examinedwhether escalation bias is likely to lead to poor marketing Journal of theAcademy of Marketing Science Volume 21, Number 3, pages 247-253.Copyright 91993 by Academy of Marketing Science. All rights of reproduction in any formreserved. ISSN 0092-0703. decisions. We first provide a brief review of theliterature. This is followed by a series of experiments to test whetherescalation bias occurs in marketing decision making. Our hypothesis was thatescalation bias would occur for marketing decisions.

BRIEF EVALUATION OF PRIOR RESEARCH

When people make a decision, they often developa commitment to it. This seems especially likely for people inorganizations. For example, Fox and Staw (1979) showed that "trappedadministrators" showed a higher level of commitment to a previous courseof action than "nontrapped administrators." If decisions turn outpoorly, the "escalation of commitment" hypothesis suggests that they mayinvest further so as to make the decision turn out well. Conlon and Parks(1987) called this retrospective rationality. In simple terms, decision makersare often unwilling to take a loss. Staw (1981, p. 577) commented that: Becauseit is often possible for persons who have suffered a setback to recoup theirlosses through an even greater commitment of resources to this same course ofaction, a cycle of escalating commitment can be produced. Staw and Ross (1987a)and Garland (1990) summarized a variety of reasons for the existence ofescalation bias. For example, managers may be affected by partial reinforcementand have difficulty accepting information that rewards have disappeared.They might also distort negative information through perceptual defensemechanisms. Organization-JAMS 247 SUMMER, 1993