Gain/loss asymmetry and riskless choice: Loss aversion in choices among job finalists

Tversky and Kahneman (1991) extended the notions of reference points and gain/loss asymmetry to multiattribute choice tasks having alternatives that cannot be assigned explicit probabilities. Two experiments were designed to test these propositions in the context of choices among job finalists. In the first experiment, undergraduates (N = 82) were provided with attribute information on two job candidates, along with (incomplete) attribute information on a previous job incumbent. The results showed that the relative standing of the job finalists was evaluated from the reference point of the incumbent's performance on the non-missing attributes. Candidates representing the status quo (i.e., no change in performance on the non-missing attributes) were preferred over candidates representing a loss. However candidates representing a gain were not preferred over candidates representing the status quo. In the second experiment, undergraduates (N = 74) received assessor training and then watched a demonstration of performance in one of two work samples. Participants told that the demonstration was one of excellent performance strongly preferred the status quo candidate (i.e., the one that performed equally well on the demonstration work sample). However, participants told that the example was one of poor performance were indifferent between the status quo and gain candidate. We conclude that the gain/loss asymmetry commonly found in risky choice can also be found in riskless choice tasks such as choosing among job finalists. (C) 1996 Academic Press.