Two experiments tested the hypothesis that framing biases in decision making would affect more strongly individuals with relatively low levels of need for cognition (NC). Participants were classified as high or low NC based on responses to a standard scale and subsequently were exposed to one of two framings of a choice problem. Different choice problems were used in each experiment, modeled after those developed by Kahneman and Tversky. Experiment 1 employed a monetary task and Experiment 2 a medical decision-making task. Consistent with expectations, framing effects on choice were observed in both experiments, but only for low NC participants. High NC participants were unaffected by problem framing, showing that they were less susceptible to attempts to alter their frame of reference.
KEY WORDS framing biases; need for cognition; individual differences have identified a variety of biases that are classified as 'framing effects', typically involving two versions of a choice problem that vary in relatively subtle ways, such as linguistic changes, yet elicit strikingly different choices among participants. A frequently cited example is the classic 'Asian disease problem', in which research participants are asked to choose between two treatments whose outcomes differ in riskiness but are of equal expected value -one which is guaranteed to save the lives of 1/3 of those stricken, and one which carries a 113 probability of saving everyone and a 2/3 probability of saving nobody. Under these circumstances, most people choose the former option (72% in the 1981 study by Tversky and Kahneman). However, by simply rephrasing the choice problem so that the language focused on lives to be lost rather than lives to be saved, Tversky and Kahneman (198 1) reported a complete reversal of preferences, such that 78% now favored the risky option.
A plethora of studies have replicated these 'framing effects' to varying degrees in the years since (see Schneider et af., 1995, for a review). Although the original theorists do not suggest that the effects are likely to be stronger or weaker under any particular circumstances (Fischhoff, 1983;, subsequent research has made clear that a number of moderating variables exist. Our understanding of framing effects has been significantly advanced by the CCC 0894-3257/96/040283-08 0 1996 by John Wiley & Sons, Ltd.