between L and S and to price both prospects by stating the smallest immediate
cash payment for which they would be willing to exchange each prospect
(Tversky, Slovic, and Kahneman 1990). Because the payoffs and the prices
again are expressed in the same units, compatibility suggests that the long-term
prospect (offering the higher payoff) will be overvalued in pricing relative
to choice. In accord with this hypothesis, subjects chose the short-term prospect
74percentofthetimebutpricedthelong-termprospectabovetheshort-term
prospect 75 percent of the time. These observations indicate that different
methods of elicitation (for example, choice and pricing) can induce different
weightings of attributes that in turn give rise to preference reversals.
26.5.2 Relative Prominence
Another psychological mechanism that leads to violations of procedure invari-
ance involves the notion of relative prominence. In many cases, people agree
that one attribute (for instance, safety) is more important than another (such as
cost). Although the interpretation of such claims is not entirely clear, there is
evidence that the attribute that is judged more important looms larger in choice
than in pricing (Tversky, Sattath, and Slovic 1988). This is theprominence hy-
pothesis. To illustrate this notion, consider two programs designed to reduce the
number of fatalities due to traffic accidents, characterized by the expected re-
duction in the number of casualties and an estimated cost. Because human lives
areregardedasmoreimportantthanmoney,theprominencehypothesispre-
dicts that this dimension will be given more weight in choice than in pricing.
When given a choice between programs X and Y (see accompanying table), the
great majority of respondents favored X, the more expensive program that saves
more lives.
Expected Number of Casualties CostProgram X 500 $55 million
Program Y 570 $12 million
However, when the cost of one of the programs is removed and subjects are
asked to determine the missing cost so as to make the two programs equally
attractive, nearly all subjects assign values that imply a preference for Y, the
less expensive program that saves fewer lives. For example, when the cost of
program X is removed, the median estimate of the missing cost that renders the
two programs equally attractive is $40 million. This choice implies that at $55
million, program X should not be chosen over program Y, contrary to the
aforementioned choice. Thus, the prominent attribute (saving lives) dominates
the choice but not the pricing. This discrepancy suggests that different public
policies will be supported depending on whether people are asked which pol-
icy they prefer or how much, in their opinion, each policy ought to cost.
Further applications of the prominence hypothesis were reported in a study
of people’s response to environmental problems (Kahneman and Ritov 1993).
Several pairs of issues were selected, where one issue involves human health
or safety and the other protection of the environment. Each issue includes a
brief statement of a problem, along with a suggested form of intervention, as
illustrated.
612 Eldar Shafir and Amos Tversky