have been examined with only a small subset examined in any one study. The predominant method of assessing the meaning of a probability expression has been to ask subjects to provide a percentage from 0 to 100 that corresponds to it(Reagan, Mosteller, and Youtz, 1989). Results consistently show a large degree of between-subject variability(Budescu and Wallsten, 1985). This varıabiliıty has been found to be inversely related to the distance from the center of the scale (Wallsten, ef al, 1986). The variability has been found to be lower in groups from homogeneous backgrounds(Brun and Teigen, 1988), and expressions embedded in context exhibit more varlability than those in isolation(Beyth-Marom, 1982).

Results reported in the psychology literature also show a lack of symmetry(Budescu and Wallsten, 1985). That is, probabilities assıgned to mirror-ımage pairs such as"probable" and "improbable" do not sum to 100 percent(Lichtenstein and Newman, 1967). Negative probability expressions tend to be further away from the 50 percent midpoint than the related positive expressions(Reagan, Mosteller, and Youtz, 1989).

The variability in responses to uncertainty expressions can be due to(a) the vague meaning of the expressions or(b) the fact that the meaning of probability terms is not constant across individuals. The results of Wallsten, ef al(1986) support the notion that non-numerical probability expressions convey vague uncertainties; the vagueness can be attributed to the use of a verbal expression and not to the perceived uncertainty. They suggest that people not only understand uncertainty expressions as representing amounts of probability but also as representing degrees of confidence in that probability.

Phillips and Wright(1975) introduced the notion that culture can influence the cognitive

processes involved in probability assessment. They hypothesized that English people(who have