where F is a cumulative distribution function. Hence, with random variables we can quantify the uncertainty in ordering with a convolution integral, Equation (9.2a). Second, let us assume that the uncertainty in rank arises because of ambiguity. For example, suppose we are trying to rank people’s preferences in colors. In this case the ranking is very subjective and not reducible to the elegant form available for some random variables, such as that given in Equation (9.2a). For fuzzy variables we are also able to quantify the uncertainty in ordering, but in this case we must do so with the notion of membership. A third type of ranking involves the notion of imprecision (Dubois and Prade, 1980). To develop this, suppose we have two fuzzy numbers, I ∼ and J ∼ . We can use tools provided in Chapter 12 on the extension principle to calculate the truth value of the assertion that fuzzy number I ∼ is greater than fuzzy number J ∼ (a fuzzy number is defined in Chapter 4):