Consider a finite set of alternatives, and an associatedcollection of random variables representing some given relevant property of eachalternative, such as its utility, cost or reliability. Suppose that there is a processthat selects exactly one of these alternatives, an alternative with an extreme(maximal or minimal) value. In economics, discrete-choice analysis based onrandom utility theory is the most well-known modelling approach to such aprocess. Among practitioners, it is widely believed that in order to analyze suchsituations, one needs to resort to particular parametric forms of the underlyingprobability distributions of the random variables. We show that this is notso. Indeed, parametric forms impose unnecessary theoretical and empiricalconstraints and may hide more general properties. For the special case ofstatistical independence, we further provide a characterization of the invarianceproperty that all random variables, representing the relevant property of thealternatives, have the same distribution, conditional upon being chosen.