The role played by nonstatistical procedures, in contrast to multivariant statistical approaches, in analyzing faculty salaries is discussed. Multivariant statistical methods are usually used to establish or defend against prima facia cases of gender and ethnic discrimination with respect to faculty salaries. These techniques are not applicable, however, when the number of individuals is small or when it is necessary to draw conclusions concerning a single individual. The problem then becomes one of conducting an impartial investigation of a claim made by a single individual assigned to a rather small department. The technique of counterparting, or cohort analysis, can be used to show evidence of salary or promotion discrimination (or its absence) within a unit as small as a department. The idea behind counterparting in salary studies is that, for each individual to be studied, the researcher selects a group of people who are identical to that person with respect to all the commonly available characteristics that affect salary, except for the single characteristic on which the discrimination claim is based. Factors affecting faculty salaries are discussed, along with events that cause a salary change to occur. A hypothetical example of a salary case is included. (SW)