The need to find effective ways to counter racial, ethnic, gender, and other forms of discrimination in the public work force has been an important item on our policy agenda for a considerable number of years. This concern has led to the pursuit of programs emphasizing equal employment opportunity and affirmative action and has given rise to a significant amount of empirical research. Studies have addressed questions related to the growth of minority and female employment, the impact of various approaches to equal opportunity, and the relationship between the employment of women and minorities and the implementation of policy outcomes consistent with the interests of those groups (Cayer and Sigelman, 1980; Cornwell and Kellough, 1994; Eisinger, 1982a, 1982b; Hindera, 1993a, 1993b; Kellough, 1989, 1990a, 1990b; Kellough and Rosenbloom, 1992; Kim, 1993; Lewis, 1988; Meier, 1975, 1993a, 1993b; Meier and Stewart, 1992; Mladenka, 1989a, 1989b; Nachmias and Rosenbloom, 1973; Riccucci, 1987; Rosenbloom, 1977, 1980, 1984; Saltzstein, 1979, 1986, 1992; and Stein, 1986). Answers to these important questions are tied to the nature of analytical techniques applied in the research and the general state of methodological development in public administration. One analytical device that can be useful in addressing issues associated with minority and female employment is the Measure of Variation (MV) introduced by Nachmias and Rosenbloom (1973). The MV is a measure of integration defined as heterogeneity along racial, ethnic, or gender lines. Shortly after its introduction, it was used in empirical work by Grabosky and Rosenbloom (1975) and Rosenbloom (1977), and more recently it has been applied in two other studies (Kellough 1990a; Kim 1993). As with any analytical technique, however, the MV can be misunderstood and misused. A case in point is the recent article by Guajardo (1996) that purports to assess the reliability and validity of the W. Because this work proceeds from an assumption that the MV is an indicator of minority representation, it is grounded on a basic misinterpretation of the measure. Furthermore, the study obscures the essential meanings of measurement reliability and validity through the use of a variety of exercises that are not well conceived and in most instances are not appropriate for the analysis undertaken. In short, Guajardo's assessment rests on a misapplication of several fundamental methodological concepts.(1) Such inaccuracies in the presentation of methodological techniques promote confusion and ultimately may adversely affect our ability to comprehend important issues. Understanding the MV Because Guajardo's difficulties derive in part from a concept misunderstanding, it is essential that we begin by setting out clearly what the MV index measures and how it operates. Rosenbloom (1977, 9) provides a particularly succinct description: [The MV] measures the heterogeneity of specified social characteristics found in a group of people and transforms this value into an index. It varies from 0 to 1.0, and the closer it Is to 1.0 the more equally represented are the groups under analysis. In the case of racial/ethnic heterogeneity, the MV is constructed by determining how many racially or ethnically divergent pairs of employees can be observed in an organization given the racial and ethnic composition of the work force and then dividing that number by the maximum number of such differences that could theoretically occur if all racial and ethnic groups were represented equally.(2) Mathematically, the numerator for the MV is expressed as: [summation of] [f.sub.i] [f.sub.j] where f equals the number of individuals with a particular social characteristic i, j and 1 [is not equal to] j. The denominator is represented by the following expression: n (n - 1) / 2 ([f.sup.2] / n) where n equals the number of different social characteristics considered (e. …