Various measures of variation for categorical data which have been introduced suffer from the limitation that their numerical values generally appear to be unreasonable. This may cause their use to give misleading results and poor data discrimination. In this paper, two coefficients of variation for categorical data which seem to have appropriate numerical properties are introduced. One of the coefficients is appropriate for nominal data and one for ordinal data.
References
1.
AgrestiA.Analysis of ordinal categorical data. New York: Wiley, 1984.
2.
AllisonP. D.Measures of inequality. American Sociological Review, 1978, 43, 865–880.
3.
BerryK. J.MielkeP. W.Jr.Assessment of variation in ordinal data. Perceptual and Motor Skills, 1992, 74, 63–66. (a).
4.
BerryK. J.MielkeP. W.Jr.Indices of ordinal variation. Perceptual and Motor Skills, 1992, 74, 576–578. (b).
5.
KvålsethT. O.Nominal versus ordinal variation. Perceptual and Motor Skills, 1989, 69, 234. (a).
6.
KvålsethT. O.Odds measure of qualitative variation. Perceptual and Motor Skills, 1989, 68, 830. (b).
7.
KvålsethT. O.Range measure of qualitative variation. Perceptual and Motor Skills, 1989, 68, 1282. (c).
