Researchers in the social sciences tend to analyze two dimensional frequency tables by means of Karl Pearson's Chi-square test of homogeneity. This can be improved upon if the dependent variable represents an ordered qualitative characteristic. In particular, one can substitute the Kruskal-Wallis analysis of variance on ranks for the more familiar Karl Pearson test. An example is provided of this modification for a planned and post hoc analysis.
Get full access to this article
View all access options for this article.
References
1.
Dagenais, F. and Marascuilo, L. A.Perception of the success of integration in a multiracial high school among social groups defined by social interaction. Journal of Educational Research, 1981, 75, 78-86.
2.
David, H. A.International Encyclopedia of the Social Sciences: Nonparametric Statistics. Volume 11, pp. 175-201. David L. Sills (Ed.) New York: Macmillan Company and the Free Press, 1968.
3.
Kendall, M. G.Rank Correlation Methods. Third edition. London: Griffin, 1962.
4.
Kirk, R. E.Experimental Design: Procedures for the Behavioral Sciences. Belmont, California: Brooks/Cole, 1968.
5.
Kruskal, W. H. and Wallis, W. A.Use of ranks in one-criterion variance analysis. Journal of the American Statistical Association, 1952, 47, 583-621.
6.
Lehmann, E.Nonparametrics: Statistical Methods Based on Ranks. San Francisco: Holden-Day, 1975.
7.
Marascuilo, L. A. and McSweeney, M.Nonparametric post hoc comparisons for trend. Psychological Bulletin, 1967, 67, 401-412.
8.
Maxwell, A. E.Analysing Qualitative Data. London: Methuen, 1961.
9.
Yates, F.The analysis of contingency tables with groupings based on qualitative characteristics . Biometrika, 1948, 35, 176-181.
