Deviance probabilities: Determination of judgmental bias within Kendall’s coefficient of concordance data
Tóm tắt
Kendall’s coefficient of concordance is reviewed, with particular concern for nonsignificance. A statistic is presented that allows determination of whether one judge’s rankings come from the same population as the other judges’ rankings. This is ∑/D/, the sum of absolute differences in ranks that N objects receive from any two judges. Frequency distributions of ∑/D/, computed for N = 3-10 objects, were used to obtain the probability of a given /∑/’s being greater than a certain constant under the null hypothesis. This cumulative probability density function for incremental ∑/D/s is tabled. Obtained probabilities that one judge’s ranking came from the same population as all other judges’ rankings may be calculated from Bayes’ theorem. The statistic ∑/D/, associated probabilities, and comparison of one judge’s rankings with collective rankings are based on the raw data of Kendall’s W and operationally allow identification of judgmental bias.
Tài liệu tham khảo
Bradley, J. V. Distribution-free statistical tests (WADD Tech. Rep. 60-661). Wright-Patterson AFB, Ohio: Aerospace Medical Division, August 1960.
Buckalew, L. W., & Pearson, W. H. Determination of critical observed frequencies in chi square. Bulletin of the Psychonomic Society, 1981, 18, 289–290.
Buckalew, L. W., & Pearson, W. H. Critical factors in the chisquare test of independence: A technique for exploratory data analysis. Bulletin of the Psychonomic Society, 1982, 19, 225–226.
Ferguson, G. A. Statistical analysis in psychology and education (5th ed.). New York: McGraw-Hill, 1980.
Siegel, S. Nonparametric statistics for the behavioral sciences. New York: McGraw-Hill, 1956.
Uspensky, J. V. Introduction to mathematical probability. New York: McGraw-Hill, 1937.