Conover, W. J., Iman, R. L. (1981). Rank transformations as a bridge between parametric and nonparametric statistics. American Statistician, 35, 124-129.  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... that these tests are not robust with respect to variance heterogeneity, cf. Lehmann (1975) Also, and more importantly, the null hypothesis for the rank transformed data may be no longer the same as for the original scale data, as pointed out by Fligner (1981) and Noether (1981) in comments on Conover and Iman (1981). Keselman and Wilcox (1999) propose to replace the hypothesis of homogeneity of the means by the homogeneity hypothesis of the trimmed means, but this concerns a di erent hypothesis. 5 We will make use of the following test statistics: i) ANOVA F test The F test, F , is given by F = N K K ....

Conover, W. J., Iman, R. L. (1981). Rank transformations as a bridge between parametric and nonparametric statistics. American Statistician, 35, 124-129.

....retain the solution quality improvements while imposing a lesser computational penalty. Table 1 : Experimental ACO results for 32 problems with and without look ahead information. We analysed our results to determine whether the quality improvements observed have statistical significance. Conover Iman s [1981] result allows us to use a parametric t test on the average ranking. The ranking was obtained from the 40 runs of the two algorithm versions for each problem. However, to avoid a too large number of ties, as is the usual practice, we have fourteen problems where 80 or more of the results ....

Conover W.J., Iman R.L., [1981], Rank transformations as a bridge between parametric and nonparametric statistics, The American Statistician, 35, 124-129.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC