A new statistic for testing uniformity of ranking data
COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION, vol.25, no.1, pp.139-156, 1996 (SCI-Expanded, Scopus)
- Publication Type: Article / Article
- Volume: 25 Issue: 1
- Publication Date: 1996
- Doi Number: 10.1080/03610919608813303
- Journal Name: COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION
- Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
- Page Numbers: pp.139-156
- Keywords: n rankings, concordance, average Kendall's tau, average Spearman's rho, Rao's diversity coefficient, Monte Carlo experiments, power comparisons
- Dokuz Eylül University Affiliated: Yes
Abstract
In this paper, as an alternative to both the average Kendall tau and average Spearman's rho, we propose a new statistic, tau*, to test the null hypothesis of complete randomness amongst the n judges each rank r objects independently. We define tau* based on Rao's ideas of the diversity coefficient of a population. Monte Carlo experiments we made show that the test based on the proposed statistic has better and desirable power properties relative to competing tests.