Abstract

Goodman and Kruskal's T b statistic is frequently employed by sociologists to measure association among categorical polytomies. The test of the null hypothesis T b = 0 is presently limited to an asymptotic approximation that fails to adjust for unbalanced marginals, yielding inaccurate probability values. This article details a nonasymptotic test of significance that adjusts for unbalanced marginals.

Keywords

Kruskal's algorithmCategorical variableStatisticNull hypothesisStatisticsMathematicsTest statisticNull (SQL)EconometricsKruskal–Wallis one-way analysis of varianceStatistical hypothesis testingCombinatoricsComputer scienceData miningMann–Whitney U test

Affiliated Institutions

Related Publications

Much Ado About Nothing

Missing data are a recurring problem that can cause bias or lead to inefficient analyses. Development of statistical methods to address missingness have been actively pursued in...

2007 The American Statistician 759 citations

Publication Info

Year
1985
Type
article
Volume
13
Issue
4
Pages
543-550
Citations
34
Access
Closed

External Links

Social Impact

Altmetric

Social media, news, blog, policy document mentions

Citation Metrics

34
OpenAlex

Cite This

Kenneth J. Berry, Paul W. Mielke (1985). Goodman and Kruskal's TAU-B Statistic. Sociological Methods & Research , 13 (4) , 543-550. https://doi.org/10.1177/0049124185013004005

Identifiers

DOI
10.1177/0049124185013004005