Rank Correlation Methods

by the late Sir Maurice Kendall, Jean Dickinson Gibbons

The new edition of this classic reference includes a wealth of information on the mathematics and practical applications of rank correlation methods. It represents the most extensive revision since the book first appeared in 1948 and provides clear explanations of the methods of rank
correlation widely used by statisticians, educators, psychologists, and others involved in analyzing qualitative material. Among the many topics covered are the Goodman-Kruskal coefficient, partial rank correlation, trend tests, and regression techniques. Real numerical examples and problems are
taken from interesting research in the social sciences. Expanded tables have also been added that include larger sample sizes and a table of distribution of Kendall's partial law. Reflecting the vast amount of research published in recent years, this edition updates the references and offers over
300 new bibliographic entries. The valuable appendix tables have been revised and expanded.

272 pages, Hardcover

First published January 1, 1970

Reviews

[Douglas · Rating 4 out of 5]

This is, in my humble view, the go-to book for rank correlation methods. I like Kendall's style because he doesn't use a lot of jargon or dumbed-down verbosity; and his mathematical methods are more informal. His understanding is straightforward and not idiosyncratic as is the case with so many modern textbooks. You will not find over generalizations resulting from the application of abstruse mathematical subjects, another recent defect which gets in the way of pedagogy. Mathematics is a handmaiden, not a master, so you can get to stuff quickly.

This book, while it is still available, is nowhere as cheap as I paid for my copy; which, as I remember, I came across in an old used bookstore in Chicago that, for some reason, had a lot of math books. It wasn't near a university which was usually the case for such stores. It was run by an old gentleman who was quiet but helpful. His collection was in mild disarray but he knew, unlike sales people today, where everything was.

Apropos this, I remember going into a chain bookstore and inquiring of a young saleslady whether she had an unabridged edition of some famous and long work--maybe it was Don Quixote, I don't remember. Besides never having heard of the work, she asked me in that mousy-nasal voice that young females use today, "What's does unabridged mean?" It was hard to keep back some kind of comeback like, "Well, I see that you're just here to collect a paycheck, right? And why don't you talk like a mature woman...like...oh!...Betty Davis? Which you've probably never heard of but from whom you may learn something beneficial, when it comes to elocution, if you start watching American Movie Classics instead of Nickelodeon." Tempting as such a response might be we all know you can't say, in the face of such incompetence and unabashed indifference, stuff like that anymore! If you do they'll be the ones calling the manager! They've all been taught that they are "winners" no matter how poorly they performed; and they are even given awards. We now see the result...

Just to taunt the reader one more time, I think I paid five bucks or so for my copy! Alas! There are, today, probably more used video game shops in Chicago than used bookstores. And it shows! This is not good! I'm glad I'm a septuagenarian as I could not bear long life in a world continuing in such trends.

[Данило Судин · Rating 5 out of 5]

Праця класика і автора коефіцієнта кореляції τ, присвячена ранговим кореляціям як таким. Серед розглянутих коефіцієнтів, слід відзначити ρ Спірмена, τ Кендала, W конкордації та u згоди. Перші три активно застосовуються і в сучасних дослідженнях, а тому для тих, хто прагне зрозуміти "механіку" цих коефіцієнтів - варто прочитати саму книгу.
Структура її доволі зручна: розділи розпадаються на дві великі групи. До першої потрапив "екстракт" висновків, а до другої - власне доведення основних тез автора. А тому читач, що не має особливого бажання чи достатнього терпіння або й математичної підготовки, може розділи другої групи опустити, зосередившись на основних здобутках автора.
Мова Моріса Кендала проста і зрозуміла: він один з класиків, що пишуть просто і доступно. Окрім того, в праці детально проговорюються деякі важливі питання рангових кореляцій, які - навіть в сучасних посібниках - є маргінальними або й відсутніми. (Наприклад, питання лінійності рангових кореляцій чи множинних рангових кореляцій). А тому ця книга, попри свій вік (майже 50 років) є досі актуальною для всіх дослідників, що послуговуються статистичними методами в своїй діяльності.