Summary
The functionϕ n (S) is linearly related tof n (r), the frequency distribution ofSpearman's coefficient for rank correlation. Exact combinatorial results forn≦8 and an asymptotic approximation are available inKendall [3].
Since the combinatorial approach is intractable forn>10 the principal purpose of this paper was to studyϕ n (S) forn>10 by the Monte Carlo method. The conclusion was reached that the asymptotic approximation is satisfactory forn>20.
Zusammenfassung
Die Funktionϕ n (S) hängt linear mitf n (r), der Verteilung desSpearmanschen Rangkorrelationskoeffizienten zusammen. Exakte Ergebnisse fürn≦8 und eine asymptotische Näherung sind inKendall [3], zu finden.
Da exakte Werte fürn>10 wegen des Rechenaufwands schwer zu erhalten sind, ist es das hauptsächliche Ziel dieses Artikels,ϕ n (S) fürn>10 mit einem Monte-Carlo-Verfahren zu berechnen. Wir sehen, daß die asymptotische Näherung fürn>20 ausreichend genau ist.
Similar content being viewed by others
References
Durstenfeld, R.: Collected Algorithms from CACM, p. 235. San Diego, California: General Atomic. 1964.
Hall, M. Jr., andD. Knuth: Combinatorial Analysis and Computers (Part 2). Amer. Math.M 72, 21 (1965).
Kendall, M.: The Advanced Theory of Statistics, Vol. 1, 2nd ed., p. 388. London: Charles Griffin and Co. 1958.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Maghsoodloo, S. An investigation by high speed sampling of the frequency distribution of rank correlation. Computing 8, 1–12 (1971). https://doi.org/10.1007/BF02234038
Received:
Issue Date:
DOI: https://doi.org/10.1007/BF02234038