Correlation

doi:10.1136/bmj.b2578

Published 1 July 2009, doi:10.1136/bmj.b2578
Cite this as: BMJ 2009;339:b2578

Endgames

Statistical quiz

Correlation

John Fletcher, clinical epidemiologist

¹ BMJ, London WC1H 9JR

jfletcher@bmj.com

The first 150 words of the full text of this article appear below.

Which of the following, if any, does not measure correlation?

a) Kendall’s ${tau}$

b) Spearman’s ${rho}$

c) Cronbach’s ${alpha}$

d) Pearson’s r

None—These are all measures of correlation. Kendall’s ${tau}$ is a non-parametric measure of correlation that involves measuringthe degree of correspondence in rank order between two variables.Spearman’s ${rho}$ is also a non-parametric measure of correlationbased on rank order. Both methods have the same statisticalpower and are based on the same underlying assumptions. Theirmagnitude is often different, however, because they are computedin different ways and measure different concepts. Kendall’s ${tau}$ is based on the probability that the rank orders are different,whereas Spearman’s ${rho}$ is based on the degree of variabilityin rank order of one variable accounted for by the other.

Spearman’s ${rho}$ is equivalent to Pearson’s (productmoment) correlation coefficient computed on rank rather thanvalue. Pearson’s correlation coefficient, r, is the mostfamiliar. It measures the degree of . . . [Full text of this article]

Add to CiteULike CiteULike Add to Complore Complore Add to Connotea Connotea Add to Del.icio.us Del.icio.us Add to Digg Digg Add to Reddit Reddit Add to StumbleUpon StumbleUpon Add to Technorati Technorati What's this?